Male C57BL/6J mice at two months of age were purchased from Shanghai SLAC Laboratory Animal Company (Shanghai, China), and this study was approved by the animal ethics committee of Shanghai Ninth People’s Hospital. The mice were fed commercial food and water under specific aseptic (SPF) conditions. For cold stimulation, thirty mice were randomly divided into two groups, cold stimulation (cold) and room temperature (normal), with fifteen mice per group. Briefly, mice in the cold group grew in incubators (Fuyilian, FYL-YS-280 L) at 4°C, while mice in the normal group were in the same type of incubators at room temperature (23°C) (36). Mice were euthanized after being exposed to cold/normal temperature for 1 d, 14 d and 28 d, and tibias were then collected.

To calculate dynamic bone histomorphometry, animals were injected with 30 mg/kg calcein (Sigma) and alizarin red (Sigma) 10 and 3 days before euthanasia. Nondecalcified tibiae were embedded in methyl methacrylate and sectioned.

After imaging with a confocal microscope, histomorphometric examination was confined to the consistent cortical region and was performed using BIOQUANT OSTEO 2019 (v19.6.60). The mineral apposition rate (MAR) and bone formation rate per bone surface (BFR/BS) were analyzed at 40× magnification from 6 representative fields per bone sample (37).

At the end of each experiment, the tibias of the mice were fixed in 4% paraformaldehyde. Samples were scanned using micro-CT (μCT 80; Scanco, Zurich, Switzerland) as previously described (38). The micro-CT parameters were as follows: voltage, 70 kV; electric current, 114 μA; and resolution, 10 μm per pixel. Three-dimensional structural parameters, including bone volume fraction (BV/TV), trabecular number (Tb.N), trabecular thickness (Tb.Th) and trabecular separation (Tb.Sp), were analyzed (38, 39).

Perkin Elmer micro-CT was applied to analyze brown adipose tissue as described in previous studies (40, 41). Briefly, mice were anesthetized with 30 mg/kg pentobarbital sodium and then scanned in the machine to measure BAT volume in the scapula. The data were analyzed using Analyze 12.0 according to the instructions to determine the BAT volume.

Mice were scanned by dual-emission X-ray absorptiometry (DXA, Hologic Discovery A) through animal models to obtain the lean mass, %lean, fat mass, %fat and whole-body weight.

Interleukin-6 (IL-6) (70-EK206/3-96, Multisciences), P1NP (Lengton, Shanghai), and CTX-1 (Lengton, Shanghai) ELISA kits were used to detect the expression levels of IL-6, P1NP, and CTX-1 in the serum, according to the manufacturer’s instructions. A microplate reader (Bioteck, Arcugnano [Vicenza], Italy) was used to determine the optical density (OD) of each well at 450 nm.

After decalcification in 10% ethylenediaminetetraacetic acid (EDTA) for 3 weeks, samples were embedded in paraffin. To observe the microstructure of the samples, 4-µm-thick sagittal sections of the medial compartment of the knee joint were cut. After TRAP staining was performed, OC surface/bone surface (Oc.N/BS) was calculated using a Bioquant system.

BAT was isolated and cultured following previously described methods (42). Briefly, 4-week-old C57BL/6 mice were euthanized. Interscapular BAT was collected, minced, and digested in collagenase digestion buffer (DMEM, 1 mg/ml collagenase I, 1% FBS). Preadipocyte cells were collected by filtering through 70 µm membranes and centrifuging. Preadipocytes were cultured to 80%-90% confluence in DMEM supplemented with 10 ng/ml bFGF (Pepro Tech), 10% fetal bovine serum (Gibco) and pen/strep (Life Technologies). Cells were subcultured every 3 days and used from passages 3 to 5. Then, preadipocytes were differentiated with DMEM containing 10% fetal bovine serum (Gibco), 10 µg/ml insulin (Sigma), 1 µM dexamethasone (Sigma), 0.5 mM 3-isobutyl-1-methylxanthine, phosphodiesterase inhibitor (IBMX, Sigma), 5 µM rosiglitazone (Sigma), and 1 nM T3 (Sigma) for 6 days until brown adipocyte formation.

To obtain conditioned medium (BAT CM), DMEM with 10% exosome-free FBS was used to culture brown adipocytes, which were collected after 48 hours. The conditioned medium was centrifuged at 300 × g for 10 min to discard cells and further centrifuged at 2,000 × g for 10 min and at 10,000 × g for 30 min to remove cellular debris and large vesicles, respectively (43, 44). After filtration through a 0.22 µm filter, conditioned medium was used to culture BMMs and BMSCs.

To obtain bone marrow macrophages (BMMs), 4-week-old C57BL/6J mice were sacrificed, and their femurs and tibias were separated under sterile conditions. Bone marrow was flushed from the mouse femurs and tibias using complete a-MEM, which contained macrophage-colony stimulating factor (M-CSF, 30 ng/mL). After resuspension, bone marrow cells were cultured in a 10-cm dish at 37°C in 5% CO₂. The medium was changed after 3 and 7 days to remove nonadherent cells. When the cellular density reached 80% confluence, BMMs were washed three times with phosphate buffered saline (PBS) and collected using 0.25% trypsin for subsequent experiments. Bone mesenchymal stem cells (BMSCs) were obtained through the same procedure but were cultured without M-CSF (45). For osteoclastogenesis, BMMs were seeded into 24-well plates (10⁵ cells/well) or 96-well plates (10⁴ cells/well) and then treated with osteoclastogenesis medium, which consisted of complete a-MEM, receptor activator of nuclear factor kappa-B ligand (RANKL, 50 ng/mL) and M-CSF (30 ng/mL). BAT CM was added or not to the osteoclastogenesis medium for 7 days until osteoclasts were formed (46). After fixing the cells in 4% paraformaldehyde for 30 min, TRAP-positive cells were stained using a TRAP staining kit (Sigma-Aldrich, 387A-1KT). ImageJ software (Media Cybernetics Bethesda, MD, United States) was used to calculate the number of multinuclear (n≥3) TRAP-positive cells in each well. To determine the bone resorption ability, BMMs were differentiated in an osteoassay stripwell plates for 8 days. Then, the cells were washed with 4% sodium hypochlorite for 10 minutes followed by three washes with double distilled water. The resorption area was calculated using ImageJ software (Media Cybernetics Bethesda, MD, United States).

BMSCs were seeded into 24-well plates and cultured until the they reached 80% confluence. Then, the medium was replaced with an osteogenesis assay kit (MUBMX-90021, Cyagen, CA, United States) with or without BAT conditioned medium (1:1) at 37°C in humidified air with 5% CO₂ for 21 days to induce osteogenesis. Bone formation was detected using alkaline phosphatase (ALP) or alizarin red staining on days 14 and 21. ALP staining was performed as follows: after washing three times with PBS and fixation in 4% paraformaldehyde for 10 min at room temperature, cultured cells were stained using the BCIP/NBT Alkaline Phosphatase Color Development Kit (Beyotime Institute of Biotechnology, Shanghai, China). All steps were strictly in accordance with the manufacturer’s instructions. After 21 days of culture, alizarin red staining was performed. Briefly, the cultured cells were washed with PBS and fixed in 4% paraformaldehyde for 30 min, and then 500 µL alizarin red dye (contained in the MUBMX-90021 kit) was added to each well and incubated at room temperature for 10 min. After washing five times with PBS, 10% cetylpyridinium chloride (500 µL) (H811089, Macklin, CA, United States) was added to each well for semiquantitative analysis. Then, the absorbance of the supernatant at 562 nm was detected after incubation for 30 min at room temperature.

TRIzol reagent (Thermo Scientific, US) was used to extract total RNA. After the concentration was measured, the total RNA was converted to complementary DNA using a Quant script RT Kit (Promega, Madison, WI, USA). To detect messenger RNA (mRNA) levels, cDNA and SYBR Premix Ex Taq Mix (Selleck) PCR in 10 µL PCRs were performed in the Real‐Time PCR System (Light Cycler 2.0; Roche Diagnostics GmbH, Mannheim, Germany). The primer sequences are shown in Table 1 .

GraphPad Prism 5.0 (GraphPad Software Inc., CA, United States) was used for statistical analysis of data. Each experiment was repeated at least three times. For animal studies, each group had at least three mice. All quantitative values are presented as the mean ± standard deviation (SD). Two−way analysis of variance (ANOVA) or Student’s t-test was used for analysis of differences. Bonferroni correction was performed for multiple comparisons. P < 0.05 was considered statistically significant.

After being exposed to 4°C for different times, mice exhibited fluctuations in bone microstructure ( Figure 1A ). As shown in Figure 1B , BV/TV and Tb.Th decreased at 14 days but recovered after 28 days of exposure. Tb.Sp in both groups increased after 28 days of exposure, but it was higher in the cold-treated group. Tb.N in both groups decreased with prolonged time and was lower in the cold-treated group.

Calcein and alizarin red double labeling were performed to determine the bone formation rate ( Figure 2A ). As shown in Figure 2C , the bone formation rate/bone surface (BFR/BS) decreased dramatically after 14 days of cold stimulation and increased at 28 days, but there were no significant changes in the mineral appositional rates (MARs) ( Figure 2B ). Osteoclast numbers were calculated using TRAP staining ( Figure 2D ), and increased osteoclast number/bone surface (Oc.N/BS) of the cold group was observed at 14 days ( Figure 2E ), which was recovered even lower than the baseline after 28 days of exposure. Levels of P1NP in the cold group, a marker of bone formation, were higher at 14 days and lower at 28 days than that in the control group ( Figure 2F ). CTX-1 was detected in the same way to confirm bone resorption ability. Expression levels of CTX-1 in the cold group were higher than that in the control group and reached their highest level at 14 days ( Figure 2G ). In addition, cold promoted the expression of IL-6, one of the most important factors induced by BAT (26), at 1 and 14 days ( Figure 2H ).

To observe the effect of cold on BAT, BAT in the interscapular region of mice was scanned and calculated. As shown in Figures 3A, C , brown adipose volume was stimulated by cold temperature, which increased in 14 days and was maintained until 28 days. Concordantly, the lean mass in both groups was higher on day 14 than on day 1, while the cold group exhibited a higher lean mass on day 28 ( Figure 3D ). Other results calculated by DXA, such as body weight, fat mass, %lean and %fat, did not change significantly ( Figures 3B , E – G ).

Brown adipose tissue cells in the stromal vascular fraction (BAT svf, BS) were cultivated in vitro. Oil red O staining was performed to test whether BAT SVF cells were differentiated into brown adipocytes (BB) ( Figure 4A ), which produce scattered lipid droplets. Western blot analysis showed that UCP1 expression was increased in the BB group ( Figure 4B ). Furthermore, brown adipocyte gene markers were dramatically increased in the differentiated group ( Figure 4C ).

Brown adipocyte conditional medium (BAT CM) was added to the osteogenesis differentiation medium of BMSCs. Staining assays showed that BAT CM strengthened ALP and alizarin red staining in BMSCs ( Figure 4D ). Semiquantitative analysis of alizarin red is shown in Figure 4E , which confirmed improved bone formation in the BAT CM group. Moreover, Alp, Ocn, and Runx2 levels were upregulated by BAT CM compared to the corresponding control treatments ( Figure 4F ).

To determine the effects of BAT CM on osteoclastogenesis, we added BAT CM to the differentiation medium of BMMs, and TRAP staining ( Figure 4G ) revealed reduced multinuclear osteoclastogenesis after 6 days of cultivation ( Figure 4I ). Meanwhile, resorption ability was assessed using the osteoassay stripwell, and the resorption area was smaller in the BAT CM group ( Figure 4J ). Additionally, Ctsk, Dcst, and Traf6 were downregulated by BAT CM ( Figure 4H ).