3D assessment of intervertebral disc degeneration in zebrafish identifies changes in bone density that prime disc disease

ARTICLE OPEN 3D assessment of intervertebral disc degeneration in zebrafish identifies changes in bone density that prime disc disease Erika Kague 1 , Francesco Turci 2 , Elis Newman 1 , Yushi Yang 2,3,4 , Kate Robson Brown 5,6 , Mona S. Aglan 7 , Ghada A. Otaify 7 , Samia A. Temtamy 7 , Victor L. Ruiz-Perez 8 , Stephen Cross 9 , C. Patrick Royall 2,10 , P. Eckhard Witten 11 and Chrissy L. Hammond 1 Back pain is a common condition with a high social impact and represents a global health burden. Intervertebral disc disease (IVDD) is one of the major causes of back pain; no therapeutics are currently available to reverse this disease. The impact of bone mineral density (BMD) on IVDD has been controversial, with some studies suggesting osteoporosis as causative for IVDD and others suggesting it as protective for IVDD. Functional studies to evaluate the influence of genetic components of BMD in IVDD could highlight opportunities for drug development and repurposing. By taking a holistic 3D approach, we established an aging zebrafish model for spontaneous IVDD. Increased BMD in aging, detected by automated computational analysis, is caused by bone deformities at the endplates. However, aged zebrafish spines showed changes in bone morphology, microstructure, mineral heterogeneity, and increased fragility that resembled osteoporosis. Elements of the discs recapitulated IVDD symptoms found in humans: the intervertebral ligament (equivalent to the annulus fibrosus) showed disorganized collagen fibers and herniation, while the disc center (nucleus pulposus equivalent) showed dehydration and cellular abnormalities. We manipulated BMD in young zebrafish by mutating sp7 and cathepsin K, leading to low and high BMD, respectively. Remarkably, we detected IVDD in both groups, demonstrating that low BMD does not protect against IVDD, and we found a strong correlation between high BMD and IVDD. Deep learning was applied to high-resolution synchrotron µCT image data to analyze osteocyte 3D lacunar distribution and morphology, revealing a role of sp7 in controlling the osteocyte lacunar 3D profile. Our findings suggest potential avenues through which bone quality can be targeted to identify beneficial therapeutics for IVDD. Bone Research (2021) 9:39 ; https://doi.org/10.1038/s41413-021-00156-y INTRODUCTION Back pain is a global healthcare concern and economic burden. It is the leading cause of years lived with disability, with estimates that over 80% of adults will suffer from symptoms at some time in their lives 1 . Intervertebral disc degeneration (IVDD) is one of the major causes of back pain symptoms, one of the first pathological signs, and a potential target for intervention 2 . As the global population ages, a substantial increase in morbidity due to degenerative diseases and back pain is expected; 3 this increase emphasizes the importance of improving our understanding of the causes of IVDD and its relationship with other common degenerative conditions to revise or further develop novel therapeutic strategies. Intervertebral discs (IVDs) connect consecutive vertebral bodies; their main function is mechanical, and they act as shock-absorbing cushions 4 . Mammalian discs have an exquisite three-dimensional (3D) architecture comprised of the outer annulus fibrosus (AF) made from fibrocartilage and an inner region, the nucleus pulposus (NP). The NP is derived from the notochord; it contains chondrocyte-like cells and a gelatinous matrix composed of collagen type 2 and proteoglycans that are vital for the function of the discs and serve as afluid-filled shock-absorbing cushion. IVDs are avascular; nutrients are delivered through diffusion at the endplates and AF 5,6 . While fissures and bulging AF are commonly found during aging, progressive loss of proteoglycans, water, and nutrient content leads to irreversible cellular and mechanical changes in the NP 2 . Aging, spine deformities, injuries, diseases, and genetic factors are involved in the pathogenesis. Several studies have highlighted some of the genetics of IVDD 7 , which include genes involved in bone and cartilage homeostasis (i.e., collagens, metalloproteinases, vitamin D, GDF5, and IL-6) 7–9 . While genome-wide association studies (GWAS) have identified an association of a few loci with back pain, IVDD endophenotypes, larger samples, replication, and functional studies are needed to improve reliability and to identify genes associated specifically with IVDD 7,8,10,11 . The association between IVDD and bone mineral density (BMD) has been controversial. It has long been proposed that reduced bone quality leads to progressive endplate degradation and spondylosis, ultimately culminating in increased disc degeneration 12 . However, Received: 25 September 2020 Revised: 22 March 2021 Accepted: 7 May 2021 1 School of Physiology, Pharmacology and Neuroscience, Biomedical Sciences, University of Bristol, Bristol, UK; 2 School of Physics, HH Wills Physics Laboratory, University of Bristol, Bristol, UK; 3 Centre for Nanoscience and Quantum Information, University of Bristol, Bristol, UK; 4 Bristol Centre for Functional Nanomaterials, University of Bristol, Bristol, UK; 5 Department of Anthropology and Archaeology, University of Bristol, Bristol, UK; 6 Department of Mechanical Engineering, University of Bristol, Bristol, UK; 7 Clinical Genetics Department, Human Genetics and Genome Research Division, Center of Excellence for Human Genetics, National Research Centre, Cairo, Egypt; 8 Instituto de Investigaciones, Biomedicas de Madrid, and Ciber de Enfermedades Raras (CIBERER), Madrid, Spain; 9 Wolfson Bioimaging Facility, Biomedical Sciences, University of Bristol, Bristol, UK; 10 School of Chemistry, University of Bristol, Bristol, UK and 11 Evolutionary Developmental Biology, Department of Biology, Ghent University, Ghent, Belgium Correspondence: Erika Kague ([email protected]) or Chrissy L. Hammond (chrissy.hammo[email protected]) www.nature.com/boneres Bone Research ©The Author(s) 2021 1234567890();,: patients with low BMD, despite having higher risks of vertebral body fractures, seem to show reduced rates of IVDD 13–15 .Accordingtothis hypothesis, IVDD would be delayed by osteoporosis, an aging-related degenerative and debilitating condition that affects millions of people worldwide 16 . However, the long-term effects of osteoporosis or poor bone quality are far from resolved, as other studies have reported a positive association with disc degeneration 17–19 ,and treatments used for osteoporosis, such as alendronate or calcitonin retarded IVDD in ovariectomized rats 18,20–22 . On the other hand, a positive association of higher lumbar spine BMD with IVDD has been suggested from cadaveric studies 23 , a UK Twin Study, and populational studies 24,25 . Functional studies to understand the relationship between the genetics of BMD and IVDD integrity could clarify this relationship and aid in the identification of new therapeutic opportunities. Zebrafish (Danio rerio) provide an attractive teleost model to study adult skeletal conditions, including those related to osteoporosis, osteoarthritis, and spinal deformities 26,27 .Zebrafish carrying mutations in genes associated with osteoporosis and osteogenesis imperfecta exhibit frequent fractures in the ribs and low BMD 28–31 . In teleosts, the connections of the endplates of vertebral bodies are similarindesignandfunctiontomammalian IVD (hereafter called IVD). Centrally located vacuolated notochord cells with the same function as NP (hereafter called the NP) are surrounded by a strong composite ligament that functions like the AF (hereafter called the AF) 32 . The ligament is composed of, from inside to outside, collagen type II embedded in a matrix resembling cartilage, elastin, and collagen type 1 fiber bundles 33 . In adult teleosts, vacuolated notochord cells turn into fibrous, keratinized, connective tissue (the septum and notochord strand) and extracellular vacuoles 32,34 .Disc damage and subsequent vertebral body fusion have been reported in farmed Atlantic salmon, suggesting that zebrafish could be a potential degenerative model for IVD 35 . We have shown frequent spinal deformities in aged zebrafish, including spinal curvatures, osteophytosis, and changes akin to those of osteoarthritis 36 . Recently, Monma et al. showed reduced trabecular volume and number, similar to the reductions observed in aging humans 37 .Moreover, genetic zebrafish models displaying BMD fluctuations are attractive tools for functional evaluation of the interplay between BMD and IVDD. Here, we report for the first time IVDD in adult zebrafish. We describe microstructure alterations of the vertebral bone, showing local changes in bone mineral distribution, the osteocyte lacunar profile, and collagen fiber organization, demonstrating that deterioration of bone quality in aged fish leads to increased fragility, as is seen in individuals with osteoporosis. In the IVD, we characterized changes to bone and soft tissue involving collagen fibers of the AF, dehydration, fibrosis (scarring tissue), and cellular changes in the NP. To address the controversy over the relationship of BMD with IVDD, we analyzed the spines from zebrafish with low BMD (sp7 −/− ) and high BMD by genetically manipulating cathepsin K. Interestingly, both zebrafish mutant lines showed premature IVDD, contradicting the idea that low BMD is a protective factor in IVDD. A positive correlation between low BMD and the incidence of IVDD and between high BMD and the incidence of IVDD was observed, suggesting a U-shaped model in which deviation in either direction from normal BMD increases the risk of developing IVDD. By analyzing changes in bone quality that precede the detection of BMD fluctuations, we suggest that therapeutics targeting improving bone quality can potentially help in the prevention and treatment of disc disease. RESULTS Radiographic signs of IVDD in the aging zebrafish vertebral column Radiographic signs of IVDD in humans typically include osteophytosis, endplate sclerosis, and disc space narrowing 38 . To identify progressive changes in the spine during aging that could act as markers for IVDD, we analyzed microcomputed tomography (μCT) image data of 1- (n=36), 2- (n=16) and 3- (n =34) year-(y)-old wild-type (wt) zebrafish, followed by spinal abnormality classification based on severity (Fig. 1a, b). With increasing age, zebrafish exhibited increased severity of vertebral fusion, misalignments, osteophytes, endplate sclerosis, uneven disc spacing, IVD narrowing and disc calcification (Fig. 1b, c). Fifty-seven percent of all fusions and disc narrowing occurrences were found between vertebrae in the rib-bearing abdominal region. Degenerative changes in the vertebral centra can be detected by simple radiographs or using Alizarin Red S staining (Supplementary Fig. 1). To understand whether vertebral bodies display shape changes during aging, we performed twodimensional (2D) morphometric analysis on the third transitional vertebra from a subset of young and aged fish (n=8) using lateral μCT images. Shape variation confirmed modification during aging (P=0.043). Interestingly, the highest deformation (standard deviation from the normal shape) corresponded to sclerotic endplates (Supplementary Fig. 2). Thus, aging zebrafish develop radiographic signs of IVDD. Elevated bone density and increased risk of fractures in aged zebrafish spines: are the bones of aged zebrafish osteoporotic? To study BMD and IVDD using zebrafish, we analyzed μCT image data with grayscale calibration for BMD, comparing 3D volumetric renders of young (1 year) and aged (3 years) zebrafish. Regions of higher density corresponded to the greatest morphometric variations at endplates, while lower density was observed in the middle of the centra (Fig. 1d). Increased centra tissue mineral density (TMD) was detected when the same centrum was compared among spines (P=0.004 7) (Fig. 1e). We analyzed a higher number of centra in aging zebrafish by automated centra segmentation (Fig. 1f and Supplementary Fig. 3). Due to the complex morphology of the abdominal region leading to difficulties in separating the centra from the ribs and the highest mechanical load of the caudal vertebral bodies 39 (similar to the human lumbar spine), we excluded those centra in the abdominal region from our analysis. We analyzed 644 vertebral centra from 3month, 1-, 2and 3-year-old zebrafish (63, 291, 83, and 207 centra, respectively). Although the number of vertebral centra analyzed was different between the groups, potentially causing statistical bias in the automated analysis, we observed a gradual increase in bone density from 3-month-old to 3-year-old samples (Fig. 1g), among other measurements retrieved (Supplementary Fig. 4). The heterogeneous distribution of bone tissue minerals is an indicator of poor bone quality and fracture risk 40,41 . We hypothesized that uneven BMD would alter the material and structural properties of aged bone and result in increased fragility. The standard deviation within samples suggested an increased mineral heterogeneity distribution in aged spines (Fig. 1i). To test the performance of the vertebral column, we applied compressive force to a section of the spine (anterior to posterior axis) ex vivo using a material testing stage (MTS) within a μCT machine; this approach allowed us to monitor the point of failure of the vertebral column in a protocol previously described 42 . We tested young (1 y) and aged (3 y) samples (n=3 for each group). While young spines failed at 15 N, aged spines failed at 3.33 N (P=0.005) (Fig. 2a, d and Supplementary Fig. 5G). The results demonstrate that a focal increase in bone density colocalizes with deformed endplates in aging zebrafish; uneven mineral density distribution supports poor bone quality, leading to increased fragility. Our results corroborate human populational studies demonstrating that degenerative changes of the lumbar spine may lead to falsely elevated BMD values and cause underdiagnosis of osteoporosis 43 . Despite increased bone density at zebrafish endplates, the bones of aged zebrafish are osteoporotic. Impact of bone density in disc degeneration E Kague et al. 2 Bone Research (2021) 9:39 Young zebrafish with low BMD show premature radiographic IVDD The study of aging zebrafish to disentangle the association between BMD and IVDD is complex, as our data indicate increased BMD associated with IVDD in aging fish; furthermore, increased fragility in aged bones, resembling osteoporosis, was also associated with IVDD in aging zebrafish. To probe this relationship, we analyzed the vertebral column and IVD of young fish with genotypes associated with reduced bone density. Sp7 is a major osteogenesis transcription factor expressed in osteoblasts 44,45 . Genome-wide association studies (GWAS) have identified associations between SP7 and BMD 46 , between SP7 and lumbar spine BMD and between SP7 and estimated BMD (musculoskeletal a 1 2 3 5 64 d HA/(g·cm -3 ) HA/(g·cm -3) 0.4 Young Aged 1.3 b Young Aged AV PV hgf Centra segmentation Severe Unaffected Mild Moderate Mild/ Moderate c e Fusions Misalignments Osteophytosis Sclerosis Narrowing Calcification 1 year 2 years3 years Young (1 year) Aged (3 years) 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 P=0.004 7 1.2 1.75 1.50 1.25 1.00 0.75 0.50 0.25 1.0 0.8 Relative bone density Relative bone density std 0.6 3 months 1 year 2 years 3 years 3 months 1 year 2 years 3 years Fig. 1 Progressive abnormalities found in aged zebrafish vertebral columns. a3D rendering from μCT images of young (1 year) and aged (3 years) spines. AV anterior vertebra, PV posterior vertebra, IVD (arrow). Scale bar =500 μm. bFrequent changes observed at the IVD. 1: normal IVD; 2: osteophytes (pink arrow), vertebral misalignment and IVD narrowing (green arrow); 3: endplate sclerosis; 4: sclerosis (pink arrow) and IVD narrowing (green arrow); 5: sclerosis and fusion (pink arrow); 6: sclerosis and IVD calcification (pink arrow). Scale bar =100 μm. cHeat map graph showing spinal morphological changes classified by severity during aging (1 year n=36 (42% females, 58% males), 2 years n=16 (57% females, 43% males), and 3 years n=34 (55% females, 45% males)). Average fish standard lengths (measured from tip of the head to the last vertebral column): 1 year =3.22 cm (0.19 SD), 2 years =3.41 cm (0.18 SD), and 3 years =3.46 cm (0.23 SD). d3D rendering from μCT images of young and aged fish, color coded to show bone mineral density changes. The selected area of the spine (dashed box) is magnified, as shown on the right of the panel. Higher density colocalizes with regions of sclerosis and deformities (arrowhead) in the aged spine. Scale bar =500 μm. eTMD (tissue mineral density) retrieved from the third thoracic vertebrae in young (1 year) and aged (3 years) fish. Nonparametric, two-tailed, Mann–Whitney test; data are the mean and SD. Pvalues are indicated. f3D volume rendering from a μCT image of wt fish showing an individual vertebral centrum segmented by computational automation. gRelative bone density from the vertebral centra in aging fish (3 months to 3 years). Average standard lengths: 3 months =2.6 cm (0.18 SD); 3 years =3.46 cm (0.23 SD). The notch plot was scaled by the average value from the 3-year centra. hWithin-sample standard deviation in bone density. The notch plot was scaled by its average value from the 3-year centra. c,eGenerated in Prism 8. g,hGraph was generated in Python Impact of bone density in disc degeneration E Kague et al. 3 Bone Research (2021) 9:39 knowledge portal, mskkp.org). The SP7 locus was also identified in a meta-analysis evaluating gene variants associated with BMD across the lifespan 47 . Mutations in SP7 cause recessive osteogenesis imperfecta (OI type XII), characterized by generalized osteopenia, recurrent fractures, bone deformities and bent bones 48–50 . Zebrafish sp7 −/− show frequent fractures in the ribs, skull abnormalities (Wormian bones) and tooth phenotype, similar to those found in human patients 31,51 . By performing μCT image analysis of young (3 months, mo) sp7 −/− zebrafish, we observed a dramatic reduction in TMD in comparison to wt siblings (p= 0.034) (Fig. 2b, e and Supplementary Fig. 5E). Interestingly, endplate sclerosis, calcification and broadening of the IVD (Fig. 2b, c and Supplementary Fig. 5D) suggest premature bone degenerative changes. The compressive loading experiment for the section of sp7 −/− spine (3 mo) and comparative sibling samples showed that sp7 −/− areas bent more and suddenly failed at 14.67 N (P=0.05) (Fig. 2d and Supplementary Fig. 5F, G). Bent long bones have been described in a human patient homozygous for a recessive mutation in SP7 (c.1052delA, p.(Glu351Glyfs*19)) 48 , while vertebral column curvature has also been reported 49,50 , IVD wt 3 mo sp7-/- 3 mo wt 1 year wt 3 years 0 5 10 15 20 Falirure point/N P= 0.008 P= 0.05 P= 0.005 a f Compression (N =0) Failure (N>0) Anterior Posterior Frontal Lateral Region 1 Region 2 1 2 * HA/(g·cm -3 ) HA/(g·cm-3) 0.2 1.0 wt (3 mo) sp7 -/-(3 mo) ec b d wt (3 mo) sp7-/- (3 mo) 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 P = 0.035 7 1 year 1 year Fig. 2 Altered vertebral column biomechanics in aged and sp7 −/− zebrafish show increased bone fragility. aIntact vertebral column motion segments (composed of three vertebrae surrounded by trunk musculature) were placed in a material testing stage (MTS), followed by μCT imaging at increasing compression forces. An example of vertebral column segments before compression (N =0) and after failure (N > 0) is shown. Anterior and posterior orientations of the zebrafish are annotated. b3D volumetric rendering from μCT images of 3-month-old wt and sp7 −/− zebrafish color coded for tissue mineral density (TMD). The dashed box shows the magnified region presented on the right of the panel. Scale bars =500 μm. cVolumetric rendering from μCT images of 1-year-old wt and sp7 −/− zebrafish. Note IVD calcification (arrowhead) and abnormal shapes of the centra (arrow) in sp7 −/− zebrafish. Scale bars =500 μm. dFailure points during compressive forces of 3-month-old (3 mo) sp7 −/− zebrafish, wt zebrafish siblings, 1-year-old wt siblings, and 3-year-old wt siblings (n=3 per group). One-way ANOVA and post hoc Holm–Sidak’s multiple comparisons test were used; data are the mean and SD. Pvalues are indicated. The graph was generated in Prism 8. Scale bars =500 μm. eTMD of 3-month-old (3 mo) wt and sp7 −/− zebrafish (wt n=5 (3 males +2 females), sp7 −/− zebrafish n=3 (2 males +1 female)). Standard fish lengths: wt =2.43 cm (0.2 SD); sp7 −/− =2.17 (0.4 SD). Nonparametric, two-tailed, Mann–Whitney test; data are the mean and SD. Pvalues are indicated. fRadiograph of a 16-year-old male with a frameshift mutation in SP7 (c.1052delA). Frontal and lateral images are displayed. Regions 1 and 2 are magnified and displayed on the right. Lumbar spinal curvature (asterisk), biconcave vertebrae (pink arrows), the spread of IVD, and signs of IVD calcification (cyan arrowhead) are shown Impact of bone density in disc degeneration E Kague et al. 4 Bone Research (2021) 9:39 changes have not been described to date. We contacted the Egyptian child with the SP7 mutation reported in 2010; 48 now 15 years old, the patient displayed 2–3 fractures per year and was unable to walk or stand. Radiographs of the patient’s trunk showed mild spinal curvature (kyphoscoliosis), abnormal vertebral shape (biconcave wedging, “cod shape”), endplate sclerosis, more areas affected by IVD and signs of calcification (Fig. 2f and Supplementary Fig. 6), similar to our findings in zebrafish. Therefore, the sp7 −/− vertebral column recapitulates changes observed in a human SP7 patient. Our results indicate an association of SP7, low BMD, and the incidence of IVDD. Mineral distribution heterogeneity, reduced mineralization robustness, and reduced bone quality in aged bones substantiate osteoporosis in aged zebrafish The biomechanical properties of bone are affected not only by the total amount of minerals but also by the distribution and organization of minerals within the nonmineralized matrix 33,52–54 . We imaged young (1 y), aged (3 y), and sp7 −/− (1 y) vertebrae using synchrotron radiation-based μCT (SRCT) (0.33–0.68 μm isotropic resolution, n=3fish per group) (Fig. 3a), allowing observation of subtle 3D changes to the IVDs (Fig. 3b). Virtual cross-sections through the centra revealed distinct mineral distribution patterns among the groups studied (Fig. 3b). Quantification of % mineralization robustness, here defined as the relative amount of high grayscale values in a sample (Fig. 3b, red color), showed a tendency towards lower mineralization robustness in aged spines and a significant reduction in sp7 −/− vertebrae (1 y) when compared with young wt vertebrae (1 y) (P=0.007 9) (Fig. 3c). While wt and sp7 −/− vertebrae (1 y) displayed coefficients of variation (CVs) within samples of 4% and 5.8%, respectively, the aged group showed a CV of 11.95%, indicating mineral distribution heterogeneity. Mineral distribution and bone mechanical properties also depend on the organization of collagen, the most abundant component of the extracellular matrix. To examine collagen organization, we performed picrosirius red staining on paraffin sections and quantified collagen fiber thickness (Fig. 3d, e). Aged samples failed to maintain collagen organization; the matrix in aged samples showed a relative increase in thinner fibers in comparison to 1-year-old wt fish, similar to the immature collagen organization seen in 3-month-old wt fish (Fig. 3d, e). sp7 −/− fish (1 y) displayed a relative reduction in thick collagen fibers (P= 0.000 4) but no significant quantitative changes in thinner fibers, suggesting that these fish failed to develop normal bone structure and composition (Fig. 3d, e). Therefore, heterogeneous mineral distribution accompanied by deterioration of collagen organization within the bone matrix in aged samples suggests that bone quality is subtly impaired prior to the changes observed in BMD. Characterization of the 3D osteocyte lacunar profile in zebrafish bones demonstrates the mechanosensitivity regulatory role of sp7 BMD changes could alter bone mechanical loading and influence IVDD. Osteocytes are the most abundant bone cells, strategically positioned within the bone matrix in cavities called lacunae, interconnected through cell extensions housed in small canals and canaliculi. This communication network allows osteocytes to sense mechanical loading of the bone matrix 55,56 . Variation in the 3D parameters of lacuna morphology mirror osteocyte mechanosensitivity, observed in aging human populations 57 and in bone conditions, such as osteoporosis 55 . Osteocytes detect microfractures and initiate targeted bone remodeling but also participate in bone remodeling in the bone surface by signaling to osteoblasts and osteoclasts 58 . We explored whether variations in osteocyte lacunae morphology would be observed in aged samples and whether this observation could help to improve the present understanding of the bone mechanical properties of sp7 −/− .We used deep learning to automate the 3D analysis of osteocyte lacunae from the SRCT image dataset (Supplementary Fig. 7A), enabling rapid and efficient retrieval of lacunae number/bone volume, shortest distance from the lacuna to the bone surface and from one lacuna to the proximal lacuna, lacunae volume, orientation and sphericity (Supplementary Fig. 7B). Lacunae showed an arrangement with a specific orientation (Fig. 4a), similar to that reported for wt zebrafish 59 . This arrangement was lost in sp7 −/− (1 y) (Fig. 4b). No significant differences were observed in cells/bone volume during aging (P=0.925 7). However, sp7 −/− fish showed a dramatic reduction in cells per μm 3 of bone (P=0.000 5) (Fig. 4c), and also showed an increase in the shortest average distance between lacunae (P=0.001 8) (Fig. 4e). As bone is formed, osteocytes are trapped in the newer bone matrix, more distant from the bone surface. Shallower lacunae were found in sp7 −/− fish (P=0.028 2) (Fig. 4d). While no significant changes in the mean volume of the lacunae were identified during aging, our results showed a subpopulation of very small cells in sp7 −/− fish (~20 μm 3 ), which was significantly different when evaluating the average volume of lacunae (Fig. 4f, g). We calculated the sphericity of the lacuna shape (perfect circle =1), highlighting the higher sphericity in sp7 −/− fish compared to wt fish (P< 0.000 1) (Fig. 4h). While the reduced cellularity observed in sp7 −/− fish fits with the function of the gene in osteoblast to osteocyte differentiation 44 , we also demonstrated the role of sp7 in regulation of the 3D profile of lacunar organization within the bone. This is suggestive of an impaired ability to sense mechanical load to regulate bone homeostasis. Interestingly, aged bone did not display significant changes in the lacunar profile, suggesting that IVDD in aging samples is independent of the lacunar profile. Zebrafish IVDD is characterized by the accumulation of scar tissue, dehydration and defective cellular organization Having characterized the changes in bone morphology occurring during IVDD, we next wanted to study cellular changes in the discs (NP and AF). For this, we performed further histological analysis of young (1 y), aged (3 y), and sp7 −/− (1 y) zebrafish. Overall IVD morphology was assessed by toluidine blue staining of sequential sagittal sections throughout the spine. The NP of aged samples exhibited accumulation of fibrotic tissue (scarring) and disorganized cells (vacuolated and nonvacuolated), similar features to those observed in sp7 −/− fish (Fig. 5b). To observe the 3D organization of the fibrotic tissue, we used phosphomolybdic acid as a contrast agent for μCT image analysis (5 μm resolution). We confirmed the 3D attachment of the fibrotic tissue with the bony walls and notochord sheath region, which suggests an involvement of the notochord sheath layer during the process of degeneration (Fig. 5c). As the main components of the notochord string are keratin and collagen type I (Supplementary Fig. 8A), we performed pankeratin staining, which labels keratin (red/orange), collagen (including collagen type I, yellow), and glycosaminoglycans (blue) (Fig. 5b). In discs from young zebrafish, we observed keratin (red) in the notochord strand (middle section). Keratin was evident throughout the scar tissue of aged NPs but not in young sp7 −/− fish. Similar to degeneration in mammals, dehydration of the NP was detected in sp7 −/− zebrafish and dramatically in aged zebrafish (pankeratin, blue). To check whether dehydration led to collapse of the vacuolated notochord cells and cell death, we used antibody staining for cadherin in the NP vacuolated cells. Interestingly, the dramatically dehydrated discs in aged zebrafish showed decreased cellularity and a lack of vacuolated cells, while sp7 −/− zebrafish (1 y) displayed focal areas of collapsed NP vacuolated cells with altered cellularity; in other areas, the NP vacuolated cells in sp7 −/− zebrafish appeared normal (Fig. 5b). These results indicate that NPs in aged samples are mostly fibrous, while in sp7 −/− samples, premature degeneration is observed and accompanied by cellular changes. The AF showed stretched and disordered collagen fibers in aged samples and, remarkably, in sp7 −/− fish (Fig. 5b, toluidine blue). However, the AF maintained Impact of bone density in disc degeneration E Kague et al. 5 Bone Research (2021) 9:39 b a 12 34 SRCT 3D render: Fine IVD abnormalities dc Young SRCT 3D render SRCT orthoslice (mineral distribution) 100% 0% Aged 100%0% sp7-/- 100% 0% Picro-sirius red Agedsp7-/- Young HighLow e 3 mo 1 y 3 y sp7-/- 1 y 3 mo 1 y 3 y sp7-/- 1 y 3 mo 1 y 3 y sp7-/- 1 y 0 20 40 60 80 100 Mean intensity P= 0.000 4 P= 0.032 0 20 40 60 80 Mean intensity P= 0.000 1 P= 0.033 8 P= 0.001 3 0 10 20 30 40 50 Mean intensity P< 0.000 1 P< 0.000 1 P< 0.000 1 P< 0.000 1 1 y 3 y sp7-/- 1 y 0 10 20 30 40 50 60 70 80 90 100 Mineralization robustness/% P= 0.007 9 P= 0.06 Fig. 3 SRCT reveals subtle bone morphological abnormalities and alterations in mineral density distribution. a3D volumetric rendering from μCT images showing fine endplate abnormalities. 1: normal IVD; 2: endplate sclerosis (blue arrow) and IVD calcification (pink arrow); 3: sclerosis (blue arrow), uneven edges and AF calcification (pink arrow); 4: sclerosis (blue arrow), AF calcification (pink arrow); osteophyte (dashed blue arrow). Scale bars =50 μm. b3D volumetric rendering from μCT images showing IVD internal changes in aged and sp7 −/− zebrafish. Calcification depicted by pink arrow (SRCT 3D rendering). Note, young zebrafish with IVD and disc misalignment shows points of IVD calcification. Virtual sections (orthoslices) color coded for bone density (SRCT orthoslice). Dashed box regions are magnified and displayed on the right. Scale bars =50 μm. cQuantification of mineralization robustness (%) in wt young (1 y), wt aged (3 y), and sp7 −/− 1 y zebrafish (n =3 per group, average from 4 sections per fish). One-way ANOVA and post hoc Holm–Sidak’s multiple comparisons test were used; data are the mean and SD. Pvalues are indicated. dParaffin sections stained with picrosirius red showing the thickness of collagen fibers (thin =blue/ green and thick=red/orange). Scale bars =50 μm. eQuantification of collagen fiber thickness from picrosirius red staining (blue =thin, green =medium, red =thick) (1 y, n=8; 3 y, n=8; 3 months, n=7; sp7 −/− 1y,n=6; we analyzed three sections per fish). One-way ANOVA, post hoc Tukey’s multiple comparisons test; data are the mean and SD. Pvalues are indicated Impact of bone density in disc degeneration E Kague et al. 6 Bone Research (2021) 9:39 a dcb efg Deep learning: osteocyte lacunar profile Young Aged sp7 -/- 0° 90° 0°0° 90°90° h Young Aged sp7-/- 0.000 00 0.000 02 0.000 04 0.000 06 0.000 08 0.000 10 (Lacunae/bone)/µm3 P= 0.000 5 P= 0.000 4 ns Young Aged sp7-/- Young Aged sp7-/- Young Aged sp7-/- 0 50 100 150 200 Mean volume/µm3 ns ns ns Young Aged sp7-/- 12 14 16 18 20 22 24 Closest lacunae/µm P= 0.001 8 P= 0.001 7 ns 0.0 0.5 1.0 Sphericity P< 0.000 1 P< 0.000 1 ns 35 40 45 50 55 60 65 25% lacunae volume/µm3 P< 0.000 1 P< 0.000 1 ns Young Aged sp7-/- -60 -40 -20 0 20 Distance to bone surface/µm ns P= 0.028 2 ns 0 20 40 60 80 100 P= 0.072 2 P= 0.041 2ns 90 60 30 0 90 60 30 0 90 60 30 0 Fig. 4 The osteocyte lacunar profile is unchanged in aged fishbut is dramatically compromised in young sp7 −/− fish. aThe top panel presents 3D volumetric rendering and postimaging analysis of young (1 y wt), aged (3 y wt), and sp7 −/− (1 y) (lateral view of the IVD, 0°), showing vertebral bone and segmented lacunae (labeled in green), resulting from automated image segmentation using deep learning. Note the distribution of lacunae at the endplates and along the bone, with a dramatic reduction in sp7 −/− . The bottom panel shows segmented lacunae (90° clockwise rotation from the lateral view of the IVD), color coded to show orientation (angle), with the center of the vertebral centrum as reference for orientation. Scale bars =50 μm. bViolin plot of lacunae orientation from the center of the centrum. Data are individual zebrafish (n=3). Nested one-way ANOVA, post hoc Tukey’s multiple comparisons test. Pvalues are indicated. ns nonsignificant. cViolin plot of number of lacunae per volume of bone. One-way ANOVA, post hoc Tukey’s multiple comparisons test. Pvalues are indicated. ns nonsignificant. dViolin plot of lacunae distances from the bone surface. Nested one-way ANOVA, post hoc Tukey’s multiple comparisons test. Pvalues are indicated. ns nonsignificant. eViolin plot of the distance to the closest lacunae. Data are the mean (n=3). Nested one-way ANOVA, post hoc Tukey’s multiple comparisons test. Pvalues are indicated. ns nonsignificant. fViolin plot of lacunae volume. Note the subpopulation of lacunae showing a small volume in sp7 −/− zebrafish. Data are the mean (n=3). Nested one-way ANOVA, post hoc Tukey’s multiple comparisons test. Pvalues are indicated. ns nonsignificant. gThe top 25% smallest lacunae volume from young (1 y), aged (3 y), and sp7 −/− (1 y) zebrafish were compared. One-way ANOVA, post hoc Tukey’s multiple comparisons test. Pvalues are indicated. ns nonsignificant. hViolin plot of lacunae sphericity (circularity). Data are the mean (n=3). Nested one-way ANOVA, post hoc Tukey’s multiple comparisons test. Pvalues are indicated. ns nonsignificant Impact of bone density in disc degeneration E Kague et al. 7 Bone Research (2021) 9:39 the same layers of type I and II collagens and elastin. Given the extreme phenotype of sp7 −/− , we tested whether the NP and AF would reflect degeneration or developmental abnormalities in these fish. We analyzed the expression of sp7 using whole-mount in situ hybridization for sp7 and imaged dissected spines of the sp7 reporter line Tg(Ola. Sp7:nlsGFP) zf132 . While osteoblasts at the endplates were positive for sp7, we did not detect sp7 expression in the NP (Supplementary Fig. 8B, C). We analyzed histological sections of sp7 −/− zebrafish aged from 1 to 3 months old. At 1month-old, despite the thin vertebral bones, we did not identify changes in the AF or NP. At 3 months old, disorganized AF and small IVD calcification were already observed (Supplementary 8C). Thus, we confirmed that changes in the vertebral column in sp7 −/ − fish are not developmental but are a consequence of premature degeneration. We next tested the composition of the mineralized IVD by staining cryosections of wt and sp7 −/− zebrafish to detect calcium phosphate. We confirmed that its composition is similar to hydroxyapatite (Ca 5 (PO 4 ) 3 (OH)) (Supplementary Fig. 8C). In conclusion, aged zebrafish develop IVDD in which histopathological changes include participation of the notochord sheath layer, leading to fibrotic, dehydrated, and acellular NPs—resembling those seen in human IVDD patients (collagen type I accumulation and keratin)—and destabilization of collagen fibers of the AF. Moreover, sp7 −/− fish develop premature IVDD with histopathological similarity to aged discs. 3D assessment of the zebrafish AF shows disrupted collagen fibers that prime disc herniation During disc degeneration, the collagen fibers of the AF are prone to rupture and tearing, leading to disc herniation, a frequent cause of back pain 5 . To investigate collagen fibers of the AF, we imaged whole IVDs in 3D through second harmonic generation (SHG) imaging using multiphoton microscopy. A healthy zebrafish disc is formed by a mesh of collagen fibers at the endplate (Fig. 6a, arrow); perpendicular fibers mark a boundary between the endplate and the AF (Fig. 6a, arrow), formed by smooth organized collagen fibers. Aged zebrafish, in contrast, have a heterogeneous mix of thick and thin fibers in the bone. Remarkably, in aged and sp7 −/− fish, perpendicular AF fibers were not observed, and the boundary AF/endplate appeared to be lost; thick fibers traveled through the AF (Fig. 6a, dashed arrow). We traced individual fibers using CTFire 60 (Fig. 6a, second column: fiber tracing) and detected a significant change in the frequency distribution of the angle of fibers in sp7 −/− zebrafish when analyzing the whole IVD (P=0.04) (Fig. 6b, c). We then explored whether collagen changes would lead to rupture of the AF and disc herniation. 3D volumetric renders showed swelling across the AF discs of aged zebrafish and localized bulging in sp7 −/− zebrafish (Fig. 6, green arrows), suggesting that changes in the orientation of the collagen fibers could prime herniation. Given the specific bulging phenotype of sp7 −/− fish and the differences in collagen fiber orientation, we performed scanning electron microscopy (SEM) of the IVDs of sp7 −/− and wt (1 year) fish to observe the AF in detail. Mutants showed more widespread IVD, with AF stretching, bumpy surfaces, and abrupt endplate edges (Fig. 6d). In conclusion, we showed that altered collagen fiber orientation primed bulging discs as part of the complex 3D modifications involved in IVDD in aged zebrafish and in low-BMD osteoporotic fish, demonstrating that low BMD per se does not act as a protective factor in IVDD. Young zebrafish displaying high bone density show an association with IVDD, suggesting a U-shaped model to explain the association between BMD and IVDD As increased vertebral body BMD was linked to disc degeneration 23 , we aimed to investigate young fish exhibiting high BMD. Cathepsin K is a protease expressed by osteoclasts and is important for bone resorption. Mutations in CTSK cause pycnodysostosis, a disease characterized by osteosclerosis (bone hardening and high BMD) 61 .CTSK is associated with estimated BMD and lumbar spine area (mskkp.org). Furthermore, genes associated with osteoclast function are associated with BMD and changes in aging 47 . CRISPR technology in zebrafish has been shown to be highly efficient, such that G0 skeletons display phenotypes consistent with those of homozygous mutants 62 .We used CRISPR/Cas9 to generate ctsk mosaics (crispants, crps) showing high BMD throughout the vertebral column. We analyzed 1-year-old ctsk crp by radiographs (n=25), μCt (n=8), and histological sections (n=3). We detected severe endplate sclerosis, osteophytosis and IVD calcification in 70% of injected fish, mild in 15% of injected fish and no IVD changes in 15% of injected fish. By calculating centra TMD in ctsk crps, we confirmed the significant increase in TMD in affected vertebrae (sclerosis) of crispants (G0s), from 0.6 to 0.75 g·cm −3 HA (P=0.007 7) (Fig. 7). Automated centra segmentation in comparison to aging wt showed the highest density in ctsk crps (28 centra), with a standard deviation similar to that of 3-year-old wt fish (Fig. 7c, d). We used radiographs to calculate the correlation between X-ray attenuation (pixel intensity, representative of BMD) and morphology of the vertebral column at the discs (wt =13, wt standard (the same reference wt fish added to all X-rays) =9, ctsk crp =25). Remarkably, higher bone mineral densities were systematically associated with IVD deformities (Pearson correlation coefficient = −0.85, P=7.8e −14 ) (Fig. 7e). An inverse correlation was observed between density and morphology; thus, higher density was positively associated with abnormal disc morphology, suggesting a U-shaped model linking BMD and IVDD (Fig. 7f). Ctsk crps subjected to vertebral compression displayed significant resistance to the applied force, with failure points at 30 N (P=0.032) (Fig. 7g) and reduced Young’s modulus (elasticity) (Supplementary Fig. 9A). When we sectioned ctsk crps, we revealed IVDD, fibrous NP and disrupted AF (Supplementary Fig. 9B), and the lack of the elastin layer in discs exhibiting abnormal morphology (Fig. 7h). Picrosirius red staining revealed a spectrum of colors associated with an increase in thinner fibers in ctsk crps (P=0.017 5) (Fig. 7i), signifying altered bone quality through changes in collagen fiber organization. By analyzing young fish with high BMD, we detected endplate deformities and disc calcification associated with IVDD. Therefore, our data suggest that the association between BMD and IVDD best fits a U-shaped model in which both increased and decreased BMD can exacerbate IVDD (Fig. 7g). DISCUSSION IVDD is the most common cause of back pain, a leading cause of disability worldwide and a global concern as the population ages 1,3 . Understanding the relationship between IVDD and agingrelated degenerative diseases would support drug repurposing and the development of less invasive therapeutics. Here, we showed evidence of osteoporosis in aged zebrafish. We characterized zebrafish IVDD in 3D and shed light on the relationship between low BMD and IVDD and between high BMD and IVDD. We provided evidence that disproved osteoporosis is a protective factor in IVDD and showed a positive correlation between high BMD and IVDD. Our findings suggest a U-shaped model to explain the relationship between BMD and IVDD. Genetic factors regulating bone formation, resorption and bone quality are candidates for disc disease and potential targets for therapeutics. We characterized zebrafish disc degeneration in 3D, focusing on bone, the NP, and the AF. This resulted in a more holistic appreciation of phenotypes not often analyzed together, highlighting the potential of zebrafish as a spontaneous model to recapitulate human IVDD. Aged zebrafish displayed bone morphological abnormalities comparable to those of mammals: bone sclerosis, osteophytosis, disc narrowing, misalignments and disc calcification. Cellular transformations within the NP were followed by 3D modifications in the architecture of the discs, with fibrosis Impact of bone density in disc degeneration E Kague et al. 8 Bone Research (2021) 9:39 and disc herniation, recapitulating chronic cases of human IVDD. Several animal models and mechanisms to replicate IVDD have been proposed, but due to IVDD complexity, all forms of replication have limitations. The sand rat and the chondrodystrophoid dog are the most studied species in which spontaneous IVDD occurs 63 . The mechanical load of discs in quadrupedal mammals differs from that in humans, with gravitational loading occurring perpendicular to the vertebral a b Young Agedsp7 -/- Toluidine blue Pan-Keratin Anti-cadherin cadherin DAPI wt E. contrast µCT sp7 -/- c NP NP NP AF AF AF Centrum IVD ncnens se st ev el dccffbobbo Fig. 5 IVDD histopathology underlying 3D disc changes in zebrafish. aSchematic of the vertebral segments of the zebrafish. IVDD histopathology underlying 3D disc changes in zebrafish. The centrum and IVD regions are annotated. bo, bone; cf, collagen type I fiber bundles; dc, dense collagen type I matrix; el, elastin; ev, extracellular vacuole; fb, fibroblasts; nc, notochord cells, vacuolated cells and notochord epithelial cells attached to the notochord sheath; ne, notochord epithelium; ns, notochord sheath; ob, osteoblasts; se, septum; st, notochord strand. bToluidine blue, pankeratin and pan cadherin immunostaining of young (1 y), aged (3 y), and sp7 −/− (1 y) discs. Toluidine blue of a middle section of the disc showing the NP in young, aged, and sp7 −/− discs (n=3 per group, serial sections were analyzed). Note the fibrous and disorganized NP in aged and sp7 −/− (magenta arrows) discs. Higher magnification, with a focus on the AF, is shown on the right. Note collagen bundle fibers disrupted in the aged AF and loose organization in the sp7 −/− disc (magenta arrows). Pankeratin showing bone (yellow), keratin (orange), and glycosaminoglycans (green). Note the accumulation of keratin in aged discs and dehydration (weak stain for glycosaminoglycans) in aged and sp7 −/− discs. Immunostaining for pan cadherin showing the loss of cellularity in aged discs (arrow) and disorganized NPs in sp7 −/− discs, with vacuolated cells (pink arrowhead), acellular regions (pink arrow), and cellular agglomerates (pink dashed arrow) (n=3 per group, ≥3 sections analyzed). Scale bars =50 μm. c3D volumetric rendering from enhanced (E) contrast μCT images of wt (1 y) and sp7 −/− (1 y) zebrafish. Note fibrous tissue organization in sp7 −/− zebrafish. Arrows point to an abnormal notochord sheath layer. Scale bars =50 μm Impact of bone density in disc degeneration E Kague et al. 9 Bone Research (2021) 9:39 52. Currey, J. D. The relationship between the stiffness and the mineral content of bone. J. Biomech. 2, 477–480 (1969). 53. Mammone, J. F. & Hudson, S. M. Micromechanics of bone strength and fracture. J. Biomech. 26, 439–446 (1993). 54. Cotti, S. et al. More bone with less minerals? The effects of dietary phosphorus on the post-cranial skeleton in zebrafish. Int. J. Mol. Sci. 21, 5429 (2020). 55. Milovanovic, P. & Busse, B. Phenomenon of osteocyte lacunar mineralization: indicator of former osteocyte death and a novel marker of impaired bone quality? Endocr. Connect 9, R70–R80 (2020). 56. Rolvien, T. et al. Long-term immobilization in elderly females causes a specific pattern of cortical bone and osteocyte deterioration different from postmenopausal osteoporosis. J. Bone Min. Res. 35, 1343–1351 (2020). 57. Hemmatian, H., Bakker, A. D., Klein-Nulend, J. & van Lenthe, G. H. Aging Osteocytes, and mechanotransduction. Curr. Osteoporos. Rep. 15,401–411 (2017). 58. Rochefort, G. Y. The osteocyte as a therapeutic target in the treatment of osteoporosis. Ther. Adv. Musculoskelet. Dis. 6,79–91 (2014). 59. Ofer, L., Zaslansky, P. & Shahar, R. A comparison of the structure, composition and mechanical properties of anosteocytic vertebrae of medaka (O. latipes) and osteocytic vertebrae of zebrafish (D. rerio). J. Fish Biol. 98, 995–1006 (2021). 60. Bredfeldt, J. S. et al. Computational segmentation of collagen fibers from secondharmonic generation images of breast cancer. J. Biomed. Opt. 19, 16007 (2014). 61. Gelb, B. D., Shi, G. P., Chapman, H. A. & Desnick, R. J. Pycnodysostosis, a lysosomal disease caused by cathepsin K deficiency. Science 273, 1236–1238 (1996). 62. Watson, C. J. et al. Phenomics-based quantification of CRISPR-induced mosaicism in zebrafish. Cell Syst. 10, 275–286.e5 (2020). 63. Daly, C., Ghosh, P., Jenkin, G., Oehme, D. & Goldschlager, T. A review of animal models of intervertebral disc degeneration: pathophysiology, regeneration, and translation to the clinic. Biomed. Res. Int. 2016, 5952165 (2016). 64. Wang, F. et al. Formation, function, and exhaustion of notochordal cytoplasmic vacuoles within intervertebral disc: current understanding and speculation. Oncotarget 8, 57800–57812 (2017). 65. Risbud, M. V. & Shapiro, I. M. Notochordal cells in the adult intervertebral disc: new perspectiveonanoldquestion.Crit. Rev. Eukaryot. Gene Expr. 21,29–41 (2011). 66. Chanchairujira, K. et al. Intervertebral disk calcification of the spine in an elderly population: radiographic prevalence, location, and distribution and correlation with spinal degeneration. Radiology 230, 499–503 (2004). 67. Novais, E. J. et al. Comparison of inbred mouse strains shows diverse phenotypic outcomes of intervertebral disc aging. Aging Cell 19, e13148 (2020). 68. Wickrematilake, G. W. Calcium pyrophosphate dihydrate deposition disease in young patients: two case reports. Arch. Rheumatol. 32,80–83 (2017). 69. Huxeley, T. H. Observations on the development of some parts of the skeleton of fishes. Q. J. Microsc. Sci. (continued J. Cell Sci.) 7,33–46 (1859). 70. Lleras Forero, L. et al. Segmentation of the zebrafish axial skeleton relies on notochord sheath cells and not on the segmentation clock. Elife 7, e33843 (2018). 71. Pogoda, H. M. et al. Direct activation of chordoblasts by retinoic acid is required for segmented centra mineralization during zebrafish spine development. Development 145, dev159418 (2018). 72. Kaiser, J. et al. Heterogeneity and spatial distribution of intravertebral trabecular bone mineral density in the lumbar spine is associated with prevalent vertebral fracture. J. Bone Min. Res. 35, 641–648 (2020). 73. Wang, Y. X. J. Senile osteoporosis is associated with disc degeneration. Quant. Imaging Med. Surg. 8, 551–556 (2018). 74. Khajuria, D. K. & Karasik, D. Novel model of restricted mobility induced osteopenia in zebrafish. J. Fish Biol. 98, 1031–1038 (2020). 75. Harada, A., Okuizumi, H., Miyagi, N. & Genda, E. Correlation between bone mineral density and intervertebral disc degeneration. Spine (Philos. Pa 1976) 23, 857–861 (1998). discussion 862. 76. Dequeker, J., Aerssens, J. & Luyten, F. P. Osteoarthritis and osteoporosis: clinical and research evidence of inverse relationship. Aging Clin. Exp. Res. 15,426–439 (2003). 77. Weintroub, S. et al. Osteoarthritis of the hip and fracture of the proximal end of the femur. Acta Orthop. Scand. 53, 261–264 (1982). 78. Marcelli, C. et al. The relationship between osteoarthritis of the hands, bone mineral density, and osteoporotic fractures in elderly women. Osteoporos. Int.5, 382–388 (1995). 79. Lee, S. et al. Association between low-back pain and lumbar spine bone density: a population-based cross-sectional study. J. Neurosurg. Spine 19, 307–313 (2013). 80. Lindstrom, E. et al. Nonclinical and clinical pharmacological characterization of the potent and selective cathepsin K inhibitor MIV-711. J. Transl. Med. 16, 125 (2018). 81. Reid, I. R. Short-term and long-term effects of osteoporosis therapies. Nat. Rev. Endocrinol. 11, 418–428 (2015). 82. Westerfield, M. The Zebrafish Book. A Guide for the Laboratory Use of Zebrafish (Danio rerio) (Univ. of Oregon Press, 2000). 83. Wienholds, E. et al. Efficient target-selected mutagenesis in zebrafish. Genome Res. 13, 2700–2707 (2003). 84. DeLaurier, A. et al. Zebrafish sp7:EGFP: a transgenic for studying otic vesicle formation, skeletogenesis, and bone regeneration. Genesis 48, 505–511 (2010). 85. Brunt, L. H., Begg, K., Kague, E., Cross, S. & Hammond, C. L. Wnt signalling controls the response to mechanical loading during zebrafish joint development. Development 144, 2798–2809 (2017). 86. Carrington, B., Varshney, G. K., Burgess, S. M. & Sood, R. CRISPR-STAT: an easy and reliable PCR-based method to evaluate target-specific sgRNA activity. Nucleic Acids Res.43, e157 (2015). 87. Shah, A. N., Davey, C. F., Whitebirch, A. C., Miller, A. C. & Moens, C. B. Rapid reverse genetic screening using CRISPR in zebrafish. Nat. Methods 12, 535–540 (2015). 88. De Clercq, A. et al. The external phenotype-skeleton link in post-hatch farmed Chinook salmon (Oncorhynchus tshawytscha). J. Fish. Dis. 41, 511–527 (2018). 89. Bensimon-Brito, A. et al. Revisiting in vivo staining with alizarin red S-a valuable approach to analyse zebrafish skeletal mineralization during development and regeneration. BMC Dev. Biol. 16, 2 (2016). 90. Kellgren, J. H. & Lawrence, J. S. Radiological assessment of osteo-arthrosis. Ann. Rheum. Dis. 16, 494–502 (1957). 91. Bouxsein, M. L. et al. Guidelines for assessment of bone microstructure in rodents using micro-computed tomography. J. Bone Min. Res. 25, 1468–1486 (2010). 92. Kague, E. et al. Scleraxis genes are required for normal musculoskeletal development and for rib growth and mineralization in zebrafish. FASEB J. 33, 9116–9130 (2019). 93. Schindelin, J. et al. Fiji: an open-source platform for biological-image analysis. Nat. Methods 9, 676–682 (2012). 94. Ronneberger, O. F. P. & Brox, T. U-Net: convolutional networks for biomedical image segmentation. Lecture Notes in Computer Science 9351,1–8 (2015). 95. Rueden, C. T. et al. ImageJ2: ImageJ for the next generation of scientific image data. BMC Bioinformatics 18, 529 (2017). 96. Arganda-Carreras, I. et al. Trainable_Segmentation: Release v3.1.2.Zenodo https:// doi.org/10.5281/zenodo.59290 (2016). 97. Cross, S. MIA v0.14.13. Zenodo https://doi.org/10.5281/zenodo.3976622 (2020). 98. Legland, D., Arganda-Carreras, I. & Andrey, P. MorphoLibJ: integrated library and plugins for mathematical morphology with ImageJ. Bioinformatics 32, 3532–3534 (2016). 99. Chollet, F. Keras. GitHub Repository (2015). 100. Abadi, A. et al. Tensorflow: Large-scale machine learning on heterogeneous systems. Preprint at https://arxiv.org/abs/1603.04467 (2015). 101. Doube, M. et al. BoneJ: free and extensible bone image analysis in ImageJ. Bone 47, 1076–1079 (2010). 102. Rohlf, F. J. TpsDig2. Digitise landmarks and outlines version 2.17. Department of Ecology and Evolution, State University of New York (2013). 103. Haines, A. J. & Crampton, J. S. Improvements to the method of Fourier shape analysis as applied in morphometric studies. Paleontology 43, 765–783 (2000). 104. Hammer, O., Harper, D. A. T. & Ryan, P. D. Paleontological statistics software package for education and data analysis. Paleontologia Electron. 4, 9 (2001). 105. Chen, X., Nadiarynkh, O., Plotnikov, S. & Campagnola, P. J. Second harmonic generation microscopy for quantitative analysis of collagen fibrillar structure. Nat. Protoc. 7, 654–669 (2012). 106. Schneider, C. A., Rasband, W. S. & Eliceiri, K. W. NIH image to ImageJ: 25 years of image analysis. Nat. Methods 9, 671–675 (2012). Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons. org/licenses/by/4.0/. © The Author(s) 2021 Impact of bone density in disc degeneration E Kague et al. 16 Bone Research (2021) 9:39