Sequential yap1 and wwtr1 expression in the musculoskeletal system

Yap1 mRNA is expressed widely in early zebrafish development (Fig 1A–1K) and clearly as slow and fast myogenesis are initiated, respectively, in presomitic pre-adaxial cells and mediolateral stripes of the anterior border cells within each forming somite (Fig 1E and 1G; [79]). When adaxial and posterior somite cells become postmitotic en route to myogenesis, yap1 mRNA is downregulated (Fig 1E–1H), both as adaxial cells undergo terminal differentiation and as presomitic cells become somitic and upregulate myod1 mRNA (Fig 1E,1G, and 1H). By 24 hpf, yap1 mRNA has diminished in terminally-differentiated slow and fast muscle fibres, but remains present in the dermomyotomal cells located on the lateral surface of the myotome, which include precursor cells for later muscle growth (Fig 1J). As muscle precursors undergo terminal differentiation into fibres, yap1 is replaced by wwtr1 mRNA (Fig 1E’,1G’, and 1J’) in cells that have down-regulated myod1 mRNA (Fig 1H’ and 1I’). A similar pattern of yap1 mRNA in superficial precursors and wwtr1 mRNA in differentiated fibres is observed during pectoral fin myogenesis (Fig 1K and 1K’). Thus, yap1 and wwtr1 show sequential expression during early myogenesis.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Expression of yap1 and wwtr1 mRNAs during early development.

(A,A’) Schematic indicating the location of antisense probe on each mRNA transcript. In situ mRNA hybridisation for yap1 mRNA (B-K) or wwtr1 mRNA (B’-K’) shown in lateral whole mount, anterior to top (B,B’-D,D’, F,F’, J,J’, K,K’), dorsal flat mount, anterior to top (E,E’, G,G’, H,H’) or transverse cryo-section (B,B’-D,D’, J, J’, K, K’). Maternal yap1 mRNA augmented by zygotic expression at high stage, contrasted with uniform wwtr1 mRNA zygotic expression (C,C’). (E) At 6ss, yap1 mRNA is widespread in forebrain (bracket), notochord (white arrowhead) and lateral somite stripes (black arrowheads), but less apparent in adaxial slow muscle precursors (black arrows). (E’) Wwtr1 mRNA appears more restricted to mid/hindbrain (bracket) and adaxial cells (black arrows), but is absent from notochord (white arrowhead). (F,G) At 10-12ss, notochord (white arrowhead), forebrain (bracket), tail bud (white asterisk) and lateral stripes in nascent somites (black arrowheads) contain yap1 mRNA, but expression appears diminished in more mature somites. (F’,G’) Accumulation of wwtr1 mRNA is observed in adaxial cells (black arrows), stripes in lateral somites (black arrowheads) and mid/hindbrain (bracket), but not in notochord (white arrowhead) or tail bud (white asterisk). (H) At 15ss, myod1 mRNA with yap1 mRNA localises in the anterior half of the somites, alternating with myod1 mRNA (black arrowheads) and notochord (white arrowhead). (H’) Conversely, wwtr1 mRNA co-localises with the myod1 mRNA in the posterior half of each somite (black arrowheads) and adaxial cells (black arrow). (I) Schematic indicating location of flat-mount (I) and 20 μm transverse cryosections (II and III) in J and J’. I’. Schematic indicating location of 15 μm transverse cryosections showing wwtr1 and myod1 mRNA in somitic regions at 32 hpf. (J, J’) At 24 hpf, low level yap1 mRNA was present in dorsal tip (yellow arrowhead) of somite (outlined in yellow dots), whereas wwtr1 mRNA was detected throughout the somitic myotome (yellow asterisks). Both yap1 and wwtr1 mRNAs were present in the eye (white asterisks) and head (brackets). Yap1 mRNA persisted in the tail bud (black asterisk). (K, K’) At 50 hpf, yap1 mRNA persisted at dorsal and ventral somite extremes and was present in heart (inset, black arrowhead) and pectoral fin (red arrowheads) whether viewed from anterior or in 20 μm transverse cryosection (red arrow). Yap1 and wwtr1 mRNAs persisted in the head (white asterisks), but both were reduced in the myotome (yellow asterisks). In pectoral fin, wwtr1 mRNA was detected in two regions consistent with the differentiating muscle masses (red arrows and arrowheads). Bars = 100 μm, except K and K’ 50 μm.

https://doi.org/10.1371/journal.pgen.1012172.g001

Yap1 mRNA also accumulates in the nascent notochord before being down-regulated throughout its length by 24 hpf (Fig 1E,1F,1G and 1J). Wwtr1 mRNA, however, was not detected in the notochord or in nascent fin cartilage (Fig 1E’,1F’,1G’ and 1K’).

Myod is required to suppress yap1 and induce wwtr1 mRNA in somites

Skeletal myogenesis in zebrafish is driven by the myogenic regulatory factors (MRFs) Myf5, Myod, Myogenin and Mrf4/Myf6 [11,80–82]. Whereas yap1 mRNA precedes expression of the myogenic transcription factor myod1, wwtr1 mRNA is expressed after myod1 mRNA in differentiated muscle deep within the myotome (Fig 1H,1I and 1H’,1I’). The time course of wwtr1 mRNA accumulation matches that of myog and myf6 [82]. To test the relationship between myogenesis and yap1/wwtr1 expression, embryos lacking each MRF gene were subjected to in situ mRNA hybridisation for yap1 and wwtr1 and compared to their non-mutant siblings. Loss of myod1 function delays differentiation of fast muscle [11,83]. Congruently, myod1 mutants show reduced wwtr1 expression and persistence of yap1 mRNA in the lateral somite (S1A and S1B Fig). As expected, expression of yap1 and wwtr1 was unaltered in other regions of myod1 mutant embryos (S1A and S1B Fig). Mutation of myog does not prevent muscle terminal differentiation, but reduces fusion into multinucleate fibres [80]. Yap1 mRNA appeared unaltered in myog mutants, whereas wwtr1 mRNA was somewhat reduced in muscle tissue (S1C and S1D Fig). Loss of function of either myf5 or myf6 had no discernible effect on yap1 or wwtr1 mRNA accumulation (S1E-S1H Fig). These observations support the view that a switch from expression of Yap1 to Wwtr1 accompanies zebrafish skeletal muscle terminal differentiation.

Mildly impaired larval movement in wwtr1 mutant

To test the role of Yap1 and Wwtr1 in the musculoskeletal system, each gene was mutated using TALEN-mediated genome editing in the coding region of the first coding exon (S2A and S2B Fig). Among various nonsense alleles obtained of each gene, a trend in nonsense-mediated mRNA decay (NMD) was observed depending on nonsense tail length. One class of mutants had long nonsense tails and showed strong NMD (S2C- S2E Fig). Mutants shifting to the other nonsense reading frame resulted in a short nonsense tail and much less NMD (S2F and S2G Fig). Yap1 mRNA was not detectably upregulated in wwtr1 mutant larvae (S2H Fig). Wwtr1 mutants, when crossed onto a membrane-EGFP reporter to permit confocal analysis [84], did not show an obvious morphological defect before 5 dpf and muscle growth appeared to be normal (Figs 2A and S3A). However, fewer wwtr1kg^169/kg169 mutants (we use the form wwtr1kg¹⁶⁹ for such homozygote mutants hereafter) survived to adulthood than their co-habiting siblings (S3E Fig, n = 165, Χ² p = 0.017). To ask if the lack of Wwtr1 affected muscle function, we used two movement assays. At 4 dpf, larvae from a wwtr1kg^169/+ in-cross were anaesthetised and tested for their response to electrical stimulation by measuring the maximum angle of evoked trunk movement. Wwtr1kg¹⁶⁹ mutants moved less than their siblings (p = 0.02; Fig 2B). However, wwtr1kg¹⁶⁹ mutants did not show a slower swim speed response to touch at either 5 or 11 dpf (Fig 2C and 2D) and grew normally at later stages (see below, Fig 4B). Whereas electrically-evoked tail movement likely reflects a high force regime in which movement may be limited by altered trunk structure or extracellular matrix deposition, swimming speed may represent a lower force regime under which a motility defect is not apparent. These results show that Wwtr1 is not essential for muscle growth but is required transiently for optimal muscle function during early larval stages.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Wwtr1 mutants are defective in evoked motility.

(A) Muscle growth analysed in sibling wild type (wt), heterozygote and wwtr1kg¹³³ mutant bred on the ß-Actin:membEGFP background and reared at 28.5ºC until 5 dpf and then at 26.5ºC to the indicated age. Wwtr1kg¹³³ mutants and their siblings have similar muscle growth. Note that apparent gaps (asterisks) between fibres reflect reporter mosaicism, not absence of fibres. Bar = 50 μm. (B) Fish movement in response to electrical stimulation under anaesthetic was reduced in wwtr1kg¹⁶⁹ mutants compared to siblings. Bar = 500 μm. (C,D). Swim velocity in response to touch stimulation at 5 dpf (C) and 11 dpf (D). Number of fish analysed is indicated on columns. Significant statistical results are shown (A-D) for Kruskal-Wallis tests with Dunn’s post hoc comparisons, with the exception of the highly significant main effect of age is in A.

https://doi.org/10.1371/journal.pgen.1012172.g002

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Yap1kg¹⁵¹ mutants are defective in larval movement.

Muscle properties analysed in yap1 mutants and their non-mutant siblings on the ß-Actin:membEGFP background reared at the indicated temperature from 70% epiboly until prim 10 (B, see schematic) or 5 dpf-equivalent (A,C-E) and 26.5ºC thereafter (E). (A) MZyap1kg¹³⁷ or MZyap1kg^137/kg152 mutants and their heterozygote siblings grown at non-permissive temperature have similar muscle growth. Lack of signal in occasional fibres due to mosaic transgene expression (asterisks) is distinguished from broad loss of fluorescence (#) in oedematous MZ mutants at 5 dpf. Bar = 50 μm. (B) Yap1kg¹⁵¹ mutants grown as schematised had myotome volume indistinguishable from their heterozygous and wt siblings. (C-E) Yap1kg^151/+ in-cross grown at permissive temperature from 70% epiboly. Symbol shape indicates lay. Bar = 500 μm. (C) Yap1kg¹⁵¹ mutants at 6 days (equivalent to ~4.25 dpf) swim away from touch stimulation slower than their siblings (upper panel), even when corrected for shorter total length (lower panel). (D) When stimulated by electrical pulses, yap1kg¹⁵¹ mutants at 7 days (equivalent to 5 dpf) move their trunk through a smaller angle in comparison to their siblings. (E) Swimming velocity of yap1kg¹⁵¹ mutants at 13 days (equivalent to 11 dpf) is variable between lays. Symbol shapes represent different lays/experiments (3 lays C, 2 lays D,E). Number of fish analysed is indicated on columns. Significant statistical results are shown (A-E) for Kruskal-Wallis tests with Dunn’s post hoc comparisons.

https://doi.org/10.1371/journal.pgen.1012172.g003

Yap1 mutant fish are defective in movement

We next assessed the phenotype of homozygous yap1 mutants observing, as previously reported, temperature-dependent variability [25,85]; S3B and S3C Fig). At the normal rearing temperature of 28.5 ºC, some yap1 mutants had reduced eyes, pericardial oedema and shortened or curled bodies (S3B and S3C Fig). When two lays were reared to adulthood at 28.5 ºC, no mutants were found among 97 surviving siblings (Χ², p = 1 x 10^-8). Reared at a higher, restrictive temperature of 32 ºC, eye, body, oedema and lethality phenotypes were fully penetrant in mutants by the equivalent of 5 dpf, whereas their siblings were unaffected (S3C Fig). Conversely, rearing at the lower permissive temperature of 20.5 ºC supressed these gross morphological phenotypes and mutants survived beyond 5 dpf-equivalent (7 days). When such surviving mutants were subsequently reared through the nursery at 26.5 ºC in tanks with their non-mutant siblings, approximately half survived to 3 months of age (S3D Fig; n = 423 survivors, Χ² p =6 x 10^-6). Husbandry and legal considerations prevented rearing at 20.5 ºC beyond 5 dpf-equivalent. Surviving mutant females bred poorly. Nevertheless, some maternal-zygotic (MZ) yap1 mutant larvae obtained, when reared at the permissive temperature until 7 days, were motile and survived to adulthood. Thus, yap1 is dispensable for early myogenesis.

Failure to thrive in the nursery and breed might reflect poor muscle function leading to low food intake. So we investigated muscle at pre-feeding stages by analysing muscle growth in MZyap1kg¹³⁷ mutants or MZyap1kg^137/kg152 trans-heterozygotes on a plasma membrane-EGFP reporter background (Fig 3). Analysis of myotome volume and muscle fibre number revealed no defect in maternal-zygotic (MZ) yap1 mutants compared to their heterozygous siblings (Fig 3A). Even mutants showing oedema grew muscle normally, despite becoming sick and necrotic at 5 dpf (Fig 3A). To avoid the problem of the early oedema and other phenotypes, yap1kg¹⁵¹ mutant embryos were grown at the permissive low temperature from 70% epiboly to the prim 10 stage (3 days of incubation; see Materials and Methods for explanation of rearing and staging at low temperatures) and then shifted to the restrictive higher temperature during early larval muscle growth. At 5 days of incubation, the equivalent of 4 dpf stage, no difference in myotome growth was detected between yap1kg¹⁵¹ mutants and their siblings (Fig 3B). At 7 days of incubation, 5 dpf-equivalent, no defects in sarcomere structure were observed either in live larvae or after immunodetection of α-actinin (S4 Fig). Wwtr1 mRNA was not detectably upregulated in yap1kg¹⁵¹ mutants, either in larvae or in adult muscle (S2H Fig). Thus, Yap1 is not required for muscle growth at pre-feeding stages.

To ask if the lack of Yap1 affects muscle function, we next looked at yap1kg¹⁵¹ mutants using the two movement assays. Siblings from yap1kg^151/+ in-crosses were reared at the permissive temperature until 6 days (4 dpf-equivalent) and measurement of swim velocity in response to touch revealed that yap1kg¹⁵¹ mutant swam slower than their siblings (p ≤ 0.04; Fig 3C; hereafter fish grown at 20.5ºC from 70% epiboly to 5 dpf-equivalent and then transferred to 26.5ºC are staged in days, not dpf). At 7 days, maximal evoked trunk movement in yap1kg¹⁵¹ mutant fish was also reduced compared to siblings (p ≤ 0.003; Fig 3D). Neither defect could be accounted for by altered size of mutants (S5A and S5B Fig). When yap1kg¹⁵¹ mutants were assayed for swim speed at 13 days the defect was less apparent (Fig 3E). One lay of yap1 mutants suggested persistence of a movement defect, whereas another lay did not (Fig 3E). We conclude that yap1 mutants have significant motility defects in early larval stages that appear more severe than those in wwtr1 mutants.

Cooperativity between Yap1 and Wwtr1

Given the transient nature of the motility defects in yap1 and wwtr1 single mutants, we asked if these paralogous genes have redundant function. As reported previously with other alleles, yap1kg¹⁵¹;wwtr1kg¹⁶⁹ double mutants were not viable, showing a failure of tailbud outgrowth with misshapen somites (S6A Fig), and death at around 1 dpf, which is suggested to arise from a periderm defect [78]. Analysis of muscle development in double mutants did not reveal any defect in myogenesis, either when analysing expression of MRF genes at the 16 somite stage (16ss; S6B Fig), or in terminal differentiation and morphogenesis of slow and fast muscle at 18ss (S6C Fig). At the 24ss, shortly before their death, fast muscle tissue was well differentiated although of aberrant morphology in yap1kg¹⁵¹;wwtr1kg¹⁶⁹ double mutants (S6D Fig). Thus, despite their early and sequential expression within the myogenic lineage, neither Yap1 nor Wwrt1 is required for early myogenesis.

We next asked if loss of one allele of wwtr1 or yap1 on a null mutant background of the other gene would affect the muscle growth at later stages. Even when reared at the restrictive temperature, at 2 dpf no difference in myotome size was apparent in yap^kg151/+;wwtr1kg¹⁶⁹, yap^kg151;wwtr1kg^169/+ or its MZ yap^kg151;wwtr1kg^169/+ counterpart (S6E and S6F Fig). MZ analysis could not be performed on wwtr1 mutants because eggs from wwtr1 mutant females could not be fertilised, as previously suggested [86]. Despite their good initial myogenesis, yap1kg¹⁵¹;wwtr1kg^169/+ fish showed more severe and earlier defects than their yap1kg¹⁵¹ siblings elsewhere in the body (S3A Fig), and yap1kg^151/+;wwtr1kg¹⁶⁹ larvae developed mild cardiac oedema and failed to survive to adulthood (S3A and S3F Fig). Thus, yap1 and wwtr1 show partial genetic redundancy.

Yap1 mutant fish develop kyphoscoliosis

To investigate the role/s yap1 and wwtr1 play in later development, surviving single mutants were assessed further (Fig 4). Adult yap1kg¹⁵¹ mutants were significantly shorter and lighter than their siblings, although Fulton’s condition factor (k = Weight/Length³; [87] was not reduced in mutants (Figs 4A and S6C). A similar size reduction was observed in yap1kg¹³⁷ and yap1kg¹⁵² mutants (S7B and S7D Fig). In contrast, the length, weight and k factor of wwtr1kg¹⁶⁹ mutants were similar to their siblings (Figs 4B and S7A). Microtomographic (CT) examination of yap1kg¹⁵¹ siblings raised at the permissive temperature for 7 days (5 dpf-equivalent) revealed severe kyphoscoliosis in some surviving adult yap1kg¹⁵¹ mutants (Fig 4C). Histological analysis revealed no obvious muscle defect in yap1 mutants, whether kyphoscoliotic or not (Fig 4D).

To assess further the kyphoscoliotic phenotype in yap1kg¹⁵¹ mutants, a lay was raised to 5.5 months and kyphoscoliosis was again observed in some mutants (Fig 4E). From 86 genotyped fish, all 17 surviving yap1kg¹⁵¹ mutants and 17 wildtype siblings were closely analysed by alizarin red staining; 3/17 mutant fish had a severe kyphoscoliotic phenotype and 4/17 fish were less severely curved (Fig 4E). Ten mutant fish were straight and similar to their wildtype siblings, although in general shorter in length (Fig 4E). About a third of mutants died before adulthood (Figs 4E,4F and S3D; n = 128 surviving mutants out of 763 fish of 3 to 5.5 months of age, Χ² p = 4 x 10^-6) and about a third of these surviving yap1 mutants exhibited kyphoscoliosis (n = 21/66 mutants, Χ² p = 1 x 10^-12; Fig 4E and 4F). Commensurately, whole spine length and vertebral length measured in the caudal region was shorter in the straight yap1kg¹⁵¹ mutants (Fig 4G and 4H). The dorsoventral height of the posterior end of the vertebral centrum was reduced in mutants (Fig 4H). Thus, loss of Yap1 function leads to a variety of vertebral phenotypes including smaller vertebrae which, in some individuals, develops into a strong kyphoscoliosis and failure to thrive.

To ask if kyphoscoliosis was more penetrant in MZ mutants, a lay from a yap1kg¹⁵¹ mutant female with a heterozygous yap1kg^151/+ male were reared at the permissive temperature until 5 dpf-equivalent and then at 26.5ºC until 27 days. Kyphoscoliosis occurred at a similar rate to zygotic mutants, with 3/10 MZyap1kg¹⁵¹ mutants exhibiting a bent spine (Fig 5A), suggesting that maternal Yap1 does not protect zygotic yap1 mutants from kyphoscoliosis.

When the forming spine of larval fish raised at the permissive temperature was analysed with the live bone stain calcein [88], kyphoscoliosis was apparent in 5/12 mutants at 18 days (Fig 5B), developing from a more mild kyphoscoliosis that was already visible in some larvae at 13 days (Fig 5C). We conclude that about a third of yap1kg¹⁵¹ mutants develop kyphoscoliosis from as early as 13 days (11 dpf-equivalent).

Investigating the cause of kyphoscoliosis

Three reported causes of kyphoscoliosis in zebrafish are defective notochord development, alterations of cerebrospinal fluid flow in the neural canal and perturbation of the Reissner’s fibre. We investigated each but failed to find evidence of defects in yap1 mutants (S8–S10 Figs).

Multiple mutant zebrafish lines that have disorganised notochord cells develop kyphoscoliosis [89–91]. Yap1 mRNA is present early within the notochord (Fig 1E–1G). We therefore examined yap1kg¹⁵¹ mutants reared at the permissive temperature until 3 days (48 hpf-equivalent). No defects were noted in notochords of six mutants examined by compound microscopy for size and vacuolation at 3 dpf (S8 Fig), nor in >200 embryos resulting from four separate yap1kg^151/+ in-crosses. Vertebral centrum defects frequently arise in mutants with notochord defects [92]. Our observations at 13 days failed to reveal defects in vertebral centrum formation (Fig 5C), further arguing against an early notochordal origin for the kyphoscoliosis.

Murine Yap1 deficiency causes hydrocephalus [93]. Defective cilia in cerebral ventricles lead to abnormal flow of cerebrospinal fluid that, in turn, has been suggested to cause both hydrocephalus and kyphoscoliosis [94,95]. Scanning electron microscopy of the rhombencephalic ventricle of kyphoscoliotic and non-kyphoscoliotic yap1kg¹⁵¹ mutants and their wild type siblings, failed to reveal either hydrocephalus or ciliary defects (S9 Fig). We conclude that Yap1-dependent kyphoscoliosis is not accompanied by a visible defect in ventricular ciliary morphology.

We did not examine ciliary function in the spinal canal, which is required for proper Reissner’s fibre formation. Kyphoscoliosis can arise in zebrafish with defects in the Reissner’s fibre, located within the spinal canal [96,97]. However, immunodetection of the Reissner’s fibre revealed its presence at the 5 dpf-equivalent stage in yap1 mutants grown at the permissive temperature (S10 Fig).

Another cause of kyphoscoliosis in zebrafish is loss of Dstyk activity, which also reduces notochordal collagen expression and shortens the body axis by 2 dpf [98]. We did not observe either a shortening of the body axis or reduction in notochordal col9a1b or col8a1a mRNA in yap1kg¹⁵¹ mutants (Fig 6A and 6B). However, yap1kg¹⁵¹ mutants did display focal losses of strong col8a1a mRNA accumulation in the hypochord in all seven mutants genotyped (Figs 6B and S11). Other genes, such as fast myosin heavy chain genes fmyhc1.2 and fmyhc2.1, which are regionally expressed along the body axis, showed no defects (Fig 6C and 6D). We conclude that Yap1 is essential for normal patterning of hypochord development but how this early phenotype relates to later growth defects and kyphoscoliosis is unclear.

Larval growth requires Yap1

Growth of yap1 mutants was studied in more detail. Fish from a yap1kg^151/+ in-cross were reared at the low permissive temperature until 7 days to allow early survival and their growth under optimal husbandry conditions followed at 26.5ºC in our professional aquarium. Yap1kg¹⁵¹ mutant fish analysed at 13 days were already smaller in total fish length, spine length, head size and tail fin calcified area (Fig 7A-7E). Although fish with bent spines were not particularly short at 13 days (Fig 7B), by 18 days scoliotic fish were amongst the shortest in the lay (Figs 5B and 7F). At 13 days, spine and total fish lengths were reduced similarly (Fig 7G). Thus, Yap1 is required for normal growth of the larval skeleton.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. Adult yap^kg151 have kyphoscoliosis and/or smaller vertebrae.

Heterozygous in-crosses of yap1kg^151/+ reared at the permissive 20.5ºC temperature (A,C-H) or wwtr1kg^169/+ reared to 28.5ºC (B) until 5 dpf-equivalent and then at 26.5ºC to the indicated age, points plotted with jitter to avoid overlap. (A) Reduced growth of yap1kg¹⁵¹ mutants. Some yap1kg¹⁵¹ mutant individuals showed a dramatic tail truncation not observed in non-mutant (inset). (B) At 3 months, adult wwtr1kg¹⁶⁹ mutants have similar weight and length to their wild type and heterozygous siblings but survive poorly (Χ², n = 165, p = 0.02). (C) Computerised Tomography (CT) scans of adult fish at 4.5 months viewed from lateral and dorsal. (D) NADH tetrazolium reductase stain showing comparable oxidative/slow (arrows), intermediate (arrowheads) and glycolytic/fast fibres (asterisks) between wild type and yap1kg¹⁵¹ mutant muscle in superficial (left) and medial (right) myotome transverse cryosections taken at 6 months. (E) Lateral view of wholemount alcian red stained 5.5 month old skeletons. (F) Frequency of curved spine within sibling 4.5-5.5 month adult fish is greater in surviving mutants from four separate yap1kg^151/+ in-crosses compared to their siblings (Χ², n = 340, p = 1 x 10^-11). (G) Spine length of 5.5 month sibling wt and yap1kg¹⁵¹ fish. (H) Caudal vertebrae length, height and posterior height of 5.5 month straight and curved yap1kg¹⁵¹ fish. Statistics show significant results of Kruskal-Wallis tests with Dunn’s post hoc comparisons (G,H). Bars: A,C and E = 5 mm; D = 100 μm. Number of fish analysed is indicated on columns.

https://doi.org/10.1371/journal.pgen.1012172.g004

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. Kyphoscoliosis arises in yap1kg¹⁵¹ mutants from 13 days.

(A) Larval fish at 27 days deriving from a female yap1kg¹⁵¹ x male yap1kg^151/+ cross reared at 20.5 ºC until 7 days (5 dpf-equivalent) and then at 26.5 ºC. Wholemount fish before (upper) and after (lower) skeletal stain for bone (red) and cartilage (blue). Note reduced dorsoventral extent in severely kyphoscoliotic mutant (right). (B,C) Several examples of larval fish deriving from a yap1kg^151/+ in-cross reared at 20.5 ºC until 7 days and then at 26.5 ºC stained live with calcein. Lateral views of 18 days (B) and 13 days (C). Numbers indicate fraction of fish of each genotype with morphology shown. Bars = 1 mm.

https://doi.org/10.1371/journal.pgen.1012172.g005

Download:

• PNG
  larger image
• TIFF
  original image

Fig 6. Altered hypochord marker distribution in yap1kg¹⁵¹ mutant.

In situ mRNA hybridisation for collagen genes implicated in kyphoscoliosis and fast myosin heavy chain genes involved in axial patterning in sibling embryos reared from a yap1kg^151/+ in-cross at non-permissive temperature (28.5ºC) that were photographed and subsequently PCR genotyped. (A,B) Collagen extracellular matrix genes col9a1b (A) and col8a1a (B) are unchanged in 36 hpf notochord (arrows) of mutants compared to their siblings. Col8a1a mRNA is abundant in sibling hypochord throughout the axis (B, arrowheads, shown magnified in insets), but lacking in several segments in mutants (B, brackets). Additional col8a1a-stained embryos of each genotype are shown in S11 Fig. (C,D) In contrast, the axially-patterned localisation of fast myosin heavy chains fmyhc1.2 (D, trunk) and fmyhc2.1 (E; tail) is unaltered at 48 hpf. Numbers of (individuals with the pattern shown)/ (total genotyped individuals) are displayed on each panel. Bars = 100 μm.

https://doi.org/10.1371/journal.pgen.1012172.g006

Download:

• PNG
  larger image
• TIFF
  original image

Fig 7. Yap1 is required for normal larval growth in a group.

Yap1kg^151/+ in-crosses were reared at 20.5°C from 70% epiboly to 7 days (5 dpf-equivalent) and then at 26.5°C until 13 days (11 dpf-equivalent) or 18 days (16 dpf-equivalent). Larvae were stained with calcein. (A) Schematics of measurements made (white). (B-G,H,M,N) Total fish length (B,F,N), head area (C), spine length (D), tail fin calcified area (E), Spine length divided by fish length ratio (G), number of calcified vertebrae (H) and myotome 17 volume/total fish length cubed (M). (I,J,K) After segregation into groups containing 28 or 29 calcified vertebrae, total fish length (I) and head area (J), but not tail calcified area (K), were reduced in yap1 mutants compared to wt siblings. (L) Volume of myotome 17 in 13-14 day fish reared in group nursery environment (clear bars) or individually with unlimited food (striped bars). (N) Length of fish at 5, 7, 9, 11 and 13, 15, 17 and 18 days reared individually with unlimited food (striped bars) or in group nursery environment (clear bars). Statistics: Kruskal-Wallis tests with Dunn’s post hoc comparisons. Green bars wt siblings, blue bars yap1kg^151/+ siblings red bars yap1kg¹⁵¹. Red circle symbols represent curved spine fish (B-F,H-K). Symbol shape represents different lays/experiments (L,N). Number of fish analysed is indicated on columns.

https://doi.org/10.1371/journal.pgen.1012172.g007

Yap1kg¹⁵¹ mutant fish appeared slightly delayed in development, as mutants had on average one fewer calcified vertebrae than their siblings (p = 0.0005 Mann-Whitney; Fig 7H). When larvae were stage-matched according to their number of calcified vertebrae, yap1kg¹⁵¹ mutants still had a growth defect compared to their siblings, with shorter total length and head area (p ≤ 0.02 Mann-Whitney; Fig 7I and 7J). These reductions were not more pronounced in bent compared to straight mutants (Fig 7I and 7J compare circle and triangle symbols). In contrast, tail fin calcification appeared unchanged after stage-matching mutants and siblings (Fig 7K), suggesting this feature is indirectly dependent on Yap1. These findings suggest that yap1 mutants have a gross skeletal growth defect independent of a subtle developmental delay.

The skeletal reduction in yap1kg¹⁵¹ mutants was accompanied by defects in muscle growth. The myotome volume measured between 13 and 14 days was reduced in mutants housed with their siblings (p = 0.01 Mann-Whitney; Fig 7L). Fish housed alone grew more slowly despite ad libitum feeding and had yet to show a muscle defect at the same age (Fig 7L). Although the yap1kg¹⁵¹ mutant fish reared in groups had smaller myotome volumes, this size difference was removed if we divided it by body length cubed, to measure their overall condition [99,100], an analogue of Fulton’s k; Fig 7M). Therefore, despite a clear skeletal defect in yap1 mutants, there is little indication of a disproportionate muscle defect that might account of the skeletal changes.

To ask when the growth defect arose, fish reared under the permissive low temperature regime until 5 dpf-equivalent (7 days) were measured at 5, 7, 9, 11 and 13 days. Under standard husbandry conditions, zebrafish get nutrients entirely from their yolk for the first 5 dpf, at which point they start to feed but yolk is not fully consumed until at least 10 dpf. After the first length measurement, the fish were raised in a light cycling incubator at 28.5°C in individual 35 mm wells in 6 well plates to enable longitudinal study of each fish. Fish were fed rotifers as they would be in the aquarium from 7 days. Surprisingly, these fish remained small and yap1kg¹⁵¹ mutants showed no difference in length or muscle volume from their siblings even at 13 days, nor did any have kyphoscoliosis (Fig 7L and 7N). However, when the experiment was repeated and fish raised individually through 15 and 17 days, yap1kg¹⁵¹ mutants began to show a growth defect (Fig 7N). We conclude that Yap1 is required for normal larval growth during the mid-larval period.

Temperature sensitivity of yap1 putative null alleles

Four separate yap1 early-truncating mutant alleles occurring in distinct exons and isolated in different laboratories, yap1^sa25458, yap1^mw48, yap1^zf2210 and yap1kg¹⁵¹, have now shown temperature sensitivity, with higher temperatures being non-permissive [25,85,101]; the current study). Although none of these alleles is a deletion, they are likely to reveal the null phenotype. Yap1kg¹⁵¹ causes translational termination early in the first coding exon, shows nonsense-mediated mRNA decay, and lacks evidence of genetic compensation. Moreover, as previously reported with other yap1 alleles that lack Yap1 immunoreactivity, at the non-permissive temperature yap1kg¹⁵¹;wwtr1kg¹⁶⁹ dual loss of function is embryonic lethal during late somitogenesis stages, whereas loss of yap1 function alone leads to cardiac and yolk sac oedema, small eyes and death in the early larval stage in both zygotic and MZ mutants [25,78]. At the permissive temperature, almost all yap1kg¹⁵¹ mutants appear wild type at 5 dpf. Loss of wwtr1 function, on the other hand, leads to viable adults despite the reported cardiac and other defects [25,48,78,86,102–104].

Temperature sensitivity of null mutants has been reported in bacteria and yeast [105–107]. In animals, however, ts alleles are generally missense single amino acid substitutions affecting protein folding [108]. Interestingly, a truncating loss of function mutation (possibly not null) in scrib, which interacts genetically with yap1, is also temperature sensitive [85]. We suggest that temperature-sensitivity of unknown modifier gene/s or physiological processes account for variable penetrance of yap1 mutant alleles. The role of Yap1 in the response to environmental temperature merits further investigation.

Yap1 is required for larval motility and growth

Normal Yap1 function is essential to avoid early developmental defects, including oedema and eye defects. When yap1kg¹⁵¹ mutant fish are grown at non-permissive temperatures, the larvae do not survive. Pericardial oedema is frequently attributed to cardiovascular defects and Yap1 has been extensively implicated in cardiac development [109–113]. Pericardial and yolk-sac oedema has also been attributed to kidney or liver defects [114,115] and yap1 has been reported to play roles in both kidney and liver development [116–118]. Interestingly, the earliest defect observed in yap1kg¹⁵¹ mutants is an uneven axial distribution of col8a1a mRNA in the hypochord, an axial midline structure underlying the notochord. Hypochord signals, such as Vegfc, have been implicated in lymphangiogenic and vascular patterning [119], defects in which might contribute to the cardiovascular phenotype of mutants. However, gross vascular flow or morphology defects were not observed at 36 hpf; further study is warranted. Thus, the data are consistent with roles of Yap1 in early development of a variety of organs.

When grown at the low permissive temperature until the 5 dpf-equivalent stage, fish lacking zygotic Yap1 have a motility defect but normal body and muscle size and, even when shifted to non-permissive conditions at 5 dpf-equivalent, some go on to survive to adulthood. Nevertheless, following Yap1 inactivation from 5 dpf-equivalent, yap1kg¹⁵¹ mutants show an early growth deficit, which amounts to about a 10% reduction in length, head and tail size. A subset of such yap1 mutants show no morphological or obvious behavioural defects, but have a slight delay in vertebral calcification and growth. These ossification defects correlate with the reduction in body length. Various murine Yap1/Wwtr1 mutant combinations show reduced overall growth [54], as we observe in yap1 mutants. The findings suggest that the growth defect in yap1kg¹⁵¹ mutants arises as early as 13 days and persists into adulthood, accompanied by poor survival.

The mechanism underlying the motility defect is unclear because Yap1 protein is expressed in multiple tissues. Histological and size analysis of the yap1 mutants at early larval stages show myotome size and cellularity are similar to wild type siblings, even though early motility is required for muscle growth [18]. Although the function of larval muscle is poor, adult muscle appears histologically normal except for the reduced size of the fish. TEAD transcription factors can mediate Yap1 effects on transcription and a recently isolated tead1a mutant appeared ‘slimmer’ than wild type; however scoliosis was not reported [120]. Resolution of the cell autonomy and mechanism of Yap1 action will await tissue-specific genetic manipulation.

Yap1 helps maintain vertebral column symmetry

Yap1 is required to ensure symmetrical vertebral column morphogenesis. A kyphoscoliotic spinal phenotype of variable severity arises when mutant larvae are grown at the permissive temperature until 5 dpf-equivalent stage and the non-permissive temperature thereafter. Whereas the only early cartilage or bone defect in the backbone of larvae was a slight delay in calcification, some individual mutants showed a mildly curved spine at 13 days and by 1 month of age kyphoscoliosis was apparent in around a third of zygotic yap1 mutants. Similar defects were not observed in wwtr1kg¹⁶⁹ mutants, yap1kg^151/+ heterozygotes or yap1kg^151/+;wwtr1kg^169/+ dual heterozygotes even if grown at the non-permissive temperature throughout their development. Analysis of yap1 MZ mutants revealed a similar phenotype to the zygotic mutants; about a third of MZyap1kg¹⁵¹ mutants showed kyphoscoliosis, suggesting that the maternal contribution does not affect the penetrance of the zygotic phenotype.

Yap1 has been extensively implicated in mechanosignalling [50,121,122]. A correlation between mechanical influences and kyphoscoliosis has long been proposed [34,35,37–39,123]. Once an initial defect in the spine occurs, generally from an unknown cause, worsening of the defect often occurs during adolescent growth spurts. In the 1880s, this observation engendered a hypothesis, the Hueter-Volkmann ‘Law’, which stated that vertebral growth is retarded by mechanical compression of the growth plate and stimulated by reduction of compression [34,39] thereby explaining the exacerbation of scoliosis during growth under asymmetric gravitational load. Studies on rat tails show that mechanical force can modulate or correct vertebral growth [35]. The kyphoscoliosis observed in the yap1kg¹⁵¹ mutants is first observed from 13 days and gradually worsens in some individuals, during zebrafish adolescence [124]. However, this process cannot be attributed to gravitational positive feedback effects because larvae are neutrally buoyant. Perhaps, in the absence of yap1, mechanical force is not transmitted or sensed correctly along the spine, causing a curve that, once present, progresses to kyphoscoliosis in about a third of the mutant fish.

About a third of yap1kg¹⁵¹ mutants also die during rearing. We note, however, that the frequency of kyphoscoliotic mutants did not obviously decrease with age, suggesting either that mutants die for reasons unrelated to kyphoscolioisis or that, as kyphoscoliotic mutants perish, additional mutants develop kyphoscoliosis during their adolescence.

As with humans, the heterogeneous genetic background in zebrafish yap1 mutants may explain the variable penetrance and time of onset of kyphoscoliosis. Alternatively, subtle epigenetic or environmental variables may affect the expression of the yap1 mutant phenotype. Such effects may account for the variable survival rate between different lays.

Poor larval survival without Wwtr1

Zebrafish wwtr1kg¹⁶⁹ mutants, like those previously reported [25,86,118,125–127] do not recapitulate the failure to undergo skeletal calcification reported in wwtr1 morphants [73]. This is in contrast to murine manipulations in which Wwtr1 mutant mice are born significantly smaller than siblings with minor defects in ossification [54].

Despite the viability and unaltered size of zebrafish wwtr1kg¹⁶⁹ mutants, they survive less well than their siblings, even in the protected environment of an aquarium. The wwtr1^bns35 mutant allele has been shown to cause defects in cardiac trabeculation [126], which might enhance larval death. Mice lacking Wwtr1 die between birth and weaning, possibly from kidney defects [54,70]. As we confirm that yap1;wwtr1 double mutants show lethally severe early defects, it seems that balance of functional requirement for Yap1 and/or Wwtr1 in individual tissues has varied during vertebrate evolution. We also note that yap1kg^151/+ heterozygosity may prevent survival of wwtr1kg¹⁶⁹ mutants. Clearly, some Wwtr1 function(s) may be essential for effective survival in the wild.

Cell autonomy of Yap1 action in kyphoscoliosis unresolved

As we have analysed putative null yap1 and wwtr1 mutations, the cellular origin of the observed defects remains unknown. All components of the neuromusculoskeletal system, nerve, muscle, tendon, cartilage, bone and associated vessels, or a combination of them, are candidates for the site(s) of action of Yap1. Despite their early poor motility, the earliest putative precursor to kyphoscoliosis that we could discern in yap1kg¹⁵¹ mutants was at 11–13 days in spinal bone itself, where a delay in calcification was accompanied by mild spinal curvature in some individuals. Vertebrae arise from cells of the notochord sheath, somitic sclerotome and neural crest, but no defects in these tissues were observed at earlier stages. An osteogenesis imperfecta (OI) bone phenotype correlating with the extent of reduction of Yap1 and Wwtr1 in bone precursors has been reported in murine conditional mutants [55]. Murine cartilage-specific deletion of Yap1 and Wwtr1 has also shown subtle morphological bone defects [58]. Moreover, Yap1 and Wwtr1 have opposing effects at different stages of limb osteoblast differentiation [59]. Thus, vertebral precursor cells remain prime candidates for a cell autonomous origin of the defect.

The striking segmental interruptions of the hypochord observed in 36 hpf yap1 mutants suggest a possible vascular or lymphatic origin for kyphoscoliosis. We note, however, that the hypochord defect did not cause any obvious vessel defects and was fully penetrant at the non-permissive temperature. Further studies are required to determine whether and how similar defects in a subset of embryos reared at the permissive temperature might contribute to either the motility defect or the later vertebral anomalies and kyphoscoliosis.

Many zebrafish scoliotic phenotypes arise from defects in early notochord vacuolation, elongation and extracellular matrix deposition that lead to subsequent defective of ossification [98,128]. Yap1 mutants lack the poor notochord vacuolation and body length reduction at embryonic stages and calcein vertebral centrum defects at 15–20 dpf observed in dstyk and MZptk7a mutants and vangl2 knockdowns. Thus, the yap1 vertebral column phenotype arises at the mid larval stage after apparently normal early notochord formation.

The early motility defect is also a possible cause of kyphoscoliosis. Defects arising in the central nervous system are poor candidate kyphoscoliosis triggers because yap1 mutants showed early motility defects in response to direct electrical stimulation, which is thought to act peripherally [129]. So altered muscle-derived force was a prime trigger candidate, particularly because severe craniofacial cartilage and bone defects can arise from early muscle defects caused by mutation of genes acting cell autonomously in muscle [11]. Given the strong and sequential expression of yap1 and wwtr1 during myotomal myogenesis and evidence that Yap1 can affect murine myogenesis [57,66,130,131], we hypothesised that early muscle defects in yap1 mutants might cause force asymmetry leading to kyphoscoliosis. However, an extensive analysis of early muscle failed to reveal a morphological or histological defect in mutants, despite their poor motility. Moreover, yap1kg¹⁵¹ mutants show normal early muscle growth even when shifted to the non-permissive temperature during the early growth phase. Yap1 has also been shown to affect tendons [132–134], but given that tendons are small in 5 dpf zebrafish, we have not analysed this tissue.

Recessive variants in MYH3, the gene encoding the skeletal muscle-specific embryonic myosin heavy chain, cause spondylocarpotarsal synostosis (SCTS), characterised by vertebral fusions and scoliosis [135]. Ablation of the murine Myh3 gene similarly causes severe muscle defects with altered muscle Yap1 signalling and leads to scoliosis in surviving mice [5]. Our finding that loss of yap1 function can cause kyphoscoliosis strengthens the evidence for the involvement of Yap1 signalling in the aetiology of MYH3-associated SCTS and possibly other scolioses. Bharadwaj et al. further suggested that Myh3 ablation caused muscle defects and scoliosis through Yap1 activation because both phenotypes were mitigated by CA3, a drug that can block Yap1 action [5]. These observations raise the hypothesis that inhibition of Yap1 function cell autonomously within muscle mitigates scoliosis by its clear beneficial effects on muscle development. However, yap1 mutant fish do not show a severe muscle defect, yet have scoliosis. Although zebrafish do not contain an Myh3 orthologue, we observed no alteration in expression of regionally-expressed fmyhc1.2 and fmyhc2.1 fast myosin genes that might have contributed to kyphoscoliosis. Moreover, in fish, it is loss of yap1 function rather than gain, that promotes scoliosis. Thus, defining the cellular site and mechanism of Yap1 action required to prevent kyphoscoliosis is a priority.

Familial Kyphoscoliotic Ehlers-Danlos Syndrome arises from rare mutations in genes implicated in collagen matrix formation (PLOD1 and FKBP14; [136]). GWAS studies of adolescent idiopathic scoliosis (AIS) have yielded hits near genes involved in neuromuscular (EPHA4/PAX3, LBX1, SOX6 and GPR126) and skeletal (PAX1, SOX9, TBX1, MEOX2) development and in matrix biology (BNC2, FBN1), but to date account for a small fraction of heritability [3,137]. A YAP1-BNC2 interaction is implicated in matrix remodelling in response to mechanosignals [138]. We speculate that Yap1 function mediates mechanosignalling to regulate cell-matrix interactions in one or more tissues of the developing neuromusculoskeletal system and may contribute to some forms of AIS.

Lastly, a number of mutations affecting sensorimotor function have yielded kyphoscoliotic phenotypes in zebrafish, including some affecting cilia and the Reissner’s fibre [95,97]. While the cell types requiring this ciliary function are unclear, it has been suggested that defects in fluid circulation in the neural ventricles can underlie both Reissner’s fibre defects and scoliosis [94,96]. Although Yap1 has not been implicated in ciliary function, we investigated this possibility by analysing cilia in the ventricle and the Reissner’s fibre but failed to observe any morphological defect. Whether other neural system defects deriving from loss of Yap1 function, but acting through altered muscle or hypochord functionality, underlie kyphoscoliosis remains to be determined. While the mechanism leading to kyphoscoliosis in some individual yap1 mutants remains unclear, we hypothesise that Yap1 may act to suppress positive mechanofeedback leading to asymmetric bone growth.

Ethics statement

All experiments were performed in accordance with licences held under the UK Animals (Scientific Procedures) Act 1986 and later modifications after approval by King’s College London Animal Welfare and Ethical Review Board and the UK Home Office and conforming to all relevant guidelines and regulations.

Zebrafish lines and maintenance

Genetically-altered Danio rerio myf5^hu2022 [82], myf6 kg¹²⁶ (new, 3 bp deletion creating STOP (UAG) at position aa34: predicted null), myod1^fh261 [11,139], myog^kg125 [80], wwtr1kg¹³³ (new, TALEN: 1 bp deletion T83, creating 4 aa nonsense tail then STOP; predicted null), wwtr1kg¹⁶⁹ (new, TALEN: 8 bp deletion 82–89 CTTTTTAA, 105 aa nonsense tail; predicted null), yap1kg¹³⁷ (new, TALEN: 19 bp deletion 106–124 AAAAACACCATCGTCCCCC, 4 aa nonsense tail; predicted null), yap1kg¹⁵¹ (TALEN: 5 bp deletion 111–115 CACCA, 84 aa nonsense tail; predicted null), yap1kg¹⁵² (TALEN: 11 bp deletion 110–120 ACACCATCGTC, 82 aa nonsense tail; predicted null), Tg(Ola.Actb:Hsa.HRAS-EGFP)^vu119 (ßActin:membEGFP; [140]), Tg(actc1b:mCherryCAAX)^pc22 and Tg(actc1b:LIFEACT-EGFP)^pc21 [141] were kept on AB background, reared at King’s College London on a 14/10h light/dark cycle [142]. Adults were kept at 26.5°C and embryos/larvae were kept at 28.5°C (or 20.5°C or 32.0°C where stated) in the dark until 5 dpf-equivalent stage. To avoid developmental problems, including increased death rates [143], caused by rearing fish at non-standard temperatures during early embryonic stages, all fish were reared at 28.5ºC until 70% epiboly (~8 hpf) and then shifted to alternate temperatures as described. For clarity, the age of fish reared at non-standard temperatures are described by their absolute age in days, rather than by standard dpf stages at 28.5ºC (S1 Table). Pilot experiments showed equivalent development of length, pigmentation and yolk consumption at 5 dpf @ 28.5ºC and 7 days @ 20.5ºC. Fish raised past 5 dpf-equivalent developmental stage (7 days if raised at 20.5°C) in the incubator were kept separately in 12 or 24 well plates on a 12/12h light/dark cycle fed with fresh rotifer food daily.

Genotyping

Adult fish fin-clips or whole anaesthetised embryos/larva were dissolved in 50 μl 25 mM NaOH, 0.2 mM EDTA for 1 hour at 95°C and neutralized with 50 μl of 55 mM Tris HCl pH 8.0. Myf5^hu202 was genotyped by PCR using primers CATTGTCTCCAATGGGCCTGCAAAGCTCG and GGATCTCTACCTTGGGGAGGCGTTG annealed at 60°C, followed by restriction digest with TaqI (NEB, R0149S) at 65°C for 4 hours to over-night, yielding uncut wt band (175 bp) and cut mutant bands (145 and 30 bp). Myf6 kg¹²⁶ was genotyped by PCR using primers GGGCACCAGAAGGCCTATTG and GGTTGTATGTGTAAGGGTCAGT GTC annealed at 59°C, digested with RsaI (NEB, R0167) at 37°C for 4 hours to over-night yielding cut wild-type bands (250 and 396 bp), uncut mutant band (646 bp). Myod^fh261 was genotyped by PCR using primers GGACCCCAGGCTTGTTC and GTTGGATCTCGGACTGGA annealed at 56°C, digested with BsaXI (NEB, R0609) at 37°C for 4 hours to over-night, yielding uncut wild-type band (397 bp) and cut mutant bands (260, 119 and 18 bp). Myog^kg125 was genotyped by PCR using primers TCAGAAACACCCACAAACGCTCAC and GCAGGCCCAGGGGAGACACT annealed at 54°C, digested with EcoRV-HF (NEB, R3195) at 37°C for 4 hours to over-night, yielding cut wild-type bands (197 and 164 bp) and uncut mutant band (361 bp). Wwtr1kg¹³³ and wwtr1kg¹⁶⁹ were genotyped by PCR using primers CGGCCATTTTAATCGAAGTTTGTT and CTGTAGGGACGCCGGAGATGAGC annealed at 57°C followed by sequencing (Genewiz) with the latter primer. Yap1kg¹³⁷, yap1kg¹⁵¹ and yap1kg¹⁵² were genotyped by PCR using primers TCTTTTTGGGTTGTTTTGGATTA and GGCTCTGGCGGCGTGAA annealed at 57°C, followed by sequencing with the former primer. An alternative High Resolution Melting (HRM) method was also used to genotype yap1kg¹⁵¹ using the primers ACCGATCTGGAGGCTCTTTT and GTCTGGCAGCTTTCTCAACC and wwtr1kg¹⁶⁹ using primers GTGATCCATGTCGCCAAAGACT and GCGGCATATCCTTGTTC. Both HRM methods used the standard 60°C annealing temperature with Melt Dr (Thermo Fisher Scientific, 4415450) reagent in an Applied Biosystems ViiA 7 with the HRM set-up in duplicate 10 μl reaction volumes.

Generation and characterisation of yap1 and wwtr1 mutants

Yap1 and Wwtr1 mutants were generated by TALEN (Transcription activator-like effector nuclease) genome editing. Two RNA guide TALEN arms were designed to target within the first exon of each gene, 5′-TAACGCTGTGATGAACCC-3’ and 5’-CCCCTTCCGTGCCGATG-3’ (exon 1 of yap1) and 5′-TGGACACGGATCTGGAGG-3’ and 5’-CATGAACCCGAAACCGA-3’ (exon 1 of wwtr1) using Zifit software [144, 145]. TALEN arms were constructed using cloning by the REAL Assembly Plasmid Method [145]. Capped TALEN mRNAs were injected into single cell embryos to induce FokI cuts at the target site. Alleles were selected at F1 that carried mutations causing premature stop codons after a nonsense tail. All the alleles described for each gene have a similar phenotype, but yap1kg¹⁵¹ and wwtr1kg¹⁶⁹ were eventually favoured due to their high levels of nonsense mediated decay. Repeated outcrossing to ‘wildtype’ AB did not reveal changes in phenotype to F4.

Quantitative reverse transcriptase-polymerase chain reaction (qRT-PCR) was performed on RNA extracted from four sibling individual 7–8 dpf-equivalent larval trunk/tails of each genotype (heads were used for genotyping) or from dissected adult muscle of the indicated number of genotyped sibling individuals. Larvae were sonicated on ice in 100 μL TRI Reagent (Sigma T9424) and left at room temperature for 5 minutes before 1-Bromo-3-chloropropane (50 μL, Sigma B9673 was added, incubated for 10 minutes at room temperature, microfuged at 4°C for 10 minutes and RNA purified from the aqueous upper phase by phenol/chloroform extraction. Adult muscle (~50 mg) was processed in 500 μL of TRI Reagent using a Tissue Ruptor (Qiagen 9002755) and purified using RNeasy Mini kit (Qiagen 74106). The cDNA synthesis and qPCR were performed as described [146], using primers listed in S2A Table.

In situ mRNA hybridization (ISH), immunodetection, histology and imaging

ISH was performed as described [147]. Digoxigenin-labelled probe for yap1 was made by PCR from 24 hpf cDNA with the primers TAATACGACTCACTATAGGGAGAACGCCGCCAGAGCCAAAGTCC and GGATCCATTAACCCTCACTAAAGGGAAGTGCCGCCATCCTGCTCCAT. Probes for wwtr1 were made from Image Clone 9037619 (Sourcebioscience). Primers used to make probes for col8a1a, col9a1b, fmyhc1.2 and fmyhc2.1 are listed in S2B Table. Embryos were immersed in glycerol and imaged on a Leica MZ16F with LED light attachment, Olympus DP70 camera and DP Controller software.

Immunodetection was performed as previously described [148] with antibodies Rabbit anti-GFP (1:500, Torrey Pines, ABIN110592), A4.1025/sarcomeric myosin heavy chain (MyHC; 1:10, Abcam, ab37484), F59/slow MyHC (1:5, Santa Cruz Biotechnology, sc-32732), F310/fast myosin light chain (1:5 DSHB), Mouse anti-alpha-actinin (1:500 Sigma A7811), GαMIgG(H + L)Alexa Fluor488 (1:1000 Invitrogen A-11001), GαMIgA-FITC (1:1000 Serotec 5104-3104F), GαMIgG1AlexaFluor555 (1:1000 Invitrogen A-21127), GantiR IgG (H + L) Alexa Fluor488 (1:1000 Invitrogen A-11008) and Hoechst 33342 (1:2000 New England Biolabs 4082S). Larvae older than 2 dpf were lightly fixed with 2% paraformaldehyde in PBS for 30 minutes at room temperature. Embryos/larvae were mounted in low melting point agarose and imaged on a Zeiss LSM Exciter confocal microscope using Zeiss 20 × /1.0 NA dipping objective, analysed using Fiji (NIH, ImageJ2 2.9.0/1.53t) or ZEN 2009 (Zeiss) software. Myotome volume and fibre number measurements followed [84]. NADH tetrazolium reductase staining was performed as described [80].

Immunodetection of the Reissner’s fibre was performed as described in [97] on bleached larvae at 5 dpf-equivalent using 3 day incubation for the AFRU rabbit primary antibody (a generous gift of Dr. M. Monserrat Guerra, Universidad Austral de Chile) and 3 days for the secondary antibody. Some larvae were transected to improve access. Individuals were countrerstained with Hoechst 33342 in the secondary step.

Skeletal assays

Alizarin Red and Alcian Blue staining was performed as previously described [149]. Measurements of spine and caudal vertebrae were performed using Image J version 1.53. Measurements on caudal vertebrae C15-C20 were averaged for each fish. Calcein staining of live fish was as described [150]. Micro Computerized Tomography (CT) scans of 4.5 month frozen fish supported by gauze were made in the Centre for Craniofacial Regenerative Biology (KCL) using a Scanco MicroCT 50 (300–350 ms exposure, 70 kV, 114 μA, 10 μm resolution).

Scanning electron microscopy

Skulls were opened with forceps, the brain removed and bisected with a sapphire knife, fixed in 2% (v/v) glutaraldehyde in 0.1 M sodium cacodylate buffer (pH 7.3) overnight at 4°C, and processed by the KCL Centre for Ultrastructural Imaging.

Movement assays

The larval tail movement assay was carried out as previously described [146], and tail angle of movement was measured in Fiji (NIH) software. Velocity swim assay conducted as previously described [151], recorded at early larval stages on Leica MZ16F with LED light attachment, Olympus DP70 camera and DP Controller software or with an iPhone 13. Fish velocity was measured using Tracker (http://physlets.org/tracker) software, by averaging the second and third time intervals at the early (4–5 dpf) larval stage, but the second to fourth time intervals at the later (11 dpf) larval stage because motility persisted longer than in younger larvae. The first interval was omitted as it was unknown when within it the fish began to move.

Statistics

Unless otherwise stated, parametric and non-parametric tests appropriate to the data were performed using GraphPad Prism version 10.6.1. Χ² tests were performed in Excel for Mac 16.101.3. Raw count data underlying the current work is available in S1 Data.