Introduction

Advances in cancer treatment, such as chemotherapy, have contributed to a steady rise in cancer survivors over the past 50 years1. As of January 2022, an estimated 18.1 million people are cancer survivors in the United States alone and this number is expected to continue growing1,2. While improving disease-free survival, chemotherapy has been linked to accelerated biological aging and frailty3,4,5,6,7. A decrease in bone mineral density, deterioration of bone microarchitecture, and an increased risk for fractures are common consequences of chemotherapy8,9,10,11,12,13.

After development and growth have ceased, bone is continually remodeled via the work of osteoclasts, giant multinucleated cells specialized in resorbing bone, and osteoblasts which produce bone matrix and replace the bone removed by the osteoclasts. Bone loss occurs when the amount of bone resorbed is larger than the amount of new bone formed14. The loss of bone with chemotherapy has been attributed to both an increase in osteoclasts and a decrease in osteoblasts15,16,17,18,19,20. Chemotherapy administration to breast cancer patients, the most common cancer among women, often causes bone loss21. Cancer incidence increases with aging when women experience a large amount of bone loss due to aging itself combined with loss of estrogens at menopause8,14,21,22,23,24. However, breast cancer also impacts many young women21,24. Because bone regenerates throughout life, it is unclear whether the consequences of chemotherapy during youth remain with aging1,10,13,25,26. Administration of diverse chemotherapy drugs such as doxorubicin, paclitaxel cyclophosphamide, etoposide, cisplatin, or carboplatin to disease-free young mice decreases bone mass15,16,17,18,19,20. Nonetheless, the mechanisms responsible for chemotherapy-induced bone loss remain poorly understood. Chemotherapy drugs increase inflammation, DNA damage, and apoptosis27,28. Some chemotherapy regimens also cause cellular senescence—a process in which cells stop dividing and become resistant to apoptosis29,30. The cell cycle arrest in senescent cells is due to the upregulation of cyclin-dependent kinase (CDK) inhibitors such as Cdkn2a (p16) and Cdkn1a (p21). Senescent cells also secrete cytokines, chemokines, and proteases, known as the senescence-associated secretory phenotype (SASP)31. Chemotherapy increases the expression of SASP factors in various tissues, including bone, of young mice17,29,32,33,34,35,36,37.

Administration of senolytics—drugs that kill senescent cells—along with genetic approaches in mice have been used to evaluate the contribution of cellular senescence to tissue degeneration due to aging and other causes38,39. Several types of senolytics have been described, including the natural product piperlongumine (PL) and the combination of dasatinib (D), a tyrosine kinase inhibitor and quercetin (Q), a naturally occurring flavonoid with diverse biological activities38,39,40. As in other tissues, the number of senescent cells increases in the aged skeleton including bone, joints, and intervertebral discs41,42. Administration of D + Q to aging mice eliminates senescent bone cells, increases bone mass, and attenuates the degeneration of intervertebral discs41,42. D + Q also attenuates the deleterious consequences of chemotherapy in brain, kidney, ovary, and heart35,37,43,44. In addition, genetic ablation of p16-expressing cells has implicated senescence in chemotherapy-induced bone loss17.

A combination of doxorubicin (Adriamycin®), cyclophosphamide (Cytoxan®), and docetaxel (Taxotere®), known as TAC45,46, is commonly administered as a breast cancer treatment. In this study, we examined whether administration of the senolytics D + Q or PL attenuates the effects of TAC on bone mass. We also determined whether the skeletal consequences of chemotherapy remain long after cessation of treatment.

Materials and methods

Ethical approval and statements

Animal studies reported in this manuscript complied with the Animal Research: Reporting of In Vivo Experiments (ARRIVE) guidelines 2.0 (https://arriveguidelines.org/arrive-guidelines) to ensure methodological rigor and reproducibility of methods and findings. All animal procedures (protocol #4178) were reviewed and approved by the Institutional Animal Care and Use Committee of the University of Arkansas for Medical Sciences in compliance with the Public Health Service Policy on Humane Care and Use of Laboratory Animals, the National Institutes of Health Guide for the Care and Use of Laboratory Animals (2011) as well as the American Veterinary Medical Association Animal Euthanasia Guidelines (2020). All the experiments and methods were performed in accordance with the relevant guidelines and regulations.

Drug preparation

Doxorubicin (2 mg/kg; Hikma Pharmaceuticals, London, UK), cyclophosphamide (50 mg/kg; Amneal Pharmaceuticals, Bridgewater, NJ, USA), and docetaxel (8 mg/kg; AuroMedics Pharma LLC, USA) solutions were prepared with sterile physiological saline (0.9% sodium chloride). On the day of injection drugs were diluted and mixed. Dasatinib (5 mg/kg; LC Laboratories, Woburn, MA, USA) and quercetin (50 mg/kg; Sigma-Aldrich, St. Louis, MO, USA) were prepared in 30% polyethylene glycol (PEG 400) on the day of administration. Piperlongumine (Sigma-Aldrich, St. Louis, MO, USA) was prepared in dimethyl sulfoxide (DMSO; Fisher Scientific™) on the day of administration. Doses of chemotherapeutic agents were adapted from Flanigan et al.47 using the body surface area normalization method. D + Q41 and PL48 doses were based on previous studies.

Animal studies

Female C57BL/6 and BALB/cJ mice were obtained from the Jackson Laboratory (Bar Harbor, ME, USA). Mice were housed in specific-pathogen-free conditions in groups of up to five per cage and were given one week to acclimatize to the facility before any experimental procedures. Animal facility was maintained at an ambient temperature of 23–25 °C with 30–70% humidity and a 12-h light–dark cycle. The mice had access to water and standard chow, containing 14% protein and 4% fat ad libitum. Four independent experiments were performed.

In experiment 1 (Fig. 1), mice were administered a combination of doxorubicin and cyclophosphamide (AC) with or without the senolytic D + Q. Specifically, 16-week-old C57BL/6 mice were randomly distributed into four treatment groups based on spine bone mineral density (BMD) measurements: saline + vehicle, AC + vehicle, saline + DQ, and AC + DQ. BMD was determined in live mice using dual-energy X-ray absorptiometry with a PIXImus densitometer (GE Healthcare Lunar, software v2.0) as previously described49. AC or saline was administered via intraperitoneal (IP) injection once a week for three cycles. D + Q or vehicle solution was administered via oral gavage 24 h after each chemotherapy cycle. Mice were sacrificed one day after the last injection. In experiment 2 (Fig. 3), mice were administered AC, as above, for four cycles, followed by docetaxel once a week for four cycles via IP injection. D + Q or vehicle solution was administered after each AC and T chemotherapy cycle. Weekly administration of D + Q was maintained for an additional five weeks after cessation of chemotherapy. Mice were sacrificed one week after the last D + Q administration. Experiment 3 (Fig. 5) followed a similar experimental protocol to experiment 2, with D + Q being replaced by PL. In experiment 4 (Fig. 7), 19-week-old BALB/cJ mice were administered the TAC regimen, as described in experiment 2, and sacrificed 6 or 12 months after cessation of treatment. All mice were euthanized by exposure to CO2, followed by cervical dislocation. No mice were excluded from the analysis. However, one femur damaged during tissue collection was omitted from Fig. 3b.

Fig. 1
figure 1

D + Q failed to prevent early bone loss caused by AC. (a) Schematic of experimental design created with BioRender.com. (b) Percent change calculated from measurements taken the day before the beginning of treatment and before sacrifice. (c) μCT analysis of trabecular bone volume over tissue volume (BV/TV) and (d) trabecular microarchitecture in fifth lumbar (L5) vertebrae after sacrifice (n = 9/group). (e) Quantitative RT-PCR analysis of mRNA from the sixth lumbar (L6) vertebrae. Bars represent mean ± SD; p-values by 2-way ANOVA with Tukey’s multiple comparisons test. The most relevant comparisons between individual groups are displayed.

Full size image

Micro-computed tomography (μCT) analysis

Following euthanasia, lumbar vertebrae (L5) and right femora were dissected, cleaned of adherent soft tissues, and fixed overnight in 10% Millonig’s solution (Leica Microsystems, Germany). Bones were then dehydrated in a graded ethanol series (70–100%) and preserved in 100% ethanol until analysis. Micro-CT analysis was performed using a µCT 40 system (Scanco Medical AG, Switzerland) to determine the microarchitecture of the bones, as previously described50,51. 3D voxel images (1024 × 1024 pixels) were acquired using the following scan settings: 12 µm isotropic voxel size, 55 kVp X-ray tube potential, 145 µA X-ray intensity, and 220 ms integration time. Images were processed using Gaussian filtration (sigma = 0.8, support = 1) and a 200 mg/cm3 threshold to segment bone tissue. The thickness of the femoral cortex was measured at the mid-diaphysis (18 slices) and the distal metaphysis (50 slices). Femoral trabecular bone was measured at the distal femur, starting 8–10 slices away from the growth plate and continuing towards the diaphysis (151 slices) to avoid the primary spongiosa. The entire vertebrae were scanned with a transverse orientation and trabecular bone analyses were performed on contours of cross-sectional images, excluding cortical bone and the primary spongiosa. Trabecular bone volume per total bone volume fraction (BV/TV), trabecular thickness, trabecular number, and trabecular separation were calculated using 3D standard microstructural analysis. All nomenclature, symbols, and units follow the American Society for Bone and Mineral Research recommendations52.

Reverse transcription-quantitative polymerase chain reaction (RT-qPCR)

Following euthanasia, the tibiae, right femora, L6 vertebrae, liver, and kidney were dissected. Cortical bone pieces were prepared from the tibiae and femora by removing the periosteum with a scalpel, cutting both bone ends, and flushing the marrow with a 23-gauge needle using sterile PBS. The cortical bone pieces, L6 vertebrae, liver, and kidney were then snap-frozen in liquid nitrogen and stored at − 80 °C for subsequent RNA isolation. Frozen specimens were pulverized with a multi-well tissue pulverizer (BioSpec Products, Inc. USA) followed by total RNA extraction with TRIzol reagent (Thermo Fisher Scientific Inc, USA) according to the manufacturer’s instructions. Complementary DNAs were generated from 1 μg of RNA using the High-Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific Inc, USA). Real-time qPCRs were performed using TaqMan™ Gene Expression Master Mix and gene specific TaqMan™ (Thermo Fisher Scientific Inc, USA) probe and primer sets in a QuantStudio 3 PCR System (Thermo Fisher Scientific Inc, USA). The TaqMan™ probe and primer sets used were the following: Acp5 (Mm00475698_m1); Alpl (Mm00475831_m1); Bglap (Mm03413826_mH); Ccl2 (Mm00441242_m1); Ctsk (Mm00484039_m1); Cxcl12 (Mm00445553_m1); Cdkn1a (Mm00432448_m1); Cdkn2a (Mm00494449_m1); Cd38 (Mm01220906_m1); Il-1α (Mm99999060_m1); Il-1b (Mm00434228_m1); Il-6 (Mm00446190_m1); Hmgb1 (Mm00849805_gH); Mmp-13 (Mm00439491_m1); Mrsp2 (Mm00475528_m1); and Tnfsf11 (Mm00441908_m1). Transcripts were calculated relative to the housekeeping gene ribosomal protein S2 (Mrps2) and shown as fold change using the 2−∆∆Ct method53.

Statistical analysis and reproducibility

Sample numbers were based on previous studies and were not statistically predetermined. Data visualization and statistical analyses were performed using GraphPad Prism 10.0 software (GraphPad Software, Inc. USA). In every graph, each dot represents one mouse. The normality of data distribution was assessed by D’Agostino-Pearson, Anderson–Darling, Shapiro–Wilk, and Kolmogorov–Smirnov tests. Variance homogeneity was tested using F-test (P > 0.05) for two groups and Spearman’s test for more than two. P values between two normally distributed groups were determined by unpaired two-tailed Student’s t-test. Welch’s correction was applied for unequal variances. An unpaired two-tailed Mann–Whitney test was used for non-normal data. Comparisons between multiple groups were done by 2-way analysis of variance (ANOVA) with Tukey’s post-hoc test or Kruskal–Wallis with Dunn’s correction for normal and non-normal distribution, respectively. A p-value threshold to determine significance was not used as recommended by the guidelines (2022) of the American Statistical Association54. We followed the ARRIVE guidelines 2.0 (https://www.nc3rs.org.uk/arrive-guidelines) for transparent, reproducible, and complete reporting. Raw data is available in Supplementary Data S1.

Results

Chemotherapy causes rapid loss of bone mass associated with markers of cellular senescence

Previous studies have shown that genetic ablation of p16-expressing cells attenuates the loss of bone mass following doxorubicin administration to mice17 implicating cellular senescence as mediator of the bone loss with chemotherapy. To examine whether senolytics could impact the early effects of chemotherapy on bone mass, we administered AC once a week for a total of three administrations in combination with vehicle or D + Q (Fig. 1a). Mice were sacrificed two weeks after initiating the treatments. No significant body weight changes were detected with any of the treatments (Fig. 1b). Micro-CT measurements in vertebrae revealed that AC treatment decreased trabecular bone volume by approximately 20%, using individual group comparisons (Fig. 1c) or 2-way ANOVA (Supplementary Table S1). Trabecular bone parameters such as thickness, number, and separation were not affected by AC treatment (Fig. 1d). However, 2-way ANOVA analysis (Supplementary Table S1) indicated that the overall effect of AC treatment on trabecular bone volume was likely due to a subtle decrease in trabecular number and an increase in trabecular separation. D + Q did not alter the decrease in bone mass caused by AC (Fig. 1c).

The loss of bone mass with chemotherapy has been associated with a decrease in osteoblasts and an increase in osteoclasts15,16,17,18,19. Expression of osteoblast and osteoclast markers mRNA from whole vertebral bone showed that Bglap (osteocalcin) and Alpl (alkaline phosphatase) was reduced by AC, while the osteoclast markers Acp5 (Trap) and Ctsk (cathepsin K) were not impacted (Fig. 1e). D + Q did not alter the decrease in Bglap or Alpl caused by AC. Two-way ANOVA analysis (Supplementary Table S1) indicated that, overall, AC decreased the expression of the osteoblast marker genes, while the osteoclast markers were decreased (Acp5) or unchanged (Ctsk). Previous work shows that loss of bone mass with chemotherapy still occurs in the presence of a bisphosphonate, used to halt bone resorption55, indicating that a decrease in osteoblasts and bone formation are critical mediators of the effects of chemotherapy.

We next examined the mRNA levels of several genes associated with senescence and SASP. These markers were chosen based on compilations of senescence gene sets such as SenMayo56 and previous evidence of their up-regulation in different tissues, including bone, in response to other chemotherapy regimens17,29,32,33,34,35,36. In mRNA from whole lumbar vertebrae, Cdkn1a, Ccl2, and CxCl12 were increased with chemotherapy. D + Q did not attenuate the increase in these markers. The expression of other senescence markers such as Cdkn2a, Il-6, Il-1a, Tnfsf11, and Mmp13 was unchanged, while CD38 was decreased with AC (Fig. 2a). We also examined senescence markers in the cortical portion of long bones. These samples are devoid of the majority of bone marrow and enriched in osteocytes, which are former osteoblasts embedded in the bone matrix. Osteocytes are terminally differentiated, long-lived, and the most abundant cells in bone57. The cortical bones of femur and tibiae also contain other cell types, including osteoblasts, osteoclasts, as well as other mesenchymal, hematopoietic, and endothelial cells58,59. In osteocyte cultures from aged mice, an increase in the expression of Cdkn1a and Cdkn2a was correlated with in situ markers of osteocyte senescence, such as senescence-associated distension of satellites60. AC increased the expression of Cdkn1a, Cdkn2a, Mmp13, and Ccl2. D + Q attenuated the upregulation of Cdkn1a but not the other markers (Fig. 2b). AC had no effect on the other markers of senescence except for a decrease in CD38. D + Q attenuated the inhibition of CD38 by AC.

Fig. 2
figure 2

Senescence markers altered by chemotherapy and D + Q. Quantitative RT-PCR analysis of mRNA from (a) sixth lumbar (L6) vertebrae, and (b) cortical bone from femur and tibia (n = 8–9/group). Bars represent mean ± SD; p-values by 2-way ANOVA with Tukey’s multiple comparisons test. The most relevant comparisons between individual groups are displayed.

Full size image

D + Q had no effect on the loss of bone mass caused by TAC

We next examined whether D + Q could attenuate the loss of bone mass with chemotherapy at later time points. To this end, mice received the full TAC regimen and the administration of D + Q was continued for five additional weeks after termination of the chemotherapy regimen. Mice were sacrificed 14 weeks after the beginning of the treatment (Fig. 3a). No significant changes were seen in femoral cortical thickness in TAC treated mice (Fig. 3b). TAC decreased trabecular bone volume by approximately 28% in vertebrae (Fig. 3c). Together with the findings described in Fig. 1, this indicates that majority of bone loss with this regimen of chemotherapy occurs during the first 2 weeks of treatment. The loss of trabecular bone volume was due to a decrease in trabecular number associated with an increase in trabecular separation (Fig. 3d). Trabecular thickness was not altered by chemotherapy. D + Q had no effect on cortical or trabecular bone mass in mice treated with vehicle or TAC. Analysis of mRNA from whole vertebral bone revealed that the osteoblast and osteoclast markers were not altered by TAC at the end of the experiment. Furthermore, D + Q had no impact on the markers of bone remodeling (Fig. 3e).

Fig. 3
figure 3

D + Q did not alter the loss of bone mass caused by TAC. (a) Schematic of experimental design created with BioRender.com. Quantitative μCT analysis of (b) cortical thickness at the midshaft (diaphysis) or the distal end (metaphysis) of the femur; and (c) trabecular bone volume over tissue volume (BV/TV) and (d) microarchitecture in fifth lumbar vertebrae. (e) Quantitative RT-PCR analysis of mRNA from the sixth lumbar vertebrae. Bars represent mean ± SD; p-values by 2-way ANOVA with Tukey’s multiple comparisons test. The most relevant comparisons between individual groups are displayed.

Full size image

No markers of cellular senescence are seen in bone after long-term chemotherapy treatment

We next examined whether the markers of cellular senescence caused by chemotherapy remained or were altered after long-term chemotherapy. Surprisingly, we found that senescence markers in mRNA from L6 vertebrae were unchanged in TAC treated mice compared to saline treated controls (Fig. 4a). Similar findings were obtained when we examined mRNA from cortical bone shafts (Fig. 4b). It has been proposed that senescent cells in some tissues can indirectly impact other tissues via systemic effects61. Because in the long-term treatments, we could not find senescent markers in bone, we examined whether these could be present in other tissues such as liver and kidney, which have been shown to exhibit senescence cells with chemotherapy29,32,33,34,35,36,37. Levels of Cdkn1a were increased in both tissues from mice treated with TAC, with a higher magnitude in liver (Fig. 5a,b). None of the other senescence-related genes examined were impacted. However, examination of the impact of TAC in the liver immediately after the termination of treatment revealed a sevenfold higher expression of Cdkn1a, while a twofold increase was seen for Cdkn2a. Cd38, Il-6, and Hmgb1 also showed modest increases (Fig. 5c). These, along with the results obtained in bone tissue, suggest that the presence of senescent cells with TAC is transient.

Fig. 4
figure 4

Markers of cellular senescence in bone 7 weeks post TAC. (a) Quantitative RT-PCR of cellular senescence markers in cortical bone and (b) whole sixth lumbar vertebrae from mice treated with saline + vehicle and TAC + vehicle of the experiment described in Fig. 3 (n = 5–6/group). Bars represent mean ± SD; p-values by unpaired two-tailed Student’s t-test; with Welch’s t-test for Tnfsf11 (a) and Mann–Whitney test for Il-1a (a).

Full size image
Fig. 5
figure 5

Markers of cellular senescence in soft tissues following TAC. (a) Quantitative RT-PCR of mRNA isolated from kidney of mice treated with saline + vehicle and TAC + vehicle of the experiment described in Fig. 3, and (b) in livers from mice of all 4 groups. (c) mRNA isolated from livers of mice sacrificed 1 day after the last TAC administration (n = 6/group). Bar graphs represent mean ± SD; p-values by unpaired two-tailed Student’s t-test; with Welch’s correction for Cdkn1a (c) and Cdkn2a (a,c); Mann–Whitney test for Il-1α (a); and two-way ANOVA with Tukey’s test for (b).

Full size image

Piperlongumine had no effect on the loss of bone caused by TAC

To investigate whether senolytics, other than D + Q, could prevent the bone loss caused by TAC, we administered PL to 16-week-old female mice, similar to the experiment described in Fig. 3. Briefly, PL was administered with each chemotherapy cycle and continued for five additional weeks after termination of the chemo regimen. TAC caused loss of body weight and this effect was attenuated by PL (Fig. 6a). Consistent with the studies above, TAC decreased trabecular bone volume; however, no changes in trabecular microarchitecture could be detected, most likely due to the smaller sample size (Fig. 6b). Nonetheless, 2-way ANOVA analysis indicated that the overall effect of AC treatment on trabecular microarchitecture was similar to the one seen in the previous experiments (Supplementary Table S1). PL did not attenuate the damaging effects caused by TAC on trabecular bone.

Fig. 6
figure 6

Piperlongumine did not alter the loss of bone mass caused by TAC. (a) Body weight of mice during treatment. (b) Quantitative μCT analysis of trabecular bone volume over tissue volume (BV/TV) in the fifth lumbar vertebrae, and (c) trabecular microarchitecture. Bars represent mean ± SD; p-values by two-way ANOVA followed by Tukey’s multiple comparisons test. The most relevant comparisons between individual groups are displayed.

Full size image

Chemotherapy during adulthood does not accelerate skeletal aging

Chemotherapy causes both short- and long-term side effects. While it is well-established that chemotherapy causes rapid effects on bone mass in young mice15,16,17,18,19,20, it remains unknown whether these effects persist throughout life or whether chemotherapy early in life can exacerbate skeletal aging. Here, we examined bone mass up to 1 year after cessation of TAC administration when mice were 20-month-old. At this age, female C57BL/6 exhibit extensive age-associated bone loss62. To avoid possible confounding effects caused by aging, we used female BALB/cJ mice which are resistant to age-associated bone loss at least up to 20 months of age63,64. Nineteen-week-old mice were administered saline or TAC, as described above (Fig. 3a). Half of the mice were sacrificed after 35 weeks and the other half after 62 weeks of the initiation of treatment (Fig. 7a). As seen in the previous experiments with C57BL/6J, TAC administration had no effect on cortical bone mass of BALB/cJ mice (Fig. 7b). Both the individual group comparisons and the 2-way ANOVA, indicate that TAC caused loss of trabecular bone volume both at the distal femoral metaphysis (Fig. 7b) and vertebrae (Fig. 7c) due to a decrease in trabecular number (Fig. 7d) (Supplementary Table S1). The effects on bone mass were milder compared to the ones seen in C57BL/6J mice. Nonetheless, the magnitude of bone loss due to TAC was similar between the mice examined at the two time points.

Fig. 7
figure 7

Low bone mass persists up to twelve months after cessation of TAC. (a) Schematic of experimental design created with BioRender.com. Quantitative μCT analysis of femur (b) cortical thickness at midshaft (left), and trabecular bone volume over tissue volume (BV/TV) at the distal end (right); (c) fifth lumbar vertebrae BV/TV, and (d) trabecular microarchitecture. Data represents mean ± SD; p-values using two-way ANOVA followed by Tukey’s multiple comparisons test for (b, left) and (d). Kruskal–Wallis with Dunn’s correction for (b, right) and (c). The most relevant comparisons between individual groups are displayed.

Full size image

We also examined markers of cellular senescence in mRNA collected from the L6 vertebrae, cortical bone of the femur and tibiae, kidney, and liver in mice sacrificed 62 weeks after initiation of treatment. TAC treatment did not alter the expression of the majority of senescence-related genes (Fig. 8a–d), with the exception of an increase in Cd38 in L6 vertebrae and Mmp13 in kidney.

Fig. 8
figure 8

Cellular senescence was not detected up 12 months after cessation of TAC. Quantitative RT-PCR of senescence marker genes in (a) sixth lumbar (L6) vertebrae, (b) femur and tibia cortical bone, (c) kidney, and (d) liver from mice described in Fig. 7a (n = 5/group). Bar graphs represent mean ± SD; p-values by unpaired two-tailed Student’s t-test; with Welch’s correction for Il-6 (d) and Mann–Whitney test for Cdkn2a (b,c), Hmgb1 (c), and Mmp13 (d).

Full size image

Discussion

To investigate the contribution of cellular senescence to the deleterious effects of chemotherapy on bone mass, we conducted studies employing the TAC regimen, widely used to treat breast cancer38, combined with administration of senolytics such as D + Q or PL38,39,40. Both D + Q and PL have been extensively used to eliminate senescent cells in multiple tissues35,37,43,44,48,65. However, neither D + Q nor PL were effective in antagonizing the effects of chemotherapy on bone. Our findings also revealed that the mechanisms that cause early loss of bone with chemotherapy do not persist long-term.

We show here that senescence markers in both vertebra and long bones were transiently up-regulated early during chemotherapy treatment. Anti-cancer therapies, such as chemotherapy, can cause senescence in both cancer and healthy cells66,67,68,69. These drugs cause DNA damage which, in turn, promotes ATM-CHK2 and ATR-CHK1 kinase-mediated activation of the interconnected p53-RB pathways to cause senescence70,71. For example, the topoisomerase inhibitor doxorubicin, one of the drugs included in the present studies and widely utilized for the treatment of a variety of cancers, dysregulates re-ligation of DNA strands after supercoil unwinding72. This leads to massive DNA damage and increased expression of p53 and its downstream target Cdkn1a (p21), subsequently inducing senescence71. Administration of doxorubicin alone to healthy mice increases markers of senescence in multiple tissues such as skin, lung, liver, heart, and bone. To date, no effects of docetaxel or cyclophosphamide in inducing senescence in non-cancer cells have been described in mice.

We found that D + Q attenuated some markers of senescence in bone. However, the two distinct types of senolytics used in our studies could not protect bone from the deleterious effects of TAC. In contrast, administration of D + Q is equally effective in clearing senescent cells and improving bone mass with aging and radiation-induced bone loss41,73. Furthermore, D + Q has been used extensively in pre-clinical models to show that intermittent treatment attenuates age-related pathologies such as neurodegeneration, hepatic steatosis, insulin resistance, pulmonary fibrosis, and skeletal muscle dysfunction, among others38. While much less is known about effects of PL, administration of this senolytic enhances cognitive function in aging mice65, in a murine model of Alzheimer’s disease47, and in mouse models of cognitive impairment induced by chemotherapy. A plausible explanation for the lack of effect of senolytics is that the increase in senescent cells by TAC is modest and does not constitute the primary cause of bone loss.

In all the experiments performed, we could not detect changes in bone mass in the femoral cortex. Similar to our findings, the vast majority of studies showing effects of chemotherapy in rodents report decreases in trabecular but not cortical bone mass15,16,17,18,19,55,74. The reasons for the different responses of the two bone compartments to chemotherapy remain unclear but are unlikely to be related to cellular senescence. Indeed, genetic ablation of p16 expressing cells or administration of D + Q to old mice increases both cortical and trabecular bone, suggesting that the presence of senescent cells in bone is deleterious to both bone compartments41. A possible explanation might rely on the rapid but transient effects of chemotherapy in suppressing bone formation, causing an early loss of trabecular bone microarchitecture. We show here that most of the bone loss occurs within two weeks of treatment. Likewise, two weeks of administration of cisplatin to mice is sufficient to decrease trabecular bone mass55. This is in line with the well-established findings that chemotherapy causes rapid and severe disruption of bone marrow populations, followed by acute myelopoiesis in both human and mice. These effects are associated with a transient increase in pro-inflammatory cytokine and chemokines which occurs during the first two weeks after drug administration and returns to baseline afterwards20,75. Importantly, the effects of TAC do not seem to accelerate bone loss in aged mice, supporting the idea that long-lived cells in bone, such as osteocytes and osteoblast precursor cells are not overtly impacted.

A caveat to our studies is that up-regulation of cell cycle inhibitors and SASP components do not conclusively define cellular senescence30,76. Nonetheless, up-regulation of Cdkn2a has been associated with bone cell senescence caused by aging60,77,78, radiation17,73, and chemotherapy17. Furthermore, the elimination of senescent bone cells using genetic or pharmacological approaches is always associated with a decrease in the expression of Cdkn2a41,79,80. It remains possible that TAC causes a type of cellular senescence that is characterized by markers distinct from the ones we used. This is, however, unlikely due to previous findings that DNA damage caused by doxorubicin alone or by focal radiation up-regulates Cdkn2a; and that ablation of p16-expressing cells or treatment with D + Q, respectively, attenuates the skeletal effects caused by these agents17,41,73. Elimination of p16-expressing cells also attenuates cardiac toxicity and the decline in physical activity caused by doxorubicin81.

Our results support the idea that cellular senescence is not a major contributor to the loss of bone mass with TAC administration. It is important to note that the apoptotic response to chemotherapy is dominant in many cancers and that most chemotherapies are unable to induce senescence in a significant fraction of cancer cells in vivo27,82. Thus, a decrease in osteoblast number due to apoptosis is a likely explanation for the rapid effect of chemotherapies. In support of this premise is the evidence that many TUNEL-positive cells are present on the bone surface as early as 2 days after administration of 5-FU83. Moreover, mice in which apoptosis is inhibited by genetic ablation of Bax and Bak in cells of the osteoblast lineage have increased trabecular bone mass50. Interestingly, cortical bone is unaffected in these mice indicating that osteoblast/osteocyte apoptosis has a distinct contribution to trabecular versus cortical bone mass homeostasis. Thus, a pro-apoptotic effect in osteoblastic cells due to DNA damage could potentially explain both the decrease in bone formation and the predominant effect of chemotherapy on trabecular bone. Future research with mice in which osteoblastic cell apoptosis is inhibited should elucidate the contribution of cell death to the skeletal effects of chemotherapy.