The approvals of the animals and experimental procedures obtained access to the Ethics Committee of the First Affiliated Hospital of Guangzhou University of Chinese Medicine (GZUCM) (approval number: 20130425). Rats were fed in specific pathogen-free standard conditions which conform to the Guide for the Care and Use of Laboratory Animals, published by Institute of Laboratory Animal Resources (U.S.).

ZGW was purchased from the Beijing Tongrentang Pharmaceutical Factory (State Food and Drug Administration approval number: Z11020735). It comprised eight herbal medicines: Radix Rehmanniae (24 g), Semen Cuscutae (12 g), Radix Achyranthis Bidentatae (9 g), Carapax et Plastrum Testudinis (12 g), Cornu Cervi Pantotrichum (12g), Rhizoma Dioscoreae (12 g), Fructus Corni (12 g), and Fructus Lycii (12 g). Before the experiment, the drug was dissolved in normal saline, and a magnetic stirrer was stirred to help in dissolving. After dissolution, it was used for the animal experiment. For the cell experiment, ZGW solutions were filtered and sterilized at a 0.22-μm pore size for later use.

Three-month-old female Sprague-Dawley rats (n = 54) were obtained from the animal experiment center of Southern Medical University (Guangzhou, China), housed in the First Affiliated Hospital of GZUCM (SYXK [Yue] 2018-0092). The weight of rats at baseline was 220 ± 30 g. During the entire experiment, the rats were provided with food and water adequately.

After 1 week of adaptive feeding, the rats were randomly divided into three groups: the control group (CON) received vehicle intervention to eliminate systematic deviations caused by intervention factors; the dexamethasone (DEXA) group received subcutaneous injection of dexamethasone 0.6 mg/kg body weight twice a week for 3 consecutive months; and the ZGW group (ZGW) received oral administration of ZGW (1.62 g/kg), once a day for 3 consecutive months. The dose selection of ZGW was based on the recommended daily dose for humans (18 g/kg), and the equivalent conversion between animals and humans was based on the body surface area. This method of calculating the rat oral administration dose is similar to previous studies (9, 20).

Rats in the CON, DEXA, and ZGW groups underwent euthanasia for the next experimental analysis at the end of the fourth week (M1, N = 6/group), eighth week (M2, N = 6/group), and twelfth week (M3, N = 6/group). Lumbar 2 (L2) specimens were separated for microcomputed tomography (micro-CT) and biomechanical analysis. L4 specimens were fixed with 4% paraformaldehyde for the next HE staining. Before assessing the AKP activity, blood samples were collected, centrifuged for serum, and stored at -80°C. Transmission electron microscopy was employed to detect autophagolysosomes from osteoblast L5 specimens. L6 specimens were stored at -80°C for real-time qPCR.

The software (μCT80 Evaluation Program v. 6.5-1; Scanco Medical, Brüttisellen, Switzerland) provided by the micro-CT system was employed to quantify the three-dimensional image and parameters of the L2 vertebral bodies, which was similar to previous studies (21, 22). The scanning conditions were 55 kV and 80 μA, and the spatial resolution was 14 μm. Vertebral cancellous bone was selected as the volume of interest. The microstructural parameters of cancellous bone were analyzed using micro-CT analysis software to determine the relative bone volume (BV/TV, %), structural model index (SMI), trabecular number (Tb. N), thickness (Tb. Th), and separation (Tb.Sp), and volumetric bone mineral density (vBMD). Three-dimensional images are obtained by reconstruction analysis.

A material testing machine (ElectroPuls E1000 Testing System; Instron Corp., Norwood, MA) was employed for bone biomechanics analysis. Both end plates of vertebral bodies and their appendages were excised to form a central cylinder with parallel planes at both ends. The vertebrae were then examined along the longitudinal axis in this machine at a speed of 1 mm/min. Then the bone biomechanics parameters, including compressive stiffness, strength, and displacement as well as energy absorption capacity, were analyzed.

L4 vertebral bodies were fixed with 4% paraformaldehyde for 48 h and then decalcified with 10% EDTA for 21–28 days. The samples were dehydrated with graded alcohol, embedded in paraffin, and then cut into 5-μm slices along the coronal plane using a paraffin microtome. Sections were stained with HE and observed under a microscope (BX53, Olympus Corporation, Tokyo, Japan).

Rat serum AKP activity was analyzed using a kit (Nanjing Jiancheng Biological Products Co., Ltd., Nanjing, China). The serum alkaline phosphatase (AKP) levels in rats were determined using a visible spectrophotometer (Thermo Fisher, Waltham, MA, USA).

The L5 samples were fixed with 0.2% glutaraldehyde at room temperature for 2 h, in 1% osmium tetroxide water for 1 h, and then stained in 2% uranyl acetate water for 1 h in the dark place. After dehydration through an alcohol gradient, the samples were embedded and sliced into 80-nm slices. The sections were stained with uranyl acetate and lead citrate and then observed with a transmission electron microscope (Philips CM, Amsterdam, Netherlands).

The expression levels of let-7f, autophagy-associated genes (mTORC1, Beclin-1, ATG12, ATG5, and LC3), Runx2, and Cathepsin K (CTSK) were evaluated by RT-qPCR. Total RNA was extracted from L6 vertebrae or BMSCs using the TRIzol Reagent (Thermo, USA). The PrimeScript™ RT Master Mix Kit (Takara, Shiga, Japan) was used to reverse transcribe RNA to cDNA. Quantitative analysis was performed under a Bio-Rad qPCR instrument using the SYBR Premix Ex Taq™ II Kit (Takara, Japan). Actin or U6 was used as the internal standard for mRNA and miRNA, respectively. The relative expression of each gene was quantified using the 2^−ΔΔCt method. The primer sequences are shown in Supplementary Table 1.

For osteoblast formation, we used rat bone marrow-derived mesenchymal stem cells (BMSCs). The separation and culture of rat BMSCs were performed as previously described (4). Briefly, we selected 7-day-old mice, dissected the femur in a sterile manner, and used a syringe needle to flush out bone marrow solutions. After centrifugation, the cell solutions were resuspended in α-MEM containing 10% fetal bovine serum (all from Gibco, Grand Island, NY, USA). The cells were cultured at 37°C in an incubator containing 5% CO₂. Osteogenic differentiation was induced in rat BMSC osteogenic differentiation medium (Cyagen Biosciences, Santa Clara, CA, USA). After 7 days’ osteogenic induction, osteoblasts were fixed with 4% paraformaldehyde and stained using an BCIP/NBT Alkaline Phosphatase (ALP) kit (Beyotime, Shanghai, China).

A PMIR-Report luciferase vector containing a let-7f binding site in the ULK2 3′-UTR was constructed. A mutant plasmid was used to measure binding specificity. Next, pGL3 luciferase reporter vectors were inserted into the let-7f promoter containing the ULK2 binding site. Then, we transfected luciferase vectors into osteoblasts. Luciferase activity was examined using the Luciferase Reporter Assay Kit (Promega, Madison, WI, USA).

BMSCs were lysed with RIPA lysis buffer (Beyotime) containing 1% protease inhibitor and 1% PMSF phenylmethanesulfonyl fluoride (PMSF) for 30 min to inhibit protein degradation and centrifuged at 12,000 rpm for 10 min, and the supernatant was transferred to a new tube. The BCA Protein Assay Kit (Beyotime) was used to determine the protein concentration; the Gel Preparation Kit (Beyotime) was used to perform SDS-PAGE. The separated proteins were transferred to polyvinylidene fluoride (PVDF) membranes (Millipore) and blocked with QuickBlock™ Western Primary Dilution antibodies (Beyotime) overnight at 4°C temperature. Anti-β-actin primary antibody (1:5,000, mouse, ab6276, Abcam, Cambridge, MA, USA), ULK2 (1:1,000; Rabbit; DF9890; Affinity, Cincinnati, OH, USA), ATG5 (1:1,000, rabbit, #12994, Cell Signaling Technology (CST), Danvers, MA, USA), ATG12 (1:1,000, rabbit, #4180, CST), Beclin1 (1:1000, rabbit, #3495, CST), and LC3 (1:1000, rabbit, #4108, CST) were used as the primary antibodies following incubation with the membranes for 12 h at 4°C. After washing the membranes with TBST sufficiently, the membranes were incubated with the corresponding secondary antibody for 2 h at room temperature and then washed three times with TBST. The protein level was determined by BeyoECL Moon enhanced chemiluminescence (Beyotime).

The data were analyzed with SPSS 19.0 software. For quantitative analysis, all results are expressed as mean ± standard deviations (SD). Statistical significance was evaluated by one-way analysis of variance (ANOVA) with the Student–Newman–Keuls (SNK) test for post-hoc analysis. Two-sided p-values < 0.05 were considered statistically significant.

Micro-CT showed that there was a small amount of trabecular bone thinning and damage in the DEXA group at each time point, while the ZGW group had more trabecular bone and more dense distribution of trabecular bone ( Figure 1A). Microarchitecture data of L2 cancellous bone of vertebrae examined by micro-CT displayed that the DEXA group showed low BV/TV, Tb.N, Tb.Th, and vBMD and significantly high SMI and Tb.Sp at all the time points compared to those in the CON group. Compared with the DEXA group, the ZGW group displayed increased BV/TV, Tb.N, Tb.Th, and vBMD at each time point and decreased SMI and Tb.Sp at each time point ( Figure 1B).

Compressive strength, compressive stiffness, compressive displacement, and energy absorption capacity were employed to evaluate the bone strength of GIOP rats. Compared to the CON group, the DEXA group displayed reductions in compressive strength at each time point and energy absorption capacity at M2 and M3. Meanwhile, the DEXA group (compared to the CON group) showed a decreased trend in compressive stiffness and compressive displacement at all the time points, although no significant differences were observed. Compared with the DEXA group, the ZGW group exhibited significantly increased compressive strength at each time point, high compressive stiffness, and energy absorption capacity at M3. Simultaneously, compared to the DEXA group, ZGW showed an increased trend in compressive stiffness at M1 and M2, compressive displacement at all the time points, and energy absorption capacity at M1 and M2, although the differences were not significant ( Figure 2).

HE staining was employed to assess the morphological changes in the L4 trabeculae of the various groups. At each time point, the DEXA group displayed increased trabecular spacing, trabecular absorption, perforation, fracture, and poor trabecular continuity compared to the CON group. Compared with the DEXA group, the ZGW group manifested improvements in vertebral trabecular bone ( Figure 3).

ELISA was employed to detect the blood serum AKP activity of each group at different time points. Compared to the CON group, the DEXA group exhibited significantly decreased serum levels of AKP at M2 and M3. Compared with the DEXA group, the ZGW group manifested significantly increased serum levels of AKP at M2 and M3 ( Figure 4).

By transmission electron microscopic observation, at each time point, the DEXA group exhibited more cytosolic autophagosomes, which were abundant in the osteoblasts, than in the CON group. Compared with the DEXA group, the ZGW group showed a significantly opposite trend in autophagic characteristics ( Figure 5).

The expressions of let-7f, autophagy-associated genes (mTORC1, Beclin-1, ATG12, ATG5, and LC3), Runx2, and CTSK were detected at the mRNA level to explore how ZGW might function. After DEXA intervention, let-7f expression was downregulated at M2 and M3. Furthermore, the DEXA group exhibited a significantly downregulated mTORC1 (a negative autophagy regulatory gene) mRNA expression at all the time points, significantly upregulated Beclin-1, ATG5, and LC3 mRNA expressions at all the time points, and ATG12 mRNA expression at M2 and M3. Meanwhile, bone formation-related gene Runx2 expression was downregulated at M3, whereas bone resorption-related gene CTSK mRNA expression was significantly upregulated at M2 and M3 ( Figure 6).

After ZGW intervention, let-7f expression was significantly upregulated at M2 and M3 and was upregulated to some extent at M1 but did not differ significantly from that of the DEXA group. Compared to the DEXA group, the ZGW group exhibited a significantly downregulated ATG5 expression at all the time points, Beclin-1 expression at M2 and M3, ATG12 mRNA expression at M2, and LC3 mRNA expression at M3. Simultaneously, Runx2 expression was significantly upregulated at M3, whereas CTSK expression was significantly decreased at M3 ( Figure 6).

To explore the regulatory effect of ZGW on let-7f in osteogenic differentiation of BMSCs, we conducted gain-and-loss-of-function experiments by transfecting let-7f mimics of the inhibitor into BMSCs. The effect of overexpression or silencing was verified by qPCR ( Supplementary Figure 1A). CCK8 was utilized to evaluate the effect of ZGW on BMSC proliferation. 10 μg/ml was the maximum concentration that did not affect BMSC proliferation, so we used 10 μg/ml ZGW as the optimum concentration in the next experiment ( Supplementary Figure 1B). By ALP staining, we indicated that let-7f mimics could promote BMSC osteogenic differentiation, whereas the let-7f inhibitor suppressed BMSC osteogenic differentiation. After ZGW intervention, ZGW partially reversed the decline in osteogenic differentiation induced via let-7f silencing ( Figure 7).

To examine whether let-7f binds to the 3′UTR of ULK2 (a positive autophagy regulatory gene), we conducted luciferase reporter assays by establishing a whole length of the ULK2 3′UTR, which was considered as the ULK2 wild-type (WT) 3′UTR. The ULK2 mutant (Mut) 3′UTR was established by binding sequence mutation. Following let-7f mimic transfection into BMSCs, the ULK2 relative luciferase activity was suppressed compared with that in WT (Figure 8). However, in the mutant (Mut) groups, the activity of ULK2 relative luciferase has no significant differences among the different groups. These data indicated let-7f in the regulation of ULK2 3′UTR ( Figure 8).

To further examine the effect of ZGW on let-7f in osteoblast differentiation of BMSCs, we employed Western blot to verify the relationship among let-7f, autophagy, and ZGW. We indicated that let-7f mimics could repress the autophagy-positive regulatory proteins (ULK2, ATG5, ATG12, Beclin1, and LC3) compared with the control mimic group. The let-7f inhibitor could increase the autophagy proteins, especially ATG12, Beclin1, and LC3, compared to the control inhibitor group; however, ZGW partially reversed let-7f silencing-induced increased autophagy ( Figure 9).