CoCrMo‐Nanoparticles induced peri‐implant osteolysis by promoting osteoblast ferroptosis via regulating Nrf2‐ARE signalling pathway

Abstract Objectives Aseptic loosening (AL) is the most common reason of total hip arthroplasty (THA) failure and revision surgery. Osteolysis, caused by wear particles released from implant surfaces, has a vital role in AL. Although previous studies suggest that wear particles always lead to osteoblast programmed death in the process of AL, the specific mechanism remains incompletely understood and osteoblast ferroptosis maybe a new mechanism of AL. Materials and Methods CoCrMo nanoparticles (CoNPs) were prepared to investigate the influence of ferroptosis in osteoblasts and calvaria resorption animal models. Periprosthetic osteolytic bone tissue was collected from patients who underwent AL after THA to verify osteoblast ferroptosis. Results Our study demonstrated that CoNPs induced significant ferroptosis in osteoblasts and particles induced osteolysis (PIO) animal models. Blocking ferroptosis with specific inhibitor Ferrostatin‐1 dramatically reduced particle‐induced ferroptosis in vitro. Moreover, in osteoblasts, CoNPs significantly downregulated the expression of Nrf2 (nuclear factor erythroid 2‐related factor 2), a core element in the antioxidant response. The overexpression of Nrf2 by siKeap1 or Nrf2 activator Oltipraz obviously upregulated antioxidant response elements (AREs) and suppressed ferroptosis in osteoblasts. Furthermore, in PIO animal models, the combined utilization of Ferrostatin‐1 and Oltipraz dramatically ameliorated ferroptosis and the severity of osteolysis. Conclusions These results indicate that CoNPs promote osteoblast ferroptosis by regulating the Nrf2‐ARE signalling pathway, which suggests a new mechanism underlying PIO and represents a potential therapeutic approach for AL.


| INTRODUC TI ON
Total hip arthroplasty (THA) serves as one of the most successful orthopaedic surgeries that are widely used for the treatment of severe degenerative, post-traumatic, and other end-stage diseases of the hip joint. 1 Nevertheless, periprosthetic aseptic loosening and subsequent osteolysis always result in THA failure and surgical revision. 2,3 The main cause of revision surgeries is due to wear particles, which are created from implanted prostheses. 4 Wear particles can lead to variable adverse local tissue responses, which disrupt the homeostasis between bone formation and resorption. Many different types of cells, including osteoclasts, osteoblasts, macrophages, fibroblasts, and lymphocytes, contribute to the occurrence of osteolysis to varying degrees. 5,6 Numerous researches have focused on osteoclasts and macrophages, which are mainly related to increased bone resorption during osteolysis. 7 Among the cell lines stimulated by wear particles, osteoblasts are one of the earliest cells exposed to wear particles; however, the influence of wear particles on osteoblasts has not attracted enough attention. 8 Osteoblasts are of crucial importance in the secretion of extracellular matrix (mainly type I collagen) in vivo, which are the exclusive cells engaged with bone formation. 9, 10 Wear particles have been shown to have adverse effects on the proliferation and survival of osteoblasts. 11 The mechanism underlying the interaction between wear particles and osteoblasts remains unclear.
Ferroptosis, a novel form of regulated cell death pathway, is characterized by the accumulation of lethal lipid reactive oxygen species (ROS). 12,13 It is inconsistent with the classical apoptotic cell death programmes, including apoptosis, necroptosis, and other classical non-apoptotic cell death programmes, in terms of morphological, biochemical, and genetic studies. 14 Current research has proved that ferroptosis can be induced by the blockade of cystine/glutamate system Xc − activity, downregulation of glutathione peroxidase 4 (GPX4), and an increase in lipid ROS. 15,16 System Xc − (composed of SLC7A11, solute carrier family 7, and SLC3A2, solute carrier family 3 member 2) is a transmembrane cysteine-glutamate antiporter that specifically transfers extracellular cystine to intracellular glutamate as the origin of glutathione (GSH), which is the major cellular antioxidant. 17 GPX4, another core regulator of ferroptosis, reduces organic hydroperoxides and lipid peroxides through the consumption of GSH. [18][19][20] The inhibition of GPX4 and System Xc − can result in an uncontrolled increase in lipid peroxidation, resulting in ferroptosis. 21 Moreover, acyl-CoA synthetase long-chain family member 4 (ACSL4), a vital factor in metabolism-associated diseases, has been linked to the promotion of the esterification of arachidonoyl (AA) and adrenoyl into phosphatidylethanolamine (PE), which is strongly associated with ferroptosis and critically determines the sensitivity of the cells to ferroptosis. 22,23 Many diseases have been reported to be related to ferroptosis, including Alzheimer's disease, 24 carcinogenesis, 25 intracerebral haemorrhage, 26 traumatic brain injury, 27 stroke, 28 and ischaemia-reperfusion injury. 22 However, the relationship between ferroptosis and particles induced osteolysis (PIO) is still unknown.
Nrf2, the nuclear factor erythroid 2-related factor 2, is of crucial significance to the antioxidant response that protects cells from different kinds of cell death-induced oxidative stress, including ferroptosis. 29,30 Nrf2 binds its promoter to target genes to regulate cellular redox homeostasis, 17,31 which contains antioxidant response elements (AREs), such as heme oxygenase-1 (HO-1) and quinone oxidoreductase-1 (NQO-1). Numerous studies have shown that Nrf2 can be continually degraded by Kelch-like ECH-associated protein 1 (Keap1), and the suppression of Keap1 activates the Nrf2-ARE pathway under various oxidative stress conditions. 29,32 The ROS content around the prosthesis increases significantly during wear-particlemediated osteolysis. 33,34 However, how Nrf2 regulates the process of wear-particle-mediated osteolysis requires further study.
In the present study, we hypothesized osteoblast ferroptosis was an important reason for PIO. We used the most common clin-

| Nanoparticle preparation and characterization
CoCrMo nanoparticles (CoNPs), which had an average particle diameter of 90.87 nm, were ground and supplied by Nanjing Kester Metal Products Company. CoNPs were stocked at a concentration of 50 mg/ml in phosphate-buffered saline (PBS). They were resuspended well in PBS after sonication for 15 min. The CoNPs were autoclaved prior to treatment. The stock solution of CoNPs was diluted to a specific concentration and ultrasonicated for 15 min prior to being exposed to the cells. A field emission scanning electron microscope (Merlin, Carl Zeiss AG, Germany) and transmission electron microscope (JEM-2000, Japan) were used to determine the morphology of the CoNPs. The size distribution was determined by dynamic light scattering (DLS) using a BI-200SM multiangle dynamic/static laser scattering instrument (Brookhaven, USA). The chemical composition of the CoNPs was performed using X-ray photoelectron spectroscopy (XPS) with a photoelectron spectroscopy system (PHI 5000 Versa Probe II, ULVAC-PHI).

| Cell culture
MC3T3-E1 cells were obtained from the Cell Bank of Chinese Academy of Sciences (Shanghai, China). MC3T3-E1 cells were maintained in α-MEM containing 10% FBS, 100 μg/ml streptomycin, and 100 U/ml penicillin in a cell incubator with 5% CO 2 at 37°C.

| RNA-seq
The MC3T3-E1 cells were stimulated by CoNPs (50 μg/ml) for 24 h, and total RNA was obtained using TRIzol Reagent. The extracts were screened and amplified using PCR and sequenced using the Illumina HiSeq 4000 instrument. For bioinformatics analysis, raw reads which contained the adapter or had low quality (Q-value ≤20) were deleted and then located in the mouse-related genome using HISAT2 (version 2.1.0). Samtools were used to sequence the resulting files, and HTSeq (version 0.9.0) analysis was performed to determine the count of every gene. 35 Here, we used FDR <0.05 and log 2 (fold change) >2 as thresholds to define substantial differentially expressed genes (DEGs). We performed Gene Ontology (GO) and Kyoto Encyclopaedia of Genes and Genomes (KEGG) pathway analyses for further study. Gene set enrichment analysis (GSEA) and bioinformatics analysis were carried out by OmicStudio tools (https:// www.omics tudio.cn/tool).

| Periprosthetic osteolytic bone tissues
Periprosthetic osteolytic bone tissue was obtained from patients who underwent aseptic loosening after THA. Cases of local infection were carefully ruled out. The clinical data and preoperative Xray examinations of the patients are shown in Table 1 and Figure 3A.
Samples from patients with developmental dysplasia of the hip (DDH) who underwent THA were used as controls. Unloose specimen was collected from a patient who underwent THA 5 years back and needed further surgery for mechanical loosening due to a tumbling accident. Samples for cases 1-4 were collected from patients  for 2 h. The optical density (OD) at 450 nm was measured by a microplate photometer.

| Western blotting
We used western blotting to analyze the protein expression.
Proteins were extracted from MC3T3-E1 cells through ice-cold radioimmunoprecipitation assay lysis buffer after different treatments. The BCA protein assay (Beyotime) was applied to determine ImageJ software was adopted to quantify western blot.

| Nuclear and cytoplasmic extraction
MC3T3-E1 cells were treated with varying concentrations (0, 25, 50, 75, and 100 μg/ml) of CoNPs for 24 h. To extract cytoplasmic along with nuclear proteins, a nuclear and cytoplasmic extraction kit (Beyotime) was adopted to process cell pellets in line with the instructions of the manufacturer.

| ROS analysis
To

| Lipid peroxidation assay and GSH assays
To assess the end product of lipid peroxidation and malondialdehyde (MDA), we used the lipid peroxidation (MDA) assay kit (Jiancheng, Nanjing, China). The cells were homogenized in MDA lysis buffer on ice and centrifuged at 13,000 × g for 3 min. The MDA-TBA adduct was generated when the MDA in the cell lysate reacted with thiobarbituric acid (TBA) solution, of which the chromaticity (OD = 532 nm) was proportional to the amount of MDA present. We measured the relative GSH concentration in cell lysates employing a glutathione assay kit (Sigma, USA) in accordance with the manufacturer's instructions.

| Immunofluorescence staining
After

| Particle -induced osteolysis animal model
A surgically induced calvaria resorption model using CoNPs was used to study the pathogenesis of osteolysis. In summary, 25 10-week-old male C57BL/6 mice were obtained from Slac Laboratory Animal Co.
Ltd. and housed in a specific pathogen-free facility at the Laboratory Animal Department of Shanghai General Hospital. The mice were di-

| Micro-CT analysis
After harvesting, the mouse calvaria were analyzed with high-

| Histological assessments
All C57BL/6 mice were euthanized 2 weeks after surgery. The mouse calvaria were fixed in 4% PFA for 24 h, decalcified with 10% EDTA for one month, and embedded in paraffin. The calvaria was sliced into 4μm-thick sections and stained with haematoxylin and eosin (H&E).

| Statistical analysis
All data represent means ± SD of triplicate independent experiments. Data analysis was performed by SPSS software (version 22.0; SPSS, Chicago, USA). T-tests or a one-way ANOVA were used to determine the differences between various groups. ( *,# indicate p < 0.05, **,## indicate p < 0.01, ns indicates not significant).

| Physical characteristics of CoNPs
Scanning electron microscopy (SEM) image ( Figure 1A) and transmission electron microscopy (TEM) image ( Figure 1C) performed the appearance of CoNPs, which varied in size and were basically spherical. Figure 1D shows further size distribution analysis using a Nano Sizer. The size of particles ranged between 29.30 nm and 204.27 nm, and the mean particle diameter was 90.87 nm ( Figure 1E). The chemical composition of the CoNPs was performed using XPS, and the main composition of CoNPs was Co2p, Cr2p and Mo2d, as shown in Figure 1B Figure 2F). 37 In brief, our RNA-seq analysis results suggest that ferroptosis occurs in MC3T3-E1 cells after treatment with CoNPs.

| Activation of ferroptosis and suppression of osteoblast in peri-implant bone tissue of patients with aseptic loosening
Preoperative radiographs of the patients are shown in Figure 3A.
Previous studies have shown that the degree of PIO assessed by intraoperative discovery was consistent with preoperative X-rays and the implantation time of the revision patient.

| CoNPs-induced osteoblasts ferroptosis in vitro
For sake of exploring the role of CoNPs in osteoblasts, the effect of CoNPs stimulation on the survivability of osteoblasts was verified.
As shown in Figure 4A, with the increase of CoNPs stimulation time or concentrations, the number of osteoblasts decreased gradually.
In subsequent studies, we selected the appropriate CoNPs concentration and stimulation time to treat osteoblasts. Transmission electron microscopy images of cell mitochondria showed the occurrence of ferroptosis. As presented in Figure 4B, compared with controlled-treated osteoblasts, cells treated with CoNPs (50 μg/ ml) showed smaller, crumpled, and fractured mitochondria, and the cristae of mitochondria disappeared. 41 Cells treated with CoNPs (100 μg/ml) changed more prominently. Several key factors, such as SLC7A11, GPX4, ACSL4, and COX2, are considered as key proteins involved in ferroptosis regulation. 12 Hence, to assess ferroptosis sensitivity after MC3T3-E1, the cells were treated with CoNPs in vitro.
We verified the content of the proteins under different concentrations of CoNPs (0, 25, 50, 75 and 100 μg/ml). Figure 4C and D dis-  Figure S2A and B. In addition, according to Figure 4E and Figure S2C, real-time PCR analysis of Slc7a11, Gpx4, Acsl4, and Ptgs2 presented a similar trend as that observed in Figure 4C and Figure S1. Furthermore, lipid peroxidation, which can result in ROS accumulation, is one of the hallmarks of ferroptosis. 42 We used fluorescent probes, DCFH-DA and C11-BODIPY, to detect the content of ROS and lipid peroxidation in osteoblasts by flow cytometry. 43 As shown in Figure 4F and was detected, which is the end product of lipid oxidation. 36 As shown in Figure 4G, CoNPs caused an obvious upregulation in MDA levels in osteoblasts, similar to RSL3. 44 Finally, the GSH content in osteoblasts was quantified after treatment of CoNPs ( Figure 4H). There was a clear downregulation in CoNPs-treated osteoblasts compared with that in control-treated cells. In conclusion, the data showed that CoNPs triggered ferroptosis in osteoblast cells in vitro.

| Ferrostatin-1 ameliorates the ferroptosis induced by CoNPs in vitro
Ferrostatin-1 (Ferr-1), the inhibitor of ferroptosis, which could rescue the injury specifically, was used to determine the appearance of ferroptosis in osteoblasts after treatment of CoNPs. 45 Previous studies have shown that Ferr-1 can suppress ferroptosis in different cells. 46,47 Ferr-1 can prevent ferroptosis by directly inhibiting the production of reactive oxygen species. 12 CCK8 kit was used to confirm the concentration of Ferr-1 that suppressed osteoblast proliferation and viability ( Figure S3A). Compared with the control group, up to 4 μM/ml Ferr-1 had no obvious cytotoxic effect on osteoblasts.
Another CCK8 assay was performed to verify whether Ferr-1 could ameliorate the ferroptosis in osteoblasts induced by CoNPs. As shown in Figure 5A, pre-treating osteoblasts with Ferr-1 (1 μM/ml) for 12 h before treatment with CoNPs distinctly improved the cell viability compared with that of the cells treated directly with CoNPs.
The results from the western blot assay show that CoNPs treatment significantly downregulated the content of SLC7A11and GPX4 and upregulated the content of ACSL4 and COX2, while Ferr-1 inhibited the degradation of SLC7A11 and GPX4 and upregulated the expression of ACSL4 and COX2 in Figure 5B and C. Moreover, as presented in Figure 5D, we performed real-time PCR analysis of Slc7a11, Gpx4, Acsl4, and Ptgs2. The analysis showed the same results as Figure 5B and C. Furthermore, light microscopy showed that compared with control-treated osteoblasts, cells treated with CoNPs (50 μg/ml) shrank and their cell membranes ruptured. However, Ferr-1 rescued cell morphology and suppressed the ferroptosis induced by CoNPs ( Figure 5E). Next, we analyzed cellular ROS and lipid peroxidation in osteoblasts using DCFH-DA and C11-BODIPY, respectively. As shown in Figure 5F and I, the change in fluorescence showed that CoNPs treatment obviously increased the cellular ROS and lipid peroxidation, whereas Ferr-1 decreased the generation of ROS and lipid peroxidation in osteoblasts pre-treated with Ferr-1 for 12 h. We further quantified the end product of lipid oxidation, malondialdehyde (MDA), and GSH content. As shown in Figure 5G and H, Ferr-1 reduced the production of MDA and suppressed the degradation of GSH. Collectively, Ferrostatin-1 suppressed the ferroptosis induced by CoNPs in vitro.

| The expression of Nrf2 decreased in the osteoblasts after treatment with CoNPs
Nrf2, the nuclear factor erythroid 2-related factor 2, has been con-

| CoNPs-induced osteoblasts ferroptosis via regulating Nrf2-ARE signalling pathway
This study shows that CoNPs-induced osteoblast ferroptosis and SiRNA2 of Keap1 was selected for the next study. As shown in Figure S3B, Oltipraz had no significant toxicity on osteoblast viability until cell viability decreased with exposure to up to 20 μM.
Treatment with CoNPs (50 μg/ml) significantly upregulated the protein content of ACSL4, and decreased the protein expression of GPX4 and SLC7A11, but pre-treatment with Sikeap1, Oltipraz, or Sikeap1 plus Oltipraz reversed the tendency ( Figure 7B and C).
Moreover, confocal microscopy revealed that Oltipraz reversed the decrease in GPX4 expression in osteoblasts stimulated by CoNPs. A combination of Ferr-1 and Oltipraz could be more effective ( Figure 7E). Furthermore, Oltipraz obviously improved the cell viability ( Figure 7D

| Suppressing ferroptosis and activating Nrf2 signalling pathway ameliorated CoNPs-induced osteolysis in vivo
Our analysis confirmed that CoNPs downregulated the Nrf2-ARE signalling pathway and induced osteoblast ferroptosis in vitro. To further investigate the impact of osteoblast ferroptosis in the pathogenic process of osteolysis, we established a calvaria resorption model with male C57BL/6J mice, which is a frequently used model for the study of AL. 11 As mentioned above, we divided the mice into five groups. There were five C57BL/6J mice in each group. Micro-CT with 3-dimensional reconstruction was used to examine the osteolysis induced by CoNPs. Figure 8A shows that CoNPs-induced bone loss was suppressed Ferrostatin-1 plus Oltipraz relieved osteolysis more effectively ( Figure 8B). As shown in Figure S3C and D, the treatment of

| DISCUSS ION
THA is generally accepted as the most effective surgical method for trauma, degenerative arthritis, and other end-stage joint diseases.
Nevertheless, periprosthetic osteolysis and subsequent AL caused by wear particles frequently lead to the failure of THA and surgical revision, which greatly add to the social burden. 2  Although existing studies have demonstrated that osteoblasts are damaged during PIO, this research has been limited to in vitro studies, and the mechanism is not fully clear. 10 In this study, we found that wear particles caused osteoblast ferroptosis in the process of PIO, which may be a new therapeutic target in AL.
Previous studies have found that there are a variety of programmed cell death mechanisms for primary human osteoblasts and osteoblast-like cells stimulated by wear particles, including apoptosis and autophagy. 35 In this study, RNA-Seq analysis revealed that CoNPs downregulated the expression of GPX4 in osteoblasts, which is of crucial importance in reg- Considering the appearance of ferroptosis and the reduction in the number of osteoblasts, we focused on elucidating the relationship between ferroptosis and wear particle-stimulated osteolysis, and investigating the effect of suppressing ferroptosis as a potential approach for the treatment of PIO. The transcription factor Nrf2 has been proved to regulate cellular antioxidant response by expressing many genes that counteract oxidative and electrophilic stresses. 33  autophagy is also involved in the process of PIO induced by programmed death of osteoblasts, and the systematic dissection of different pathways requires further investigation.

| CON CLUS IONS
In conclusion, our results highlight that CoCrMo-Nanoparticles induced peri-implant osteolysis by promoting osteoblast ferroptosis via regulating Nrf2-ARE signalling pathway ( Figure 9). We carried F I G U R E 9 The schematic illustration of CoCrMo-Nanoparticles induced peri-implant osteolysis by promoting osteoblast ferroptosis via regulating Nrf2-ARE signalling pathway out RNA-Seq transcriptome analysis of MC3T3-E1 osteoblasts, and the results showed that osteoblast ferroptosis could be a vital reason for imbalanced bone metabolism. Marked characterization of ferroptosis was performed in osteoblasts in vitro and in vivo. In addition, all present data clearly show that CoNPs induce PIO by promoting osteoblast ferroptosis through the regulation of the Nrf2-ARE signalling pathway. More importantly, blocking ferroptosis with Ferrostatin-1 and Oltipraz significantly ameliorated osteolysis induced by implanted particles. In summary, this study may imply a prospective mechanism underlying wear-particle-induced osteolysis and represent a potential therapeutic strategy for treating aseptic loosening.

ACK N OWLED G EM ENTS
This work was financially supported by the National Natural Science Research project (18sjkjgg18).

CO N FLI C T O F I NTE R E S T
The authors declare no conflicts of interest for this work.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data included in this study are available upon request from the corresponding author.