A novel approach for the prevention of ionizing radiation-induced bone loss using a designer multifunctional cerium oxide nanozyme

The disability, mortality and costs due to ionizing radiation (IR)-induced osteoporotic bone fractures are substantial and no effective therapy exists. Ionizing radiation increases cellular oxidative damage, causing an imbalance in bone turnover that is primarily driven via heightened activity of the bone-resorbing osteoclast. We demonstrate that rats exposed to sublethal levels of IR develop fragile, osteoporotic bone. At reactive surface sites, cerium ions have the ability to easily undergo redox cycling: drastically adjusting their electronic configurations and versatile catalytic activities. These properties make cerium oxide nanomaterials fascinating. We show that an engineered artificial nanozyme composed of cerium oxide, and designed to possess a higher fraction of trivalent (Ce3+) surface sites, mitigates the IR-induced loss in bone area, bone architecture, and strength. These investigations also demonstrate that our nanozyme furnishes several mechanistic avenues of protection and selectively targets highly damaging reactive oxygen species, protecting the rats against IR-induced DNA damage, cellular senescence, and elevated osteoclastic activity in vitro and in vivo. Further, we reveal that our nanozyme is a previously unreported key regulator of osteoclast formation derived from macrophages while also directly targeting bone progenitor cells, favoring new bone formation despite its exposure to harmful levels of IR in vitro. These findings open a new approach for the specific prevention of IR-induced bone loss using synthesis-mediated designer multifunctional nanomaterials.


Introduction
Radiotherapy exposes patients to controlled levels of ionizing radiation (IR) and is an effective and essential component of cancer care and management. Due to its high calcium content, bone tissue is estimated to absorb 30-40% more radiation than other tissues, making it a common site for serious ancillary tissue damage in cancer survivors [1]. Ionizing radiation increases cellular oxidative damage, causing an imbalance in bone turnover that is primarily driven via heightened activity of the bone-resorbing osteoclast, resulting in fragile, osteoporotic bone. The risk of suffering an osteoporotic insufficiency fracture is multi-dependent, but can increase by 7.1% [2], 55.7% [3] and 89% [4], due to a variety of factors (e.g., gender, menopausal state, age and comorbidities), exposure parameters (e.g., dose per fraction, total dose) [5], and the cancer itself [6]. Because of IR-induced cell dysfunction, the subsequent bone healing response to fracture is often impaired or absent, resulting in up to 75% of patients unable to self-repair fractures and a further 43% requiring amputation to resolve the complication [7,8]. Although a fundamental and necessary therapeutic tool, the burden of IR-induced damage to healthy bone is a persistent and substantial source of functional impairment, pain and morbidity. The exact pathogenesis of IR-induced bone loss has not been discovered, and despite extensive advances in the research and development for finding nontoxic, safe, and effective prophylactic countermeasures, the FDA has approved only amifostine. Animal models have demonstrated amifostine protects against bone damage [9][10][11][12] however, its major drawback is when administered at the high doses required for radioprotection, the drug is toxic, and patients become hypotensive, with both upper and lower GI disturbances [13,14]. This has resulted in frequent adverse clinical events and decreased efficacy [15,16]. Despite subsequent progress made to improve its effectiveness, none of the strategies have resolved the issue of its toxicity/side effects, hence excluding it from general clinical use [17]. Thus, no effective osteopreventative agent exists and patients are instead advised to modify their daily habits to reduce the likelihood of suffering from these complications. Therefore, new insights into IR-induced dysfunction could be of great clinical and therapeutic importance and motivates the continued search for an osteoprotective treatment against bone tissue loss.
The absorption of IR by living cells can directly disrupt atomic structures through the ionization of deoxyribonucleic acid (DNA) and other cellular targets, i.e., causing loss of electrons (oxidation) and simultaneous addition of excess electrons (reduction), with ensuing lesions within the molecular structure [18][19][20][21]. This can bring about single and double strand breakage, and tandem lesions including crosslinks to DNA, and functional damage to the cell. Radiation also damages cells via the indirect effect of water radiolysis, culminating in the formation of excessive reactive oxygen species (ROS), which are a major concern due to their detrimental cellular effect [22]. The energy deposition to water causes intracellular bursts of ROS, able to microdistribute extracellularly, and far enough to cause a "bystander effect" to cells within the local proximity, damaging critical molecules through protein carbonylation, lipid peroxidation, and enhancing rates of spontaneous gene mutations and neoplastic transformation [20,[23][24][25][26]. The species produced by the radiolysis of water include radical products (eaq , OH • , H • ), and molecular products (H 2 and H 2 O 2 ) as well as protons and hydroxyl ions [27]. In the presence of oxygen, eaq and H-atom (H • ) radicals are rapidly converted to superoxide anion radical/hydroperoxyl radicals (O 2 •-/HO 2 • ), among other organic radicals that are also formed [28]. It is estimated that about two-thirds of IR damage to DNA in mammalian cells is caused by the hydroxyl radical (OH • ), as it reacts amply with organic biomolecules found in living organisms [26,29]. The O 2 •radical is also a major culprit of free radical-mediated toxicity as it is easily generated, and through various pathways or chain reactions, is a primary precursor in the formation of various other harmful ROS [19]. Hydrogen peroxide (H 2 O 2 ) is a neutral species believed to have more toxic potential (often ascribed to its kinetic stability and ability to permeate the cell membrane) than O 2 •and HO 2 • is able to directly generate OH • radicals via the Fenton reaction [30,31]. In addition to attacking DNA, ROS initiate the most devastating effect of oxidative stress, which is membrane lipid peroxidation [32]. Thus, both direct and indirect IR-induced dysfunction are initiated via a series of molecular and biochemical signalling events that occur during or shortly after IR exposure. These events are responsible for most of the harmful effects of radiation that occur thereafter [23,33].
In vitro bone tissue studies demonstrate that human bone marrowderived mesenchymal stem cells (hBMSCs) to exhibit a dosedependent decrease in proliferation and a reduced capacity for osteogenic differentiation following exposure to IR [34][35][36]. Osteoblasts display growth inhibition and reduced bone mineral deposition [37,38]. Macrophages when either directly [39] or indirectly damaged by IR, exert their bystander effect by releasing high concentrations of proinflammatory cytokines that serve to further suppress osteoblastic activity and stimulate osteoclastogenesis and bone resorption [40][41][42]. The combined dysfunction caused by IR to BMSCs, osteoblasts, immune cells and endothelial cells [43], represents the primary contributor for increased bone loss following exposure to IR [44]. Downregulating excess ROS and DNA damage to cells during and immediately following IR exposure, may limit the subsequent dysfunction and tissue damage observed.
The properties that rare earth metals endow, makes them a remarkable strategic resource and thus the focus of this study. The chemistry of rare earth metals differs from other main group metals because of the nature and occupation of the 4f orbitals, which in turn, imparts unique catalytic, magnetic and electronic properties [51]. These unusual properties can be exploited to create new technologies that are not possible with transition and main group metals. Unique nanozymes such as cerium oxide nanoparticles (CeONPs), are a new generation of artificial enzymes that have received much attention because of their exemplary nanozymatic activities, low toxicity and ability to easily and drastically adjust their electronic configurations in response to changes in the bioenvironment [52,53]. These properties are derived from quick and expedient interconversion of the oxidation state between Ce 4+ (fully reduced) and Ce 3+ (fully oxidized). The CeONPs feature oxygen vacancies, or defects, in the lattice structure, which arise through loss of oxygen and/or electrons, when alternating between CeO 2 and CeO 2-x during redox reactions. Through this dual and regenerative role as an oxidation and reduction catalyst, many studies have shown that CeONPs possess multiple antioxidant-enzyme-like activities, including SOD, CAT, and peroxidase-like activities. As such, they are expected to scavenge almost all types of noxious reactive species under suitable conditions, outperforming endogenous antioxidants [29,32,51,54] ( Supplementary Fig. S1).
Here, we describe a novel approach by which rats originally susceptible to IR-induced bone damage, can become resistant to its toxic effects after treatment with a synthesis-mediated designer multifunctional nanozyme composed of cerium oxide. Most notably, the nanozyme repressed IR-induced inflammation and osteoclastogenesis, and in parallel, liberated osteoblastogenesis in vitro, culminating in a bone structure able to maintain its architecture and strength in vivo. The mechanistic novelty revealed in this study is based on the heightened presence of Ce 3+ surface sites, versatile shifts in electronic configuration and the presence of oxygen vacancies (defects) on the nanozyme surface; features that encourage significant overexpression of cytosolic and mitochondrial SOD, while neutralizing increased amounts of ROS (i.e., O 2

•-
, H 2 O 2 and OH • ). Our data also supports the additional mechanistic role of nanozyme-induced hypoxia, which may serve a dual role of reducing ROS formation in conjunction with oxygen extraction,

Synthesis and cellular internalization of two distinct nanozymes
The low formation energies of surface oxygen vacancies are important for oxidation, and the localization of charge as Ce 3+ state provide power for reduction, altering the electronic configuration, catalytic properties and response to ROS [55,56]. Here, we adjusted the relative fractions of Ce 3+ and Ce 4+ surface sites (CeONPs 3+/4+ ) respectively, to form two particle formulations (CeONP 60/40 rich in reduced-state cerium sites, and reduced-state lean CeONP 20/80 ); thereby altering the electronic configuration, catalytic properties and response to ROS. Particles from each formulation were observed to be roughly spherical in morphology, suggesting a truncated octahedral morphology with predominately {111} surface facets terminated by {100} facets (Fig. 1[A, C]). Selected area electron diffraction (SAED) images confirm particle crystallinity for CeONP 60/40 and CeONP 20/80 with ring (halo) patterns denoting nanoscale dimensions of primary crystallites and spacing indexed to the cerium oxide crystal structure Fig. 1 [B, D]). This morphology is characteristic of cerium oxide with a fluorite (Fm3m, ICSD: 55384) crystal structure, and it is confirmed for both formulations by indexing the selected area electron diffraction patterns. The morphology and average particle size (3-5 nm for CeONP 60/40 and 5-7 nm for CeONP 20/80 ); (Fig. 1[A] and 1[C], respectively) are common to their respective syntheses and were chosen to allow for a more direct determination of the effects from unique cerium formulations and surface chemistry on the tested bio-system [57][58][59]. The surface (zeta) potential of the CeONP 60/40 particles was 25.4 ± 0.6 mV with a value of 44.0 ± 7.98 mV measured for the CeONP 20/80 particles.
X-ray photoelectron spectroscopy (XPS) measurements ( Fig. 1[E, F]) were performed to characterize nanomaterial redox state distributions and surface chemistry. High resolution XPS scans over the Ce3d binding energy region allow determination of redox state distribution/fraction between Ce 3+ and Ce 4+ [60]. Specifically, spectra are de-convoluted into cerium spin-orbit coupled doublets (3d 5/2 and 3d 3/2 ), with redox-state-specific multiplet peaks further identified for each spin-orbit. The relative amounts from each state were calculated as integrated peak area ratios [Ce 3+ = A Ce3+ /ΣA i ] where A i is the integrated area of the peak "i". From this calculation, the concentration of Ce 3+ was found to be 61.8% in CeONP 60/40 samples ( Fig. 1[E]) and 20.1% in CeONP 20/80 samples ( Fig. 1[F]).
Studies noting the approximately inverse relationship between nanoceria formulations' SOD and CAT activities are well-represented in the literature (35,41,42). While the detailed mechanisms informing this relationship remain a subject of current interest [63], the observed dependence on measured Ce 3+ is well-supported (positive correlation with SOD activity, negative correlation with CAT activity) [58]. A high SOD activity was observed for the CeONP 60/40 formulation (61.8% Ce 3+ according to XPS measurements/analysis), in corroboration with earlier published studies ( Fig. 1[G]) [58,61,62]. Conversely, CeONP 20/80 particles show less SOD activity ascribed to lower measured Ce 3+ (20.1%). In compliment to respective SOD activities, a higher catalase-memetic activity was noted for CeONP 20/80 samples, whereas negligible catalase-mimetic activity was observed for CeONP 60/40 . As shown in Fig. 1[H], the cellular internalization of FITC-labeled CeONPs into hBMSCs was confirmed using flow cytometry analysis. The mean fluorescence intensity increased significantly after CeONP uptake (p < 0.01), indicating internalization into the cell after a 24h period. Both CeONP 60/40 and CeONP 20/80 were internalized in similar amounts.  (Fig. 1[I]) but that the interaction of any H 2 O 2 species is more favorable in the presence of Ce 3+ (CeO 2-x - Fig. 1[J]). This is particularly noticeable for the products of dissociation of H 2 O 2 compared to molecular adsorption of H 2 O 2 . Ce 3+ is therefore a stronger binding and scavenging site compared to Ce 4+ . The presence of Ce 3+ and associated oxygen vacancy does not allow molecular H 2 O 2 to be stabilized on {100} surfaces; barrier-less dissociation of molecular species such as water has also been reported [55]. The adsorption process that sees ROS adsorbing directly onto surface oxygen vacancies enables a surface "healing" process too, as surface Ce ions increase their oxygen coordination [64,65] to resemble the stoichiometry of a CeO 2 stoichiometric surface. The "healing" of oxygen vacancies is also accompanied by a scavenging process whereby the ROS (i.e., OH, OO and OOH) are trapped into surface defect sites with a strong interaction energy as shown in Fig. 1 These results in Figure K Fig. 2[A]. A reduction in ROS was observed following the supplementation of cells with 10 μg/mL of either CeONP 60/40 or CeONP 20/80 , confirming their antioxidant properties. The intense red fluorescence stain observed also confirms the presence of metabolically active mitochondria. A reduction in intensity is noted in the cells exposed to radiation without CeONP treatment. ROS generation in the non-irradiated cells was also evaluated, and data shows no significant differences among groups ( Supplementary Fig. S3). Flow cytometry analysis further demonstrated that supplementation with either CeONP 60/40 or CeONP 20/80 , significantly reduced intracellular superoxide anion generation (p < 0.001) ( Fig. 2[B]).

Fig. 2[C]
shows that pre-treatment of hBMSCs with CeONP 60/40 at a concentration of 10 μg/mL followed by IR exposure, resulted in significant upregulation of SOD1 (p < 0.001) and SOD2 (p < 0.01) expression at 24 h, when compared with the control (0 μg/mL) group. No significant difference was found when the CeONP 20/80 group was compared with control cells and, remarkably, significantly increased levels of both SOD1 and SOD2 expression was observed in the CeONP 60/40 treated cells when compared with the CeONP 20/80 group (p < 0.0001 and p < 0.01, respectively). To support this, SOD1 Western blot analysis also indicated increased SOD1 protein expression following 10 μg/mL CeONP 60/40 treatment (p < 0.01) when compared with the 0 μg/mL group (Fig. 3 [D]), while catalase activity showed no difference among all tested groups ( Fig. 3[E]). In Fig. 1  Western blot analysis of SOD1. Endogenous α-Tubulin expression was shown as control. **p < 0.01.
[E] Catalase activity was measured using an Amplite® Fluorimetric Catalase Assay Kit.
[J] Mineral deposition by cells was quantified following X-Ray exposure in treated and untreated cells.
[K] HIF-1α expression. The level of HIF-1α expression was measured in primary hBMSCs 24h following exposure to 7 Gy. Protein levels were determined using a HIF1α human ELISA kit. The CeONPs were replenished in the media following irradiation. Experiments were carried out in triplicate. All values are given as the mean ± SD. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

CeONP 60/40 and CeONP 20/80 reduce radiation-induced cellular senescence
Cellular senescence occurs naturally as a result of aging as well as through pro-senescence stressors, such as ionizing radiation, resulting in the release of a proinflammatory secretome [70]. Radiation-induced cell senescence was evaluated using β-gal staining. As shown in Fig. 3[E], β-gal-positive staining intensified following exposure to ionizing radiation, however, qualitative observations showed fewer senescent cells in both the 1 μg/mL and 10 μg/mL CeONP 60/40 and CeONP 20/80 groups after 28-days of culture.

CeONP 60/40 and CeONP 20/80 do not affect hBMSC proliferation and cytoskeletal morphology following IR-induced damage
Higher or lower levels of ROS can induce delays in different phases of the cell cycle, even in the absence of DNA damage [23]. We evaluated the effect of CeONPs on proliferation and cytoskeletal damage and found no differences between groups at 3-and 5-days post-IR ( Supplementary  Fig. S5[A-H]). This result was similar to our previous studies (34,42).

CeONP 60/40 promotes a greater osteogenic response
Next, we explored the ability of hBMSCs to differentiate into osteoblasts. To compare the protective effect of CeONP 60/40 and CeONP 20/80 on irradiation-induced functional damage, hBMSCs were pre-treated with 0, 1 or 10 μg/mL of CeONPs for 24 h before exposure to 7 Gy of irradiation. CeONP treatment within osteogenic differentiation media continued post-IR exposure. By day 7 post-irradiation, and as shown in On day 28 of culture, Alizarin red S staining was used to evaluate levels of mineralization. Following pre-treatment, irradiation and supplementation with either CeONP 60/40 or CeONP 20/80 , resulted in an increase in bone mineral deposition in the 1 μg/mL and 10 μg/mL CeONP 60/40 group, and 10 μg/mL CeONP 20/80+ group when compared with the 0 μg/mL group (Fig. 3[I]). Remarkably, the supplementation of CeONPs 60/40 at a concentration of 10 μg/mL resulted in a significant >4fold (p < 0.0001) increase in new bone deposition, when compared with the limited mineral deposition formed in the non-CeONP-given irradiated cells (Fig. 3[J]). The level of mineralization in each group was also evaluated in non-irradiated and CeONP-treated cells and results showed the significant effect that all CeONP treated groups had on increasing osteogenesis and bone mineral deposition ( Supplementary Fig. S6[A,  B]).

Both CeONP 60/40 and CeONP 20/80 increase intracellular levels of the hypoxic regulator HIF1α
Next, and to elucidate why CeONP 60/40 and not CeONP 20/80 significantly increased bone mineral deposition, we considered whether the increased oxygen extraction activity of Ce 3+ vacancies may induce transient hypoxia [57]. The hypoxic regulator HIF1α protein is a positive regulator of bone formation [71,72]. Following exposure to IR, our findings show that both formulations of CeONPs significantly increased intracellular levels of the HIF1α protein (p < 0.001 in both groups) when compared with control, IR exposed and non-CeONP treated cells. No significant difference was found when the CeONP 60/40 nanozyme was compared with CeONP 20/80 (Fig. 3[K]).

CeONP 60/40 and CeONP 20/80 do not affect RAW 264.7 proliferation and cytoskeletal morphology following IR-induced damage
Following IR-injury, macrophages are recruited to the irradiated site [73] and when under oxidative stress, the phenotype of the macrophage is key to whether they secrete anti-inflammatory cytokines that promote osteoblastogenesis, or pro-inflammatory cytokines that through RANKL, promote osteoclastogenesis and osteoclastic activity, leading to bone resorption [40][41][42]. Our previous study demonstrated a reduction in ROS levels within RAW macrophages following CeONP treatment [59].
To investigate the effect of both CeONP 60/40 and CeONP 20/80 on RAW 264.7 macrophages, cell proliferation and changes to their actin filaments was investigated on days 1 and 3 following irradiation. However, no significant differences were found between IR only and nanozyme (CeONP 60/40 and CeONP 20/80 )-treated groups ( Supplementary Fig. S7 [A-H]).

CeONP 60/40 and CeONP 20/80 repress multinucleated giant cell and osteoclast marker expression following IR-induced cell damage to the macrophage
Radiation-induced multinucleated giant cell (MNGC) formation is a hallmark symbol indicating chronic inflammation; a condition that promotes osteoclastogenesis and increased osteoclastic activity [74]. Our previous study [59] showed CeONP 60/40 was able to regulate the pro-and anti-inflammatory response and MNGC formation. Here, we assessed the influence of heightened Ce 3+ activity on the macrophage response, and found that both nanozymes were able to suppress the formation of MNGCs (Fig. 4[A-D] and Supplementary Fig. S8). On day 3 and in the irradiated cell group that received no CeONP treatment, the formation of multinucleated giant-like cells was pronounced. In contrast, fewer of these giant cells were evident in the groups that were supplemented with either 10 μg/mL CeONP 60/40 or 10 μg/mL CeONP 20/80 . Further quantification of changes in cell area between groups demonstrated significantly larger cells in the 0 μg/mL IR-exposed group when compared with the 10 μg/mL CeONP 60/40 and 10 μg/mL CeONP 20/80 groups (p = 0.01 in both groups).
To investigate this further, tartrate resistant acid phosphatase (TRAP) staining was used to confirm the formation of active osteoclastlike cells. Macrophages were cultured in the absence of osteoclastic differentiation factors, and as shown in Fig. 4[G-H], irradiation directly induced the formation of "radiation-associated macrophages", which are characterized by the formation of multinucleated TRAP-positive-like cells with high expression of pro-inflammatory and osteoclast-related markers. On day 3, and following supplementation of cells with either treatment of IR-induced toxicities [41,75].

CeONP 60/40 appears to be a non-toxic nanozyme
Our in vitro work showed that the CeONP 60/40 formulation increased binding of the highly damaging OH • radical, bolstered protective cellular SOD activity and augmented osteoblastogenesis more effectively than CeONP 20/80 . For this reason, we decided to investigate the effect of CeONP 60/40 in animals and following exposure to 3 fractions of 8 Gy irradiation (total exposure of 24 Gy). A flow chart of the in vivo study is shown in Fig. 5[A]. A consideration was whether CeONPs when given at a dose of 4 mg/kg via a single tail vein injection twice during the study period, elicits a toxicologic response. Our results showed no histological indications of acute toxicity in the liver, spleen, or kidney 14days post-treatment. Importantly, a protective response by CeONP 60/40 was observed in these organs following IR exposure ( Supplementary  Fig. S9).

CeONP 60/40 radioprotects cells within the bone marrow niche by reducing DNA damage
Notably, the level of DNA damage to cells within the bone marrow was markedly reduced in the nanozyme-treated rats compared to healthy age-and sex-matched rats (9 weeks old) and following exposure to damaging levels of IR (Fig. 5[B]). IR exposure increased DNA damage to cells and that the mean comet length (p < 0.01) and tail length (p < 0.0001) in irradiated and CeONP-given animals were significantly lower than the IR only group of animals. These results indicate that injection of the CeONPs markedly reduces IR-induced DNA damage to cells within the bone marrow.

CeONP 60/40 reduces expression of the osteoclastogenesis factor, RANKL, while also reducing osteoclast number and activity
To investigate changes in the number of active osteoclasts between groups and the role of RANKL, we used immunohistochemical analysis to determine RANKL expression and TRAP staining to measure active osteoclasts from transverse histology sections prepared through the femoral condyle. As shown in Fig. 5[C], the number of RANKL + cells increased in irradiated rats as compared with the non-IR group of rats. In contrast, the expression of RANKL was markedly reduced following 4 mg/kg of CeONP treatment. These results suggest a modulatory role of CeONPs on the osteoclast formation factor, RANKL. Further, and as shown in Fig. 5[D], exposure to irradiation significantly increased the number of TRAP + cells, while no or few TRAP + staining was identified in the non-irradiated group. Remarkably, the number of TRAP + osteoclasts were significantly lower in the rats injected with CeONPs both on days 7 (p = 0.001) and 14 (p = 0.01) (Fig. 5[D]). The serum level of bone resorption marker, CTX-1, was also significantly reduced following 4 mg/kg of CeONP treatment (( Fig. 5[E]). Our findings demonstrate that CeONPs markedly suppress bone resorption by osteoclasts following exposure to harmful levels of IR.

CeONP 60/40 reduces irradiation-induced cell senescence
To further clarify whether CeONPs protect bone cells against IRinduced cellular senescence, we analyzed senescence-associated β-gal (SA-β-gal) activity. Immunostaining results indicated that in vivo expression of β-gal substantially increased after irradiation and by day 7, minimal β-gal activity was observed in the non-irradiated group of animals, with a more pronounced response in those animals who had been exposed to IR-only. In contrast, animals exposed to IR with no CeONP treatment, showed evidence of β-gal activity. However, β-gal activity   Supplementary Table S1. All values are given as the mean ± SD. **p < 0.01, ***p < 0.001. [J] Representative 3D reconstruction images of the proximal tibia via nano-Ct scanning.
was reduced when compared with the IR-only animals, both on days 7 and 14 (Fig. 5[F]).

CeONP 60/40 reduces bone and trabecular volume loss and increases the mechanical properties of ultimate stress σ u and fracture strength σ f in cortical bone
Next, we evaluated whether the nanozyme-treated rats were able to maintain their trabecular and cortical bone structure and strength. A transverse section through the center of the femoral condyle was prepared and bone structure was assessed histologically and following H&E staining in samples obtained 7-and 14-days post-irradiation. Rats who underwent irradiation and received no CeONP treatment developed osteoporosis, as indicated by thinner and shorter trabeculae, reduced trabecular connectivity with increased porosity and bone marrow adiposity by day 7, which had further exacerbated by day 14. In contrast, animals receiving both radiation and CeONP therapy, maintained a healthy bone structure and morphology (Fig. 5[G]). Quantification of BA/TA% (Fig. 5[H]) demonstrated a significant reduction in bone area in the non-CeONP treated IR-exposed animals at 7-and 14days when compared to IR-exposed CeONP-treated animals (p < 0.05 and p < 0.01 respectively).
Stress-displacement deformation behavior when under 3-point bending test conditions was used to examine the mechanical parameters of ultimate stress σ u and fracture stress σ f in the tibial mid-shaft of bone in each experimental group. Here, we show the drastic impact that IR has on reducing bone strength, making it more susceptible to insufficiency fracture. Remarkably, and on day-7, each of these bone strength parameters (fracture stress p < 0.01, and ultimate stress p < 0.01) significantly increased in animals following exposure to irradiation and when given CeONP treatment, compared with the IR-exposed and non-CeONP treated animals. It is important to note that no significant differences were found when animals in the healthy non-irradiated group were compared with those in the IR-exposed, CeONP-treated group (Fig. 5[I]). Load displacement curves, mean fracture and ultimate loads measured and the biomechanical parameters used are presented in Supplementary Fig. S10&S11 and Supplementary Table S1. Additionally, nano-CT analysis demonstrated bone volume loss, while 4 mg/kg of CeONP-treated rats showed reduced bone volume loss when compared to X-Ray group (Fig. 5[J).

Discussion
The use of CeONPs against radiation-induced damage has been reported in our cell-based osteogenesis study [69] and in other non-bone related studies both in vitro [76] and in vivo [77][78][79][80]. However, the designer tuning of the Ce 3+ /Ce 4+ ratio towards a specific and protective function, especially in the preservation of osteogenesis following an IR insult in vivo, is unknown. Here we reveal the critical contribution of increased Ce 3+ surface sites in protecting bone against IR-induced damage in vitro and in vivo and present new insights into the use of a designer artificial nanozyme for the specific prevention of radiation-induced bone loss.
The SOD-and CAT-mimetic activities of CeONPs have not been described here, however, Self et al. [61], and Celardo et al. [52], provide excellent reviews. Our findings support several other studies [52,61,81] and confirm CeONPs designed with a higher fraction of Ce 3+ surface sites provide enhanced SOD-mimetic activity, while the Ce 4+ sites dominate catalase-mimetic activity [51,82]. Further, and using EPR, we show that Ce 3+ surface sites offer specificity for O 2 •to a greater degree than Ce 4+ sites. Our findings confirm the role of oxygen vacancies in directly binding H 2 O 2 and its dissociative forms confirming previous studies [81,83]. However, here we highlight the important and superior participation of Ce 3+ sites in the adsorption including the increased neutralization of O 2

•-
, H 2 O 2 and OH • . Our ab initio data also supports a stronger interaction of Ce 3+ with ROS compared to Ce 4+ .
Our in vitro results confirmed that both nanozymes successfully reduced the IR-induced accumulation of ROS within hBMSCs after 24h, and significantly reduced DNA damage and cellular senescence. This indicates the Ce 3+ /Ce 4+ ratio is mechanistically less relevant here, and through successfully preventing and scavenging ROS, it is conceivable the nanozymes subsequently prevent undesirable perturbations in mitochondrial homeostasis that lead to senescence 28 days later [84]. Interestingly, the DNA tail length in the CeONP 20/80 group was further reduced when compared with cells supplemented with CeONP 60/40 , suggesting it may confer a superior level of protection. The mechanisms for this remain unclear and as such, warrants future investigation. Importantly, IR-induced DNA damage within the bone marrow niche and cell senescence were both substantially reduced in animals who received CeONP 60/40 treatment.
Similarly, our findings show the Ce 3+ /Ce 4+ ratio to be mechanistically less relevant when regulating macrophage activity and upregulating HIF1α expression in hBMSCs following IR exposure. Macrophages play an important role in bone homeostasis and their early infiltration contributes a critical role in IR-induced disease [85][86][87]. Elevated expression of osteoclastogenic-and inflammation-related factors have been reported in in vitro [39] and in vivo [88][89][90] following IR. Notably, CeONPs are able to downregulate inflammatory genes including, IL-1β, IL-6 and TNFα via decreased ROS and downregulation of NF-κß [91][92][93]. Our in vitro findings show that IR exposure directly augmented osteoclast-like differentiation in macrophages, as indicated by an increased number of RANKL + and TRAP + cells. Importantly, pre-treatment with the nanozymes significantly inhibited inflammation and osteoclastogenesis-related marker expression while also reducing MNGC formation. The presence of MNGCs has also been suggested to cause chronic inflammation leading to failed osteogenesis and increased osteoclastic differentiation and activity [74]. Increased HIF1α is reported to regulate inflammation and polarize macrophages towards the pro-healing M2 phenotype via NF-κß [94][95][96][97][98]. Further, HIF1α is a positive regulator of bone formation, increasing osteoblastogenesis, enhancing bone defect healing in vivo [99][100][101], while selectively downregulating osteoclastogenesis [102][103][104], thus favouring bone deposition. Our findings reveal both nanozymes significantly upregulated HIF1α 24 h following exposure to IR. Das and colleagues [57] demonstrated that CeONPs activate HIF1α through modulation of intracellular O 2 levels; first extracting O 2 and then liberating it in the catalytic cycle: CeO 2-x + 0.5xO 2 ⇆ CeO 2 . Thus, increased levels of HIF1α may offer a role in reducing inflammation while simultaneously promoting osteogenesis and repressing osteoclastogenesis. This was validated in part, by a reduction in RANKL + and TRAP + cells in our in vivo IR model. The direct effect of CeONPs on osteoclastic behavior is emerging. Through scavenging or generating ROS and via NF-κß, a study by Yuan et al. [105] reported CeONPs facilitated osteoclast formation at lower concentrations, but inhibited osteoclastogenesis at higher concentrations in vitro. In contrast, CeONPs have also been reported to elicit no effect on osteoclast activity in vivo [106]. While our findings support a significant reduction in both TRAP + active osteoclasts on the bone surface, and CTX-1 protein levels with serum following IR, and nanozyme treatment in vivo, this study did not investigate the direct response to osteoclasts. Future studies will seek to further clarify its role in regulating osteoclast function and activity.
Further in vitro analysis revealed a disparity in the efficacy of the two nanozymes. An emerging protective mechanism observed in our study, was that CeONP 60/40 exclusively and significantly increased expression of both cytosolic SOD1 (Cu/Zn-SOD) and mitochondrial SOD2 (Mn-SOD) in hBMSCs 24h following IR exposure. SOD is indicated in preventing apoptosis and precancerous cell changes, maintaining mitochondrial integrity, protecting enzymes, membranes, microsomes, DNA [49,[107][108][109][110][111] and in reducing IL-1β and TNFα production [49]. Notably, protecting lung tissue [112,113], the oral cavity [114,115] and the esophagus [116,117] from IR-induced damage. Both SOD1 and SOD2 have been suggested as therapeutic targets for bone disorders.
Sod1-deficient mice exhibit reduced enzymatic collagen cross-linking, low bone turnover and develop significant bone fragility [118]. Increased levels of SOD2 is vital for bone metabolism and in suppressing ROS and thus osteoclastic differentiation [119][120][121]. SOD upregulation may inhibit RANKL-induced osteoclastogenesis, while simultaneously enhancing osteogenesis [76,122]. While our findings serve to further support its protective role, the mechanism of heightened Ce 3+ -SOD related activity remains undiscovered. SOD is a metalloenzyme and requires a redox active transition metal in the active site for activity against O 2 •-. SOD proteins have been shown to bypass the dismutation cycle and use external redox equivalents (i.e., alternative metal species) [123]. However, it is important to mention that metal ion promiscuity is highly irregular. Thus, we cautiously speculate that CeONPs may provide an available, alternative, and facile redox equivalent and that the Ce 3+ sites offers greater selectivity and binding affinity to both SOD1 and SOD2 or confers higher stability or a more beneficial charge transfer efficiency, potentially affording greater reaction kinetics [124].
Importantly, we also show more CeONP 60/40 -induced mineralized nodule formation compared to CeONP 20/80 in vitro. To support this, our data confirmed a significant increase in ALP, COL1, and OCN gene expression, and significantly increased ALP protein expression in response to supplementation with CeONP 60/40 compared with CeONP 20/ 80 . The role of CeONPs in upregulating osteogenic protein expression during differentiation has been previously reported [69,125]. Pathway analysis by Luo and colleagues [126], showed that CeONPs enhanced the nuclear translocation of β-catenin and activated the canonical Wnt pathway by promoting sequence similarity 53 member B/simplet expression. This supports our findings as the Wnt/β-catenin signaling pathway regulates osteogenic differentiation, and bone formation, and Wnt pathway activation contributes to increased expression of downstream osteoblast-related genes including, OCN, ALP and COL1 [127]. Further to this, our findings reveal the importance of increased Ce 3+ surface sites in modulating cell fate and increasing bone deposition in vitro. Interestingly, our results are in contrast to Li et al. [128], where a CeONP coating was applied to a titanium surface using magnetron sputtering. The authors report the increased osteogenic differentiation of rat hMSCs in vitro and bone formation in vivo when a higher Ce 4+ , and not Ce 3+ fraction, was investigated. It was speculated this balance in Ce valence may have delivered a system more effective in scavenging ROS, thereby promoting osteogenic differentiation and bone formation. However, here it is important to consider that IR exposure induced a significant increase in highly damaging O 2

•-
, H 2 O 2 and OH • , and our data suggests that when under these physiological conditions, an increased fraction of Ce 3+ surface sites may be critical in selectively scavenging these ROS, potentially restoring redox balance, and healthy cell function. Nevertheless, and taken together, these findings indicate that manipulation of the Ce 3+ /Ce 4+ ratio can modulate bone formation and that further research will be needed to identify the exact mechanism/s involved.
Our in vivo work confirmed that CeONP 60/40 given i.v to rats significantly reduced IR-impaired architectural decline and bone tissue loss and that non-CeONP treated rats developed fragile, osteoporotic bone. Remarkably, animals who received the nanozyme prior and during IRexposure, demonstrated a significantly higher mean fracture stress and ultimate stress, with values similar to the non-irradiated, control animals. Although bone area was maintained in CeONP-given, and IRexposed animals, differences in bone formation rates in vivo were not investigated. Therefore, any nanozyme-induced increase in new bone formation was not assessed in this study. However, our findings indicate that the nanozyme is able to deliver a radioprotective effect and the heightened IR-induced susceptibility to insufficiency fracture was mitigated in this model.
In conclusion, we have demonstrated that rats treated with a nanozyme results in bone tissue able to maintain its architecture, volume and strength in vivo, despite exposure to harmful levels of IR. Analyses revealed the nanozymes were multifaceted and able to prevent intracellular ROS accumulation, DNA damage, senescence, reduce proinflammatory and pro-osteoclastogenic markers while increasing proosteogenic HIF1α protein release in vitro. However, the nanozyme possessing increased Ce 3+ sites provided important and superior protection through the presence of increased oxygen vacancies leading to stronger interaction with ROS and increased neutralization of O 2

•-
, H 2 O 2 and OH • . We also reveal Ce 3+ exclusively targeted pro-osteogenic cytosolic and mitochondrial SOD overexpression. Overall, CeONPs, may represent a novel multifunctional therapeutic strategy for mitigating IR-induced damage (Fig. 6), and nanomaterials with a further increased trivalent fraction, may also have a therapeutic schema for IR-induced osteoporosis.

Synthesis of a nanozyme designed for greater relative SOD activity (CeONP 60/40 )
Synthesis was performed based on a previously published protocol [129]. Cerium (III) nitrate hexahydrate (99.999% purity; Sigma Aldrich) was added to 50 mL of deionized water and allowed to dissolve completely. The Ce 3+ ions were converted to a highly hydrated, cerium (IV) oxide through addition of 3% hydrogen peroxide to a pH below 3.5 and a final cerium concentration of 5 mM. From here, the solutions were left standing away from light and aged for up to 8 weeks. Aging was performed to allow degradation of excess hydrogen peroxide, by catalytic surface reactions, and for equilibration of particle phase composition/surface character (aging effects).

Synthesis of a nanozyme designed for greater relative CAT activity (CeONP 20/80 )
A forced hydrolysis technique was used to generate nanoparticles tuned to possess a lower fraction of Ce 3+ surface sites, relative to Ce 4+ . Synthesis was performed based on a previously published protocol [57]. Specifically, 1.24 g of cerium (III) nitrate hexahydrate (99.999% purity; Sigma Aldrich) was stirred in 50 mL of water for 1 h followed by titration with 30% ammonium hydroxide (ACS grade, Alfa Aesar) to force precipitation of nanocrystalline cerium (hydro-)oxide over 4 h of stirring. The solution was then centrifuged at 10,000 rpm to collect the sedimented particles and the sediment was washed 3x with de-ionized water (to purify/isolate nanomaterial products and promote oxidation to cerium oxide). Particles were then re-suspended in fresh de-ionized water and ultra-sonicated for 20 min to disperse stable particles. Solutions were left standing overnight and any further sediment was removed by manually collecting supernatant of well-suspended particles. Particles were used without further modification.

Characterization of CeONP 60/40 and CeONP 20/80 and analysis of SOD and CAT mimetic activity
Physicochemical characterization of CeONP 60/40 and CeONP 20/80 samples was performed using high resolution transmission electron microscopy (HRTEM; Philips Tecnai operating at 300 kV). X-ray photoelectron spectroscopy (XPS) was performed using a Thermo Scientific ESCALAB 250Xi spectrometer in an ultra-high vacuum (UHV) chamber (4 × 10 − 9 Torr). The radiation source was from a monochromatic Al-Kα cathode (binding energy: 1486 eV). The beam spot size was 650 μm and C1s peak at 284.6 eV was used as a base for binding energy calibration within an experimental error of ±0.2 eV. The quantitative determination of SOD-mimetic activity was performed using a SOD assay kit (Dojindo Molecular Technologies, kit #S311: SOD Assay Kit -WST). A xanthine/xanthine oxidase reaction system was used to estimate the superoxide anion scavenging activity. Similarly, the catalase-mimetic activity was determined using an Amplex red-based hydrogen peroxide assay kit (Invitrogen, Cat. No. # A22188, Carlsbad, CA). Each assay was performed per manufacturer instructions. All catalytic activity measurements were obtained using 2 mM CeO 2 concentrations. Particle colloidal stability and related surface charging were determined via zeta potential measurements using a Zetasizer Nano (Malvern Instruments). Amine-reactive fluorescein 5(6)-isothiocyanate (FITC) labeling techniques and flow cytometry were used to confirm the cellular uptake of CeONPs at 24 h after CeONP treatment as previously described [69].

Characterization of the speciation of CeO 2 and CeO 2-x using density functional theory calculations
Calculations were performed using the VASP code [130,131], using a plane-wave basis set with cutoff 500eV and the projector augmented wave (PAW) approach [132,133]. The frozen core is [He] [135], where the coulombic (U) and exchange (J) parameters are treated as independent variables. We chose U = 5 eV and J = 0 eV, making the Liechtenstein and the Dudarev [136] methods equivalent. This methodology accounts for the presence of the localized Ce 3+ states, and has been commonly used in the wider literature [55,64,65,137,138]. All calculations were spin polarized and 3D boundary conditions were used. The minimised bulk unit cell of CeO 2 retain the space group 225 symmetry and has a lattice constant of 5.498 Å, which overestimates the experimental value of 5.411 Å [139] but is in line with previous literature [55,64,65,137,138]. The electronic and ionic convergence criteria were 1 × 10 − 5 eV and 1 × 10 − 3 eV Å − 1 . The Brillouin zone was sampled using a Γ-centred 5 × 5 x 5 k-point grid. Surfaces were generated using the METADISE code [140] and modelled using the slab method [141] in which two identical surfaces are created via the introduction of a vacuum gap perpendicular to the surface. A vacuum gap of 15 Å was used to minimize the interaction between images. The {100}, {110} and {111} slabs with a (2 × 2) expansion of the primitive surface unit cell were investigated. The {100} and {110} were 7 layers (28 CeO 2 units) and the {111} were composed of 5 layers (20 CeO 2 units). To remove the dipole of the {100} slab, half of the surface oxygen atoms were moved from one to the other side to the slab. The Brillouin zone sampled using a Γ-centred 2 × 2 × 1 grid, with the third vector perpendicular to the surface plane. Relaxation of the atomic structure for all atoms was deemed to have converged when the forces were below 1 × 10 − 2 eV Å − 1 . All slab calculations used symmetric introduction of oxygen vacancies and adsorbates on both sides of the slab, thus ensuring that the surfaces were identical, and the cell had no net dipole moment. As the number of configurations for adsorbed species on surfaces of ceria is extremely large, we have limited the choice to structures that maximize the hydrogen bonding between the adsorbate and the surface. The isolated H 2 O 2 was simulated at the Γ point in a cubic cell of side 10 Å and with the same convergence criteria as the surface calculations. The energy of adsorption (E Ads ) of an adsorbate on ceria surfaces is calculated as , where E Slab+Adsorbate , E Slab and E Adsorbate are the energies of the adsorbate adsorbed onto the slab, of the bare slab, and of the adsorbate, respectively, and 2 accounts for the adsorption on adsorbates on both sides of the slab.

Characterization of the isotropic and anisotropic activity and selectivity of CeONP 60/40 and CeONP 20/80 to O 2
•-Isopropanol (IPA) and dimethyl sulfoxide (DMSO) were purchased from VWR. HPLC-grade water and potassium superoxide powder were purchased from Sigma-Aldrich. This spectrum was generated from dissolving KO 2 in a 1:2:2 water/DMSO/IPA solution which forms a glass when cooled to 77 K. Each sample was initially at 40 mM KO 2 but were allowed to sit for 45 s to allow the O 2 ∸ to react with the CeONPs and/or solvent before being rapidly frozen in liquid N 2 , quenching any further reactions.
Following our previous work on superoxide [66] and cerium nanoparticles [142], solutions of H 2 O, DMSO and IPA in a 1:2:2 ratio, respectively, were prepared using aqueous CNPs with surface Ce 3+ /Ce 4+ ratios both high (WB-CNPs) and low (TH-CNPs) and HPLC-grade H 2 O. The initial pH of the solutions was lowered to 3 to slow CNP precipitation (pH values rapidly rose upon KO 2 dissolution to a final value of 9, leading to precipitation of the CNPs after approximately 1 min). A control was prepared similarly using only HPLC-grade H 2 O. KO 2 powder stored in anhydrous conditions under N 2 gas was then dissolved and the solution was allowed to sit for 45 s before being frozen in liquid nitrogen. A Bruker EMXPplus-9.5/2.7/P/L X-band continuous wave EPR spectrometer was used for all EPR measurements and all O 2 ∸ samples were run under identical conditions (100 kHz field modulation, 320 G scan range centred at 3280 G, 3.2 G modulation amplitude and microwave power of 45 dB at 77K).

7.
In vitro studies using hBMSCs and murine RAW 264.7 macrophages following exposure to ionizing radiation

Culture and irradiation of cells
Human bone marrow-derived mesenchymal stromal cells (ATCC® PCS-500-012™) were maintained in Dulbecco's Modified Eagle's Medium (DMEM; Thermo Fisher Scientific, USA) containing 10% fetal bovine serum (FBS; Thermo Fisher Scientific, USA), and 1% penicillin/ streptomycin (Thermo Fisher Scientific, USA) in a humidified incubator with 5% CO 2 at 37 • C. Murine-derived macrophages (RAW 264.7 cells, ATCC® TIB-71™) were maintained in DMEM supplemented with 10% heat inactivated FBS, and 1% (v/v) penicillin/streptomycin in a humidified incubator containing 5% CO 2 at 37 • C. To investigate the protective effect of CeONP 60/40 and CeONP 20/80 on IR-induced cellular damage, cells were pre-treated at a dose of either 1 μg/mL or 10 μg/mL of CeONPs 24 h prior to IR exposure. The dosage of CeONP 60/40 or CeONP 20/80 were selected based on our previous study [59]. After removal of the CeONPs through washing with phosphate buffered saline (PBS) thrice, cells were subjected to irradiation. Cells were exposed to an IR (X-ray) dose of 7 Gy, a 160 kV tube voltage, 4 mA tube current, at a distance of 30 cm between the source and the surface (SC 500 smart controller, KIMTRON, USA). After X-ray or mock-X-ray exposure, the media containing either the hBMSCs or RAW264.7 cells, were replenished with 1 μg/mL or 10 μg/mL of CeONP 60/40 or CeONP 20/80 . Cells in the control group received no CeONP treatment.

Quantitative real-time reverse-transcription polymerase chain reaction (qRT-PCR) in hBMSCs and murine RAW 264.7 macrophages
Here we used qRT-PCR to investigate the gene expression of endogenous antioxidant factors in hBMSCs, and the pro-inflammatory and pro-osteoclastic gene expression in murine macrophages following IR-induced damage. We determined the protective effect of CeONP 60/40 or CeONP 20/80 to these cells. Total RNA was extracted using a PureLink™ RNA Mini Kit (12183018A, Thermo Fisher Scientific, USA) and genomic DNA contamination removed from samples using DNase (PureLink™ DNase Set, 12185010, Thermo Fisher Scientific, USA). RNA concentration was measured using a NanoDrop 8000 spectrophotometer (NanoDrop technologies). 500 ng of total RNA was used as a template for reverse transcription and cDNA synthesis was performed using Super-Script™ III First-Strand Synthesis SuperMix (18080400, Thermo Fisher Scientific, USA). qRT-PCR was performed using Fast SYBR™ Green Master Mix (4385612, Thermo Fisher Scientific, USA) on an ABI Prism 7500 Thermal Cycler (Applied Biosystems, Foster City, California, USA). The primers used in this study were KiCqStart™ Primers and were purchased from Millipore Sigma. The fold change of relative mRNA expression was calculated using the comparative Ct (2 − ΔΔCT ) method. We used these predesigned qRT-PCR assays for the analysis of Catalase (Gene ID:

Analyses of CeONP 60/40 and CeONP 20/80 protection against IRinduced DNA damage in hBMSCs
To investigate cellular DNA damage following exposure to radiation and the effect of the nanozyme formulations, a comet assay® (4250-050-K, R&D Systems, USA) was performed as previously described [69], and on day 3 after irradiation. In brief, the cells were detached gently using a cell scraper, pelleted and then resuspended in ice cold PBS, free of Ca ++ and Mg ++ ions. A 50 μL cell suspension was mixed with 500 μL molten LM Agarose at 37 • C. 50 μl of the mixture was then pipetted immediately onto a CometSlide™. After a 10 min incubation period in the dark and at 4 • C, the slides were immersed in a lysis buffer overnight at 4 • C. Cells were then immersed for 20 min in a freshly prepared alkaline unwinding solution in the dark and at room temperature before performing gel electrophoresis at 21 V for 30 min. The slides were then washed twice in deionized H 2 O, followed by rinsing once in 70% ethanol. Samples were stained using 100 μL of diluted SYBR™ Green I Nucleic Acid Gel Stain for 10 min in the dark. After rinsing briefly in water, slides were dried at 37 • C and images were taken using a confocal laser scanning microscopy (Zeiss, USA).

Analyses of CeONP 60/40 and CeONP 20/80 on osteoclastic formation using murine RAW 264.7 macrophages
Tartrate resistant alkaline phosphatase (TRAP) staining was used to determine the formation of osteoclasts from macrophages. Following radiation-induced damage, RAW 264.7 macrophages were cultured in DMEM containing 10% FBS and supplemented with or without CeONPs in a humidified incubator containing 5% CO 2 at 37 • C for 3 days. Cells were fixed with 4% paraformaldehyde for 20 min. Images were captured using an inverted phase microscope (BZ-X800E, Keyence, USA).

Analyses of CeONP 60/40 and CeONP 20/80 on the osteogenic differentiation of hBMSCs and mineralization following IR-induced damage
After CeONP 60/40 or CeONP 20/80 pre-treatment and following exposure to 7Gy of radiation, the response of the hBMSCs to osteogenic induction was performed using DMEM containing 10% FBS supplemented with the osteogenic components; 2 mM β-glycerophosphate, 100 μM Lascorbic acid 2-phosphate, 10 nM dexamethasone (Sigma-Aldrich, USA). The CeONPs were introduced at a dose of either 1 μg/mL or 10 μg/ mL and the media was changed every 3 days. To determine the cell response during osteogenic differentiation, alkaline phosphatase (ALP) activity was measured using ALP staining and Western blot, gene expression of osteogenic-related markers was evaluated by qRT-PCR, and mineral deposition was quantified using alizarin S red staining. ALP staining was performed at 14 days and according to the manufacturer's instructions (K2035-50, BioVision, USA). In brief, after removing the culture medium, cells were washed in 300 μL washing buffer, and stained with ALP staining solution for 30 min at 37 • C. After rinsing twice with washing buffer, images were captured using an inverted phase microscope (BZ-X800E, Keyence, USA). Gene expression of osteogenic-related markers was evaluated by qRT-PCR based on our previous study [69]. ALP (NBP2-67295, Novus Biological, USA) and α-Tubulin (NB100-690SS, Novus Biologicals, USA) were used for quantifying ALP expression. ALP Western blot was performed based on our previous study [69]. Alizarin red S staining (A5533-25G, Millipore Sigma, USA) was performed to quantify the deposition of mineral noduli after 28 days of culture. In brief, cells were fixed using 4% paraformaldehyde for 20 min at room temperature. After fixation, cells were washed with deionized H 2 O and incubated with 2% pH 4.1 alizarin red S solution for 20 min at room temperature. The samples were air-dried at room temperature. Images were captured using an inverted phase microscope (BZ-X800E, Keyence, USA). To quantify mineralization, the noduli were dissolved using 10% cetylpyridinium chloride and analyzed by determining Optical Density (OD) values at 562 nm.

Analysis of CeONP 60/40 and CeONP 20/80 on HIF1α expression in hBMSCs
To evaluate the levels of and HIF-1α expression, a HIF1α human ELISA kit (EHIF1A, Thermo Fisher Scientific, USA) was used according to the manufacturer's instructions. In brief, hBMSCs were pre-treated with 10 μg/mL of CeONPs for 24 h. After the removal of CeONPs through washing with PBS thrice, cells were subjected to irradiation.
After X-ray exposure, hBMSCs were replenished with 10 μg/mL of CeONPs in osteogenic induction medium and cultured in a hypoxia incubator (5% oxygen) for 2 h. The absorbance was read at 450 nm using a microplate absorbance spectrophotometer (Synergy HTX, USA). The concentrations of HIF1α were calculated using the standard calibration curve.

8.
In vivo studies and analysis of CeONP 60/40 in 9-week-old SAS Sprague-Dawley rats and following IR-induced tissue damage

Power calculation and randomization of rat allocation to experimental group
The predesigned primary endpoint in the rat studies was to record the effect of CeONP 60/40 in protecting against trabecular and cortical bone tissue loss following exposure to radiation-induced damage. Our power analysis suggested that when using six rats in each experimental group, we would have 80% power to detect a biologically significant effect in bone loss, and therefore we used six animals in this study. Animals were randomly assigned to each experimental group and assessments of the outcomes from these experiments were blinded to the investigators.

In vivo study design, CeONP 60/40 administration, and exposure of the rat hind limb to ionizing radiation
Thirty-six male SAS Sprague Dawley rats aged 8-9 weeks and weighing ~200g, were used in the in vivo study. Animals were acclimatized for a period of 1 week before commencement of the experiment. The handling of the animals was approved by the Institutional Animal Care and Use Committee of University of Central Florida (2020-48). All animal treatments were performed as per the CITI protocols for the University of Central Florida and NIH guidelines. Rats were randomly divided into three experimental groups: (1) mock-X-ray control group, (2) X-ray only group and, (3) X-ray + CeONP 60/40 given group. Rats were euthanized at either 7-or 14-days following the first of three doses of radiation. To prepare the animals for exposure to radiation, general anaesthesia was induced and maintained with 2% isoflurane and an oxygen flow rate of 0.3 mL/min. The main body of the rats were shielded by a radiation protection lead blanket (MPS-S, Z&Z Medical, USA), and the hind limbs of the rat were subjected to local fractionated irradiation (8 Gy/each time) on days 1, 3 and 5 of the study (a total of 24 Gy was applied), at a 160 kV tube voltage, a 4 mA tube current, and at a distance of 30 cm between the source and the surface of the animal (SC 500 smart controller, KIMTRON, USA). Rats received CeONP 60/40 at a dose of 4 mg/kg while suspended in sterile saline and 24 h prior to the first dose of radiation (day zero). A second dose of CeONP 60/40 was given on day 4. The CeONPs 60/40 were administered intravenously and via tail vein injection.

Analyses of CeONP 60/40 organ toxicity
In the animals euthanized on day 14, sections of the liver, pancreas and kidney were obtained and prepared for histological analysis using standard techniques [143]. Hematoxylin and eosin staining was performed according to the manufacturer's instructions (ab245880, Abcam, USA).

Analyses of CeONP 60/40 and in vivo DNA damage to cells within the bone marrow niche
On day 7 after irradiation, DNA damage to bone marrow cells was detected using the comet assay and as described above.

Whole-mount histological staining
Dissection of the femora was undertaken for detection of bony elements by hematoxylin and eosin (H&E) staining. Standard histology methods were used [143]. Briefly, samples were fixed, dehydrated in graded ethanol, and embedded in paraffin wax. Five μm thick sections were prepared from paraffin-embedded samples, followed by deparaffinising and rehydrating. Then, the tissue sections were stained with H&E solution. Longitudinal sections were prepared through the distal femoral condyle. Images were captured using an inverted phase microscope (BZ-X800E, Keyence, USA). To quantify bone area (%), trabecular bone area (BA) to total tissue area (TA) was calculated from 5 random slides in each group at the same magnification (×10) using Imagej software.

Analyses of CeONP 60/40 on the cellular expression of RANKL
Immunohistochemical staining was performed according to the Abcam protocol (https://www.abcam.com/ps/pdf/protocols/ihc_p. pdf). Briefly, sections were deparaffinised in xylene, hydrated in descending alcohol. Slides were boiled in 10 mM sodium citrate buffer (pH 6.0) for 10 min to retrieve antigenicity and then cooled down at room temperature for 20 min. The endogenous peroxidase activity was inactivated by a hydrogen peroxide blocking reagent (ab64218, Abcam, USA) for 10 min. After washing in PBS, sections were treated with a protein block (ab64226, Abcam, USA) at room temperature for 30 min. Immunostaining was performed overnight with the following primary antibodies diluted in antibody diluent (ab64211, Abcam, USA) at 4 • C overnight: RANKL (NB100-56512, Novus Biologicals, USA). Goat anti-Mouse IgG Secondary Antibody [HRP Polymer] (VC001-025, Novus Biologicals, USA) was used as secondary antibody. The slides were stained with SignalStain® DAB Substrate Kit (#8059; Cell Signaling Technology, USA) and counterstained with hematoxylin. All stained sections were dehydrated in graded alcohol baths of increasing concentration, cleared in xylene, and coverslipped with VectaMount™ AQ mounting medium (NC9354983, Fisher Scientific, USA). Images were captured using an inverted phase microscope (BZ-X800E, Keyence, USA).

Analyses of CeONP 60/40 on osteoclastic activity
Tartrate resistant alkaline phosphatase staining was performed to determine osteoclastic activity using the standard naphthol AS-BI phosphate post-coupling method. Briefly, tissue sections were rehydrated and incubated with TRAP staining solution containing 0.2 M sodium acetate buffer (pH 5.0), 50 mM L-(+) Tartaric acid, 0.5 mg/mL naphthol AS-MX phosphate, and 1.1 mg/mL Fast Red TR Salt 1,5-naphthalenedisulfonate salt (ab146351, Abcam, USA) for 1-3 h at 37 • C. Nuclei were then counterstained with hematoxylin for 5 min before mounting with VectaMount™ AQ mounting medium (NC9354983, Fisher Scientific, USA). Images were captured using an inverted phase microscope (BZ-X800E, Keyence, USA). To further investigate the effect of CeONP 60/40 on bone resorption, bone resorption marker C-terminal end of the telopeptide of type I collagen (CTX-1) was measured in rat sera by ELISA (NBP2-69077, Novus Biologicals, USA) per manufacturer instructions.

Analyses of CeONP 60/40 on cellular senescence
Frozen tissue sections (20 μm) from decalcified samples were used for detection of β-gal activity using the methods as described above. Nuclear fast red (50-317-51, Electron Microscopy Sciences, USA) was used for counterstaining.

Nano-CT scanning
Tibia was prepared for high-resolution -ray computed tomography (CT-scanning) using GE V|TOME|X M 240 Nano CT scanner (General Electric) at the University of Florida (Gainesville, FL, USA) with a 180 kv x-ray tube with a diamond-tungsten target and with the following settings: 75 kV, 150 mA, a 0.5 s detector time, averaging of three images per rotation and a voxel resolution of 12.4 μm 3D models of the trabecular network within the proximal tibia were created using a 3D Slicer (v4.11.20210226; Brigham and Women's Hospital and Massachusetts Institute of Technology). The DICOM files were imported, and a label map created. A threshold was used to automate the segmentation process and a smooth crushing tool used to manually clean the segments.

Three-point bending analyses of SAS rats
Immediately after the dissection, tibiae retrieved on day-7 were stored at − 20 • C. Three-point-bending tests were performed using a universal testing machine within 1 month of freezing (Criterion® 43, MTS, Minnesota, USA). Each tibia was loaded to failure at a displacement rate of 0.02 mm/s. The distance between the support bars of the 3point-bending fixture was 8 mm and each tibia was positioned horizontally with the anterior bow lying superiorly ( Supplementary  Fig. S10). A vertical force was applied to the tibial mid-shaft using a 3 mm diameter loading roller until failure occurred. Load-displacement curves were obtained, and the mechanical properties were calculated as follows: where σ is the stress (Pa), F is the applied load (N), L = 0.008 is the span distance between the supports (m), c o is the outer radius of the tibia's midshaft (m), and I is the moment of inertia (m 4 ) calculated as follows: where c i is the inner radius of the tibia's midshaft (m). The morphological parameters of three bones from each group were obtained following computed tomography (CT). Micro-computed tomography (μCT) scans were performed using a cone beam scanner (GE Phoenix Nanotom-M™, Waygate Technologies). Formalin fixed tibiae retrieved on day-7 were placed in 15 mL Eppendorf tubes and imaged at a 90 kV source voltage, 110 μA source current (mode 0) using a tungstendiamond target with a 500 ms exposure time at 7-9 μm isotropic voxel resolution (depending on femur size). Data was collected for 1080 projections over 360 • (0.33 • steps) with three averaged images per rotation position. Four cross-sections were selected from the same region of the proximal, mid and distal tibia, and both inner and outer diameters were calculated from each cross-section. The average value of inner-to-outer diameter from each cross-section were calculated to determine the inner radius. Ultimate stress, and fracture stress were subsequently quantified.

Statistical analyses
All numerical data is presented as mean ± standard deviation (SD). Statistical analysis was carried out using GraphPad Prism (version 8.0, US) and groups were compared using one-way analysis of variance (ANOVA) and a post-hoc Mann Whitney U test. p values < 0.05 were considered significant.

CRediT author statement
All authors made substantial contributions to conception and design, acquisition of data, or analysis and interpretation of data.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.