Eldecalcitol is more effective for promoting osteogenesis than alfacalcidol in Cyp27b1-knockout Mice

Calcium (Ca) absorption from the intestinal tract is promoted by active vitamin D (1α,25D3). Vitamin D not only promotes Ca homeostasis, but it also inhibits bone resorption and promotes osteogenesis, thus playing a role in the maintenance of normal bone metabolism. Because 1α,25D3 plays an important role in osteogenesis, vitamin D formulations, such as alfacalcidol (ALF) and eldecalcitol (ELD), are used for treating osteoporosis. While it is known that, in contrast to ALF, ELD is an active ligand that directly acts on bone, the reason for its superior osteogenesis effects is unknown. Cyp27b1-knockout mice (Cyp27b1−/− mice) are congenitally deficient in 1α,25D3 and exhibit marked hypocalcemia and high parathyroid hormone levels, resulting in osteodystrophy involving bone hypocalcification and growth plate cartilage hypertrophy. However, because the vitamin D receptor is expressed normally in Cyp27b1−/− mice, they respond normally to 1α,25D3. Accordingly, in Cyp27b1−/− mice, the pharmacological effects of exogenously administered active vitamin D derivatives can be analyzed without being affected by 1α,25D3. We used Cyp27b1−/− mice to characterize and clarify the superior osteogenic effects of ELD on the bone in comparison with ALF. The results indicated that compared to ALF, ELD strongly induces ECaC2, calbindin-D9k, and CYP24A1 in the duodenum, promoting Ca absorption and decreasing the plasma concentration of 1α,25D3, resulting in improved osteogenesis. Because bone morphological measurements demonstrated that ELD has stronger effects on bone calcification, trabecular formation, and cancellous bone density than ALF, ELD appears to be a more effective therapeutic agent for treating postmenopausal osteoporosis, in which cancellous bone density decreases markedly. By using Cyp27b1−/− mice, this study was the first to succeed in clarifying the osteogenic effect of ELD without any influence of endogenous 1α,25D3. Furthermore, ELD more strongly enhanced bone mineralization, trabecular proliferation, and cancellous bone density than did ALF. Thus, ELD is expected to show an effect on postmenopausal osteoporosis, in which cancellous bone mineral density decreases markedly.

In living organisms, 1α,25D 3 is transported in the plasma when bound with vitamin D-binding protein (DBP) to reach target tissues [1], such as the bones, kidneys, parathyroid glands and small intestine, where it binds to the vitamin D receptor (VDR), which belongs to the intranuclear steroid hormone receptor super family [2]. VDR, when bound to 1α,25D 3 , forms a dimer with retinoid X receptor before binding with vitamin D responsive element (VDRE) in the target gene promoter to mutually interact with various transcriptional coactivators and basal transcription factors to activate promoters inducing target gene expression [3][4][5]. Thus, 1α,25D 3 plays important roles in balancing Ca metabolism in living organisms by modulating target gene expression and promoting active Ca absorption in the intestinal tract as well as Ca reabsorption in the kidneys, osteogenesis, and bone resorption [6]. The plasma 1α,25D 3 concentration is normally maintained in the range of 20-70 pg/mL by CYP27B1. Plasma Ca, P, parathyroid hormone (PTH) or calcitonin level induce CYP27B1 expression in the kidneys, increasing 1α,25D 3 production [7]. In turn, 1α,25D 3 inhibits VDR-mediated CYP27B1 expression at the transcription level to create a negative feedback loop [8].
Thus, the production volume of 1α,25D 3 is strictly controlled. Not only does 1α,25D 3 promote active Ca absorption from the intestinal tract, it also inhibits PTH secretion [9], suppresses bone resorption, and promotes osteogenesis in order to maintain normal bone remodeling and bone quantity. However, as 1α,25D 3 strongly promotes intestinal Ca resorption, when administered in pharmacological quantities, there is a risk of hypercalcemia [10]. Therefore, alfacalcidol (ALF), a prodrug of 1α,25D 3 , is mainly used clinically and has long been the most widely used prodrug for this purpose in Japan. Because ALF is metabolized in the liver to 1α,25D 3 via 25-hydroxylation, it has been used clinically as a therapeutic agent for treating diseases involving abnormal vitamin D metabolism, such as chronic kidney failure and vitamin D-dependent rickets. In addition, it has become widely applied as a vitamin D formulation for osteoporosis. ALF reportedly increases bone density, inhibits bone fractures, and suppresses the breakdown of the trabecular structure [11][12][13].
In recent years, vitamin D formulations improving bone metabolism have been developed for the treatment of osteoporosis. Eldecalcitol (ELD) was discovered through in-vivo screening using ovariectomized rats (OVX rats) as an osteoporosis model animal.
ELD is a compound with a functional group introduced at 2β of 1α,25D 3 [14]. Unlike ALF, ELD is not need to be metabolized; it is an active vitamin D formulation in which the compound itself acts as an active ligand [15]. ELD exhibits lower "binding affinity " for VDR than 1α,25D 3 , but stronger binding affinity to DBP [16,17]. As this strong binding affinity to DBP increases stability in the plasma, it features a long plasma half-life [17,18]. Its PTH secretion-inhibitory effects reportedly are weaker than those of 1α,25D 3 [19].
In elderly rats and OVX rats, ELD increases thoracic vertebra bone density and bone strength [18]. Further, ELD increases bone quantity and osteogenesis speed in normal and OVX rats [20]. These studies using osteoporotic animals have shown that ELD offers superior bone quantity-increasing effects. Clinical experiments have indicated that ELD inhibits bone resorption in a dosage-dependent manner without affecting osteogenesis, thereby increasing thoracic vertebra bone density [21]. However, the mechanism underlying the superior osteogenic effects of ELD remains unclear. In particular, because it has been reported that the plasma 1α,25D 3 concentration decreases upon ELD administration, it is unknown whether its superior osteogenic effects are direct effects or are mediated by decreased plasma 1α,25D 3 [22].
Cyp27b1 -/mice have been bred by multiple research teams and are used in studies of the physiological functions of CYP27B1 [23][24][25][26]. Cyp27b1 -/mice exhibit marked hypocalcemia and high PTH levels after weaning. This results in decreased growth, with osteodystrophy involving bone hypocalcification and growth plate cartilage hypertrophy [23][24][25][26]. In Cyp27b1 -/mice, although VDR is expressed normally, 1α,25D 3 deficiency occurs because 1α-hydroxylation of 25D 3 by CYP27B1 does not occur. Accordingly, Cyp27b1 -/mice are an effective animal model for analyzing the physiological actions of exogenously administered active vitamin D derivatives in an environment without endogenous 1α,25D 3 . Therefore, to clarify the osteogenic effects of ELD in an environment not affected by endogenic 1α,25D 3 , we used Cyp27b1 -/mice in this study. In addition, we investigated the characteristics of ELD osteogenesis in comparison with ALF, which is clinically applied as a therapeutic agent for osteoporosis.

Rearing conditions and body-weight measurements
All mice were weaned at 3 weeks of age. After weaning, they were allowed to freely feed on a diet of solid food (Diet 11+D) and deionized water. The mice were raised in an enclosure with controlled temperature and humidity (23 ± 1°C, 60 ± 1%). The mice were raised until 9 weeks of age (administration period: 6 weeks). Body weight was measured once per week starting directly after weaning (3, 4, 5, 6, 7, 8 and 9 weeks of age).

ALF and ELD dosing solutions
ALF and ELD dosing solutions were formulated by dissolving 100 μL of ethanol containing 10 μg/mL of ALF or ELD into 7.9 mL of a medium-chain triglyceride solution. The vehicle dosing solution was prepared similarly using 100 μL of ethanol solution. The solutions were administered orally 3 times per week to achieve a dosage of 0.25 μg/kg body weight.

Reverse-transcription quantitative PCR
Total RNA was isolated from the mouse duodenum and liver using Isogen (Nippon Gene, Tokyo, Japan), in accordance with the manufacturer's protocol. First-strand cDNA was generated using AMV reverse transcriptase (Takara Bio, Kusatsu, Japan).

Soft X-ray imaging of the femur and tibia
Mice were anesthetized and then euthanized by means of cardiac plasma collection.
The femurs and tibias were collected. SOFTEX soft X-ray imaging equipment (CMB-2; SOFTEX, Kanagawa, Japan) was used to acquire soft X-ray images of the femurs and tibias.

Femur and bone morphological measurements
Parameters related to bone structure, bone shape, and bone resorption were analyzed in non-decalcified specimens of femurs obtained from female mice after 6 weeks of administrations. Total bone density, cortical bone density, and cancellous bone density of the femoral metaphysis, as well as cortical bone density and cortical bone thickness in the diaphysis were measured by peripheral quantitative computed tomography (pQCT). Horizontal, vertical and twisting bone strength in the femur were measured by pQCT. Two-dimensional (2D) and three-dimensional (3D) structural analyses of femoral trabeculae were conducted by micro-CT. The pQCT and micro CT analyses were performed at Kureha Special Laboratory (Fukushima, Japan).

Alcian blue staining
ATDC5 cells were plated in 12-multiwell plates and cultured in chondrogenic medium with 10 -7 -10 -10 M 1α,25D 3 , ALF, ELD, or 10 -6 -10 -9 M 25D 3 using the method described above. The cells were rinsed with PBS and fixed in 95% methanol for 20 min, and then stained with 0.1% alcian blue 8GX (FUJIFILEM Wako Pure Chemical) in 0.1 M HCl overnight. The stained cells were rinsed with PBS three times and observed by microscopy.

Transfection and luciferase activity assays
To asses the activity via the VDRE, ATDC5 cells (2 × 10 5 ) were suspended in 2 mL of medium and transfected with 0.5 μg of pGVB2-rat Cyp24a1 luciferase reporter vector

Role of vitamin D derivatives in megalin-knockdown cells
Stealth

Statistical analysis
Data are expressed as the mean ± SEM. The number of specimens is n = 5. Means were compared using an unpaired Student's t-test or Dunnett's test and Tukey-Kramer's honestly significant difference (HSD) test. P < 0.05 was considered significant.

Effects of ALF or ELD administration on body weight changes and calcium metabolism in Cyp27b1-knockout mice
We monitored body weight in Cyp27b1 +/+ and Cyp27b1 -/mice that were administered ALF, ELD, or vehicle for 6 weeks after weaning. From five weeks of age, Cyp27b1 -/mice exhibited significantly lower body weight than the Cyp27b1 +/+ mice.
In contrast, in Cyp27b1 -/mice administered ALF or ELD, body weight increased to levels similar to those observed in Cyp27b1 +/+ mice (Fig 1A). Although a tendency for hypocalcemia was observed in Cyp27b1 -/mice treated with vehicle, the administration of ALF or ELD significantly increased plasma Ca concentration. In the ALF and ELD administration groups, both Cyp27b1 +/+ and Cyp27b1 -/mice exhibited normal or slightly high calcium Ca concentration as of 6 weeks of age ( Fig 1B). Plasma PTH concentration was markedly higher in Cyp27b1 -/vehicle-administered mice than in Cyp27b1 +/+ vehicle-administered mice. In Cyp27b1 -/mice administered ALF or ELD, plasma PTH was significantly decreased as compared to vehicle-treated mice, and at levels similar to those noted for Cyp27b1 +/+ mice ( Fig 1C). Next, we measured the duodenal mRNA expression of ECaC2 and calbindin-D 9 k, which encode calcium metabolism-modulating proteins. In Cyp27b1 +/+ mice administered ELD, mRNA expression levels of ECaC2 and calbindin-D 9 k in the duodenum are significantly increased as compared to those in vehicle-and ALF-administered mice. In Cyp27b1 -/mice administered ALF or ELD, mRNA expression levels of both genes tended to increase (Fig 1D,E). In addition, we measured mRNA expression levels of CYP2R1 and CYP27A1, which encode enzymes involved in vitamin D metabolism in the liver.
CYP2R1 and CYP27A1 mRNA expression levels did not exhibit any significant changes as a result of ALF or ELD administration (Fig 1F, G). Both Cyp27b1 +/+ and Cyp27b1 -/mice administered ALF or ELD exhibited significantly increased CYP24A1 mRNA expression compared to the respective vehicle-treated groups. Particularly marked increases were noted in the ELD administration groups (Fig 1H).

Cyp27b1-knockout mice administered ALF or ELD
The femurs and tibias of Cyp27b1 -/mice administered vehicle were shorter than those of Cyp27b1 +/+ mice administered vehicle, and hypertrophy was noted in the epiphyseal region. In contrast, the external appearance of the bones of Cyp27b1 -/mice that were administered ALF or ELD was nearly the same as that of the bones of Cyp27b1 +/+ mice (Fig 2A). Soft X-ray images of femurs and tibias demonstrated that in the Cyp27b1 -/mice administered vehicle, the center was black and hollow compared to that in Cyp27b1 +/+ mice, with an unclear shape, indicating hypocalcification. White, clear X-ray images similar to those obtained for Cyp27b1 +/+ mice were observed in the Cyp27b1 -/mice administered ALF or ELD, suggesting that the hypocalcification of Cyp27b1 -/mice had been improved. In addition, although the femoral bones were significantly shorter in Cyp27b1 -/mice administered vehicle than in Cyp27b1 +/+ mice administered vehicle, the bones were longer in Cyp27b1 -/mice administered ALF or ELD, with lengths being similar to those in Cyp27b1 +/+ mice (Fig 2B).

Histological staining of the femurs of Cyp27b1-knockout mice administered ALF or ELD
Femur calcification in Cyp27b1 +/+ and Cyp27b1 -/mice was evaluated using von Kossa staining. Areas stained black are calcification sites in which Ca has been deposited. In Cyp27b1 -/mice administered vehicle, there were less Ca sedimentation sites in both cortical and cancellous bone than in Cyp27b1 +/+ mice, suggesting hypocalcification. Bone calcification to the same degree as that seen in Cyp27b1 +/+ mice was observed in Cyp27b1 -/mice administered ALF or ELD (Fig 3A). Cartilage formation was evaluated in Cyp27b1 +/+ and Cyp27b1 -/mice by using toluidine blue staining. Areas stained purple are cartilage cells. In Cyp27b1 -/mice administered vehicle, cartilage in the epiphyseal region and growth plate was markedly increased compared to Cyp27b1 +/+ mice administered vehicle, with hypertrophy and irregularity noted in the growth-plate cartilage layer. The hypertrophy and irregularity in the growth-plate cartilage area were completely resolved by ALF or ELD administration ( Fig 3B). Villanueva staining was used to evaluate osteoid formation in Cyp27b1 +/+ and Cyp27b1 -/mice. Areas stained reddish purple are osteoid. No differences in osteoid quantity were noted between Cyp27b1 +/+ mice treated with vehicle, ALF, or ELD.
However, osteoid tended to be present in a larger amount in Cyp27b1 -/mice administered vehicle. In contrast, osteoid quantity decreased to the level observed in Cyp27b1 +/+ mice upon administration of ALF or ELD (Fig 3C). We evaluated bone formation speed in femurs of Cyp27b1 +/+ and Cyp27b1 -/mice by double calcein labeling. Double calcein labeling was clearly observed in Cyp27b1 +/+ mice administered vehicle, ALF, or ELD, indicating that bone formation was occurring normally. In contrast, calcein staining spread throughout the entire bone in Cyp27b1 -/administered vehicle, demonstrating that bone formation did not occur at a normal speed. This abnormal bone formation speed was improved to the level in Cyp27b1 +/+ mice in the ALF or ELD administration groups of Cyp27b1 -/mice ( Fig 3D).
In both Cyp27b1 +/+ and Cyp27b1 -/mice, overall bone density in the femoral metaphysis was significantly increased by ELD administration (Fig 4A). Although cortical bone density was significantly lower in Cyp27b1 -/mice administered vehicle than in Cyp27b1 +/+ mice administered vehicle, it increased significantly in the ALF and ELD administration groups, to the levels observed in Cyp27b1 +/+ mice (Fig 4B).
Cancellous bone density increased significantly in both Cyp27b1 +/+ and Cyp27b1 -/mice by administration of ELD and ALF. (Fig 4C). Femoral diaphysis cortical bone density and thickness were significantly lower in Cyp27b1 -/than in Cyp27b1 +/+ mice administered vehicle, but they increased to the level observed in Cyp27b1 +/+ mice upon administration of ALF or ELD. In Cyp27b1 -/mice administered vehicle, although horizontal bone strength, vertical bone strength, and twisting bone strength were significantly lower than in Cyp27b1 +/+ mice, bone strength increased to approximately the same level as that in Cyp27b1 +/+ mice in the ALF and ELD administration groups ( Fig 4D, E, F). Bone strength increased particularly upon ELD administration.

Cyp27b1-knockout mice administered ALF or ELD
2D micro-CT image analysis revealed a lack of trabecular bone formation in the Cyp27b1 -/mice administered vehicle, with a markedly small amount of calcified bone, which appeared white on micro-CT images. In contrast, ALF or ELD administration restored trabecular bone formation to the level observed in Cyp27b1 +/+ mice. In particular, cancellous bone formation was stronger in both Cyp27b1 +/+ and Cyp27b1 -/mice administered ELD than in Cyp27b1 +/+ mice administered vehicle (Fig 5A). 3D image analysis revealed poor cancellous bone formation and hypocalcification in Cyp27b1 -/mice administered vehicle. In contrast, trabecular bone formation and calcification were at the levels as observed in Cyp27b1 +/+ mice in Cyp27b1 -/mice treated with ALF or ELD. In particular, the ELD administration group of not only Cyp27b1 -/-, but also Cyp27b1 +/+ mice exhibited clearly enhanced trabecular bone formation ( Fig 5B).  (Fig 6B). We found that vitamin D congeners and derivatives inhibited mesenchymal cell differentiation into chondrocytes.

Effects of ALF or ELD on VDR-dependent transcriptional activity in ATDC5 cells
ATDC5 cells were used to evaluate the effects of the vitamin D derivatives on CYP24A1 transcriptional activity. 1α,25D 3 , used as a control, increased the transcriptional activity of CYP24A1 in a concentration-dependent manner.
Transcriptional activity was not observed upon treatment with ALF and 25D 3 , but marked transcriptional activity was induced by ELD at 10 -8 M (Fig 7A). Next, the presence or absence of DBP on VDR mediated transcriptional activity was evaluated.
The effect of ALF was not influenced by DBP, but 1α,25D 3 increased VDR-mediated transcription activity, depending on the concentration of DBP. However, the transcriptional activity induced by ELD and 25D 3 decreased depending on the concentration of DBP (Fig 7B). Since ELD and 25D 3 bind strongly to DBP, we evaluated the involvement of megalin, an uptake pathway of 25D 3 bound to DBP. In cells in which megalin was knocked down by approximately 90% through RNAi ( Fig   7C), the effects of 1α,25D 3 and ALF were not affected, while VDR-mediated transcription activity induced by ELD and 25D 3 was significantly decreased (Fig 7D) .

Discussion
We used Cyp27b1 -/mice, an effective model for analyzing the action of vitamin D derivatives without any influence of endogenous 1α,25D 3 , to investigate the effects of ALF and ELD administration on vitamin D metabolism. First, we evaluated the effects of ALF or ELD administration on the delayed growth observed in Cyp27b1 -/mice. The administration of ALF or ELD restored the body weight of Cyp27b1 -/mice to approximately the level in Cyp27b1 +/+ mice. The hypocalcemia observed in Cyp27b1 -/mice was also improved and plasma calcium was maintained at a normal level by ALF or ELD administration. In both Cyp27b1 +/+ and Cyp27b1 -/mice, the plasma Ca concentration in the ELD-treated groups was relatively high, but within the normal range. In Cyp27b1 -/mice, 1α,25D 3 deficiency and hypocalcemia caused secondary hyperparathyroidism, with plasma PTH reaching markedly high levels [23,26]. When ALF and ELD were administered, plasma PTH decreased markedly. In a previous study, when 1α,25D 3 was administered to Cyp27b1 -/mice, hyperparathyroidism was improved [27]. ALF and ELD both correct hyperparathyroidism via the same process [27]. By promoting active Ca absorption in the intestinal tract and kidneys, 1α,25D 3 maintains Ca homeostasis in the plasma. Active Ca absorption occurs when Ca flows in through the epithelial Ca channel ECaC2, after which Ca is transported into cells by proteins such as calbindin-D 9k . The VDRE is found upstream of vitamin D-response gene promoters, and its expression is controlled at the transcriptional level by 1α,25D 3 [28,29]. Therefore, we measured the mRNA expression levels of ECaC2 and calbindin-D 9k in the duodenum to further investigate the Ca metabolism-modulating effects of ALF and ELD. In accordance with a previous report [30], administering ALF to mice did not induce ECaC2 or calbindin-D 9k expression in the intestinal tract.
However, ELD increased the mRNA expression of ECaC2 and calbindin-D 9k in both Cyp27b1 +/+ and Cyp27b1 -/mice, most likely because ELD is an active ligand that binds directly with VDR. Even in experiments using ATDC5 cells, ELD increased VDR-mediated transcriptional activity in a concentration-dependent manner, similar to 1α,25D 3 . (Fig 7A). These results demonstrated that the administration of ALF or ELD made it possible to maintain the plasma Ca in Cyp27b1 -/mice within the normal range, because it enhanced active Ca absorption mediated by ECaC2 and calbindin-D 9k expression in the duodenum.
It has been reported that ELD reduces plasma 1α,25D 3 and increases plasma 24,25D 3 in experimental animals [22]. This appears to be because ELD induces CYP24A1, which metabolizes 1α,25D 3 and 25D 3 into inactive forms. Thus, lowering the plasma When we investigated the potential influence of ALF and ELD on mRNA expression of CYP2R1 and CYP27A1 in the liver, no effects were noted. Because studies using Cyp27b1 -/mice raised on high-calcium feed and Vdr-knockout mice revealed that the expression of CYP2R1 and CYP27A1 is similar to that in Cyp27b1 +/+ mice [26], it appears that 1α,25D 3 and 25D 3 do not control CYP2R1 and CYP27A1 expression. The mRNA expression of CYP24A1 in the kidneys of Cyp27b1 -/mice was significantly lower than that in Cyp27b1 +/+ mice. However, CYP24A1 expression was strongly induced by ALF and ELD administration. Similar effects were noted in Cyp27b1 +/+ mice. Because VDRE is present upstream from the CYP24A1 promoter [31], it appears that, as with 1α,25D 3 , ELD interacts with the VDR to induce CYP24A1 expression.
Moreover, because of its high DBP bindability, ELD is stable in the plasma and circulates longer than 1α,25D 3 [32]. Therefore, it appears that ELD more strongly induces CYP24A1 expression than ALF. Like ELD, it has been reported that the compound 2α-hydroxypropoxy-1α,25D 3 (O2C3) having a 3-hydroxypropoxy group at the 2α position is more difficult to metabolize at CYP24A1 than 1α,25D 3 [33], which may have allowed for the prolonged effects of ELD. Based on the above, the significant effects of ELD that were noted even in Cyp27b1 +/+ mice indicated that ELD can promote Ca absorption and improve osteogenesis in patients with osteoporosis. By using Cyp27b1 -/mice in our experiment, we were able to demonstrate that ELD exhibits Ca metabolism-regulating effects without being influenced by endogenous 1α,25D 3 . A study in OVX rats showed that ELD is superior to 1α,25D 3 in terms of increasing lumbar vertebrae and femur bone density, and inhibiting bone resorption [24]. However, ELD administration decreases the plasma 1α,25D 3 concentration [25], and this decrease may mediate the action of ELD on bones. Previous studies using OVX rats were unable to exclude the influence of 1α,25D 3 to analyze the actions of ELD alone. Therefore, in order to clarify the actions of ELD alone on bone, excluding the effects of 1α,25D 3 , we analyzed the osteogenic actions of ALF and ELD in Cyp27b1 -/mice. Externally, the femurs and tibias in vehicle-treated Cyp27b1 -/mice were observed to have hypertrophy in the epiphyseal region and reduced bone length. However, the administration of ALF or ELD improved these abnormalities to a state similar to that observed in Cyp27b1 +/+ mice. Bone samples of Cyp27b1 -/mice exhibited hypocalcification, hypertrophy and irregularity in the growth-plate cartilage, and increased osteoid formation, as well as poor osteogenesis speed and clear lesions, indicating imperfect osteogenesis. These abnormalities were completely restored by ALF and ELD administration, and osteogenesis occurred at the level observed in Cyp27b1 +/+ mice. Osteogenic abnormalities involving bone length and bone density in Cyp27b1 -/mice can be sufficiently improved with Ca supplementation [26]. However, hypocalcification and growth-plate cartilage hypertrophy cannot be completely normalized with Ca supplementation alone. Accordingly, the fact that osteogenesis occurred at the same level as in Cyp27b1 +/+ mice due to ALF or ELD administration indicates that not only Ca metabolism was maintained, but 1α,25D 3 was metabolized and generated from ALF as well as the direct actions of ELD on the bone. Previously, it has been reported that when ATDC5 cells, which can differentiate into cartilage cells, are cultured to differentiate into cartilage cells, the addition of 1α,25D 3 markedly inhibits such differentiation [34]. We found that when ATDC5 cells were treated with not only 1α,25D 3 , but also ALF and ELD, ATDC5 cell proliferation and differentiation were inhibited in a dose-dependent manner (Fig 6). We also investigated whether the presence or absence of DBP affects VDR transcriptional activity. The results showed that while the effect of ALF was not at all affected by DBP, 1α,25D 3 increased VDR transcriptional activity when DBP was present. However, ELD and 25D 3 decreased VDR-mediated transcriptional activity in a concentration-dependent manner (Fig 7B).
Because ELD and 25D 3 are thought to strongly bind to DBP, we evaluated the involvement of megalin, an uptake route for 25D 3 bound to DBP. Cells in which megalin had been knocked down to approximately 90% using RNAi were found to not be affected by 1α,25D 3 or ALF. However, VDR-mediated transcriptional activity significantly decreased for ELD and 25D 3 (Fig 7D). Based on the above results, it appears that ELD strongly binds with DBP to act on cartilage cells mediating megalin so as to maintain their normal proliferation and differentiation. When we used non-decalcified specimens to measure parameters related to bone structure, osteogenesis, and bone resorption, we found that Cyp27b1 -/vehicle-treated mice exhibited decreased bone quantity, decreased trabecula width and number, abnormally widened trabecular spaces, and decreased calcification and osteogenesis.
However, ALF or ELD administration improved these abnormalities to the normal levels of these parameters measured in Cyp27b1 +/+ mice (S1 Table). In the Cyp27b1 -/-ELD administration group, in particular, bone resorption surface, osteoclast count, and osteoclast surface tended to decrease more than during ALF administration, and it was found that bone quantity and trabeculae were significantly increased, while the osteogenic rate was maintained. This suggests that ELD more strongly inhibits bone resorption than ALF, resulting in superior osteogenesis. Similarly, in a previous, the femurs of mice administered ELD exhibited decreased osteoclast counts [35]. Moreover, as the lumbar vertebrae of OVX rats administered ELD were reported to exhibit greater decreases in osteoclast surface than those administered ALF, ELD appears to have strong bone resorption-inhibitory effects [36]. ELD was found to promote femur cancellous bone density in both Cyp27b1 +/+ and Cyp27b1 -/mice more strongly than ALF. Moreover, the fact that the density-increasing action of ELD on cancellous bone was also observed in Cyp27b1 +/+ mice demonstrated that ELD exhibits particularly strong effects on cancellous bone. This may be due to differences in bone turnover between cortical and cancellous bone. Although no remodeling occurs in the cortical bone of mice and rats, bone resorption by osteoclasts and osteogenesis by osteoblasts occurs on the trabecular surface of cancellous bone, implying that remodeling occurs continuously [37]. This suggests that ELD directly acts on the osteoclasts and osteoblasts in cancellous bone, where more bone remodeling occurs than in cortical bone, to reduce the osteoclast count and inhibit bone resorption, thereby promoting osteogenesis by osteoblasts. The particularly strong effects of ELD compared to ALF may have been due to ELD not only having high bone selectivity [21] ( Fig 7A), but also higher DBP-bindability than ALF and a long plasma half-life [16][17][18][19].
Previously, it has been reported that 1α,25D 3 markedly inhibits c-Fos expression in osteoclast precursor cells, thereby suppressing osteoclast differentiation [38]. Therefore, ELD may also directly act on osteoclast precursor cells via this mechanism to inhibit bone resorption [39]. Accordingly, the direct action of ELD on osteoblasts is likely to have induced the expression of factors that promote osteogenesis, such as osteocalcin (Fig 8). In this study, we have not evaluated the effect of ELD on mini-modeling, a bone formation without prior bone resorption that has been shown in rat trabecular bone treated with ELD [40]. By using Cyp27b1 -/mice, it seems that the osteogenic effect of ELD on mini-modeling could be analyzed in more detail. Although bone strength was markedly decreased in Cyp27b1 -/mice treated with vehicle, the administration of ALF and ELD allowed these animals to maintain bone strength at a level similar to that of Cyp27b1 +/+ mice (Fig 4). In the Cyp27b1 -/-ELD administration group, in particular, bone strength tended to be more strongly increased than in the ALF administration group. In terms of factors related to bone strength, the trabecular structure is important in addition to bone density. Thinner trabeculae in cancellous bone result in more fragile bones. Using 2D and 3D micro-CT analyses of femoral trabeculae, we found that trabecular bone formation and calcification were markedly decreased in Cyp27b1 -/mice administered vehicle. However, both were maintained at the same level as in Cyp27b1 +/+ mice upon ALF and ELD administration.
In the ELD administration group in particular, trabecular bone formation was clearly enhanced in not only Cyp27b1 -/mice, but in also Cyp27b1 +/+ mice, demonstrating that ELD strongly increases trabecular formation and trabecular connectivity. These results suggest that ELD acts more strongly on cancellous bone, in which remodeling occurs and turnover is faster than in cortical bone, to promote osteogenesis and inhibit bone resorption. Osteoporosis, which commonly occurs in postmenopausal women, develops as decreases in female hormones that act to inhibit bone resorption and cause bone resorption to become relatively dominant. Because postmenopausal osteoporosis involves particularly marked decreases in the trabecular count in cancellous bone [41], ELD may be a very effective therapeutic agent for the disease due to this selective action on cancellous bone and promotion of bone density and trabecular bone formation.
We administered ELD to weaned mice and evaluated the effects on osteogenesis, revealing that it exhibits highly superior osteogenesis effects. This indicates that ELD is also an effective agent for treating vitamin D-dependent rickets type I.         (D) VDR-dependent transcriptional activity of CYP24A1 in megalin-knockdown ATDC5 cells. In megalin-knockdown ATDC5 cells, while 1α,25D 3 and ALF had no effect, ELD and 25D 3 significantly lowered CYP24A1 transcriptional activity.  Table S1. Measurement of bone morphology of Cyp27b1 +/+ or Cyp27b1 -/mice administered ALF or ELD for 3 or 6 weeks. *P < 0.05 vs. Cyp27b1 +/+ mice administered vehicle, Student's t-test. # P < 0.05 vs.