The botanical drugs used to prepare the SX-1 decoctions were purchased from the First Affiliated Hospital of Xiamen University. The ingredients are listed in Table 1. All herbs were pooled together and boiled twice with pure water. The two decoctions were mixed together and concentrated to 1.383 g/ml. This decoction is a hospital preparation. The treatment dosage for rats was 6.25 times that for patients based on the body weight (Nair et al., 2018). For instance, a patient needs 155 g herbs per 70 Kg body weight every day. Then, a rat needs 1.383 g/100 g herbs, which is the low dose of SX-1 (SXL). The high dose (SXH) is twice the low dose. In the same way, the daily dose of fluoxetine required for an adult (20 mg/70 Kg) was converted into the dose of fluoxetine for rats (0.1786 mg/100 g). Fluoxetine is diluted with pure water.

Wistar rats (300–350 g, Shanghai SLAC Laboratory Animal Co., Ltd.) were gavaged with the prepared low-dose SX-1 (Rat SX serum, n = 20) or distilled water (Rat con serum, n = 20) for 7 days. Abdominal aortic blood was collected 1 h after intragastric administration on day 7. Before blood collection, rats were intraperitoneally injected with chloral hydrate at a dose of 0.3 ml/100 g. After anesthesia, the abdominal cavity of the rats was opened surgically to find the abdominal aorta. Disposable blood collection needle and negative pressure blood collection vessel were used to collect blood from the abdominal aorta. and the serum were extracted from blood with the centrifuge at 3,000 rpm for 15 min. The rat SX serum and rat con serum were mixed well respectively, then sterilized in a 55°C water bath for 30 min and stored at -20°. l-Glutathione Reduced was purchased from APExBIO.

All herbs of SX-1 were crushed with a mixer mill for 60 s at 60 Hz. 100 mg of sample was added to 500 μl of extracted solution which dissolved in 80% methanol containing 10 μg/ml of internal standard. After 30 s vortex, the samples were homogenized at 45 Hz for 4 min and sonicated for 1 h in ice-water bath. After placing 1 h in -40°C, the samples were centrifuged at 12,000 rpm (RCF = 13,800 (×g), R = 8.6 cm) for 15 min at 4°C. Finally, the supernatant was obtained and put in a fresh 2 ml tube for LC-MS/MS analysis. 400 μl of plasma sample were added to 40 μl of hydrochloric acid (2 mol/L), then the mixture was vortexed for 1 min and followed by incubated for 15 min at 4°C. The vortex and incubat cycle were repeated for 4 times. Add 1.8 ml acetonitrile, then the mixture was vortexed for 5 min and the samples were centrifuged at 12,000 rpm (RCF = 13,800 (×g), R = 8.6 cm) for 5 min at 4°C. 1,600 μl of supernatant was transferred to a fresh tube and nitrogen dried. The dried samples were reconstituted in 150 μl of 80% methyl alcohol containing 10 μg/ml of internal standard by vortex for 5 min. The constitution was then centrifuged at 12,000 rpm (RCF = 13,800 (×g), R = 8.6 cm) for 5 min at 4°C, and 120 μl of supernatant was transferred to a fresh glass vial for LC/MS analysis.

LC-MS/MS analysis of supernatant of SX-1 was performed on an Agilent ultra-high performance liquid chromatography 1290 UPLC system with a Waters UPLC BEH C18 column (1.7 μm 2.1*100 mm). The column temperature was set at 55°C and the sample injection volume was set at 5 μl. The flow rate was set at 0.5 ml/min. The mobile phase consisted of 0.1% formic acid in water (A) and 0.1% formic acid in acetonitrile (B). The multi-step linear elution gradient program was as follows: 0–11 min: 85–25% A; 11–12 min: 25–2% A; 12–14 min: 2–2% A; 14–14.1 min: 2–85% A; 14.1–15 min: 85–85% A; 15–16 min, 85–85% A. An Q Exactive Focus mass spectrometer coupled with an Xcalibur software was employed to obtain the MS and MS/MS data based on the IDA acquisition mode. During each acquisition cycle, the mass range was from 100 to 1,500, and the top three of every cycle were screened and the corresponding MS/MS data were further acquired. Sheath gas flow rate: 45 Arb, Aux gas flow rate: 15 Arb, Capillary temperature: 400°C, Full ms resolution: 70,000, MS/MS resolution: 17,500, Collision energy: 15/30/45 in NCE mode, Spray Voltage: 4.0 kV (positive) or -3.6 kV (negative).

LC-MS/MS analysis of rat serum was performed on an UHPLC system (Vanquish, Thermo Fisher Scientific) with a Waters UPLC BEH C18 column (1.7 μm 2.1*100 mm). The flow rate was set at 0.4 ml/min and the sample injection volume was set at 5 μl. The mobile phase consisted of 0.1% formic acid in water (A) and 0.1% formic acid in acetonitrile (B). The multi-step linear elution gradient program was as follows: 0–3.5 min: 95–85% A; 3.5–6 min: 85–70% A; 6–6.5 min: 70–70% A; 6.5–12 min: 70–30% A; 12–12.5 min: 30–30% A; 12.5–18 min: 30–0% A; 18–25 min: 0–0% A; 25–26 min: 0–95% A; 26–30 min: 95–95% A. An Orbitrap Exploris 120 mass spectrometer coupled with an Xcalibur software was employed to obtain the MS and MS/MS data based on the IDA acquisition mode. During each acquisition cycle, the mass range was from 100 to 1,500, and the top four of every cycle were screened and the corresponding MS/MS data were further acquired. Sheath gas flow rate: 30 Arb, Aux gas flow rate: 10 Arb, Ion Transfer Tube Temp: 350°C, Vaporizer Temp: 350°C, Full ms resolution: 60,000, MS/MS resolution: 15,000, Collision energy: 16/38/42 in NCE mode, Spray Voltage: 5.5 kV (positive) or -4 kV (negative).

Male Wistar rats (8 weeks old, 250–280 g) were purchased from Shanghai SLAC Laboratory Animal Co., Ltd. All the rats were housed under the controlled environment with a room temperature at 22 ± 2°C, a 12 h light/dark cycle, and with unlimited access to water and food. All animal experiments were carried out in accordance with the rules for the Laboratory Animal Ethics Committee of Xiamen University.

One week after the acclimation of the environment, the rats were randomly divided into the control group (n = 10) and CUMS model group (n = 40). For the control group, five rats were kept in each cage, and for the model group, each rat was housed in a single cage. CUMS was applied to the model group rats for 4 weeks. Different physical stimulation was applied to the CUMS model rat daily for 4 weeks, as summarized in Table 2. Four weeks later, the CUMS rats were further randomly divided into four groups: 1) CUMS rats treated with vehicle (CUMS), 2) CUMS rats treat with SX-1 at low dose (CUMS + SXL), 3) CUMS rats treat with SX-1 at high dose (CUMS + SXH), and 4) CUMS rats treat with Fluoxetine (CUMS + Flu). There are ten rats in each group (n = 10). SX-1 and Fluoxetine were administrated by gastric gavage with the volume of 0.1 ml/10 g for 6 weeks. Meanwhile, a small amount of stimulation was continued. After 6-week treatment, sucrose preference experiment and open field test were arranged to carry out. At the end of the experiments, the rats were anesthetized with 4% chloral hydrate. Blood and liver were collected under anesthesia. The lower right edge of the hepatic lobar was collected and fixed in 4% Paraformaldehyde (PFA), and the rest of the liver was cut into small pieces and frozen in a cryopreserved tube with liquid nitrogen, and then stored at -80 C.

Sucrose preference experiment is a behavioral experiment to measure whether animals lack pleasure. When given the choice of drinking sucrose solution or water, rodents will choose to drink sucrose solution. Animals that don’t show this preference are thought to exhibit behavior similar to anhedonia (Papp et al., 1991). Sucrose preference experiment consists of two parts: adaptation training and test. During the training, two bottles of 1% (w/v) sucrose solution were put into each cage in the first 24 h, and then one of the bottles was replaced with pure water in the next 24 h. After adaptation, the rats were fasted for 24 h, and then the baseline test was started for 12 h. In the test, the rats could only choose two bottles weighed in advance, one is 1% (w/v) sucrose solution, and the other is pure water. After fasting for 12 h, the two bottles were taken away and weighed. The total liquid consumption, sugar consumption and pure water consumption were recorded. The formula of sugar preference index is as follows: Sugar   preference   index   % = sugar   water   consumption / ( sugar   water   consumption + pure   water   consumption ) × 100 %

One of the most commonly used behavioral tests for rodents is the open field test, which is one of the common methods to observe whether animals have the desire to explore (Sturman et al., 2018). The open field reaction box has a gray inner wall, with a digital camera mounted directly above it. The vision field of the camera covered the entire open field. The rats were allowed to adapt to the environment first. The central region and the peripheral region are set up in the small animal recording behavior analysis system I Topscanlite (Clever Sys). At the beginning of the experiment, the rats were gently placed in the center of the open field with their heads facing one of the four corners of the box. The timing was 5 min from the time when the rats entered the open field box. The system will record the residence time of the rats entering the central area and the total travel distance. At the end of each experiment, urine and stool were removed with 70% alcohol and dried with an air blower.

AML12 cells (Cellcook, Guangzhou, China) were cultured in DMEM/F-12 medium (BasalMedia, Shanghai, China) containing 10% fetal bovine serum (FBS, Gibco, Australia), 1% ITS liquid media supplement (Sigma, I3146), 1% Penicillin-Streptomycin (BasalMedia, Shanghai, China) and Dexamethasone 40 ng/ml in a 37°C 5% CO2 saturated humidity incubator. The normally cultured AML12 cell lines were centrifuged at 1,000 rpm for 5 min, and 10⁵ cells/well were placed in a 12-well plate and 5*10³ cells/well were placed in a 96-well plate and incubated at 37°C.

AML12 cell culture was conducted in the following groups: a normal control group, rat con serum groups at [10%, 20%, 40%, 80%], rat SX serum groups at [10%, 20%, 40%, 80%], l-Glutathione (Glu) groups at [10 μM, 20 μM, 40 μM, 80 μM], stattic groups at [5 μM, 10 μM, 20 μM], cobalt chloride (CoCl₂) groups at [200 μM, 400 μM, 800 μM]. Glu used in this study was provided by APExBIO (Houston, USA; CAS No. 70-18-8; Molecular formula: C₁₀H₁₇N₃O₆S; Molecular weight: 307.32); l-Glutathione Reduced is an endogenous antioxidant, with the ability to scavenge oxygen-derived free radicals, and stattic was provided by Selleck (Shanghai, China; CAS No. 19983-44-9; Molecular formula: C₈H₅NO₄S; Molecular weight: 211.19), and CoCl₂ was provided by Sigma (Darmstadt, Germany; CAS No. 7646-79-9; A string у Split spinning: CoCl₂; Molecular weight: 129.84). After 80% cell adhesion, different concentrations of rat con serum and rat SX serum were added. The system of each pore was 100 µl. The experiment was repeated three times independently. After different time points of drug action, 10 μl cell counting kit-8 (CCK8) detection solution (MedChemExpress, Shanghai, China) was added to each pore, and the reaction time was 2.5 h at 37°C. A blank control group was set up and treated in the same way without anything added. Optical density (OD) was measured at 450 nm. The results were calculated using the following equation: Cell survival rate (%) = OD value of experimental group/OD value of non-drug group × 100%.

Serum levels of ALT and alanine aminotransferase (AST) were measured by ALT and AST assay kit (IFCC) (140120008, 140220006, Shenzhen Mindray Bio-Medical Electronics Co., Ltd.) using an automatic biochemical detector (BS-240VET, Shenzhen Mindray Bio-Medical Electronics Co., Ltd.).

Serum levels of corticosterone were measured by commercial Elisa kits according to the manufacturer’s protocols (201074, R&D, U.S).

Hepatic IL-6 level was measured by ELISA. Briefly, the rat livers were homogenized with RIPA buffer and centrifuged at 10,000 rpm for 15 min at 4°C. The supernatant was collected and used for determination of the protein levels of IL-6, using commercial ELISA kits (ELR-IL-6-1, Ray Biotech, U.S) following the manufacturer’s protocols.

Total RNA was extracted from liver tissue using total RNA Extraction Kit (Tiangen, No. DP419). CDNA was synthesized from total RNA (2 μg) using fastking RT Kit (with gDNase) reverse transcription Kit (Tiangen, No. KR116-02). Real time quantitative PCR (qPCR) was performed using SYBR Green PCR Master Mix (Tiangen, No. FP205-02) and the forward primer of IL-1β was CAC​CTC​TCA​AGC​AGA​GCA​CAG, the reverse primer of IL-1β was GGG​TTC​CAT​GGT​GAA​GTC​AAC; the forward primer of IL-18 was TGG​AGA​CTT​GGA​ATC​AGA​CC, the reverse primer of IL-18 was GGC​AAG​CTA​GAA​AGT​GTC​CT; the forward primer of TNFα was AAA​TGG​GCT​CCC​TCT​CAT​CAG​TTC, the reverse primer of TNFα was TCT​GCT​TGG​TGG​TTT​GCT​ACG​AC, the forward primer of β-actin was CTG​TGT​GGA​TTG​GTG​GCT​CT, the reverse primer of β-actin was CAG​CTC​AGT​AAC​AGT​CCG​CC. Gene expression data were analyzed by fluorescence quantitative PCR (Roche Lightcycle 480II, USA).

The rat livers and AML12 cells were lysed by RIPA buffer, separated by 10% polyacrylamide gel electrophoresis and then transferred to polyvinylidene fluoride (PVDF) membrane. After blocking in 5% Bovine Serum Albumin (BSA), the membrane was incubated with the phospho-JAK2 (Tyr931) (1:1,000, affinity), total JAK2 (1:1,000, proteintech), phospho-STAT3 (Tyr705) (1:1,500, abcam), total STAT3 (1:2000, abcam), IL-6 (1:500, bioworld), IL-18 (1:2000, bioworld), IL-1β (1:500, sino biological), TNFα (1:1,000, sino biological), HIF-1a (1:1,000, Novusbio), HIF-2a (1:1,000, Cell Signaling Technology), VEGFa (1:1,000, Novusbio) primary antibody seperately overnight at 4°C, and then was incubated with HRP conjugated secondary antibody for 1 h at room temperature. Western blots were displayed by advanced enhanced chemiluminescence kit and then semi-quantified by Clinx (Shanghai Qinxiang Scientific Instrument Co., Ltd).

The liver tissue was fixed in tissue fixative, embedded in paraffin, and then sectioned (5 μm) for HE and Immunohistochemistrical staining. HE staining was used to observe the morphological changes of the liver. Hepatic fibrosis was visualized by Immunofluorescence staining of Col I. The monoclonal antibody against Col I (1:500, Servicebio) and secondary goat anti-rabbit IgG-HRP antibody (1:300, Servicebio were used respectively, and the positive signals were observed under the florescence microscope.

All data are shown as the mean ± standard error (SEM). Two tailed unpaired t-test was used to compare the two groups of data. Multiple groups of random measurement data were tested by one-way ANOVA. After conforming to the normality test, the data were continued to be used to test homogeneity of variance between groups, using least-significant difference LSD for multiple comparisons on the assumption of equal variance. Repeated measures ANOVA was applied to the data detected at multiple time points such as weight and depression-like behaviors. SPSS 25 software (IBM, United States) was used to analyze the data. The difference was considered statistically significant when p < 0.05.

The quality control of SX-1 and drug-containing serum were investigated by UHPLC-QE-MS. The UHPLC-QE-MS chromatogram is shown in Figure 1 and Figure 2. Several ingredients were identified. As shown in Figure 1 and Figure 2, sixteen compounds were distinguished respectively.

Four-week CUMS significantly reduced the rat body weight (p<0.001), which was maintained through the following 6-weeks treatment. SX-1 significantly restored the CUMS-resulted decreased body weight at both low and high doses (p = 0.004 and p = 0.032, respectively), while Flu had no effect (p = 0.259; Figures 3A,B). CUMS provoked obvious depression-like behaviors, indicated by monitoring of sucrose preference and total activity distance in open field test (both p<0.001), which were improved by SX-1 and Flu (p<0.001 and p = 0.01; Figure 3C) and SXL respectively (p = 0.021; Figure 3D). SX-1 also tended to recover the exploration time length in central area, although without statistical significance (p = 0.928 or p = 0.812; Figure 3E). Consistently, CUMS significantly increased the serum levels of stress-related hormone corticosterone compared with the control group, which were completely reversed by SX-1 but not Flu (p = 0.001, p<0.001, p = 0.003 and p = 0.051, respectively; Figure 3F).

As an indicator of hepatic injury, serum levels of ALT were significantly increased in CUMS rats compared to that of the control rats (p = 0.003), which were completely reversed by SX-1 but not Flu (p < 0.001 and p = 0.063; Figure 4A). The serum levels of AST were of no difference among groups (p = 0.142, p = 0.184, p = 0.409 and p = 0.073, respectively; Figure 4B). HE staining showed that CUMS resulted in obviously disordered hepatic sinus and lobule structure, with typical balloon-like changes, which were improved by SX-1 treatment (Figure 4C).

CUMS induced a significant deposition of Col I in the liver. As shown in Figures 4A,D large number of red fluorescence surrounded the nucleus in the CUMS group, that is, the pathological manifestation of Col I deposition in the extracellular matrix. It also indicated the development of hepatic fibrosis. After SX-1 and Flu treatment, the red fluorescence in the picture decreased significantly. According to the statistics of fluorescence area by ImageJ software, SX-1 and Flu can significantly reverse the pathological manifestations of liver fibrosis (p < 0.001) (Figures 4D,E).

At both mRNA and protein levels, CUMS up-regulated the inflammatory factors such as IL-1β, IL-18 and TNFα in liver tissues, which were tended to be restored by SX-1, although without statistical significance (both p>0.05; Supplementary Figure S1). Critically, compared with the control rat, CUMS rats showed increased levels of hepatic IL-6 protein, which were reversed by SX-1 and Flu treatment in Elisa test (both p<0.05; Figure 5A left). However, in WB, the protein level of IL-6 decreased significantly in SXH compared with CUMS (p = 0.027; Figure 5A right). Accordingly, the phosphorylation of JAK2 and STAT3 were significantly up-regulated by CUMS (p = 0.044 and p = 0.027, respectively), indicating the activation of IL-6/JAK2/STAT3 signaling, which were significantly improved by SXH treatment (p = 0.046 and p = 0.005; Figures 5B–D).

In vivo, we also observed that CUMS induced the expression of HIF-2α in the liver rather than HIF-1α (Supplementary Figure S2A, B). HIF-2α was only expressed in specific cells, such as hepatocytes. At the same time, acute hypoxia rapidly induced HIF-1α and HIF-2α proteins accumulate in cells. However, with the prolongation of hypoxia time, the stability of HIF-1α mRNA decreased and its protein expression decreased, while HIF-2α protein expression continued to increase (Bartoszewski et al., 2019). Therefore, we tried to explore the mechanism of hypoxia, but VEGFa protein could not be activated (Supplementary Figure S2C), and the mechanism was not clear. Therefore, in order to further verify whether SX-1 could improve CUMS induced liver injury through IL-6/JAK2/STAT3 signaling pathway, we used CoCl₂ induced cell hypoxia model in vitro, that cellular hypoxia increases the expression of inflammatory factors (Tang and Ran, 2018; Gu et al., 2021). In particular, HIF-2α increases the protein expression of IL-6, and it is HIF-2a rather than HIF-1a that further causes liver lipid accumulation (Qu et al., 2011).

Firstly, we applied CCK8 method to explore the cell concentration of drugs used in this validation experiment, such as SX-1, CoCl₂, stattic and l-Glutathione (Glu). Compared with the control group, 10% rat serum had no effect on cell viability (p = 0.126; Figure 6A). Therefore, 10% rat con serum was used as the control in the subsequent groping for a suitable concentration of SX-1. Compared with 10% rat con serum, 10% SX-1 acting on cells for 24 h showed a protective effect (p<0.001; Figure 6B). Similarly, 40 μM Glu had a significant protective effect on cells compared with the control group for 24 h (p = 0.028; Figure 6C). While CoCl₂ inhibited cell growth in a concentration dependent manner. When the concentration of CoCl₂ was 400 μM, the cell viability was significantly lower than that of the control group for 48 h (p = 0.020; Figure 6D); while 10 μM stattic began to reduce cell viability when it was applied to cells for 48 h (p<0.001; Figure 6E).

Then, we established AML12 cells hypoxia model with CoCl₂, which increased HIF-2α and IL-6 level (p = 0.045 and p = 0.046, respectively), and then treated the cells with SX-1 drug-containing serum. SX-1 can significantly reduce HIF-2α and IL-6 level (p = 0.017, Supplementary Figure S3; p = 0.049, Figure 7A). Next, we used Stattic, a small molecule inhibitor that effectively inhibits STAT3 activation and nuclear translocation, to explore whether SX-1 has the effect of reducing STAT3 activation. Under the action of CoCl₂, the level of p-STAT3 was significantly upregulated compared with the DMSO group (p = 0.004). However, after SX-1 and stattic intervention, both of them significantly reduced the expression of p-STAT3 (p = 0.002 and p = 0.003; Figure 7B). In accordance, we found that CoCl₂ induced nuclear pyknosis and cell swelling in AML12 cells compared with the control group under electron microscope, while SX-1 and stattic intervention alleviated nuclear pyknosis and cell swelling to some extent (Figure 7C). This suggests that p-STAT3 is a key target of SX-1, which plays a role in saving hepatocyte injury by reducing the expression of p-STAT3.