Mitigating Aging and Doxorubicin Induced Bone Loss in Mature Mice via Mechanobiology Based Treatments

Abstract

Aging leads to a reduced anabolic response to mechanical stimuli and a loss of bone mass and structural integrity. Chemotherapy agents such as doxorubicin exacerbate the degeneration of aging skeleton and further subject older cancer patients to a higher fracture risk. To alleviate this clinical problem, we proposed and tested a novel mechanobiology-based therapy. Building upon prior findings that i) Yoda1, the Piezo1 agonist, promoted bone growth in young adult mice and suppressed bone resorption markers in aged mice, and ii) moderate tibial loading protected bone from breast cancer-induced osteolysis, we hypothesized that combined Yoda1 and moderate loading would improve the structural integrity of adult and aged skeletons in vivo and protect bones from deterioration after chemotherapy. We first examined the effects of 4-week Yoda1 (dose 5 mg/kg, 5 times/week) and moderate tibial loading (4.5 N peak load, 4 Hz, 300 cycles for 5 days/week), individually and combined, on mature mice (~ 50 weeks of age). Combined Yoda1 and loading was found to mitigate age-associated cortical and trabecular bone loss better than individual interventions. As expected, the non-treated controls experienced an average drop of cortical polar moment of inertia (Ct.pMOI) by −4.3% over four weeks and the bone deterioration occurred in the majority (64%) of the samples. Relative to no treatment, loading alone, Yoda1 alone, and combined Yoda1 and loading increased Ct.pMOI by +7.3%, +9.5%, +12.0% and increased the % of samples with positive Ct.pMOI changes by +32%, +26%, and +43%, respectively, suggesting an additive protection of aging-related bone loss for the combined therapy. We further tested if the treatment efficacy was preserved in mature mice following two weeks (six injections) of doxorubicin at the dose of 2.5 or 5 mg/kg. As expected, doxorubicin increased osteocyte apoptosis, altered bone remodeling, and impaired bone structure. However, the effects induced by DOX were too severe to be rescued by Yoda1 and loading, alone or combined, although loading and Yoda1 individually, or combined, increased the number of mice showing positive responsiveness by 0%, +15%, and +29% relative to no intervention after doxorubicin exposure. Overall, this study supported the potentials and challenges of the Yoda1-based strategy in mitigating the detrimental skeletal effects caused by aging and doxorubicin.

Graphic Abstract

1. INTRODUCTION

A healthy adult skeletal system undergoes continuous remodeling to maintain the mechanical competence required for weight bearing and body movement. Under the mechanical and biochemical cues, osteoblastic bone formation and osteoclastic bone resorption are usually coupled and balanced [1,2]. With aging, such balance is disrupted, leading to decreased bone mass, deteriorated bone structure, and more prevalent osteoporosis and fracture, which are commonly seen in post-menopausal women [3,4]. Aging also increases the risk of cancers [5]. Cancer treatments such as chemotherapy, despite being lifesaving, adversely affect bone and marrow cells [6,7], further exacerbating the problem of skeletal fragility in older cancer patients [8].

The standard care for osteoporotic skeletal fragility includes pharmaceutical and physical interventions [9,10]. The anti-resorptive and pro-anabolic medications such as Zoledronate, Denosumab, Teriparatide, and Romosozumab are effective in treating postmenopausal osteoporosis [11] and managing adverse skeletal events for metastatic cancer patients [12]. However, long-term and/or high-dose use of antiresorptives could have rare but severe skeletal side effects such as atypical fracture and necrosis of the jaw [13]. Rapid bone loss often occurs upon the discontinuation of the biologic medications [14,15]. In contrast, physical exercise is well-recognized for its safety and skeletal benefits and thus regularly prescribed to osteoporotic patients [16] and cancer patients and survivors [17]. Aerobic exercise, including walking, stair climbing/descending, or jogging, is effective in reversing the decline of bone mineral density (BMD) in postmenopausal women [18], and increasing the survival of breast and colorectal cancer patients [19]. In mice with breast cancer, mechanical loading enhances cancellous bone density and prevents bone resorption in a loading intensity-dependent fashion [20,21]. Evidence shows that mechanical loading leads to the opening of Cx43 hemichannels in osteocytes, which in turn hinders tumor formation and development [22]. However, exercise could be challenging for older adults with health problems such as cancer. Furthermore, aged skeletons are less responsive to exercise due to diminished mechanosensing, impaired cellular function, and altered mechanotransduction pathways [23,24]. Therefore, safe and effective treatments to improve aged skeletons and those exposed to chemotherapy are in urgent need.

The recently discovered mechanosensitive Piezo ion channels could act as a new target for interventions [25]. Piezo1 channels, abundantly expressed in bone, are required for loading-induced bone formation [26] and are involved in osteoblast-osteoclast crosstalk [27]. Yoda1, a small molecule agonist of Piezo1 [28], acts as a molecular wedge that reduces the mechanical threshold of Piezo1 activation [29]. Systemic delivery of Yoda1 promotes bone formation in young mice [26] via upregulating downstream Wnt/β-catenin and Yap/Taz pathways [30,31], similar to mechanical loading [32,33]. However, the effects of Yoda1 and mechanical loading on the aging skeleton and after chemotherapy are poorly understood.

We hypothesize that combined Yoda1 and moderate loading would improve the structural integrity of adult and aged skeletons in vivo and protect bones from deterioration after chemotherapy. To test the hypothesis, we performed three experiments in which mature mice were subjected to Yoda1 and unilateral tibial loading individually or combined, with or without doxorubicin, a commonly used cancer chemotherapy agent [34]. Using in vivo sequential microcomputed tomography (microCT) and histological/molecular examination, we found that combined Yoda1 and loading (2–4 weeks) mitigated aging-associated bone loss better than loading alone but could not fully overcome the negative effects of doxorubicin on bone structure, gene transcripts, and cellular apoptosis. Our findings demonstrated the potential and the need for further improvements of the mechanobiology-based strategy in addressing skeletal fragility associated with aging and breast cancer metastasis.

2. METHODS

2.1. Animals and Experimental Groups

Both male and female C57BL/6J mice (Jackson Laboratory, Bar Harbor, ME, USA) were maintained in the University of Delaware’s animal facility following the NIH Guidelines for the Care and Use of Laboratory Animals. The Institutional Animal Care and Use Committee (IACUC) approved all animal experimental protocols.

To examine the adult or aged skeletal responses to Yoda1 and tibial loading, mice of both sexes (13 male and 10 female) at the age of 50 ± 8 weeks (ranging from 36 to 58 weeks) were randomly selected to receive injections of either Yoda1 (2.5 mg/kg body weight, 7 males and 5 females) or vehicle (DMSO, 6 males and 5 females), and then subjected to unilateral tibial loading on the left tibiae while the right tibiae acted as non-loaded controls (Fig. 1A, first row). The Yoda1 and tibial loading treatment was administered 5 times weekly for 4 weeks. To further investigate how chemotherapy affected the skeletal responses of mature mice to Yoda1 and tibial loading, additional mice at the age of 50 ± 4 weeks received six injections of doxorubicin at a dose of 5 mg/kg body weight (High DOX) or 2.5 mg/kg (Low DOX) over two weeks (Fig. 1A, second and third rows). The two doses were chosen to cover the varying dosages reported in the literature [35]. The High DOX mice received either Yoda1 (5 mg/kg, n = 8 [4 males and 4 females]) or vehicle (DMSO, n = 3; all females) while unilateral tibial loading was applied on their left tibiae 5 times per week for 3 more weeks. In addition, an age-matched control group (n = 4; all females) receiving no doxorubicin, no Yoda1, and no tibial loading was also included in parallel with the High DOX groups (Fig. 1A). However, among the 11 High DOX treated mice, four males receiving Yoda1 and one female receiving vehicle died before the completion of the experiment. To reduce the potential toxicity, the Low DOX mice were given a one-week rest period after DOX injection, before they received Yoda1 (2.5 mg/kg, n = 8 [6 males and 2 females]) or vehicle injections (DMSO, n = 6 [4 males and 2 females]) as well as unilateral tibial loading on their left tibiae 5 times per week for 2 more weeks (Fig. 1A). Calcein with its high binding affinity to mineralizing bone surfaces was injected intraperitoneally at a dose of 20 mg/kg body weight 10 and 3 days prior to sacrifice. There was approximately 60–72 hours delay between the last Yoda1/loading treatment and tissue harvesting.

Figure 1. Experiment design of the three in vivo experiments to study the effects of Yoda1 and/or tibial loading on mature mice (~50 weeks of age) with or without exposure of doxorubicin (DOX).

(A) The timeline of application of DOX, Yoda1, and tibial loading including the dosages and microCT scheduled at baseline (Week 0), Week 2, and Week 4 or 5. (B) Up to three regions of interest (ROIs) at the proximal end, mid-diaphysis, and the tibial-fibular junction were chosen to analyze the cortical and trabecular bone changes using the sequential scans.

2.2. Yoda1 Administration

Yoda1 was administered to the mice via intraperitoneal injection one hour prior to the start of tibial loading (Section 2.3). Based on our previous solute transport experiments [36,37], the one-hour circulation time was sufficient for a small molecule like Yoda1 (355 Da) to penetrate bone and activate the Piezo1 channels in embedded osteocytes. Yoda1 powder (CAS#448947-81-7, Sigma-Aldrich) was dissolved in sterile DMSO to make a concentrated stock solution (12.5 mg/mL) and stored in −30°C as small aliquots. The working solution (0.5 mg/mL) was freshly made on injection days by diluting the stock solution with sterile PBS containing 30% w/v 2-hydroxy propyl beta cyclodextrin (2-HPBC, CAS#128446-35-5, Sigma-Aldrich). The use of 2-HPBC increased the solubility of Yoda1 in aqueous solutions. Vehicle DMSO (Veh) solution was diluted in 2-HPBC, similar to the Yoda1 working solution, with a final DMSO concentration of 4% v/v. The injection volumes of Yoda1 and DMSO vehicle solutions were adjusted weekly based on the body weight of the mice.

2.3. In vivo Tibial Loading

Under isoflurane anesthesia (2–3% v/v), the mice were subjected to unilateral tibial loading with the loading parameters (4.5 N peak load, 4 Hz, 300 cycles per day, 5 days per week) as in our previous study [38]. The peak loading was shown to induce moderate surface strains (~630 microstrains) in young mature mice [38]. To alleviate potential pain and stress, the animals received buprenorphine (0.05 mg/kg body weight) once a week.

2.4. Doxorubicin Administration

Doxorubicin powder (Cat# J64000, Thermofisher) was dissolved in sterile DMSO to create a stock solution (50 mg/mL), which was aliquoted and subsequently diluted in sterile DI water to prepare working solutions (0.5 or 1 mg/mL). The mice received six cycles of intraperitoneal injections of doxorubicin at a dose of 2.5 mg/kg or 5 mg/kg over week 1 and week 2. These chemotherapy doses were equivalent to 7.5 and 15 mg/m² in humans [39] and did not exceed the maximum tolerated dose reported for mice [35]. The control mice received equivalent volumes of DMSO (2% v/v) at the same frequency.

2.5. In vivo microCT Scanning and Imaging Analysis

To monitor the changes in bone structure and mineral contents in response to treatments, three sequential microCT scans per mouse were performed at the beginning, middle, and end of the experiments (noted as Week 0, Week 2, Week 4/5, Fig. 1A). The mice were anesthetized via isoflurane (2–3% v/v), and the lower legs were extended and stabilized in a built-in holder of a Skyscan^® 1276 scanner (Bruker, Kontich, Belgium), also utilizing the scanning parameters reported in our previous studies [38]. In brief, multiple projections were captured when the camera was rotated around the mouse legs for 180 degrees at a step of 0.8 degrees using the following imaging settings: 7 μm voxel size, 2000 × 2000 pixels per scan, 900 ms exposure time, X-ray 200 mA current and 50kVp, and a 0.5 mm aluminum filter [38]. Each scan lasted 4–5 minutes, and the x-ray exposure was limited to 500–600 mGy. Two hydroxyapatite phantoms were included to calibrate and measure bone and tissue mineral density.

Reconstructed 3D structures of both tibiae were then generated using NRecon^® software (Bruker), which were separated and registered to a vertically aligned tibia using our custom Python codes as previously described [38]. Our three regions of interest (ROI) consisted of 350 slices (2.45 mm thickness) of bone metaphysis below the growth plate of the proximal tibia (denoted as Prox. End), 50 slices (0.35 mm thickness) at a location 3.15 mm proximal of the tibial-fibular junction (denoted as Mid-Diaphysis), and 50 slices (0.35 mm thickness) immediately proximal of tibial-fibular junction (denoted as TF Junction) (Fig. 1B). These three ROIs allowed us to analyze the spatial responses to the interventions at the sites containing both trabecular and cortical bone, cortices under relatively high bending moment from loading, or mainly cortical bone.

Histomorphometric analysis of cortical and trabecular bone parameters was performed following the application of global thresholds (55/256) using the CT Analyzer 1.17© (Bruker). The outcome measures reported included cortical polar moment inertia (Ct.pMOI), cortical area (Ct.Ar), cortical thickness (Ct.Th) and trabecular bone volume (Tb.BV/TV), number (Tb.N), thickness (Tb.Th), and spacing (Tb.Sp). Furthermore, taking advantage of the repeated scans of the same bones, we were able to identify bone formation and resorption at later time points by registering to and comparing with the baseline scans. A voxel being non-bone at baseline but becoming bone later was defined as new bone, while an initial bone voxel becoming non-bone was defined as bone loss.

The above analysis required the three sequential scans to be of high quality and aligned well. Some mice had to be excluded from the analysis when one or two scans were missing due to unexpected early death of the animals or when the scans were compromised by motion artifacts. It happened more frequently for the DOX-treated mice because a lower isoflurane concentration (2% v/v) was used during scanning to account for their relative weakness. In the end, the microCT data were analyzed and reported for all mice used in the four-week No DOX experiment, the most of age-matched five-week non-treated mice (three females), selected High DOX mice (four females receiving Yoda1 and two females receiving DMSO), and selected Low DOX mice (three males receiving Yoda1 and one male and one female receiving DMSO).

2.6. Dynamic Histomorphometry and immunohistochemistry assays

To assess cellular responses to Yoda1 and tibial loading with and without chemotherapy, the following histological staining and immunohistochemistry assays were performed on either cryosections of undecalcified bone samples or paraffin sections of decalcified samples after they were fixed in 4% v/v paraformaldehyde for two days. All samples were embedded with the medial side facing the cutting surface.

To assess Calcein bone labels, we used both loaded and nonloaded tibiae of No DOX mice treated with Yoda1 or Veh. Due to the small number of DOX-treated mice, we used the nonloaded humerus (Low DOX mice) or tibial distal ends with much smaller loading (High DOX mice) to examine the effects of DOX and Yoda1 (but not loading) on bone formation. The samples were immersed in 30% (w/v) glucose solution for two days and embedded in OCT^® mounting medium for cryo-sectioning as published [40]. Longitudinal sections (8 μm in thickness) were obtained using adhesive cryo-films designed for sectioning undecalcified bones (Cryosection Laboratory, Japan). Whitefield and fluorescent images of the sections were captured using an EVOS 7000 Imager (Thermo Fisher Scientific, Waltham, MA) with a 10x objective. The bone surface and the Calcein labels were quantified manually using the free Fiji ImageJ software (U. S. National Institutes of Health, Bethesda, Maryland, USA, https://imagej.nih.gov/ij/), and the periosteal and endosteal mineralizing surface (Ps.MS/BS; Ec.MS/BS) were derived.

In addition, the cellular activities of osteoblasts and osteoclasts were further studied via staining their signature enzymes such as alkaline phosphatase enzyme (ALP) and tartrate-resistant acid phosphatase (TRAP) using the commercial staining kits (Sigma-Aldrich) [40]. For the NO DOX treated samples, TRAP staining was performed on paraffin sections using the protocol previously published [38]. Briefly, the paraffin slides were rehydrated in graded alcohols and incubated in prewarmed TRAP staining media for 1 h, followed by washing (5 min), counterstaining (1 min), dehydration in graded alcohols (5 min), and coverslip mounting. Since enzymes like TRAP and ALP appeared better preserved in cryosections, the TRAP staining of DOX treated samples was later conducted using cryosections, following the published protocol [40] including rehydration (1 h), staining (15 min), washing (5 min), counterstaining (1 min), and dehydration in graded alcohols. The ALP staining of all samples was performed on cryosections, followed by 1h rehydration, 15 min incubation in the ALP staining buffer (100 mM Tris, 50 mM MgCl2, 100 mM NaCl), 15 min incubation with the ALP substrate buffer (ALP buffer, Fast Red TR Salt, Naphthol AS-MX), and washing, counterstaining, dehydration, and coverslip mounting as described above.

To investigate the apoptosis of osteocytes following the various treatments, terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) staining was performed on paraffin sections of distal tibiae (No DOX groups), proximal tibiae (High DOX groups), and full tibiae (Low DOX groups). The bone samples were decalcified in 14% ethylenediaminetetraacetic acid (EDTA, pH 7.4) for 4 weeks at 4°C and embedded in paraffin. Sequential sagittal sections (5 μm thickness) were collected, air dried for 12 hours, stained using a commercial kit (VB-4005D, VitroVivo Biotech^®) [38] and counterstained with 0.02% Fast Green for better contrast.

The entire sample, bone and marrow, were imaged with a 10X objective for TRAP and ALP staining and a 40X objective for TUNEL staining. Using the autofocus and stitch function of the microscope, a tile image was readily obtained of the regions of interest (bone cortex and marrow). Osteoclastic activity was assessed via TRAP staining on both the endosteum and periosteum. We calculated and reported the percentage of TRAP positive (+) surface (stained red) over the endosteal, or periosteal surface shown in the sections. Osteoblastic activity was measured using the same procedure and reported as the percentage (%) of ALP positive (+) surface. The TUNEL positive (+) and negative (−) osteocytes were identified with the brown stain of the DAB and blue hematoxylin counterstain, respectively, under a Zeiss microscope with the adjustable focus on the nuclei. We counted the number of positive and negative cells on the entire bone cortex and then calculated the % TUNEL+ cells.

2.7. RT-PCR gene expression

In a subset of animals, femoral bone shells were harvested from DOX treated and age-matched control mice at Week 2 and Week 5 by removing bone marrow with a centrifuge in a RNAnase free environment, and snap-frozen in liquid nitrogen and stored at −80°C. The total RNA was extracted with TRIzol (Life Technologies, NY) via the manufacturer’s guidelines. cDNA was prepared using the first strand cDNA synthesis kit (Life Technologies). Using the Taqman assays (Applied Biosystems), the RNA expression levels of signature bone markers such as Wnt1 (Mm01300–555_g1), Tnfsf11 (Mm00441906_m1), Tnfrsf11b (Mm00435452_m1), Sost (Mm00470479_m1), Acp5 (Mm00475698_m1) were measured and reported relative to a housekeeping gene, ribosomal protein S2 (Mrps2, Mm00475529_m1) as published [26].

2.8. Statistical Analysis

Since microCT data passed the normality test, one-way ANOVA was used for analyzing multiple groups at the same time point, while one-way repeated measure ANOVA analysis was performed for the same group at different time points. The Tukey post-hoc test was chosen for conservativeness to identify groups with different means. For histology and PCR data with small sample sizes, the non-parametric Mann-Whitney tests or Kruskal-Wallis tests followed with Dunn’s multiple comparisons were used to detect differences between two groups or multiple groups, respectively. All the statistical analysis was performed in the GraphPad Prism version 9.0.0 for Windows (GraphPad Software, San Diego, California, USA).

3. RESULTS

3.1. Combined Yoda1 and loading mitigated age-associated cortical and trabecular bone loss in mature mice by increasing bone formation and suppressing bone resorption.

As expected, the aging mature mice demonstrated steady decline in both cortical and trabecular bone indices measured via sequential microCT scans at the tibial proximal end and mid-diaphysis (Fig. 2). There was no sex difference for the change in properties under each group, hence the male and female data were pooled (Supplemental Figs. 1S, 2S). Relative to the baseline (Week 0), the age-matched non-treated group (Veh-Nonloaded) showed negative changes with time in the mean values of Ct.pMOI, Ct.Ar, Ct.Vol, Tb.BV/TV, and Tb.Th in the proximal end (Figs. 2B–2F) and similar decline of cortical bone (Ct.pMOI, Ct.Ar, and Ct.TMD) in the mid-diaphysis (Figs. 2J–2L). It is noted that the temporal changes of Ct.pMOI from the baseline, reported herein to confirm the aging-associated bone loss over the four weeks of experiment, were −7.0% and −1.6% of the baseline values for the proximal end and mid diaphysis ROIs (average −4.3%, Table 1). In addition, a majority of the non-treated samples (64% = 1–36%) showed negative changes in Ct.pMOI (Table 1). These changes were consistent with the mature bone phenotype expected for the studied mice. Due to the high-resolution repeated scans, bone remodeling activities, including bone formation and resorption volumes, were clearly detected during the periods of two and four weeks (Fig. 2O).

Figure 2. Combined Yoda1 and loading mitigated the loss of cortical and trabecular bone in mature No DOX mice.

(A-H) The tibial proximal end, cortical and trabecular envelopes (A) were analyzed for Δ Ct.pMOI (B), Δ Ct.Area (C), Δ Ct.Vol (D), Δ Tb.BV/TV (E), Δ Tb.Th (F), Ct.Formation.Vol (G), and Ct.Resorption.Vol (H) on Week 2 or 4 relative to Week 0. Yoda1 and tibial loading combined but not alone reversed bone degradation with aging. (I-O) At the mid-diaphysis (I), similar age-related decline was confirmed in Ct.pMOI, Ct.Ar, Ct.TMD (J, K, L), which was not affected by tibial loading alone. Yoda1 or Combined Yoda1 and tibial loading reversed the decline by promoting bone formation (M) and suppressing bone resorption (N). Bone formation (green) and resorption (red) on periosteal and endocortical surfaces are shown (O). (*: p < 0.05; **: p < 0.01; ***: p < 0.001 using one way ANOVA with Tukey postdoc tests).

Table 1.

Effects of tibial loading and/or Yoda1 on aging bone

Group	Sample size	Proximal End		Mid Diaphysis		Average of both locations
		%ΔCt.pMOI	% samples (ΔCt.pMOI > 0)	%ΔCt.pMOI	% samples (ΔCt.pMOI > 0)	%ΔCt.pMOI	% samples (ΔCt.pMOI > 0)
Veh-Nonloaded	11	−7.0% (7.1%)	27%	−1.6% (5.2%)	45%	−4.3%	36%
Veh-Loaded	11	4.6% (15.7%)	73%	1.3% (5.2%)	64%	3.0%	68%
Yoda1-Nonloaded	12	3.7% (15.3%)	50%	6.6% (8.3%)	75%	5.1%	63%
Yoda1-Loaded	12	7.4% (9.0%)	75%	7.9% (11.9%)	83%	7.6%	79%

Note: %ΔCt.pMOI was obtained as (Week 4-Week 0)/Week 0 and presented as mean (standard deviation).

The combined Yoda1 and tibial loading resulted in more broad improvements on the cortical and trabecular bone parameters at both mid-diaphysis and proximal end in comparison with no treatment, and individual interventions (Fig. 2, Table 1). Compared to the Veh-Nonloaded group, the Yoda1-Loaded group not only reversed the decline of Ct.Ar, Ct.Vol, Tb.BV/TV,and Tb.Th but also positively and significantly increased these parameters at the proximal tibiae (p = 0.01 to 0.06, Figs. 2C–2F) and mid-diaphysis (p = 0.01 to 0.05, Fig. 2J–2K). Combined Yoda1 and loading for four weeks resulted in the highest increase of the Ct.pMOI (+7.6%) and the highest percentage of samples with increasing Ct.pMOI (79%, Table 1). In comparison, we did not detect significant difference between Veh-Loaded and Veh-Nonloaded groups on the measurements (Fig 2), but an increased Ct.pMOI (+3.0%) and a higher % of samples with positive response (68%) were found in Veh-Loaded group, relative to the decreased Ct.pMOI (−4.3%) and the 36% positive responding rate in the Veh-Nonloaded group (Table 1). The Yoda1-Nonloaded group showed significant increases in the mid-diaphysis Ct.pMOI, Ct.Ar, and Ct.TMD vs. Veh-Nonloaded group (Figs. 2J–2L), as well as the increased Ct.pMOI (+5.1%) and the 63% positive responding rate (Table 1). Additional bone parameters from the microCT scans such as Tb.N and Tb.Sp can be found in the Supplemental Fig. 3S.

The dynamic tracking and comparison of the cortexes revealed that, relative to the non-treated group, the Yoda1-Loaded group showed a consistent elevated bone formation (p < 0.05, 0.001 on Week 2 and Week 4, respectively, Fig. 2G) and suppressed bone resorption (p < 0.05 on Week 2, Fig. 2H) in the proximal tibiae. Similar trends were observed at the mid-diaphysis including increased Ct.pMOI and Ct.Ar changes (p < 0.01, 0.05 on Week 2 and Week 4, respectively, Figs. 2J–2K), bone formation (p < 0.05 on Week 2, Fig. 2M), suppressed bone resorption (p < 0.05 on Week 2, Fig. 2N). Notably, the Yoda1 treated groups showed various improvements in Ct.pMOI, Ct.Area, Ct.TMD, and Ct.Formation relative to Veh treated groups (p < 0.01 to 0.05, Figs. 2J–2M).

3.2. Combined Yoda1 and loading led to elevated mineralization and osteoblastic activity in the periosteal surface.

Histological examinations of 2D slices of cryo and paraffin sections confirmed that Yoda1 combined with tibial loading tended to promote osteoblastic bone formation and suppress osteocyte apoptosis (Fig. 3). Using cryo-sections of undecalcified tibiae, calcein bone labels administrated 3 and 10 days prior to sacrifice were examined via fluorescence (green, Fig. 3A) along with bone surfaces under brightfield (not shown). At the epiphyseal trabecular bone region, the four groups (Veh-Nonloaded; Veh-Loaded; Yoda1-Nonloaded; Yoda1-Loaded) had similar mineralizing surface (Tb.MS/BS). Consistent with the Week 4 microCT data (Fig. 2), the Yoda1-Loaded group showed significantly higher Tb.BV/TV, higher Tb.N, and smaller Tb.Sp than the Veh-Nonloaded controls (Fig. 3A). At the mid-diaphysis, the Yoda1-Loaded group showed higher mineralizing Ps.MS/BS than the Veh-Loaded group at the periosteum, while no significance was detected at the endosteum (Fig. 3A). Consistent with the calcein labeling data, Yoda1-Loaded group showed a trend of higher ALP-positive surface (indication of osteoblastic activity) on the periosteal surface than Veh-Loaded and Yoda1-Nonloaded groups (p = 0.3, 0.4, Fig. 3B). There was no difference of ALP-positive surface on the endosteum among the four groups (plot not shown). Using paraffin sections, we found that Yoda1 alone or with loading tended to lower the TRAP-positive surface on the endosteum than the Veh-Loaded group (p = 0.2, 0.4, Fig. 3C), while limited TRAP staining was found along the periosteal surface (not shown). Osteocyte apoptosis showed a reducing trend in the Yoda1-Loaded group relative to the non-treated control (p = 0.09, Fig. 3D).

Figure 3. Histological examination of the effects of Yoda1 and tibial loading on bone and bone cells.

(A) Calcein bone labels (green) at the epiphyseal trabecular bone region and tibial cortices in undecalcified cyro-sections. Yoda1-Loaded group showed higher epiphyseal Tb.BV/TV, Tb.N, and smaller Tb.Sp than nontreated control and a trend of elevated Ps.MS/BS. (B) ALP staining showed a trend of higher % ALP+ surface (indication of osteoblastic activity) on the periosteum. (C) Yoda1 alone or with loading tended to reduce # osteoclasts on endosteal surface than Veh-Loaded group. (D) Osteocyte apoptosis (%TUNEL+) was significantly decreased in Yoda1-Loaded group vs. non-treated control. One way ANOVA and Tukey tests were used in Panel A, and Kruskal-Wallis tests followed with Dunn’s multiple comparisons used in Panels B-D. (*: p < 0.05; **: p < 0.01, ep: epiphyseal; gp: growth plate; Ps: periosteal surface; Ec: endocortical surface; OC: osteoclasts; OCY: osteocytes).

3.3. Doxorubicin tended to reduce bone structural properties, decrease bone formation, and increase bone resorption in mature mice

To evaluate the skeletal effects of two-week administration of DOX only, we examined the changes of cortical bone structural properties such as Ct.pMOI, Ct.Ar, Ct.Th, Ct.Formation.Vol, and Ct.Resorption.Vol by comparing two microCT scans captured on Week 0 and Week 5 in the age-matched control mice and Low and High DOX treated mice (Fig. 4). There was no sex difference in these measures. Hence, we combined the male and female data for the Low DOX batch (Fig. 4S). At this age, the Ct.pMOI, Ct.Ar, and Ct.Th increased slightly for the control mice at the proximal end (Fig. 4A), the mid-diaphysis (Fig. 4B), and TF junction (Fig. 4C). However, treatment of Low and High DOX tended to blunt or arrest such increases with p values ranging from 0.07 to 0.6 in comparison with the No Dox group (Fig. 4). Without doxorubicin, bone formation volumes were greater than bone resorption volumes, contributing the positive changes of Ct.pMOI at the three examined locations (Fig. 4). However, with the increasing dose of DOX treatment, bone formation tended to decrease while bone resorption elevated (Fig. 4). Due to the small sample size, only a trend and no significant difference was found (p = 0.2 to 0.8, Fig. 4A). An overlay of the two scans revealed bone resorption (red) and bone formation voxels (green) within the marrow cavity and the endosteal and periosteal cortices (Fig. 4D).

Figure 4. Dose-dependent decline of bone structure in mature mice receiving doxorubicin (DOX).

Changes of Ct.pMOI, Ct.Ar, Ct.Th, Ct.Formation.Vol, Ct.Resorption.Vol were quantified between Week 5 and Week 0 at (A) the proximal end, (B) mid-diaphysis, and (C) tibia-fibular (TF) junction for age-matched control mice (−), and mice receiving Low DOX (+) or High DOX (++). Representative 3D images showed the spatial distributions of bone resorption (red) and formation (green) at the three sites for the groups. P values were reported using the Kruskal-Wallis tests followed with Dunn’s multiple comparisons.

3.4. Combined Yoda1 and loading did not rescue DOX-induced bone degradation but increased the number of bones showing improved structure.

We combined the Low and High DOX samples to analyze the effectiveness of loading and Yoda1 on protecting bone from DOX-induced bone deterioration. In control mice without the DOX exposure, 100% bone samples showed positive increase in Ct.pMOI (0.01–0.04 mm⁴, ~14%) from Week 2 to Week 5 at the three sites (Fig. 5). Such increases were blunted or even reversed in DOX-treated mice as the average values of ΔCt.pMOI became smaller or even negative (Fig. 5). The effects of DOX were significantly detected at the mid-diaphysis and TF junction with p values ranging from 0.001 to 0.05 (Fig. 5A–5C). No significant effects of tibial loading and Yoda1 were detected, possibly due to the small sample sizes. However, comparing the four groups with DOX exposure, the relative Ct.pMOI changes over the three cortical sites was −0.3%, 0.3%, 0.8%, and 1.4%, for the non-treated, loaded, Yoda1, and combined Yoda1 and loading groups, respectively (Fig. 5D). The percentage of bones showing positive Ct.pMOI change (indicative of no bone loss) were 42% (non-treated group), 42% (loaded alone), 57% (Yoda1 alone), and 71% (combined, Fig. 5D). Thus, loading alone, Yoda1 alone, and combined Yoda1 and loading increased the positively responsive bones by 0%, 15%, and 29% relative to no intervention after DOX exposure.

Figure 5. The effects of tibial loading and Yoda1 on cortical bone structure and remodeling following DOX exposure.

The changes of Ct.pMOI on Week 5 relative to Week 2 when DOX injections concluded were quantified at the proximal end (A), the mid-diaphysis (B), and TF Junction (C). The % of Ct.pMOI change relative to Week 2 and % of samples showing positive changes (indicative of no bone loss) averaged for the three ROIs were shown for each group (D). Representative images show bone formation (green) and resorption (red, E). Low and High DOX samples were pooled in the analysis. (*: p < 0.05, ***: p < 0.001 using one way ANOVA followed with Tukey postdoc tests).

3.5. Dose- and envelope-dependent cellular effects of doxorubicin and Yoda1 on aging bone

To examine the histological responses to Yoda1 following a two-week low or high dose of DOX exposure, longitudinal sections of cortical bones were quantified for ALP staining (osteoblast activity), Calcein bone labels (bone mineralization), TRAP staining (osteoclast activity), and TUNEL staining (osteocyte apoptosis, Fig. 6).

Figure 6. The dose-dependent effects of DOX and Yoda1.

on bone formation marker ALP (A), bone mineralization (B), bone resorption marker TRAP (C), and osteocyte apoptosis (D). Dox was administrated for two weeks at low (+) or high (++) concentration for two weeks; Ec: endocortical surface. Kruskal-Wallis and Dunn’s multiple comparison results were shown above the horizonal lines, while the Mann Whiteney test p values shown below the x labels within <>.

No significant difference was detected among the five groups on the ALP positive staining at the periosteal surface (not shown) and endocortical surface (Fig. 6A). However, there was a trend of decreased %ALP+Ec.Pm seen in High DOX+Vehicle group (40%) relative to No DOX group (67%, p = 0.17, Fig. 6A). Yoda1 treatment did not significantly affect the mineralizing surface in comparison with the Vehicle treatment after Low DOX or High DOX exposure (p = 0.28 or 0.53, Fig. 6A). While a significant suppression of the Ec.sL/BS was observed in the High DOX+Yoda1 (~28%) vs. Low DOX+Yoda1 groups (60%, p = 0.04), Yoda1 treatment promoted bone mineralization in the Low DOX mice relative to the vehicle controls (p = 0.04) but not in the High DOX mice (Fig. 6B). Osteoclastic activities on the endocortical surface (%TRAP+ Ec.Pm) showed a significant reduction in Low DOX+Yoda1 vs. No DOX groups (18% vs 47%, p = 0.04) and an obvious increase in High DOX+Yoda1 vs. Low DOX+Yoda1 (p = 0.007, Fig. 6C). Relative to vehicle, Yoda1 decreased the %TRAP+Ec.Pm in the Low DOX groups (trend p = 0.09) and elevated %TRAP+Ec.Pm in the High DOX groups (trend p = 0.13, Fig. 6C). Very faint TRAP signals were detected on the periosteal surface in the paraffin-embedded sections (Fig. 6C). The elevated TRAP and ALP staining in the High DOX+Yoda1 group demonstrated a high turnover state at the endocortical envelope (Fig. 6A and 6C). Furthermore, High DOX induced five times higher osteocyte apoptosis (80%) than the No DOX control (15%, p = 0.02), while Yoda1 did not have significant effects on Low DOX groups and showed a decreasing trend in High DOX groups relative the vehicle controls (p = 0.27, Fig. 6D).

3.6. Time- and dose-dependent transcript responses to doxorubicin and Yoda1 treatment in aging bone

Transcripts involved in bone remodeling were altered by DOX in a dose and time-dependent manner while Yoda1 did not have significant effects (Fig. 7). For the Wnt1 mRNA expression, there was no acute change at Week 2 induced by Low or High DOX compared with No DOX control (p ≥ 0.7). On Week 5, Yoda1 did not affect Wnt1 expression regardless of the dose of DOX, although Wnt1 showed a decreasing trend from Week 2 to Week 5 in the No DOX group (p = 0.2), Low DOX group (p = 0.07), but not in the High DOX group (Fig. 7A). For Sost, we detected an acute decrease in response to High DOX (a trend on Week 2, p = 0.06), a sustained decrease in response to Low DOX (a trend on Week 5, p = 0.1, Fig. 7B), as well as a significant age-related decline in the Low DOX group (p = 0.01, Week 2 vs. Week 5, Fig. 7B). It is noted that Sost expression was lower in the Low DOX + Yoda1 group than the normal control group on Week 5 (p = 0.07, Fig. 7B). For the TRAP-encoding Acp5 transcript, it also showed a decreasing trend or pattern from Week 2 to Week 5 for the No DOX group (p = 0.06), Low DOX group (p = 0.04), and High DOX group (p = 0.2, Fig. 7C). However, High DOX induced showed a nearly 6-fold acute increase of Acp5 (a trend, p = 0.1), which dropped to the control level on Week 5 (Fig. 7C). Yoda1 treatment tended to increase Acp5 in the No DOX group (p = 0.06) but did not show significant effects on Low DOX group and High DOX group by Dunn’s multiple comparison (p = 0.4) while the effects were marginal significant using Mann Whitney comparisons (p = 0.1, Fig. 7C). The ratio of transcripts (Tnfsf11/Tnfsf11b) encoding RANKL and OPG was elevated acutely on Week 2 for the High DOX group (a trend, p = 0.1), which was subsequently reduced to the No DOX level at Week 5 (Fig. 7D). Although three-week Yoda1 treatment did not significantly alter the transcripts of Wnt1, Sost, Acp5, and Tnfsf11/Tnfsf11b ratio in cortical bone shells when compared with the vehicle treatment, decreased Acp5 (p = 0.05) and Tnfsf11/Tnfsf11b (p = 0.03) were found in the High DOX + Yoda1group vs. the No DOX+Yoda1 group (Fig. 7).

Figure 7. Cortical femoral gene transcripts in response to two-week Low or High DOX and subsequent three-week Yoda1 treatment.

(A) Wnt1 (positive regulator of bone formation); (B) Sost (negative regulator of bone formation); (C) Acp5 (Trap, positive regulator of bone resorption); (D) Ratio of Tnfsf11/Tnfsf11b (Rankl/Opg) indicating osteoclastogenic capacity. There are six bars in each plot, with the first two representing transcripts at Week 2 and the other four bars representing transcripts at Week 5. The Mann Whitney tests were performed for Vehicle vs. DOX on Week 2, Week 2 vs. Week 5, and Yoda1 vs. vehicle on Week 5, with p values indicated using “<>”. The Kruskal Wallis tests followed by Dunn’s comparisons were used for multiple group comparisons on Week 5, with p values shown above the horizontal lines.

4. DISCUSSION

The goal of the present study was to test the efficacy of a novel mechanobiology-based intervention in treating aging and chemotherapy-induced bone loss, which negatively affects many older cancer patients and survivors. Anti-cancer chemotherapeutics are essential treatments for these patients, but they also cause debilitating side effects on the human body, including the skeletal system [41]. Preserving the skeletal mechanical competence in aged subjects receiving chemotherapy is thus an urgent clinical need. Recent studies have demonstrated various skeletal benefits of targeting the mechanosensitive Piezo1 channels via mechanical loading and agonist Yoda1 in young and mature mice [26,42] which led to our hypothesis that Yoda1 in combination with moderate tibial loading could mitigate bone loss induced by aging and doxorubicin. The in vivo findings from the present study demonstrated, for the first time, that combined Yoda1 and loading not only prevented the cortical and trabecular bone loss with aging but also better maintained the skeletal integrity in doxorubicin-treated aging mice compared to no treatment, loading alone, or Yoda1 alone. The results were clinically relevant, given that mature mice used herein represented the high-risk population of interest better than the young rodent models in previous studies [43,44,45]. Despite the small bone changes in adult subjects, our adoption of in vivo sequential microCT scans increased the power to detect structural changes caused by aging, doxorubicin, and the proposed interventions.

Overall, the data partially supported the original hypothesis regarding the dual therapy’s efficacy in mitigating aging-related bone loss. The data demonstrated the additive skeletal benefits of combined Yoda1 and loading in aging mice but did not fully overcome the adverse effects of chemotherapy as originally hypothesized. However, the dual therapy was successful in increasing the percentage of mice showing positive skeletal responses. The therapeutic potentials of the dual therapy in protecting bone from aging were first shown using 50-week-old mice. Using the sequential microCT scanning, we observed a decline of cortical and trabecular bone structure, bone volume, and bone formation in aged mice over four weeks, similar to a previous study in which both male and female C57BL/6J mice exhibited a loss in strength and BMD in an age-dependent manner [46]. It is well documented that aged skeleton shows impaired mechanical responses including dampened bone formation [23] and dysregulated bone remodeling [47]. Despite the age-related mechanotransduction deficits [48], the moderate tibial loading, after being applied for four weeks, was found to fully reverse the loss of cortical structural rigidity (Ct.pMOI) during aging. It was likely that the repeated bouts of tibial loading (expected surface strain ~600 με) [38] activated anabolic genes such as Wnt1 when a single bout failed [23]. We further tested and confirmed the skeletal benefits of Yoda1 (the Piezo1 agonist) and found additional protection from aging-associated bone loss when Yoda1 and tibial loading were combined. The observed skeletal effects after the three treatments showed similar trends such as increased bone formation and decreased bone resorption, although the effect magnitudes changes varied. Given that the dual treatment provided better outcomes including reduced osteocyte apoptosis (a tendency) than the individual interventions, tibial loading and Yoda1 might have activated multiple overlapping signaling pathways [49,50], which could collectively regulate the function and crosstalk of bone cells [51]. Although the specific signaling pathways remain to be determined, it is noted that one of the earliest cellular responses to mechanical loading and Yoda1 is the acute intracellular calcium influx, which could activate various downstream pathways that have been implicated in the skeletal mechano-responses [23,25–31,43].

Accumulating evidence supports the crucial roles of mechanosensitive Piezo1 channels in skeletal development, maintenance, and mechano responses. The Piezo1-Bmp2 pathway has been found to regulate the differentiation and fate determination of bone marrow mesenchymal stem cells (MSCs) in osteogenesis [52–54]. Applying high hydrostatic pressure or Yoda1 on MSCs increased the BMP2 expression and accelerated the osteogenic differentiation of primary MSCs [55–59]. In young adult mice, Li et al. (2021) reported that Piezo1 activation by Yoda1 increased the transcription and protein expression of Wnt1 through nuclear translocation of Yes-associated protein (YAP) and the transcriptional coactivator with a PDZ-binding domain (TAZ) in osteocytes [26]. Furthermore, Piezo1 activation in cultured osteocytes suppressed Sost expression, the negative regulator of osteoblastic bone formation and an important component of the Wnt/β-catenin pathway [60]. Given that deletion of Piezo1 from osteoclasts did not impact bone mass [27], we believe that Piezo1 is not essential for osteoclast function and osteoclasts are unlikely to respond to Yoda1 directly. Instead, Li et al. (2023) found that Yoda1 treatment increased the expression of Tnfrsf11b/OPG (an inhibitor of osteoclast formation) in both young mice and old mice (to a less degree), and the mechanism was via the activation of calcium/calmodulin/mTOR signaling pathway in osteocytes [42]. The finding further supported the osteocytes-osteoclasts crosstalk via the OPG/RANKL axis [51]. Please note that we did not detect the effects of Yoda1 on Wnt1, Sost, Acp5 (TRAP), and Tnfrsf11/ Tnfrsf11b ratio in our limited PCR samples. It was possible that we missed the changes because the samples were not harvested in time and the effects diminished rapidly with time in aged mice [23]. Future studies are needed to identify the intracellular pathways activated by combined loading and Yoda1.

We further investigated the intricate interplay between chemotherapy (doxorubicin), Yoda1 administration, and tibial loading, especially their dose- and site-dependent effects on bone integrity in mice. Our observations confirmed the dose- and site-dependent decline of bone structure in mice treated with DOX, as reported in the literature, showing increased bone resorption and reduced bone formation and BMD [41,8]. Doxorubicin is categorized as an anthracycline antibiotic, a group of chemotherapy drugs known for their ability to insert themselves between DNA strands, disrupt DNA unwinding and separation, hinder the activity of helicase and topoisomerase enzymes, and generate free radicals through redox cycling, leading to not only its high effectiveness in killing cancers but also adverse effects on other organs such as heart, muscle, liver, and kidney [61]. In fact, it was the negative effects on bone and marrow [6–8] that motivated this investigation. Using aged mice, our study confirmed the prior observations including cortical and trabecular bone loss, dramatic increase of osteocyte apoptosis, suppressed osteoblast activity, and elevate transcripts of TRAP and RANKL/OPG that promote osteoclast formation. As elucidated in previous animal and human studies [62,63], free radicals induced by doxorubicin could drive osteoclast formation and bone resorption, which might have been further amplified by the concurrent osteocyte apoptosis via upregulated RANKL/OPG [51].

To alleviate the adverse effects of doxorubicin, exercise regimens have been extensively tested in human trials and animal studies, which overwhelmingly confirm the cytoprotective effects of exercise on heart, muscle, kidney, and liver via increasing endogenous antioxidants, normalizing mitochondrial dysfunction, and reducing proapoptotic signaling [61]. Furthermore, aerobic and strength exercise showed improved bone outcomes in a meta-analysis of 26 randomized controlled trials including 2728 cancer patients at risk of bone loss due to cancer and cancer treatment [64]. Preserved or increased bone mineral density was reported in three trials with 347 breast cancer patients treated with doxorubicin or other chemotherapy [17,65,66]. A review of preclinical studies on doxorubicin-induced cardiomyopathy suggested that exercise exerts protective actions through various mechanisms such as reduced accumulation of doxorubicin in myocardial tissue as well as the modulation of oxidative stress response and apoptotic stimuli, and that exercise prior to chemotherapy appears to be more cardioprotective [67]. In this study, we did observe that moderate tibial loading was skeletal-protective from doxorubicin-induced toxicity in agreement with the literature [64]. Please note that our tibial loading was initiated immediately after the completion of doxorubicin injections, which likely induced fluid flow in the lacunar-canalicular system and promoted the clearance of doxorubicin from the bone tissue [68]. However, future studies are needed to investigate the skeletal effects and the underlying mechanisms when exercise is administrated prior to or during the chemotherapy exposure.

Although this study filled a gap in the literature and revealed the interactions among Yoda1, loading, and chemotherapy on bone health, the cellular and molecular mechanisms remain to be determined. The systemic delivery of Yoda1 could prime the Piezo1-responsive cells such as osteoblasts, osteocytes, and marrow cells to respond to acute mechanical stimulation in a more robust fashion, leading to the long-term benefits such as resistance to the doxorubicin toxicity, and reduced bone resorption. Our in vivo results demonstrated the increasing benefits when Yoda1 was combined with loading in mitigating doxorubicin-induced bone loss, which was in agreement with previous in vitro experiments. Lin et al. (2022) demonstrated that low-magnitude high-frequency vibration (0.3 g, 60 Hz, 1 h) and Yoda1 (10 μM, 2 h) on cultured osteocytes increased the nuclear translocation of YAP and reduced osteoclastogenesis potentials via RANKL/OPG [30].

There are several limitations of this study. Although our chosen doses were within the tolerated doses reported in the literature [35,67], some of the experimental mice could not withstand the side effects of doxorubicin possibly due to the relatively old age used in our study (~50 weeks), resulting a small sample size and a limited number of assays. Increasing the sample size and statistical power is needed to provide more definitive results. Another limitation was the 60–72 hours delay in collecting the samples after the last session of intervention, during which the mechanosensitive transcripts, if there were any downstream of loading and Yoda1 treatment, might have died down as shown in 12-month-old mice [23]. Doxorubicin was chosen in the present study because of the relatively abundant clinical and preclinical data. Whether the findings herein could be translated to the other commonly used chemotherapeutic agents like cisplatin (an alkylating agent) and taxol (an anti-microtubule agent) [69] requires further examination. We suspect that the skeletal benefits of Yoda1 and mechanical loading could be present for the other chemo drugs if there are remaining viable bone cells and that the effect sizes may also vary depending on the drugs’ dose and toxicity on the skeletal system. Given the importance of the bone marrow microenvironment in cancer metastasis [70,71] and bone homeostasis [72], future investigation should focus on the effects of Yoda1, loading, and chemotherapy on bone and marrow compartments.

Despite its limitations, this animal study yielded new in vivo findings that are clinically relevant and supported the further fine-tuning of mechanobiology-based strategies to improve bone health in the context of aging and cancer. In particular, the longitudinal high-resolution micro computed tomography allowed spatiotemporal quantifications of the skeletal responses to mechanical loading, Yoda1, and chemotherapy. Our findings collectively suggest that i) Yoda1 treatment has the potential as a therapeutic intervention to mitigate age-related bone deterioration and ii) combining Yoda1 and mechanical loading offers better skeletal protections following chemotherapy. The results agree with the general observations that exercise improves the well-being and quality of life for cancer patients. Further studies are warranted to elucidate the underlying mechanisms to further improve the skeletal protective strategies for aging cancer patients and survivors.

Highlights.

• Aging and chemotherapy induce bone loss and increase fracture risk

• A dual therapy combining Yoda1 and moderate tibial loading was tested in vivo

• The dual therapy showed additive benefits in mitigating bone loss with aging

• % of bones with improved structural integrity was highest for the dual therapy

• Therapy did not counter effects of doxorubicin but increased positive responses