The study protocol was approved by the ethics committee of the First Affiliated Hospital of Hainan Medical University (Haikou, China). All animal experimental procedures were carried out in accordance with the guidelines laid down by the National Institute of Health (United States) and the National Laws and Regulations.

Female Sprague Dawley (SD) rats aged 16 weeks (Animal Laboratory Center of Chongqing Medical University, Chongqing, China) were housed under standard laboratory conditions (12-h light/dark cycle, temperature 22 ± 2°C, and humidity 55 ± 5%) with free access to water and standard rodent diet for a 7-day adaptation period. A total of 62 animals were randomly divided into OVX group (bilateral OVX, n = 50) and control group (sham operation, n = 12). Anesthesia was administered by subperitoneal injection with pentobarbital sodium (30–50 mg/kg). Among the OVX group, two rats died during the bilateral OVX surgery because of deep anesthesia. For the sham operation in the control group, the ovaries were exteriorized and then replaced back in the abdominal cavity. We observed the postoperative recovery, diet, wound healing, and daily activities after OVX. All rats were housed under standard laboratory conditions with free access to water and a standard rodent diet. All of them gradually obtained complete recovery and good wound healing.

Drugs included necrostatin-1 (Nec-1, Sigma-Aldrich, United States) and benzyloxycarbonyl-Val-Ala-Asp (zVAD, MP Biomedicals, Solon, OH, United States), which were dissolved in 10% dimethyl sulfoxide (DMSO, Sigma, United States) for subsequent use. zVAD and Nec-1 were used to specifically block cell apoptosis and necroptosis, respectively (Cui et al., 2016b; Ormsby et al., 2019). After 4 weeks of recovery after OVX, 48 rats in the OVX group were randomly assigned into four subgroups: OVX + vehicle group, OVX + zVAD group, OVX + Nec-1 group, and OVX + zVAD + Nec-1 group. Nec-1 (1.65 mg/kg/d) (Linkermann et al., 2012), zVAD (1.0 mg/kg/d) (Baumann et al., 2009), or an equal volume of 10% DMSO as the vehicle was injected intraperitoneally into the rats once per day for 4 weeks in the corresponding groups. The animals in the control group were administered with the vehicle solution. The rats were sacrificed 8 weeks after the operation by cervical dislocation because the peak time of osteocyte death was reported to be at 8 weeks after OVX (Cui et al., 2016a).

Both the right and left tibial plateaus were dissected, and the soft tissue was removed from the bone. The bone marrow was rinsed out from the tibial plateau by repeated washing with PBS. Only the bone matrix of the tibial plateaus were stored in liquid nitrogen for subsequent western blot analysis (n = 6 rats per group). Some tibial plateaus were fixed in 10% neutral buffered formalin for 24 h for micro–computed tomography (micro-CT) analysis and histological analysis. The tibial plateaus used for the histological analysis were decalcified in 15% ethylenediaminetetraacetic acid (EDTA, pH 7.4) for 4 weeks and then subjected to the transferase dUTP nick end labeling (TUNEL) and immunohistochemistry reaction for detection of caspase-3. Other tibial plateaus were fixed in 2.5% glutaraldehyde for 24 h at 4°C and decalcified in 15% EDTA (pH 7.4) for 4 weeks for transmission electron microscopy (TEM).

To analyze the bone microarchitecture, the tibial plateaus were scanned with micro-CT (viva CT40, SCANCO Medical AG, Zürich, Switzerland) using the following conditions: 15 µm resolution, 70 kVp, 114 μA, and 250 ms integration time. The volume of interest (29 × 29 × 29 µm³) was selected using a semiautomatic contouring method. 3D images were constructed using a series of planar transverse gray-value images. The microarchitectural parameters including bone mineral density (BMD), bone volume fraction (BV/TV), trabecular thickness (Tb.Th), trabecular number (Tb.N), and trabecular separation were calculated automatically using the micro-CT software.

The tibial plateaus frozen in liquid nitrogen were pulverized and homogenized in ice-cold buffer (Beyotime, Nantong, Jiangsu, China). The total protein concentration was determined with the bicinchoninic acid protein assay kit (Beyotime, Nantong, Jiangsu, China) according to the operating manual. The protein levels were determined by measuring the absorbance at 280 nm. Approximately 80 µg of protein was separated by SDS-PAGE and transferred to the polyvinylidene difluoride membrane (EMD Millipore, United States), which was then blocked with 5% nonfat dry milk solution for 1.5 h at room temperature and incubated overnight at 4°C with the following primary antibodies: anti-RIPK3 polyclonal antibody (ab152130, Abcam, Inc., Cambridge, MA, United States, 1:1000 dilution), anti-cleaved caspase-3 polyclonal antibody (ab2302, Abcam, Inc., Cambridge, MA, United States, 1:1000 dilution), anti-TNF-α monoclonal antibody (ab205587, Abcam, Cambridge, MA, United States, 1:1000 dilution), or anti-β-actin monoclonal antibody (sc-47778, Santa Cruz Biotechnology, CA, United States, 1:1000 dilution). Western blotting was performed routinely as previously described (Cui et al., 2016b). The protein bands of interest were visualized using an external cavity laser detection kit (Beyotime, Nantong, Jiangsu, China). The protein expression was quantitated by densitometry using Image Lab version 2.1 (Bio-Rad) and quantitative densitometric values for each kind of protein were normalized to β-actin.

The bone matrix of the tibial plateaus was sliced into 4-µm sections for immunofluorescence and in situ fluorescence TUNEL staining. After three washes with 0.1 M PBS, the sections were incubated with 20 μg/ml proteinase K at 37°C for 15 min, treated with 0.1% Triton X-100, blocked with 10% goat serum, and incubated with polyclonal antibody against cleaved caspase-3 (Cell Signaling Technologies, Danvers, MA, United States; 1:100 dilution) overnight at 4°C, followed by anti-mouse IgG (EarthOx, San Francisco, United States). After three washes with PBS, the sections were analyzed by using the in situ cell death detection kit (Roche, Basel, Switzerland) following the manufacturer's protocol for the TUNEL method. All the sections were counterstained with 4′,6-diamidino-2-phenylindole (DAPI). Finally, the number of total cells and TUNEL-positive cells with immunostaining after caspase-3 (with and without cleaved caspase-3 immunostaining) were counted in three to five noncontiguous high-power fields (×200 magnification) for each specimen under a laser scanning confocal microscope (Leica TCS SP2, Wetzlar, Germany). The percentage and number of TUNEL-positive and cleaved caspase-3–positive cells were calculated by an experienced pathologist who was blinded to the experimental conditions.

The tibial plateau (1 mm³) was fixed in 2.5% glutaraldehyde for 24 h at 4°C, washed with 0.1 M PBS (pH 7.4), and demineralized in 15% EDTA for 4 weeks at room temperature. After three washes with PBS, tissue fragments were postfixed in 2% osmium tetroxide for 1 h, block-stained with 2% uranyl acetate, embedded in epoxy resin, and sliced into 80-nm sections which were subsequently stained with uranyl acetate and lead citrate and finally observed for the ultrastructures by using a transmission electron microscope (Hitachi-7500, Hitachi, Tokyo, Japan).

Murine long bone osteocyte Y4 (MLO-Y4, 2.5×10⁴ cells, Cell Bank of the Chinese Academy of Sciences, Beijing, China) were plated in 12-well plates and cultured in DMEM/F12 (Gibco Life Technologies, Carlsbad, CA, United States) supplemented with 10% fetal bovine serum (FBS; Gibco Life Technologies, Carlsbad, CA, United States) in a humidified incubator containing 5% CO₂ at 37°C. The cells were pretreated with DMSO (1%), Nec-1 (30 mmol/L), zVAD (25 mmol/L), or Nec-1 (30 mmol/L) + zVAD (25 mmol/L) for 30 min at 37°C, and then treated with TNF-α (100 ng/ml) for 24 h (Zhu et al., 2016). Flow cytometric analysis was performed using a FITC Annexin V apoptosis detection kit (BD Biosciences, San Diego, CA) according to the manufacturer's protocol. The cells were harvested, washed twice with PBS, and resuspended in 100 µL binding buffer. Cells were incubated for 15 min at room temperature in the dark after adding 5 µL FITC Annexin V and propidium iodide (PI), and then diluted with 400 µL of binding buffer. The cells were analyzed by flow cytometry (BD Biosciences, San Jose, CA). The cells in the four different quadrants were analyzed and interpreted as Q1: necrosis (Annexin V−/PI+), Q2: late apoptosis (Annexin V+/PI+), Q3: early apoptosis (Annexin V+/PI−), and Q4: viable cells (Annexin V−/PI−) (Malik et al., 2020). The percentages of late apoptosis plus necrosis and early apoptosis were calculated. Then, the cells (5×10⁶) were harvested and prefixed with 2.5% glutaraldehyde for 2 h. These cells were observed by TEM as described previously. The percentage of dead cells was calculated in six noncontiguous fields (×6,000 magnification) in each group.

Data were presented as mean ± standard deviation (SD). Statistical analysis was performed using SPSS software version 22.0. The significant differences between groups were evaluated using ANOVA followed by the least significant difference or Dunnett's post hoc test. The p value of <0.05 was considered statistically significant.

About 8 weeks after OVX in SD rats, bone microarchitectures were measured by micro-CT (Figure 1). Figure 1A demonstrates the 2D reconstruction of the tibial plateaus and the BV/TV (Figure 1B), Tb.N (Figure 1C), Tb.Sp (Figure 1D), and Tb.Th (Figure 1E) in each group were compared. Compared to that in the control group, OVX led to significantly decreased BV/TV (Figure 1B), Tb.N (Figure 1C), and Tb.Th (Figure 1E) and increased Tb.Sp (Figure 1D), confirming that OVX resulted in substantial loss of bone mass and osteoporosis (p < 0.05). Compared with the vehicle group, BV/TV (Figure 1B), Tb.N (Figure 1C), and Tb.Th (Figure 1E) were remarkably increased in the Nec-1 group, zVAD group, and Nec-1 + zVAD group (p < 0.05), accompanied by reduced Tb.Sp (Figure 1D). In terms of BV/TV, the Nec-1 group, zVAD group, and Nec-1 + zVAD group led to approximately 57% increase, 44% increase, and 73% increase than vehicle group, respectively (Figure 1B). These three groups were also associated with about 24% increase, 19% increase, and 26% increase in Tb.N relative to the vehicle group (Figure 1C). For Tb.Sp, we found approximately 27% decrease in Nec-1 group, 39% decrease in zVAD group, and 41% decrease in the Nec-1 + zVAD group compared to the vehicle group (Figure 1D). Regarding Tb.Th, these three groups results in about 30% increase, 29% increase, and 40% increase than vehicle group (Figure 1E). These results suggested that treatment with Nec-1, zVAD and especially Nec-1 + zVAD could improve bone mass as compared to the vehicle group.

DNA fragments produced by dead cells in the bone matrix of the tibial plateau were detected by the TUNEL assay (Figure 2A). The percentage of TUNEL-positive cells in the bone matrix of the tibial plateau in the vehicle group was greater than in the control group (Figure 2B). The percentage of TUNEL-positive cells was reduced significantly after treatment with zVAD or Nec-1 alone, and the significance of reduction became more robust in the Nec-1 + zVAD group. It was important to emphasize that the number of TUNEL-positive cells in OVX rats treated with Nec-1 or Nec-1 + zVAD was significantly lower than in rats treated with zVAD alone (Figures 2A,B). Furthermore, anti-active caspase-3 staining was conducted to identify cell apoptosis (Günther et al., 2011; Liedtke et al., 2011; Vanlangenakker et al., 2012). These results indicated that both necroptosis and apoptosis led to the death and loss of osteocytes, and cell necroptosis had a bigger role in the loss of osteocytes than cell apoptosis.

To further examine the morphological characteristics of dead osteocytes, we observed the ultrastructures of osteocytes in the bone matrix of the tibial plateaus using TEM at 8 weeks after OVX. The typical morphological features of necroptotic osteocytes included swelling of the organelles, rupture of the plasma membrane, chromatin clumping, and intracellular vacuoles, which were seen in the vehicle group and zVAD group. Nuclear condensation of apoptotic cells was observed in the Nec-1 group (Figure 3). These indicated that osteocyte death was mainly caused by necroptosis rather than apoptosis at 8 weeks after OVX.

RIPK3 and cleaved caspase-3 were the important markers to identify necroptosis and apoptosis of cells, respectively. Their protein expression levels were measured in the bone matrix of OVX rats by using western blotting. The results showed that RIPK3 and caspase-3 protein levels were increased in the bone matrix of OVX rats compared with those in sham-operation rats (Figure 4A). Treatment with Nec-1 significantly reduced the RIPK3 level (Figure 4B), while the caspase-3 level was substantially decreased in the zVAD group than in the Nec-1 group (Figure 4C). This confirmed that both necroptosis and apoptosis contributed to the death and loss of osteocytes.

TNF-α is mainly produced in hematopoietic precursor cells and is one important inflammatory factor which induces PMOP (Liu et al., 2020; Lu et al., 2020). TNF-α could directly enhance the expression of the receptor activator of nuclear factor κΒ ligand (RANKL) in osteocytes and promote osteoclast formation (Kitaura et al., 2020). We detected the expression of TNF-α in the bone marrow of OVX rats in each group by western blotting and found that TNF-α expression was significantly lower in the control group and increased sharply in the vehicle group (Figures 5A,B). Although treatment with zVAD or Nec-1 showed a decreasing trend in TNF-α protein expression compared to the vehicle group, no statistical difference was seen between them.

Estrogen deficiency increased the expression of TNF-α in the bone marrow of OVX rats, and we explored the influence of TNF-α abnormality on the death of osteocytes. Therefore, we cultured MLO-Y4 cells in order to compare necroptosis with apoptosis for osteocyte death. Flow cytometry and TEM analyses revealed that the cell death rate of MLO-Y4 cells stimulated by TNF-α was significantly increased. Compared with the control group, early rate of cell death in the TNF-α group was significantly increased, and zVAD effectively reduced the early death rate of MLO-Y4 cells, but Nec-1 had no significant effect on the early death rate (Figures 6A,D). The late death rate in the TNF-α group was also significantly higher than in the control group. Both Nec-1 and zVAD remarkably reduced late necrosis and apoptosis of MLO-Y4 cells, and Nec-1 could result in a significantly lower rate of cell death than did zVAD (Figures 6A,E). By contrast, the incidence of necroptotic cells was greater than that of apoptotic MLO-Y4 cells in response to TNF-α stimulation (Figure 6B, C, F). TNF-α combined with zVAD significantly decreased the rate of apoptosis but increased the incidence of necroptosis. Furthermore, TNF-α combined with Nec-1 only blocked necroptosis but did not change the rate of apoptosis (Figure 6B, C, F). In addition, Nec-1 plus zVAD sharply reduced the incidence of apoptotic and necroptotic MLO-Y4 cells compared to vehicle intervention. These results confirmed that the death of osteocytes after TNF-α stimulus was caused by necroptosis and apoptosis, and cell necroptosis had a bigger role in the loss of osteocytes than did cell apoptosis. These findings are consistent with the results in vivo.