Targeting Bortezomib to Bone Increases Its Bone Anabolic Activity and Reduces Systemic Adverse Effects in Mice

Abstract

Bortezomib (Btz) is a proteasome inhibitor approved by the FDA to treat multiple myeloma. It also increases bone volume by promoting osteoblast differentiation and inhibiting osteoclastogenesis in mice. However, Btz has severe systemic adverse effects, which would limit its use as a bone anabolic agent. Here, we designed and synthesized a bone-targeted form of Btz by conjugating it to a bisphosphonate (BP) with no antiresorptive activity. We report that BP-Btz inhibited osteoclast formation and bone resorption and stimulated osteoblast differentiation in vitro similar to Btz. In vivo, BP-Btz increased bone volume more effectively than Btz in three mouse models: untreated wild-type mice, mice with ovariectomy, and aged mice with tibial factures. Importantly, BP-Btz had significantly less systemic side effects than Btz, including less thymic cell death, sympathetic nerve damage, and thrombocytopenia, and it improved survival rates in aged mice. Thus, BP-Btz represents a novel anabolic agent to treat conditions, such as postmenopausal and age-related bone loss. Bone targeting is an attractive approach to repurpose approved drugs to treat skeletal diseases.

Introduction

Osteoporotic fracture is a serious health issue in the elderly, leading to decreased quality of life and increased health care costs. Developing new therapies to increase bone mass and strength is a critical clinical need. Repurposing existing drugs to promote fracture healing, especially in the elderly, is an attractive shorter-term approach to address this unmet need. Mesenchymal stem/progenitor cells (MPCs) have been tested in animal models and humans with fractures with promising results.⁽¹⁾ MPC homing to the fracture site, expansion, and differentiation into osteogenic cells are three important cellular processes that actively participate in fracture healing.^(2,3) A factor that regulates all these processes would be an ideal drug candidate. We reported that MPCs from mice deficient in several members of the Nedd4 subclass of ubiquitin E3 ligases have increased migration, growth, and osteoblast (OB) differentiation by increasing the stability of Runx2, JunB, and CXCR4 proteins via the ubiquitin-proteasome system.⁽⁴⁾ MPCs from aged mice have increased proteasomal degradation of Runx2 and JunB and decreased OB differentiation.^(5-7) Thus, we and others hypothesized that the selective inhibition of proteasomes in skeletal cells may increase bone volume and promote fracture healing by improving MPC differentiation to OBs in the elderly.

Bortezomib (Btz, marketed as Velcade) is an FDA-approved proteasome inhibitor for the treatment of patients with multiple myeloma.⁽⁸⁾ It also promotes MPC-OB differentiation by inhibiting the turnover of positive regulators of osteoblastogenesis and inhibits osteoclast (OC) formation by inhibiting the RANKL/NF-κB/TRAF6 signaling pathway.^(9,10) Because of its positive effects on OBs and negative effects on OCs, Btz and other proteasome inhibitors have been considered as very attractive candidates for the development of bone anabolic agents. Several studies have reported that Btz increases bone volume and promotes fracture healing in young or adult normal or ovariectomized mice.^(10,11)

However, Btz has significant side effects, including peripheral neuropathy and thrombocytopenia,⁽¹²⁾ which restrict its use as a bone anabolic drug. In mice, Btz causes death of thymic cells,⁽¹¹⁾ thrombocytopenia,^(13,14) and damage to dorsal root ganglia (DRG).^(15,16) These adverse effects result from the systemic distribution of the drug. Thus, generation of a bone-targeted Btz could meet a critical unmet clinical need by increasing the relative local drug concentration in the skeleton to capitalize on its anabolic and antiresorptive activities, while reducing systemic side effects.

Initial approaches to target drugs to bone with bisphosphonates (BPs) have previously led to limited success with noncleavable linkages that can often interfere with drug action. For example, Agyin and colleagues previously synthesized such BP-Btz analogs, which inhibited myeloma cell growth in vitro, but their in vivo effects were not tested.⁽¹⁷⁾ Recent attempts to target estrogen analogs⁽¹⁸⁾ and a prostaglandin EP4 receptor agonist^(19,20) to bone through a carbamate linker have apparently been more successful. These chemical linkages allow targeting and subsequent releasing an active agent at the bone surface, and this approach improved the efficacy of estradiol to inhibit bone resorption, while limiting side effects, such as endometrial hyperplasia.⁽¹⁸⁾ They facilitated adequate delivery to bone and the slow release of prostaglandin in an anabolic model. However, they typically linked drug “warheads” to a bioactive BP (eg, alendronate), making it difficult to determine if the bioactivity came from the drug or the BP^(17,19,20) or if they had competing or synergistic effects. Thus, new approaches to study the selective delivery of Btz or similar proteasome inhibitors to bone are needed.

Here, we linked Btz to a BP residue that has no antiresorptive activity but retains the strong bone-binding characteristic of bisphosphonates. This allows a clear separation of the bioactivity of therapeutic concentrations of the drug being tested (=Btz) from those of the BP. We demonstrate that our BP-Btz conjugate has higher efficacy than Btz to increase bone volume in normal and ovariectomized mice and promote bone fracture healing in aged mice, with less systemic side effects. Our study is the first in vivo report of the generation of a clinically used bone-targeted proteasome inhibitor and provides a platform for repurposing bone anabolic drugs to treat diseases associated with bone loss.

Materials and Methods

Design and synthesis of a bone-targeted bortezomib analog

To design the bisphosphonate-bortezomib (BP-Btz) conjugate that could deliver Btz to bone via a BP moiety, we used amino-methylene bisphosphonic acid. In this BP analog, the amino group was attached directly to the BP moiety, rendering it inactive rather than using an amino alkyl BP with a 3- to 6-carbon chain, such as alendronate. A diethanolamine fragment was then attached to generate a 1,3,2-dioxazaboroyl component upon coupling with Btz. This component, as the putative released BP, displayed no relevant antiresorptive activity (Fig. 1A). We linked Btz to the above BP analog using a boronate ester (Fig. 1B). The rationale for this approach is that the BP targets Btz to skeletal surfaces (in highest concentration at the highest bone turnover sites) where Btz is released within the acidic environment of bone resorption lacunae generated by active OCs. This results in increased “depot-like” prolonged local Btz concentrations and less or no systemic adverse effects because of a slow release of Btz (Fig. 1C). The molecular weight of the BP containing the linker structure is 366.06. Because Btz has a molecular weight of 384.34 and BP-Btz is 795.32, we typically used half the mg/kg doses of Btz to achieve equimolar concentrations. An injection of 0.6 mg/kg Btz was given as a dose that is similar to that given to humans (1.3 mg/M^{2 (21)}) and one that has been used in most mouse studies.^(1,10) To compare the efficacy of BP-Btz to Btz, we either used the same amount in weight or in equimolar doses (Fig. 1D). Because the BP moiety of BP-Btz does not have antiresorptive activity, we anticipated that the bioactivity of BP-Btz is most likely derived from the proteasome inhibitory activity of the Btz component. Detailed chemical steps for generating BP-Btz has been described in our recent publication.⁽²²⁾

Fig. 1.

The BP-Btz conjugate. (A) Structure and formula of BP-linker (green circle) in which modification of the BP for the linker is indicated by a black line and Btz (blue box). (B) BP-Btz (red box), which comprises the BP-linker and Btz in a form of HBr. (C) Principle of bone-targeted Btz. (D) Molecular weight and molar doses of Btz and BP-Btz used in this study.

Animal studies

Three types of mice were used for various experimental purposes. All mouse surgery procedures and outcome measures were performed according to standard operating procedures (SOP) used in the Center for Musculoskeletal Research (CMSR) at the University of Rochester Medical Center. (1) Ub^G76V-GFP mice⁽²³⁾ were used for in vitro experiments to test the inhibitory effect of Btz on proteasomal degradation of Ub^G76V-GFP proteins. Three-month-old Ub^G76V-GFP mice were purchased from Jackson Laboratory (Bar Harbor, ME, USA; stock #008111) and used for bone marrow (BM) cell isolation. (2) C57BL/6J female mice were used for in vivo experiments. Three month-old C57BL/6J female mice (=WT mice) were purchased from Jackson Laboratory (stock#000664). (a) Mice (n = 4 mice/group) received the same amount (0.6 mg/kg/ip injection) of PBS vehicle, BP, Btz, or BP-Btz daily for 3 weeks to determine effects of BP-Btz on basal bone volume and OB differentiation. (b) Mice received sham surgery or ovariectomy (OVX).⁽²⁴⁾ In brief, sham surgery was performed by identifying both ovaries and OVX was performed by surgically removing the ovaries. Mice (n = 8 mice/group) received an equimolar dose of BP, Btz, or BP-Btz, in which Btz and BP were 0.6 mg/kg, whereas BP-Btz was 1.2 mg/kg, based on their molecular weights described in Fig. 1D. Treatment was started at 4 weeks post-surgery for 4 weeks to determine effects of BP-Btz on OVX-induced bone loss and OB differentiation. (c) Mice (n = 4 mice/group) received an equimolar dose of BP, Btz, or BP-Btz daily for 3 weeks to examine systemic adverse effects of BP-Btz on non-bone tissues. (3) Aged C57BL/6J mice (18-month-old, male) were obtained from the National Institute of Aging and received open tibial fracture surgery.^(25-28) Briefly, a 6-mm-long incision was made in the skin over the anterior tibia after anesthesia. A sterile 27G needle was inserted through the proximal tibial articular surface into the BM cavity, temporarily withdrawn to facilitate midshaft transection using a scalpel, and reinserted to stabilize the fracture, which was confirmed by X-ray. The incision was closed with sutures. Mice received buprenorphine SR, 0.5 mg/kg, SQ to control pain and were divided into five groups (n = 6–8 mice/group). Mice in group 1 received PBS vehicle. Mice in groups 2 to 4 received equimolar doses of BP, Btz, and BP-Btz. Mice in group 5 received a one-third lower molar dose of BP-Btz. All mice were treated on days 1, 3, and 5 post-fracture surgery to determine the effects of BP-Btz on the early phase of fracture healing when MPCs play a critical role.⁽²⁹⁾ All the animal procedures used in this study were approved by Animal Care and Use Committee of University of Rochester.

Cell cultures

(1) OC formation. A total of 5 × 10⁴ BM cells from WT mice were cultured in 96-well-plates with M-CSF and RANKL. After multinucleated cells were observed under a microscope, the cells were fixed, stained for TRAP activity to identify OCs (TRAP+ cells containing >3 nuclei), and counted, as described previously.⁽⁷⁾ (2) Bone resorption. A total of 1 × 10⁵ BM cells from Ub^G76V-GFP mice were seeded on bovine cortical bone slices and cultured for 10 to 12 days, as described previously.⁽³⁰⁾ Then the GFP-expressing OCs on bone slices were viewed using a Zeiss (Thornwood, NY, USA) LSM 510 confocal fluorescence microscope. Fluorescence in OCs was quantified by measuring the area and intensity of GFP staining in the entire bone slice using Image Pro Plus 6.0 software, and the results were expressed as integrated optical density (IOD): IOD = area × intensity of fluorescence. The bone slices were then fixed and stained for TRAP activity. Osteoclast numbers were counted and OCs then were removed by brushing the bone slices, which were stained with 1% toluidine blue to highlight resorption pits for measurement of pit areas. The area of resorption pits per bone slice was assessed, as described previously.⁽³⁰⁾ (3) OB differentiation. Bone-derived mesenchymal stem cells (MSCs) were isolated using a recently published protocol,⁽³¹⁾ and human MSCs were purchased from Lonza (Basel, Switzerland; catalog #PT-2501). Cells were cultured in 60-mm dishes at 2 × 10⁶ cells/dish in α-MEM culture medium containing 10% FCS with or without 50 μg/mL ascorbic acid and 10 mM β-glycerophosphate. Media were changed every 4 days for 12 days. At the end of the culture period, cells were fixed in formalin and stained for alkaline phosphatase (ALP) activity. The ALP+ area was measured, as described previously.⁽³²⁾

Micro-CT

Femurs or fractured tibias were dissected free of soft tissue, fixed overnight in 10% buffered formalin, and scanned at high resolution (10.5 μm) on a VivaCT40 μCT scanner (Scanco Medical, Bruttisellen, Switzerland) using 300 ms integration time, 55 kVp energy, and 145 μA intensity. For trabecular bone analysis of long bone, the region of interest (ROI) was selected from the distal femoral metaphysis adjacent to the metaphyseal growth plate and extended 100 slices (600 μm) proximal. A standardized Scanco threshold for femur trabecular images was 240 (=369 mg HA/cm³). For assessment of the mineralized callus, the threshold was chosen using 2D evaluation of several slices in the transverse anatomic planes so that mineralized callus was identified, but surrounding soft tissues were excluded. The Scanco threshold was 260 (=416 mg HA/cm³). Three-dimensional images were generated using a constant threshold of 275 for all samples. Trabecular bone parameters, including bone volume per tissue volume (BV/TV), trabecular thickness (Tb.Th), trabecular number (Tb. N), and trabecular spacing (Tb.Sp), or mineralized callus bone volume were assessed using Scanco analysis software.⁽³³⁾

Biomechanical testing

For the OVX studies, femurs were wrapped in saline-soaked gauze and stored at −80°C until analysis. Specimens were thawed to room temperature, rehydrated, and subjected to a three-point bending test using the Instron Dynamight 8841 servo-hydraulic materials testing device (Instron, Norwood, MA, USA) with Bluehill software. In brief, femurs were positioned across a custom-built apparatus with the posterior side facing downward. Each femur was preloaded with 1N, and the crosshead was lowered at 0.5 mm/min until specimen failure.⁽³⁴⁾ For the fracture studies, after mouse euthanization on day 28 post-fracture surgery, the stabilizing needles were removed carefully from tibias to avoid any disturbance of the architecture of healing fracture callus and then were stored at −80°C until analysis. The tibial ends were potted in polymethylmethacrylate and placed on an EnduraTec system (Bose Corporation). A rotation rate of 10/s was used to twist the samples to failure or up to 80°. Maximum torque, maximum rotation, and torsional rigidity were analyzed based on a CMSR SOP.⁽³⁵⁾

Histology and histomorphometric analysis

Legs, including femora and tibias, were fixed in 10% buffered formalin, decalcified in 10% EDTA, and embedded in paraffin for sectioning. Paraffin sections (4 μm) were stained with H&E for general histology or Alcian blue/hematoxylin (ABH) for cartilage and woven bone. Adjacent sections were also stained for TRAP activity to identify OCs. Histomorphometric analyses were performed on sections cut at three levels in each sample using CMSR SOPs.^(36,37) In brief, stained sections were numbered blindly and were converted to digital images using an Olympus VS120 whole slide imaging system (Olympus, Center Valley, PA, USA). A full static bone histomorphometric analysis in long bone, including BV/TV (%), Tb.Th (μm), Tb.N (/mm), Tb.Sp (mm), osteoblast surface/bone surface (Ob.S/BS; %), osteoclast surface/bone surface (Oc.S/BS; %), and eroded surface/bone surface (ES/BS; %), were performed in the secondary spongiosa area using the automated algorithm with a Visiopharm Image Analysis System as we previously described.⁽³⁸⁾ The fat vacuoles within secondary spongiosa and diaphysis of femoral sections were measured using Image Pro Plus 6.0 software. The percentage of fat vacuoles area over the tissue (secondary spongiosa and diaphysis) area was calculated. For dynamic bone formation data, mice received calcein injection (20 mg/kg, ip) on day 6 and day 1 before death. Femora were fixed in 10% formalin and applied for frozen embedding and sectioned. Pictures were taken by a fluorescence microscope (Olympus BX51, Center Valley, PA, USA). Bone formation parameters, including bone formation rate (BFR), mineral apposition rate (MAR), and double-labeled surface/bone surface (dLS/BS), were assessed using Bioquant Osteo software (Bioquant Image Analysis Corp., Nashville, TN, USA) based on the ASBMR guideline.⁽³⁹⁾ For fracture study, woven bone area and cartilage area in external callus were analyzed using the automated algorithm developed by the CMSR members with a Visiopharm Image Analysis System (Westminster, CO, USA).⁽²⁸⁾

Distribution of BP labeled with IRDye800 (=800CW-ZOL)

Two sets of mice were used to examine the distribution of BP conjugates in fracture callus or in calcified extraskeletal tissues. 800CW-ZOL (BioVinc, Pasadena, CA, USA; catalog #BV551001) was administrated by retro-orbital injection. For fracture callus, 3-month-old WT mice (no-fracture or at 5 days post-fracture) received 800CW-ZOL (1 μg) and were subjected to an IVIS SpectrumCT preclinical in vivo imaging system (PerkinElmer, Waltham, MA, USA) at 24 hours to measure NIR fluorescence in legs before they were euthanized. The epi-fluorescence intensity in metaphyses and diaphyses was measured. Mice were then euthanized and fractured tibias were fixed in 10% buffered formalin and embedded in OCT. Undecalcified frozen sections (7 μm) were cut and the location of 800CW-ZOL was visualized and imaged by fluorescent microscopy with a xenon NIR filter (830/50 nm). Images were merged with bright-field images taken at the same location.

Flow cytometry

Thymi were dissected, mashed with a syringe plunger, and passed through a cell strainer (100 μm, BD Biosciences, San Jose, CA, USA; catalog #352360). A single-cell suspension was stained with PE/Cy5-anti-mouse CD3 monoclonal antibody (eBioscience, San Diego, CA, USA; catalog #35-0031-82) and FITC-anti-mouse B220 monoclonal antibody (eBioscience, catalog #11-0452-82) for 30 minutes and subjected to flow cytometric analysis using a 12-color LSRII (BD Biosciences). Results were analyzed by Flowjo7 software (FLOWJO, LLC, Ashland, OR, USA).

Electron microscopic analysis of dorsal root ganglia

Animals were euthanized under general anesthesia and perfused with 4% paraformaldehyde in PBS. L₄ to L₅ dorsal root ganglia (DRG) were harvested, as previously described.⁽¹⁶⁾ Semi-thin (1.0 to 2.0 μm) sections were stained with toluidine blue and examined under a Nikon light microscope to determine the appropriate area for electron microscopic analysis. Based on the light microscopic findings, ultrathin sections (70 nm) were prepared from selected tissue blocks and examined with a Hitachi - Science & Technology (Berkshire, UK) H-7650 transmission electron microscope (EM) that has an Erlangshen 11 megapixel digital camera and Gatan software for imaging and morphometric analysis. Myelin thickness and myelinated axon diameter were measured on EM images, according to a published method.^(40,41)

Routine blood cell counting

Blood was collected from the retro-orbital sinus and stored in a microtainer (BD, 365974) for less than 1 hour before analysis. Complete blood count values were acquired using the Scil Vet ABC Plus hematology analyzer (Scil Animal Care Company, Gurnee, IL, USA).

Statistical analysis

All results are given as means ± SD. Statistical analysis was performed using GraphPad Prism 5 software (GraphPad Software Inc., San Diego, CA, USA). Comparisons between two groups were analyzed using a 2-tailed unpaired Student’s t test. One-way ANOVA and Dunnett’s post hoc multiple comparisons were used for comparisons among three or more groups. Any p values less than 0.05 were considered statistically significant.

Results

BP-Btz inhibits osteoclastogenesis, binds to bone matrix, and stimulates osteoblast differentiation

Btz inhibits OC formation and increases OB differentiation.^(1,9) To determine if the addition of a BP moiety with the boronate ester linkage to Btz affects its inhibitory effects on osteoclastogenesis, we examined the effect of different doses of Btz and BP-Btz on OC formation in plastic culture dishes using BM cells from WT mice. Btz and BP-Btz inhibited OC formation at similar concentrations (Fig. 2A).To investigate if BP-Btz can bind to bone mineral and inhibit OC bone resorption, we used BM cells from mice with an ubiquitin/proteasome system reporter (Ub^G76V-GFP) in which their Ub-GFP protein undergoes constitutive proteasome degradation in the absence of a proteasome inhibitor.⁽²³⁾ We pre-incubated bone slices with 1 μM BP, Btz, and BP-Btz overnight, removed drug solutions, and washed the bone slices extensively with PBS. We then seeded BM cells onto the bone slices and performed OC resorption assays. We found that Ub-GFP protein degradation, OC formation, and bone resorption were inhibited when cells were cultured on bone slices pre-incubated with BP-Btz but not Btz or BP (Fig. 2B). To compare the stimulatory effect of BP-Btz and Btz on OB differentiation, we treated murine BM MPCs (Fig. 2C) or human mesenchymal stem cells (Fig. 2D) with BP-Btz or Btz and found that both promoted OB differentiation with similar efficacy, as assessed by ALP staining. Together, these data show that as an intact conjugate, BP-Btz has biological effects similar to those of Btz on OBs and OCs when it is added directly to culture medium. BP-Btz, but not Btz, bound to bone slices and later were able to exert its anti-OC activity.

Fig. 2.

BP-Btz inhibits osteoclastogenesis, binds to bone matrix, and stimulates osteoblast differentiation. (A) Bone marrow (BM) cells from C57BL/6J WT mice were treated with BP-Btz or Btz with M-CSF and RANKL for 2 days. OCs were identified by TRAP staining. n = 5 wells. Three experimental repeats. One-way ANOVA with Dunnett’s test. *p < 0.05 versus 0. (B) Bone slices were pre-incubated with BP-Btz, Btz, or BP overnight and washed with PBS. BM cells from Ub^G76V-GFP mice were cultured on bone slices with M-CSF and RANKL for 9 days. (a) GFP⁺ cells were observed under fluorescence microscopy and GFP signal intensity was quantified. (b) Bone slices stained for TRAP activity in OCs. OC numbers per slice were counted. (c) Cells were removed from slices. Toluidine blue staining was performed, and pit areas per slice were counted. n = 5 slices. Three experimental repeats. One-way ANOVA with Dunnett’s test. *p < 0.05 versus other groups. (C) Murine BM cells from WT mice or (D) human MSCs were cultured in OB-inducing medium with BP-Btz or Btz for 4 days. Cells were stained for ALP. ALP⁺ areas were measured. n = 4 wells. Two experimental repeats. One-way ANOVA with Dunnett’s test. *p < .05 versus 0. All data are means + SD.

BP-Btz has higher bone anabolic efficacy than Btz in WT mice

A critical difference between BP-Btz and Btz is that BP-Btz is designed to accumulate at higher concentrations in the skeleton in vivo, thereby exerting most of its bioactivity locally in bone, whereas Btz exerts its bioactivity systemically. We hypothesized that BP-Btz would increase bone volume and stimulate OB differentiation more effectively than Btz. To examine this, we treated 7-week-old female C57BL/6J WT mice with vehicle or the same amount (by weight) of BP (0.6 mg/kg), Btz (0.6 mg/kg), or BP-Btz (0.6 mg/kg) daily for 3 weeks (Fig. 3A). This Btz treatment regimen has been shown previously to increase bone volume and promote fracture healing in mice.^(1,10) Micro-CT analysis showed that Btz-treated mice had higher tibial bone volumes than BP-treated mice and that values were significantly higher than in Btz-treated mice (Fig. 3B; Supplemental Fig. S1). Histomorphometric analysis of H&E-stained sections confirmed that BP-Btz-treated mice had the highest bone volumes and OB surface (Fig. 3C, D). TRAP-stained sections indicated that both Btz and BP-Btz treatment reduced OC parameters to a similar extent (Fig. 3C, D). Calcein-labeled sections revealed that both Btz and BP-Btz treatment increased in vivo bone formation parameters. Impressively, BFR in mice treated with BP-Btz is significantly higher than that from the mice receiving Btz. Other bone formation parameters in mice treated with BP-Btz also tended to be higher than those from Btz-treated mice (p = 0.057 for dLS/BS; p = 0.063 for MAR) (Fig. 3D). To determine if BP-Btz has a greater effect on OB differentiation than Btz, we examined CFU-F and CFU-ALP+ colony formation using BM stromal cells from BP- Btz-, Btz-, or BP-treated mice. Cells from BP-Btz-treated mice formed significantly more CFU-F+ and CFU-ALP+ colonies than cells from either Btz- or BP-treated control mice (Fig. 3E). It is noteworthy that in this set of experiments, the molar concentration of BP-Btz was half that of Btz (molecular weights 795.32 versus 384.24), indicating that BP-Btz has significantly higher bone anabolic efficacy than Btz in WT mice.

Fig. 3.

BP-Btz has greater bone anabolic effects than Btz in mice. (A) WT mice received daily administration of BP, Btz, BP-Btz of Vehicle for 3 weeks. Bone volume and OB differentiation were examined at the end of the experiment. n = 4–5 mice/group. (B) Representative of micro-CT images and tibial BV/TV (%). Scale bar = 300 μm. One-way ANOVA with Dunnett’s test. *p < 0.05. (C) Representative H&E-, TRAP-stained, and double calcein labeling sections. Scale bar = 500 μm. (D) Histomorphometric analyses of bone volume, OB, and OC parameters. One-way ANOVA with Dunnett’s test. *p < 0.05. (E) BM cells were cultured in the basal growth or OB-inducing medium. CFU-F colonies were identified by methylene blue staining and CFU-ALP⁺ colonies were identified by ALP staining. One-way ANOVA with Dunnett’s test. *p < 0.05. All data are means + SD.

BP-Btz protects against ovariectomy-induced bone loss more effectively than Btz

OVX is the most commonly used mouse model for the evaluation and development of new drugs to treat postmenopausal osteoporosis. Because BP-Btz increased bone volume in mice at a dose that was half of that of Btz, we investigated if BP-Btz could prevent OVX-mediated bone loss with a lower-frequency dosing regimen than Btz (Fig. 4A). We treated OVX-mice with daily administration of Btz as a positive control and 2 times/week administration of BP-Btz or Btz+BP as comparison groups. We used equimolar concentrations of BP-Btz (1.2 mg/kg) and Btz (0.6 mg/kg, a dose that has been used for Btz in multiple in vivo mouse studies by other investigators^(1,10) and by ourselves in Fig. 3). A total of five groups of mice were tested. The groups 1, 2 and 4 were various controls, including sham-operated mice treated with saline (Sham), OVX mice treated with saline as OVX plus vehicle control (Veh, 2 times/week), and OVX mice treated with 0.6 mg/kg Btz as Btz-positive controls (Btz daily). Groups 3 and 5 were test groups that received treatment 2 times/week, including OVX mice treated with BP-Btz or Btz+BP. Mice were treated at 4 weeks post-OVX, by which time most sex steroid deficiency–induced bone loss has occurred, for 4 weeks. Histomorphometric analysis of H&E- and TRAP-stained sections revealed that, compared with Sham mice, Veh-treated OVX mice had significantly lower bone volumes, higher OB and OC parameters, and more marrow fat (Fig. 4B, C). As a positive control, daily administration of Btz reversed decreases in bone volume, OCs, and marrow fat and returned them to the levels found in the Sham group. Btz increased OBs. Interestingly, mice treated 2 times/week with BP-Btz, but not with Btz+BP, had levels of increased bone volume and OBs and decreased OCs and marrow fat similar to those in the Btz daily group, despite the lower frequency and total dose of BP-Btz administered (Fig. 4B, C). Biomechanical testing using three-point bending indicated that the mechanical stress indexes, including stiffness, yield load, and energy to yield, were decreased in the OVX-Veh group compared with the Sham group. All of these indexes were rescued in the Btz daily and BP-Btz groups but not in the Btz+BP group (Fig. 4D). Furthermore, there was no significance of Veh control versus Btz+BP BP treatment in the OVX mice (Fig. 4C, D). To determine if BP-Btz affects OB differentiation, we examined CFU-F and CFU-ALP+ colony formation using BM stromal cells from these mice. Cells from the OVX+Veh group formed fewer CFU-F and CFU-ALP+ colonies than cells from the Sham mice. Cells from Btz daily, BP-Btz, and BP+Btz groups formed significantly more CFU-F and CFU-ALP+ colonies than cells from OVX+Veh-treated mice. Cells from the Btz daily and BP-Btz groups formed more CFU-F and CFU-ALP+ colonies than those from the Sham group (Fig. 4E). Taken together, these data show that BP-Btz more effectively restores OVX-induced bone loss and promotes OB differentiation than Btz.

Fig. 4.

BP-Btz restores OVX-induced bone loss more effectively than Btz. (A) WT mice received OVX or sham surgery. OVX mice were treated for 4 weeks with Veh, Btz, BP-Btz, or BP+Btz, starting at 4 weeks post-surgery. Sham mice were treated with Veh. Bone volume and strength and OB differentiation were examined at the end of the experiment. n = 8 mice/group. (B) Representative H&E- and TRAP-stained sections of left femoral bones. Scale bar = 500 μm. (C) Histomorphometric analyses of bone volume, OB, and OC parameters, and fat cell area. One-way ANOVA with Dunnett’s test. *p < 0.05. (D) Biomechanical testing of right femora. One-way ANOVA with Dunnett’s test. *p < 0.05. (E) CFU-F and CFU-ALP+ colony formation as in Fig. 3D. One-way ANOVA with Dunnett’s test. *p < 0.05. All data are means + SD.

BP-Btz more effectively enhances fracture repair in aged mice than Btz

Osteoporotic fractures occur commonly in the elderly, and aged subjects more often suffer from adverse effects of drugs than younger subjects. Btz promotes bone fracture healing in young mice^(1,10) and when given at the early phase of fracture repair, it enhanced healing and MPC-OB differentiation in young mice.⁽⁴²⁾ To examine if BP-Btz has similar or better efficacy as Btz to enhance fracture repair in aged mice, we treated aged mice (18 months old) on days 1, 3, and 5 post-fracture surgery with Veh or the equimolar dose of BP, Btz, or BP-Btz. We also treated one group of mice with a threefold lower molar dose of BP-Btz. We examined fracture healing by micro-CT and histology on day 14 and by biomechanical testing on day 28 (Fig. 5A). Micro-CT analysis showed that compared with Veh or BP-treated mice, BP-Btz and Btz increased callus bone volume (Fig. 5B). Histomorphologic analysis of ABH-stained sections revealed increased woven bone and cartilage areas (Fig. 5C), and biomechanical testing indicated improved bone strength (Fig. 5D) in BP-Btz- and Btz-treated mice. Importantly, low-dose (one-third) BP-Btz promoted bone fracture repair in these aged mice with similar efficacy as Btz. We treated young mice (3-month-old, C57BL/6J male mice) with BP-Btz using the same regimen as in aged mice and found that, similar to old mice, BP-Btz also increased bone volume in fracture callus. The increase is slightly but significantly higher than that in old mice (153% ± 1.2% in young mice versus 131% ± 4.5% in old mice, p = 0.0231). These data suggest that bones of young mice may take up more BP-Btz than those of old mice (p = 0.023, Supplemental Fig. S2).

Fig. 5.

BP-Btz enhances fracture repair in aged mice. (A) Eighteen-month-old WT mice received open tibial fracture surgery in both legs and were treated with Veh, BP, Btz, or BP-Btz on days 1, 3, and 5 post-fracture. Mice were euthanized 2 weeks or 4 weeks later and tibias were analyzed. (B) Callus volume by micro-CT. (C) Representative ABH-stained sections and histomorphometric analysis of woven bone and cartilage areas in the fracture calluses. Scale bar = 1 mm. Data were collected at 2 weeks post-treatment. n = 6–8 mice/group. One-way ANOVA with Dunnett’s test. *p < .05. (D) Bone strength by biomechanical testing. Data were collected at 4 weeks post-treatment. n = 3–5 mice (=6–10 legs)/group. One-way ANOVA with Dunnett’s test. *p < .05. All data are means + SD.

To determine if the BP-conjugated drug accumulates at the bone fracture site where high bone remodeling occurs, we injected the NIR-labeled BP (800CW-ZOL) intravenously and examined its distribution in the fractured and sham legs of the mice 24 hours later (Fig. 6A). As reported previously, BPs bind in higher concentration on bone surfaces with active bone remodeling,⁽⁴³⁾ and we observed a similar distribution of strong 800CW-ZOL signal intensity in the metaphyseal regions of both fractured or non-fractured legs. Impressively, markedly increased 800CW-ZOL signal intensity was detected in the diaphyseal fracture sites compared with the same region in nonfractured legs or in the metaphyses (Fig. 6B, C). Frozen sections of undecalcified bone showed much higher 800CW-ZOL signal intensity in the woven bone in calluses than in the trabecular bone of metaphyseal and epiphyseal regions (Fig. 6D).

Fig. 6.

Distribution of a NIR-labeled BP in fracture callus. (A) WT mice with no fracture or at 5 days post-fracture surgery received an NIR-labeled BP (800CW-ZOL) via intravenous administration. Mice were scanned by an IVIS in vivo imager 24 hours later and then were euthanized. Tissues were subjected to frozen sectioning. (B) Representative IVIS images showing the distribution of 800CW-ZOL in legs. (C) The epi-fluorescence intensity in tibial metaphyses (indicated by yellow circle in B) and diaphysis (indicated by green circle in B). Data are means + SD of four legs. Student’s t test. *p < 0.05. (D) Representative images taken using an NIR filter showing the location of 800CW-ZOL (pseudo-color pink) in fracture callus (upper panel) and nonfracture area (lower panel) on frozen sections of a fractured tibia. n = 3. Scale bar = 125 μm.

BP-Btz has less systemic adverse effects than Btz in WT mice

A critical difference between BP-Btz and Btz is that the BP directs Btz to bone in vivo, thereby facilitating its bioactivity locally, whereas Btz acts systemically. In patients, peripheral neuropathy and thrombocytopenia are major adverse effects of Btz.⁽⁸⁾ In animals, Btz causes death of thymic cells,⁽¹¹⁾ thrombocytopenia,^(13,14) and damage to dorsal root ganglia (DRG).^(16,44) Thus, we used changes in thymus, platelet counts, and DRG as surrogates to evaluate systemic toxicities of BP-Btz in mice. We treated 7-week-old C57BL/6J WT mice with an equimolar dose of BP, Btz, and BP-Btz daily for 3 weeks. Body weight was measured weekly, and thymus, DRG, and megakaryocytes were examined at the end of experiment (Fig. 7A). No differences in body weight were detected among BP-, Btz-, and BP-Btz-treated mice (data not shown). However, thymic weights and thymocyte numbers were significantly lower in the Btz-treated mice than in the BP-treated mice, in which these parameters were normal. Similarly, the numbers of CD3+ T cells and CD19+ B cells were also decreased in Btz-treated mice but not in BP-Btz-treated mice (Fig. 7B). More importantly, EM examination revealed significantly lower myelin thickness and axon diameter in DRG from Btz-treated than BP control mice, whereas these indexes were normal in BP-Btz-treated mice (Fig. 7C, D).

Fig. 7.

BP-Btz has less systemic adverse effects than Btz in WT mice. (A) WT mice received daily administration of BP, Btz, or BP-Btz for 3 weeks and were euthanized 1 day post-treatment. (B) Thymus weight, total cell number, T cell number, and B cell number. n = 4 mice. One-way ANOVA with Dunnett’s test. *p < 0.05 versus other groups. (C) EM images of cross sections of sciatic nerves showing changes in myelin thickness and myelinated axons. Scale bar = 5 mm. (D) Myelin thickness and myelinated axon diameter measured on EM images. n = 20 fields. One-way ANOVA with Dunnett’s test. *p < 0.05 versus other groups. (E) Megakaryocytes were counted on H&E-stained sections of tibias and femora. n = 4 mice/group. All data are mean + SD. (F) WT mice received Veh, Btz, or BP-Btz once. Blood platelet numbers were measured at different times. n = 6 mice. One-way ANOVA with Dunnett’s test. *p < 0.05 versus Veh. All data are mean + SD. (G) Aged mice (18 months old) were treated with test agents, as in Fig. 5A. Survival rates by a Kaplan–Meier plot. n = 5 mice in Veh-, 5 mice in BP-, 12 mice in Btz-, 8 mice in BP-Btz-low-, and 10 mice in BP-Btz-Eq. molar-treated group.

Thrombocytogenesis occurs in the BM where platelets are formed and released by megakaryocytes into the blood-stream.⁽¹³⁾ Thus, because BP-Btz is targeted to bone, it is possible that may cause BM toxicity, resulting in more severe thrombocytopenia than Btz. To examine this possibility, we counted megakaryocyte numbers in BM sections from the above mice and found that neither Btz nor BP-Btz affected megakaryocyte numbers (Fig. 7E). Furthermore, we compared blood platelet counts in WT mice before and 24, 48, and 96 hours after they received equimolar concentrations of high-dose Btz (2 mg/kg) or BP-Btz (4 mg/kg), a dosing regimen used previously to test the effects of Btz on blood platelet toxicity.^(13,45) As reported, Btz significantly reduced blood platelet counts at 24 to 48 hours, and these returned to pretreatment stage after 72 hours. BP-Btz had no significant effects on blood platelet counts (Fig. 7F).

Most Btz studies in mice were performed in young or adult mice.^(1,10) In our fracture study described in Fig. 6, we found that 60% of aged mice were dead within 5 days from the start of Btz treatment. In contrast, neither BP-Btz nor BP treatment caused a significant number of deaths in aged mice (Fig. 7G). Of note, Btz did not cause death in young fractured mice using the same administration regime (death: 0 of 5 mice). Together, the data represented in Fig. 7 strongly demonstrated that BP-Btz has less systemic adverse effects than Btz in WT mice.

Discussion

Btz is a first-line drug used to treat patients with multiple myeloma with significant response rates in the setting of either newly diagnosed or relapsed/refractory disease.⁽⁴⁶⁾ Its mechanism of action is to reversibly inhibit the mammalian 26S proteasome.⁽⁴⁷⁾ in vivo mouse studies reported bone anabolic effects of Btz by promoting MPC-OB differentiation and inhibiting OC formation.^(10,11) However, systemic side effects limit the possibility of repurposing Btz as a bone anabolic drug. In the current study, we generated a bone-targeted Btz (BP-Btz) by conjugating Btz to a BP moiety with no antiresorptive activity using a novel chemical linkage. We demonstrated for the first time that our BP-Btz conjugate more effectively restores OVX-induced bone loss and promotes bone fracture repair in aged mice with a lower total dose and dosing frequency than Btz. Importantly, we observed that at equimolar concentrations, BP-Btz had less systemic adverse effects than Btz, including a lack of neuronal toxicity in dorsal root ganglia, no significant effects on immune cells in the thymus, and reduced death rates in aged mice. Thus, BP-Btz and perhaps our bone-targeting platform represent a novel therapeutic approach to repurpose drugs used clinically for the treatment of patients with osteoporosis and fracture or other bone diseases.

Targeting drugs to bone by linking them to a BP residue has been a very attractive idea for many years because of the potential benefits of lower systemic toxicities, lower dosage requirements, and the potential for increased efficacy. Recently, both estradiol-BP and prostaglandin EP4 receptor agonist-BP conjugates were synthesized and tested in mice, and these showed improved bone anabolic effects for basal or OVX-induced bone loss versus non-BP analogs.^(18-20) BP-nanoparticle-Btz have also been generated. So far these have displayed similar bone anabolic effects and increased OB function in mice to unconjugated Btz.⁽⁴⁸⁾ These preclinical reports demonstrated the technical feasibility of generating bone-targeting drugs using a BP as a carrier. However, there are issues that have not yet been addressed in these reports. For example, the main side effects of Btz in humans are peripheral neuropathy and thrombocytopenia. It will be important to determine that these side effects can be reduced in humans by bone targeting Btz. In the clinic, patients who suffer from Btz-induced neuropathy often have abnormal sensations in feet or hands, such as tingling or numbness, which cannot be modeled in mice. However, EM examination revealed nerve damage in DRG of Btz-treated rats^(16,44) and Btz causes thymic cell death in mice.⁽¹¹⁾ Using these outcome measures, we demonstrated that BP-Btz had only minor adverse effects relative to Btz in mouse DRGs and in the thymus. Thrombocytopenia is another indicator of Btz toxicity. Platelets are derived from megakaryocytes in the BM. BP-Btz might worsen Btz-induced thrombocytopenia because Btz is released from it in the bone microenvironment. However, this does not appear to occur because there were no obvious differences in megakaryocyte numbers between Btz- and BP-Btz-treated mice and, unlike Btz, BP-Btz did not cause thrombocytopenia. Thus, our findings are supportive of potential translation of BP-Btz conjugates to the clinic because they are the first report demonstrating significantly reduced systemic toxicities by targeting Btz to bone.

We used an NIR-labeled BP as a surrogate to visualize the distribution of BP-Btz and found higher fluorescent BP signal intensity in fracture callus than on other bone surfaces (Fig. 6). This suggests that more BP enters the callus through its enhanced vasculature and binds to active bone-forming surfaces. Because BP-Btz will likely bind to calcified and calcifying bone surfaces and act locally, it should have higher efficacy than Btz to increase bone repair. However, to confirm that Btz is released from the BP-Btz at bone locally, we will need to use chemical approaches, such as LC–MS/MS, to measure released Btz. Another issue is that we do not know if new BP-Btz conjugates with slower or faster release rates might improve efficacy. These studies are under active study by our research team.

An important issue is the association between BP-Btz uptake and bone turnover because bone turnover is likely to be lower in aged human patients. Thus, BP-Btz could be more effective in young individuals with a higher bone turnover than in older subjects. Our data in BP-Btz-treated young mice indicate that young mice may take up slightly more BP-Btz than old mice (Supplemental Fig. S2). However, since the deposition of BP-Btz in fracture callus is more than that on trabecular bone surfaces (Fig. 5D), even in old mice, the potential effect of aging on the BP deposition may not affect the overall efficacy of BP-Btz on fracture repair of aged mice. Furthermore, BPs are first-line drugs used to treat patients with osteoporosis, which justifies the use of a BP bone-targeted approach in the aged population with a relatively low bone turnover rate.

In the current study, we investigated the effect of short-term administration of BP-Btz on fracture healing in aged mice. We did not examine the bone volume in unfractured limbs after BP-Btz treatment because we treated fractured mice only 3 times, on days 1, 3, and 5 post-fracture surgery. The rationale for early-short treatment is to avoid the influence of Btz on osteoclasts, a similar rationale that Dr Matthew J Hilton used for his Notch inhibitor fracture study.⁽²⁸⁾ It was reported that in young WT mice, Btz increases bone volume after 3 weeks of daily administration,⁽¹⁾ which we also show in Fig. 3. Thus, it is very likely that longer treatment is needed to see any bone anabolic effect of Btz in mice, especially aged mice. A better experiment would be to compare the anabolic bone effect of BP-Btz in young and old mice. However, because BP-Btz is unlikely to have an application for treatment of osteoporosis in humans and the high cost associated with use of aged mice, we did not perform this experiment.

Our overall hypothesis is that the BP will bring Btz to the bone surface (particularly at high turnover sites) where Btz is released locally. Released Btz will affect bone cells and other nearby cell types via similar molecular mechanisms as free Btz but with a stronger effect because higher concentration of locally released Btz will be generated and sustained in bone longer. Higher OB numbers (Figs. 3 and 4) in BP-Btz-treated mice than in mice treated with free Btz indicate that higher/longer Btz exposure (achieved via BP-Btz) had a stronger stimulatory effect on OB differentiation. To determine if BP-Btz provides a higher/longer Btz presence in bone, we examined the expression of total ubiquitinated proteins in bone samples isolated from BP-Btz or free Btz-treated mice at 4 and 48 hours after drug administration. The rationale for this experiment is that if BP-Btz provides a higher/longer level of Btz in bone, we should detect more ubiquitinated proteins in bones from BP-Btz-treated mice than those from Btz-treated mice at later time points, based on the mechanism of action of Btz: a proteasome inhibitor that prevents degradation of ubiquitinated proteins in the proteasome. Our preliminary data show more total ubiquitinated proteins in bone samples from BP-Btz-treated mice than in bone samples from Btz-treated mice, whereas Btz and BP-Btz have similar effects on the expression of ubiquitinated proteins in lung tissue. Thus, a higher/longer level of Btz may serve as a potential mechanism of BP-Btz (data not shown). It is also possible that a higher/longer Btz exposure affects molecular events in cells differently than in cells that undergo a lower/shorter Btz exposure. We can test this possibility using cells isolated from BP-Btz or Btz-treated mice and performing bioinformatics analysis in future studies.

Another question is how does BP-Btz attach to bone slices in vitro (Fig. 2B) without prior bone formation? The BP that we have used has a very high affinity for hydroxyapatite and thus binds tightly to mineralized surfaces, and this has been demonstrated in the literature. It also has a very low affinity for the farnesyl farnysyl synthase to which nitrogen-containing and other OC-inhibiting BPs bind in OCs. Recently, there have been multiple reports of BPs binding to bone slices and other crystalline forms of hydroxyapatite in vitro, and the binding mode has been demonstrated with X-ray diffraction techniques. Although other diagnostic agents label bone by incorporating into formation sites, the BPs bind and chemisorb to calcium atoms in hydroxyapatite (calcium phosphate) surface. Other groups have reported binding studies in vitro, confirming this binding.⁽⁴⁹⁾ In vivo, it was reported with the radiolabeled drug by Sato and colleagues⁽⁵⁰⁾ that BPs also bind to resorption surfaces. It was also shown with fluorescent BPs in vivo⁽⁵¹⁾ that BPs can label formation, resorption, and neutral or quiescent surfaces, although again formation surfaces incorporate the most drug. The BP-based technetium bone scan agent data may be misleading as they are ionically linked and the technetium may, therefore, wash away at resorption sites but get incorporated with BP at formation surfaces.

Btz has many biological effects and it affects multiple cell types, which may influence bone volume directly or indirectly. We only examined changes of OB differentiation using MPCs from BP-Btz-treated mice because of the importance of OB-mediated bone formation in maintaining or increasing bone mass. Nevertheless, the delayed repair of fractures in the elderly is mostly associated with reduced MPC-OB differentiation. Our study provides proof of principle that Btz can be targeted to this cell population and enhances fracture repair, raising the possibility that it could be developed as a strategy to enhance fracture repair.

In summary, we have developed a novel bone-targeted Btz conjugate by linking Btz to a bone-binding BP with no antiresorptive activity and have shown that it binds to bone matrix, increases bone volume, and enhances fracture healing with greater efficacy than Btz. Importantly, we have also provided experimental evidence to demonstrate that the bone-targeting approach markedly reduces peripheral neuropathy induced by Btz, a major limitation of its overall efficacy in patients. The scientific premise of our study is that a similar approach may enable the generation of other bone-targeted anabolic bone agents that have positive effects on OB function in vitro and in vivo but that cannot be used in humans to treat bone disorders because the doses that would be required would cause severe systemic adverse effects on other cell types. This bone-targeting strategy could be utilized to repurpose other drugs for the treatment of fractures and common bone diseases, including osteoporosis.