Zebrafish as an Emerging Model for Osteoporosis: A Primary Testing Platform for Screening New Osteo-Active Compounds

1Osteoporosis (OP) is a degenerative bone disease that affects around 27.6 million people over the age of 50 in the 27 European Union (EU27) countries alone. As average life expectancies increase, it is predicted that the annual cost of treating OP in the EU will rise from €37 billion in 2010 to €46.5 billion by 2025. OP is characterized by a reduction in bone mineral density (BMD), reduction of bone mass (BM), and a decrease in the trabecular volume of long bones; resulting in brittle bones that are more prone to fracture. The underlying mechanism behind OP is a dysregulation of bone homeostasis; with decreased bone anabolism (decreased activity of osteoblasts and osteocytes) and increased catabolism (enhanced osteoclast activity). Successful treatment of OP should therefore increase bone anabolism and decrease catabolism to reinstate the equilibrium in bone homeostasis. While therapeutic options are increasing, all but one available therapies aim to reduce bone resorption. However, as osteoclast and osteoblast activity are coupled, anti-resorptives can negatively affect anabolic osteoblast activity and may not fully restore bone architecture. The only injectable osteoanabolic compound, teriparatide, is an analog of the parathyroid hormone. However, it is not an ideal long-term therapy option as, not only is it expensive, long term exposures in rat increase susceptibility to osteosarcoma limiting treatment duration (currently 2-years) in OP patients. Thus, an ideal treatment plan should focus on both strengthening bones using an osteoanabolic compound, combined with use of an anti-resorptive treatment (also ideally non-invasive) to maintain bone integrity, few such options exist. Currently, a major bottleneck in the development of new pharmaceuticals is the collection of primary functional data on new biological drug targets with osteo-anabolic capacities.

2The twinning of genetic information with mechanistic data is key for development of new treatments. For example, familial studies on high bone mass (HBM) cases led to the discovery of mutations in SOST (Sclerostin). Further mechanistic data generated in model systems showed that SOST acts negatively on the WNT signaling pathway and led to the development of a novel antibody treatment Romosozumab (approved in 2018 for clinical use), which blocks SOST activity. With the advent of genome-wide association studies (GWAS), and efficient whole-genome/exome sequencing (WGS/WES) data mapping there has been a sizeable increase in availability of human genetic data from cohort studies for musculoskeletal conditions including OP, high bone mass (HBM), and osteoarthritis (OA). Recent large cohort studies, such as UK-Biobank, have identified many new loci that contain novel osteogenic factors. For example, the UK-Biobank data yielded 518 loci associated with changes in BMD using heel ultrasound data. Currently, there is a substantial gap in translating these human genetic findings to model systems in which the mechanism by which these genes act on the skeleton can be defined, where hypotheses can be tested, and ultimately define new putative drug targets that can be assessed with pharmacological agents. Because the skeletal system involves complex interactions between different cell and tissue types, genes and mechanical stimuli it is difficult to recapitulate features of OP in a petri dish. However, traditional rodent models are expensive to genetically manipulate. Zebrafish (Danio rerio) could therefore bridge this gap by offering fast genetic manipulation and complex tissue interactions required to model complex diseases such as OP.

3Zebrafish are vertebrates and show strong similarities in their skeletal physiology to mammals. They are highly fecund and a single pair of fish can lay up to 300 eggs a week, which develop externally and are translucent. They show conservation of 70% of all genes and 85% of disease genes with humans. However, the main advantage of zebrafish for functional genetic studies is their genetic tractability, as constructs that modify the genome can be injected directly into embryos at the single cell stage. This has allowed the generation of transgenic lines that allow dynamic imaging of all the cells of the developing skeletal system in live larvae (Table 1) and in more recent years allowed genome editing strategies to be employed. In this review we set-out these different approaches and how developing and adult zebrafish can be used to study bone mineralization, bone content formation, and osteoblast-osteoclast interactions in a whole animal context. We also discuss future prospects for drug screening pipelines in zebrafish which may confer advantages over other pre-clinical model systems.

4Zebrafish are genetically high amenable and new ways to manipulate the genome are constantly being added to the zebrafish genetic toolbox, which includes knockout, knock-down and, DNA insertion strategies. The external development of the embryos allows tools targeting genes of interest to be microinjected directly in embryos at the 1-cell stage and hundreds of embryos can readily be injected in a morning. Acute knockdown of gene expression can be achieved either by targeting mRNA with antisense RNA morpholino (MO) molecules that stably bind the target mRNA to block translation or splicing through steric hindrance. MOs offer a rapid method to assess the phenotype of a gene of interest during early development. However, they can only be used to study developmental processes occurring over the first 4 or 5 days of development, which limits their utility in skeletal studies as mineralization occurs from 4 days of development. While concerns have been raised about MO veracity as morphants frequently show more severe phenotypes than stables mutants generated for the same gene. This is due to a transcriptional compensation response for chronic loss of a gene as has been shown in mouse, cultured human cell lines, plants, and zebrafish models. Thus, while MOs have a role, their use has been largely supplanted by use of genome editing strategies.

5Traditionally, zebrafish mutant lines have been generated by forward genetic screening; using mutagens [e.g., N-ethyl-N-nitroso urea (ENU)] to induce random point mutations in offspring that were then screened for phenotypes of interest. The expansion of the zebrafish genetic toolkit with zinc-finger nucleases (ZFN), Transcription Activator-Like Effector Nucleases (TALEN), and Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR)/Cas9 reverse genetic strategies, which, in combination with a fully sequenced genome, allow tailored gene-specific mutagenesis in the zebrafish. Gene function can be studied in genetic knockouts by generating insertion/deletion (indel) mutations leading to premature stop codons, deleting whole exons containing important protein domains and generate new stable mutant lines (Figure 1A). Moreover, the CRISPR/Cas9 protocol is so efficient that the F0 injected fish (crispants) can be used to study loss of gene function in these crispants, despite them carrying mosaic mutations (i.e., not every cell carries a mutation and more than one mutation may be present) (Figure 1B). Single base gene editing (knock ins) using modified Cas9 enzymes or supplying a DNA template for the endogenous homologous recombination machinery initiated by a double stranded break allows to introduce specific genetic changes to model specific human disease mutations in zebrafish orthologs.

6Zebrafish in common with higher vertebrates, have both dermal/intramembranous ossification, in which bone is formed de novo directly by osteoblasts, and chondral/endochondral ossification in which bone forms by progressively replacing a cartilaginous template. Although zebrafish have thinner bones than terrestrial vertebrates, with fewer embedded osteocytes and little trabeculation, all of the relevant skeletal cell types and modes of regulation are conserved between zebrafish and higher vertebrates. This, importantly for the study of OP, includes osteoblast and osteoclast coupling and regulation of bone remodeling.

7A major advantage of using zebrafish to probe the mechanism of bone homeostasis is that cell behavior can be visualized dynamically in vivo. Zebrafish larvae are translucent and develop rapidly, and early skeletal processes can be dynamically visualized in the living fish through use of fluorescent transgenic reporter lines marking these cell types (see Table 1 for examples). Formation of the craniofacial skeleton occurs early, with the first cartilaginous structures of the jaw forming by 2 days post fertilization (dpf), the first skeletal joints are formed and mobile by 3 dpf, by 5 dpf, hypertrophic chondrocytes, marked by col10a1a, are seen in some elements from 5 dpf, and first osteoblasts surrounding the cartilage and forming bone matrix by 7 dpf. The first intramembranous bones, such as the cleithrum, anterior notochord, and operculum, are visible in the craniofacial skeleton from 72 hpf. While skeletal development occurs early, true remodeling through the combined activity of osteoblasts and osteoclasts does not commence until the second week of development as osteoclasts (marked by Cathepsin-K (Ctsk) or TRAP) are not formed until day 10–12. Unlike mammals, mononucleated osteoclasts as well as multinucleated cells are present and actively resorb bone.

8There are many transgenic lines available to mark musculoskeletal tissues, these include reporter lines which label cells or signaling pathway activation by driving expression of proteins in the cytoplasm, targeted to the nucleus, or plasma membrane, and lines that tag proteins. Reporter lines mark cell types by using a tissue specific promotor, responsive elements from a signaling pathway, or transcription factor binding sites controlling expression of a fluorescent protein (Table 1). For example, to study bone homeostasis, osteoblasts and osteoclasts can both be labeled in vivo, using osteoblast reporters such as sp7, and osteoclast reporters such as ctsk, so that their numbers, location and activity monitored in living bone tissue either longitudinally, in response to drug treatment, genetic mutation, or environmental stimuli (Table 1). Relevant to research into OP, osteoclasts can be specifically temporally activated by use of a heat shock promoter driving RANK ligand (rankl) expression; such that following a period of immersion in water at 39°C, osteoclast activity, labeled with the blue fluorescent protein (CFP), is increased resulting degradation of the bone matrix and in an osteoporotic phenotype of low BMD. A simple Alizarin Red S (AR) staining, which marks calcium phosphate crystals and fluoresces strongly in the red channel (580 nm wavelength), allows a rapid assessment of ossified elements in live or fixed fish. In combination with transgenic lines, endochondral ossification in the lower jaw (Figure 2A) and intramembranous bone formation in the operculum (Figure 2B) can be easily visualized compared to traditional rodent models.

9The zebrafish adult skeleton is relatively complex and once fully formed by around 2 months is composed of 74 ossified cranial elements (compared with 22 in humans), 28–31 vertebrae; 4 cervical, 10–11 thoracic vertebrae, and 15–16 separated vertebrae in the tail region and fins (pectoral, dorsal, anal (ventral), and caudal). As in larvae, live AR and Calcein staining, or use of transgenic lines, allows easy detection of superficially located calcified elements in the skull, elasmoid scales, and fins using a simple fluorescent microscope. Deeper tissues can be imaged by multiphoton microscopy in small juveniles. However, bones located more internally (e.g., vertebrae and ribs) in large adults are difficult to visualize using this method. Post-mortem staining of bone (AR) and cartilage [Alcian blue (AB)] is a cost-effective way to analyse these structures for adult skeletal abnormalities (Figure 3A) and has been used in forward genetic screens to obtain detailed skeletal morphology information.

10Recent advances in X-ray based imaging: radiographs, micro-computed tomography (μCT), and synchrotron equipped μCT technologies (SR-μCT), and their subsequent downstream imaging processing, opened avenues to assess the adult zebrafish skeleton. The major advantage of using these X-ray imaging techniques is that they are non-destructive and can be used in the intact fish, allowing the samples to be used for other purposes, such as histology. Radiographs give two-dimensional (2D) images of the zebrafish skeleton at relatively low resolution (Figure 3B), permitting the visualization of bone elements and a broad evaluation of changes in the skeleton, radiographs can be used to image live anesthetized fish permitting longitudinal analysis of the skeleton over time. Higher resolution (2 μm voxel size) and three-dimensional (3D) assessment of the zebrafish skeleton can be achieved by using μCT (Figures 3C–E). As fish bone, like that of mammals, is composed of hydroxyapatite crystals, quantification of BMD can be performed by comparison to phantoms, which are samples of known hydroxyapatite content. Additionally, treatment with agents to improve contrast such as silver nitrate (AgNO3) or iodine, allow detection of juvenile (less dense) bone and of soft tissues such as muscle and cartilage.

11Very detailed data on bone micro-architecture can be achieved with SR-μCT. This technique can yield a spatial resolution of 100 nanometres on tissue samples and visualize fine bone structures at a cellular level including the vasculature in mineralized bone, osteoclast resorption pits, and osteocyte lacunae. As the size and resolution of data sets increase, the bottleneck in the process is frequently data analysis. Commercially available software packages such as “boneJ” are tailored for CT data analysis, and recently open source user friendly software have become available to process μCT data from zebrafish scans. For example, the “FishCut” software processes whole-body μCT scan datasets and applies semi-automated analysis algorithms. The current version segments the axial skeleton, then generates values for the surface area of vertebrae and centrae, and calculates BMD and mineralized thickness in a semi-automated fashion.

12An increasing number of zebrafish genetic mutants in skeletally relevant genes have been shown to recapitulate human bone disease. These have provided insight into the dynamic regulation of bone formation, mineralization, and remodeling. We have included a list of zebrafish skeletal mutants in Table 2. While there are currently few models for OP, there are various zebrafish mutant lines that accurately model human skeletal dysplasias, including collagenopathies and forms of osteogenesis imperfecta, which are characterized by brittle bones and frequent low-impact bone fractures. Autosomal dominant mutations COL1A1 and COL1A2 genes predominantly affect glycine-X-Y (Gly-X-Y) repeat domains that result in collagen α1(I) and α2(I) heterotrimer maturation defects, causing fragile bone matrix and insufficient mineralization. The Gly-X-Y mutations lead to impaired hydroxylation and defects in collagen maturation in the endoplasmic reticulum (ER), which is also conserved in zebrafish. The autosomal dominant chihuahua (chi) zebrafish mutant, was identified in a forward genetic screen using radiography. Linkage mapping identified a mutation causing a glycine to aspartate amino acid substitution in a conserved Gly-X-Y repeat of col1a1a (zebrafish col1a1 is duplicated). Note that in contrast to mammals, zebrafish type-I collagen is constituted by three different α chains [α1 (col1a1a), α3 (col1a1b), α2 (col1a2)] due to duplication. chi/+ zebrafish display phenotypes resembling those seen in humans, including a shortened axial skeleton, with irregular radiodensity, uneven mineralization, and brittle bones that fracture easily (especially ribs). Transmission electron microscopy revealed that chi/+ fish show signs of ER stress. The ER trapping of insufficiently hydroxylated oligotrimerized α1(I)/α2(I)/α3(I) collagen leads to lower extra-cellular collagen maturity, abnormally shaped and thinner vertebrae bodies, areas of higher calcium content, different local mechanical properties, and reduced osteocyte number. Osteogenesis imperfecta has a broad disease spectrum in the clinic, and recent comparative studies of multiple mutant alleles for col1a1a, col1a1b, col1a2, and also bmp1a (described later) and plod2 described a diversity of skeletal phenotypes (Table 2) with brittle bones as the common feature.

13The zebrafish sp7/osterix mutant has been shown to model human osteogenesis imperfecta caused by recessive damaging mutations in SP7. This mutant showed uneven mineralization, severe fractures caused by minimal impact, and misshapen bones. Moreover, rare craniofacial characteristics caused by impaired SP7 function, such as wormian bones, reported in human patients carrying mutations in SP7 were also observed in zebrafish.

14Another example of a zebrafish mutant that recapitulates patient phenotype is the bmp1a mutant frilly fins (frf). In humans a damaging missense mutation in the BMP1 signal peptide causes brittle bones in an osteogenesis imperfecta pedigree. frf mutants showed normal osteoblast number, but pericellular pro-collagen processing (C-pro-peptide removal) defect leading to mineralization defects in the axial skeleton and fin rays.

15Collagenopathies, such as Stickler Syndrome, have also been successfully modeled in zebrafish. We have recently reported a col11a2 zebrafish mutant showing specific traits of the human disease which include thicker collagen fibers and degradation of type-II collagen in zebrafish larvae leading to compromised jaw shape, mechanical properties and movement of the jaw leading to premature OA. In many skeletal dysplasias zebrafish not only model the human condition but allow mechanistic insight into how genetic changes lead to the cellular changes that underpin the disease symptoms. As such zebrafish offer exciting prospects for delivering functional studies in new osteoporotic genetic loci.

16Zebrafish are capable of regeneration many tissues and organs including the heart, lens, and pancreas. They also show regeneration of skeletal tissues following amputation of the tail fin (lepidotrichia) or removal of elasmoid scales. As the fins and scales are translucent, and readily imaged they allow cells and their calcified matrix to be visualized in detail using standard fluorescent microscopes (Figure 4A). After amputation of a ray fin (typically a caudal fin), a wound healing response results in the formation of an epimorphic blastema which regenerates all affected tissues of the amputated organ, including bone, in a controlled fashion. Following this inflammation response, osteoblasts undergo dedifferentiation and proliferate to contribute to the blastema. These juvenile osteoblasts then secrete matrix with intermediate properties between cartilage and bone and are later remodeled as mature bone by matured osteoblasts and recruited osteoclasts (Figure 4B). These fins can also be injured via cryo-injury by placing a −196°C knife perpendicularly to the caudal fin rays allowing to study the bone resorption response. These techniques offer great perspectives to compare bone formation and bone remodeling, in an adult context.

17Using transgenic lines and in vivo staining methods, such as AR (fluoresces red, 545 nm excitation, 580 nm emission) and Calcein (fluoresces green, 495 nm excitation, 515 nm emission), which binds to calcified matrix, the dynamics of bone formation can be visualized by using a fluorescent stereomicroscope in a regenerating caudal fin of a living fish. This allows longitudinal analysis by following regeneration rate and volume, since AR stains fully mineralized bone and Calcein binds to newly deposited bone matrix (Figure 4C).

18The utility of fin regeneration assays to test bioactive compounds has been demonstrated by treating regenerating fins with the glucocorticoid prednisolone. Following treatment bone formation was reduced, and furthermore, both osteoblast number and subsequent bone deposition and osteoclast recruitment was reduced in these fins. Interestingly, skull injury repair is less affected following prednisolone treatment, this is similar to mammals. Treatment of fins with Botulinum toxin (Botox) leads to a reduction in bone mineralization and regeneration following amputation, comparable to the situation in mammals where fracture repair is impaired following Botox induced paralysis.

19A major issue with OP is increased fracture risk due to weaker bone structure, and therefore identification of therapeutics that can improve fracture healing is desirable. Zebrafish show a fracture healing response, including callus formation (Figure 4D), with strong similarities to that of mammals. Fractures can be induced in zebrafish fins using simple pressure applied externally to the fin. As the fin has around 300 bony rays, multiple fractures can be induced in a single fin. A fracture callus is formed and de novo bone formation is initiated 2 days post-injury accompanied by an increased expression of osteoblast genes such as runx2 and sp7/osx. As the fin is flat, the fracture repair process can be dynamically tracked at cellular resolution using transgenic lines (Table 1) or by labeling bone formation with AR and Calcein (Figure 4D). As for regeneration, it is possible to add pharmacological agents to the regenerating tissue, allowing potential osteoanabolic compounds to be tested for beneficial effects in fracture repair in vivo.

20The body of zebrafish is covered with elasmoid scales made of calcified dermal bone harboring osteoblasts and osteoclasts (Figures 5A–C). The calcified matrix is composed of a plywood structure of collagen fibrils, which are easily visualized with second harmonics generation microscopy (Figure 5D). Scales are embedded in, and grow from, the dermis and shed and replace naturally throughout life of the fish. As scales are part of the exoskeleton they are easy to collect from an anesthetized fish. Each flat scale is subdivided in four regions by its morphology: anterior, lateral, central, and central with epidermis (Figure 5A). The anterior region is attached to the skin and does not grow or form new bone. The lateral area is characterized by its curved ridges (circuli), whereas the central area has linear trenches. Within the lateral circuli and central grooves newly mineralized matrix is formed by osteoblasts and degraded by osteoclasts. The posterior area has increased osteoblast number and bone is continuously deposited (Figures 5A,B). Osteoblasts in different regions of the scale express different markers of maturity. As the scale contains living cells, including nerve and vascular endothelial cells, their use offers an opportunity to study bone cell behavior in a mature context.

21As larvae are small and develop in water, it is possible to grow larvae in multi-well format with the addition of water-soluble compounds to their growth media for easy uptake. Zebrafish have been used extensively for high-throughput screening using larvae and now drugs are used in clinical studies that were first identified in zebrafish. A great example is the identification of the kinase inhibitor dorsomorphin (BMP type-1 receptor (BMP1R) antagonist) to treat lymphoma which was discovered in an early embryogenesis phenotype screen using 7,500 small-molecules. Another example used semi-automated imaging strategy of Calcein stained larvae exposed to a small-compound library identifying 6 catabolic and 2 anabolic compounds that alter notochord mineralization (Table 3). Thus, when fluorescent compounds are twinned with fluorescent reporters for osteoblasts (e.g., sp7:gfp with AR) (Figure 2B), it will allow assessment of osteoblast number and activity in a semi-high content setting using plate imaging microscopy. When these assays are combined with high efficiency CRISPR/Cas9 genome engineering strategies, it will open avenues to test compounds of interest that could alter disease causing mutations deteriorating effects. Thus, this comprehensive approach will also offer opportunities to develop compounds for personalized medicine. For OP research it may be more advantageous to focus on adult skeletal assays to allow assessment of osteoclast activity (bone catabolism) simultaneously with an assessment of osteoblasts (bone anabolism). An example of pharmaco-genetics improving brittle bones, is when type-I collagen secretion in the bone matrix is ameliorated by treating chi/+ mutants with 4 phenyl butyrate (4PBA) compound.

22Zebrafish elasmoid scales are bony plates that are small and contain bioactive osteoblasts and osteoclasts (Figures 5B,C). These therefore offer huge potential as a primary pharmacological screening tool for skeletal compounds. The scales can be cultured for 72-h post-harvesting during which they can arrayed in multi-well plates and exposed to pharmacological compounds. As the scale is thin, osteoblasts are accessible to osteogenic factors, and have been demonstrated to react in a dose dependent manner to BMP-2. To model an OP-like phenotype, individuals can easily be exposed to prednisolone / dexamethasone (glucocorticoid pathway), ferric ammonium citrate, or metabolically with a high fat or glucose diet, see also Table 3. In the context of glucocorticoid induced OP (GIOP), the bisphosphonate Alendronate reverses the effects of prednisolone on ex vivo cultured elasmoid scale bone, which showed a reduction in osteoclast activity (measured by TRAP) and an increase in bone anabolism (measured by alkaline phosphatase activity); the same response as in mammals. As fat metabolism has been implicated with OP, a small fatty acid derivative library was used on GIOP adult fish. Biochemical assays on scales derived from these fish showed that cannaboid receptor 2 binding anandamide and N-linoleoylethanolamine (N-LLEL) fatty acids drive osteogenesis by stimulating alkaline phosphatase (ALK) activity.

23A WNT-pathway compound library was tested to identify new osteo-anabolic compounds using an assay in which luciferase was expressed under control of the sp7 promoter allowing a quantitative readout of osteoblast activity (Figure 6). This screen identified three osteo-anabolic (Table 3) and 15 osteo-catabolic compounds from 85 trial compounds. This library contained five previously published compounds tested in vivo, and nine tested in vitro mammalian bone progenitor cell lines. Strikingly, this scale luciferase assay was able to reproduce the effect of all in vivo tested compounds and about half of all in vitro tested compounds. These studies demonstrate the exciting potential that scale assays represent for testing of skeletal compounds relevant to OP in a cost-effective manner.

24Recently, there has been a substantial expansion in the quantity of high-quality genetic data from large-scale human genomic and transcriptomic studies that contain potential osteo-anabolic factors. Here we describe a potential screening pipeline that makes use of the genetic tractability and imaging in zebrafish to offer a relatively low cost, high-throughput option compared to traditional in vitro and in vivo models (Figure 7).

25After identification of several candidate genes/drug targets from human genetic studies, the pipeline consists of two experimental arms that can be carried out simultaneously to generate primary pre-clinical data to validate the putative drug targets. Using genome editing, loss-of function studies can be performed in transgenic backgrounds to test the effect of the gene of interest on the developing skeleton or on mineralization, and simultaneously allowing safety testing for deleterious effects on other tissues or organs. For example, using CRISPR/Cas9 editing, it is possible to generate hundreds of mosaic zebrafish mutants within 3–4 weeks (includes the generation of the targeting reagents), which is difficult to achieve in other available systems, such as cultured chondrocytes and osteoblasts (differentiation of these takes multiple weeks). With CRISPR/Cas9 editing it is also feasible to study the specific human disease mutation in zebrafish, as long as it is in a conserved coding region. These fish can be grown to adulthood and germline mutations identified allowing more detailed studies on the mature skeleton to be performed (Figure 7).

26In addition to genetic studies, pharmacological assessment of the identified putative drug target can be performed. By using water-soluble compounds, or lipid soluble compounds dissolved in DMSO, screens in a multi-well format can be performed using ex vivo culture of elasmoid scales. As a single adult fish has around 200 scales, this assay allows testing of many compounds, including control compounds (e.g., osteo-anabolic (alendronate) and -catabolic (prednisolone), on scales harvested from a single individual, reducing intra-individual variation. Therefore, this technique offers a platform to generate a primary read-out of novel osteo-active compounds in the context of homeostasis in a mature tissue. Additionally, this ex vivo technique will reduce the number of (potentially harmful) compounds being exposed to living fish, therefore contributing to ethical refinement and reduction of experimental animal use, but also reducing associated costs. As this scale assay reduces the number of putative osteogenic compounds substantially, these positive compounds can be further validated (along with safety testing) on developing transgenic larvae. These larvae would be plated out at 3 larvae per well and the compounds added from 3 days of development, with high-content imaging used for preliminary assessment of the effects of each compound and more detailed analysis including dose response followed up for validated positive hits. Further downstream tests, such as fin regeneration or fracture assays, can further reduce the number of compounds as such that only high-confidence compounds will be assessed in tetrapod pre-clinical studies (Figure 7).

27If desired, the two experimental arms can be performed simultaneously, so that stable mutants are being generated during the compound testing phase. This opens the possibility to perform pharmacogenetic experiments in a relatively short time frame to validate the effects of putative drugs on specific disease mutations to see if they can “rescue” the disease phenotype (Figure 7). Together, zebrafish offer the potential in future to bridge the gap between human genetic hits, and fast functional validation.

28The zebrafish is a well-established increasingly used animal model for studying various diseases including (congenital) metabolic bone diseases. As zebrafish have historically been mainly used for its fast-embryonic development properties to better understand disease onset, zebrafish aging studies have only recently been conducted to model age-related diseases such as OA and OP. OP is an emerging field in zebrafish modeling and more research is needed to fully establish an OP-like phenotype as it was previously determined in its teleost cousin medaka. The advantageous properties as set-out in this review should be further exploited to benefit drug development for OP. Zebrafish show the appropriate response to increased mechanical loading, where the cellular (transcriptional) response initiates increased bone formation and mineralization in the loaded bone elements that are easily quantified. However, since zebrafish and mammalian bone morphology show some differences, a pharmacological assay should particularly focus on the complex tissue and osteoblast-osteoclast interactions that underpin OP pathology. As traditional rodent and in vitro co-culture both have limitations to pursue large-scale drug discovery in a genetic context, zebrafish can take the place as a primary testing platform and therefore opening avenues to work toward gene specific compound discovery that have been identified as risk factors in human genetic studies. After primary safety testing, these identified compounds can be further tested in mammalian OP models to determine the effect on BMD, bone strength, and trabeculation. Fully exploiting these opportunities by using zebrafish as a primary screening model will open exciting avenues to perform pharmacogenetics for OP on a larger scale.

29Zebrafish procedures were approved by the University of Bristol Animal Welfare and Ethical Review Body (AWERB) and performed in accordance with a UK Home Office project license.

30DB and EK generated the figures for the manuscript. DB, EK, and CH researched and drafted the manuscript together.

31The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.