2.1 Chemicals and reagents

EC and EF were purchased from Bozhou Pharmaceutical Company, ltd. (Anhui, China) and were morphologically authenticated by Professor Xue. The voucher specimens were also deposited in the herbarium of the Inner Mongolia Medical University (#20190822 and #20190823). Chloral hydrate (CCl₃CH(OH)₂,), Gushukang granules, and saline (NaCl) were obtained from Ze Sheng Biotechnology Company (Hohhot, China). Methanol (CH₃OH) and formic acid (HCOOH) (MS grade) was purchased from Fisher Scientific Corporation (Loughborough, UK). Ultra-high-purity water (18.2 MΩ) was acquired from an ALH-600-U purification system (Chongqing, China). Enzyme-linked immunosorbent assay kits for osteocalcin, serum phosphorus, and alkaline phosphatase (ALP) were obtained from Hua Lian Biotechnology Institute (Wuhan, China).

2.2 Plant extraction preparation and determination before animal experiments

The final preparation of EC and EF extracts was carried out as described in our previous paper (Guo et al., 2020a). EC and EF were crushed to fine powder (over 40 mesh screens), and 200 g of the powder was reflux-extracted twice with 1.88 L of 64 % methanol for 2 h. After filtration, the filtrate was collected and concentrated. Methanol was removed from the desorption solution by rotary evaporation at 40℃followed by freeze-drying at −80℃. The extraction yields of EC and EF were 18.0 % and 33 %, respectively. The dried powders were stored in a refrigerator at − 20℃ and dissolved in an appropriate amount of distilled water or saline before HPLC-Q-Exactive/MS analysis or animal experiment. Thus, distilled water and saline were selected as solvent control in HPLC-Q-Exactive/MS analysis and animal experiment, respectively.

The fingerprintings of EC and EF in different batches were achieved by high performance liquid chromatography (HPLC) (Supplementary Fig. S1), and several ingredients of EC and EF extracts were identified by high-performance liquid chromatography-quadropole-Exactive-mass spectroscopy (HPLC-Q-Exactive/MS) (Thermo Fisher Scientific Inc., MA, USA). In order to confirm the structure of these compounds more accurately, ten components including chlorogenic acid were verified by comparing the retention time and molecular features with standard substances. The main components in the EC and EF extract samples are listed in Supplementary Table S1. The contents of main components in EC and EF extracts sample are shown in Supplementary Table S2.

2.3 Animal experiments

Twelve-week-old Wister rats, 190–220 g in weight, were purchased from the animal experiment center of the Inner Mongolia Medical University. All rats were kept in a clean area for 1 week to adapt to the environment. Subsequently, they were equally randomly assigned to six groups (six rats per group): a normal group (normal), an ovariectomized group (OVX), a sham group (sham), a Gushukang-treated group (GT), an EC-treated group (EC), and EF-treated group (EF). All rats except those in the normal group and the sham group underwent bilateral ovariectomy. Bilateral laparotomy was carried out in the sham group. Seven days after surgery, all rats were subjected to keratinization experiments involving the vaginal epithelium to evaluate whether the models were successfully established. The animal study was reviewed and approved by the Animal Ethics Committee of Inner Mongolia Medical University (Reference: SCXK2015-0001).

After the OVX model was successful built, all rats were gavaged with treatment once a day: the normal, OVX, and sham group rats were given 2 mL of 0.9 % saline daily; the GT group rats were given Gushukang granules (105.1 mg/kg/d, which was the clinical equivalent dose) as an positive control. The EC group was given 0.36 g/kg/d of EC extract (equivalent to 2.0 g/kg/d of raw medicine), and the EF group was given 3.3 g/kg/d of EF extract (equivalent to 10.0 g/kg/d of raw medicine). After 8 weeks of treatment administration, all rats were fasted 12 h, and then anesthetized by intraperitoneal injection of 10 % chloral hydrate, and blood samples were obtained. The serum was separated by centrifugation at 3,500 rpm for 10 min at 4 °C, and supernatants were stored at − 80 °C. The left tibia was also removed from each animal, and these samples were fixed in formalin to await analysis.

2.4 Analysis of bone mineral density and trabecular microarchitecture

X-ray microtomography was used to analyze bone mineral density (BMD) and trabecular microarchitecture. All tibias were fixed with 70 % ethanol and subjected to X-ray microtomography with an isotropic voxel size of 10 μm. Tomographic images were acquired at an integration time of 250 msec with 500 projections during the full 360° rotation. Three-dimensional reconstructions were generated with the following parameters: smoothing was set to 3; ring artifacts reduction was set to 5; and beam hardening correction was set to 30 %. BMD, trabecular separation (Tb × Sp), trabecular number (Tb × N), percent bone volume (BV/TV), and structure model index (SMI) were all determined by analyzing a specific region of interest that was chosen by setting the same coordinates in the tibia growth plate for each sample.

2.5 Preparation of serum samples

The serum samples were removed from a refrigerator set at − 80 °C and thawed to 4 °C. A 200-µL quantity of serum was spiked into methanol at a volume ratio of 1:3. The mixture was vortex for 3 min, stood for 10 min, and was centrifuged at 4 °C and 13,000 rpm for 10 min to remove protein. The supernatant was transferred to a new centrifuge tube and dried using a vacuum centrifugal concentrator (CVE-3000; EYELA, Tokyo, Japan). Subsequently, the residue was reconstituted in 100 μL of methanol and centrifuged at 13,000 rpm for 10 min at 4 °C before HPLC-Q-Exactive/MS analysis.

The quality control (QC) sample was obtained by mixing equal volumes (10 μL) of each test serum sample, and it was prepared by the same method described above to remove the protein. One QC sample was injected into a HPLC-Q-Exactive/MS system after 10 test samples.

2.6 Metabolomics analysis based on HPLC-Q-Exactive/MS

HPLC-Q-Exactive/MS was used to analysis metabolomics. Analytes extracted from serum samples were separated on a HSS T3 column (Waters Acquity UPLC HSS T3 column; 2.1 × 50 mm, 1.8 μm) at a column temperature of 35 °C. A gradient elution program at the flow rate of 0.4 mL/min was established as follows: 5 % A up to 15 % A (0–1.7 min); 15 % A up to 17 % A (1.7–3.0 min); 17 % A (3.0–3.3 min); 17 % A up to 25 % A (3.3–8.0 min); 25 % A up to 30 % A (8.0–9.7 min); 30 % A up to 35 %A (10.6–14.1 min); 55 % A maintained at 14.1–14.6 min; 55 % A up to 100 % A (14.6–15.1 min); 100 % A maintained at 15.1–17.0 min; 100 % A down to 5 % A (17.0–18.1 min); and maintained at 5 %A to 20 min. The mobile phases were composed of methanol (A) and 1 % formic acid water (B).

Optimized parameters for the Q-Exactive/MS were as follows: Spray voltage was set to 4 kV for the positive ion mode and 3.2 kV for the negative ion mode. The sheath gas flow rate was 40 L/min for the positive ion mode and 35 L/min for the negative ion mode. The auxiliary gas flow was 2 L/min for both ion mode. The auxiliary gas temperature was set 350℃ and the capillary temperature was 300 V for both ion modes. Data were collected in full scan/dd ms²mode, and the mass range was 100–1,100 m/z.

2.7 Metabolomics data processing

Total ion chromatograms were pre-processed with Compound Discover software (version 3.0, Thermo Fisher Scientific Inc.). After peak alignment, peak filter, peak extraction, and automatic integration, an Excel table including accurate mass data, retention time, peak area, ChemSpider ( https://www.chemspider.com ) prediction compound results, and mzCloud ( https://www.mzcloud.org/) prediction compound results were formulated. The raw data were listed in Supplemental Data Set_5 (acquisition data in positive mode) and Supplemental Data Set_6 (acquisition data in negative mode). These datasets were analyzed with pattern-recognition methods using the software package simca-p (version 14.1, Umetrics, Umea, Sweden). The response variables were scaled and centered to Pareto variance to normalize the skewed distributions. Moreover, features with>50 % of values missing were removed. To eliminate the effect of intersubject variability and to identify endogenous metabolites that contributed significantly to the classification, linear combinations of X variables orthogonal to the Y vector were removed by orthogonal projections to latent structures (OPLS)–discriminant analysis (DA). Cross-validation was used to estimate the robustness and predict abilities of established model, which was validated in more detail by 200 permutation tests. The metabolites with variable importance in the projection (VIP) scores > 1.0 in the OPLS-DA and a P value < 0.05 by Student’s t test were examined and selected for discrimination power according to multiple statistical criteria. Identification of metabolites was achieved by comparing molecular features (accurate mass and MS/MS spectra) with metabolomics libraries, such as the Human Metabolome Database (HMDB) (https://www.hmdb.ca/) and mzCloud, or with the standard substances.

2.8 Statistical analysis

Data relating to osteocalcin, serum phosphorus, ALP, BMD,Tb × Sp, Tb × N, BV/TV, and SMI were described as the means ± standard deviations (SDs). The comparisons between six groups were made using one-way ANOVA. Mann-Whitney test was used to determine the significance of differences between two groups (Vetter and Mascha, 2018), which was performed with IBM SPSS 22.0 software (Chicago, USA), and a P value < 0.05 was regarded as statistically significant.

3.1 Composition differs slightly between EC and EF

Analysis of the main components types of EC and EF were conducted to test the hypothesis of alternative herbal sources. The chemical profiling of EC and EF were performed with HPLC-Q-Exactive MS using a total of 27 compounds characterized from EC extracts and 24 compounds detected from EF (Supplementary Table S1), 10 of which were unambiguously identified using authentic standards. The fingerprints of EC and EF samples were obtained and reported in our previous study. In that study, the common peaks in 23 batches of EC and EF samples were defined by the Similarity Evaluation System software (version 2004A; China) and included chlorogenic acid, geniposidic acid, genipioside, genipin, and pinoresinol diglucoside. The concentrations of geniposidic acid, geniposide, genipin, and pinoresinol diglucoside in EC were approximately-five times those in EF. The anti-osteoporosis activities of ten main components contain both in the EC and EF were also tested according to previous report (Zheng, et al. 2019). As shown in Fig. S2-S3, the formation of osteoclasts were inhibited following the treatment of chlorogenic acid, PDG, geniposide, deacetylasperulosidic acid, geniposidic acid, eucommiol and cryptochlorogenic acid. The number of osteoclasts and TRAP positive multinuclear cells were also decreased after these compounds treatment. Considering that the component determines the effect (You et al., 2018.), a dose of EF that is five times the dose of EC would eliminate any difference in efficacy resulting from ingredient contents.

3.2 EC and EF have the same influences on serum markers of ovariectomized rats

To evaluate if EF could be an alternative to EC, we surgically ovariectomized 12-week-old rats, a method that represents the most commonly used model for osteoporosis (Chevalier et al., 2020), and we fed them intragastrically with EC and EF extracts. Ovariectomy led to elevated expressions of osteocalcin, serum phosphorus, and ALP (Fig. 1A-1C) compared with the control group. Conversely, GT decreased the levels of osteocalcin, serum phosphorus, and ALP compared with the OVX group (Fig. 1A-1C). GT, as a traditional Chinese medicine (Wang et al., 2020) for osteoporosis, was positive control drug used to evaluate the efficacy of EC and EF in osteoporosis. Considerable differences were detected in the content of osteocalcin, serum phosphorus, and ALP between the GT group and the OVX group. EC and EF extracts also prevented changes to the levels of osteocalcin, serum phosphorus, and ALP caused by ovariectomy (Fig. 1A-1C). No differences were found in these serum index expressions between the EC and EF group, which suggests good substitutability between EC and EF exist with regard to regulation of these serum markers.

Close

Fig. 1

Effects of Eucommia cortex (EC) and Eucommia folium (EF) on biochemical markers levels of bone metabolism in ovariectomized (OVX) rats: (A) osteocalcin; (B) serum phosphorus; (C) ALP. Values are expressed as the mean ± standard deviation (SD); n = 6. *p < 0.05, ^**p < 0.01 compared with the OVX group. ^#p < 0.05, ^##p < 0.01 compared with the sham group.

Export to PPT

3.3 EC and EF have the same therapeutic effects on trabecular microarchitecture of ovariectomized rats

Estrogen depletion leads to a decrease in BMD and alternation in the trabecular microarchitecture, which can increase the incidence of fracture (Kreipke et al., 2016). We evaluated the effects of estrogen deficiency on rats, using female OVX rats. This surgery caused a small, thin, and sparse trabecular region in the tibia of OVX rats (Fig. 2A)—a result that was consistent with a previous report (Kinney et al., 1995). Conversely, in the tibia of sham and normal rats, the trabecular regions were neatly and densely arranged (Fig. 2A). Ovariectomy led to a decrease in BMD, Tb × N, and BV/TV and an increase in Tb × Sp and SMI (Fig. 2B-2F). EC and EF treatment prevented the trabecular region loss caused by ovariectomy (Fig. 2A), as assessed by computed tomography. This finding was consistent with the elevation of BMD, Tb × N, and BV/TV (Fig. 2B, 2D, 2E) in the EC- and EF-treated OVX rats compared with the OVX controls. Ovariectomy-induced increases in the Tb × Sp and SMI of tibias was also prevented in the EC- and EF-treated OVX rats (Fig. 2C, 2F). Amelioration of BMD, Tb × N, BV/TV, Tb × Sp, and SMI changes in EC-treated OVX rats was similar to changes seen in EF-treated OVX rats, indicating the same therapeutic effect on trabecular microarchitecture.

Close

Fig. 2

The region of interest (ROI) image and bone parameters analysis in tibia in different rats: (A) ROI image of sham, normal, ovariectomized (OVX), Gushukang-treated (GT), Eucommia cortex–treated (EC), and Eucommia folium–treated (EF) groups. (B) Comparison of bone mineral density (BMD) in six groups. (C) Comparison of trabecular number (Tb × N) in six groups. (D) Comparison of trabecular separation (Tb × Sp) in six groups. (E) Comparison of percent bone volume (BV/TV) in six groups. (F) Comparison of structure model index (SMI) in six groups. Values were expressed as the mean ± standard deviation (SD); n = 6. *p < 0.05, ^**p < 0.01 compared with the OVX group. ^# p < 0.05, ^##p < 0.01 compared with the sham group.

Export to PPT

3.4 Key endogenous components of EC and EF have gradual and similar changes in metabolomics studies

3.4.3

3.4.3 Metabolic regulations of EC and EF trended the same in tryptophan and 46 other metabolites

As shown in Table 1, the metabolic regulations of EC and EF were same in tryptophan and in 46 other metabolites. A heatmap was constructed after identification of the differential metabolites (Fig. 3) to visualize the main changes of metabolites in different groups; color differences were used to exhibit the metabolite alterations occurring in different groups. For example, the level of platelet-activating factor in OVX group was much higher than that of the sham group, so the color of platelet-activating factor in the OVX group on the heat map is red. Conversely, the color of platelet-activating factor in the EC and EF groups is blue, because the intensity of platelet-activating factor was down regulated in the EC and EF groups. The comparison of the top 10 VIP values for differential metabolites between the OVX group and the sham group are shown in Fig. 4; among 10 predicted metabolites, seven metabolites were regulated to a normal state in the serum of EC- and EF-treated rats. Deeper analysis of the differentially expressed metabolites suggested that nine of the top 10 differential markers in the EC and EF groups were not different, which may be compelling evidence for the similarity of the metabolite alterations by EC and EF on OVX models.

Close

Fig. 3

Heatmap based on the relative levels of top ten variable importance in the projection (VIP) value differential metabolites in different groups. Color key indicates metabolite expression: red: upregulated, blue: downregulated.

Export to PPT

Close

Fig. 4

Comparison of the top ten variable importance in the projection (VIP) value differential metabolites in six groups. Values areexpressed as the mean ± standard deviation(SD); n = 6. *p < 0.05, ^**p < 0.01 compared with the ovariectomized (OVX) group. ^#p < 0.05, ^##p < 0.01 compared with the sham group. ^Δp < 0.05, ^ΔΔp < 0.01 compared with the Eucommia cortex(EC) group.

Export to PPT

The differences between the EC and EF of l-Norleucine and l-Phenylalanine may be related to the difference ingredients contained in the two parts. According to the results of this experiment, EF contains rutin, while EC does not. As previous reported (Lee et al., 2020; Xiao et al., 2019), rutin could prevent the ovariectomy-induced osteoporosis in rats, and inhibit osteoporosis. Amino acid metabolism (including l-Norleucine and l-Phenylalanine) plays an important role in the pathogenesis of osteoporosis (Ohata et al., 1970). Therefore, the presence of rutin has a certain regulatory effect on amino acid content, which in turn causes differences in amino acid content after administration of EC and EF.

3.5 Main anti-osteoporotic metabolic pathway affected similarly by EC and EF

According to the methods mentioned above, the differential metabolites between the EC and OVX groups and those between the EF and OVX groups were screened out and identified. The score plots and S-plot of the OPLS-DA model were shown in Supplementary Material Fig. S5 and Fig. S6. Functional analysis showed that the differential metabolites between the EC and OVX groups were primarily involved in phenylalanine metabolism; phenylalanine, tyrosine, and tryptophan biosynthesis and more. However, the most significantly different pathways between the EF and OVX groups were phenylalanine metabolism; phenylalanine, tyrosine, and tryptophan biosynthesis. Detailed results of the pathway analysis are listed in Supplemental Data Set_7 and Supplemental Data Set_8, and a summary is shown in Fig. 5. The pathway with − log (P) values > 2 and a path impact threshold > 0.2 was identified as the main metabolic pathway. This pathway was affected by EC and EF and involved phenylalanine metabolism; phenylalanine, tyrosine, and tryptophan biosynthesis; ether lipid metabolism. And the main regulated pathways affected by the two groups are basically the same. These main regulated pathways may be the main contributors to the anti-osteoporosis effect of EC and EF. We speculate that the mechanisms of EC and EF in the treatment of osteoporosis may be similar.

Close

Fig. 5

Signaling pathway analysis based on the potential biomarkers of Eucommia cortex (EC) (A) and Eucommia folium (EF) (B) in the treatment of osteoporosis. The size and color of each circle are based on the pathway impact value and the p value, respectively.

Export to PPT