Senescent Cells Linked to Bone Structure Damage and Loss from Frequent Parathyroid Hormone Treatment

Both aged and young mice received PTH administration at two distinct frequencies (low‐frequency [5 times/week] and high‐frequency [10 times/week], 200 μg/kg/week for 4 weeks) to clarify the differences in the effects of PTH on bone microstructure depending on the aging status of the mice (Figure 1A, Table 1). Bone turnover in rats is two to three times higher than that in humans (Takakura et al. 2016; Yamamoto et al. 2016). Thus, PTH was administered to mice 10 times a week to mimic the daily PTH administration and five times a week to simulate the weekly PTH regimen in clinical settings in humans.

In both aged and young mice, an increased frequency of PTH administration led to a similar increase in bone volume fraction (BV/TV) and trabecular number (Tb.N) while decreasing trabecular separation (Tb.Sp) (Figure 1B–D). However, in aged mice, high‐frequency PTH administration was associated with a decrease in trabecular and cortical bone thickness (Ct.Th) and cortical bone mineral density (BMD), and a significant increase in cortical bone porosity (Ct.Po) (Figure 1E–H, Figure S1).

In contrast, in young mice, low‐frequency PTH administration resulted in trabecular bone thickening, whereas high‐frequency administration did not lead to thinning of either trabecular or cortical bone. In addition, even with high‐frequency PTH administration, no cortical bone porosity was observed in the femoral diaphysis, and bone mineral density (BMD) remained unchanged (Figure 1E–H, Figure S1). Mechanical testing was performed to assess whether these quantitative changes correlated with alterations in bone strength. It revealed that low‐frequency PTH administration increased the failure load in both aged and young mice; in contrast, high‐frequency PTH administration further improved the failure load only in young mice but not in aged mice (Figure 1I). These results show that the effect of PTH on bone microstructure and strength differed between young and aged mice.

Given the observed differences in the bone microstructure between young and aged mice after PTH administration, serum bone metabolism markers were investigated. Amino‐terminal propeptide of type I collagen (P1NP) and tartrate‐resistant acid phosphatase 5b (TRAcP5b) levels were elevated after the 4‐week PTH treatment in both aged and young mice; however, in aged mice, the increase in TRAcP5b was significantly greater than that in young mice (Figure 2A,B). Tartrate‐resistant acid phosphatase (TRAP) staining showed that the increase in the number of osteoclasts due to the increased frequency of PTH administration was greater in aged mice than in young mice (Figure 2C,D). These findings were consistent with the results of osteoblast surface per bone surface (Ob.S/BS) and osteoclast surface per bone surface (OC.S/BS) as assessed by hematoxylin and eosin (HE) staining (Figure S2A–C). These results show that the activity of osteoclasts is enhanced by PTH administration in aged mice compared to that in young mice.

Next, we investigated the association between bone microstructural changes induced by PTH administration and the accumulation of senescent cells in aged and young mice, using polymerase chain reaction (PCR) and immunohistochemistry (IHC). PTH administration induced cellular proliferation, as evidenced by Ki‐67‐positive staining, predominantly in mesenchymal stem cells (MSC) and osteoblast‐lineage‐enriched cells in both young and aged mice (Figure S3C). In aged mice, the expression levels of p16INK4a (cyclin‐dependent kinase inhibitor 2A) and SASP factor genes—IL1α (interleukin‐1 alpha), MMP13 (matrix metallopeptidase 13), and Ccl5 (C‐C motif chemokine ligand 5)—were significantly increased by PTH administration in a frequency‐dependent manner (Figure 2E–K). IHC staining for p16INK4a showed increased localization of senescent cells in the high‐frequency PTH group, with prominent accumulation in regions adjacent to trabeculae (Figure 2L).

In contrast, in young mice, the number of senescent cells and the expression levels of p16INK4a and SASP genes were not increased by PTH administration (Figure S4A–I). These results revealed distinct differences in senescent cell accumulation between aged and young mice in response to PTH administration.

To directly investigate the relationship between PTH administration and senescent cell accumulation, mouse calvarial osteoblasts were treated with PTH. Given that the mechanistic Target of Rapamycin Complex 1 (mTORC1) is a pivotal promoter of cellular senescence, IHC was performed to detect the downstream markers of the mTORC1 pathway, including phosphorylated S6 kinase (p‐S6K) and phosphorylated eukaryotic translation initiation factor 4E‐binding protein 1 (p‐4EBP1), to confirm their activation following high‐frequency PTH administration (Figure 3A). Western blot analyses further demonstrated the direct effect of PTH on mTORC1 activity in osteoblasts, as evidenced by the phosphorylation of S6K and 4EBP1 under PTH treatment with a concentration of 100 nM (Figure 3B). Consistent with previous reports, phosphorylation of extracellular signal‐regulated kinase (ERK) and phosphoinositide 3‐kinase (PI3K) upregulation were observed upstream of mTORC1 (Marino and Bellido 2024; Martin et al. 2021; Yamamoto et al. 2007). These findings suggest that an increased frequency of PTH administration in old mice, where aging stresses such as replicative stress accumulation can induce osteoblast senescence through the activation of the mTORC1 pathway (Herr et al. 2024).

Senescent osteoblasts are involved in bone remodeling and cortical bone thinning (Ukon et al. 2024). The observed changes in bone microstructure and the localization of senescent cells adjacent to trabeculae in aged mice following PTH administration suggest that PTH administration promotes the accumulation of senescent osteoblasts.

To verify that the senescent cells that accumulated in the bone following PTH administration were osteoblast‐lineage‐enriched cells, double immunostaining was performed on osteopontin (OPN) and p16INK4a. The results confirmed a significant increase in the number of OPN‐ and p16INK4a‐positive cells after high‐frequency PTH administration. These findings suggest that PTH administration induced osteoblast‐lineage‐enriched senescence in aged mice and that the accumulation of senescent osteoblast‐lineage‐enriched cells may be related to different responses to PTH treatment between aged and young mice.

Next, we investigated whether the combination of senolytic drugs, which selectively eliminate senescent cells, with PTH administration can eliminate senescent cells and prevent the deterioration of bone microstructure associated with PTH administration in aged mice. The senolytic drug combination of dasatinib and quercetin (D + Q) was co‐administered with PTH 10 times per week to aged mice over a 4‐week period (Figure 4A and Table 2). The frequency of PTH administration was determined based on the observed differences in trabecular and cortical bone morphology, as well as the accumulation of senescent cells between young and aged mice. Microstructural analysis using micro‐computed tomography (CT) revealed that the combination of PTH and D + Q reduced cortical porosity compared to PTH alone (Figure 4B,C). In contrast, no significant changes were observed in the trabecular bone and Ct.Th following the combination treatment (Figure 4D–H). Finite element analysis (FEA) revealed that co‐administration of D + Q with PTH increased spine failure load compared to PTH treatment alone (Figure S5C). Simultaneous administration of PTH and D + Q reduced the expression levels of p16INK4a and SASP‐related genes in the vertebrae compared to PTH administration alone (Figure 4I–N). IHC staining for p16INK4a and OPN indicated a decrease in the number of senescent osteoblast‐lineage‐enriched cells in the group undergoing combined PTH and D + Q treatment compared to that in the PTH alone group (Figure 5A,B). In the analysis of serum bone metabolism markers, co‐administration of D + Q did not alter P1NP levels but significantly decreased TRAcP5b levels (Figure 5C,D). The expression of tumor necrosis factor superfamily member 11 (Tnfsf11), whose transcript is a receptor activator of nuclear factor‐κB ligand (RANKL), was increased by PTH and exhibited a trend toward reduction with combined D + Q administration (p = 0.09). The expression of Tnfrsf11b, whose transcript is osteoprotegerin (OPG), was significantly upregulated with the combined PTH and D + Q treatment (Figure 5E,F). These findings were consistent with the results of OC.S/BS, as assessed by HE staining. Specifically, the increase in osteoclasts induced by PTH was suppressed by the co‐administration of D + Q (Figure S2D). Such results indicate that the accumulation of senescent cells induced by PTH administration, along with the activation of osteoclasts mediated by the SASP, contributes to the increase in cortical porosity observed in aged mice following PTH treatment.

Finally, we investigated whether the accumulation of senescent cells induced by PTH administration contributes to the rapid bone loss observed following the discontinuation of PTH treatment (Leder et al. 2009) and whether the clearance of senescent cells can mitigate bone loss caused by deterioration of the bone microstructure in trabecular and cortical bones. In aged mice, an increase in BV/TV was observed during the 4 weeks of PTH administration; however, the subsequent 4 weeks of vehicle treatment alone resulted in a rapid decline in BV/TV, returning to levels comparable to those of the control group (Figure 6A,B). Co‐administration of D + Q during the PTH treatment period and for 4 weeks after PTH withdrawal effectively maintained the increased BV/TV during the 4‐week PTH treatment and prevented the loss of BV/TV following discontinuation of PTH treatment (Figure 6B). Furthermore, D + Q administration mitigated the deterioration of both trabecular and cortical bone microstructural parameters, including BV/TV, Tb.N, Tb.Sp, and Ct.Po after the discontinuation of PTH treatment (Figure 6B–H). Although a mild improvement in BV/TV was observed following 8 weeks of D + Q treatment, no significant alterations were detected in other bone structural parameters (Figure S7B–G). Conversely, in young mice, BV/TV did not decrease with the 4 weeks of vehicle administration following the initial 4‐week PTH treatment (Figure S6A,B). No significant differences in bone structural parameters were observed between the vehicle and D + Q co‐treatment groups (Figure S7H–M). Similarly, other bone microstructural parameters in young mice did not deteriorate following PTH discontinuation, and D + Q administration did not induce significant changes in these parameters (Figure S6B–H). These findings suggest that rapid bone loss following the discontinuation of PTH administration in aged mice is at least partly driven by the accumulation of senescent osteoblast‐lineage‐enriched cells and that bone loss can be mitigated by preventing the accumulation of senescent cells.

Three‐month‐old (young) and 18‐month‐old (aged) C57BL/6J male mice were obtained from Charles River Laboratories Japan Inc. (Kanagawa, Japan). The mice were maintained on a normal chow diet with a 12‐h light/dark cycle.

A high‐resolution micro‐CT scan (Skyscan 1272, Bruker, Billerica, MA) was used at a resolution of 5 μm per voxel. Parameters, including bone volume, tissue volume, trabecular and cortical thickness, trabecular separation, cortical porosity (Harrison et al. 2020), and mineral density, were analyzed using CTAn software (Bruker). Mineral density calibration was performed using a bone calibration phantom. Three‐dimensional reconstructed images were obtained using CTvox software (Bruker). The analysis area of the vertebrae was from a point 0.3–0.6 mm proximal to the distal growth plate, and the area of the femur was 6–8 mm proximal to the distal growth plate. Briefly, cortical bone analysis was performed using a region of interest defined as a 2‐mm segment starting 6 mm proximal to the distal femoral growth plate (Piemontese et al. 2017).

Total RNA was isolated using the TRIzol reagent (Thermo Fisher Scientific, Waltham, MA). Bone (vertebrae, femurs, and tibiae) preprocessing for in vivo assays was performed, as previously described (Farr et al. 2016; Stern et al. 2012). Briefly, we used femoral, tibial, and lumbar bones from mice for RNA isolation. Bones were minced into small pieces and subjected to two sequential 30‐min collagenase digestions (Sigma‐Aldrich, St. Louis, MO). Total RNA was extracted from the remaining bone pieces. These digestions contain osteoblast‐ and osteocyte‐enriched cells (Farr et al. 2016; Stern et al. 2012). Complementary DNA was synthesized using ReverTra Ace (TOYOBO, Osaka, Japan) and used as a template for qPCR with FAST SYBR Green Master Mix (Thermo Fisher Scientific). Target mRNA levels were normalized to the reference gene expression. The median threshold cycle was compared with that of the control sample, and the fold difference between the reference and target genes was calculated. Glyceraldehyde‐3‐phosphate dehydrogenase (GAPDH) was used as an internal control for in vivo assays. The primer sequences used for RT‐qPCR are listed in Table S1.

A three‐point bending test was conducted using a mechanical testing machine (Instron‐5565; Instron, Norwood, MA) to investigate the biomechanical strength of the femoral bone. The femurs were placed on a pedestal with the anterior side facing downward. A constant vertical compression load (3 mm/min) was applied to the midpoint of each specimen until failure. Load and displacement data were acquired every 10 ms. From the load–displacement curve, the ultimate load at failure (N; maximum load that the bone could bear) was determined.

FEA was performed to estimate the mechanical properties of mouse lumbar vertebrae and femora based on micro‐CT imaging data. Lumbar vertebrae (L4) and femora were scanned using high‐resolution micro‐CT (Skyscan 1272, Bruker; voxel size: 5 μm). The acquired 3D datasets were segmented using a global thresholding method to distinguish mineralized bone from soft tissue. Image processing, segmentation, and conversion to finite element meshes were performed using TRI/3D‐BON‐FCS64 Bone Morphometry Software (RATOC System Engineering Co. Ltd., Tokyo, Japan). For vertebral analysis, a linear elastic and isotropic material model was used. Uniaxial compression was applied to the superior endplate while the inferior endplate was fully constrained. Failure load and energy to failure were computed. For femoral analysis, FE models were generated under three‐point bending test conditions, consistent with the in vitro setup. To reproduce the mechanical boundary conditions accurately, the femur was fixed at two support points located 4 mm proximal and 4 mm distal to the mid‐diaphysis. A downward displacement was then applied at the central loading point to simulate bending. Estimated failure load and failure energy were calculated and compared across groups.

Quantitative assessment of bone formation and resorption surfaces was performed on HE‐stained decalcified sections. Bone‐forming surfaces were identified based on the presence of cuboidal osteoblasts lining the trabeculae, while bone‐resorbing surfaces were defined by multinucleated osteoclasts in resorption lacunae. The proportions of these surfaces were quantified by manual counting performed by a bone histomorphometry specialist blinded to the experimental groups.

Primary osteoblasts were obtained from the calvaria of neonatal mice (postnatal age 2–3 days) (Jonason and O'Keefe 2014). Briefly, calvaria were minced and digested four times in alpha‐Minimum Essential Medium (α‐MEM) (Nacali Tesque, Kyoto, Japan), 0.176% NaHCO₃, and 0.008% collagenase type II (ThermoFisher Scientific) at 37°C for 15 min. Supernatants from the third and fourth digestions were collected. After centrifugation, cell suspensions were incubated in 10‐cm dishes with α‐MEM, 10% Fetal bovine serum (FBS), and 1% penicillin–streptomycin at 37°C with 20% O₂ and 5% CO₂. After 1 week, confluent cells (passage 0) were used for experiments.

The effects of PTH on osteoblast proliferation and gene expression were examined in vitro. Murine calvarial osteoblasts were seeded in six‐well plates at a density of 1.0 × 10⁵ cells per well. Cells were maintained in α‐MEM, supplemented with 10% FBS and 1% penicillin/streptomycin solution. Cells were incubated at 37°C in humidified incubators with 5% CO₂. The cells were treated with 100 nM PTH for 24 h. After treatment, the cells were harvested for protein isolation. Protein concentrations were measured using bicinchoninic acid (Thermo Fisher Scientific), according to the manufacturer's protocol. The cell lysates were separated on 4%–12% Bis‐Tris gels (Life Technologies, Carlsbad, CA) and transferred to polyvinylidene difluoride membranes (Nippon Genetics, Tokyo, Japan), which were then incubated in Tris‐buffered saline containing 5% skim milk and Tween 20 (TBS‐T) at 22°C. The blocked membranes were incubated with primary antibodies in Can Get Signal Solution 1 (Toyobo) at 4°C overnight, after which they were incubated with secondary antibodies in Can Get Signal Solution 2 (Toyobo) at room temperature for 1 h. After washing with TBS‐T, the immunoreactive bands were visualized using the ECL Plus Western Blotting Detection System kit (Thermo Fisher Scientific) and ChemiDOC Touch (Bio‐Rad, Hercules, CA). The following antibodies were used for western blotting at the indicated dilutions: rabbit monoclonal anti‐4E‐BP1 (#9644, 1:1000, Cell Signaling Technology), rabbit monoclonal anti‐phospho‐4E‐BP1 (#2855, 1:1000, Cell Signaling Technology), rabbit monoclonal anti‐p70 S6 kinase (#9202, 1:1000, Cell Signaling Technology), rabbit monoclonal anti‐phospho‐p70 S6 kinase (#9234, 1:1000, Cell Signaling Technology), rabbit monoclonal anti‐p44/42 MAPK (Erk1/2) (#4695, 1:1000, Cell Signaling Technology), rabbit monoclonal anti‐phospho‐p44/42 MAPK (Erk1/2) (#4370, 1:1000, Cell Signaling Technology), rabbit monoclonal anti‐PI3 Kinase p110α (#4249, 1:1000, Cell Signaling Technology), and rabbit polyclonal anti‐β‐Actin (#4967, 1:1000, Cell Signaling Technology).

The excised samples were fixed in 10% neutral‐buffered formalin, decalcified with ethylenediaminetetraacetic acid, embedded in paraffin wax, and cut to a 3.5‐mm thickness. Bone sections were deparaffinized and dehydrated for immunostaining. The antigens were activated in 10 mM citrate buffer at 80°C for 30 min. After quenching the endogenous peroxidase activity with methanol containing 3% H₂O₂ for 10 min, the sections were blocked with Blocking One (Nacalai Tesque) for 20 min at room temperature. Sections were then incubated with a primary antibody overnight at 4°C, followed by incubation with a horseradish peroxidase‐labeled secondary antibody for 1 h. Finally, the labeled sections were stained with Histofine Simple Stain Mouse MAX PO (Nichirei Bioscience, Tokyo, Japan) and counterstained with hematoxylin.

For immunofluorescence, the bone sections were deparaffinized, rehydrated, permeabilized for 30 min with 0.3% Triton X‐100 in PBS, and blocked with Blocking One (Nacalai Tesque) at room temperature for 20 min. Blocked sections were incubated with primary antibodies overnight at 4°C. After incubation with the primary antibody, the sections were washed with PBS and incubated with the appropriate Alexa Fluor‐conjugated secondary antibodies (1:1000, Thermo Fisher Scientific) diluted in PBS for 1 h at room temperature. Sections were washed with PBS before mounting with ProLong Diamond Antifade Mountant with 4′,6‐diamidino‐2‐phenylindole (DAPI) (Thermo Fisher Scientific). The following primary antibodies were used: rabbit monoclonal anti‐Cyclin Dependent Kinase Inhibitor 2A (CDKN2A)/p16INK4a (ab211542, 1:250, Abcam, Cambridge, UK), rabbit monoclonal anti‐phospho‐S6 ribosomal protein (Ser240/244) (#5364, 1:500, Cell Signaling Technology), rabbit monoclonal anti‐phospho‐4EBP1 (1:800), mouse Leptin R antibody (AF497, 1:250, R&D Systems), mouse Dentin Matrix Protein 1 (DMP‐1) antibody (AF4386, 1:250, R&D Systems), Runt‐related Transcription Factor 2 (RUNX2) rabbit monoclonal antibody (#12556, 1:500, Cell Signaling Technology), anti‐osteocalcin antibody (ab309521, 1:500, Abcam), anti‐Ki67 antibody (ab15580, 1:500, Abcam), mouse Interleukin‐1 alpha (IL‐1α) antibody (AF‐400‐NA, 1:250, R&D Systems), and goat anti‐osteopontin antibody (AF808, 1:100, R&D Systems, Minneapolis, MN). Secondary antibodies used were donkey anti‐rabbit IgG conjugated to AF 594 (A21207, Invitrogen, Carlsbad, CA) and donkey anti‐goat immunoglobulin G (Zhu et al. 2015) conjugated to AF 488 (ab150129, Abcam). BZ‐X800 (Keyence, Osaka, Japan) was used to observe and capture the fluorescence images. The number of CDKN2A/p16INK4a and osteopontin‐positive cells in five randomly selected images of 270 × 360 μm in size from each section was counted automatically.

Serum was collected from anesthetized mice via cardiac puncture and stored at −80°C for biochemical assays. P1NP (ng/mL) and TRAcP 5b (U/L) were quantified using a Rat/Mouse P1NP enzyme immunoassay kit (Immunodiagnostic Systems, Boldon, UK) and Mouse TRAcP 5b ELISA kit, respectively.

Statistical analyses were performed using GraphPad Prism version 10.2.3 (GraphPad Software, San Diego, CA). All data are presented as mean ± standard error of the mean (SEM). One‐way and two‐way analysis of variance (ANOVA) were used to determine the statistical significance, as indicated in the legends. Statistical significance was set at p < 0.05. The sample sizes for each experiment are described in the corresponding figure legends.

All mice were cared for in strict compliance with and approval from the Institutional Animal Committee of our institution (approval number: 04‐082‐003) and restrictedly followed ARRIVE (Animal Research: Reporting of In Vivo Experiments) guidelines and the National Institutes of Health Guide for the Care and Use of Laboratory Animals.