More Bone with Less Minerals? The Effects of Dietary Phosphorus on the Post-Cranial Skeleton in Zebrafish

Dietary phosphorus (P) is essential for bone mineralisation in vertebrates. P deficiency can cause growth retardation, osteomalacia and bone deformities, both in teleosts and in mammals. Conversely, excess P supply can trigger soft tissue calcification and bone hypermineralisation. This study uses a wide range of complementary techniques (X-rays, histology, TEM, synchrotron X-ray tomographic microscopy, nanoindentation) to describe in detail the effects of dietary P on the zebrafish skeleton, after two months of administering three different diets: 0.5% (low P, LP), 1.0% (regular P, RP), and 1.5% (high P, HP) total P content. LP zebrafish display growth retardation and hypomineralised bones, albeit without deformities. LP zebrafish increase production of non-mineralised bone matrix, and osteoblasts have enlarged endoplasmic reticulum cisternae, indicative for increased collagen synthesis. The HP diet promotes growth, high mineralisation, and stiffness but causes vertebral centra fusions. Structure and arrangement of bone matrix collagen fibres are not influenced by dietary P in all three groups. In conclusion, low dietary P content stimulates the formation of non-mineralised bone without inducing malformations. This indicates that bone formation and mineralisation are uncoupled. In contrast, high dietary P content promotes mineralisation and vertebral body fusions. This new zebrafish model is a useful tool to understand the mechanisms underlying osteomalacia and abnormal mineralisation, due to underlying variations in dietary P levels.


Introduction
Phosphorus (P) is an essential element for a wide variety of biological processes. It plays a key role in cellular metabolism, cell signalling, and the composition of phospholipid membranes and nucleic acids. For all vertebrates, P is crucial for mineralisation of the skeleton, bone, dentin, enamel/enameloid and mineralised cartilage. Naturally our ideas about bone mineral metabolism are influenced by insights that we have obtained from the mammalian (human) model. Vertebrates must control plasma calcium (Ca) within very narrow limits and mammals involve their bone to maintain plasma Ca levels.

T0 Animals Display a Mineralised Vertebral Column without Any Anomaly
At the beginning of the experiment (T0), control specimens were analysed by Alizarin red S whole mount staining to establish the degree of vertebral column mineralisation and possible malformations. Bird and Mabee [40] serve as reference for normal skeletal development in zebrafish. In addition several publications define malformations of the zebrafish axial skeleton [41][42][43]. The absence of any of the described malformation is regarded as normal in this study. None of the analysed samples presented malformations such as vertebral centra fusion, vertebral centra compression, curled or supernumerary arches (Figure 1b). Histological analysis of sagittal sections further confirmed that vertebral bodies do not show malformations and that the notochord sheath, perinotochordal membrane bone, and arches are properly mineralised (Figure 1c).
A comparison of the standard length (SL) reveals that zebrafish treated with the low P (LP) diet are significantly smaller compared to the controls (regular P diet, RP) and high P (HP) diet treated animals after one month of treatment ( Figure 1a and Table 1). This difference becomes more evident after two months of treatment. In contrast, HP animals have significantly increased SL compared to LP at both timepoints and compared to RP after two months of treatment.  Animals treated with the low P diet (LP) are smaller than controls (RP) and high P diet (HP) treated animals, whereas HP zebrafish present a significantly increased standard length. Mann-Whitney test, *: p < 0.05; **: p < 0.01; ***: p < 0.001. (b) T0 zebrafish, prior the beginning of the experiment, stained with Alizarin red S shows normally developed vertebral column and forming vertebral bodies. No vertebral column malformations, nor vertebral body fusion or compression are present. Scale bar: 500 µm. (c) Notochord sheath, perinotochordal membranous bone and neural (na) and haemal (ha) arches are mineralised in T0 animals, as shown by Von Kossa/Van Gieson staining on sagittal sections. Vertebral bodies are normally shaped and spaced. High magnification panels show (c') vertebral endplates with osteoid (black arrowheads) and intervertebral ligament (white arrowhead, see Figure 6a for details), (c") haemal arch with non-mineralised collagen matrix (black arrowhead). Mineralised bone: brown (black arrows); pigment: black (white arrows), non-mineralised collagen matrix/osteoid: red (black arrowhead). Scale bar: 200 µm.

Mineralisation of Endoskeleton and Dermal Skeleton Is Arrested under Low P Conditions
After two months of treatment, X-rays of vertebral columns from LP individuals show reduced radiodensity compared to vertebral columns of RP and HP animals (Figure 2a). To better investigate the LP phenotype, bone mineralisation levels were evaluated on whole mount Alizarin red S stained specimens. In comparison to the control group (RP diet), LP animals display an overall low-mineralised endoskeleton, including vertebral body centra and neural and haemal arches and spines. After one and two months under low P conditions, the majority of the animals have non-mineralised vertebral body endplates and largely non-mineralised neural and haemal arches. Conversely, the HP diet shows enhanced mineralisation of vertebral body endplates, neural and haemal arches. These structures are high mineralised after one and two months of dietary treatment in all the HP animals analysed (Figure 2b-e, Supplementary Table S1). Compared to controls, LP and HP fish do not show a completely homogenous phenotype after one month of dietary treatment: a small percentage (8%, four fish out of 51) of LP animals shows fully mineralised vertebral body endplates and some HP individuals (19%, six fish out of 31) present non-mineralised endplates. However, all HP animals have fully mineralised vertebral body endplates after two months of HP diet (Supplementary Table S1). The histological analysis on non-demineralised sagittal sections of the vertebral column confirms these findings. Von Kossa staining for P allows the precise distinction between mineralised and non-mineralised bone (osteoid), comparable to the whole mount Alizarin red S staining for Ca. Both techniques show large amounts of non-mineralised collagen matrix at the rim of vertebral body endplates in LP animals. Moreover, sections stained with Von Kossa show that LP vertebral bodies are surrounded by large amounts of non-mineralised bone matrix, and vertebral centra bone trabeculae and arches present a similar phenotype (Figure 2f). In contrast, vertebral body endplates and arches in control animals (RP diet) have narrow osteoid layers, indicative for fast mineralisation. The HP animals display an extremely thin osteoid layer at the vertebral endplates and fully mineralised arches (Figure 2e). The extent of bone matrix mineralisation level coincides with the dietary P content (Supplementary Table S1).
Similar to vertebrae, mineralisation of the fin endoskeletal support elements (pterygiophores or radials) [40] is affected after one and two months by dietary P. LP animals display a low or intermediate extent of bone matrix mineralisation, RP animals display intermediate mineralisation levels and HP animals have fully mineralised radials (Figure 3 and Supplementary Table S1).
Dietary P affects also the mineralisation of the dermal skeleton as evident from the analysis of the lepidotrichia, which are paired fin ray segments that mineralise [44]. Depending on dietary P content, lepidotrichia show progressively increasing mineralisation levels in the dorsal, anal, and caudal fin. After two months of treatment, the LP animals show low or intermediate lepidotrichia mineralisation. In contrast, in RP and HP animals the fin ray segments are fully mineralised (Figure 3 and Supplementary  Table S2).

Dietary P Has No Effect on Vertebral Morphology but HP Animals Have Fused Vertebral Centra
Detection of vertebral column abnormalities was performed on whole mount Alizarin red S stained specimens. After one or two months of treatment, none of the dietary groups display bending of the vertebral column (kyphosis, scoliosis, or lordosis). Vertebral bodies present in general a normal shape and size (but see below, vertebral fusion in HP animals). On the contrary, deformities of neural and haemal spines are present in most LP animals (Figure 2b), suggesting that the non-mineralised collagen matrix is easily deformable by muscle contraction.
No deformities of vertebral body centra occur in LP and RP animals except a few cases of vertebral fusion and compression at both analysed timepoints ( Table 2). Zebrafish treated for two months with the HP diet present an increased frequency of vertebral body fusions. More than a quarter, 28%, of the analysed specimens present at least one vertebral centra fusion. The increased occurrence of fusions in HP zebrafish is statistically significant (p = 0.006). Interestingly, six out of nine animals that suffer from vertebral fusion have multiple fusions in the caudal region [40] ( Table 2 and Figure 2a). This suggests that high dietary P supply might promote vertebral body fusions.

High Dietary P Is Associated with a Higher Stiffness in the Vertebral Endplates
To assess if changes in dietary P influence the mechanical properties of the vertebral tissue, nanoindentation was performed ( Figure 4a). In the vertebral endplate regions, a significantly higher elastic modulus is noted in HP zebrafish with 18.48 ± 3.18 GPa compared to LP zebrafish with 12.77 ± 3.68 GPa (p = 0.004), and a trend towards higher elastic modulus in HP compared to RP zebrafish with 15.57 ± 4.30 GPa (p = 0.073). In the central region, the elastic modulus is similar in all dietary groups with 11.44 ± 3.05 GPa in LP, 13.47 ± 2.01 GPa in RP, and 14.11

LP Individuals Have Increased Bone Matrix Formation and Highly Active Osteoblasts
Synchrotron based X-ray tomographic microscopy scans of a representative, similar-sized, specimen from each of the dietary groups reveal microstructural differences in the vertebrae of treated animals which are difficult to assess on whole mount Alizarin red S stained specimens. Although the analysed vertebral bodies in the caudal region [40] of the vertebral column of LP, RP, and HP individuals have a similar length and height, the length of arches and spines varies, with a maximum value in the HP group and a minimum value in the LP group ( Figure 5 and Supplementary  Table S3). Instead, the total vertebral body bone volume, calculated as the volume of vertebral body centra plus haemal and neural arches and haemal and neural spines, strongly differs between the representatives of the three diet groups. The total bone volume reaches a maximum in the individual from the LP group and a minimum in the HP animal. The RP individual has an intermediate bone volume. A pronounced increase in non-mineralised bone matrix is observed in the LP specimen compared to the RP and HP animals ( Figure 5 and Supplementary Table S3). Extensive nonmineralised bone matrix is localised at both vertebral endplates, at the neural and the haemal arches and at the neural and haemal spines in the LP animal. Volume data confirms that the volume of the newly formed bone in the LP vertebra is larger compared to the HP vertebra. In the individual from the HP group, endplates, arches, and spines are completely mineralised but thinner than in RP vertebral bodies (

LP Individuals Have Increased Bone Matrix Formation and Highly Active Osteoblasts
Synchrotron based X-ray tomographic microscopy scans of a representative, similar-sized, specimen from each of the dietary groups reveal microstructural differences in the vertebrae of treated animals which are difficult to assess on whole mount Alizarin red S stained specimens. Although the analysed vertebral bodies in the caudal region [40] of the vertebral column of LP, RP, and HP individuals have a similar length and height, the length of arches and spines varies, with a maximum value in the HP group and a minimum value in the LP group ( Figure 5 and Supplementary  Table S3). Instead, the total vertebral body bone volume, calculated as the volume of vertebral body centra plus haemal and neural arches and haemal and neural spines, strongly differs between the representatives of the three diet groups. The total bone volume reaches a maximum in the individual from the LP group and a minimum in the HP animal. The RP individual has an intermediate bone volume. A pronounced increase in non-mineralised bone matrix is observed in the LP specimen compared to the RP and HP animals ( Figure 5 and Supplementary Table S3). Extensive non-mineralised bone matrix is localised at both vertebral endplates, at the neural and the haemal arches and at the neural and haemal spines in the LP animal. Volume data confirms that the volume of the newly formed bone in the LP vertebra is larger compared to the HP vertebra. In the individual from the HP group, endplates, arches, and spines are completely mineralised but thinner than in RP vertebral bodies (Figures 2e,f and 5).    Osteoblasts are active and present a high number of endoplasmic reticulum (ER) cisternae (black arrows), which are enlarged in low P diet (LP) treated animals (asterisks) compared to controls (RP) and high P diet (HP) treated animals, indicative of increased bone matrix production. ECM: extracellular matrix, N: nucleus. Scale bar: 1 µm. (c) Higher magnification of collagenous bone matrix located at the vertebral endplates. Collagen type I fibres in the immediate vicinity of osteoblasts (OB) have similar diameters among the three dietary groups, as well as collagen fibres located at a distance from the osteoblasts, within the extracellular matrix, indicative of unaltered fibre maturation. Black arrowheads: fibres in the vicinity of the osteoblasts with small diameters; white arrowheads: fibres at a distance from osteoblasts with large diameters. Scale bars: 200 nm.

Collagen Type I is Unaltered in All Dietary Groups
Electron microscopy was used to analyse collagen type I fibres in the bone growth zone of the vertebral endplates.

Discussion
This study describes the primary effects of low and high dietary phosphorus (P) content in juvenile zebrafish. P is a critical element for several biological processes, including hard tissue mineralisation. An extended period of dietary treatment with different P levels affects growth, bone formation, bone mineralisation and bone mechanical properties. In particular, low dietary P (LP) level causes growth retardation but also increases non-mineralised bone matrix production, as indicated by histology and synchrotron X-ray tomographic microscopy. Endoplasmic reticulum cisternae of osteoblasts are enlarged, indicative for increased collagen synthesis, in line with the observed increase of bone matrix production. Conversely, high dietary P (HP) level increases growth, bone

Discussion
This study describes the primary effects of low and high dietary phosphorus (P) content in juvenile zebrafish. P is a critical element for several biological processes, including hard tissue mineralisation. An extended period of dietary treatment with different P levels affects growth, bone formation, bone mineralisation and bone mechanical properties. In particular, low dietary P (LP) level causes growth retardation but also increases non-mineralised bone matrix production, as indicated by histology and synchrotron X-ray tomographic microscopy. Endoplasmic reticulum cisternae of osteoblasts are enlarged, indicative for increased collagen synthesis, in line with the observed increase of bone matrix production. Conversely, high dietary P (HP) level increases growth, bone mineralisation and stiffness and promotes vertebral body fusions. Collagen post-translational modification, structure and arrangement of collagen fibres in the bone matrix are not influenced by dietary P content.

Skeletal Mineralisation Arrest under Low P Conditions
The arrest of bone matrix mineralisation without the stop of bone matrix production is the primary effect of the LP diet on the skeleton of juvenile zebrafish. Structure and shape of the non-mineralised bone are normal. It can thus be defined as bone according to De Ricqlès et al. (1982): "mineralisation in bone can be missing alone or in combination with other components. Nevertheless, for reasons dealing with composition, homology, origin and function the tissue should be recognised as bone" [46]. The LP zebrafish model recapitulates the bone phenotype typical of P deficiency that is also observed in other vertebrates. Indeed, mammals, including humans and rats, and teleost fish under dietary P deficiency, show bones with reduced radiodensity and an increased amount of osteoid at the level of epiphyseal plates [15], endosteal bone [14], pharyngeal bone [13], and vertebral centra [10]. Staining for mineral detection (whole mount Alizarin red S for Ca, Von Kossa on sections for P) reveals the lack of bone mineralisation in LP animals. Endoskeletal elements, such as vertebral body centra, haemal and neural arches, and elements of the dermal skeleton, such as fin rays, are equally affected (Figures 2 and 3, Supplementary Tables S1 and S2). Given that growing juvenile zebrafish were exposed for two months to low dietary P content, the presence of hypomineralised bones was expected. Still, bones are not only hypomineralised, but new bone matrix is formed and this matrix has no minerals. Our findings match previous studies that analysed the consequences of dietary P deficiency in Atlantic salmon (Salmo salar) [10,12,20] and in Nile Tilapia (Oreochromis niloticus) [13]. Moreover, the observed phenotype resembles the bone phenotype detected in murine models of heritable hypophosphatemia [47][48][49], a disorder related to low P levels in the blood [19,50]. Likewise, patients suffering from rickets present hypomineralised bones [15].
Conversely, in animals in this study that received a diet with regular P content (RP), bone mineralisation was normal and in line with previous studies that traced zebrafish skeletal mineralisation [40]. All bone structures analysed in the RP group present a small amount of non-mineralised bone identifiable as osteoid. Bone elements from HP fish are even further mineralised, the osteoid is extremely narrow.
This study shows that a low P diet arrests mineral deposition equally in endo-and dermal skeletal elements. The mineralised endoskeleton evolved much later than the mineralised dermal skeleton [51]. The latter comprises teeth, scales and fin rays [35,52,53]. In the present study, the degree of dermal fin ray mineralisation coincides with the degree of vertebral body mineralisation (Figures 2 and 3, Supplementary Tables S1 and S2). Likewise, a zebrafish mutant strain called nob (no bone) that completely lacks bone mineralisation presents non-mineralised dermal and endoskeletal elements to the same extent [54]. That fin rays can serve as indicators to track skeleton mineralisation has immediate applications. It will allow monitoring the mineralisation status of the overall skeleton related to dietary P content in vivo, using vital mineral staining for fin rays [55]. Such a non-invasive method avoids animal sacrifice in the context of low or high dietary P treatment or in other experiments which trace skeletal mineralisation.

A Functional and Unaltered Axial Skeleton Despite Low Dietary P
In teleosts, particularly in farmed salmonids, dietary P deficiency has been linked to vertebral column malformations such as vertebral body compression and fusion [20][21][22][23][24]. In this study, low dietary P content for two months does not cause vertebral centra deformities in juvenile zebrafish. Despite the lack of bone mineralisation, vertebral centra have a normal shape without alterations that would foreshadow vertebral body compression or fusion [56,57]. Different from the unaltered centra, neural and haemal spines are twisted in LP animals. This phenotype has been described as sign of P deficiency in farmed teleost species such as Atlantic salmon, haddock (Melanogrammus aeglefinus), and halibut (Hippoglossus hippoglossus) [58][59][60]. Similarly, low-mineralised neural and haemal spines in LP zebrafish have an undulated shape (Figure 2b), yet without signs of fracture. Indeed, the collagen-based bone matrix alone is a very tough material. The toughest known vertebrate bones are deer antlers, which can flex without damage due to their low degree of mineralisation [61,62]. Notably, vertebral centra and arches and their spines are developmental modules, meaning that the control of their development is to a large degree independent [63][64][65]. This could explain why spines are twisted but vertebral centra are not affected in LP zebrafish. In laboratory zebrafish strains, undulated spines also occur linked to conditions other than P deficiency such as increased rearing density [43] or disturbed somite formation [66]. In addition, it has been suggested that spine deformities relate to musculature impairment [67,68]. The comparison of the results from this study with other studies that encountered malformed spines can, however, be difficult. Other studies have used different species or different zebrafish strains, different rearing conditions and different diet formulations. Costa et al. [69] tested the effect of six diets with different P levels on the zebrafish skeleton, but the diet composition was different from the one used in the present study. The diets contained poultry visceral meal and soy bean oil, whereas the diet used in this experiment contains krill meal, fish meal, and fish oil. Moreover, the inorganic P source in the present study is monoammonium phosphate (MAP), whereas dicalcium phosphate (DCP) was used by Costa et al. [69]. Solubility and digestibility, and thus the bioavailability of MAP, are considerably higher than DCP [70,71]. This and other dietary ingredients could explain why our LP zebrafish do not develop vertebral centrum deformities, or other deformities encountered by Costa et al. [69], such as severe bending of the vertebral column or craniofacial malformations. The phenotype of the control group (RP) in the present study, 1.0% total P based on MAP supplement, equals the phenotype obtained by Costa et al. [69] with the diet containing 1.85% total P based on DCP supplement.
The absence of vertebral centra malformations in the LP zebrafish group is in line with what is observed in recent studies on Atlantic salmon. In two different experiments, animals in their early seawater phase received P-deficient diets (50% of the total P requirement) for 10 weeks and 17 weeks. Like LP zebrafish in the current study, Atlantic salmon developed bone without minerals but no vertebral column deformities [10,12]. It can of course not be excluded that a prolonged low P period would eventually generate skeletal malformations in growing zebrafish.
How can the absence of malformations in LP zebrafish be explained? From a functional point of view, the notochord alone in the absence of vertebral bodies can act as efficient axial skeleton. This is the case in teleost fish that hatch as embryos [72,73] and in basal adult osteichthyans [74]. Members of several stem-ward groups, which comprise large animals such as dipnoans (lungfishes), coelacanths (crossopterygians) and sturgeons (chondrosteans, up to six meters in length), have a continuous non-constricted notochord as functional axial skeleton and do not develop mineralised vertebral centra [74][75][76]. Also, the nob zebrafish mutant strain that completely lacks bone mineralisation, shows correctly patterned but non-mineralised vertebral body anlagen [54]. Moreover, several species of deep sea fish are characterised by extremely low-mineralised skeletons and low-mineralised vertebral centra [77][78][79]. Thus, a functional and healthy notochord that supports the axial skeleton may compensate for non-mineralised vertebral centra as it compensates for absence in other osteichthyans. Indeed, this study demonstrates that a two months period under low P conditions does not cause any morphological alteration of the vertebral centra in zebrafish, except osteomalacia. Internal vertebral centra structures appear unaltered on histological sections. Intervertebral ligaments and the notochord tissue in the intervertebral space of LP zebrafish (a region called intervertebral disk by Schaeffer [80]) remain intact. This is an important observation because, also in teleosts, vertebral body malformations typically start with alterations of the intervertebral disk [56,57,81]. These findings strengthen the idea that P deficiency alone is not a primary cause of vertebral column abnormalities in zebrafish, and that other or additional factors trigger the development of malformations. Other factors that are currently discussed to cause vertebral column malformation in zebrafish and other teleost species are rearing temperature, excess swimming, increased rearing density or dietary vitamin A supply [42,43,82,83].

Is There a Link between Excess P Content and Vertebral Fusion?
In the present study, zebrafish from the HP group present an increased frequency of vertebral centra fusion. This suggests that high rather than low dietary P content could be a causative factor for vertebral body fusion in zebrafish. Up to now, little is known about the effects of dietary P excess on the development of skeletal malformations in teleosts [59]. In mammals, however, ectopic mineralisation can be caused by excess dietary P intake. Humans with excess dietary P ingestion develop abnormal mineralisation of the vascular tissue [27][28][29]. Dogs treated with high P diets developed mineralisation in the kidney [30][31][32][33] and increased accretion of cortical bone [34]. Also metabolic disorders that increase plasma P levels can cause anomalous mineralisation of soft tissues, as reported in patients that suffer from chronic kidney disease (CKD) [29,84] and in mice models of genetic diseases that cause hyperphosphatemia. Fgf23 is a hormone released by bone cells that down-regulates renal P reabsorption. Murine models with mutations of Fgf23 or in genes encoding proteins involved in Fgf23 modifications, suffer from hypermineralisation adjacent to the growth plate in the primary spongiosa and hyperdense femur bones [85,86], similar to what observed in our HP zebrafish.
Pathological mineralisation that affects the vertebral column has been reported for zebrafish of the enpp1 mutant strain. The lack of the ectonucleotide pyrophosphatase/phosphodiesterase-1 (Enpp1) reduces pyrophosphate, a mineralisation inhibitor generated by osteoblasts. The bones of juvenile enpp1 mutants are hypermineralised and vertebral centra fuse [87], a phenotype similar to vertebral fusions in HP zebrafish. Notably, the increased frequency of vertebral body fusions in HP zebrafish is only diagnosed after two months. This suggests that the prolongation of the dietary treatment is required before an effect can be observed. Further studies are required to clarify the exact mechanisms of vertebral body fusion related to high dietary P content.
Regarding the mechanical properties of the bone formed under the dietary treatment, a higher elastic modulus (increased material stiffness) was observed in the hypermineralised vertebral endplates of HP zebrafish compared to LP zebrafish. HP zebrafish also showed a tendency towards a higher elastic modulus compared to RP zebrafish. This suggests that not only a physiological increase in mineralisation, but also a dietary P-induced increase in mineralisation leads to a higher stiffness of bone [12,62,88]. It could be hypothesised that the increased stiffness of the vertebral centra likely increased the mechanical load on the intervertebral space while swimming, causing compression and tension of the intervertebral ligaments in the caudal region [89]. Tension is a well-known trigger for the mineralisation of ligaments and tendons [90,91]. Thus increased tension could trigger the mineralisation of the intervertebral ligaments, consequently leading to centra fusion [57].

Less Minerals but More Bone Production by Osteoblasts
Synchrotron X-ray tomographic microscopy allows the identification of mineralised and non-mineralised bone in the zebrafish vertebral bodies at a high resolution. The representative LP individual shows a vertebral body with an increased total bone volume in comparison to control and HP animals. The increase in bone volume is ascribed to a considerable increase of non-mineralised bone in the growth zone of the vertebral body endplates, neural and haemal arches and spines. Considering that animals of equal size were used from each group, it is intriguing to note that LP zebrafish show increased bone matrix formation. An increased production of collagen, consistent with the increased bone volume, suggests that osteoblasts are highly active at producing collagen matrix and this could explain the presence of enlarged endoplasmic reticulum cisternae in LP zebrafish [92,93]. The intensified matrix production does not influence collagen type I synthesis and post-translational modifications at the structural level. Collagen type I post-translational modification appears normal in LP zebrafish, as suggested by similar electrophoretic migration of α(I) chain bands in all dietary groups. Moreover, the progressive increase of collagen fibre diameter in the secreted bone matrix reflects normal fibrils maturation and aging [94]. Our observations agree with studies on Nile tilapia that show increased osteoid formation on pharyngeal bone in P-deprived animals [13]. Likewise, hypophosphatemic rats present increased osteoid width [14]. In mammals, osteoid undergoes several chemical modifications, designated as maturation, prior to mineralisation [95]. It has been suggested that increased osteoid production in P-depleted rats relates to a decreased rate of osteoid maturation, indicating a delay in the onset of mineralisation [14]. In Atlantic salmon, however, the non-mineralised bone formed under P-deficient conditions can mineralise completely if the animals receive a P-sufficient diet [12]. This, together with the normal post-translational modifications of collagen type I and the normal ultrastructure of collagen fibrils in LP zebrafish, argue in favour of normal bone matrix maturation also under low P conditions.
Maintaining bone mechanical stability could be a possible explanation for the increase of bone matrix production in LP zebrafish. As Ca and P contents in bone are always linked [10,12], the mechanical properties of bone change in accordance with the bones' mineral content [62], as also shown in this study. LP zebrafish could increase collagen secretion to compensate for the lack of minerals and reduced stiffness. Osteocytes are mechanosensitive cells that regulate bone formation and bone resorption in response to mechanical load. Upon mechanical stimuli, osteocytes are activated and produce signalling molecules that increase the activity of osteoblasts [8]. In particular, prostaglandins secreted by osteocytes stimulate bone formation in response to mechanical load in vivo [96,97]. As the newly secreted matrix cannot mineralise due to insufficient P levels, stronger mechanical stimulation of osteocytes inside the soft non-mineralised bone could trigger an increased activity of the osteoblasts under LP conditions.

Zebrafish and Ethical Statement
Wild type AB zebrafish were obtained from European Zebrafish Research Center (Eggenstein-Leopoldshafen, Germany). Zebrafish embryos were kept in petri dishes in fish water (1.2 mM NaHCO 3 , 0.01% instant ocean, 1.4 mM CaSO 4 , 0.0002% methylene blue) at 28 • C until 7 days post-fertilisation (dpf), then housed in ZebTEC semi-closed recirculation housing systems (Techniplast, Buguggiate, Italy) at 28 • C, pH 7.5 and conductivity 500 µS on a 14/10 light/dark cycle. Zebrafish from 7 to 21 dpf were fed three times a day alternating commercial dry food (ZM000, Zebrafish Management Ltd., Winchester, UK) and brine shrimp (Artemia cysts, Zebrafish Management Ltd., Winchester, UK). Fish were then fed for another week three times a day with the dry regular phosphorus (RP) diet (Table 3, see also the next section: experimental diets), until 28 dpf, to adjust them to this type of dry food. The nutrition trial started at 28 dpf: fish were randomly divided in three groups, grown in identical tanks with a density of 10 fish/L and fed three times a day with a low P (LP) diet, a regular P (RP) diet and a high P (HP) diet, respectively (Table 3). Samples were collected before the start of the experiment (T0 samples, 28 dpf) and after one and two months of dietary treatment (two and three months post-fertilisation, respectively). Fish were euthanised by tricaine (3-amino benzoic acidethylester) overdose (0.3%) and fixed for further analyses. The experiments were conducted in the centralised animal facility of the University of Pavia (Pavia, Italy). The experimental protocol was approved by the Italian Ministry of Health (Approval animal protocol No. 260/2020-PR, 26 March 2020).

Experimental Diets
Three experimental diets were formulated to have a total P content of 0.5%, 1.0% and 1.5%, termed LP diet, RP diet and HP diet, respectively (Table 3). P content for the control diet, RP, was based on the total dietary P requirement of 0.6-0.8% for different teleost species (Rainbow trout Oncorhynchus mykiss, Atlantic salmon Salmo salar, Pacific salmon Oncorhynchus sp., Carp Cyprinus carpio, European sea bass Dicentrachus labrax) [98]. LP and HP diets were formulated to have, respectively, a drastic reduction and an excess of total P content. Monoammonium phosphate (MAP) was used as dietary inorganic P supplement. MAP has a high P bioavailability and high P retention efficiency in teleost fish [71]. In order to keep all diets equal in nutrients, except for P concentration, MAP replaced the inert filler diatomaceous earth (Diamol, Imerys, Denmark). The experimental diets were formulated by SimplyFish AS (Stavanger, Norway, www.simplyfish.no) and produced by extrusion with subsequent crumbling to a suitable particle size by the Danish Technological Institute (Taastrup, Denmark, https://www.dti.dk). The P content of the product was verified at the University of Hohenheim (Stuttgart, Germany, https://www.uni-hohenheim.de) and determined with 5.04 g/kg diet, 9.84 g/kg diet and 14.64 g/kg diet for the LP, RP, and HP diet, respectively.

Morphometric Analysis
Fish were euthanised by tricaine overdose and lateral images were acquired with a M165FC stereomicroscope (Leica, Wetzlar, Germany) connected to a DFC425C digital camera (Leica, Wetzlar, Germany). The standard length (SL), described as the distance from the most anterior tip of the upper jaw to the most posterior region of the body where caudal fin rays insert [99], was measured using the ImageJ software (NIH, Bethesda, MD, USA) (T0 n = 96; one month treatment: LP n = 70, RP n = 34, HP n = 42; two months treatment: LP n = 59, RP n = 63, RP n = 47).

Whole Mount Staining with Alizarin Red S
Zebrafish of 28 dpf (n = 8), one month treated (LP n = 51, RP n = 21, HP n = 31) and two month treated (LP n = 41, RP n = 48, HP n = 32) fish were euthanised by tricaine overdose, fixed for 24 h in 4% paraformaldehyde (PFA) in 1× phosphate-buffered saline (PBS) at 4 • C and were stained according to the following protocol: 1.5% H 2 O 2 in 0.25% KOH (2 h); distilled H 2 O (dH 2 O) (5 min); 0.01% Alizarin red S in 0.5% KOH (12 h); 1% KOH (2 h); 25% glycerol in 0.75% KOH (2 h); 50% glycerol in 0.5% KOH (2 h); 75% glycerol in 0.25% KOH (2 h); 100% glycerol (modified from [100]). Fish were analysed and imaged using an Axio Zoom V16 stereomicroscope (Carl Zeiss, Oberkochen, Germany) with oblique illumination equipped with a 5MP CCD camera. Lateral images of stained fish were used to quantitatively analyse mineralisation levels of vertebral endplates. The vertebral body in the transition region, described as the first caudal vertebra possessing elongated unfused haemal arches, drastically shortened ribs and absence of haemal spine [40], was considered for measuring the total vertebral length and the non-mineralised endplate length. Endplates represent the growth zone of vertebral centra, where the newly formed bone is deposited. Thus, endplates provide a valuable location to characterise bone formed during the dietary treatment. The non-mineralised endplate was expressed as a percentage of the total non-mineralised endplate length over the total vertebral length: (A + A')/B (Figure 2c). Vertebral endplates with a non-mineralised percentage value greater than 10% were classified as low-mineralised, between 3% and 10% as intermediate mineralised, and less than 3% were considered fully mineralised. Fin rays were classified as low-mineralised if more than two segments were non-mineralised, intermediate mineralised if one or two segments were non-mineralised and fully mineralised if all segments were mineralised. Mineralisation levels of neural and haemal arches and dorsal and anal pterygiophores were qualitatively evaluated as low, intermediate or high depending on Alizarin red S distribution in the bone (Supplementary Figure S1).

Transmission Electron Microscopy
Specimens treated for two months (n = 1 fish per dietary group) were euthanised by tricaine overdose and fixed for 24 h in 2.5% PFA, 1.5% glutaraldehyde, 0.1 M sodium cacodylate buffer (pH 7.4) and 0.001% CaCl 2 at 4 • C. Fixed fish were decalcified in 0.1 M EDTA for 14 days. The decalcification solution was changed every 3 days. Specimens were subsequently rinsed in 0.1 M sodium cacodylate buffer with 10% saccharose and then post-fixed for 2 h in 1% OsO 4 solution in 0.1 M cacodylate buffer containing 3% saccharose. After rinsing in buffer, samples were dehydrated in a series of graded ethanol solutions and embedded in epon epoxide medium [103]. Semi-thin 1 µm sections were cut on a Microm HM360 microtome (Marshall Scientific, Hampton, NH, USA), stained with toluidine blue at pH 9 for 2 min (0.5% toluidine blue, 1% Na 2 B 4 O 7 in dH 2 O), rinsed with H 2 O and mounted with DPX. For transmission electron microscopy (TEM) analysis, ultrathin sections (about 70 nm) of the region of interest were prepared on an UltracutE ultramicrotome (Reichert-Jung, Buffalo, NY, USA), contrasted with uranyl acetate and lead citrate [104] and analysed with a Jeol JEM 1010 transmission electron microscope (Jeol Ltd., Tokyo, Japan) operating at 60 kV. Microphotographs were taken with a Veleta camera (Emsis, Muenster, Germany). TEM images were used to measure the diameter of collagen type I fibres in proximity of the osteoblasts (n = 300 per dietary group) and in the extracellular matrix (n = 300 per dietary group). Diameters were measured on fully transversely sectioned fibres using ImageJ software (NIH, Bethesda, MD, USA). Analysis of collagen fibres was based on previously established protocol [36], the number of fibres analysed exceeds the established one.

Synchrotron X-ray Tomographic Microscopy
Fish treated with the experimental diets for two months (n = 1 fish per dietary group) were euthanised by tricaine overdose, fixed for 24h in 2.5% PFA, 1.5% glutaraldehyde, 0.1 M sodium cacodylate buffer (pH 7.4) and 0.001% CaCl 2 at 4 • C and were dehydrated in a graded series of ethanol. Representative specimens of the dietary groups with similar size were selected by whole mount Alizarin red S staining of the abdominal region of the vertebral column. Synchrotron X-ray tomographic microscopy of the caudal region of the vertebral column [40] was performed at the TOMCAT beamline (X02DA) (Swiss Light Source (SLS), Paul Scherrer Institut (PSI), Villigen, Switzerland, www.psi.ch/sls). Scans were acquired with 1501 projections over 180• at 16 keV with an exposure time of 150 ms using a 10× objective resulting in an effective voxel size of 0.65 µm [105]. Radiographs were phase-retrieved using the Paganin algorithm, tomographically reconstructed and subsequently analysed using Amira 3.1.1 software (TermoFisher, Waltham, MA, USA). The mineralised parts of the vertebrae were identified applying a constant greyscale-threshold to all samples, the non-mineralised parts were manually selected on each projection. The resulting segmentations and virtual sections of the 10th caudal vertebral body [40] in LP, RP and HP fish were used to visualise the mineralised and non-mineralised parts of the vertebra, neural and haemal arches. Segmentations were also used to measure vertebral body parameters (length, height), arch and spine length, and the volumes of mineralised and non-mineralised parts of the vertebrae (Figure 5b).

Nanoindentation
Mechanical properties were assessed in the first four caudal vertebrae of 1 fish per dietary group according to previously established protocols [106]. Briefly, fish were embedded in PMMA, ground coplanar until the sagittal plane was exposed, and polished with 3 µm and 1 µm diamond suspension followed by final polishing with 0.05 µm aluminium-oxide suspension. Indentations were performed in depth-sensing continuous stiffness mode with a final depth of 300 nm. Indentations were placed in the proximal and distal vertebral body endplates and the central region of vertebrae (Figure 4a). The nanoindenter (Nano Indenter G200 equipped with a Berkovich diamond tip, Keysight Technologies, Santa Rosa, CA, USA) was calibrated on fused silica before and after each measurement. Based on the Oliver and Pharr method [107] and by applying a Poisson's ratio of 0.3, the mechanical properties elastic modulus (E) and hardness (H) in the different vertebral regions were determined using in-house software (NanoSuite, Keysight Technologies, Santa Rosa, CA, USA).

Collagen Extraction from Bone
Bones were dissected following sacrifice from animals fed with the experimental P diets for two months (n = 2 fish per dietary group). Bones were defatted for 6 h in 0.1 N NaOH at 4 • C and then decalcified for 48 h in 0.5 M EDTA (pH 7.4) at 4 • C. The pepsin-soluble collagen fraction (PSC) was obtained by digesting tissues with 0.1 mg/mL pepsin in 0.5 M acetic acid at 4 • C for 48 h. The PSC was precipitated by 0.9 M NaCl in 0.5 M acetic acid overnight at 4 • C [108] and quantified using Sircol Soluble Collagen assay (Biocolor, Carrickfergus, UK). Equal amounts of collagen from each sample were loaded on 6% SDS-Urea-PAGE in non-reducing condition. Gels were stained overnight with 0.08 M picric acid, 0.04% Coomassie Brilliant Blue R250, rinsed in water and recorded with Versadoc3000 (Bio-Rad, Hercules, CA, USA).

Statistical Analysis
Quantitative variables are expressed as mean ± standard deviation, categories are expressed as percentages. Statistical comparison of the standard length values was based on the non-parametric Mann-Whitney test. Differences in the occurrence of skeletal malformations and in bone mineralisation levels were evaluated by means of Chi-squared test or the Fisher's exact test. Statistical analysis was performed using Past 4.01 software [109]. Comparisons of mechanical properties was performed using ANOVA followed by a Bonferroni post hoc test. A p-value less than 0.05 was considered significant.

Conclusions
In this experiment, a new zebrafish model for low and high dietary P levels demonstrates that low P levels in the diet have no negative effect on bone matrix formation, although new bone matrix remains non-mineralised. Moreover, no vertebral centra malformations occur, indicating that other factors may trigger the development of skeletal deformities. In contrast, high dietary P levels lead to increased bone mineralisation, increased bone stiffness and fusion of vertebral centra. Neither the lack of mineralisation, nor the high mineralisation affect collagen post-translational modifications, as expected given that ER resident enzymes do not depend on P [45]. Increased production of normal collagen matrix is observed in animals from the LP group: organic matrix is continuously produced and collagen fibres mature despite the arrest of mineralisation. The current findings therefore support the idea that bone matrix secretion and bone mineralisation are uncoupled processes, and explain why non-mineralised bone produced under P deficiency conditions can mineralise completely when adequate dietary P is provided [12]. This new bone mineralisation zebrafish model can be used in biomedical research to obtain insights into bone mineralisation pathologies related to high and low mineralisation degree. The findings of this study may also benefit aquaculture research as a model for the effects of dietary P supply in farmed teleosts.