Chemicals and reagents
CTX was purchased from Jiangsu Hengrui Medicine Co., Ltd. (Lianyungang, China). The ginseng sample (JSPACM-03-1) was collected from Jilin province, the authentic origin of China, and authenticated by Dr. Song-Lin Li. The voucher specimen was deposited in Department of Metabolomics, Jiangsu Province Academy of Traditional Chinese Medicine. Reference compounds, including ginsenoside Re, Rg1, Rf, Ro, Rb1, Rc, Rb2 and Rd, were purchased from Sichuan Victory Co. Ltd. (Chengdu, China). Murine 4T1 (mouse mammary carcinoma) cell was purchased from Keygen Biotech Co., Ltd. (Nanjing, China), and FD-4 were obtained from Sigma (St. Louis, USA). Antibodies for ZO-1, Occludin and E-Cadherin were obtained from Abcam Inc. (CA, USA), and for ZO-2, NFκB and Nrf2 were purchased from Santa Cruz Biotechnology (CA, USA). All other reagents were obtained from commercial sources.
Preparation of CTX and ginsenosides
CTX was dissolved in physiological saline for intravenous (i.v.) administration, and ginsenosides sample was prepared according to our previous protocol  and suspended in 2% CMC-Na solution for intragastric (i.g.) administration. The composition and structural information of the ginsenosides were determined by our published UPLC-QTOF-MS method . The contents of Re, Rg1, Rf, Ro, Rb1, Rc, Rb2 and Rd were 2837μg/g, 6784μg/g, 1043μg/g, 1935μg/g, 1937μg/g, 1043μg/g, 811μg/g and 1225μg/g, respectively.
Animal experimental design
Female ICR Mice (weight 18–22g) were supplied by The Chinese University of Hong Kong (Hong Kong SAR, China) or Comparative Medicine Centre of Yangzhou University (Yangzhou, China) and maintained in a pathogen-free environment, air-conditioned at 24 ± 2℃ with a standard 12h light–dark cycle. Five mice were housed in per cage and allowed access to distilled water and standard pellet diet ad libutum. Animal experiments were approved and performed in accordance with the guidelines of Institutional Animal Care and Use Committee at the Chinese University of Hong Kong (Ref No. (16-710) in DH/HA&P/8/2/1 Pt.63) or Animal Ethics Committee and Institutional Animal Care and Use Committee at Jiangsu Province Academy of Traditional Chinese Medicine (AEWC-20150705-4).
Mice were randomly divided into 6 groups, which were BLK, CTX, CTGS, GS, FT-CTX and anti-CTGS groups. Every group contains 10 animals, all mice were subcutaneously inoculated murine mastadenoma 4T1 cells (2 × 106) in the right armpit, and treatments were performed when tumors reached 200 ± 30mm3. The dosage regimens of all groups were shown in Figure 8. The fresh feces solution of GS was prepared referring to the published protocol , and broad-spectrum antibiotics including vancomycin (100mg/L), neomycin (200mg/L), ampicillin (200mg/L), and metronidazole (200mg/L) were dissolved in sterile drinking water according to previous protocol  and changed day.
Tumor Growth Evaluation and Immunity Assessment
Mouse behavior was observed every day, body weight and tumor size were measured every 5d. On the first day of the treatment, blood samples of the first four groups mice were collected for analysis at 4h after CTX administration, and on the twentieth day of the treatment, blood, feces, intestine, 4T1 mastadenoma tissues, spleen and thymus of all mice were collected after humanely sacrificed. The tumor volume, survival rate, tumor inhibition rate as well as thymus and spleen indexes were calculated. Besides, serum IL-2, IL-6, INF-γ and IL-17 were measured using an assay kit according to the manufacturers' recommendations.
Histological and Immunohistochemical Assessments
Fixed hepatic tissues were embedded in paraffin, cut into 5μm sections, stained with H&E or stained with primary antibodies and mounted with HRP conjugated secondary antibody, and observed using a light microscope.
Diarrhea Assessment and Intestinal Permeability Assay
For diarrhea assessment, stool was scored on day 20, with 0 indicating normal, 1 indicating slight diarrhea (slightly wet and soft stool), 2 indicating moderate diarrhea (wet and unformed stool), and 3 indicating severe diarrhea (watery stool with severe perianal staining) as previously described . For intestinal permeability assay, each animal was gavaged with 6mg/kg FD-4 after last dosing. Serum samples were obtained by cardiac puncture at 4h after oral administration , and fluorescence intensity of each serum sample was measured by DTX 880 Multimode Detector (Beckman Coulter, CA, USA).
Western Blot Analysis
Intestinal or tumor tissue homogenates were prepared with RIPA Buffer. Denatured total protein was separated by 10% sodium dodecyl sulfate polyacrylamide gel electrophoresis and transferred onto a polyvinylidene difluoride membrane. The membrane was blocked by 5% fetal bovine serum and incubated with appropriate dilutions of primary antibody, followed by incubation with rabbit anti-mouse or goat anti-mouse HRP-conjugated secondary antibody. The intended protein was detected using ECL kit and normalized to the corresponding GAPDH expression.
Intestinal microbiota analysis
Changes of intestinal microbiota profiles in response to treatments were determined by analyzing bacterial DNA of feces following our protocols . Fastq files generated through Illumina were demultiplexed and quality filtered, OTUsGS assigned using SILVA database, and the data processed through Mothur 1.40.1.
Data are expressed as means ± standard deviation. To compare samples collected from any groups, a non-parametric paired Student’s t-test (Wilcoxon matched-pairs signed rank test) was performed. All the univariate statistical analyses were performed using either with SPSS software (version 24.0) or GraphPad Prism (version 7) for windows. All statistical tests were two-sided and differences were considered statistically significant if p < 0.05.