Resolving primary pathomechanisms driving idiopathic-like spinal curvature using a new katnb1 scoliosis model

Summary Idiopathic scoliosis (IS) refers to abnormal spinal curvatures that occur in the absence of vertebral or neuromuscular defects. IS accounts for 80% of human spinal deformity, afflicts ∼3% of children worldwide, yet pathogenic mechanisms are poorly understood. A key role for cerebrospinal fluid (CSF) homeostasis in zebrafish spine development has been identified. Specifically, defects in cilia motility of brain ependymal cells (EC), CSF flow, and/or Reissner fiber (RF) assembly are observed to induce neuroinflammation, oxidative stress, abnormal CSF-contacting neuron activity, and urotensin peptide expression, all associating with scoliosis. However, the functional relevance of these observations to IS remains unclear. Here we characterize zebrafish katnb1 mutants as a new IS model. We define essential roles for Katnb1 in motile ciliated lineages, uncouple EC cilia and RF formation defects from spinal curvature, and identify abnormal CSF flow and cell stress responses as shared pathogenic signatures associated with scoliosis across diverse zebrafish models.


INTRODUCTION
Abnormal curvatures of the spine, known as scoliosis, have been pervasive throughout human history (Naderi et al., 2007;Vasiliadis et al., 2009). The most prevalent type, idiopathic scoliosis (IS), manifests as a rotational spinal curvature that develops in the absence of gross vertebral malformations or other physiological defects, often during periods of intense growth linked to adolescence (Horne et al., 2014;Cheng et al., 2015). IS afflicts 3%-4% of school-aged children, yet our understanding of pathogenic mechanisms underlying IS remains clouded in large part by its genetic heterogeneity (Kesling and Reinker, 1997;Wise et al., 2008;Cheng et al., 2015;Grauers et al., 2016;Peng et al., 2020). Recent whole exome sequencing (WES) studies and multi-ethnic GWAS meta-analyses have associated variants in musculoskeletal collagen and cartilaginous extracellular matrix genes with IS susceptibility for a subset of cases (Haller et al., 2016;Khanshour et al., 2018). However, it is estimated that >95% of the total genetic variance underlying IS remains to be discovered (Kou et al., 2019). This fact underscores a critical need to better understand IS pathobiology.
Surprisingly, spinal curvatures rank among the most common deformities observed in teleost fish, and utilization of zebrafish genetic IS models has spearheaded efforts across multiple research groups to elucidate the biological underpinnings of spinal curve progression (Hayes et al., 2013(Hayes et al., , 2014Grimes et al., 2016;Van Gennip, Boswell and Ciruna, 2018;Zhang et al., 2018;Rose et al., 2020;Terhune et al., 2020;Troutwine et al., 2020;Gray et al., 2021;Wang et al., 2022). Pioneering work with zebrafish protein tyrosine kinase 7a (ptk7a) mutants characterized the first genetically defined developmental model of IS and identified a key role for cerebrospinal fluid (CSF) homeostasis in spine development, linking motile cilia-driven CSF flow defects with idiopathic-like spinal curvatures (Hayes et al., 2013(Hayes et al., , 2014Grimes et al., 2016;Van Gennip, Boswell and Ciruna, 2018). Notably, there is some evidence that cilia and CSF flow defects may also be linked to human IS (Patten et al., 2015;Oliazadeh et al., 2017;Baschal et al., 2018;Terhune et al., 2020;Wang et al., 2020;Algin et al., 2022). iScience Article associated with the disrupted formation of Reissner fiber (RF, a proteinaceous filament that threads through ventricles of the brain and spinal cord) were linked to sspo IS phenotypes. However, as defects in bulk CSF flow and RF aggregation are shared across sspo, ptk7a, and other zebrafish IS models Troutwine et al., 2020), the primary cause of scoliosis remains unclear. Downstream of CSF homeostasis and RF deficits, abnormal CSF-contacting neuron (CSF-CN) activity, urotensin neuropeptide (Urp) expression, neuroinflammatory signals, and oxidative stress have all been associated with axial curvature (Bö hm et al., 2016;Cantaut-Belarif et al., 2018, 2020Sternberg et al., 2018;Van Gennip, Boswell and Ciruna, 2018;Zhang et al., 2018;Lu et al., 2020;Orts-Del'Immagine et al., 2020;Rose et al., 2020). The interdependence and functional relation of these phenotypes to idiopathic-like scoliosis remains to be determined.
To gain further insights into the etiopathogeneses of spinal curvature, we characterize zygotic katanin p80 subunit mutants (katnb1 mh102/mh102 )  as a new zebrafish IS model. We define essential roles for Katnb1 in motile cilia and choroid plexus (ChP) development. Through this analysis, katnb1 mutants functionally uncouple brain ependymal cell (EC) cilia, urp expression, and RF formation defects from spinal curvature. Rather, our studies identify physiological stress responses downstream of CSF homeostasis defects as biological mechanisms associated with idiopathic-like scoliosis that are shared across diverse zebrafish IS models.

katnb1 mutants develop late-onset spinal curvatures that model human IS
Katanin is a microtubule severing heterodimer comprised of catalytic p60 and scaffolding p80 subunits, with the p80 subunit (KATNB1) functioning to localize Katanin to the centrosome (McNally et al., 1996;Hartman et al., 1998). In single-celled organisms and in vertebrates, Katanin has been found to play essential roles in flagella motility and ciliogenesis (Dymek et al., 2004;Sharma et al., 2007;Mirvis et al., 2019). Based on human KATNB1 mutations linked to microlissencephaly, studies utilizing cell culture and mouse models also revealed the functional importance of KATNB1 in centriole biogenesis, Shh signaling, and neural progenitor development Mishra-Gorur et al., 2014), whereas analysis of maternal-zygotic (MZ) katnb1 mutant zebrafish demonstrated a critical role for Katnb1 in early embryonic patterning and viability .
In contrast to strong MZ katnb1 mutant phenotypes, zygotic katnb1 mh102/mh102 mutant zebrafish display no overt morphological abnormalities at embryonic stages, owing to maternally contributed RNAs . However, at late larval and early juvenile stages (7-8 mm standard length; approximately 3 weeks post fertilization [wpf]), we observed that zygotic katnb1 mh102/mh102 mutants develop obvious axial curvatures ( Figure 1). In order to characterize vertebrae formation, we used the vital dye calcein, which binds to calcified skeletal structures (Du et al., 2001), and observed no obvious vertebral segmentation defects prior to or post-onset of curve formation ( Figures 1A-1D). Axial curvatures emerge only during late juvenile stages of development and could first be detected in some katnb1 mh102/mh102 mutants at 17 dpf (days postfertilization; $7 mm length; Figures 1D and 1D 0 ). The majority of mutants develop curves between 21 and 23 dpf, and scoliosis is fully penetrant by 30 dpf.
To better visualize spine deformities, we performed microcomputed tomography (mCT) of 3-month-old (3mo) adults ( Figures 1E-1H) and quantified the position, magnitude, and orientation of spinal curvatures iScience Article ( Figure S1). Within the dorsal-ventral (DV) or sagittal plane, the position and orientation of curves are well conserved along the rostral-caudal axis of katnb1 mh102/mh102 mutants ( Figures 1G and 1I), beginning with a stereotypic kyphotic curvature correlating with the positioning of the swim bladder (Parichy et al., 2009) followed by a compensatory lordotic curve in the opposite direction. The most caudal spinal curvatures in katnb1 mh102/mh102 mutants are more variable in nature, perhaps free from biomechanical constraints of underlying organ systems.
Spinal curvatures in the medio-lateral (ML) or coronal plane are less severe than sagittal curves and vary in both their position and left versus right chirality along the rostral-caudal axis ( Figures 1H and 1K). Although no significant difference in scoliosis severity was observed between male and female katnb1 mh102/mh102 mutants, cumulative spinal curve measurements trended higher in females ( Figures 1J and 1L). Overall, katnb1 mh102/mh102 mutant phenotypes recapitulate many hallmarks of human IS, including an adolescent age of onset, scoliosis progression in the absence of obvious congenital vertebral malformations, and characteristic patterns of spinal curvature. Although the shape and positioning of idiopathic-like curves differ between human IS patients and zebrafish katnb1 mutant models, curve patterns likely reflect biomechanical forces imposed by species-specific organ anatomy (Schlö sser et al., 2017).
katnb1 scoliosis phenotypes are associated with motile cilia and CSF flow defects Bulk CSF flow defects have previously been associated with scoliosis in both ptk7a and sspo zebrafish IS models (Grimes et al., 2016;Van Gennip, Boswell and Ciruna, 2018;Rose et al., 2020). To determine whether similar mechanisms underly spine curvature in katnb1 mh102/mh102 mutants, we injected fluorescent dye into the brain ventricles of mutant and sibling control fish and monitored CSF flow into the spinal cord. At 2 wpf, prior to scoliosis onset, katnb1 mh102/mh102 mutants and sibling controls exhibit similar, slow rates of CSF flow ( Figure 2I). However, at 3 wpf, corresponding with onset of spinal curvature, katnb1 mh102/mh102 mutants exhibit a significant reduction in CSF flow as evidenced by restricted distribution of dye along the  iScience Article spinal cord 2 h post-injection ( Figure 2J). These results indicate that bulk CSF flow defects are conserved across ptk7a, sspo, and katnb1 zebrafish IS models.
We next investigated whether katnb1 mh102/mh102 mutants also display defects in EC mono-or multi-cilia formation, which are believed to regulate CSF flow and have been linked with scoliosis in ptk7a hsc9/hsc9 and other zebrafish scoliosis models (Grimes et al., 2016;Konjikusic et al., 2018;Olstad et al., 2019;D'Gama et al., 2021;Wang et al., 2022). Scanning electron microscopy (SEM) of brain rhombencephalic ventricles from 3mo control animals revealed a dense, polarized bed of EC multi-cilia ( Figure 2A). In contrast, 3mo katnb1 mh102/mh102 mutants exhibited a complete lack of multi-cilia as well as most mono-cilia expected to line brain ventricles ( Figure 2B). katnb1 mh102/mh102 mutants presented with a more severe cilia loss than that described for ptk7a hsc9/hsc9 mutants (Grimes et al., 2016) but did not exhibit hydrocephalus, which often accompanies EC multi-cilia defects and abnormal CSF flow (Abdelhamed et al., 2018;Figures S2A-S2C). Notably, observed cilia defects were tissue and stage dependent. katnb1 mh102/mh102 mutant embryos exhibited normal left-right patterning (a trait commonly disrupted in motile cilia mutants [Essner et al., 2005]); normal differentiation of pronephric motile cilia; as well as normal formation, orientation, and motility of spinal cord floor plate cilia ( Figure S3).

Defects in EC cilia are not specifically linked to spinal curvature
Despite obvious parallels in pathogenic mechanisms associated with scoliosis in katnb1 and ptk7a IS models (CSF flow defects associated with defects in foxj1a-positive motile cilia lineages), katnb1 mh102/mh102 mutants demonstrate milder scoliosis phenotypes. Quantitative analysis of Cobb angles (a measure of curve severity) revealed that, compared with ptk7a hsc9/hsc9 fish, katnb1 mh102/mh102 mutants exhibit a significant reduction in curve severity, specifically in the ML plane (p = 0.0014; Figures 1MÀ1P). Furthermore, although curve formation in ptk7a hsc9/hsc9 mutants begins around 14 dpf (Hayes et al., 2014;Rose et al., 2020), katnb1 mh102/mh102 mutants developed scoliosis later in development, on average by 21 dpf. Thus, katnb1 mh102/mh102 animals demonstrate later onset of apparent curvature and less severe spinal curvatures than ptk7a hsc9/hsc9 mutants, despite presenting more severe EC cilia defects at 3 months of age (Grimes et al., 2016;Figures 2A and 2B).
To determine whether phenotypic variances were caused by differences in the timing of EC cilia loss, we performed SEM analysis of rhombencephalic ventricles from katnb1 mh102/mh102 , ptk7a hsc9/hsc9 , and heterozygous control siblings from earlier stages of development. At 6wpf, control fish exhibited well-developed EC multi-cilia throughout the ventricle, although not as dense as 3mo stages ( Figure 2C). In contrast, katnb1 mh102/mh102 mutants exhibited a clear loss of EC multi-cilia formation at this stage ( Figure 2D). Strikingly, ptk7a hsc9/hsc9 mutants exhibited sparse multi-cilia bundles lining the ventricle that appeared similar to control animals ( Figure 2E). These results indicate that milder scoliosis phenotypes in katnb1 versus ptk7a mutant zebrafish cannot be explained by later loss of EC cilia in katnb1 mh102/mh102 mutants.  iScience Article scoliosis, we performed whole mount immunohistochemistry on 21-and 30-day old katnb1 mh102/mh102 mutant versus control brains and screened for alterations in polyglutamylated tubulin (PolyETub) staining, which is a reliable marker for cilia on foxj1-expressing cells in zebrafish (Bré et al., 1994;D'Gama et al., 2021).
Notable differences were observed in both the forebrain choroid plexus (fChP) and rhombencephalic choroid plexus (rChP), which are highly ciliated structures found on the dorsal surface of the third and fourth ventricle (in zebrafish) and part of the circumventricular organ (CVO) network that is critical for generating and secreting components of the CSF (Henson et al., 2014;Guerra et al., 2015). In 21 dpf control fish, both ChPs appeared highly ciliated with a prevalence of mono-ciliated cells ( Figures 4A-4C, 4G and 4I). However, 21 dpf katnb1 mh102/mh102 mutants exhibited lower numbers of ChP mono-cilia ( Figures 4D-4F, 4H, and 4J).
By 30 dpf, both fChP and rChP have developed bundles of multi-cilia in control animals ( Figures 5A and 5B), especially on the anterior edge of the fChP and on the dorsal-medial telencephalon, corresponding to the tela choroidea (Nieuwenhuys, 2011;Folgueira et al., 2012;D'Gama et al., 2021). In striking contrast, 30dpf katnb1 mh102/mh102 ChPs do not form multi-cilia but instead display circular aggregations of PolyETub staining within anterior fChP and dorsal-medial telencephalic regions ( Figures 5C and 5D). In addition, there were significantly fewer and shorter mono-cilia in the rChP of 30dpf katnb1 mh102/mh102 mutants compared with controls (Figure S4). Taken together, these observations define a critical period in the ciliation and differentiation of normal zebrafish ChPs, identify defects in foxj1-positive ciliated cells that arise just as katnb1 mh102/mh102 mutants develop   Although observed ChP cilia defects correlate with scoliosis in katnb1 mh102/mh102 mutants, they may not be causative. Indeed, scoliosis in sspo mutant zebrafish develops in the absence of obvious cilia defects and has instead been linked to defects in RF formation. Furthermore, RF formation defects have also been observed in ptk7a and other zebrafish IS models associated with irregularities in motile-cilia-driven CSF flow Troutwine et al., 2020), raising the possibility that loss of RF formation may ultimately drive scoliosis in these mutants.
Therefore, CSF flow defects and scoliosis in katnb1 mh102/mh102 mutants are not associated with loss of RF formation, albeit may involve decreased functionality or aberrant aggregation of Reissner's substance. These observations challenge current theories regarding the etiopathogenesis of scoliosis (see Bagnat and Gray, 2020;Ringers and Jurisch-Yaksi, 2020), demonstrating that the presence of RF is not sufficient to keep the spine straight. Rather, the specific and ectopic accumulation of Sspo below the forebrain and rhombencephalic ChPs suggests that the primary defect in katnb1 IS models relates to abnormal ChP development and/or function with correlative Sspo accumulation subsequent to onset of curvature. To investigate the molecular consequences of ChP defects, ectopic Sspo accumulation, and disrupted CSF flow on zebrafish brain and spine development, we performed bulk RNA-seq and differential gene expression analysis of dissected 30-dpf brains and 21-dpf trunk/tail segments isolated from scoliotic katnb1 mh102/mh102 mutants and their control katnb1 mh102/+ siblings. Of note, there was no evidence of dysregulated urp1 or urp2 urotensin neuropeptide gene expression in katnb1 mh102/mh102 mutants (Tables 1, S1, and S2). This differs from other studies that suggest reduced Urp expression, downstream of abnormal RF-mediated CSF-cN activity, may ultimately drive axial curvature in zebrafish IS models (Bö hm et al., Sternberg et al., 2018;Zhang et al., 2018;Cantaut-Belarif et al., 2020;Lu et al., 2020;Orts-Del'Immagine et al., 2020). Notably, transcriptomic analysis of sspo dmh4/+ mutant brains, which exhibit Sspo secretion and RF formation defects, demonstrated a significant reduction in urp2 expression (Rose et al., 2020; Table 1). Conversely, bulk RNA-seq analysis of ptk7a hsc9/hsc9 mutant brains, which exhibit ectopic accumulation of RF substance , demonstrate a significant increase in urp2 expression (Table 1). Therefore, differential gene expression analysis of dissected brain tissue, which includes the hindbrain/spinal cord junction-an area demonstrated to contain urp positive CSF-cNs (Quan et al., 2015) and an intact RF , can speak to activity of the RF-urp signaling axis. The absence of dysregulated urp expression in katnb1 mh102/mh102 mutants further suggests that RF assembly/function is normal in this model and provides intriguing evidence that an intact RF-Urp signaling axis is not sufficient to keep the zebrafish spine straight.
As an alternative explanation to the etiopathogenesis of scoliosis, investigations of ptk7a and sspo IS models implicate neuroinflammatory and oxidative stress responses that arise downstream of CSF iScience Article homeostasis defects as necessary and sufficient to drive spinal curvature (Van Gennip, Boswell and Ciruna, 2018;Rose et al., 2020). Although 30-dpf katnb1 mh102/mh102 mutant brains do not display an obvious inflammatory response at the transcriptional level, there is strong evidence for cell stress: Metascape enriched ontology cluster analyses revealed significant downregulation of pathways involved in oxidative phosphorylation and response to oxidative stress ( Figure 8A), and although upregulated pathways were predominantly associated with chromatin remodeling, significant dysregulation of unfolded protein response and cellular response to stress signatures were observed ( Figure 8B). Within the trunks and tails of 21-dpf katnb1 mh102/mh102 mutants, differential gene expression analyses also identified significant cell stress indicators including upregulated autophagy, downregulated RNA processing, and lipid metabolic processes, in parallel with dysregulated cilia and muscle function signatures ( Figures S5A  and S5B).   iScience Article DISCUSSION We find that katnb1 mh102/mh102 mutant zebrafish exhibit defining attributes of human IS, including juvenile age of onset, scoliosis that progresses in the absence of congenital vertebral malformations, and characteristic patterns of spinal curvature. We demonstrate that Katnb1 function in foxj1a-positive motile-ciliated lineages is necessary for normal spine development and when deficient, leads to scoliosis. Further, we show that scoliosis onset in katnb1 mh102/mh102 mutants associates with bulk CSF flow defects-a phenotype conserved with ptk7a and sspo zebrafish IS models (Van Gennip, Boswell and Ciruna, 2018;Rose et al., 2020). Investigation into the timing of EC cilia defects in both katnb1 and ptk7a IS models uncouples EC cilia function from observed phenotypes. Rather, ciliogenesis defects in the developing forebrain, and rhombencephalic ChPs of katnb1 mh102/mh102 mutants were found to correlate temporally with CSF flow defects and scoliosis onset, suggesting that the primary deficit in katnb1 IS models may relate to abnormal ChP development and/or function-a model supported by ectopic aggregations of Sspo protein localized specifically below fChP and rChP in katnb1 mh102/mh102 mutants. Article Surprisingly, motile-cilia and CSF flow defects do not disrupt RF formation or urp gene expression in katnb1 mh102/mh102 mutants, providing intriguing evidence that an intact RF-Urp signaling axis is not sufficient to keep the zebrafish spine straight. However, strong cellular and metabolic stress responses were observed in the brains and bodies of scoliotic katnb1 mh102/mh102 mutants. Cellular stress can be initiated by multiple factors, including mitochondrial and electron transport chain disruptions and high levels of ROS (Dehdashtian et al., 2018) that can lead to an unfolded protein response and endoplasmic reticulum stress (Xiao and Loscalzo, 2020). Endoplasmic reticulum stress has also been shown to contribute to inflammatory responses and autophagy (Menu et al., 2012;Mou et al., 2020;Thangaraj et al., 2020). These signatures are all present in katnb1 mh102/mh102 mutant tissues, although specific cells/tissues of origin remain to be determined. Notably, recent studies have identified neuroinflammatory and oxidative stress signals as being necessary and sufficient to drive spinal curvature downstream of CSF homeostasis defects in ptk7a and sspo mutant zebrafish (Van Gennip et al., 2018;Rose et al., 2020).
Together with published data on ptk7a and sspo IS models (Van Gennip et al., 2018, our data for katnb1 mutants identify cell stress responses downstream of CSF flow defects as common pathogenic signatures associated with idiopathic-like scoliosis across multiple zebrafish IS models. How these physiological cues ultimately disrupt spine development and their functional relationship to human IS pathogenesis remains to be determined. However, further understanding the role for Katnb1 in ciliogenesis and the functional consequences of katnb1 mutations on ChP differentiation promise to provide valuable new insights into fundamental mechanisms governing axial morphogenesis.

Limitations of the study
The functional role for Katnb1 in motile cilia formation and the identity of polyglutamylated structures observed in the ChPs of katnb1 mutant embryos remain to be determined.
Although katnb1 mutants demonstrate clear ChP ciliation phenotypes, at this point it is still difficult to fully resolve which foxj1a-positive cell types underlie the scoliosis phenotype and whether observed cilia defects in the choroid plexus are causative of the phenotype.
Downstream of katnb1 motile cilia defects, the specific and causal roles for observed oxidative stress and unfolded protein responses, chromatin/epigenetic dysregulation, autophagy, and metabolic defects in spinal curve formation and progression remain to be determined.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:

Materials availability
There are MTA restrictions to the availability of zebrafish lines generated in this study, due to Institutional policies on the distribution of biological reagents.
Data and code availability d RNA-seq data have been deposited at NIH NCBI and are publicly available as of the date of publication. Accession numbers are listed in the key resources table. Microscopy data reported in this paper will be shared by the lead contact upon request.
d This paper does not report original code.
d Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

EXPERIMENTAL MODEL AND SUBJECT DETAILS Zebrafish
Established zebrafish husbandry protocols were adhered to and performed in accordance with Canadian Council on Animal Care (CCAC) guidelines. Wild type zebrafish from AB and TU strains were used. Embryos from natural mattings were grown at 28 C. When required, experimental animals were euthanized with tricaine (500 mg/L; MS-222/MESAB), followed by submersion of anesthetized fish in ice water for several minutes. As laboratory zebrafish strains do not utilize a chromosomal sex determination mechanism and sex differentiation does not initiate until after $3 weeks post fertilization (Kossack and Draper, 2019), we cannot report sex for our embryonic and juvenile studies. Sex was reported for adult Cobb angle measurements.

Calcein staining
Staining protocols were followed according to published reports (Du et al., 2001). To prepare Calcein solution, 10 mg of dry calcein powder (Sigma Cas# 154071-48-4) was dissolved in 50 mL of MilliQ water, with pH adjusted to 7.2. Selected fish were transferred to a petri dish filled with above solution. Fish were incubated in dish for 50 min. Solution was removed and fish were transferred to breeding tank filled with system water for 50 min. Animals were anesthetized and imaged on AxioCam MRc 1.4MP Colour Microscope Camera (Zeiss). After sacrificing, larval fish were subsequently genotyped.

Microcomputed tomography (mCT)
Fish were fixed in neutral buffered 10% formalin (Sigma) at 4 C overnight and then mounted in 1% low melt agarose (Sigma) in a plastic vial. Specimens were scanned for 4-8 min using SkyScan 1275 high resolution Micro-CT scanner (Bruker) with the X-ray power at 45-50 kV and 200 mA current and reconstructed with 17.74 mm isotropic resolution. 30 dpf samples were scanned for 26 min, Xray power 50 kV, 200 mA current, and 13.69 mm isotropic resolution. The images were analyzed using CTVox software (Bruker), and ImageJ Schindelin et al., 2015). Graphs were generated in Graphpad Prism 9.

Scanning electron microscopy (SEM)
Experimental animals were euthanized with tricaine (500 mg/L), followed by submersion of anesthetized fish in ice water for several minutes. Once euthanasia was assured, 3-week, 6-week, and 3-month-old brains were immediately dissected in cold PBST (13 phosphate saline buffer +0.25% TritonX iScience Article anesthetized in a dilute solution of Tricaine. Anesthetized fish were transferred onto an agarose injection plate with anesthetizing agent and system water. Ventricle location was identified using the diamondshaped pigment pattern on the top of the head, and a glass injection capillary was inserted into that region. Approximately 10 nL of non-toxic violet highlighter ink (Takase et al., 2013) was injected into the ventricle. Injected fish were imaged at 2 h post injection on an Axio ZoomV16 microscope (Zeiss).

Transgenesis
Entry plasmids were generated through BP recombination into respective pDONR plasmids (Invitrogen) and then shuttled into standard Tol2 kit Gateway compatible vectors using LR recombination methods to create the final transgenes (Kwan et al., 2007). To generate Tg(floxed-foxj1a::katnb1) zebrafish, previously generated p5E-foxj1aP (Grimes et al., 2016), pME-katnb1-stop (extraction of katnb1 using Gateway primers (detailed below) from a pME18SFL3 eukaryotic expression vector containing Danio rerio katanin p80 subunit B1 full length cDNA obtained from the National Institutes of Health Zebrafish Gene Center), and p3E-PolyA were recombined into pDEST Tol2LGSR transgenesis vector. This previously described transgenesis vector is Cre-excisable and permits conditional removal of transgenes with a visual readout of lens change colour (Grimes et al., 2016;Van Gennip, Boswell and Ciruna, 2018).
Embryos were injected at the one cell stage with 25 pg of assembled transgene (plasmid) and 25 pg of Tol2 transposase RNA and screened at 48 hpf for transgenesis marker expression. Imaging of reporter expression was performed on an Axio Zoom.V16 (Zeiss). Embryos with strong reporter expression were grown to adulthood and crossed with wildtype fish to establish stable F1 lines. Subsequent F1 lines harboring the transgene were bred into katnb1 mh102/+ and maintained in katnb1 mh102/mh102 fish.

Cobb angle statistical analysis
Lines were drawn parallel to the top and bottom most displaced vertebrae for each curve. The Cobb angle was then measured as the angle of intersection between lines drawn perpendicular to the original 2 lines (Van Gennip, Boswell and Ciruna, 2018). Analysis was conducted using ImageJ Schindelin et al., 2015). Cobb angle measurements for lateral and dorsal curvatures were summed to obtain a combined Cobb angle measurement for each fish. Results were graphed and statistical significance was calculated using a student's t-test using GraphPad Prism 9.

Choroid plexus cilia quantification
Et(cp:EGFP) sj2 fish (Henson et al., 2014) were crossed into katnb1 mh102/+ and then offspring were intercrossed to generate heterozygotes and mutants. Brains from these crosses were dissected, fixed, and stained with immunofluorescent antibodies polyglutamylated tubulin GT335 antibody and GFP polyclonal antibody (described above). Brains were imaged on an LSM 710 confocal microscope (Zeiss) at 633. Images were processed using ImageJ Schneider et al., 2012;Schindelin et al., 2015) and Z-stack composites were used for analysis. To determine cilia number, a set area of the choroid plexus was used. Cilia in the given area were counted using ImageJ. To determine cilia length, cilia within the above specified area of the ChP were measured using ImageJ. Cilia length for each sample were then averaged. Results were graphed and statistical significance was calculated using a student's t test using GraphPad Prism9.