NHQXW was provided by Beijing Tong Ren Tang Chinese Medicine Co. Ltd. NHQXW is composed of 27 herbal medicines including artificial Niuhuang, artificial antelope hom, artificial forest musk abelmosk, panax ginseng radix, atractylodes macrocephala radix, angelicae sinensis radix, paeoniae radix, chuanxiong rhizoma, bupleuri radix, glycyrrhizae radix, smilacis glabrae rhizoma, platycodonis radix, saposhnikoviae radix, asini corii colla, cinnamomi cortex, ophiopogonis radix, ampelopsis radix, borneolum syntheticum, typhae pollen, armeniacae semen amarum, Jujubae fructus, sojae semen praeparatum, dioscoreae rhizoma, scutellariae radix, bubali cornu, and zingiberis rhizoma. NHQXW was made with GMP standard and extracted by methanol for quality control analysis. To obtain a powder for treatment or quality control purposes, NHQXW was blended. The methanol concentration for extraction was set at 85%. The ultrasonic extraction method was applied for active compound dissolution using an ultrasonic bath (Kunshan Ultrasonic Instrument Co. Ltd., China). The condition of ultrasonic extraction was set at 30 min under room temperature with power at 150 W. For quality control, we analyzed the content of bilirubin, a representative component of artificial Niuhuang, by using an oxalic acid solution and methylene chloride under ultrasonic conditions. These extraction processes were repeated in two cycles. After centrifugation, the supernatant was filtered before LC analysis.

Quality control analysis was conducted by using a high-performance liquid chromatography (HPLC) system (ThermoFisher Scientific, UltiMate 3000) according to our previous studies (Du et al., 2020). The compound separation was performed by using the ACE 5 AQ-C18 column (5 μm, 4.6 mm × 250 mm, Advanced Chromatography Technologies). We selected liquiritin, baicalin, paeoniflorin, and bilirubin for quality control study according to Chinese Pharmacopoeia and the reported neurogenic bioactivities (Zhao et al., 2008; Fan et al., 2019; Wang et al., 2021). For detecting liquiritin, baicalin, and paeoniflorin, we used the mobile phase condition optimized as follows: 0–8 min, 19% acetonitrile (ACN); 8–35 min, 19%–50% ACN; and 35–36 min, 50%–100% ACN. The water phase contained 0.085% phosphoric acid (H₃PO₄). For analysis of bilirubin, the water phase was 90% ACN containing 0.1% formic acid and 10% water with 1% acetic acid for 20 min. Column temperature and sample temperature were kept at 35°C and 20 C, respectively. Each injection volume per test was 10 µL. The detector ultraviolet (UV) wavelength was set at 283 nm for the detection of liquiritin, baicalin, and paeoniflorin and 450 nm for bilirubin, and the scan range for DAD was 190–450 nm. To build calibration curves for quantification, we dissolved standards with 90% methanol at 1 mg/mL, and the standard mixture was mixed at 0.33 mg/mL for each compound. The standard solution of bilirubin was prepared at 120 μg/ml as the maximum concentration. The standard curves were established by using standard solutions with standard protocols.

C57BL/6N mice were purchased from the Centre for Comparative Medicine Research, the University of Hong Kong with ages of 6–8 weeks. The animal experiment and process were approved by the University Committee on the Use of Live Animals in Teaching and Research (CULATR no. 4001-16). Mice were given ad libitum access to food and water.

Several behavioral tests were applied to assess the emotional depressive- and anxiety-like behaviors, including the tail suspension test (TST), forced swimming test (FST), and open field test (OFT) as in our previous work (Du et al., 2021). The behavior tests were performed 24 h after the last treatment. The TST and FST were performed in the morning and the interval of each mouse between different tests was 30 min. Briefly, in the TST, the mouse tail was suspended from the top of the box for 6 min to record the behavior. In the whole test, we recorded the immobility time to evaluate the learned helplessness of mice. In the FST, the mice were placed into a water tank (30 cm × 20 cm) for 6 min. The temperature was kept at 23°C∼25°C. The mobility time was recorded in a cylinder filled with water. The activations including strong body shaking, movement of the limbs, and trying to reach the wall were captured. The moving time of the mice in the FST was recorded as the mobility time. In the FST, the first 2 min was recognized as habituation, and the mobility time in the last 4 min was applied for the analysis (Gao C. et al., 2018). In the OFT, the mice were placed into a transparent Plexiglas cube with the size of 43 cm × 43 cm with a center zone (20 cm × 20 cm). In the test session, the mice were placed in the corner and allowed 2 min to recover from handling. The open field would be cleaned with 75% ethanol before the test. The movement was recorded for 8 min using video tracking software. In the sucrose splash test (SST), 10% sucrose solution was used to spray on the back of the mice. The grooming time, including licking, biting, or scratching, was calculated over a 10-min period. The novelty-suppressed feeding test (NSFT) was conducted the next day after 24 h of food deprivation. NSFT was performed in an open field with a single food pellet in the center of the arena. The weight of the food was recorded to calculate the food consumption. The latency of feeding and the consumption weight of the food pellet were recorded over 10 min. All the videos were recorded with video tracking software Smart 3.0 system (RWD, Delawares).

Human embryonic stem cells (hESCs, WiCell) were cultured with the protocols described in our previous study (Tsoi et al., 2022). In brief, H7 cells were cultured in a Matrigel-coated 6-well plate (Corning). The culture medium was feeder-free TeSR-E8 medium. At the first 24 h of H7 seeding, Rho-kinase inhibitors were supplemented into the culture medium. After that, the culture medium was changed into fresh medium for every 24 h. NSCs were induced from hESCs after 2–3 days of recovery. The induction medium consisted of two separated mediums, the first of which was N2, a non-essential amino acid, and GlutaMax in DMEM/F12 medium. The other medium was a NeuroBasal medium supplemented with B27, and two SMAD inhibitors: SB431542 and LDN193189 (Watanabe et al., 2005; Chambers et al., 2009). Baicalin, paeoniflorin, and liquiritin (Pusi BioTech), three representative compounds, were prepared in DMSO at a concentration of 1 mg/ml as the stock solution. For the treatment, these compounds were diluted to 10 μg/ml. As liquiritin revealed its neurogenesis-promising effects, we tested whether the bioactivity is related to modulating TrkB signaling. The cells were co-treated with liquiritin and a TrkB selective inhibitor cyclotraxin B (Cyc-b, MCE) at a dosage of 100 nM. The control group was treated with the same volume of DMSO as the vehicle control.

For immunofluorescence staining, brain samples from different groups were post-fixed with 4% PFA for 48 h, followed by dehydration in 30% sucrose solution at 4°C. After full dehydration, the brain tissues were embedded in O.C.T. and stocked in −80°C till to use. The entire length of the DG was cut into 30 μm sections as frozen slices and stored at −20°C. For in vivo NSCs imaging, the tissue sections were attached to Poly-D-Lysine coated microscope slides (0111500; GmbH and Co. KG). Antigen retrieval was conducted by using citrate acid buffer (pH 6.0) at boiling conditions. After a PBS wash, the brain slices were permeabilized using PBS containing 1% Triton X-100 (PBST) and blocked by 5% goat serum for 1 h at room temperature. As for the BrdU staining, the samples were incubated with 2 N HCl for 30 min followed by 0.1 M borate buffer (pH 8.5) neutralizing. The samples were incubated with primary antibodies including anti-BrdU (Abcam, 1:200), anti-doublecortin (DCX, Cell Signaling Tech, 1:400), anti-NeuN (Cell Signaling Tech, 1:400), anti-MAP2 (Cell Signaling Tech, 1:400) and anti-β Tubulin (Tuj1, Santa Cruz, 1:400). The sections were then stained with fluorochrome-conjugated secondary antibodies, and the nucleus was counterstained with DAPI and mounted with antifade medium (Dako). The fluorescent images were obtained by a confocal microscope (Zeiss LSM 800, Core facility in LSK Faculty of Medicine, HKU) and analyzed by Zeiss software. As for in vitro NSCs imaging, the NSCs were seeded to the Matrixgel coated slides at a density of 50,000 cells/well before 3 days of imaging. The processing methods of slides imaging were similar to methodology in the tissue section. To analyze the positive fluorescent results quantitatively, after obtaining the confocal image with z-stack, we adopted the method named “maximum intensity project” to convert the 3D image to a 2D image with an increased intensity for calculation (Du et al., 2023). Then, we calculated the BrdU/DCX or BrdU/NeuN dual positive cells colocalized with the nucleus as newborn immature or mature neurons in the brain section. To avoid an individual difference in cell counting, the positively stained cells were calculated by the authors, who had no information, in a group setting to avoid personal bias.

As in our previous works, the expression levels of targeted proteins were analyzed by Western blot (Du et al., 2021; Gao et al., 2022). The proteins were extracted from the hippocampal zone of the ischemic brain site by using radioimmunoprecipitation (RIPA) buffer plus with 1% protease and phosphatase inhibitor cocktails (Cell Signaling Tech). A protein assay kit (Thermo Fisher Scientific) was used for quantification of the loading contents of the protein samples. The protein samples (20 μg per well) were separated by sodium dodecyl sulfate-polyacrylamide (SDS-PAGE) gel electrophoresis and transferred on 0.45 µM polyvinylidene fluoride (PVDF) membranes for blotting. The membrane was blocked by 5% bovine serum albumin solution for 1 h at room temperature in a rotation and immunoblotted with primary antibodies following HRP-conjugated secondary antibodies. The primary antibodies applied for the blotting analysis were listed as follows: BDNF (abcam, ab108319, 1:1000), TrkB (Cell Signaling Tech, 4603, 1:1000), Phospho-MEK1/2 (S217/221) (Cell Signaling Techn, 9121, 1:1000), MEK1/2 (Cell Signaling Tech, 9122, 1:1000), Phospho-P44/P42 ERK (T202/T204) (Cell Signaling Tech, 4370, 1:1000), P44/P42 ERK (Cell Signaling Tech, 4695, 1:1000), CREB (Cell Signaling Tech, 4820, 1:1000), Phospho-CREB (S133) (Cell Signaling Tech, 9198, 1:1000), and β-Actin (Cell Signaling Tech, 3700, 1:3000). After incubation with these primary antibodies, the samples were incubated with the secondary antibodies overnight at 4°C with rotation. The membranes were visualized using a chemiluminescent ECL select kit (GE Healthcare) detected by a Gel Doc system (Bio-Rad) and analyzed using Image Lab software (Bio-Rad). The relative expression levels of BDNF and TrkB were calculated by dividing their respective volumes by the volume of β-Actin. For the phosphorylated proteins, the relative expression levels were determined by comparing the volume of phosphorylated proteins to the volume of total proteins on the same membrane. For the calculation of p-MEK1/2/T-MEK1/2, the volume of p-MEK1/2 (S217/221) was quantified in phosphorylation compared to the volume of total MEK1/2. The volume of proteins was calculated using Image Lab software (Bio-Rad).

After NSCs induction and drug treatment for 10 days, the cells were dissociated by using Accutase (Gibco) and the isolated cells were fixed with 4% PFA for 15 min, followed conjugated fluorochromes staining overnight, including PE anti-Nestin (Biolegend, 1:200), APC anti-Tubulin β 3 (Biolegend, 1:200) and anti-rabbit DCX (Cell Signaling Tech, 1:2000). Alexa Flour 488 anti-rabbit secondary antibody (Thermo Fisher Scientific, 1:2000) was used to conjugate DCX for flow cytometry analysis. We detected the conjugated fluorochromes using a NovoCyte Quanteon flow cytometer (Agilent, Core facility in LSK Faculty of Medicine, HKU) for data collection, and data analysis was performed using FlowJo software.

All the data were presented as mean ± S.E.M. and analyzed using Prism 8 (GraphPad). Statistical analysis was performed using one-way ANOVA followed by Dunnett’s multiple-comparison test for multiple group comparisons. For the behavioral results of the animal studies, the samples were collected from two separate batches of mice experiments. The results of the Western blot were conducted with 5–6 samples from each group and repeated twice. The in vitro experiments were performed with three biological replicates and repeated twice. The sample sizes, degree of freedom, F values, and p-values were provided. The normal control group was the mice without model establishment, and it was compared to the vehicle-treated depression model group to provide information on successful model establishment. The vehicle-treated depression model was compared with the NHQXW treated depression mice to give statistics information between the treatment group and the depression group. The p-value of <0.05 was recognized as statistical significance.

The identification and quantification of chemicals are essential for TCM studies to ensure repeatable and consistent results. According to Chinese Pharmacopoeia and previous reports, liquiritin, baicalin, and paeoniflorin could be representative components for quality control of NHQXW (Zhao et al., 2008; Fan et al., 2019; Wang et al., 2021). Calculus bovis (C. bovis) is a key component of NHQXW and bilirubin is the representative compound in C. bovis responsible for its neuropharmacological effects (Yu et al., 2020). In our previous works, baicalin could improve hippocampal neurogenesis in a chronic CORT stress depression mouse model (Gao C. et al., 2018). Hence, we performed bioactive components-based quality control studies of NHQXW using HPLC. As shown in Figures 1A, B, liquiritin, baicalin, and paeoniflorin had good separation and linearity in standard solutions and samples using HPLC analysis. The correlation coefficients indicated a good linear correlation in regression analysis (R² > 0.999) for all constituents in the ranges of 10 μg/mL to 333.3 μg/mL. These results suggest that the method should be reliable with high sensitivity (Figures 1C–E). The content of liquiritin, baicalin, and paeoniflorin was 3.1264 ± 0.0238, 3.0851 ± 0.0255, and 1.0669 ± 0.0058 mg/pill, respectively in NHQXW (3 g per pill). Given that bilirubin has different chemical characteristics from other active compounds, we developed a method to successfully determine the concentration of bilirubin in NHQXW, showing a good linear correlation with R² at 0.9999 using DAD detector (Figures 1F, H), and the content of bilirubin was 4.6326 ± 0.0831 mg/pill in NHQXW.

We first evaluated the anti-depressant activities of NHQXW by using a CRS induced depression mouse model. To induce restraint stress, we placed the mice into a 50 mL polypropylene conical tube for 6 h per day continuously for 30 days. Fluoxetine (FLX) treatment was regarded as the positive control. The first day of the restraint session was recognized as D0. NHQXW or FLX were orally administrated to the mice at D0 and continuously treated for 30 days (Figure 2A). The treatment dosage of NHQXW was 20 mg/g/day, an equivalent dose for human subjects that mimics the clinical use of NHQXW for antidepressant treatment. The TST, FST, OFT, SST, and NSFT were applied to assess the depressive-like behaviors. The restraint stress mice revealed typical anxiety-like behaviors with decreased grooming frequency (Figures 2B–G). In SST, the treatment of NHQXW significantly increased grooming frequency in the restraint stress mice (Figure 2B), the effects of which were similar to the effects of FLX. In the TST and FST, NHQXW treatment significantly attenuated immobility time, the effect of which was also similar to FLX treatment (Figures 2C, D). In the OFT, both the NHQXW and FLX treatment groups spent more time exploring the central field than the vehicle-treated depression mice (Figure 2E). In the NSFT, both the NHQXW treatment group and the FLX group had a shorter latency to food and higher food consumption than the vehicle-treated depressive mice (Figures 2F, G). Taken together, NHQXW attenuated the depressive-like behaviors in the restraint stress mice.

We also assessed the role of NHQXW on CCS mice. As a serotonin reuptake inhibitor, FLX is commonly used for anti-depressive treatment and we also adopted FLX as the positive treatment for chronic stress depressive mice. However, FLX could trigger severe withdrawal reactions in depression treatment (Coupland et al., 1996; Fava et al., 2015). Thus, we further evaluated the anti-depressive effects of NHQXW on the CCS model with both therapeutic and withdrawal protocols. The therapeutic protocol and withdrawal protocol were designed to investigate the anti-depression effects of NHQXW on a short-term and long-term basis (Figure 3A). After 15 days of CORT stress, NHQXW and FLX were administered to the mice for 25 days as the therapeutic strategies. As expected, the CORT-stressed mice showed abnormal depression-like behaviors in different behavior tests including the SST, TST, FST, OFT, and NSFT (Figures 3B–G). The treatment of NHQXW for 30 days significantly attenuated the depressive- and anxiety-like behaviors in the CORT mice in those behavioral tests. Similar to the CRS mice, NHQXW also remarkedly increased the grooming frequency of the CORT mice in the SST (Figure 3B). Furthermore, NHQXW significantly increased the mobility time of the subjected CORT mice in the TST and FST (Figures 3C, D). In the OFT, we recorded the central exploring time in the open field and found that NHQXW and FLX increased the exploring times of the mice in the center of the field (Figure 3E). In the NSFT, we tested the feeling of fear in a novel space. Similar to FLX, the NHQXW treatment significantly ameliorated food consumption and the first latency to the food (Figures 3F, G). We also determined the depressive behaviors with withdrawal protocol. The mice were administrated NHQXW and FLX for 14 days and then these treatments were stopped in the period between day 15 and day 28 until the end of CORT induction. In the FST, NHQXW treatment significantly improved the mobility time in the CORT stress mice even with the withdrawing protocol for 2 weeks (Figure 3H). Similar effects were also found in the OFT. Even after withdrawing NHQXW treatment for 2 weeks, the CORT mice that received NHQXW treatment still revealed an increase in the exploring time in the OFT (Figure 3I). Furthermore, the FLX treatment had no significant effect on the abnormal behaviors in the FST and the OFT (Figures 3H, I). These results indicate that NHQXW had anti-depressive and anti-anxiety effects in the CCS mice. Particularly, NHQXW prevented withdrawal-associated rebound symptoms in the chronic CORT stress mice.

We next tracked the effect of NHQXW on modulating the proliferation and differentiation profiles of NSCs in the hippocampus area in both the restraint stress model and the CCS depression mouse model. BrdU was applied to label the proliferation of the NSCs in the hippocampal DG. We performed dual immunofluorescent experiments to co-label the expression of doublecortin (DCX) and neuronal nuclear protein (NeuN) in the region of SGZ to detect the differentiation potentials of endogenous NSCs. The distribution of positive stained NSCs was calculated at the XYZ planes of the confocal image. As such, the double positive staining of BrdU and DCX (BrdU/DCX) was used to identify newborn immature neurons with high differentiation potentials, while BrdU and NeuN dual-positive staining (BrdU/NeuN) was regarded as the newborn mature neurons. The restraint stress model showed significantly decreased BrdU and BrdU/DCX positive populations in the hippocampal area (Figures 4A–C). Treatment with NQHXW markedly elevated the BrdU and BrdU/DCX positive cells, indicating a promising induction of the NSC proliferation and differentiation respectively. The NQHXW treatment had similar effects to FLX in the restraint stress mouse model. Furthermore, the CORT depressive mice also revealed to decrease newborn immature neurons (BrdU/DCX) and newborn mature neurons (BrdU/NeuN) in the hippocampal area, indicating the inhibition of NSCs proliferation and differentiation respectively (Figures 4D, E). Both NHQXW and FLX significantly improved the BrdU/DCX positive cells (Figures 4E, H) and BrdU/NeuN positive cells (Figures 4D, G) in the DG region. Moreover, we also detected the differentiation potential of NSCs in the chronic CORT stress mice (Figures 4J–L). The ratio of BrdU/DCX positive cells in total newborn NSCs was markedly increased compared with the ratio in the CCS mice indicating the differentiation potential of newborn NSCs was improved after NHQXW treatment (Figure 4I). The 2-week NHQXW treatment significantly enhanced the BrdU/DCX and BrdU/NeuN dual-positive cells in the SGZ region of the CORT-treated mice. In contrast, the treatment of FLX for 2 weeks had no remarkable effect on the BrdU/DCX and BrdU/NeuN dual-positive cells in the SGZ region (Figures 4J–L). These results reveal that NHQXW promoted hippocampal neurogenesis in chronic stress mice.

The BDNF-associated signaling pathway plays a crucial role in regulating NSC differentiation for neurogenesis (Martinowich et al., 2007; Brunoni et al., 2008; Bathina and Das, 2015; Miranda et al., 2019). BDNF and its downstream signaling (TrkB, BDNF receptor, and CREB) participate in multiple brain repair aspects including neurogenesis, synaptic plasticity, and synaptic transmission, subsequently affecting depression development (Taliaz et al., 2010; Yu and Chen, 2011). Thus, we explored the effects of NHQXW on the BDNF/TrkB/ERK/CREB signaling pathway in the CCS-exposed mice (Figure 5). After being exposed to chronic CORT stimulation, the vehicle mice revealed the downregulated expression of BDNF, TrkB, p-ERK (T202/T204), p-MEK1/2 (S217/221) and p-CREB (S133) in the hippocampus area (Figures 5A–F). These changes were completely abolished by the treatment with NHQXW (Figures 5A–F). As a positive control, the treatment of FLX also blocked CCS stimulation on the expression of BDNF, p-ERK, and p-CREB but had no effect on p-TrkB in the hippocampus (Figures 5E, F). Taken together, NHQXW could regulate the BDNF/TrkB/ERK/CREB pathway to induce hippocampal neurogenesis in chronic stress-associated depression.

We investigated the bioactivities of paeoniflorin, liquiritin, and baicalin, at a dosage of 10 μg/mL, in promoting neurogenesis by using in vitro cultured human NSCs. After 10 days of NSC induction and compound treatment, the expansion and differentiation of NSCs were determined by immunostaining of Nestin/Tuj1 and DCX respectively. The treatments of liquiritin and paeoniflorin significantly increased nestin/Tuj1 positive cells, indicating promising effects on the proliferation (Figures 6A, B). Meanwhile, treatment of liquiritin significantly increased the number of DCX-positive cells, suggesting the promotion of NSC differentiation (Figures 6C, D). Baicalin and paeoniflorin showed no obvious effect on the induction of DCX-positive cells. Furthermore, treatments of baicalin had no effect on the induction of Nestin/Tuj1 and DCX-positive cells. Thus, liquiritin could be a representative ingredient of NHQXW with neurogenesis-promising effects. We then investigated whether the neurogenic effects of liquiritin could be related to the modulation of TrkB, the downstream signaling of BDNF. The CORT stimulation remarkably inhibited the rates of MAP2-positive cells in the cultured NSCs. Furthermore, liquiritin treatment abolished the inhibitory effects of CORT on neurogenesis (Supplementary Figure S1). Furthermore, co-treatment with TrkB inhibitor Cyc-B abolished the effects of liquiritin on the MAP2-positive cells. These results suggest that liquiritin contributes to the effects of NHQXW on the induction of BDNF signaling and the promotion of neurogenesis under CORT stimulation. Thus, liquiritin and paeoniflorin could be used as representative compounds to promote neurogenic efforts, and subsequently they are suitable for use in a bioactivity-guided quantity control study. Of note, baicalin treatment did not show the neurogenesis-promising effects in our in vitro results.