Fndc3a (Fibronectin Domain Containing Protein 3A) influences median fin fold development and caudal fin regeneration in zebrafish by ECM alteration

We investigated potential functions of Fndc3a during caudal fin development and regeneration in zebrafish. Reduced function interferes with correct epidermal cells structure and implies a role during vertebrate extremity development. Abstract Inherited genetic alterations are often found to be disease-causing factors of patient phenotypes. To unravel the molecular consequences of newly identified factors functional investigations in vivo are eminent. We investigated molecular functions of FNDC3A (Fibronectin Domain Containing Protein 3A; HUGO), a novel candidate gene for split-hand/foot malformations (SHFM) in humans, by utilizing zebrafish (Danio rerio) as a vertebrate model. Patients with congenital SHFM display prominent limb malformations, which are caused by disturbance of limb development due to defects in apical ectodermal ridge (AER) establishment and maintenance. Initial gene expression and protein localization studies clarified the presence of fndc3a in developing and regenerating fins of zebrafish. For functional studies we established a hypomorphic fndc3a mutant line (fndc3awue1/wue1) via CRISPR/Cas9, exhibiting phenotypic malformations and changed gene expression patterns during early stages of median fin fold development. Furthermore, fndc3awue1/wue1 mutants display abnormal collagen localization, actinotrichia breakup and cellular defects in epidermal cells during caudal fin development. The observed effects are only temporary and later result in rather normal fin development in adults. In accordance with early fin development, proper caudal fin regeneration in adult fndc3awue1/wue1 mutants is hampered by interference with actinotrichia formation and epidermal cell abnormalities. Investigation of cellular matrix formation implied that loss of ECM structure is a common cause for both phenotypes. Our results thereby provide a molecular link between Fndc3a function during both developmental processes in zebrafish and foreshadow Fndc3a as a novel temporal regulator of epidermal cell properties during extremity development in vertebrates.


Introduction 1
Patients suffering from SHFM display prominent limb malformations, lacking the central 2 autopod rays, resulting in syndactyly, aplasia and/or hypoplasia of the phalanges, 3 metacarpals and metatarsals, and median clefts of hands and feet (Duijf et al., 2003;4 Gurrieri and Everman, 2013). Several genetic loci and causes have been associated with this 5 inherited disease, for example mutations in TP63 (OMIM *603273), DLX5 (OMIM *600028) 6 or DLX6 (OMIM *600030). A large number of animal experiments and cell culture studies 7 further clarified the underlying genetic network, but still unresolved patient cases disclose the 8 necessity to identify novel contributing molecular factors. 9 We identified FNDC3A as a new potential candidate gene in pathogenesis of human non-10 syndromic SHFM by exome sequencing of a consanguineous family from Syria with four 11 affected individuals (unpublished data). FNDC3A consists of up to nine fibronectin type III 12 domains, which are a common feature of a large number of extracellular proteins acting by 13 modulation of different signaling pathways (Cheng et al., 2013; Zhu and Clark, 2014). 14 FNDC3A has initially been described to be overexpressed in human odontoblasts (Carrouel 15 et al., 2008). Functional experiments in Symplastic spermatids (sys) knockout mice indicated 16 that FNDC3A is essential for cell adhesion between spermatids and Sertoli cells, resulting in 17 sterile males (Obholz et al., 2006). Besides these two studies the developmental function of 18 FNDC3A is still unknown and no association to extremity development or regeneration has 19 been described yet. The purpose of this study was to investigate potential functions of 20 Fndc3a during extremity development and regeneration in zebrafish. 21 Differences between tetrapod limb and fin development in ray-finned fishes (Actinopterygii) 22 are obvious in anatomy and structure, but both are homologous appendages and thought to 23 share common genetic features (Yano and Tamura, 2013). Also the development of pectoral 24 fins in fish species is assumed to closely resemble extremity development in higher 25 vertebrates, by sharing common molecular signals arising from a structure called the apical 26 involved "molecular players" have been described to date, not all factors necessary for 27 correct ECM assembly in the regenerating caudal fin have been identified yet. We propose 28 Fndc3a as a novel player in both fin development and regeneration. 29

1
The FNDC3A orthologous gene in zebrafish, fndc3a, is located on zebrafish chromosome 15 2 (ENSEMBL Zv9: 3,066,162-3,114,443 reverse strand; ENSDARG00000067569; ZFIN ID:  3 ZDB-GENE-030131-7015) and encodes in 29 exons for a transcript of 3501 bp. The 4 corresponding zebrafish 1166aa Fndc3a protein (ENSDART00000097261) consists of 4 5 transmembrane domains and 9 fibronectin type III domains. Two different predicted transcript 6 variants have recently been reported for fndc3a (GenBank: XM_021466300.1, 7 XM_021466301), encoding for proteins of 1247 and 1217aa. Both transcript variants are 8 highly similar and differ only in a 30aa stretch at the N-terminus. 9 Phylogenetic and syntheny analyses showed that the FNDC3A gene is highly conserved 10 throughout vertebrate evolution and orthologues are not duplicated in ray-finned fish species 11 (data not shown). Amino acid alignments resulted in an up to 57% amino acid identity with 12 95% coverage, indicating a high level of conservation between human and zebrafish 13 proteins. Two fndc3a paralogues can be identified in the zebrafish genome: fndc3ba 14 To resolve the spatiotemporal expression of fndc3a during zebrafish development, we 22 performed RNA in-situ hybridization experiments. fndc3a transcripts were detected spatially 23 restricted to the tail bud region and the ventral fin fold mesenchyme from 14hpf onwards (hpf 24 = hours post-fertilization; Fig. 1). Expression of fndc3a at later stages of embryonic 25 development is mainly present in distinct brain regions; e.g. mid-hindbrain boundary, 26 rhombencephalon, and mesencephalon; as well as in pectoral fins, in the notochord, in 27 somites and in the caudal median fin fold ( Fig. 1A and Fig. S8A). Restricted expression in the 28 developing tail bud region at later stages (>16hpf) could be observed in the median fin fold, 29 in Kupffer´s vesicle, and in cells of the chordo neural hinge region by longer staining times 30 (Fig. 1B). 31

>Fig. 1 32
Detection of Fndc3a protein localization was performed via immunofluorescence by 33 application of a human FNDC3A antibody. This experiment showed similar regional 34 localization of Fndc3a in Kupffer´s vesicle, the median fin fold region, and the caudal 1 neuronal hinge during different stages of embryonic development consistent with RNA in-situ 2 hybridization (24hpf: Fig. 1C and 48hpf: Fig.1D). The two most noticeable differences 3 between RNA and protein localization were a spotted pattern of Fndc3a in notochord cells 4 and in chevron shaped stripes between somite boundaries (higher magnification images in 5  phenotype ranged from minor fin shape changes in the majority of affected fish, up to axis 16 shortening and stronger caudal fin deformations (Fig. 2E). We did not observe changes in 17 phenotype severity or appearance rates in subsequent, homozygous generations, excluding 18 a potential stronger maternal zygotic effect in fndc3a wue1/wue1 or mitigation of the phenotype. 19 The observation of variable adult phenotypes, incomplete phenotypic penetrance and 20 temperature sensitivity of the embryonic phenotype led us to the assumption that 21 fndc3a wue1/wue1 mutants are hypomorphic. Further investigations of the fndc3a wue1/wue1 mutant 22 line via qPCR were performed to investigate potential nonsense mediated decay of fndc3a 23 mRNAs. We quantified fndc3a mRNA expression in three independent embryo groups with 24 two independent primer pairs ( Fig. 2F; primer sequences are given in Table S1). Relative to 25 AB controls fndc3a transcripts in fndc3a wue1/+ and fndc3a wue1/wue1 mutants were reduced to 26 approximately 60% in heterozygotes and correspondingly to 30% relative expression in 27 homozygotes. This experiment showed a reduction, but not a complete loss, of fndc3a 28 mRNA in the fndc3a wue1/wue1 mutants and thereby implies a potential residual function of 29 Fndc3a in these mutants. Actinotrichia fibers in fndc3a wue1/wue1 mutants were still present, but displayed obvious 34 structural alterations and signs of breakdown in the ventral caudal fin (control: 0/10; fndc3a wue1/wue1 : 7/12). While control fish at 52hpf showed radiant symmetrical arrangement of 1 Col2a in the actinotrichia fibers of the developing caudal fin, fndc3a wue1/wue1 embryos partly 2 lack these structures and depicted unstructured, crumbled collagen fibers in the fin 3 mesenchyme (control: 0/7; fndc3a wue1/wue1 :12/14) . High levels of remaining Col2a in 4 fndc3a wue1/wue1 mutants could be detected in apical cells at the fin border (arrows in Fig.4B), 5 which were also visible as distinct cells in the DIC microscopy (Fig. 4A). of pectoral fins ( Fig. 1; Fig. S8A). A potential effect of fndc3a reduction on pectoral fin 10 development was investigated by phenotypic observation (Fig. S8B) and Col2a as well as 11 TP63 immunofluorescence staining ( >Fig. S10 28 To test this hypothesis, we performed regeneration experiments on adult caudal fins with 29 control and fndc3a wue1/wue1 fish (Fig. 5 and Fig. S9). Two remarkable observations were made: 30 First, regenerates looked opaque, disorganized and tubercular extensions attached to the 31 epidermal layer of the fin regenerates were eminent between 4dpa and 6dpa (days post 32 amputation) in fndc3a wue1/wue1 mutants (arrowheads in Fig. 5A; Fig. S9B  loosely attached cells in the outer epidermal layer of the fndc3a wue1/wue1 regenerates. These 2 cells were additionally investigated by immunofluorescence and depicted high levels of Col2a 3 at the regenerative front (arrows in Fig. 5B). TEM analysis further clarified that these cells 4 were still attached to the epidermal cell layer and incorporate electronic dense material in 5 their Golgi apparatus and in intracellular vesicles (Fig. 5E). In accordance with this 6 observation, fndc3a expression in fin regenerates could be detected in the distal wound 7 blastema at 4dpa and 6dpa (Fig. S9C). Localization of Fndc3a protein in regenerates was 8 confirmed by immunofluorescence in the epidermal cell layers of regenerates (Fig. 5C). 9 Second, actinotrichia fibers at the tip of 4dpa regenerates looked disorganized and clumped 10 in fndc3a wue1/wue1 mutants (arrows in Fig. 5A). In 4dpa regenerates of control individuals TEM 11 analysis of formed actinotrichia fibers showed bundles of stratified actin fibers in close 12 proximity to the basal membrane cells (Fig.5F). In contrast, fndc3a wue1/wue1 mutants lack these 13 prominent compact actinotrichia fibers and only depicted loose filaments adjacent to the 14 membrane layer. Subsequent investigation of epidermal cells by TP63 staining in 15 regenerates 4dpa showed interference with normal regenerate structure and disturbance of 16 epidermal cells in fndc3a wue1/wue1 mutant regenerates (Fig. 5D). This indicates altered 17 epidermal organization as a potential reason for the observed effects in regenerates after 18 reduced Fndc3a function, similar to the processes observed in the median fin fold during 19 early development. 20

>Fig.5 21
Similar to the temperature dependency of the hypomorphic fndc3a wue1/wue1 phenotype during 22 fin development, the observed effects during fin regeneration could be enhanced by keeping 23 fish at a raised temperature of 32°C during the phase of regeneration (incubation at 24°C in 24 Fig. S9A and B; incubation at 32°C in Fig 5A and Fig. S9D). Although prominent cellular 25 abnormalities were detected during the first days past amputation (dpa), the investigated 26 control and mutant fish showed no significant differences in overall tail length growth in the 27 first 10dpa and only at 6dpa a difference in regenerate length was detected (Fig. S10). All 28 fins grew normally to their former size after ~14dpa (data not shown), but small changes in 29 fin morphology, e.g. cooped fin rays and loss of intersegmental tissues, could be observed in 30 fndc3a wue1/wue1 mutants 6 weeks past amputation (wpa; control: 1/10; fndc3a wue1/wue1 : 8/10; 31 Thus, we assume that Fndc3a function during median fin fold development and caudal fin 9 regeneration is provoked by cell shape or by ECM alterations in epidermal cells. 10 To follow-up on this assumption we analyzed cell membrane structure and the ECM in the 11 median fin fold of control and fndc3a wue1/wue1 mutants by investigating F-actin (Phalloidin 12 staining) or β-catenin localization (Fig 6). Cell boundaries of control embryos displayed a 13 dense, stereotypical assembly of epidermal cells in the ventral median fin fold at 22hpf ( Besides the effects on median fin fold development we additionally investigated potential 32 ECM changes in regenerates of fndc3a wue1/wue1 mutants by F-actin ( Fig. 6C and 6D) and β-33 catenin staining (Fig. 6E and 6F). Comparison between controls and fndc3a wue1/wue1 mutants 34 clarified that similar distinctive features, i.e. altered cell matrix and appearance of cavities 35 within the tissue, were also observed in regenerates between 2 and 8dpa. F-actin depicted altered regenerate border shapes of fndc3a wue1/wue1 mutants (dashed white line Fig. 6C) and 1 appearance of regions showing cellular alterations (white arrows Fig. 6C). These obviously 2 fragmented structures within the blastema appeared 2dpa and could be detected until 6dpa. 3 High resolution microscopy clarified, that fragmented structures in 2dpa regenerates are 4 cavities within the blastema (white arrowheads Fig. 6D). Cellular alterations were also 5 detected in cells at the regenerative front of fndc3a wue1/wue1 mutants at 4 to 8dpa by β-catenin 6 immunostaining (Fig. 6E). Irregular blastema borders 4dpa were detected distinctly with the 7 same staining (white arrows and close-up pictures in Fig. 6E). High resolution imaging also 8 revealed β-catenin negative, detached cells outside of the regenerate (white arrowheads in 9 Fig. 6F) and cavities (white arrows in Fig. 6F). Later stages of caudal regeneration did not 10 seem to be compromised by reduced Fndc3a level, as fin regenerates of fndc3a wue1/wue1 11 mutants were able to grow to similar fin lengths as control fish at 10dpa (Fig. S10). Our 12 observations thereby indicate a partial loss of cellular structure and loss of adhesion within 13 the blastema during early stages of caudal fin regeneration in fndc3a wue1/wue1 mutants. 14 >Fig. 6 are known to interact with prominent ECM proteins (Akiyama, 1996)  likely disruption of correct ECM structure in basal epidermal cells is the consequential 8 underlying cellular mechanism responsible for the observed fin malformations. Our results 9 demonstrate a cellular link between median fin fold development and caudal fin regeneration 10 due to the necessity for correct cell shape and tissue cohesion in both processes via Fndc3a. 11 Beyond this, our zebrafish experiments now suggest that Fndc3a can influence TP63 12 positive epidermal cells by altering cell shape or cell adhesion during extremity development. 13 Thus, Fndc3a can be functionally linked to known SHFM genes supporting a potential 14 pathogenic relevance in SHFM phenotypes. Switzerland). We used a fndc3a cDNA fragment of 595bp size (used primers: 10 zf_fndc3a_ribo_fwd2 and zf_fndc3a_ribo_rev2) to synthesize a specific anti-sense RNA 11 probe. Sense probes were synthesized as negative control for each anti-sense probe and 12 were used under the same reaction conditions. Primers used for probe cloning are listed in 13 Table S1. 14 15

Immunofluorescence and histology 16
Immunofluorescence was performed on embryos, cryosections or on whole regenerating fins

13
Scale bars for embryo overview: 250µm; scale bars for tail magnifications: 100µm.        1 Images either show maximum intensity projections (30 to 40 single z-slices; z-distance: 1.5µm) or a 2 representative higher resolution single z slice.