A nonsecosteroidal vitamin D receptor ligand with improved therapeutic window of bone efficacy over hypercalcemia

Authors


Abstract

Vitamin D3 analogues were shown to be beneficial for osteoporosis and other indications, but their narrow therapeutic window between efficacy and hypercalcemia has limited their clinical utility. A nonsecosteroidal, tissue-selective, orally bioavailable, vitamin D receptor (VDR) ligand was ascertained to be efficacious in bone while having modest calcemic effects in vivo. This compound (VDRM2) potently induced Retinoid X Receptor alpha (RXR)-VDR heterodimerization (EC50 = 7.1 ± 1.6 nM) and induced osteocalcin promoter activity (EC50 = 1.9 ± 1.6 nM). VDRM2 was less potent in inducing Ca2+ channel transient receptor potential cation channel, subfamily V, member 6 (TRPV6) expression (EC50 = 37 ± 12 nM). VDRM2 then was evaluated in osteopenic ovariectomized (OVX) rats and shown to dose-dependently restore vertebral bone mineral density (BMD) from OVX to sham levels at 0.08 µg/kg per day. Hypercalcemia was observed at a dose of 4.6 µg/kg per day of VDRM2, suggesting a safety margin of 57 [90% confidence interval (CI) 35–91]. 1α,25-dihydroxyvitamin D3 [1α,25(OH)2D], ED71, and alfacalcidol restored BMD at 0.030, 0.0055, and 0.046 µg/kg per day, respectively, whereas hypercalcemia was observed at 0.22, 0.027, and 0.23 µg/kg per day, indicating a safety margin of 7.3, 4.9, and 5.0, respectively (90% CIs 4.1–13, 3.2–7.7, and 3.5–6.7, respectively). Histomorphometry showed that VDRM2 increased cortical bone area and stimulated the periosteal bone-formation rate relative to OVX at doses below the hypercalcemic dose. By contrast, ED71 increased the periosteal bone-formation rate only above the hypercalcemic dose. VDRM2 suppressed eroded surface on trabecular bone surfaces at normal serum calcium dosage levels, suggesting dual anabolic and antiresorptive activity. In summary, vitamin D analogues were more potent than VDRM2, but VDRM2 had a greater safety margin, suggesting possible therapeutic potential. © 2010 American Society for Bone and Mineral Research

Introduction

Bone degenerative conditions such as osteoporosis occur in a substantial proportion of the elderly population. Osteoporosis encompasses a heterogeneous group of disorders that represents a major risk for bone fractures and a substantial burden on the health care system. Billions of dollars are spent annually on medical care for the treatment of osteoporosis. Clinically, osteoporosis is characterized by diminished bone mass (reduced bone mineral density) and loss of spatial architecture, resulting in decreased bone strength and increased risk of fracture.

While a number of antiresorptive agents, including calcitonin, bisphosphonates, estrogen, and selective estrogen receptor modulations (SERMs), prevent further bone loss, they do not rebuild bone once it has been lost.1, 2 The first US Food and Drug Administration (FDA)–approved anabolic bone-building agents for the treatment of osteoporosis is recombinant human parathyroid hormone(1-34) [rhPTH(1-34)], also known as teriparatide. Teriparatide builds bone mass, restores bone architecture, and reduces the risk of vertebral and nonvertebral bone fractures in osteoporotic patients who are at high risk of fracture.3 However, as a peptide, teriparatide requires daily injections, which may be difficult for many elderly patients.4 As a result, a need persists for a bone-building agent that is orally bioavailable with reduced side effects.

Previously, vitamin D analogues were shown to increase vertebral bone mineral density (BMD) and to prevent vertebral and nonvertebral fractures in postmenopausal women with osteoporosis and in women living in residences for the elderly.5–8 However, vitamin D analogues were shown to dose-dependently elevate calcium in sera and in urine, leading to concerns about hypercalcemia and hypercalciuria in humans5, 6, 9–13 and in animals.14–18

We hypothesized that a synthetic nonsecosteroidal organic molecule may have potential advantages over current therapies because it would be orally bioavailable, would be active in both women and men, would have the potential to build bone (especially cortical bone), and could have bone-quality advantages relative to bisphosphonates. Additionally, a vitamin D receptor ligand might have other potential benefits in the elderly, such as improved neuromuscular function and skin quality, anti-inflammatory/immunomodulation properties, and prevention of breast, colon, and prostate cancer.13, 19–27 We describe in vitro and in vivo characteristics of VDRM2, which has a greater safety margin in terms of bone efficacy versus hypercalcemia than traditional secosteroidal vitamin D analogues.

Methods

Cell culture, transfection, and receptor-binding assays

Functional assays were developed in an effort to understand characteristics of VDRM2 and vitamin D analogues. For the human retinoid X receptor-human vitamin D receptor (hRXR-hVDR) heterodimerization assay with human osteoblast-like osteosarcoma cells, SaOS-2 cells were maintained in DMEM (Gibco, Carlsbad, CA) supplemented with 10% fetal bovine serum (FBS; Gibco) and plated at 5000 cells per well in a 96-well plate. The next day, cells were transfected using 0.5 µL of FuGENE (Roche Diagnostics Corp., Indianapolis, IN), 100 ng of luciferase reporter vector pFR-LUC (Stratagene, Santa Clara, CA), and 10 ng each of pVP16-VDR-LBD and pGal4-RXRα-LBD expression vectors per well. Cells were treated with the ligand 24 hours after transfection, and luciferase activity was quantitated the next day using Steady-Glo luciferase detection reagent (Promega, Madison, WI).

Activation of the osteocalcin (Ocn) promoter by compounds were evaluated in a rat osteoblast-like osteosarcoma cell line (ROS17/2.8) stably expressing rat osteocalcin promoter (1.154 kb) fused to a luciferase reporter gene. The development of the stably transfected ROS17/2.8 cell line (RG-15) containing Ocn-Luc has been described.26 Confluent RG-15 cells maintained in DMEM/F-12 medium (3:1) containing 5% FBS and 300 µg/mL of G418 at 37°C were trypsinized (0.25% trypsin) and plated into white opaque 96-well cell culture plates (25,000 cells per well). After 24 hours, cells (in DMEM/F-12 medium containing 2% FBS) were treated with the indicated concentrations of the compounds. After 48 hours of treatment, the medium was removed, and cells were lysed with 50 µL of lysis buffer (Luciferase Reporter Assay System, Roche Diagnostics Corp.) and assayed for luciferase activity using the Luciferase Reporter Gene Assay Kit (Boehringer, Mannheim, Germany). Aliquots (20 µL) of cell lysates were pipetted into wells of white opaque microtiter plates (DYNEX Technologies, Chantilly, VA) and placed in an automated injection MLX microtiter plate luminometer. The luciferase reaction mix (100 µL) was injected sequentially into the wells. The light signals generated in the reactions were integrated over an interval of 2 seconds, and the resulting luminescence values were used as a measure of luciferase activity (relative units).

C2BBe1 cells (CRL-2101, American Type Culture Collection, Manassas, VA) were maintained in DMEM (ATCC 30-2002) supplemented with 10% FBS and insulin-transferrin-selenium-G supplement (Invitrogen, Carlsbad, CA). For TRPV6 expression analysis, cells were plated in 96-well tissue culture treated plates (40,000 per well) in differentiation medium (DMEM + 5% charcoal-stripped FBS and insulin-transferrin-sodium selenite supplement (ITS)). Cells were allowed to differentiate for 6 days, with medium replacement every other day, before compound treatment. The medium was removed, and serial dilutions of 1α,25(OH)2D3, alfacalcidol, ED71, and VDRM2 in differentiation medium were added. After 24 hours, the medium was removed, and cell plates were sealed and frozen at −80°C until assayed. The RNeasy System (Qiagen, Valencia, CA) was used for RNA extraction and reverse transcription was performed using a High-Capacity cDNA Archive Kit (Applied Biosystems, Foster City, CA). Quantitative real-time polymerase chain reaction (PCR) was performed using predesigned Taqman assays (Applied Biosystems) to measure human TRPV6 (Accession Number NM_018646.2) and GAPDH (Accession Number NM_002046.3) mRNAs.

Competitive binding was evaluated using the fluorescence polarization–based Polarscreen Vitamin D Receptor Competitor Assay (Invitrogen). After discussions with the vendor, the ligand-binding domain of the human vitamin D receptor (VDR; GST-VDR-LBD, Invitrogen) was substituted for the full-length VDR supplied in the kit. Following the manufacturer's protocol, concentration response curves of 1α,25(OH)2D3, alfacalcidol, ED71, and VDRM2 were combined with 0.7 nM of VDR-LBD and 1 nM of the proprietary fluorescent tight-binding VDR ligand Fluormone VDR Red (Invitrogen). After 2.5 hours of incubation at room temperature, plates were read on an EnVision 2103 using the Optimized Bodipy TMR FP Dual Emission Label set (Perkin Elmer, Waltham, MA). The Kd for the Fluormone VDR Red ligand was determined by saturation binding to be 0.37 nM. Ki was determined using classic one-site receptor-binding parameters.

Osteopenic ovariectomized (OVX) rat studies

Aged (6-month-old) virgin, virus-antibody-free, ovariectomized Sprague-Dawley rats (Harlan Sprague Dawley, Inc., Indianapolis, IN, USA) were maintained on a 12-hour light/dark cycle at 22 °C with ad libitum access to food (TD 89222 with 0.5% Ca and 0.4% P, Teklad, Madison, WI, USA) and water. Bilateral ovariectomies were performed, except for sham-OVX animals, which served as age-matched, sham-vehicle controls. OVX rats were allowed to lose bone for 1 month, weighed, and randomized into treatment groups (n = 8 per group). Sham- OVX controls (Sham) and OVX vehicle controls were orally administered vehicle (1% NaCMC, 0.5% sodium lauryl sulfate (SLS)), whereas experimental animals were orally administered VDRM2 (Fig. 1) or another molecule every day for the following 2 months, as indicated in specific figures and tables. Sera were collected by cardiac puncture under isofluane anesthesia 1 day after the last dose for analysis of serum Ca. Total serum Ca was analyzed using an automated clinical chemistry analyzer (Hitachi 912, Tokyo, Japan) in a Good Laboratory Practice (GLP) facility. Animals were necropsied (isoflurane, pneumothorax) 1 day after the last dose. All studies were approved by the animal health committee at Lilly to ensure compliance with NIH and institutional IACUC guidelines.

Figure 1.

Compounds. Chemical structures are shown for VDRM2 relative to ED71, alfacalcidol, and 1,25(OH)2D3, which are more conventional vitamin D3 analogues.

Tissue collection and QCT analyses

From the 8-week OVX rat studies, L4 to L6 vertebrae and left femora were excised, cleaned of soft tissue, and stored in 50% ethanol/saline at 4°C. Lumbar vertebrae and femora in 50% EtOH/saline (2F7124, Baxter, Deerfield, IL) were wrapped in Parafilm (American National Co., Greenwich, CT, USA) and centered with respect to the QCT gantry (Research M, Stratec, Pforzheim, Germany). A coronal scout scan of the vertebrae and distal femoral metaphysis was generated first in 2D for positioning before analyzing in 3D. Parameters analyzed included BMD (mg/mL), bone mineral content (BMC, mg), and cross-sectional area (mm2). In addition, tibiae were excised, defleshed, and then fixed in cold buffered formalin for 2 days and then changed to 70% ethanol for histomorphometric analyses.

Histomorphometry

The proximal tibial metaphysis (PTM) and the tibial shaft (TX) were dehydrated in graded ethanol, defatted in acetone, and embedded in methyl methacrylate. Longitudinal sections of the PTM 20 µm in thickness or cross sections of the tibial-fibular junction 30 µm in thickness were prepared for analysis. Measurements were performed on the entire marrow region within the cortical shell of the PTM between 1 and 4 mm distal to the growth plate–metaphyseal junction or the TX using an Image Analysis System (Osteomeasure, Inc., Atlanta, GA, USA). Bone area, perimeter, single- and double-labeling surfaces, and eroded surface on trabecular or cortical bones were measured, and trabecular number, thickness, mineral appositional rate, and bone-formation rate–surface reference (BFR/BS) were calculated as described previously.

Biomechanical analyses

For the 8-week OVX rat studies, biomechanical properties of the femoral midshaft were ascertained using three-point bending. The femoral length was measured using calipers (Mitotoyo, Tokyo, Japan), and load was applied midway between two supports 15 mm apart. Femora were positioned so that the loading point was about 7.5 mm proximal to the distal popliteal space, and bending occurred about the mediolateral axis. Specimens were loaded in a 37°C saline bath after being submerged for 2 minutes to allow for equilibration of temperature. Load-displacement curves were recorded at a cross-head speed of 0.17 mm/s using an MTS Model 1/S materials testing machine with TestWorks 4 software (MTS Corp., Minneapolis, MN, USA). Parameters analyzed included the ultimate load Fu, stiffness, work to failure (energy), and ultimate displacement.

Midshaft and proximal femur specimens were prepared separately, before testing. Ultimate load Fu for the femoral neck was measured by mounting the proximal half of the femur vertically in a chuck at room temperature and applying a downward force on the femoral head until failure.28 The ultimate load was measured as the maximum force sustained by the femoral neck and was considered to be an estimate of femoral neck strength. All tests were conducted using the materials testing machine and analyzed using TestWorks 4 software (MTS Corp.).

Mechanical properties of L5 vertebrae were analyzed after the posterior processes were removed and the ends of the centrum were made parallel using a diamond wafering saw (Buehler Isomet, Evanston, IL, USA). Vertebral specimens were loaded to failure in compression using the materials testing device and analyzed using TestWorks 4 software (MTS Corp.). Specimens were tested in a saline solution at 37°C, after equilibration. Parameters measured included the ultimate load Fu, stiffness, work to failure (energy), and ultimate displacement. Stiffness was calculated as the maximum slope of the load-displacement curve.

Statistics

Group differences were assessed using Dunnett's corrected t test, with statistical significance defined as p < .05 (JMP Version 5.1; SAS Institute, Inc., Cary, NC). Dose-response curves were estimated using either a polynomial or sigmoidal fit (SigmaPlot, Version 9.0, Systat Software, Inc., San Jose, CA). Confidence intervals for margin-of-safety estimates were computed using the parametric bootstrap procedure (S-PLUS, Version 7.0, Insightful Corp., Somerville, MA).

Results

Competitive binding

Competitive binding was evaluated for compounds (Fig. 1) using a fluorescence polarization–based assay using the ligand-binding domain of the human VDR (VDR-LBD). 1α,25(OH)2D3 potently competed for binding with a Ki of 3.34 nM (Table 1). Alfacalcidol, a prodrug form of 1α,25(OH)2D3, competed for binding with a Ki of 23.9 nM. ED71 potently competed for binding with a Ki of 2.1 nM. With a Ki of 57.8 nM, VDRM2 was not as potent as 1α,25(OH)2D3 or ED71 (Table 1).

Table 1. Compound Effects on VDR Binding, RXR-VDR Heterodimerization, Ocn-Luc Expression, and TRPV6 transactivation
 VDR Ki, nMRXR/VDR EC50, nMOCN EC50, nMTRPV6 EC50, nM
  1. Note: Competitive-binding Ki and EC50 values for RXR-VDR heterodimerization, Ocn-Luc expression, and TRPV6 transactivation were derived by interpolation of curve fit of dose-response curves for the indicated compounds. Competitive-binding Ki values represent an average Ki ± standard error for n = 2 independent assay runs for each compound. Each compound was evaluated in each cell-based assay in triplicate for n = 2 to 108. Cell-based data are represented as mean ± SD.

VDRM257.8 ± 27.1 ± 1.61.92 ± 1.637.2 ± 11.7
1α,25(OH)2D33.34 ± 0.30.39 ± 0.420.13 ± 0.140.6 ± 0.43
Alfacalcidol23.9 ± 6.232.5 ± 7.80.40 ± 0.622.8 ± 15.1
ED712.1 ± 0.47.6 ± 3.42.8 ± 0.515.5 ± 7.7

RXR-VDR heterodimerization assay

1α,25(OH)2D3 and analogues were confirmed to be high-affinity ligands that induce heterodimerization of VDR with RXR, resulting in the formation of a RXR-VDR heterodimer-ligand complex, which is a functional mediator of vitamin D action in cells (Table 1). The RXR-VDR heterodimerization assay was developed by cotransfecting human SaOS-2 cells with the expression vectors VP16VDR-LBD and Gal4RXR-LBD, which contained a Gal4-responsive reporter gene. In this cotransfection assay, 1α,25(OH)2D3, alfacalcidol, and ED71 induced RXR-VDR heterodimerization with EC50 values of 0.39, 32.5, and 7.6 nM, respectively (Table 1). VDRM2 compared favorably with the conventional vitamin D analogues and promoted RXR-VDR heterodimerization with an EC50 of 7.1 nM.

Rat osteocalcin luciferase (Ocn-Luc) assay

Osteocalcin (Ocn) is a bone-specific protein made by osteoblasts with a vitamin D response element (VDRE) in the promoter region of the gene. Therefore, VDRM2 was evaluated for the ability to induce VDRE-dependent Ocn gene expression using an assay consisting of a rat osteoblast-like osteosarcoma cell line (ROS 17/2.8) stably expressing a luciferase reporter gene construct fused to the rat osteocalcin promoter (1.154 kb). Specifically, confluent ROS 17/2.8 cells maintained in DMEM/F-12 medium for 24 hours were treated with increasing concentrations of compounds for 48 hours. Luciferase activity from lysed cells was measured by detection of luminescence. 1α,25(OH)2D3, alfacalcidol, and ED71 induced Ocn-Luc luminescence with EC50 values of 0.13, 0.4, and 2.8 nM, respectively (Table 1). VDRM2 compared favorably with the conventional vitamin D analogues and induced the expression of Ocn-Luc in Ros17/2.8 cells with an EC50 of 1.92 nM.

TRPV6 assay

Hypercalcemia was shown to be a complication for orally bioavailable 1α,25(OH)2D3 and analogues. Epithelial cell calcium channel 2 (TRPV6) was shown to participate in Ca uptake through the intestines. Therefore, a cell-based assay was developed to predict compound effects on intestinal Ca absorption by measuring the transactivation potential of compounds on TRPV6 in C2BBe1 cells. Differentiated C2BBe1 cells were incubated with compounds to ascertain the level of TRPV6 mRNA (Accession Number NM_018646) expression, as measured using RT-PCR. This mRNA expression from compound treated cells then was normalized to the housekeeping gene GAPDH and compared with levels from untreated cells. Therefore, a compound with a higher EC50 value in this assay may be less active in intestinal Ca absorption in vivo.

1α,25(OH)2D3, alfacalcidol, and ED71 had EC50 values of 0.6, 22.8, and 15.5 nM, respectively, in this assay, indicating that 1α,25(OH)2D3 potently induces TRPV6 expression in C2BBe1 cells (Table 1). VDRM2 had modest potency to induce TRPV6 gene expression in C2BBe1 cells with an EC50 value of 37.2 nM, which compared favorably with the vitamin D analogues. The Ocn-Luc and TRPV6 data suggested that VDRM2 might be efficacious on bone, with lower likelihood of inducing hypercalcemia in vivo.

Osteopenic OVX rat assay

In vivo efficacy of VDRM2 was evaluated in osteopenic OVX rats. Each assay contained at least two controls, including sham-OVX (sham) and OVX controls treated with vehicle. QCT analyses of vertebral BMD were evaluated over past assays to show historical BMD means of sham = 625 mg/mL (Fig. 2) and Ovx = 545 mg/mL (Fig. 3) for our model. In vivo assay conditions were relatively well controlled, with two assays having sham average values and ine assay having OVX average values outside the 95% prediction intervals. Therefore, the variability of the within-assay means across assays was consistent with what would be expected based on random animal-to-animal variation. This analysis enabled us to evaluate whether a specific assay ran properly with respect to historical controls and to decide if the data could be used for structure-activity-relationship decisions affecting compounds.

Figure 2.

Sham plot. Vertebral BMD values (mg/mL) for individual sham animals (○) are shown as a function of assay number. The horizontal dashed line corresponds to the mean BMD value for all assays (sham avg. = 625 mg/mL), whereas the solid line connects the sham averages (red triangles) measured for each individual assay. The region between the dotted horizontal lines corresponds to the 95% prediction interval for average values. This analysis was useful to understand within-assay and between-assay variation and to indicate whether an assay ran properly with respect to historical sham controls.

Figure 3.

OVX Plot. Vertebral BMD values (mg/mL) for individual OVX vehicle controls (○) are shown as a function of assay number. The horizontal dashed line corresponds to the mean BMD value for all assays (OVX avg. = 545 mg/mL), whereas the solid line connects the OVX averages (red triangles) measured for each individual assay. The region between the dotted horizontal lines corresponds to the 95% prediction interval for average values. This analysis was useful to understand within-assay variation and between-assay variation and to indicate whether an assay ran properly with respect to historical OVX controls.

We next looked for a biologically meaningful level of skeletal efficacy with which to evaluate compound effects, which we reasoned might be sham-level BMD. Therefore, bone efficacy determinations were evaluated for the ability of compounds to restore lost bone from OVX BMD to the sham-level BMD. Specifically, an efficacy measure was selected that was defined as the dose of a compound that produces a sham level of BMD effect relative to the vehicle (OVX) control called the threshold efficacy dose (TED).

In addition, serum Ca levels were evaluated for OVX rats, which were found to be a highly variable parameter. In an effort to understand hypercalcemia in our animals, we evaluated serum Ca for our historical series of OVX vehicle controls, as shown in Fig. 4. Serum Ca levels were found to fluctuate depending on the set of animals, when they were assayed, the method of blood collection, and the clinical chemistry laboratory. Therefore, these data are specific for sera collected by cardiac puncture in the morning and as assayed by the same laboratory. Serum Ca levels were more variable than BMD. Previously, hypercalcemia in humans was defined as exceeding the 97.5th percentile of normal controls (International Federation of Clinical Chemistry).29 Therefore, the same criterion was applied to our animal model, with the 97.5th percentile equal to 11.2 mg/dL. The dose that reached this Ca level was selected as the hypercalcemia TED for our compounds.

Figure 4.

Serum Ca. Serum Ca values measured for individual OVX controls (○) are shown for assays conducted over the past several years. The horizontal dashed line corresponds to the mean Ca of 10.3 mg/dL for all assays. The solid line connects the assay Ca averages, which are red (triangles). The upper dotted horizontal line is the 97.5th percentile of the historical data, which is 11.2 mg/dL and corresponds to the clinical definition of hypercalcemia (International Federation of Clinical Chemistry).

VDRM2 was evaluated in our 8-week OVX rat assay five times. Figure 5 shows the composite dose-response curve for the average vertebral BMD and serum Ca data. Polynomials were used to curve fit the data and then to calculate sham-level TED for BMD and hypercalcemic dose for serum Ca. VDRM2 was observed to potently restore bone mass to sham levels in a dose-dependent manner. VDRM2 had a sham-level TED of 0.081 µg/kg, whereas hypercalcemia in these animals was not observed until 4.6 µg/kg, indicating a safety margin of 57 [Ca dose/TED, 90% confidence interval (CI) 35–91]. The BMD dose-response curve was somewhat biphasic above 1 µg/kg, increasing above sham before decreasing, as hypercalcemia was observed.

Figure 5.

Skeletal and Ca effects of VDRM2 compared with vitamin D3 analogues in the 8-week OVX rat assay. Sprague Dawley rats were permitted to lose bone owing to ovariectomy for 1 month before dosing with compounds for the following 2 months. Data for VDRM2 are plotted in panel A, which represents the average of five assays. Panel B shows the 1,25(OH)2D3 average data for four assays, panel C shows the ED71 average data for two assays, and panel D shows the alfacalcidol average data for four assays. Left panels show BMD analyzed for the vertebra by QCT. The dashed line represents the historical sham-level BMD. The filled circles represent the average for several assays, which were curve fit (solid line) using a Sigma plot. The right panel shows serum Ca as obtained by cardiac puncture at necropsy 1 day after the last dose. The filled circles represent the average for several assays, which were curve fit (solid line) using a Sigma plot. The dashed line represents hypercalcemia for OVX vehicle controls, as analyzed in Fig. 3, for this model. Plotted are mean ± SEM. Polynomials or sigmoidal curves (SigmaPlot) were used to calculate sham-level TED for BMD and hypercalcemic dose for serum Ca, which are indicated with a line drawn to the ordinate axis for each compound.

Similar analyses were conducted for vertebral BMD and serum Ca from 1α,25(OH) 2D3, ED71, and alfacalcidol, as shown in Fig. 5B, C, and D, respectively. 1α,25(OH)2D3 had a sham BMD TED of 0.03 µg/kg, with hypercalcemia observed at 0.22 µg/kg, indicating a safety margin of 7.3 (90% CI 4.1–12.5). ED71 had a sham BMD TED of 0.0055 µg/kg, with hypercalcemia observed at 0.027 µg/kg, indicating a safety margin of 4.9 (90% CI 3.2–7.7). Alfacalcidol had a sham BMD TED of 0.046 µg/kg, with hypercalcemia observed at 0.23 µg/kg, indicating a safety margin of 5.0 (90% CI 3.5–6.7). Therefore, ED71, 1α,25(OH)2D3, and alfacalcidol were more potent than VDRM2 in ovariectomized rats, but in terms of safety margin, VDRM2 > 1α,25(OH) 2D3 ≥ alfacalcidol ≥ ED71.

Biomechanics, histomorphometry, and other data for OVX rats

A possibly more rigorous assessment of skeletal efficacy is bone strength, which was evaluated postnecropsy for osteopenic OVX rats (Table 2 through 5). Femoral neck and vertebrae may be the more important sites clinically, whereas the midshaft is a cortical bone site. Bone strength was defined as the maximum force (N) that the vertebrae, midshaft, or femoral neck could sustain before failure. Bone strength is thought to be a better surrogate than BMD (or BMC) to predict clinical efficacy in postmenopausal women with osteoporosis. OVX significantly reduced strength for vertebrae, midshaft and femoral neck. As shown in Table 2, VDRM2 restored strength for vertebrae, midshaft, and femoral neck from OVX to sham levels. Dose-dependent increases in vertebral strength (69%), midshaft strength (28%), and femoral neck strength (27%) were observed relative to OVX. A biphasic effect of VDRM2 was observed on midshaft strength, with some reduction in biomechanical efficacy at the highest dose, which was above the hypercalcemic dose of 4.6 µg/kg. Nevertheless, a robust increase in bone strength was observed for VDRM2 between 0.1 and 3 µg/kg, with strength levels that compared favorably with sham.

Table 2. Biomechanical Effects of VDRM2 on the Axial and Appendicular Skeleton of Osteopenic OVX Rats
VDRM2 (µg/kg/d)Vertebral strength (ultimate load, N)Midshaft strengthFemoral neck strength
  • Note: Sprague Dawley rats were permitted to lose bone owing to ovariectomy for 1 month before dosing with compounds for the following 8 weeks. Data are mean ± SEM for two assays. Significance with respect to OVX is indicated by

  • *

    (Dunnett's t test, p < .05).

Sham301 ± 17*134 ± 4*108 ± 3*
OVX203 ± 10116 ± 397 ± 3
0.01239 ± 16124 ± 3100 ± 3
0.03252 ± 19129 ± 3110 ± 4
0.10297 ± 22*125 ± 4110 ± 3
0.30326 ± 19*138 ± 4*115 ± 5*
1.00342 ± 19*143 ± 3*123 ± 4*
3.00337 ± 26*148 ± 6*119 ± 6*
10.00343 ± 23*146 ± 3*120 ± 4*
30.00312 ± 32*134 ± 4*110 ± 4

The biomechanical effects of 1α,25(OH)2D3, ED71, and alfacalcidol also were evaluated postnecropsy, as shown in Tables 3, 4, and 5, respectively. 1α,25(OH)2D3 dose-dependently increased vertebral strength (80%), midshaft strength (14%), and femoral neck strength (18%) relative to Ovx. ED71 dose-dependently increased vertebral strength (106%), midshaft strength (20%), and femoral neck strength (32%) relative to OVX. Alfacalcidol also dose-dependently increased vertebral strength (127%), midshaft strength (28%), and femoral neck strength (22%) relative to OVX. Therefore, all four compounds compared favorably with each other in the ability to restore bone strength to sham levels at all three sites at or below their respective hypercalcemic doses. However, alfacalcidol had biphasic effects on all three sites, with some reduction of biomechanical efficacy at the highest dose, which was above the hypercalcemic dose of 0.23 µg/kg.

Table 3. Biomechanical Effects of 1,25(OH)2D3 on the Axial and Appendicular Skeleton of Osteopenic OVX Rats
1,25(OH)2D3 (µg/kg/d)Vertebral strength (ultimate load, N)Midshaft strengthFemoral neck strength
  • Note: Sprague Dawley rats were permitted to lose bone owing to ovariectomy for 1 month before dosing with compounds for the following 8 weeks. Data mean ± SEM for three assays. Significance with respect to OVX is indicated by

  • *

    (Dunnett's t test, p < .05).

Sham271 ± 12138 ± 3*109 ± 3*
OVX197 ± 8124 ± 396 ± 2
0.0003170 ± 21124 ± 4101 ± 5
0.001188 ± 10122 ± 397 ± 4
0.00322 0 ± 8125 ± 299 ± 3
0.01247 ± 12125 ± 210 0 ± 2
0.03295 ± 18139 ± 3*113 ± 2*
0.1354 ± 21*141 ± 9*101 ± 9
Table 4. Biomechanical Effects of ED71 on the Axial and Appendicular Skeleton of Osteopenic OVX Rats
ED71 (µg/kg/d)Vertebral strength (ultimate load, N)Midshaft strengthFemoral neck strength
  1. Note: Sprague Dawley rats were permitted to lose bone owing to ovariectomy for 1 month before dosing with compounds for the following 8 weeks. Data mean ± SEM for three assays. Significance with respect to OVX is indicated by * (Dunnett's t test, p < .05).

Sham316 ± 13o138 ± 3o109 ± 3o
OVX213 ± 12122 ± 397 ± 2
0.001213 ± 13123 ± 398 ± 2
0.003275 ± 14o125 ± 3101 ± 3
0.01291 ± 17o134 ± 2o109 ± 2o
0.03360 ± 21o137 ± 3o118 ± 4o
0.1439 ± 16o147 ± 2o128 ± 3o
Table 5. Biomechanical Effects of Alfacalcidol on the Axial and Appendicular Skeleton of Osteopenic OVX Rats
Alfacalcidol (µg/kg/d)Vertebral strength (ultimate load, N)Midshaft strengthFemoral neck strength
  • Note: Sprague Dawley rats were permitted to lose bone owinf to ovariectomy for 1 month before dosing with compounds for the following 8 weeks. Data mean ± SEM for two assays. Significance with respect to OVX is indicated by

  • *

    (Dunnett's t test, p < .05).

Sham340 ± 21*136 ± 4*115 ± 3*
OVX244 ± 15121 ± 3103 ± 3
0.01268 ± 16128 ± 4104 ± 3
0.03339 ± 23*129 ± 2115 ± 3*
0.1445 ± 18*149 ± 3*125 ± 3*
0.3463 ± 24*155 ± 4*119 ± 4*
1.0555 ± 22*148 ± 5*126 ± 5*
3496 ± 18*136 ± 6*90 ± 5*

We attempted to ascertain the basis to the skeletal efficacy of VDRM2 in vivo by histomorphometry. The proximal tibial metaphysis and diaphysis from the 8-week OVX rat assay were evaluated by conventional bone histormophometry for nonhypercalcemic doses (Tables 6 and 7). Relative to age-matched vehicle-treated osteopenic OVX rats, VDRM2 dose-responsively increased trabecular bone area and trabecular number while reducing eroded surface. However, VDRM2 suppressed bone-formation rate and mineralizing surface on trabecular bone surfaces at normal serum calcium dose levels (<4.6 µg/kg). At the diaphysis, VDRM2 increased mean cortical width, percent cortical area, periosteal mineral apposition rate, periosteal mineralizing surface, periosteal bone-formation rate, and total cortical area relative to OVX controls. Therefore, VDRM2 appears to function primarily as an antiresorptive on trabecular bone surfaces but stimulates bone-formation activity on cortical bone.

Table 6. Static and Dynamic Trabecular Bone Histomorphometry of VDRM2 Effects on the Proximal Tibial Metaphysis from Osteopenic OVX Rats
GroupTrabecular boneMineralizing surface (%)Mineral apposition rate (µm/d)Formation rate/bone surface (µm/d × 100)Eroded surface (%)
Area (%)Thickness (µm)Number (#/mm)
  • Note: Sprague Dawley rats were permitted to lose bone owing to ovariectomy for 1 month before dosing with vehicle or VDRM2 at the indicated doses for the following 8 weeks. Data mean ± SEM. Significance with respect to OVX is indicated by

  • *

    whereas significance with respect to sham is indicated by

  • **

    (Dunnett's t test, p < .05).

Sham14.43 ± 1.3935.35 ± 1.564.08 ± 0.3623.27 ± 2.020.91 ± 0.0520.93 ± 1.820.36 ± 0.06
Ovx4.30 ± 1.62*40.06 ± 2.771.00 ± 0.28*33.04 ± 5.650.99 ± 0.0633.40 ± 5.95*1.53 ± 0.37*
0.03 µg/kg/d3.85 ± 0.53*34.78 ± 1.781.10 ± 0.13*35.94 ± 2.43*1.01 ± 0.03*36.29 ± 2.73*0.98 ± 0.15*
0.1 µg/kg/d4.93 ± 0.82*38.49 ± 3.161.27 ± 0.20*32.47 ± 3.85*1.01 ± 0.0532.08 ± 3.18*0.77 ± 0.16***
0.3 µg/kg/d9.28 ± 1.21***39.27 ± 1.832.34 ± 0.26***15.64 ± 1.78***0.92 ± 0.0814.23 ± 1.82*.**0.94 ± 0.59
1 µg/kg/d11.50 ± 0.81***37.42 ± 1.783.06 ± 0.12***13.01 ± 0.75***0.87 ± 0.0611.36 ± 1.08***0.33 ± 0.05**
3 µg/kg/d11.61 ± 1.70**41.32 ± 2.65*2.72 ± 0.36***15.12 ± 4.47**0.80 ± 0.04**11.28 ± 2.59***0.44 ± 0.09**
Table 7. Cortical Bone Histomorphometry of VDRM2 Effects on the Tibial Diaphysis from Osteopenic OVX Rats
GroupTotal cortical area (mm2)Mean cortical width (mm)Percent cortical area (%)Periosteal mineral apposition rate (µm/d)Periosteal mineralizing surface (%)Periosteal bone-formation rate/bone surface (µm/d × 100)Endocortical bone-formation rate/bone surface (µm/d × 100)
  • Note: Sprague Dawley rats were permitted to lose bone owing to ovariectomy for 1 month before dosing with compounds for the following 8 weeks. Data mean ± SEM. Significance with respect to OVX is indicated by

  • *

    whereas significance with respect to sham control is indicated by

  • **

    (Dunnett's t test, p < .05).

Sham5.17 ± 0.17743.78 ± 6.2584.76 ± 0.650.52 ± 0.0739.73 ± 10.5724.09 ± 7.892.08 ± 0.37
OVX5.13 ± 0.14721.19 ± 6.77**82.57 ± 0.61**0.54 ± 0.0435.05 ± 5.8519.94 ± 3.7514.60 ± 2.82**
0.03 µg/kg/d5.38 ± 0.09732.09 ± 10.7082.66 ± 0.54**0.56 ± 0.0434.88 ± 5.2420.29 ± 3.3711.46 ± 3.43**
0.1 µg/kg/d5.39 ± 0.12758.46 ± 10.13*84.03 ± 0.29*0.69 ± 0.05***40.93 ± 5.7428.99 ± 4.527.97 ± 2.68***
0.3 µg/kg/d5.22 ± 0.14751.73 ± 11.62*84.88 ± 0.61*0.65 ± 0.0831.63 ± 8.2424.12 ± 9.278.27 ± 4.07**
1 µg/kg/d5.55 ± 0.19*778.13 ± 12.82***84.93 ± 0.49*0.65 ± 0.02*47.99 ± 6.2931.74 ± 4.8217.42 ± 6.59**
3 µg/kg/d5.34 ± 0.14761.59 ± 9.83*84.96 ± 0.48*0.70 ± 0.03***52.88 ± 5.51*37.38 ± 4.99*15.04 ± 3.26**

VDRM2 was well tolerated with no significant clinical signs observed in these animals; however, dose-dependent reduction in body weight from OVX to sham levels was observed for 10 µg/kg (Table 8). This loss in body weight was due to a reduction in fat mass from OVX to sham levels for 3 to 10 µg/kg. No effect of this compound on lean mass was observed. Soft tissue mineralization was not observed in these animals dosed up to 30 µg/kg based on histologic evaluation of the heart, aorta, stomach, and kidneys postnecropsy (data not shown). These data, taken together, showed that VDRM2 restored bone mass, spatial architecture, and bone strength in osteopenic OVX rats. In addition, VDRM2 was safe and well tolerated without significant clinical untoward effects in our studies.

Table 8. Other Effects of VDRM2 on Osteopenic OVX Rats
VDRM2 (µg/kg/d)Body weight (g)Fat mass (g)Lean mass (g)
  • Note: Sprague Dawley rats were permitted to lose bone owing to ovariectomy for 1 month before dosing with compounds for the following 8 weeks. Data mean ± SEM. Significance with respect to OVX is indicated by

  • *

    (Dunnett's t test, p < .05).

Sham314 ± 8*46.4 ± 3.4*223 ± 5
OVX358 ± 876.4 ± 6.4232 ± 3
0.01372 ± 1871.3 ± 10.7247 ± 7
0.03361 ± 873.4 ± 4.8236 ± 5
0.1364 ± 971.9 ± 7.0241 ± 10
0.3348 ± 865.8 ± 3.4232 ± 6
1.0355 ± 863.4 ± 6.5242 ± 6
3334 ± 942.7 ± 3.2*240 ± 7
10320 ± 9*35.2 ± 3.5*234 ± 6

Discussion

The osteopenic OVX rat is a model of postmenopausal osteoporosis that has been predictive of clinical efficacy for emerging pharmacologic agents and is required by regulatory agencies for consideration of new therapies.30, 31 In rodent models, vitamin D analogues have been shown to increase bone mass, improve spatial architecture, and strengthen bones.14–17, 32 Histomorphometry showed that vitamin D analogues can stimulate bone-formation activity on cortical and trabecular bone surfaces; however, bone-formation activity was best demonstrated at hypercalcemic doses.14, 15 Similarly in clinical studies, skeletal efficacy was demonstrated for vitamin D analogues, including reduction of nonvertebral fractures,5–7 but the realization of bone efficacy in the absence of hypercalcemia or hypercalciuria has been difficult to achieve.6–10, 33

We hypothesized that nonsecosteroidal vitamin D receptor ligands could be synthesized to preferentially modulate skeletal efficacy over hypercalcemic effects in vivo. Additionally, we hypothesized that a synthetic compound could have other potential benefits to the elderly, such as improved neuromuscular function, anti-inflammatory/immunomodulation properties, inproved skin quality, and prevention of breast, colon, and prostate cancer.13, 19–27 Therefore, compounds were evaluated for the ability to induce heterodimerization of VDR with RXR, resulting in the formation of a RXR-VDR heterodimer-ligand complex, which is a functional mediator of vitamin D action in cells. Alternatively, an RXR ligand will compete for RXR and disrupt RXR-VDR heterodimerization by driving the formation of the RXR-RXR homodimer. Therefore, specificity to potently drive RXR-VDR heterodimerization was achieved by counterscreening and eliminating compounds that potently drove RXR-RXR homodimerization (data not shown). Additionally, we evaluated compounds for the ability to stimulate expression of bone genes (osteocalcin) in a manner that compares favorably with 1α,25(OH)2D3 and ED71 in osteoblast-like cells (ROS 17/2.8). Finally, compounds were evaluated with a cell-based assay that may predict compound effects on intestinal Ca absorption by measuring the transactivation potential on TRPV6 in C2BBe1 cells. Interestingly, a novel compound was found that compared favorably with vitamin D analogues in promoting RXR-VDR heterodimerization and inducing expression of osteocalcin while having modest affinity for VDR and modestly affecting TRPV6 expression.

1α,25(OH)2D3, alfacalcidol, and ED71 are vitamin D analogues that have been investigated clinically.10 We attempted to find a synthetic molecule with a greater safety margin in terms of bone efficacy versus hypercalcemia than conventional vitamin D analogues. While the in vitro data looked interesting, in vivo efficacy was characterized by using the osteopenic OVX rat model. The in vivo data showed that VDRM2 may be a vitamin D receptor modulator (VDRM) with impressive bone efficacy, in the absence of hypercalcemia, that compares favorably with 1α,25(OH)2D3, alfacalcidol, and ED71 in terms of margin of safety. However, reviewers have pointed out correctly that we have not actually demonstrated that this compound is a VDRM, nor have we fully elucidated its mechanism of action; therefore, additional studies are under way.

During the course of these evaluations, we ascertained the importance of longitudinal evaluation of in vivo assays because the occasional assay will yield data that are not interpretable. Longitudinal assessment showed that the occasional assay yielded sham or OVX controls that were outside historical control limits. In addition, because of the wide variance observed for sham and OVX control animals, we found it necessary to repeat in vivo assays to understand reproducibility. Failure to evaluate assays longitudinally and repeat assays could lead to erroneous decisions about compound efficacy and rank order in a structure-activity-relationship analysis.

Reproducible analyses of VDRM2 in the 8-week OVX rat assay showed that it potently restored bone mass in the lumbar vertebrae, which is a trabecular bone site, with a sham-level TED of 0.081 µg/kg. At doses above 0.03 µg/kg, VDRM2 potently restored bone strength to sham levels or above in the vertebrae, femoral neck, and femoral midshaft, with the latter being a cortical bone site. Hypercalcemia in these animals was not observed until 4.6 µg/kg, indicating a therapeutic safety margin of 57-fold between bone efficacy and hypercalcemia. Histomorphometry showed that VDRM2 increased periosteal bone-formation rate, resulting in increased cortical bone properties, while functioning as a weak antiresorptive on trabecular bone surfaces compared with bisphosphonates (data not shown).

In comparative studies, 1α,25(OH)2D3 had a sham BMD TED of 0.03 µg/kg, with hypercalcemia observed at 0.22 µg/kg, indicating a safety window of 7.3. ED71 had a sham BMD TED of 0.0055 µg/kg, with hypercalcemia observed at 0.027 µg/kg, indicating a safety margin of 4.9. Alfacalcidol had a sham BMD TED of 0.046 µg/kg, with hypercalcemia observed at 0.23 µg/kg, indicating a safety margin of 5. In our hands, these conventional vitamin D analogues were not different in terms of safety margin in osteopenic OVX rats. The in vivo data taken together show that 1α,25(OH)2D3, ED71, and alfacalcidol are more potent than VDRM2, but VDRM2 has a significantly greater safety margin than these secosteroidal molecules in osteopenic OVX rats.

Alfacalcidol is a therapy for the treatment of osteoporosis in Japan and Europe, whereas vitamin D3 is approved for the treatment of osteomalacia, rickets, hypocalcemia, hypophosphatemia, and other disorders in combination with other therapies. ED71 was well tolerated at 0.75 µg/day and increased vertebral BMD in a 1-year clinical trial with osteoporotic postmenopausal women.10 If our rat data are relevant to the clinical experience, VDRM2 may have therapeutic potential to restore bone mass, spatial architecture, and biomechanical strength in the axial and appendicular skeleton of postmenopausal women with osteoporosis. VDRM2 also may have potential to treat other disorders, either alone or in combination, with other approved therapies.

Disclosures

All the authors are employees of Lilly Research Laboratories.

Acknowledgements

This study was funded by Lilly Research Laboratories.

Ancillary