wnt10a is required for zebrafish median fin fold maintenance and adult unpaired fin metamorphosis

Mutations of human WNT10A are associated with odonto‐ectodermal dysplasia syndromes. Here, we present analyses of wnt10a loss‐of‐function mutants in the zebrafish.

wnt10a belongs to the family of wingless-type WNT genes, encoding secreted cysteine-rich signaling molecules which are highly conserved in all metazoan animals. 1 WNT proteins regulate cell-to-cell interactions and play an important role both in multiple developmental processes during embryogenesis and in homeostasis in adult tissues. 2 Canonical WNTs function by inhibiting the β-catenin degradation complex, thereby allowing interaction with the nuclear transcription factors LEF/TCF and regulation of target gene expression.Mouse Wnt10a was shown to be expressed in the inner enamel knot area of the tooth bud 3 and at later developmental stages in the mesenchymal pre-odontoblast layer, 4 in skin and in placodes at the onset of hair follicle morphogenesis. 5Furthermore, chicken Wnt10a was shown to be involved in the formation of the apical ectodermal ridge (AER) during limb development. 6utations in human WNT10A underlie multiple forms of ectodermal dysplasia (ED), such as odontoonycho-dermal dysplasia (OODD), a rare autosomal recessive syndrome in which the presenting phenotype is dry hair, severe hypodontia, smooth tongue with marked reduction of fungiform and filiform papillae, onychodysplasia, keratoderma, and hyperhidrosis 7 ; Schöpf-Schulz-Passarge syndrome (SSPS), 8 which shows clinical overlap with OODD, but includes eyelid cysts as a typical sign of SSPS; and some cases of hypohidrotic ED.
The most consistent symptom in all human cases is severe oligodontia of permanent teeth. 9Therefore, various animal models have been used to study the molecular mechanism of WN10A during tooth formation, including zebrafish, in which antisense-mediated knockdown of wnt10a arrests tooth development at 5 days post fertilization (dpf), recapitulating the severe hypodontia phenotype observed in humans. 10Zebrafish Wnt10a has also been suggested, although not proven via lossof-function studies, to be the relevant ligand activating essential Wnt/β-catenin signaling during regeneration of the adult tailfin. 11Furthermore, Wnt/β-catenin signaling, activated by unspecified ligands, was shown to be necessary for regular outgrowth of adult fins in zebrafish 12 and for epithelial patterning and growth in the zebrafish embryonic median fin fold (MFF). 13,14ere, we identify Wnt10a as being essential both for the maintenance of the larval MFF and for the later, developmentally independent 15 formation of the adult median fins.The wnt10a mutant had been isolated in a phenotype-based forward genetics screen after random chemical mutagenesis and deposited in the ZFIN database as unnamed gene/mutant unm_t30922, displaying embryonic MFF defects.We show that wnt10a mutant embryos initiate MFF formation normally, but later lose specific properties of distal-most MFF cells, causing actinotrichia shrinkage and fin collapse.These findings provide new insights into Wnt10a function in the developing vertebrate skin and its appendages, including thus far unknown roles in the control of the linear growth and stabilization of actinotrichia/collagen fibrils.

| Zebrafish unm_t30922 mutants display embryonic MMF collapse and compromised adult fin formation
During a phenotype-based forward genetics screen after mutagenesis with ethylnitrosourea (ENU) performed within the frame of the Integrated Project "ZF-MODELS -Zebrafish Models for Human Development and Disease", the recessive embryonic mutant unm_t30922 t30922 (TU) was isolated, characterized by "variably absent ventral median fin" (https://zfin.org/action/genotype/view/ZDB-GENO-060608-54).Here, we have analyzed the mutant fin phenotype in more detail.Although mutant embryos form a regular MFF until 35 hours post fertilization (hpf) (Figure 1A; mutants identified via genotyping; see below), a clear, fully penetrant, and generally invariable phenotype in MFF structure is evident from 48 hpf onwards (Figure 1B).The ventral major lobe is slightly collapsed at this time point, while the minor lobe (also named pre-anal fin fold, PAFF 16 ) on the yolk sac extension is substantially collapsed.This collapse continues and at 6 dpf (Figure 1C), only spike-like remnants of the minor lobe (Figure 1D,D'') and the major lobe remain (Figure 1D',D''').In contrast, the paired pectoral fins do not seem to be altered in the mutant.The cleithrum, scapulocoracoid, and endoskeletal disc cartilages stained with alcian blue as well as the apical fin folds are present and of normal shapes and sizes in the mutant pectoral fin (Figure 1E).During metamorphosis, the embryonic and larval MFF either regresses or develops further into the fins typical for juvenile and adult fish, which in contrast to the purely actinotrichia-based (see below) embryonic and larval fin folds contain bony rays of lepidotrichia. 17t the onset of metamorphosis (13 dpf, standard length [SL] 6 mm), the MFF of mutants has completely disappeared, while remnants are still present in their siblings (Figure 1G).In addition, the caudal fin is strongly reduced (Figure 1G), as also evident after staining of lepidotrichia with alizarin red (Figure 1F), and in contrast to the rather normal morphology of the caudal fin folds at embryonic and early larval stages (Figure 1C,D',D''').After completion of metamorphosis (26 dpf, SL 11 mm), F I G U R E 1 Characterization of fin defects in unm_t30922 mutants.(A) unm_t30922 mutant embryos are initially identical to their siblings at 35 hpf and form regular median fin folds.(B) Both the minor and major lobes of the MFF have started to collapse at 48 hpf.(C-D 000 ) At 6 dpf, the MFF of the sibling has extended (C, D, D 0 ) while most of the mutant MFF has collapsed (C, D 00 , D 000 ) apart from small remnants visible in the minor lobe (arrowheads in D 00 ) and large parts of the tail fin fold (D 000 ). (E) Alcian blue stainings reveal that at 6 dpf, the pectoral fins of siblings and mutants are indistinguishable and have regular cleithrum (CL), scapulocoracoid (SC), endoskeletal disc (ED) cartilages, and a regular fin fold (FF).(F, G) Alizarin red stainings of mineralized bones (F) and images of live fish (G) reveal that at the onset of metamorphosis (15 dpf, SL 6 mm) wnt10a mutants only have remnants of caudal fins with disorganized lepidotrichia (F, G), while the minor and all other parts of the major MFF lobes have disappeared (G).Arrows in the wild-type fish point to the forming unpaired anal (af), dorsal (df), and caudal (cf) fin.(H) As juveniles (26 dpf, SL 11 mm; osterix:nuGFP transgenic 18 to label bone-forming cells) the mutants lack the unpaired dorsal (df) and anal (af) fin, while the unpaired caudal fin (cf) is strongly reduced, containing only few lepidotrichia.In comparison, the paired pectoral fins (pecf) and pelvic fins (pelf) are little affected.In addition, scales are formed as in their siblings (enlarged inserts with dashed lines).Arrowheads indicate collapsed fin folds and remnants of adult fins.MFF, median fin fold; SL, standard length.
the caudal fin of the mutants is even further reduced, with very few and shorter lepidotrichia, as also revealed by the osterix:nuGFP transgene (official name Tg(OlSp7: nlsGFP) zf132 18 ) labeling mature osteoblasts in lepidotrichia as well as in craniofacial and axial bones and in developing scales 12,19 (Figure 1H).In addition, the dorsal and anal fin, both unpaired fins which replace the larval MFF 17 are completely absent in the mutants (Figure 1H).In contrast, the paired pectoral fins, which have already been set up during embryogenesis, and the pelvic fins, paired fins that develop during metamorphosis, are only moderately affected in the unm_t30922 t30922 mutants at such juvenile stages (Figure 1H), while scales, dermal derivatives, which like most of the adult fins are formed during metamorphosis, 20 appear completely normal (dashed inserts in Figure 1H).

| unm_t30922 t30922 mutants display a mutation in wnt10a leading to a C-terminal truncation of the protein
To our knowledge, this combination of phenotypic traits in embryonic and adult fins of unm_t30922 mutants is unique and has not been reported for any other zebrafish mutant as yet.To identify the causative mutation, we applied a combination of standard meiotic mapping (with a total of 700 mutant embryos) and whole-genome sequencing (WGS) from a pool of 20 mutant embryos (see Experimental Procedures for more details).Both approaches revealed an interval on chromosome 9 (0.429 cM according to the meiotic mapping, approximately 4 Mb according to WGS) as the mutation-bearing genomic region (Figure 2A-C).According to the WGS data, this region in the mutant pool is homozygous for a point mutation (A>G) in the splice acceptor site in front of exon 4 of wnt10a (Figure 2C,D).The mutation leads to false transcript splicing.Sequencing of amplified wnt10a cDNAs from mutant embryos revealed three falsely spliced isoforms, one using an alternative splice acceptor site in the intron between exon 3 and exon 4, and two using alternative splice acceptor sites within exon 4, causing frame shifts (Figure 2D).In all three cases, this leads to the loss of the C-terminal 165 amino acid residues of the Wnt10a protein (normal size: 442 aa), including large parts of the conserved Wnt domain (aa 100-442; Figure 2D).Furthermore, quantitative reverse transcription polymerase chain reaction (qRT-PCR) showed that in mutant embryos wnt10a transcript levels are significantly reduced at 2 and 5 dpf (Figure 2E), pointing to non-sense mediated decay.Together, these data suggest that unm_t30922 is a functional wnt10a null allele.

| wnt10a mutants display impaired tooth development
A frequently described phenotype caused by the loss of wnt10a function in human 7,9,10,[21][22][23] is oligodontia/hypodontia, a congenital condition characterized by fewer than normal teeth.Loss of pharyngeal teeth has also been described for zebrafish larvae in which wnt10a transcripts were blocked with antisense morpholino oligonucleotides. 10As a first test to functionally validate the identified wnt10a mutation in our unm_t30922 t30922 mutant, we studied whether homozygotes recapitulate the oligodontia observed in wnt10a morphants.Alizarin redstained wnt10a mutant and sibling embryos were prepped and mounted for microscopic analysis.Zebrafish teeth are different from other vertebrate teeth because they are not attached to the oral jaw, but to the left and right fifth ceratobranchial arch. 24The first teeth to develop are the five ventral row (V) teeth (from 2 to 16 dpf), followed by the four mediodorsal row (MD) teeth (from 14 to 24 dpf), and finally the two teeth of the dorsal row (D) (from 24 to 28 dpf). 25At 6.5 dpf, the wnt10a mutant zebrafish larvae have normally developed 5th ceratobranchial arches (Figure 3A).However, in contrast to wild-type siblings, which already have three teeth (4V, 3V, and 5V 26 ) on each side, only one tooth is present in the mutants (Figure 3A).This is consistent with the larval tooth phenotype previously described for zebrafish wnt10a morphants. 10As adults, wnt10a mutant fish have no teeth at all and an underdeveloped fifth ceratobranchial arch (Figure 3B).
To gain first insights into the molecular and cellular mechanisms underlying the essential role of Wnt10a during zebrafish tooth development, we studied the expression of wnt10a and other known mammalian tooth regulators in wild-type and mutant embryos.8][29][30][31] First, the odontogenic epithelium thickens and starts to express several signaling molecules to induce first specification steps and a condensation of the underlying mesenchyme.The mesenchyme in turn signals back to the epithelium, which folds into itself to form a bell-like shape while growing downwards into the mesenchyme.These interactions eventually lead to the differentiation of rows of ameloblasts and odontoblasts at the epithelial-mesenchymal interface to deposit the first tooth matrix and to form the dentine cone.
In mice, Wnt10a is known to play an important role as a signaling molecule during such epithelialmesenchymal interactions to induce odontoblast differentiation and tooth morphogenesis. 4,29erforming whole-mount in situ hybridization of zebrafish embryos at 56 hpf, when the dental epithelium and underlying mesenchyme of the first forming pharyngeal tooth (4V) have started to interact, 24,25,31 we observed wnt10a expression both in the dental epithelium, characterized by co-expression of the epithelial marker p63, and in the dental mesenchyme.However, expression in the dental mesenchyme appears stronger (Figure 3C,D).Of note, also in mice, Wnt10a has been reported to be expressed in both compartments, however, in a consecutive manner, starting in the dental epithelium 3 and then shifting to the underlying mesenchyme to be continuously expressed in the developing odontoblasts. 4,32To identify the zebrafish tooth germ cells that respond to such Wnt10a signals, we took advantage of the formerly published transgenic canonical Wnt signaling responder line Tg(6xTCF:eGFP) (official name Tg(6xTcf/ LefBS-miniP:d2EGFP) isi01 33 ).GFP in situ hybridization at 56 hpf revealed strong Wnt response within the tooth germ (Figure 3E), which was particularly strong in the dental mesenchyme, but also present, although much weaker, in the surrounding dental epithelium (Figure 3F).Thus, Wnt10a might signal both across epithelial and mesenchymal compartments of the tooth germ, as well as, in a more autocrine manner, within the compartments themselves and in particular within the dental mesenchyme.
To get first insights into the correlation between Wnt10a and other known regulators of tooth development shared between mammals and fish, we also investigated the expression of ectodysplasin-A (Eda), a type II membrane protein of the Tumor necrosis factor ligand family, 34,35 the expression of fgf4, a member of the fibroblast growth factor family 28,31,36,37 and the expression of dlx2b, a Distalless-related Dlx2 homeodomain transcription factor 31,[38][39][40] in zebrafish wnt10a mutants in comparison to their wild-type siblings.Although eda 35 expression was reported to be directly correlated with wnt10a expression in whole zebrafish embryos at 48 hpf, 10 we found indistinguishable eda expression patterns and levels in the dental epithelium and dental mesenchyme of wnt10a mutant and their siblings at 56 hpf (Figure 3G).This is consistent with data obtained in mice showing expression of Eda in the dental epithelium at early stages of tooth development to be independent of Wnt/β-catenin activity 41 and suggests that Eda might act upstream or in parallel, rather than downstream of Wnt10a signaling.In contrast, we found both fgf4 and dlx2b 31 to be strongly down-regulated in the dental mesenchyme (fgf4; Figure 3H,I) and in the dental epithelium and dental mesenchyme (dlx2b; Figure 3J), respectively, suggesting that both act downstream of Wnt10a.This is consistent with former data obtained in mice, according to which FGF4 is a direct target of canonical Wnt signaling and its transcription factor LEF1. 37 dlx2b might in turn be a target of Fgf4 signaling, as in zebrafish, its expression has been reported to be strongly compromised upon pharmacological inhibition of Fgf signal reception. 31aken together, these striking similarities of the tooth phenotype between zebrafish unm_t30922 t30922 mutants, zebrafish wnt10a morphants, and mammalian Wnt10a mutants strongly suggest that the identified wnt10a mutation of the unm_t30922 t30922 mutants is causative of their abnormal tooth development.

| wnt10a is required for maintaining embryonic MFF structure and for adult fin formation during metamorphosis
To validate that the identified wnt10a mutation is also causative of the MFF phenotype of the unm_t30922 t30922 mutant, three different approaches were applied.First, we performed wnt10a knockdown with an antisense morpholino oligonucleotide (MO), as formerly used to generate the aforementioned tooth defects, 10 yielding MFF collapse from 2 dpf onwards (Figure 4A).Second, we used the well-characterized transgenic Tg(hsp70l: dkk1-GFP) w32Tg 11 line to temporarily inhibit Wnt signaling by overexpressing Dkk1 (Dickkopf1), an antagonist of the Wnt/β-catenin pathway. 42A single heat-shock of transgenic embryos at 24 hpf was sufficient to phenocopy the ventral and dorsal MFF collapse seen in the wnt10a mutants at 3 dpf (Figure 4B), indicating the involvement of Wnt/β-catenin signaling in the maintenance of the MFF structure.Finally, the MFF phenotype was rescued by generating a stable transgenic Tg(hsp70: wnt10a) fish line allowing temporally controlled forced expression of wild-type wnt10a.Functionality of this line was confirmed by heat-shocking transgenic embryos at 12 hpf for 1 h, which resulted in the loss of anterior neural identities (and an eyeless phenotype; data not shown), F I G U R E 3 unm_t30922 t30922 mutants display impaired tooth development.(A) At 6 dpf, the wnt10a siblings have alizarin red-stained 3V 1 , 4V 1 , and 5V 1 teeth on either bilateral ceratobranchial 5 arch (cb5), while the unm_t30922 t30922 mutant either has one tooth or no teeth on either side of an otherwise normally formed cb5.The cleithra (CL) of the mutant is indistinguishable from those of the siblings.(B) At 35 dpf (SL 13 mm), dissected cb5 of siblings have fully formed alizarin red-stained mineralized teeth (only 1V, 2V, 3V, 4V, and 5V are visible in this view) while the unm_t30922 t30922 mutants have no teeth and an underdeveloped cb5.Arrows indicate regular teeth, arrowheads compromised or absent teeth.Scale bars = 100 μm.(C) Dorsal view of wild-type embryo at 56 hpf, after wnt10a in situ hybridization shows that wnt10a is expressed in the tooth germs (indicated by arrows).(D) Transverse cryosection of 56 hpf wild-type embryo after wnt10a in situ hybridization and p63 immunostaining, revealing strong wnt10a expression in the dental mesenchyme (dm; white arrow) and weak wnt10a expression in the surrounding p63-positive dental epithelium (de; white arrow).(E) Dorsal view of Tg(6xTCF:eGFP) embryo at 56 hpf after transgene-encoded GFP in situ hybridization, revealing the reception of canonical Wnt signals in the tooth germs (arrows).(F) Transverse cryosection of Tg(6xTCF:eGFP) embryo at 56 hpf after transgene-encoded GFP in situ hybridization and p63 immunostaining, revealing strong canonical Wnt signal reception in the dental mesenchyme (dm; white arrow) and weak wnt10a expression in the surrounding p63-positive dental epithelium (de; white arrow).(G, H) Durcupan transverse sections of in situ hybridized unm_t30922 t30922 sibling and mutant embryos at 56 hpf.eda is strongly expressed in the dental mesenchyme (dm, arrow) and weakly expressed in the dental epithelium (de, arrow) of both the unm_t30922 t30922 sibling and mutant tooth germ (G).fgf4 is strongly expressed in tooth germ (arrow) of the unm_t30922 t30922 sibling but not in the tooth germ of the unm_t30922 t30922 mutant (arrowhead) (H).(I) Transverse cryosection of unm_t30922 t30922 sibling at 56 hpf after fgf4 in situ hybridization and p63 immunostaining, revealing strong fgf4 expression in the p63-negative dental mesenchyme.(J) Dorsal views and transverse Durcupan sections of in situ hybridized unm_t30922 t30922 sibling and mutant embryos at 56 hpf revealing strong dlx2b expression in the dental mesenchyme (dm) and dental epithelium (de) of the sibling (arrows), but an almost complete down-regulation in the mutant (arrowhead).de, dental epithelium; dm, dental mesenchyme; n, notochord; pecf, pectoral fin; y, yolk sac.comparable to the effect formerly reported for ectopic activation of other components of the canonical Wnt pathway in zebrafish. 43Daily heat-shocks of wnt10a mutants carrying the hsp70:wnt10a transgene from 24 hpf onwards rescued the 3 dpf MFF phenotype almost completely, while heat-shocks from 48 hpf onwards led to a partial rescue (Figure 4C).This demonstrates that the essential role of Wnt10a for MFF stabilization, as evident at 3 dpf, is initiated after 24 hpf and not fully completed at 48 hpf as yet.This early Wnt10a function also affects the MFF at much later developmental stages.Thus, forced transgenic expression of wnt10a from 24 hpf through 6 dpf, while not affecting the fins of wild-type fish (data not shown), perfectly rescued MFF morphology of wnt10a mutants even at 14 dpf (SL 5 mm), whereas MFF morphology was not fully recovered upon forced wnt10a expression from 48 hpf until 6 dpf (Figure 4D).The picture, however, changes dramatically for processes of fin development that occur during metamorphosis, starting after the second week of zebrafish development. 17Thus, wnt10a mutants with forced wnt10a expression from 24 or 48 hpf through 6 dpf, which displayed full or partial rescue of the embryonic MFF at 14 dpf (Figure 4D), failed to form adult fins at 31 dpf (SL 9 mm), with phenotypes similar to those of untreated wnt10a mutants (Figure 4E).Taken together, these rescue experiments indicate that wnt10a fulfills two sequential functions during different steps of fin development, which are largely independent of each other 15 : first from 1 to 6 dpf, promoting the stability of the embryonic MFF, and second during metamorphosis from 14 to 31 dpf, promoting adult fin morphogenesis.

| Wnt signaling is active during embryonic MFF formation and adult fin growth
We next aimed to determine which MFF cells express and which cells receive Wnt10a during the different stages of larval and juvenile MFF formation.During the initial phases of larval MFF development, basal keratinocytes of the bi-layered embryonic epidermis start to change their shapes from elongated to wedge-shaped profiles.8][49] Subsequently, adjacent epidermal cells on both sides of the MFF progressively lose their contact with the underlying mesoderm and instead attach to epidermal cells on the other side via cross-fibers linking the two opposing epidermal basement membranes, thereby creating a narrow sub-epidermal/dermal space and contributing to further MFF erection. 471][52] A prominent role during such early stages of MFF formation and along its entire proximodistal height (over 10 cell diameters 47 ) is played by the distal-most epidermal MFF cells, the central cleft cells and their immediate neighbors, the ridge cells.Thus, cleft and ridge cells show strongest expression of genes encoding epidermal basement membrane components such as lama5, 53 epidermal-dermal anchoring proteins such as fras1 and frem2a, 49 dermal cross fiber components such as fbn2 49 and actinotrichia components such as col1a1a, 50,51 with secreted extracellular matrix (ECM) proteins made by such distal cells also ending up in more proximal MFF domains. 49In addition, cleft cells contain a narrow and lengthy invagination (the cleft), which accommodates the distal ends of the actinotrichia, thereby possibly contributing to distal actinotrichia anchorage. 47nitially, the dermal space containing the actinotrichia is cell-free 47 ; however, fibroblast-like fin mesenchymal cells soon use actinotrichia as a migration substrate 50,54 to migrate outwards into this space, further contributing to actinotrichia and MFF stability. 52ia whole-mount in situ hybridization, wnt10a transcripts were detected in MFF fin mesenchymal cells at 28 hpf when they start to migrate distally into the newly created subepidermal space (Figure 5A,A 0 ).Fin mesenchymal cells continued to express wnt10a at 56 hpf (Figure 5B,B 0 ), when MFF collapse in mutants has become evident (Figures 1 and 4).Later during development, when the larval caudal fin starts to transit into the adult MFF, one of the processes of zebrafish metamorphosis 17 (Figure 4D), wnt10a expression was also identified in mesenchymal cells within the dermal space of the outgrowing caudal fin at 13 dpf (SL 5 mm) (Figure 5C,C 0 ).However, at no stages, we were able to detect wnt10a expression in MFF epidermal cells.Note that despite being unaffected in wnt10a mutants (Figure 1E), pectoral fins also express wnt10a at 56 hpf (Figures 3C and 5B).
To identify which MFF cells respond to Wnt10a signals, we analyzed the GFP fluorescence pattern in the Tg (6xTCF:eGFP) Wnt signaling responder line. 33At 48 hpf, GFP fluorescence can be seen in distal epidermal cells of the MFF (Figure 6A; white arrows) as well as in fin mesenchymal cells (Figure 6A, orange arrow), characterized by their arborized shape and the co-expression of RFP encoded by the GBT (gene-breaking transposon)-targeted msxc mn0245Gt locus 55 (Figure 6E).Cryosectioning of the tail of a 48 hpf 6xTCF:eGFP transgenic embryo and staining with antibodies against GFP and collagen II, one of the collagens that actinotrichia are comprised of, 51 further identified the distal-most cells as the only epidermal MFF cells responding to canonical Wnt signals (Figure 6D).This suggests that Wnt10a from fin mesenchymal cells, in addition to acting in an autocrine fashion, signals to distal-most cells of the MFF to account for embryonic and larval MFF maintenance.Of note, loss of wnt10a function in mutant embryos leads to reduced expression, and gain of wnt10a function upon Tg(hsp70: wnt10a)-driven wnt10a overexpression leads to increased expression of the Tg(6xTCF:eGFP) canonical Wnt signaling reporter.However, in both cases, responses are restricted to the minor and major lobes of the embryonic MFF, whereas the caudal and pectoral fin folds display comparably normal canonical Wnt signaling levels (Figure 6B,C).These differential responses are fully consistent with the spatially restricted fin fold collapse caused by the wnt10a mutation, affecting only the minor and major lobes of the embryonic MFF that are under essential control of Wnt10a, but not the caudal and pectoral fin folds (Figure 1C,D).The latter might be under the control of other canonical Wnt signals, which apparently cannot be replaced by Wnt10a (see Discussion).Why the increase of canonical Wnt signaling caused in the minor and major MFF lobes by Wnt10a gain-of-function is not accompanied by any morphological alterations remains unclear.Apparently, other genetic, molecular, or cellular constrains avoid the formation of larger embryonic MFF lobes.
During later steps of fin formation and at the beginning of metamorphosis, canonical Wnt signaling can be detected in an area distal to the forming lepidotrichia (stained with alizarin red) of the caudal fin (Figure 6F), where fin ray-organizing actinotrichia and actinotrichiagenerating epidermal and mesenchymal cells are located 56 and where wnt10a is expressed (Figure 5C).Of note, fully grown caudal fins in adults have been reported to lack canonical Wnt signaling in their distal regions. 57aken together, our results indicate that in the MFF of zebrafish embryos and larvae canonical Wnt signaling takes place during two distinct developmental periods, in line with the two phases of Wnt10a requirement for larval MFF maintenance and juvenile MFF metamorphosis described above (Figure 4C,D).

| MFF collapse in the wnt10a mutant is not caused by increased cell death or decreased cell proliferation
In mice, down-regulation of Wnt10a has been shown to inhibit cell proliferation and to increase apoptosis rates in embryonic palatal mesenchymal cells 58 as well as epithelial progenitor cell proliferation. 59To investigate whether similar mechanisms might underlie the MFF collapse of wnt10 mutant zebrafish embryos, we quantified cell death rates via terminal deoxyribonucleotide transferase-mediated dUTP-biotin nick end labeling (TUNEL) of wnt10a sibling and mutant embryos at 48 and 56 hpf, time points when the collapse of the MFF has already become apparent and is progressing (see Figure 1B).However, no significant difference in TUNEL-positive cells was found between wnt10a siblings and mutants (Figure 7A,B and data not shown).Consistently, pharmacological inhibition of caspase-3, an initiator of apoptosis, 60 with Z-DEVD-FMK and inhibition of p53, a tumor suppressor blocking cell cycle progression and promoting apoptosis, 61 with Pifithrin-α, from 24 to 72 hpf did not alleviate the fin fold collapse seen in wnt10a mutants (Figure 7C).Taken together, this indicates that the MFF collapse is not caused by increased cell death.To examine whether wnt10a mutants might have impaired cell proliferation, we treated embryos with Bromodeoxyuridine/5-bromo-2 0 -deoxyuridine (BrdU) from 33 to 35 hpf and fixed them at 37 hpf to identify cells undergoing DNA replication.Since the mutants could not be identified phenotypically at this early stage, BrdU incorporation was only detected in the dissected trunk and tail regions, while heads were used for genotyping.wnt10a mutants did not display fewer BrdU-positive cells in the ventral region of the major lobe at 37 hpf (Figure 7D).Unaltered BrdU incorporation rates were also obtained for mutant embryos at 48 hpf, when fin collapse started to become morphologically obvious, and after BrdU treatment from 45 to 46 hpf (Figure 7E,F).
Taken together, these data show that collapse of the MFF in wnt10a mutants is neither caused by increased cell death nor by impaired cell proliferation.

| MFF collapse of wnt10a mutants is preceded and accompanied by reduced expression of distal marker and ECM genes
Given that according to our aforementioned data obtained with the Tg(6xTCF:eGFP) line, two cells types within the developing MFF of 48 hpf wild-type embryos display the highest responses to canonical Wnt signals, the cleft and ridge cells of the distal AER and the fin mesenchymal cells (Figure 6), we investigated these two cell types in wnt10a mutant and sibling embryos with another transgenic marker, the Et(krt4:eGFP) sqet37 . 62,63At 33 hpf, before fin collapse, transgenic GFP signals in both distal epidermal MFF cells and fin mesenchymal cells are of indistinguishable strengths in wnt10a mutants and their siblings (Figure 8A).In addition, fin mesenchymal cells, which have just started to migrate distally into the newly created subepidermal space of the fin, are present in normal numbers and displayed normal shapes in the mutant (Figure 8A).At 48 hpf, however, when fin collapse has started to become evident, the mutants still have median fin mesenchymal cells of normal numbers, shapes and transgene expression levels (Figure 8A and, for cryosections, Figure 8B), whereas in MFF cleft and ridge cells of the mutants, transgene-encoded eGFP fluorescence is strongly reduced (Figure 8A,B).This suggests that fin mesenchymal cells, although normally displaying strong wnt10a expression and direct reception of canonical Wnt signals (Figures 5 and 6), are not affected by the loss of Wnt10a, whereas distal-most epidermal MFF cells are.
To further characterize these distal-most epidermal MFF cell defects, we performed in situ hybridizations at 33, 48, and 56 hpf for transcripts of known cleft and/or ridge cell marker genes.At 33 hpf, the MFF of wnt10a mutants displays strong down-regulation of the expression of the distalless-related gene dlx2a (Figure 8C), strikingly similar to the strong reduction of the expression of its paralog dlx2b in the tooth germ (Figure 3J).In light of the known essential role of the Drosophila distalless gene essential for distal appendage patterning in insects, 64 this suggests that zebrafish wnt10a is required for the specification of distal-most fates in epidermal MFF cells during the earliest stages of MFF development.This compromised specification seems to have specific consequences for the expression of some, but not all genes encoding ECM proteins involved in MFF morphogenesis.Thus, fras1, encoding a protein made by cleft cells and required for proper anchorage of the MFF basement membranes to the subepidermal dermal space, 49,65 is normally expressed in cleft cells of 33 hpf wnt10a mutants (Figure 8D), whereas the expression of its close relative frem2a, which normally is strongly expressed both in cleft and ridge cells to contribute to epidermaldermal anchorage, 49,65 is strongly reduced in 33 hpf wnt10 mutants (Figure 8E).col1a1, encoding an essential fibrillar collagen of the actinotrichia located within the dermal space of the MFF, 51,66 is normally expressed in distal MFF cells of wnt10a mutants at 33 hpf (Figure 8F), but-despite the lack of increased cell death rates at these stages (Figure 7)-become progressively reduced compared to their wild-type siblings at 48 hpf (Figure 8G) and 56 hpf (Figure 8H).Similarly, lama5, encoding a laminin required for proper MFF basement membrane formation and epithelial integrity of the MFF epidermis, 49,53 is expressed at reduced levels in distalmost MFF cells at 56 hpf, after MFF collapse has become morphologically obvious (Figure 8I).

| wnt10a mutants display a progressive reduction in the length of MFF actinotrichia and a loss of actinotrichia within the cleft of cleft cells
The MFF of fras1 and frem2a mutants is characterized by blistering underneath the basement membrane of the skin, 49 the MFF of lama5 mutants by compromised basement membrane formation and epidermal integrity, 49,53 and the MFF of col1a1a mutants by compromised actinotrichia formation and a frilly morphology of the fin. 51,66n light of the reduced expression of frem2a, lama5, and col1a1a in distal-most MFF cells of wnt10a mutants described above (Figure 8), we wondered whether wnt10a mutants display any of these phenotypic MFF traits and whether such defects might contribute to the MFF collapse phenotype.To study this, a combination of immunofluorescence (Figure 9) and transmission electron microscopy (TEM) studies (Figure 10) was applied.However, anti-laminin immunofluorescence labeling and transverse sectioning of tails of wnt10a sibling and mutant embryos (at an identical position to sections shown in Figure 8B) revealed normal basement membrane organization and normal distances between the two adjacent MFF basement membranes of wnt10a mutants at 48 hpf (Figure 9A), clearly different from the Nidogen-2a delocalization and blistering phenotype of lama5 and fras1/frem2a mutants, respectively. 49,67Consistently, TEM revealed normal basement membrane morphology and basement membrane-dermal anchorage in the MFF of wnt10a mutants at 48 hpf (Figure 10A; n = 3/3 for mutants and siblings).Taken together, this indicates that despite the reduced expression of frem2a and lama5 in mutants, Wnt10a is not required for proper MFF basement membrane formation, maintenance or anchorage and that the fin collapse phenotype must be caused by other structural defects.
In light of the reduced col1a1a expression, we also studied the MFF actinotrichia.Anti-collagen type II immunofluorescence (Figure 9B) and TEM (Figure 10A) revealed that MFF actinotrichia are present in normal numbers in wnt10a mutants, and that they are of normal thickness and structure (Figure 10A,E; n = 3/3 for mutants and siblings).However, they display a progressive reduction in their proximodistal length (Figure 9C,D).The degree of this reduction varies along the anteroposterior length of the MFF, in line with the spatial differences in the onset of MFF collapse (see Figure 1).While in siblings, actinotrichia from 48 hpf through 7 dpf become up to 25% longer, in mutants, they shrink, depending on the anteroposterior position, to almost one third of their initial length (Figure 9B-D).
9][70] For mammalian tendons (which contain comparably long and perfectly aligned collagen fibrils), collagen fibril formation takes place in specialized compartments consisting of plasma membrane channels and protrusions, named fibripositors. 71,72he cleft of cleft cells of the zebrafish MFF might represent a similar compartment, and indeed, TEM sections of wild-type embryos of 48 hpf often revealed actinotrichia that continue from the ECM of the distal tip of the cleft into intracellular cleft cell compartments (Figure 10B, regions framed in yellow; n = 3/3), strikingly similar to the formerly reported fibripositors (see fig. 6

in Ref. 71).
In cleft cells of wnt10a mutants, however, no such fibripositor-like structures were detected (n = 0/3).Rather, actinotrichia did not reach the distal-most regions of the cleft, but terminated more proximally than in wild-type siblings (Figure 10B, regions framed in blue and red, and Figure 10C,D).This suggests that the linear growth of actinotrichia normally is accomplished within fibripositor-like compartments of cleft cells, whereas in cleft cells of wnt10a mutants, this distal actinotrichia production has largely ceased, in line with the reduced expression of col1a1a (Figure 8G,H) and the arrested growth of actinotrichia (Figure 9B-D) described above.That actinotrichia in the mutant MFF even shrink (Figure 9B-D) could be due to the action of collagenases like Mmp2, the gene of which has been reported to be expressed at low levels by epidermal cells along the entire proximodistal length of the MFF of 42 hpf embryos. 73n conclusion, our data suggest that the wnt10a mutant MFF collapse is primarily caused by the loss of distal-most properties in epidermal cells of the fin fold, indicated by down-regulation of dlx2a expression in cleft cells before the onset of fin collapse.During the further course of MFF development, this leads to the progressive loss of collagen-producing and -stabilizing proteins normally generated by cleft cells and thereby to a distal shrinkage of MFF actinotrichia causing progressive MFF collapse.

| DISCUSSION
In this study, we first characterized a zebrafish lossof-function mutant in the canonical Wnt signal, Wnt10a.As reported earlier for zebrafish wnt10a morphants, 10 mouse Wnt10a mutants and human WNT10A patients, homozygous zebrafish wnt10a mutants display severe tooth malformations.In mammals, WNT10A, in addition to tooth development, is required for the formation of several other ectodermal appendages, such as hair follicles, sebaceous glands, taste buds, and sweat ducts. 5,59In this light, it is not surprising that we found zebrafish wnt10a to also be necessary for the development of the F I G U R E 8 MFF collapse in wnt10a mutant MFF involves the down-regulation of dlx2a and crucial ECM protein-encoding genes in distal MFF cells.(A) At 33 hpf (left panels), before MFF collapse becomes morphologically visible, wnt10a mutant displays unaltered expression of the Et(krt4:eGFP) sqet37 transgene both in fin mesenchymal cells and in distal-most epidermal cells of the MFF (indicated by yellow arrows).At 48 hpf (middle panels and right panels for enlarged views of regions framed in middle panels), when MFF collapse is obvious and in progress, fin mesenchymal cells of the mutant still show normal Et(krt4:eGFP) sqet37 expression, whereas transgene expression in distal epidermal cells is reduced (indicated by yellow arrowheads) in comparison to wild-type sibling (indicated by yellow arrow).(B) Transverse sections of 48 hpf Et(krt4:eGFP) sqet37 transgenics at the location indicated in middle panels of (A) by dashed lines, indicating the normal positioning and transgene expression in fin mesenchymal cells (white arrows), but the reduced transgene expression in distal epidermal cells both in the collapsing dorsal and ventral MFF of the mutant (yellow arrowheads) compared to its sibling (yellow arrows).(C-E) In situ hybridization indicates that at 33 hpf, before MFF collapse becomes morphologically visible, the expression of the distal marker gene dlx2a is strongly down-regulated in distal-most epidermal cells of the ventral MFF (C).Expression of fras1, a marker for distalmost cleft cells, 65 appears unaltered in mutants compared to their wild-type siblings (D), whereas expression of frem2a, a marker for distal ridge cells, 65 is reduced, as also obvious in transverse Durcupan sections (E).(F-I) In situ hybridization indicates that at 33 hpf (left panels), before MFF collapse becomes morphologically visible, expression of col1a1a, encoding an essential component of actinotrichia, is normally expressed in distal-most epidermal cells of the MFF of wnt10a mutants (F), while expression levels become progressively reduced at 48 hpf ((G); most obvious in minor lobe) and at 56 hpf (H), when fin collapse is in progress.A similar progressive reduction of expression in the MFF of wnt10a mutant is observed for lama5, encoding an essential component of the MFF basement membrane ((I) for 56 hpf; earlier stages not shown).Arrows indicate regular expression; arrowheads indicate less or non-detectable expression.ECM, extracellular matrix; MFF, median fin fold.
MFFs during embryogenesis and adult fins during metamorphosis, processes that involve similar events of epithelial morphogenesis as ongoing during mammalian ectodermal appendage formation.

| The role of zebrafish Wnt10a during embryonic MFF development
Former studies have shown that global pharmacological inhibition of canonical Wnt signaling leads to incorrect cell shape changes of MFF epidermal cells at 30 hpf. 13 In addition, the combined loss of Lef1 and Tcf7, transcriptional mediators of canonical Wnt signaling, has been shown to lead to a reduction in the size of the embryonic zebrafish MFF at 60 hpf. 14,74Our data identify Wnt10a as the specific canonical Wnt signal most likely required for such later, Lef1/Tcf7-dependent processes, as in wnt10a mutants, the MFF is normally set up and only stops to grow and to collapse in a progressive manner from 36 hpf onwards.
In contrast to the paired pectoral fin buds, the MFF does not grow via specific cell proliferation at the AER.[49]  Our data indicate that Wnt10a is neither required for epithelial morphogenesis underlying the initial steps of MFF formation (Figure 1), nor for the proliferation or survival of epidermal cells (Figure 7), but for particular cell specification steps in the distal-most epidermal cells of the MFF AER, the cleft and/or ridge cells (Figure 8).This is in line with data obtained in mammals according to which TCF-mediated WNT10A signaling is particularly required for the differentiation of epithelial progenitor cells in the different ectodermal tissues affected in WNT10A mutants. 59n zebrafish wnt10a mutants, cleft and ridge cells display signs of compromised cell specification before the onset of MFF collapse, affecting some (for instance, the distal-specifying gene dlx2a) but not all genes normally specifically expressed there (33 hpf; Figure 8).This initial specification failures are later (48 hpf) followed by a progressively decreased expression of genes initially expressed at normal levels, such as col1a1a (Figure 8).52]75 Such a working mechanism downstream of Wnt10a and Dlx2a is in line with data obtained for mice chondrocytes, according to which Dlx2 enhances the accumulation of type II collagen by inhibiting expression of the collagenase MMP13. 76Future studies have to investigate a potential involvement of tissue inhibitors of metalloprotease (TIMPs).Of formerly investigated zebrafish timp genes, timp2a has been described to be expressed in the MFF at 26 hpf. 77However, we failed to detect timp2a MFF expression at 48 hpf even in wild-type embryos, which display continuous actinotrichia growth (JH and MH, unpublished data).In contrast, the time course of altered col1a1a expression in cleft and ridge cells of zebrafish wnt10 mutants (Figure 8), is consistent with the time course of actinotrichia shrinkage, which starts after 33 dpf and only becomes apparent at 48 hpf (Figure 9).This, together with TEM data revealing the existence of actinotrichia-containing fibripositor-like structures 71 in wild-type cleft cells (Figure 10), but the lack of such structures and of actinotrichia in the distalmost regions of the dermal space/cleft of wnt10a mutants (Figure 10), points to a prominent and tightly regulated role of cleft and, possibly, ridge cells in the linear distal growth and stabilization of actinotrichia and thereby the entire MFF.
MFF cleft and ridge cells seem to receive such stimulatory Wnt10a signals directly.Thus, they show prominent expression of tcf7 14 and the Wnt responder transgene (Figures 5 and 6).The source of the Wnt10a signals is unclear.However, it might be the fin mesenchymal cells, which express both wnt10a and the canonical Wnt signal reception transgene (Figures 5 and 6).At 28 hpf, they start to migrate into proximal regions of the MFF and from there in a proximodistal direction towards the distally located ridge and cleft cells, 48,54 so that at 33 and 48 hpf (Figure 8A), paracrine signaling between fin mesenchymal cells and distal-most epidermal MFF cells should be possible.Fin mesenchymal cells also contribute to actinotrichia formation themselves, in particular to their lateral growth/thickness. 52However, with molecular markers, fin mesenchymal cells do not seem to respond to the loss of Wnt10a (Figure 8A and data not shown).Thus, the MFF collapse observed upon genetic ablation of fin mesenchymal cells 52 might be due, in part, to the resulting loss of Wnt10a signaling from fin mesenchymal cells to cleft and ridge cells and thereby to the loss of linear actinotrichia growth and stabilization, rather than or in addition to the resulting loss of the direct contribution of fin mesenchymal cells to lateral actinotrichia growth.
Future studies have to reveal whether similar mechanisms as during the MFF collapse of wnt10a embryos, with loss of Wnt10a signaling dlx2 and col1a1a expression in distal epidermal cells leading to actinotrichia destabilization, also underlie the normal regression of the fin fold that takes place in wild-type animals at the onset of metamorphosis in MFF regions between the forming caudal, anal and dorsal fins 17 (Figure 1G).In addition, it remains to be investigated why the most posterior region of the embryonic MFF as well as the paired pectoral fin folds of wnt10a mutants do not undergo an obvious fin fold collapse (Figure 1C-E), although they show canonical Wnt signal reception (Figure 3E for pectoral fin folds and Figure 6A-C for caudal MFF) and although they contain aligned arrays of actinotrichia. 50Of note, an arrest of pectoral fin bud outgrowth, combined with a loss of the entire embryonic MFF, including posterior most regions, has been reported for zebrafish embryos lacking Lef1 and Tcf7, two partially redundant transcriptional mediators of canonical Wnt signaling. 14,74This suggests that during AER maintenance in the pectoral fin fold and the posterior MFF, Wnt10a might function redundantly with another canonical Wnt signal such as wnt3a or wnt7a which are also expressed in the developing pectoral fin. 78

| The requirement of zebrafish Wnt10a for adult fin formation during metamorphosis
Although the posterior MFF region of wnt10a mutants remains largely unaffected during embryogenesis, it is not capable of transitioning into a proper adult caudal tail during metamorphosis (Figure 1G,H).The two other unpaired adult fins, the anal and dorsal fin, are completely absent in wnt10a mutants (Figure 1H), even when the MFF had been fully rescued up to early metamorphosis stages (see Figure 4D for 14 dpf) by temporary transgenic expression of wild-type wnt10a (Figure 4E).This points to a second and independent function of wnt10a in the transition from an embryonic, purely actinotrichia-based fin fold into unpaired adult fins, which still contain actinotrichia in distal-most regions, while their main mechanical support is provided by bony rays of lepidotrichia. 15,17,56,79Consistently, at the early stages of metamorphosis, we found wnt10a to be highly expressed in the dermal space of the caudal fin (Figure 5C) and canonical Wnt signal reception in interray domains distal of the lepidotrichia (Figure 6F), where the actinotrichia are located. 56Interestingly, genetic loss of col9a1c, which is normally expressed in mesenchymal and basal epidermal cells of the forming and regenerating adult caudal fin, leads to aberrant actinotrichia formation/maintenance in distal regions of the outgrowing caudal fins and to later lepidotrichia defects and caudal fin deficiencies 56 very similar to those shown here for the wnt10a mutant (Figure 1H).This suggests that also during the formation of adult unpaired fins, Wnt10a might act by promoting the expression of collagen genes and actinotrichia formation and stability, very similar to its role in the embryonic MFF revealed here (Figures 8-10).
In contrast, the paired adult fins, pectoral fins, and pelvic fins, although also containing actinotrichia and lepidotrichia, are rather unaffected by the loss of Wnt10a despite also expressing it (Figures 1H and 3C).The reasons for these differential responses remain unclear.Differences in the embryonic origin of the mesenchymal cells of the different fin types could play a role.1][82][83] In this light, it is tempting to speculate that only fin mesenchymal cells from the paraxial mesoderm (which themselves remain in the fins as fibroblasts 83 ) use Wnt10a to instruct themselves or epidermal cells to form actinotrichia and/or to instruct osteoblasts (which invade the adult fin anlagen only during early stages of metamorphosis 83 ) to form lepidotrichia, whereas corresponding interactions in paired fins involve other signals.Of note, like the paired fins, mesenchymal cells of the minor lobe of the MFF on the yolk tube, the PAFF, derive from the lateral mesoderm. 16If the aforementioned notion is correct, adult fins deriving from the PAFF should also be rather unaffected.However, the PAFF does not give rise to permanent fins even in wild-type animals.

| The role of zebrafish Wnt10a during embryonic and adult fin regeneration
This instructive role of Wnt10a during the formation of the embryonic MFF and the adult unpaired fins might also be re-used during fin regeneration.Thus, we found wnt10a mutant embryos to display compromised tail fin regeneration after its amputation at 2 dpf (EB and MH, unpublished results).In addition, it has previously been shown that wnt10a is expressed in the distal tip of the blastema during the early stages of the adult zebrafish tail fin regeneration. 11Furthermore, it has been shown that canonical Wnt signaling is activated in actinotrichiaforming cells located adjacent to osteoblasts of the regenerating adult caudal fin to direct osteoblast commitment and differentiation, most likely via thus far unidentified secreted factors 57 that they possibly generate as a direct response to their own stimulation by Wnt10a.In sum, this indicates that Wnt10a is required both for the de novo development and the regeneration of both the embryonic MFF and the later adult unpaired fins, most likely involving similar molecular and cellular mechanisms.

| Parallels in the roles of Wnt10a during fin and tooth development
But to what extent are such molecular and cellular mechanisms of Wnt10a function also conserved in other Wnt10a-dependent development processes, such as the formation of teeth (Figure 3) or their likely evolutionary relatives, the scales. 84During zebrafish tooth development, we found wnt10a to be expressed in both the dental mesenchyme and epithelium, similar to the expression of eda, while eda expression was unaffected in wnt10a mutants, pointing to a potential role of ectodysplasin signaling upstream of wnt10a (Figure 3).Indeed, mutants in eda or its receptor edar lack pharyngeal teeth, 85 similar to wnt10a mutants (Figure 3B).In addition, strong eda and edar mutants display compromised lepidotrichia formation and adult fin development, 85 comparable to, but not exactly identical to the later fin phenotype of wnt10a mutants (Figure 1E).However, unlike eda and edar mutants, 85 wnt10a mutants form normal scales (Figure 1H), while in reverse, eda and edar mutants do not show the embryonic MFF collapse 85 typical for wnt10 mutants (Figure 1B-D).This points to a conserved ectodysplasin-Wnt10a interaction and cooperation in some, but not all developmental processes requiring one or the other.In some cases, Wnt10a might be replaced by other canonical Wnt signals.Thus, in contrast to the specific loss of Wnt10a (Figure 1H), global inhibition of canonical Wnt signaling via forced transgenic expression of Dkk1 (compared with Figure 4) blocks early steps of scale formation, 12 as does loss of eda or edar. 85or tooth development, we found wnt10a to be required for proper expression of fgf4, encoding a Fibroblast growth factor, and dlx2b, the paralog of dlx2a, which is down-regulated in distal epidermal MFF cells of wnt10a mutants, encoding transcription factors related to the distal determinant Distalless in developing appendages of the fruitfly Drosophila melanogaster. 64,86Of note, pharmacological inhibition of FGF signaling has been reported to compromise pharyngeal tooth development, preceded by an inhibition of dlx2b expression in the tooth germs, 31 as well as scale formation 12 and early steps of embryonic MFF formation, including the downregulation of dlx gene expression 87 and the regeneration of the adult caudal fin. 88Similarly, morpholino-based knock down of dlx genes has been reported to compromise both tooth development 40 as well as embryonic MFF development. 89Together, these data point to a conserved Wnt-Fgf-Dlx pathway or co-operation during MFF formation and maintenance, tooth development, and most likely multiple other processes of epithelial organ or appendage development, while the specific paralogs of each gene family used in these processes might vary from case to case.
Future studies need to address to which extent identified potential Wnt10a (and Dlx2) target genes encoding ECM proteins or regulators, as identified here as potential drivers of embryonic MFF morphogenesis and the formation and stabilization of the collagenous actinotrichia, are also at play during such other Wnt10adependent processes.In mice, collagens, MMP collagenases, and their inhibitors Timp1, Timp2, and Timp3 are strongly expressed in epithelial-versus mesenchymalspecific manners during early steps of tooth morphogenesis, 90,91 while a possible connection to Wnt signaling and Dlx transcription factors has, at least to our knowledge, not been reported as yet.

| Zebrafish lines and Ethics statement
Zebrafish were handled in compliance with local animal welfare regulations and maintained according to standard protocols (zfin.org).Adults and embryos were grown and studied at 28 C. The following zebrafish lines were used: wild-type TL/EK; Tg(hsp70l:dkk1b-GFP) w32Tg 11 (here named Tg(hsp70l:dkk1-GFP)); Tg(OlSp7:nlsGFP) zf132 18 (here named Tg(osterix:nuGFP)); Tg(6xTcf/LefBS-miniP: d2EGFP) isi01 33 (here named Tg(6xTCF:eGFP); msxc mn0245Gt 55 ; Et(krt4:eGFP) sqet37 62,63 and unm_t30922, hereafter referred to as wnt10 t30922 mutant.This wnt10a mutant was recovered from an N-ethyl-N-nitrosourea mutagenesis forward genetics screen of wild-type Tübingen strain zebrafish performed within the frame of the ZF-MODELS Consortium (2006) (ZFIN Direct Data Submission).The wnt10a genotype was confirmed using genomic DNA and PCR as described below, or by in-crossing and subsequent phenotyping.In general, homozygous wnt10a mutant zebrafish were compared with unaffected sibling controls unless otherwise indicated.SL (in mm) of juvenile and adult fish was used to control equal development of each group.For the duration of imaging, fish were kept under anesthesia in E3 embryo medium containing 200 μg/mL tricaine (Sigma-Aldrich).
All zebrafish experiments were approved by the national animal care committees (Landesamt für Natur, Umwelt und Verbraucherschutz Nordrhein-Westfalen

| Meiotic mapping and WGS
Meiotic mapping of the fin collapse-causing mutation was essentially performed as formerly described. 49,92Heterozygous F1 carriers from WIK out-crosses were incrossed and pools of either mutant or sibling F2 progeny were subjected to bulk segregation single sequence linkage polymorphism (SSLP) analysis.Upon assignment to a linkage group, fine SSLP mapping on single-arrayed mutant embryos was used to confirm linkage and generate a broad interval on the genome.
For WGS, genomic DNA from 20 homozygous siblings and 19 homozygous fin collapse mutants (identified with mapping primers) was isolated with Nucleospin tissue columns (Thermo Fischer Scientific/Machery-Nagel) and subsequently treated with RNase A (Thermo Fischer Scientific).An Illumina TruSeq DNA LT Sample Preparation Kit was used for library preparation and the Illumina HiSeq 2000 sequencing system with paired end 100 bp (PE100) read length was used for sequencing.The reads were aligned to the zebrafish genome version GRCz1196 with bwa v0.7.17 (https://doi.org/10.48550/arXiv.1303.3997) and samtools v1.9 (https://doi.org/10.1093/gigascience/giab008).The aligned reads were deduplicated with picard v2.19.0 (http://broadinstitute.github.io/picard/) and the variants were called with GATK v4.1.1.0(https://doi.org/10.1002/0471250953.bi1110s43)and annotated with Annovar v2018apr16 (https://doi.org/10.1093/nar/gkq603).The meiotic mapping was performed by plotting the absolute difference in the allelic fractions of all SNPs identified in either the wild-type pool or the mutated pool, after excluding variants that were homozygous in both pools (allelic fraction ≥ 0.9), that had a fraction of alternate reads lower than 0.25 or that were sequenced at a depth lower than 15.We identified a total of 18,878,760 variants in the mutant pool, of which 6177 were within the mapped region on chromosome 9: 7,000,000-14,000,000.Of these, 82 variants were predicted to affect the sequence of a protein, 26 variants were homozygous and only one was not present in the control pool or in any of our previous sequencing experiments (16 WGS data sets and 11 whole exome sequencing data sets).The remaining variant was a point mutation (chr9:11685198A>G) within the splice acceptor site of intron 3 located directly in front of the start of exon 4.

| Genotyping
wnt10a mutants were genotyped taking advantage of a restriction fragment length polymorphisms (RFLP) generated by the mutation.A genomic fragment of the wnt10a gene was PCR-amplified with the primers wnt10a_FW (5 0 -CTGAGTGGGGAATGTACCTCG-3 0 ) and wnt10a_RV (5 0 -GTGTGTGGTGGGCATTCTCC-3 0 ), yielding a 526-bp amplification product which was then digested with MspI (New England Biolobs).The MspI restriction site is only present in mutant genomic DNA, yielding a 331-bp and a 195-bp cleavage product.

| Identification of mutant wnt10a cDNAs
Total RNA was isolated from identified wnt10a homozygous mutant and wild-type embryos at 48 hpf, using TRIzol reagent (Thermo Fischer Scientific) and following the manufacturer's instructions.Isolated RNA was treated with DNaseI (Roche) and cDNA was synthesized with iScript™ cDNA Synthesis Kit (Bio-Rad) according to the manufacturer's instructions.
To identify possible mutant isoforms, PCR amplification was performed with the following primers: wnt10a_exon3_FW (5 0 -ATTCACTCCAGGATGAGACT TCATA-3 0 ) and wnt10a_exon4_RV (5 0 -GTTTCTGTTGTG GGCTTTGATTAG-3 0 ). 93The products were subsequently cloned into pGEM-T Easy Vector (Promega) and 10 clones from wildtype and wnt10a homozygous mutants each were sequenced.The resulting cDNA sequences surrounding the splice site of exons 3 and 4 of the wildtype and the three mutant isoforms identified are provided in Table 1.

| Plasmid constructs and transgenesis for forced wnt10a expression
The Tol2kit multisite Gateway-based transposon system was used to make a transgenic construct from which a stable line was raised. 94Full-length wnt10a was cloned into the middle entry pME-MCS (using primers in Table 2).An LR Clonase (Invitrogen) Gateway reaction was performed with p5E-hsp70, the resulting pME-wnt10a and p3E-eGFPpA inserted into pDestTol2 CG2 to produce hsp70:wnt10a construct.This construct was co-injected with tol2 transposase RNA into zebrafish 1 cell-stage embryos to create the Tg(hsp70:wnt10a stop -eGFP) fr60Tg transgenic line.The full-length pME-wnt10a contains both start and stop codons; therefore, the C-terminal eGFP is not expressed in this transgenic embryo, so founders were screened by heart marker GFP expression.The resulting Tg(hsp70:wnt10a stop -eGFP) fr60Tg is referred to as hsp70:wnt10a.
For heat shock-induced transgene activation, embryos were transferred from 28 C to prewarmed water at 40 C for 1 h and returned to water at 28 C. Heat shock treatments were repeated every 24 h.Note: Uppercase: exon 3; uppercase and underline: exon 4; lowercase: intron 3.

| Chemical treatments
All inhibitors were obtained from Sigma-Aldrich/Merck. Stock solutions of drugs were prepared in DMSO and final dilutions were prepared in E3 embryo medium before application to the embryos.Control embryos were incubated in vehicle (DMSO) at the same dilution as the drug-treated embryos.Final concentrations were as follows: 80 μM of the caspase 3 peptide inhibitor z-DEVDfmk (Calbiochem 264155-80) 95 and 0.6 μM of the tumor suppressor p53 inhibitor Pifithrin-α (Calbiochem 63208-82-2).

| Tissue labeling procedures
Labeling of adult mineralized bone with Alizarin red 96 and labeling of embryo cartilage and mineralized bone with an acid-free alcian blue and alizarin red staining 97 were performed as previously described.
Epidermal cell proliferation was assessed by BrdU incorporation after incubating embryos from 33 to 35 hpf or from 45 to 46 hpf in 100 μg/mL BrdU (Sigma) in fish system water, followed by washing in egg water for 2 h and fixing at 37 or 48 hpf, respectively, and followed by anti-BrdU immunolabeling.
TUNEL assay (In Situ Cell Death Detection Kit, Roche) was performed on paraformaldehyde (PFA) fixed embryos following the manufacturer's instructions.Embryos treated with 300 U/mL DNaseI for 10 min before the TUNEL reaction served as a positive control, while embryos incubated in label mixture without enzyme served as a negative control.
For histological, immunofluorescence, and in situ hybridization analyses, zebrafish embryos were sacrificed by Tricaine overdose and fixed in 4% PFA for 1 h.Samples for cryosections were washed with phosphatebuffered saline (PBS) and orientated in 15% sucrose with 1.5% agarose in PBS, incubated overnight at 4 C in 30% sucrose in PBS, and mounted in tissue freezing medium (Leica).About 14 μm sections were obtained using a Leica CM1850 cryostat.
For double staining with whole-mount in situ hybridization followed by p63 antibody staining, the in situ hybridization was performed as previously described. 98amples were post-fixed in 4% PFA for 20 min at room temperature, washed three times in PBS-Tween (PBST) (0.1%) for 10 min each, blocked for 3-4 h in blocking solution (10% fetal bovine serum, 1% DMSO in PBST) and incubated with mouse monoclonal anti-P63, 1:200 dilution, overnight at 4 C.The samples were then washed four times with PBST, incubated with goat anti-chicken Alexa Fluor 488 antibody for 90 min at room temperature, 1:750 dilution, washed again four times in PBST, post-fixed in 4% PFA for 20 min at room temperature, washed four times in PBST for 10 min, embedded in 1.5% agarose and 5% sucrose solution in PBS, incubated in 30% sucrose in PBS overnight at 4 C The embedded samples were mounted in tissue freezing medium (Leica) and 10 μm cryosections were made as described above.
4.9 | Quantitative reverse transcription polymerase chain reaction Total RNA was isolated, treated with DNase I (Roche), and cDNA was synthesized as previously described (Plasmid constructs and transgenesis).qRT-PCR was performed in technical duplicates with Sybr Select Master Mix (Life Technologies, Thermo Fisher Scientific) on an ABI-Prism 7500 Fast Detect system, and relative expression levels were calculated following the ΔΔCt method and data were normalized to the expression of peptidylprolyl isomerase A-like ( ppial). 99Data are presented as fold change relative to the relevant sibling control and represent the average of at least three independent experiments.Sequences of forward and reverse primers are described in Table 2.

| Microscopy
All wholemount brightfield and fluorescence images were taken on a Leica M165 FC microscope, and confocal images were taken on a Zeiss LSM 710 microscope and processed using Fiji software. 100or TEM, 48 hpf zebrafish were anesthetized, imaged with a Leica M165 FC microscope and fixed overnight in 2% glutaraldehyde and 2% PFA in 0.1 M cacodylate buffer (pH 7.4).Fixed embryos were washed several times in 0.1 M cacodylate buffer (pH 7.3), postfixed with 1% osmium tetraoxide in 0.1 M cacodylate buffer (pH 7.3) for 1 h at room temperature (RT) in the dark, washed again in 0.1 M cacodylate buffer (pH 7.3), and dehydrated in a graded series of ethanol, gradually transferred, embedded and polymerized in Epon (20 g Epoxy, 11 g DDSA, 9 g NMA and 0.8 g DMP30).Semithin sections (500 nm) and ultrathin sections (70 nm) were cut with a Leica EM UC6 (Leica, Wetzlar, Germany) Ultracut ultramicrotome at 750 μm from the most posterior tip of the caudal fin.Semithin sections for light microscopy were stained with 1% methylene blue and 1% azur.Ultrathin sections were mounted on grids, stained for 15 min with 1.5% uranyl acetate at 37 C, washed five times in water, incubated for 4 min in lead citrate at RT, washed five times in water, and dried.Electron microscopy images were obtained with a JEOL JEM2100PLUS, 80 kV FE (Field Emission) analytical electron microscope equipped with a GATAN OneView camera.Since transverse sections do not always run completely parallel to actinotrichia bundles in fins, we used three sections per embryo to calculate an average length of dermal space devoid of actinotrichia.Three individual embryos per group were used for calculations.

F
I G U R E 2 unm_t30922 t30922 mutants carry a homozygous splice acceptor site mutation within the wnt10a gene, representing a likely wnt10a amorph.(A) Meiotic mapping placed the MFF collapse-causing mutation into a 0.429 cM interval of chromosome 9 between SSLP markers z25735 (2/1400 recombinants/meioses) and NZ79 (4/1400 recombinants/meioses).(B, C) Whole-genome sequencing (WGS) of DNA from 20 pooled mutant embryos revealed a region of homozygosity exclusively on chromosome 9 (B), which was approximately 4 Mb large (indicated by red lines in (C)) and contains a single point mutation (A>G) in the splice acceptor site before exon 4 of wnt10a (C).(D) Sequencing of cDNA from mutant embryos revealed three mutant splice isoforms, resulting from the usage of cryptic splice acceptor sites within intron 3/4 or exon 4 and leading to C-terminally truncated Wnt10a proteins lacking exon 4-encoded amino acid residues 288-442.(E) Quantitative real-time PCR (qRT-PCR) of whole embryos shows reduced amounts of wnt10a transcripts in unm_t30922 t30922 mutants at 2 and 5 dpf.

F I G U R E 4
The fin defects of unm_t30922 t30922 mutants can be phenocopied by loss and rescued by gain of Wnt10a function.(A, B) Morpholino knockdown of wnt10a (A) or inhibition of canonical Wnt signaling via temporally controlled transgenic expression of dkk1 from 24 or 48 hpf onwards (B) both phenocopy the median fin fold collapse observed in the unm_t30922 t30922 mutant embryos at 3 dpf.(C) Transgene-encoded wild-type wnt10a can fully rescue MFF collapse of unm_t30922 t30922 mutants at 3 dpf when expressed from 24 hpf onwards, but can only rescue partially when expressed from 48 hpf onwards.(D) At 14 dpf (SL 5 mm), unm_t30922 t30922 mutants lack most of their MFFs with the exception of the caudal fin.Similar to the effects at 3 dpf, mutants with transgenic expression of wild-type wnt10a from 24 to 144 hpf have regular MFFs, while mutants with transgenic expression of wild-type wnt10a from 48 to 144 hpf lack parts of their MFFs.(E) Despite the complete or partial MFF rescue at 3 and 14 dpf, mutants with transgene-driven wild-type wnt10a expression from 24 or 48 to 144 hpf have failed to develop complete adult fins at 31 dpf (SL 9 mm), resembling unm_t30922 t30922 mutants without forced wild-type wnt10a expression.Arrowheads indicate collapsed fin folds and remnants of adult fins.MFF, median fin fold; SL, standard length.

F
I G U R E 5 wnt10a is expressed in fin mesenchymal cells.wnt10a in situ hybridization of wild-type embryos.(A) At 28 hpf, wnt10a is expressed in the fin mesenchymal cells, as the cells start to migrate into the dermal space of the MFF (arrows in A', showing a magnified view of region framed in (A)).(B) At 56 hpf, wnt10a is expressed in fin mesenchymal cells throughout the entire dermal space of the MFF (arrow in B 0 , showing a transverse section at the position indicated by dashed line in (B)).(C) At the onset of metamorphosis at 13 dpf (SL 5 mm), wnt10a is strongly expressed in the dermal space of the caudal fin where lepidotrichia start to form (arrows in (C 0 ), showing a transverse section at the position indicated by dashed line in (C)).MFF, median fin fold; SL, standard length.

F
I G U R E 6 Canonical Wnt signals are received by distal-most epidermal cells and fin mesenchymal cells of the embryonic MFF.Fluorescent images of live Tg(6xTCF:eGFP) transgenics, with cells receiving canonical Wnt signals in green.(A) At 48 hpf, the responder transgene is expressed in distal-most cells of the MFF (white arrows) and in fin mesenchymal cells (orange arrow) of the caudal fin fold of wild-type embryo.(B) Compared to wild-type sibling (left panel), wnt10a mutant embryo of 60 hpf (right panel) shows reduced Tg(6xTCF: eGFP) expression in the minor and major lobes of the MFF (indicated by arrowheads), whereas expression in the caudal fin fold (indicated by orange arrow) and in the AER of the pectoral fin is comparably unaltered.(C) Compared to non-heat-shocked control (left panel), transgenic embryo with heat-shock-induced global wnt10a overexpression from 24 to 48 hpf (right panel) shows increased Tg(6xTCF:eGFP) expression in the minor and major lobes of the MFF (indicated by thick arrows), whereas expression in the caudal fin fold (indicated by orange arrow) and the AER of the pectoral fin is unaltered.(D) Transverse cryosection approximately 0.5 mm anterior of the caudal tip of the tail of 6xTCF:eGFP 48 hpf embryo, stained with antibodies against GFP and collagen II (labeling actinotrichia, red).The Wnt responder is expressed in distal-most positions of the MFF.(E) 6xTCF:eGFP, GBT0245 (msxc; marker of fin mesenchymal cells in red 55 ) double transgenic embryo revealing expression of the Wnt responder in fin mesenchymal cells of the ventral MFF at 3 dpf.(F) At the onset of metamorphosis (13 dpf, SL 5 mm), the Wnt responder is expressed distal of forming lepidotrichia (stained with alizarin red) of the caudal fin.act, actinotrichia; AER, apical ectodermal ridge; ds, dermal space; MFF, median fin fold; otv, otic vesicle; pecf, pectoral fin; SL, standard length.

F
I G U R E 7 MFF collapse in wnt10a mutant does not involve increased cell death or impaired cell proliferation.(A) TUNEL staining shows no major differences in the number of dying cells in the MFF of wnt10a mutant embryos at 48 hpf.(B) Quantification of the number of TUNEL-positive cells per embryo within region framed in (A), indicating no statistically significant increase in mutants compared to siblings.The blue dot and red square indicate values for embryos shown in (A).(C) Treatment from 24 to 72 hpf with the Caspase-3 inhibitor Z-DEVD-FMK (80 μM), the p53 inhibitor Pifithrin-α (0.6 μM), and DMSO (control) does not prevent the MFF collapse of wnt10a mutants at 72 hpf.(D, E) BrdU incorporation (in red) indicates no obvious differences in cell proliferation within the ventral MFF at 37 hpf, when MFF collapse commences (D), or at 48 hpf ((E), nuclei stained with DAPI in blue), when MFF is in progress (E).(F) Percentages of BrdU-positive cells in ventral MFF within the region shown in (E), indicating no statistically significant decrease in mutants compared to siblings.The blue dot and red square indicate values for embryos shown in (E).MFF, median fin fold.

F
I G U R E 9 wnt10a mutants display a progressive shrinkage of MFF actinotrichia.(A) Immunofluorescence of Laminin (in green, nuclei stained with DAPI in blue), transverse sections of tail region approximately 750 μm anterior of the tip of the tail of embryos at 48 hpf.Laminin shows indistinguishable protein levels and basement membrane localization in wnt10a sibling and mutant ventral MFFs.(B) Actinotrichia, visualized via Col II immunofluorescence (in red in upper row and white in lower row; nuclear DNA stained with DAPI in blue) in the ventral MFF of wnt10a mutants and siblings at 48 hpf and 7 dpf.Lower row shows magnified views of regions framed in upper row.(C, D) Graphical illustration of lengths of actinotrichia at two corresponding locations indicated in (B) in the major lobe of the ventral MFF in wnt10a siblings and mutants at 48 hpf, 54 hpf, 3 dpf, and 7 dpf.MFF actinotrichia continue to grow in the sibling, whereas they become progressively shorter in the mutant.F I G U R E 1 0 Actinotrichia of the MFF of wnt10a mutants do not extend properly into the cleft of epidermal cleft cells.All images show TEM micrographs of transverse sections through the ventral MFFs of wnt10a siblings and mutants of 48 hpf, proximately 750 μm anterior of the tip of the caudal fin.(A) In intermediate proximodistal positions, the ventral MFF mutants display regular ultrastructure, including the basement membrane between epidermal cells and the dermal space (arrowheads), basement membrane-dermal anchorage and actinotrichia.(B) In distal-most positions of the ventral MFF, cleft cells of wnt10a mutants display a slightly widened cleft (indicated by blue and red brackets for sibling and mutant, respectively).In addition, in contrast to siblings, in mutants, actinotrichia do not reach the distal tip of the cleft (distal end of actinotrichia in sibling and mutant indicated by blue and red arrowheads, respectively).Furthermore, siblings show actinotrichia extending into the interior of cleft cells, constituting "fibripositor"-like structures 71 (indicated by yellow arrow in sibling), whereas mutants do not.Blue, yellow, and red dashed rectangles of overview images frame regions shown in magnified views to their right.(C, D) Graphical illustration of quantification demonstrating that wnt10a mutant embryos do not have significantly shorter ventral MFFs compared to the siblings (C), but a statistically significant increase in the distance between actinotrichia and the distal tip of the cleft of cleft cells (D).The blue dots and red squares in (C, D) indicate values for the respective sibling and mutant shown in (B).(E) At the level of ridge cells, actinotrichia of siblings and mutants display comparable width and banded structure.MFF, median fin fold.

T
A B L E 1 wnt10a cDNA sequences wild-type and mutant isoforms.