Female Balb/c mice (n = 100, weighing 18–20 g each) and 6-week-old CD-1 mice (n = 40, half male and female) were obtained from Beijing Vital River Laboratory Animal Technology Co., Ltd. (Certificate number: SCXK (Beijing) 2017-0033). Animals were reared in specific-pathogen-free facilities under controlled light-dark exposure (12-h each of dark and light), temperatures (20–24°C), and humidity (45%–70%). Ad libitum food and water were provided. Animals were handled strictly adhering to the Guidelines for the Care and Use of Laboratory Animals. Ethical approval was obtained from the Ethics Committee of Xi Yuan Hospital of China Academy of Chinese Medical Sciences.

The herbs of BCD were purchased from Xiyuan Hospital of China Academy of Chinese Medical Science [supplied by Hebei Baicaokang Pharmaeutical Co., Ltd. (Heibei, China, Drug GMP certificate: HE20150001, Drug Manufacturing Certificate: HE20190064)]. BCD component herbs were all verified and conformed to the standard of quality in the 2015 Chinese Pharmacopoeia. Herbs were authenticated based on the following parameters: thin-layer chromatography, chemical, microscopic, and morphological identification, as well as detection of heavy metal and deleterious elements. A pharmaceutical company [Beijing Tcmages Pharmaceutical Co., Ltd. (Beijing, China, Drug GMP certificate: BJ20180373, Drug Manufacturing Certificate: BJ20180032)] was enlisted to produce BCD granules made out of a concentrated mixture of the components presented in Table 1.

BCD granule stability and quality were evaluated utilizing a high-performance liquid chromatography (HPLC) system (Agilent 1100, USA) equipped with a Waters Xbridge-C18 analytical column (4.6 × 150 mm, 5 μm). Mobile phase A indicated the acetonitrile organic phase, while mobile phase B indicated the ultrapure water (containing 0.1% phosphoric acid) aqueous phase. Mobile phases were altered based on the following: 0–5 min, 10% (phase A); 5–35 min, 10%–20% (phase A); 35–50 min, 20%–30% (phase A); 50–55 min, 30%–37% (phase A), 55–60 min, 37%–10% (phase A). The flow rate was 1.0 ml/min and the UV spectrum was calibrated to 230 nm. Specnuezhenide (CAS: 39011-92-2), cyasterone (CAS: 17086-76-9), ferulic acid (CAS: 537-98-4) and isorhamnetin-3-O-neohesperidoside (CAS: 55033-90-4) were procured from National Institutes for Food and Drug Control (Beijing, China), and hyperoside (CAS: 482-36-0), asperosaponin VI (CAS: 39524-08-8), peoniflorin (CAS: 23180-57-6) and typhaneoside (CAS: 104472-68-6) were obtained from Shanghai Standard Technology Co., Ltd. (Shanghai, China). The detection wavelength of BCD was identified based on retention time compared with reference substance.

The chemical constituents of TWP tablets (Grand Pharmaceutical Huangshi Feiyun Pharmaceutical Co., Ltd., Lot NO. Z42021212) were normalized and evaluated as Qu L.’s research (Qu et al., 2015). TWP, BCD, medroxyprogesterone acetate (Zhejiang Xianju Pharmaceutical Co., Ltd., Lot NO. H33020715), and estradiol valerate (Bayer Healthcare Co., Ltd., Lot NO. J20171038) were dissolved in distilled water containing 0.25% Tween 80 and 0.2% carboxymethyl cellulose sodium. Final concentration of TWP was 4 mg/ml, BCD was 0.27 g/ml, estradiol valerate was 0.015 mg/ml, and medroxyprogesterone was 0.09 mg/ml respectively. A 4°C refrigerator was used to store these compounds until further used.

Mice were allowed 5 days of adaptive diets and has their estrous cycles continuously observed for 14 days. Eighty female mice noted to have relatively regular estrous cycles (5–6 days) were selected, with 20 mice each designated in the hormone replacement therapy (HRT), BCD, Model, and Blank groups. In accordance with previous literature and our initial modeling experiments, mice in the HRT, BCD, and Model groups were intragastrically fed 40 mg/kg/d TWP suspension every morning at 8 a.m. for 30 days to induce POI (Lu et al., 2014; Huang et al., 2018; Yuan et al., 2019). The Blank group underwent the same procedure but with normal saline (NS) given instead. Before this experiment, we carried out the concentration of BCD screening research. Based on these findings, an oral BCD suspension at a dose of 2.67 g/kg/d was administered to the mice in the BCD group at 4 p.m. every afternoon during developing the POI mouse model. At the same time, the mice in the HRT group were treated with 4 days of oral 0.15 mg/kg/d estradiol valerate suspension, followed by the addition of 0.91 mg/kg/d medroxyprogesterone acetate suspension daily on the fourth day onwards, with one rest day from hormonal treatments every 4 days. The Blank and Model groups underwent the same procedure but received the same volume of NS instead (Figure 1A).

At the end of the experimental period, all mice models were starved overnight before being weighed and anesthetized the following morning with intraperitoneal 1% pentobarbital sodium (50 mg/kg) injections. Prior to euthanizing the animals, we extracted serum samples for further hormonal ELISA tests. The thymus, lung, cardiac, liver, spleen, kidney, uterus and ovary were then dissected to determine the organ index (organ index = organ weight/body weight × 100%). One ovary from each mouse was frozen in liquid nitrogen and stored at −80°C for further analysis while the other ovary was fixed with 4% paraformaldehyde (PFA) for histology and immunostaining.

Ten mice chosen at random from every group had their estrous cycle checked daily and body weights monitored every 4 days during the experimental period. Estrous cycle staging was done by vaginal smear cytological examination every morning between 9 a.m. and 10 a.m. Smears were obtained by flushing 20 µl of sterile NS into the vaginal cavity, spreading the flushed fluid on a glass slide, and fixing the samples with 95% ethanol for 30 min. The Wright-Giemsa Stain solution (Solarbio Science & Technology Co., Ltd., Beijing, China) was used to stain the slides, which were then rinsed with running water, air-dried, and then observed using a light microscope at ×100 magnification.

Hormonal assays to determine the levels of serum anti-mullerian hormone (AMH, CUSABIO, CSB-E13156), inhibin-B (INH-B, CUSABIO, CSB-E08151), estradiol (E₂, CUSABIO, CSB-E05109), luteinizing hormone (LH, CUSABIO, CSB-E12770), and follicle stimulating hormone (FSH, CUSABIO, CSB-E06871) were evaluated using mouse-specific ELISA kits strictly following instructions provided by the manufacturer. The sensitivity of these kits was defined as the lowest mouse concentration that could be differentiated from zero and the CV% was less than 15%. Results were based on the 450 nm absorbance.

Histopathological examination was carried out on ovarian and uterine samples fixed in 4% PFA, dehydrated with ethanol and cleansed with xylene prior to being used to produce 5 μm slices. Samples were rehydrated before being treated with hematoxylin and eosin (HE) for further morphological observation and quantification of ovarian follicles. Two independent investigators quantified follicles by counting the number of follicles on whole ovarian three serial sections. Primordial follicles were visualized using ×400 magnification while all primary, secondary and antral follicles were observed at ×200 magnification.

Ovarian tissue was homogenized in PBS to produce 10% ovarian homogenate, before being centrifuged for 10 min at 4°C and 3,000 rpm to collect the supernatant. 8-OHdG (CUSABIO, CSB-E10527), MDA (Njjcbio, A003-1-2) and CAT (CUSABIO, CSB-E14190) levels, and the activities of GSH-Px (Njjcbio, A005-1-3) and SOD (CUSABIO, CSB-E08556) in the ovarian homogenate supernatant were performed in accordance with protocols provided by the manufacturer.

Ovarian expression of the Bach 1, Nrf2, Keap 1, and HO-1 genes were evaluated using real-time qPCR. Fifty milligrams of ovarian tissues with TRIzol reagent (Invitrogen) on ice before RNA extraction that was carried out based on instructions provided by the manufacturer. Eight microliter RNA was used for 1% agarose gel electrophoresis. Twenty microliter of TIANScript reverse transcriptase (RT) mix, sequence-specific PCR primers, and RNA were combined for analysis using the Prism 7500 sequence detection system. Final results were evaluated using the ABI Prism 7500 SDS software (ABI7500, USA) and a SYBR FAST qPCR Kit Master Mix (×2) Universal kit (Biosystems). The NCBI-Primer was used to design all primer sequences (shown in Table 2). The 2^−∆∆Ct method was utilized in calculating relative gene expression levels against endogenous control gene beta actin.

Proteins were extracted from the four different mice groups. The cytoplasm and nucleus proteins were isolated using a Nucleus-cytoplasmic protein preparation kit (APPLYGEN, P1200). The BCA Protein Assay Kit (CoWin Biosciences) was used to quantify ovarian lysate protein concentrations. Proteins were separated via electrophoresis and immunoblotted onto polyvinylidenefluoride (PVDF) membranes. A 5% bovine serum albumin solution (BSA) was added to the membranes for an hour before being incubated at 4°C overnight with the following primary antibodies: Bach 1(Abcam, ab49657), Nrf2 (Abcam, ab62352), Keap 1 (Abcam, ab119403), HO-1 (Abcam, ab13248), β-actin (ZS, TA-09) and Histon H3 (all antibodies were at a dilution of 1:1,000 in blocking buffer). The next morning, we added horseradish peroxides (HRP)-conjugated secondary antibody to incubate at room temperature for another 40 min. Signals were detected by enhanced chemiluminescence and imaged with a Gel Image system ver.4.0 (Tanon, China). Protein levels were normalized first before measurement of the bands using the ImageJ 6.0 software program.

Forty 6-week-old CD-1 mice (half male and female) each weighing more than 18 g, were purchased from Beijing Vital River Laboratory Animal Technology Co., Ltd. (Certificate number: SCXK (Beijing) 2017-0033). All the mice were divided equally (n = 20 each) into BCD and control groups randomly after 3 days of adjustable feeding. The night before gavage, mice were fasted and only drank water. The BCD group mice were fed with the maximal concentration and maximal volume of BCD solution daily by intragastric administration, while control group mice underwent the same procedure but with NS instead. All mice were closely observed for 2 weeks for any potential adverse effects. Mice were weighed daily and their average daily feed intakes (ADFI) were recorded. After the 2-week observation period, mice were fasted overnight before they were anesthetized for further evaluation of their internal organs.

The SPSS 16.0 was used to perform all statistical analysis. Results were depicted in terms of mean ± SEM. One-way ANOVA or Fisher’s exact tests were used to analyze statistical differences between groups. p values of <0.05 or <0.01 were indicative of statistical significance.

HPLC-fingerprint was conducted to identify the main ingredients of BCD granules. Eight major compounds were identified and determined in BCD granules, including peoniflorin (RT, 15.050 min), ferulic acid (RT, 19.563 min), typhaneoside (RT, 21.448 min), hyperoside (RT, 23.611 min), isorhamnetin-3-O-neohesperidoside (RT, 24.541 min), specnuezhenide (RT, 29.538 min), cyasterone (RT, 30.525 min), and asperosaponin VI (RT, 52.460 min). The representative chromatography was depicted in Figure 1B.

Before TWP treatment, all mice across the four groups had approximately equal weights. Total mice body weights after TWP treatment demonstrated a downward trend in the Model, BCD, and HRT groups. However, mice weights eventually increased progressively and no significant differences were seen in all four groups (Figure 2A). Weight gain did not differ significantly across all four groups at the end of the experiment (Figure 2B). Ratios of the thymus, lung, cardiac, liver, spleen, kidney, and uterus to body weights also did not differ notably (Figures 2C–J). One exception was the decreased ovarian index in the model group in comparison to the Blank groups (p < 0.05). Conversely, the ovarian index was elevated in the BCD and HRT groups (p < 0.05). The ovarian indices of the BCD and HRT groups were not significantly different (Figure 2K).

Estrous cycle stages were determined based on cell type, number, and morphology (Figure 3A). The estrous cycles of all mice were similar prior to TWP treatment. TWP-exposed mice demonstrated increased estrous cycle disturbances in contrast to mice of the Blank group. BCD treatment restored normal estrous cycles in half of the POI mice models in the BCD group. Estrogen secretion plateaued in the HRT group while those in the Model group continuously displayed abnormal hormonal profiles.

Serum AMH, INHB, and E2 levels of Model group mice were suppressed after TWP treatment (p < 0.01, p < 0.01, p < 0.05), however, these same mice possessed higher FSH level and ratio of FSH to LH in comparison to the Blank group (p < 0.01, p < 0.05). BCD intervention resulted in raised AMH, INHB, and E2 levels in comparison to the Model group (p < 0.05, p < 0.01, p < 0.01). BCD intervention also resulted in lower FSH level and FSH to LH ratio (p < 0.01, p < 0.05). Conversely, mice in the HRT group displayed increasing levels of AMH and INHB, but decreased FSH level and FSH to LH ratio (p < 0.01, p < 0.05). The differences between INHB and E2 levels of the BCD and HRT groups achieved statistical significance (p < 0.01) (Figures 3B–G).

Figure 4A depicts histological analysis of the ovaries. TWP treatment effectively depleted numbers of dormant primordial follicles in contrast to the Blank group (p < 0.01), while HRT and BCD exposure restored them (p < 0.01) (Figure 4B). Preantral follicle numbers were significantly different between Model and BCD groups (Figure 4C). The Model group, however, had higher atretic follicles, in addition to lower antral follicles and corpus luteum in comparison to other groups (p < 0.01). Mice exposed to BCD had lower atretic follicles (p < 0.01), with higher antral follicle and corpus luteum numbers in contrast to the Model group (p < 0.01, p < 0.05). We also observed variabilities in corpus luteum and atretic follicles between the HRT and BCD groups (p < 0.05) (Figures 4D–F).

Uterine sections from the Blank group also had normal histopathological structures, specifically the adventitia, tunica muscularis, and mucous layer which made up almost two-thirds of uterine wall. There were also large numbers of normal glands. However, after TWP treatment, uterine sections from the Model group showed endometrium obvious thinning with decreased numbers of glandular organs. In comparison with the Model group, the endometrium was significantly thickened whilst a large number of glands were visible in BCD group. The histopathological structure in HRT group was also improved compared to Model group (Figure 4G).

To evaluate the impact BCD in TWP-stimulated oxidative stress, the levels of 8-OHdG, MDA, CAT, GSH-Px, and SOD in ovarian tissues were measured. After TWP treatment, Model group mice demonstrated increased 8-OHdG and MDA levels (p < 0.05, p < 0.01), but lower levels of CAT, GSH-Px and SOD compared to the Blank group (p < 0.01, p < 0.05, p < 0.01). BCD intervention resulted in decreased 8-OHdG and MDA levels, in addition to raised levels of CAT, GSH-Px, and SOD, in contrast to the Model group (p < 0.01). However, the level of MDA was decreased and SOD was increased in HRT group (p < 0.05, p < 0.01). MDA, CAT, and GSH-PX levels between BCD and HRT groups were statistically significant (p < 0.05, p < 0.01, p < 0.01) (Figures 5A–E).

Nrf2/ARE is a key signaling pathway is an important regulator of intracellular redox homeostasis. The current experiments demonstrate that TWP can induce oxidative damage, leading to depleted ovarian reserves. We utilized RT-PCR to assess various molecules involved in the Nrf2/ARE pathway. The results showed low levels of Bach 1 and Nrf2 mRNA expressions in mice of the Blank group. After TWP intervention, the mRNA expressions of Bach 1 and Nrf2 were raised in the Model, BCD and HRT groups in contrast to the Blank group (p < 0.01), although the differences between these three groups were not statistically significant (Figures 6A,B). Conversely, TWP exposure caused Keap 1 mRNA expression to be elevated (p < 0.05) and HO-1 mRNA expressions to be decreased in the Model group compared to the Blank group. Mice exposed to BCD caused lower Keap 1 mRNA expression (p < 0.05) but higher HO-1 mRNA expression in contrast to the Model group (p < 0.05). There were significant differences in terms of HO-1 mRNA expression between the HRT and BCD groups (p < 0.05) (Figures 6C,D).

Total Bach 1 and Nrf2 proteins were not significantly different between the four groups of mice. However, TWP treatment alone decreased cytoplasmic Bach 1 expression, increased nuclear Bach 1 expression (p < 0.05, p < 0.01) and increased the nuclear translocation rate of Bach 1 in the Model raised (p < 0.01). Compared to the Model group, Bach 1 expression in the nucleus and nuclear translocation rate were suppressed in the BCD and HRT groups (p < 0.01). Conversely, in the Model group, there was a rising trend of cytoplasmic Nrf2 expression but decreased nuclear Nrf2 expression (p < 0.05). There was also lower Nrf2 nuclear translocation rates (p < 0.01). After BCD and HRT treatment, Nrf2 expression in nucleus and nuclear translocation rate were raised (p < 0.05, p < 0.01) (Figures 7A–H). Furthermore, in the Model group, Keap 1 expressions were increased and decreased after BCD intervention (p < 0.05) (Figure 7I). HO-1 expression demonstrated a decreasing trend in the Model group, with the variability of HO-1 expressions between BCD and Model group achieving statistical significance (p < 0.05) (Figure 7J).

A preliminary experiment was confirmed that the dose used in the following formal toxicity tests was nonlethal. Mice of BCD group was given 5.50 g/ml BCD solution 40 ml/kg orally three times a day with an interval of 4 h between each dose. The current human clinical dosage of BCD was 2.75 g/kg. Based on the acute toxicity test, we found that the maximum oral tolerance dosage in mice was 660.38 g/kg, which was equivalent to 240 times of the human clinical dosage. During the test, none of the mice died and none developed symptoms suggestive of acute toxicity. We also demonstrated no significant difference between the two groups in terms of body weight and ADFI (Figures 8A,B). Furthermore, visual observation of internal organs revealed remarkably similar organs between the BCD and control groups (Figure 8C). Together, these results indicated that BCD exerted no significant toxicity.