Improved regeneration and de novo bone formation in a diabetic zebrafish model treated with paricalcitol and cinacalcet

Bone changes related to diabetes have been well stablished, but few strategies have been developed to prevent this growing health problem. In our work, we propose to investigate the effects of calcitriol as well as of a vitamin D analog (paricalcitol) and a calcimimetic (cinacalcet), in fin regeneration and de novo mineralization in a zebrafish model of diabetes. Following exposure of diabetic transgenic Tg(ins:nfsb‐mCherry) zebrafish to calcitriol, paricalcitol and cinacalcet, caudal fins were amputated to assess their effects on tissue regeneration. Caudal fin mineralized and regenerated areas were quantified by in vivo alizarin red staining. Quantitative real‐time PCR was performed using RNA from the vertebral column. Diabetic fish treated with cinacalcet and paricalcitol presented increased regenerated and mineralized areas when compared with non‐treated diabetic group, while no significant increase was observed in non‐diabetic fish treated with both drugs. Gene expression analysis showed an up‐regulation for runt‐related transcription factor 2b (runx2b), bone gamma‐carboxyglutamic acid‐containing protein (bglap), insulin a (insa) and insulin b (insb) and a trend of increase for sp7 transcription factor (sp7) in diabetic groups treated with cinacalcet and paricalcitol. Expression of insra and vdra was up‐regulated in both diabetic and nondiabetic fish treated with cinacalcet. In nondiabetic fish treated with paricalcitol and cinacalcet a similar increase in gene expression could be observed but not so pronounced. The increased mineralization and regeneration in diabetic zebrafish treated with cinacalcet and paricalcitol can be explained by increased osteoblastic differentiation and increased insulin expression indicating pro‐osteogenic potential of both drugs.

Prevalence of diabetes mellitus worldwide was estimated to be 135 million in 1995 and is predicted to be of 300 million in the year 2025 1 leading to an increase in patients living with the risk of developing diabetes-related complications. 2 Diabetes has been strongly associated with development of bone fractures, that begins in childhood and extends throughout life, leading to morbidity and mortality. 3 Calcitriol (1,25-dihydroxyvitamin D3-VitD) deficiency is common in chronic kidney disease (CKD) patients, leading to secondary hyperparathyroidism (SH) 4 and particularly in those diabetic patients that also undergo dialysis, there is a clear correlation with low 25-hydroxyvitamin D in serum. 5 Parathyroid hormone (PTH) stimulates bone resorption directly by activating PTH receptors in mesenchymal cells of the osteoblast lineage and indirectly by increasing differentiation and function of osteoclasts. 6 Bone loss and increased fracture risk are among the complications presented by CKD patients with SH 7,8 and are also seen in rodent models. 9 VitD, the vitamin D analog paricalcitol 10 and the calcimimetic cinacalcet 11 have been used for the treatment of SH in CKD patients with beneficial effects in lowering PTH values and also in increasing bone mass. 7,8 Paricalcitol, by selectively activating the vitamin D receptor (VDR), 12,13 and cinacalcet, by activating the calcium sensing receptor (CaSR) 12,14 in parathyroid, have been shown to be more efficient and fast in normalizing PTH levels and in reducing circulating bone turnover markers than VitD in patients with SH. 15,16 In addition, VDR and CaSR activation have been associated with increased insulin expression 17,18 and beneficial effects were reported under diabetic conditions. 19,20 Nevertheless, little is known on whether upregulation of insulin can occur in extrapancreatic tissues. In humans, extrapancreatic expression of insulin was first identified in brain 21 and then in thymus, 22 as a way for the immune system to recognize insulin, avoiding autoimmunity and b-cell destruction. Later Kojima et al. 23,24 showed the presence of cells positive for insulin RNA in the liver, adipose tissue and bone marrow in several diabetic mice models, but not in non-diabetic mice. However, this insulin expressing cells had no impact on reducing hyperglycemia in diabetic mice. Later, Kojima et al. 25 demonstrated that beside not having any impact in regulating glucose levels, this Proins/TNF-a-expressing cells had their origin in bone marrow and then migrated to several parts of the body, initiating diabetic neuropathy. 26 Cunha et al.,27 in their experiments with diabetic mice treated with streptozotocin, observed beneficial effects in the secretion of insulin by the tear film of the eye and showed that insulin was locally synthesized in the lachrymal gland. 27 With the current body of evidence, it is difficult to determine if extrapancreatic expression of insulin has any effect in glucose homeostasis, or if it serves other undetermined functions.
Zebrafish have been well established as a model for human diseases, spanning a wide range of human pathologies including genetic disorders and physiological processes that are known to be highly conserved throughout vertebrate evolution. 28 Recently it was demonstrated, under type 1 diabetic conditions, that fin regeneration was impaired in zebrafish with reduced cell proliferation and increased cell apoptosis. 29 We conducted experimental trials to understand if VitD, paricalcitol and cinacalcet, could have beneficial effects on caudal fin regeneration and bone mineralization in adult zebrafish and in operculum development of larvae under type 1 diabetic conditions. To test this hypothesis we used the transgenic Tg(ins:nfsb-mCherry) zebrafish that after being exposed to the prodrug metronidazole becomes hyperglycemic and hipoinsulinemic due to beta-cell ablation, leading to a transient state of diabetes of 10-15 days prior to beta-cell regeneration. 30,31 To understand if vitamin D analogs and calcimimetics could have positive effects on bone development and mineralization, during a transient diabetes type I state, fin regeneration and bone mineralization were assessed and expression of bone and vitamin D metabolism markers evaluated.

Zebrafish strains and maintenance
The transgenic Tg(ins:nfsb-mCherry) zebrafish line used in our experiments was kindly given by the Laboratory of Molecular Biology and Genetic Engineering, GIGA Research, Liege, Belgium. Transgenic zebrafish were maintained in a recirculating water system (Tecniplast, Buguggiate, Italy). All manipulations were performed by licensed researchers and conducted in accordance with principles and procedures following the guidelines from the Federation of Laboratory Animal Science Associations (FELASA) and in accordance with the EU and national regulations. The Tg(ins:nfsb-mCherry) line generated by Pisharat et al. 30 with a T€ ubingen AB background, contains a construct in which the nfsB gene of E. coli and the florescent protein mCherry are inserted downstream to the promoter region of the insa gene. That bacterial gene encodes a nitroreductase (NTR) enzyme, that converts prodrugs such as metronidazole (MET; Sigma-Aldrich, St. Louis, MO) to cytotoxins. By observing the loss of mCherry fluorescence after MET treatment it is possible to visualize MET dependent b-cell ablation.

Diabetes induction
Larvae at 15 days post fertilization and male and female adults with 1 year old from the Tg(ins:nfsb-mCherry) zebrafish line were anesthetized with tricaine methanesulfonate (Sigma-Aldrich, St. Louis, MO) 32 and exposed to MET either by bath or through intraperitoneal (IP) injection at the concentration of 0.05M dissolved in citrate buffer (0.05M). Corresponding control groups were left untreated under the same housing conditions. In addition, we have exposed nontransgenic siblings from a cross between a heterozygous transgenic and a wild type zebrafish, to discard potentially side effects of MET treatment in caudal fin regeneration.

Paricalcitol, cinacalcet and VitD treatments
Seventy-two hours post treatment (hpt) with MET, transgenic zebrafish were screened for loss of fluorescence due to b-cell ablation, as observed by Pisharath et al. 30 To understand if treatments could induce bone alterations in diabetic conditions, 240 larvae and 145 adults were divided into experimental groups in triplicates and exposed by immersion (larvae) or IP injection (adults) of VitD (0.001 lg/ml) (calcitriol, Sigma-Aldrich), paricalcitol (0.001 lg/ml) (zemplarV R , Abbott Laboratories, North Chicago, IL) and cinacalcet (0.05 lg/ml) (mimpara, Amgen Europe B.V., Breda, The Netherlands), mimicking the concentrations used in clinical practice. For the control groups of non-diabetic and diabetic fish we used a vehicle solution (citrate buffer 0.05M). Since cinacalcet was found to be lethal to larvae at the concentration used for adults, we have performed treatments with dilutions of the initial concentration of cinacalcet by 1:10 (0.005 lg/ml), 1:50 (0.001 lg/ml) and 1:100 (0.0005 lg/ml) and included three additional groups of diabetic and nondiabetic larvae treated with cinacalcet.

Fin amputation
After the IP injections with treatments or vehicle, the caudal fins of treated adults (n 5 8) were amputated two segments below the ray bifurcation. Both larvae and adults were maintained under treatment for 120 hours at 28.58C and fed twice a day with Artemia nauplii (EG strain, Inve, Dendermonde, Belgium).

Staining of mineralized tissue
Larvae (n 5 15) were fixed with PFA 4% for 1 hour, washed in PBS and stained in 0.01% alizarin red for 30 minutes. Adults (n 5 8) were submitted to live staining in alizarin red at a concentration of 0.01% for 15 minutes prior to observation. Adult regenerated caudal fins and larval opercula were photographed under fluorescence (546nm) in a stereomicroscope (Leica MZ9.5, Leica, Wetzlar, Germany) for identification of the calcified regions.

Quantification of operculum mineralized area in larvae
Mineralized area of opercula stained by alizarin red were measured using image J software. Results were normalized by dividing operculum area (OA) by total area of the head (HA).

Quantification of regenerated and mineralized area of adult fin
Regenerated area was determined by dividing regenerated area (REG) by stump width (STU) and mineralized area was determined by dividing mineralized area (MIN) by mean ray width (MRW) and divided by REG/STU. All quantifications were done using image J software.

Bone histology and histomorphometry
The calcified regenerated fins from each of the different groups were transferred to 70% EtOH and processed for dehydration and infiltration on a routine overnight processing schedule. Samples were then embedded in paraffin and sections with 6 mm prepared in a microtome. Before staining, sections were deparaffinized in xylene and dehydrated in an increasing gradient of EtOH. Sections were stained by von Kossa's as described elsewhere. 33 To determine fin area and thickness, the second, third and fourth hemiray of each fin were measured. Area was assessed by measuring total area of both hemirays and thickness was assessed by four longitudinal measurements of each hemiray. A detailed time course of the different procedures from metronidazole treatment to amputation and data acquisition is shown in Supplementary Figure S1.

Glucose tolerance test
To confirm that ablation of b-cells in Tg(ins:nfsb-mCherry) zebrafish led to an increase in glucose blood concentrations, we administrated a solution of glucose at a concentration of 0.1 M or vehicle by IP injection, to two groups of adult zebrafish after 72 hpt with MET. Blood glucose was monitored at 30, 60, 90, 120, 150 and 180 minutes after IP injection with glucose. The glucose levels were measured in 3 Tg(ins:nfsb-mCherry) zebrafish treated with MET and with vehicle at each time point. 6 ll of blood were collected from the caudal aorta, diluted in 2 ll of 2% heparin and rapidly transferred to a blood glucose meter Glucocard MX (Arkray A. Menarini Diagnostics, Florence, Italy). All these procedures were repeated four times with 6-7 specimens by group.

Total RNA isolation
Vertebral columns from 6 adults of each group were isolated and pooled in 2 groups (n 5 3/each) for RNA purification. The samples were placed in 1 ml of Isol-RNA Lysis Reagent (5 PRIME, Hilden, The Netherlands) and total RNA was purified according to manufacturer's protocol. RNA quantity and integrity were verified using Experion RNA Analysis Kit (BIO-RAD, Hercules, CA). , bone gammacarboxyglutamic acid-containing protein (bglap), sp7 transcriptor factor (sp7), parathyroid hormone receptor a (pthra), vitamin D receptor a (vdra), insulin a (insa), insulin b (insb) and insulin receptor a (insra) are listed in Table 1. All gene expression data were normalized against the mean of the gene expression levels of housekeeping genes ef1-alfa and 18S. To confirm extrapancreatic expression of insa and insb genes observed by qPCR, a RT-PCR reaction was performed and the identity of the amplicons confirmed by sequencing. Gene expression results are the mean of two different experiments.

Statistical analysis
All statistical analyses were performed using Stata Statistical Software and data was evaluated using the one-way ANOVA followed by Bonferroni multiple comparisons test with p < 0.05 considered statistically significant. Results are presented as means 6 standard deviation of the mean (SD).

RESULTS
As previously described, Tg(ins:nfsb-mCherry) presented loss of fluorescence in the region of the pancreas after 72 hpi with MET, confirming b-cell ablation (Supplementary Figure S2). To confirm that Tg(ins:nfsb-mCherry) exposed to MET could lead to a state of onset of diabetes, a glucose tolerance test was performed to measure how well animals are able to break down glucose, or sugar. We found that glucose concentrations in the plasma of fish treated with MET were significantly higher than in fish treated with vehicle only, with these differences being highly significant at 90, 120, 150 and 180 minutes after IP injection (Figure 1). An analysis of the mineralized area of the operculum (Figure 2A and B) of larvae from the D group revealed a significant reduction when compared with nondiabetic samples. However, the mineralized area of the operculum was significantly increased when diabetic larvae were treated with paricalcitol, cinacalcet and VitD compared with untreated diabetic larvae. Nondiabetic larvae treated with the three different treatments showed a tendency for an increase in mineralization compared with untreated nondiabetic larvae, although not significantly different due to individual variability ( Figure 2C).
Following b-cell ablation and regeneration we have quantified the regenerated and mineralized areas ( Figure 3A).  Diabetic zebrafish adults had a statistically significant impairment of fin regeneration when compared with nondiabetics. Furthermore, the diabetic and non-diabetic groups treated with paricalcitol and cinacalcet presented a significant increase in regenerated area when compared with diabetic fish, but a not significant increase in regeneration when compared with nondiabetics ( Figure 3B). In nondiabetic treated groups, the increase in regenerated area was not significant. The regenerated area of the diabetic and nondiabetic zebrafish groups treated with VitD did not present any significant differences compared with respective control groups ( Figure 3B). Wild type (WT) fish treated with MET showed no differences to nondiabetics or to control WT fish exposed to vehicle.
Quantification of the mineralized area showed that groups treated with paricalcitol and cinacalcet had no differences when compared with untreated nondiabetic fish while the diabetic fish showed a significantly reduced mineralization when compared with nondiabetic and with diabetic paricalcitol and cinacalcet treated groups, while VitD treated groups showed no differences relative to the other treated groups ( Figure 3C). Nontransgenic siblings treated with MET or vehicle (WT, WT 1 MET) showed no alterations both in regenerated and mineralized areas when compared with the ND group ( Figure 3B and C).
Histology of rays confirmed previous results, showing increased ray area and thickness ( Figure 4A) in the regenerated caudal fins of diabetic fish treated with paricalcitol and cinacalcet when compared with untreated diabetic fish, while nondiabetic treated groups showed no significant increase ( Figure 4B and C). Analysis of gene expression levels showed no significant differences relative to the expression of pthra when comparing nondiabetic and diabetic fish ( Figure 5A). The vdra expression, showed a statistically significant increase in the cinacalcet treated groups compared with all other groups ( Figure 5B). insra expression was found to be significantly down-regulated in all diabetic groups (p < 0.5) compared with nondiabetic, but the diabetic fish treated with cinacalcet showed a lower reduction in expression, with statistically higher values compared with diabetic, paricalcitol and VitD treated fish, while nondiabetic cinacalcet treated group showed increased expression compared with all other groups ( Figure 5C). sp7 expression showed a trend of reduction in all diabetic fish treated groups compared with nondiabetics, but significant downregulation was only found in diabetic compared with nondiabetic groups, no differences being observed among nondiabetic groups ( Figure 5D). Regarding runx2b, a significant up-regulation could be observed in groups treated with paricalcitol and cinacalcet ( Figure 5E). Expression of bglap was down-regulated in diabetic compared with nondiabetic fish while the paricalcitol and cinacalcet diabetic groups showed significant differences compared with untreated diabetic group. In nondiabetic fish, treated versus control groups were not significantly different. Gene expression of insa and insb were found to be up-regulated (p < 0.001) in both diabetic and nondiabetic paricalcitol and cinacalcet treated fish when compared with the other groups ( Figure 4G and H).
To demonstrate the extrapancreatic expression of both insa and insb in vertebral column, we performed an RT-PCR using cDNA of liver/pancreas, muscle, kidney, column and cleithrum/operculum from wild type zebrafish, using the same primers used for qPCR. We observed amplification of ins genes in all tissues analyzed. This result was further confirmed by sequencing the PCR amplicons which were confirmed to correspond to insulin a and b isoforms, respectively ( Figure 6).

DISCUSSION
This study demonstrated that zebrafish is a suitable model for the study of bone pathologies related to diabetes. Ablation of b-cell, by exposing Tg(ins:nfsb-mCherry) zebrafish to MET, led to loss of mCherry fluorescence in the b-cells at 72 hpi, as already described by previous authors, 30,31 and to blood glucose increase, although differences were not as accentuated as previously reported. 30,31 Beta-cell ablation was confirmed by loss of mCherry fluorescence after 72 hours of treatment with MET. Transgenic fish that did not present a total loss of fluorescence were removed from further experimental procedures. The injection of glucose in diabetic fish led to an increase in the plasma glucose levels significantly higher than in nondiabetic controls. The levels of blood glucose started to decrease after 120 minutes post treatment, although they remained significantly higher in diabetic fish while in nondiabetics the levels returned to pretreatment levels. The reduction of glucose levels in plasma observed in diabetic fish can be explained by elimination of excess plasma glucose in urine, as previously reported for other species, where after injection with levels from 6.4 mM to 25 mM glucose in tilapia the excretion of excess glucose in the urine was visible after few hours. It was reported that treated fish had a drastic increase in urine glucose levels that were the double of the values measured in plasma. 34 It has also been shown in rainbow trout that there is a renal regulation of glucose levels in the plasma with the observation that high concentrations of glucose in plasma lead to glycosuria and to elimination in the urine. 35 Moreover, it is known that small fish are considered to have lower glucose tolerance than large fish. 36 Induction of diabetes in zebrafish caused an impairment in operculum mineralization and bone growth in 15 day larvae, similar to growth retardation observed in diabetes type 1 patients 37 and diabetes mice models. 38 Treatment with paricalcitol was more efficient than with VitD in promoting an increase in the mineralized area. The concentrations of Cinacalcet in our zebrafish treatments were the same as used in clinical therapies, but were found to be toxic in larvae with a rate of mortality of 100% after 24 hours. However, lower concentrations showed reduced lethality and induced an increase in mineralized area. Regenerated fin areas of adults showed an increase in the diabetic groups treated with paricalcitol and cinacalcet when compared with diabetic group treated with vehicle, while diabetic fish treated with VitD did not present such a marked increase. There are some evidences that vitamin D analogs can have positive regenerative effects after vascular injury, as previously reported for healthy humans, diabetic mice models and conditional knockout of the vitamin D receptor mice. 39 In our work, paricalcitol had no effect in vdra expression, but we could detect an up-regulation in the group treated with cinacalcet, in agreement with previous in vitro studies with rat parathyroid glands, demonstrating that class II calcimimetics induce a stimulatory affect in Vdra expression. 40 In addition, gene expression results for both insa and insb, which were found to be overexpressed in bone in diabetics treated with paricalcitol or cinacalcet, could help explain this increase in caudal fin regeneration. Vitamin D analogs and calcimimetics have been shown to induce insulin expression and b-cell proliferation and survival, [41][42][43] and our data suggest that this up-regulation of insulin also occurs in bone cells. In the Tg(ins:nfsb-mCherry) zebrafish, it has been described that total pancreas regeneration occurs in 15 days after diabetes induction, but we do not know if this process occurred in a shorter period of time in the paricalcitol and cinacalcet treated groups, which could favor insulin signaling and glucose metabolism. In the cinacalcet treated group we could observe a significant increase in insr suggesting increased insulin signaling. In fact, VitD may have an important role in the treatment of diabetes as identified by Del Pino-Montes et al., 44 who showed that 55% of diabetic rats treated with calcitriol recovered from diabetes.
Insulin expression has also been found to be increased in several tissues under diabetic conditions in both humans and mice, 45 but in our study we could not observe such an increase in the nontreated diabetic group, at least in the vertebral column. It has been demonstrated that insulin can be almost ubiquitously expressed in human, 46 mice 45 and zebrafish 47 although at extremely low levels when compared with pancreatic insulin. Although not well understood, the function of extrapancreatic expression of insulin was associated in some studies to local needs of glucose regulation, specially under diabetic conditions, 46,48 but in other reports this phenomenon was related to the development of pathologic conditions. 29 Our results demonstrated that paricalcitol and cinacalcet can up-regulate extrapancreatic expression of insulin, including in bony tissues, like demonstrated in the vertebral column of adult zebrafish.
In humans and animal models of diabetes, hyperglycemia leads to accelerated accumulation of advanced glycation end products (AGEs), 49 promoting an inflammatory response and increased apoptosis of cells expressing the receptor of AGEs such as osteoblasts. 50 In the nontreated diabetic group we could see impairment in osteoblastic activity, since sp7, runx2b and bglap expression were found to be down-regulated compared with nondiabetic fish. In the diabetic groups treated with paricalcitol or cinacalcet, where an increase in mineralized area of the regenerated fin was observed, an up-regulation of runx2b suggests an increase in the process of osteoblastic differentiation, contributing to the process of mineralization. This is in accordance with studies in humans, indicating that VitD effects on osteoblast differentiation are mostly stimulatory and associated with increased RUNX2 expression. 51 In addition, in vitro studies with mesenchymal stem cells from human amniotic fluid have correlated calcimimetics with osteogenic differentiation and upregulation of bone markers including RUNX2. 52 The Figure 5. RNA gene expression from the vertebral column of diabetic or diabetic with treatments is altered in zebrafish; (A) Diabetic and nondiabetic treated groups showed no significant differences in pthra expression; (B) Diabetic and nondiabetic treated with cinacalcet have an increase in vdra expression compared with all the other groups; (C) Diabetic and nondiabetic groups treated with cinacalcet showed increased expression of insra; (D) The diabetic fish showed a down-regulation of sp7 compared with nondiabetics. The treated groups showed no differences relatively to untreated groups; (E) Both diabetic and nondiabetic groups treated with paricalcitol and cinacalcet showed an increase in the expression of runx2b; (F) Diabetic group showed reduced expression of bglap compared with nondiabetic fish. All the treated groups with paricalcitol and cinacalcet showed significant up-regulation relatively to the untreated diabetic group; (G, H) Diabetic and nondiabetic groups treated with paricalcitol and cinacalcet showed an increase in the expression of both insa and insb gene. Bars with different superscript letters indicate significant differences (p < 0.05). [Color figure can be viewed at wileyonlinelibrary.com] 3 Figure 6. Zebrafish presented extrapancreatic expression of both insa and insb. Expression of insa and insb could be detected in pancreas/liver, muscle, kidney and bones from vertebral column and cleithrum/operculum. principal objective of paricalcitol, cinacalcet and VitD in clinical treatment for SH is to reduce parathyroid hormone secretion. We could not observe reduced expression in pthra in all treated groups so no conclusions can be made relatively to the pth regulation of osteoclastic differentiation and bone resorption. 53 In fact, pthra results seems to support the idea that pth pathway was not altered in the treated groups because the VitD treated group did not present such a marked increase in mineralized area as observed in the other two treated groups, while having the same results for pthra. The fact that VitD acts more slowly in exerting its effects, at least when compared with paricalcitol, 54 can be one of the possible explanations for our results. Different pathways related to pth signaling, calcium metabolism or VitD induced osteoblastogenesis can be involved in the increase in bone mineralized and regenerated areas observed in the caudal fin of zebrafish under diabetic conditions treated with paricalcitol and cinacalcet. Upregulation of insulin and increased osteoblastic differentiation induced by up-regulation of runx2b can help explain our results. Both paricalcitol and cinacalcet were shown to have positive effects in promoting mineral deposition, counteracting bone loss related to diabetes, and may constitute an alternative therapy for prevention of bone related disorders observed in type I diabetes patients.

Supporting Information
Additional supporting information may be found in the online version of this article at the publisher's web-site: Figure S1. Detailed timeline of diabetes induction and treatment with experimental drugs during caudal fin regeneration. Figure S2. Confirmation of beta-cell ablation. (A) Metronidazole exposed Tg(ins:nsfb-mCherry) zebrafish lose mCherry fluorescence after 72 hours of treatment. (B) Vehicle exposed Tg(ins:nsfb-mCherry) zebrafish maintain mCherry fluorescence after treatment.