Shootins mediate collective cell migration and organogenesis of the zebrafish posterior lateral line system

The zebrafish sensory posterior lateral line is an excellent model system to study collective cell migration and organogenesis. Shootin1 is a cytoplasmic protein involved in neuronal polarization and axon guidance. Previous studies have shown that shootin1 couples actin filament retrograde flow with extracellular adhesive substrates at the leading edge of axonal growth cones, thereby producing mechanical force for the migration and guidance of axonal growth cones. However, the functions of shootin in peripheral cells remain unknown. Here we identified two novel shootin family members, shootin2 and shootin3. In zebrafish, shootin1 and shootin3 are expressed in the posterior lateral line primordium (PLLP) and neuromasts during embryonic development. A shootin1 mutant displayed a reduced speed of PLLP migration, while shootin1;shootin3 double mutation inhibited cell proliferation in the PLLP. Furthermore, our results suggest that shootin1 and shootin3 positively regulate the number of neuromasts and the number of cells in deposited neuromasts. Our study demonstrates that shootins mediate collective cell migration of the posterior lateral line primordium and formation of neuromasts in zebrafish.


Results
Identification of novel shootin genes shootin2 and shootin3. Shootin1a was originally identified in cultured rat hippocampal neurons 26 . Homologs of the rat shootin1 gene have been identified in human, mouse and zebrafish genomes 26,38 . Here, we identified, by homology search analysis, two predicted shootin1-like proteins in zebrafish (XP_690881.3 and XP_685212.4). To analyze shootin1 and shootin1-like proteins in zebrafish embryos, we cloned their cDNAs. These cDNAs encode 544-, 561-, and 655-amino acid (aa) proteins that share high similarity with mouse shootin1a (Fig. 1a and Supplementary Fig. S2). Phylogenetic and synteny analyses revealed that shootins are clustered into three separate groups and that the gene encoding the 544-aa protein is orthologous to human shootin1 ( Supplementary Fig. S3a,b). We designated the genes encoding 561-and 655aa proteins as shootin2 and shootin3, respectively ( Fig. 1a and Supplementary Fig. S2). The zebrafish shootin1, shootin2 and shootin3 genes are located on chromosome 17, 7 and 14, respectively ( Supplementary Fig. S3b). Rat shootin1a is phosphorylated at Ser101 and Ser249 by Pak1 and these phosphorylations promote axon outgrowth 32,33 . The putative phosphorylation site corresponding to Ser101 is present in all the three zebrafish shootin proteins, whereas that corresponding to Ser249 is present in shootin1 and shootin3 but not in shootin2 (arrowheads, Supplementary Fig. S2).
Shootin1-homologous sequences were identified widely in vertebrates, including fish, frogs, reptiles, birds, marsupials and mammals ( Supplementary Fig. S3a). Shootin2-homologous sequences were identified in fish, frog and Tasmanian devil (marsupial) but not in placental (eutherian) mammals ( Supplementary Fig. S3a). In contrast, shootin3-homologous sequences were identified in fish, such as cavefish, medaka, fugu and coelacanth, but not in other vertebrate lineages ( Supplementary Fig. S3a). Because a whole-genome duplication event occurred in the teleost lineage [39][40][41][42][43] , two orthologs of a gene are often identified in the zebrafish genome, whereas a single copy of the corresponding gene is present in the human genome. Synteny analysis showed that the locus containing zebrafish shootin1 is conserved widely in vertebrates, such as human, mouse, Tasmanian devil, Xenopus and zebrafish ( Supplementary Fig. S3b), whereas the locus containing zebrafish shootin2 is conserved in Tasmanian devil, Xenopus and zebrafish ( Supplementary Fig. S3b). The orders of genes in the loci containing shootin2 and shootin3 were different from those in the loci containing shootin1, suggesting that shootin2 and shootin3 are novel shootin family members.
shootin1 and shootin3 are expressed in the zebrafish lateral line system. To examine whether the shootin family members are expressed in developing zebrafish embryos, we performed RT-PCR using specific primers. All three shootin genes were expressed at 24 h post-fertilization (hpf) and 48 hpf (Fig. 1b). The transcripts of shootin1 and shootin3 were detected at 0 hpf, while we could not detect shootin2 at 0 hpf. Whole-mount in situ hybridization detected all the three shootin genes expressed in the anterior region of embryos at 36 hpf ( Fig. 1c-e). In addition, shootin1 and shootin3 were expressed in the neuromasts (arrowheads, Fig. 1c,c' ,e,e') and the PLLP (arrows, Fig. 1c,c", e, e") of the lateral line system. Zebrafish shootins interact with F-actin retrograde flow at the cellular leading edge. Previous speckle imaging analyses in vitro showed that shootin1a interacts with F-actin retrograde flow at the leading edge of axonal growth cones, and demonstrated that the shootin1a-F-actin interaction promotes growth cone migration 31,32 (Supplementary Fig. S1). We next examined whether the zebrafish shootins interact with F-actin retrograde flow. The speckle imaging assay is a useful method to monitor the retrograde flows of F-actin and clutch molecules 32,44 . As speckle imaging analysis of F-actin flow in vivo is technically difficult, we examined shootin dynamics using cultured XTC fibroblasts, a cell line established from Xenopus laevis 45 . XTC fibroblasts bear large and thin lamellipodia, and are thus particularly suitable for speckle imaging of the retrograde flows of F-actin and clutch molecules 32,44 . AcGFP-shootin1 and mRFP-actin were coexpressed in XTC fibroblasts; speckle imaging was performed as described 31 . The AcGFP-shootin1 signals moved retrogradely in the lamellipodia of XTC fibroblasts ( Fig. 2a and Movie 1), as reported in the case of mammalian shootin1a 31 . Speckles of AcGFP-shootin1 moved with those of mRFP-actin at a similar speed (AcGFP-shootin1: 1.56 ± 0.09 µm/min, n = 51 speckles; mRFP-actin: 1.57 ± 0.09 µm/min, n = 51 speckles; Fig. 2a,d, Movie 1). Speckles of AcGFP-shootin2 and AcGFP-shootin3 also underwent retrograde movement at the leading edge of XTC fibroblasts. The speed of AcGFP-shootin2 speckles was similar to that of mRFP-actin speckles (AcGFP-shootin2: 1.48 ± 0.11 µm/min, n = 54 speckles; mRFP-actin: 1.46 ± 0.11 µm/min, n = 54 speckles; Fig. 2b,d, Movie 2); the AcGFP-shootin3 speckles also moved with mRFP-actin at a similar speed (AcGFP-shootin3: 1.51 ± 0.11 µm/min, n = 53 speckles; mRFP-actin: 1.52 ± 0.09 µm/min, n = 53 speckles; Fig. 2c,d, Movie 3). These results indicate that zebrafish shootin1, shootin2 and shootin3 interact with F-actin retrograde flow at the cellular leading edge.
Zebrafish shootin1 mediates PLLP migration. The interactions of zebrafish shootin family members with F-actin retrograde flow in XTC fibroblasts raise the possibility that shootin1 and shootin3 mediate the migration of the posterior lateral line system. To assess the functions of shootin1 and shootin3 in the posterior lateral line, we generated shootin1 and shootin3 mutants using the CRISPR/Cas9 system. The shootin1 mutant allele contained an 8-bp deletion in the third exon, resulting in a frame shift and a premature stop codon after 47 aa ( Fig. 3a and Supplementary Fig. S4a,b). The shootin3 mutant allele harbored a 13-bp deletion in the second exon, resulting in a premature stop codon after 26 aa ( Fig. 3a and Supplementary Fig. S5a,b). We identified shoo-tin1 mutants by PCR-based genotyping ( Supplementary Fig. S4c) and shootin3 mutant fish by T7 endonuclease I (T7EI)-based genotyping (Supplementary Figs S5c and S6), respectively. To visualize the PLLP, we crossed the fish with SAIGFF213A;UAS:GFP transgenic fish that express GFP in the PLLP 46 . Consistent with a previous report 47 , during 32-38 hpf, PLLP migrated at 88 ± 2 μm/h in control embryos (n = 21) (Movie 4, Fig. 3b). PLLP also migrated in the shootin1 mutant embryos (Movie 5, Fig. 3b); however, the migration speed was significantly slower www.nature.com/scientificreports www.nature.com/scientificreports/ than that in the control embryos (Fig. 3c). On the other hand, no significant difference in the migration speed of PLLP was detected between control and shootin3 mutant embryos (Movies 4 and 6, Fig. 3b,c).
We also generated a shootin1;shootin3 double mutant. As in the case of shootin1 single mutant, the PLLP migration speed was significantly lower than that in control embryos and shootin3 single mutant embryos (Movies 4, 6 and 7, Fig. 3b,c). When shootin1 and shootin3 mRNAs were injected into the double mutant embryos, the reduced speed of PLLP migration was rescued to a level similar to that in the control embryos (Fig. 3b,d). Injection of shootin1 mRNA into the shootin1 single mutant embryos also rescued the reduced speed of PLLP migration ( Supplementary Fig. S7). The migrating PLLP consists of leading and trailing regions 16 ; Cxcr4b is expressed in www.nature.com/scientificreports www.nature.com/scientificreports/ www.nature.com/scientificreports www.nature.com/scientificreports/ all the cells of the PLLP, whereas Cxcr7b is expressed exclusively in the trailing cells 1 . We further examined the polarity of PLLP in mutant embryos, using Cxcr4b and Cxcr7b as markers. As shown in Supplementary Fig. S8a, cxcr4b was expressed widely in the wild-type PLLP, whereas cxcr7b was expressed only in the trailing region. Their expression patterns in the shootin1;shootin3 double mutant PLLP were similar to those in the wild-type PLLP, suggesting that PLLP polarity was not affected by the shootin1;shootin3 double mutation. Taking these data together, we conclude that shootin1 plays a key role in PLLP migration. Shootin1 and shootin3 mediate neuromast formation. Next, we analyzed the number of neuromasts in control and mutant fish after PLLP migration (48 hpf) (Fig. 4). Consistent with previous reports 17, 47 , the average number of neuromasts was 5.4 ± 0.1 (n = 15 embryos) in control embryos, and it was significantly reduced by shootin1 and shootin3 single mutations (Fig. 4a,b). The number of neuromasts was further decreased in the shoot-in1;shootin3 double mutants (Fig. 4a,b). Moreover, the injection of shootin1 and shootin3 mRNAs rescued significantly the reduced number of neuromasts in the double mutants (Fig. 4a,c). We further counted the number of cells in the first deposited neuromasts at 32 hpf by DAPI staining 18,48,49 (Fig. 4d). In control embryos, the number of cells detected in the neuromasts was 27.2 ± 0.8 (n = 14 neuromasts), which is similar to the previously reported number 18 . The shootin1;shootin3 double mutants exhibited a reduced number of neuromast cells, although no significant differences were observed between the control and single mutants (Fig. 4d,e). Furthermore, the injection of shootin1 and shootin3 mRNAs rescued partially the reduced cell number in the double mutants (Fig. 4d,f). These data indicate that shootin1 and shootin3 mediate neuromast formation, by positively regulating both the neuromast number and the number of cells in neuromasts.
shootin1;shootin3 double mutation reduces cell proliferation in the PLPP. To investigate how shoo-tin1 and shootin3 mediate neuromast formation, we quantified the number of cells in the PLLP at 32 hpf using DAPI staining 18,48,49 (Fig. 5a). In control embryos, the number of cells detected in the PLLP was 96.6 ± 3.1 (n = 10 PLLPs), which is consistent with previous data 18 . Although no significant differences were observed in the mean cell numbers in the PLLP between the control and shootin single mutant fish, the shootin1;shootin3 double mutants displayed significantly reduced PLLP cell numbers compared with the control (Fig. 5a,b). We also found that the average number of cells in the double mutant PLLP increased following the injection of shootin1 and shootin3 mRNAs (Fig. 5a,c). We further analyzed the PLLP cell number before the first neuromast deposition (24 hpf). In control embryos, the number of cells detected in the PLLP was 125.6 ± 2.7 (n = 7 PLLPs). As shown in Supplementary  Fig. S9, the cell numbers of the single and double mutant PLLP were similar to those of the wild-type PLLP at 24 hpf.
As previous studies have shown that cells in the migrating PLLP proliferate 47,50,51 , we next analyzed cell proliferation by EdU labelling (Fig. 5d). The ratio of EdU-positive proliferating cells to the total number of cells in the PLLP was 26.7 ± 1.9% (n = 11 PLLPs) (Fig. 5d,e). The ratio of EdU-positive proliferating cells was reduced significantly by the shootin1;shootin3 double mutation (Fig. 5d,e). We finally analyzed cell fate determination of hair cells by examining the expressions of their markers atoh1a and deltaA 1,3 . Expressions of atoh1a and deltaA in the PLLP and first deposited neuromast were detected in wild-type and shootin1;shootin3 double mutant embryos ( Supplementary Fig. S8b). Their expression patterns in the double mutant were similar to those in the wild-type embryo, suggesting that hair cell fate determination was not affected by the shootin1;shootin3 double mutation. Together, these data suggest that shootin1 and shootin3 positively regulate cell proliferation in the migrating PLLP.

Discussion
We report here two novel shootin family members, shootin2 and shootin3. Phylogenetic analyses identified shoo-tin1 in a wide range of vertebrates, including eutherians, marsupials, birds, reptiles, amphibians and fish. Shootin2 was found in marsupial, frog and fish genomes, while shootin3 is restricted to the fish lineage. Zebrafish shootin1 and shootin3 were expressed in the PLLP and neuromasts. The mutation in shootin1 reduced the speed of primordium migration, while shootin1;shootin3 double mutation led to reduction in cell proliferation in the primordium, neuromast number and number of cells in the neuromasts. Together, these results demonstrate key roles of shootins in collective cell migration and neuromast formation of the zebrafish posterior lateral line system.
Concerning the mechanism for shootin1-mediated PLLP migration, our data indicate that zebrafish shoo-tin1 interacts with F-actin retrograde flow at the leading edge of fibroblasts. Previous studies reported that rat shootin1a interacts with F-actin retrograde flow at the leading edge of axonal growth cones, thereby producing force for growth cone migration as a clutch molecule (Supplementary Fig. S1) [31][32][33] . A recent study also demonstrated that mouse shootin1b produces force for neuronal migration through its interaction with F-actin retrograde flow 37 . In the zebrafish PLLP, F-actin accumulates at the leading edge of the leader cells during collective cell migration 52 . Thus, we consider that zebrafish shootin1 is likely to mediate PLLP migration by interacting with F-actin at the leading edge of the leader cells (Fig. 6a,b). Although shootin3 interacted with F-actin retrograde flow, the shootin3 mutation did not have a significant effect on PLLP migration. However, we do not rule out the possibility that shootin3 functions as a clutch molecule. As a possible reason to explain the difference between the phenotypes of shootin1 −/− and shootin3 −/− fish, the efficiency of shootin3-mediated clutch coupling may be weaker than that of shootin1-mediated clutch coupling. Alternatively, the expression level of shootin3 in PLLP may be lower than that of shootin1.
The present study demonstrated that the number of neuromasts as well as the number of cells in neuromasts were reduced in the shootin1;shootin3 double mutant embryos. As the effect of the shootin1;shootin3 double mutation was larger than that of shootin1 single mutation or shootin3 single mutation, we think that shootin1 and shoo-tin3 up-regulate cooperatively the neuromast number and the number of cells in neuromasts. Concerning these results, our data also showed that the shootin1;shootin3 double mutation inhibits cell proliferation in the PLLP, suggesting that shootin1 and shootin3 promote cell proliferation there. Although the molecular mechanisms underlying the shootin-mediated cell proliferation are unclear, we expect that the increased cell proliferation in the PLLP positively regulates both the number of neuromasts and the number of the cells in neuromasts (Fig. 6a). The details of how neuromast formation is regulated by shootin1 and shootin3 remain for further analyses.  Table S1) for reverse transcription. Specific cDNAs were PCR-amplified using the following primers: shootin1-h-Ba-Ko and shootin1-t-Xb-No for shootin1, shootin2-h-Ba-Ko and shootin2-t-Xb-No for shootin2, and shootin3-h-Ba-Ko and shootin3-t-Sp for shootin3 (Supplementary Table S1). The shootin1 and shootin2 fragments were digested with BamHI and XbaI restriction endonucleases, whereas shootin3 fragments were digested with BamHI and SpeI. The cDNAs were then cloned into the BamHI-XbaI site of pCS2 and sequenced using an ABI PRISM 3130 (Applied Biosystems).

Identification of shootin genes and phylogenetic analysis. Shootin family members were identified
in the zebrafish genome using BLAST analysis. Multiple alignments and phylogenetic analysis were performed using Genetyx ver.13 (Genetyx). The coiled-coil domains of shootin proteins were predicted using SMART 54 . The phylogenetic tree was constructed using the neighbor-joining method 55 .

RT-PCR.
To analyze the expression of shootin family members during zebrafish development, RT-PCR was performed using shootin1-rt-f and shootin1-rt-r primers for shootin1, shootin2-rt-f and shootin2-rt-r primers for shootin2, and shootin3-rt-f and shootin3-rt-r primers for shootin3. We used EF1a-rt-f and EF1a-rt-r primers as a positive control. The primers used are listed in Supplementary Table S1. www.nature.com/scientificreports www.nature.com/scientificreports/ www.nature.com/scientificreports www.nature.com/scientificreports/ DNA electrophoresis. The DNA electrophoresis in Fig. 1c and Supplementary Figs S4c and S5c was performed using agarose gels. The images in Fig. 1b were cropped from full-length gel images ( Supplementary  Fig. S10). All the gel images are raw data without modification.
Fluorescent speckle microscopy. XTC fibroblasts, a cell line established from Xenopus laevis 45 , were cultured as described previously 58 and transfected with pAcGFP-shootin1 or pAcGFP-shootin3 and pmRFP-actin 35 using the X-treamGENE 9 transfection reagent (Sigma). Fluorescent speckle imaging was performed as described Generation of zebrafish shootin1 and shootin3 mutants. Zebrafish mutants of shootin1 and shootin3 were generated using the CRISPR/Cas9 system 59 . Vectors for customized guide RNAs (gRNAs) were constructed as described previously 59 . Plasmid pT7-shootin1 (ex3) was constructed by cloning the two annealed oligonucleotides shootin1-f-ex3 and shootin1-r-ex3. pT7-shootin1 (ex4) was constructed by cloning the two annealed oligonucleotides shootin1-f-ex4 and shootin1-r-ex4, and pT7-shootin3 (ex2) was constructed by cloning the two annealed oligonucleotides shootin3-f-ex2 and shootin3-r-ex2. The gRNAs and Cas9 mRNA were synthesized and injected into fertilized eggs as described previously 59 . The injected embryos were raised and crossed with the wild type. To screen shootin1 and shootin3 mutants, a T7EI assay was performed as described previously (Supplementary Fig. S6) 59,60 . We used shootin1-5′ and shootin1-3′ primers for shootin1, and shootin3-5′ and shootin3-3′ primers for shootin3 (Supplementary Table S1). The PCR products were sequenced using an ABI PRISM3130 (Applied Biosystems). As shown in Supplementary Fig. S10, RT-PCR analyses confirmed that there is no detectable expression of shootin1 in shootin1 single mutant or shootin1;shootin3 double mutant embryos. In addition, the expression of shootin3 was undetectable in shootin3 single mutant and shootin1;shootin3 double mutant embryos.
Genotyping. PCR-based genotyping was performed to identify shootin1 mutants. Primers shootin1 (ex3)-wt and shootin1 (ex3)-5′ were used for screening the wild-type allele of shootin1 exon 3, and primers shootin1 (ex3)-mt and shootin1 (ex3)-5′ were used for screening the mutant allele of shootin1 exon 3. To identify shootin3 mutations, T7EI-mediated genotyping was performed using the primers shootin3 (ex3)-5′ and shootin3 (ex3)-3′ and two different T7EI assays. The first T7EI assay was performed as described previously ( Supplementary  Fig. S6a) 59 . In the second T7EI assay, PCR products obtained from samples were mixed with those from the wild type before denaturation at 94 °C for 3 min, annealing at room temperature and digestion of the annealed products with T7EI ( Supplementary Fig. S6b) 60 . The first T7EI assay distinguished heterozygous fish from wild-type and homozygous fish, and the second T7EI assay distinguished between homozygous and wild-type fish ( Supplementary Fig. S6c) 60 .
Tissue labeling and microscopy. Embryos were fixed in 4% paraformaldehyde (PFA) in phosphate-buffered saline (PBS: 137 mM NaCl, 2.7 mM KCl, 10 mM Na 2 HPO 4 , 1.76 mM KH 2 PO 4 , pH 7.4) overnight at 4 °C. The numbers of cells in the neuromasts and PLLP were counted by DAPI staining as described previously 18,48,49 . EdU incorporation was performed as described previously 47 . Embryos were dechorionated and 30.5-hpf embryos were immersed in 500 μM EdU solution on ice for 30 min. After washing, embryos were incubated at 28.5 °C for 1 h and fixed at 32 hpf in 4% PFA overnight at 4 °C. The EdU signals were detected using a Click-iT EdU Alexa Fluor488 Imaging Kit (Invitrogen), according to the manufacturer's instructions. Embryos were embedded in 0.5-1% low melting point agarose (Invitrogen). Confocal images were captured with a Zeiss LSM700 or Zeiss LSM710 microscope and processed using Adobe Photoshop Elements 12 and Fiji 57 .
Statistical analysis. Results are expressed as mean ± standard error (SEM). Statistical analyses were performed with GraphPad Prism 7. Statistical significance was determined by the two-tailed unpaired Student's t-test. For multiple comparisons, we used one-way ANOVA with Tukey's post hoc test.