CY (cat. no. S30563) was obtained from Shanghai Yuanye Biotechnology Co., Ltd. The cell apoptosis kit (Annexin V-FITC/PI; cat. no. 556547) was supplied by BD Pharmingen (BD Biosciences). Blood routine reagent kits (Staining reagent kit, cat. no. ZY4203 and Reaction reagent kit, cat. no. ZY4224) was purchased by IDEXX Laboratories Inc. The PE-conjugated anti-TER-119 antibody (cat. no. 116207) was obtained from BioLegend, Inc.

Dry Panax ginseng was purchased from the Wanliang ginseng market (Tonghua, China). The ginseng was boiled three times to obtain the GE (6). The extracting solution of GE was eluted with different concentrations of ethanol (80-95%), before being centrifuged at 1,200 × g at 4°C for 20 min to obtain the supernatant and precipitates. The supernatant was sequentially decolorized with a decolorization resin (cat. no. D941; Tianjin Haoju Resin Technology Co., Ltd.) and activated carbon, before being filtered and dried at 40°C to obtain TG (26). The precipitates were sequentially deproteinized, decolorized and dialyzed to remove components with a molecular weight of <3,500 Da to obtain the TP (27). The composition and content of ginsenosides and polysaccharides in the GE, TG and TP have been reported in a previous study (6,26). The ginsenoside content of GE and TG is 4.39 and 81.09%, respectively. By contrast, the polysaccharide content of GE, TG and TP is 72.28, 4.68 and 89.79%, respectively (26).

A total of 50 male 4-week-old Kunming mice (weight, 18-20 g) were purchased from Liaoning Changsheng Biotechnology Co., Ltd. [animal license no. SCXK (Liao)-2015-0001]. The mice were maintained under the conditions of controlled light (12 h light/dark), temperature (25±1°C) and humidity (60%±5%), with ad libitum access to food and water. After 5 days of acclimation, mice were randomly divided into the following five groups (n=10/group): i) The normal group (Control); ii) the group treated with CY (Model); iii) the CY + GE group (GE, 1.0 g/kg); iv) the CY + TG group (TG, 0.25 g/kg); and v) the CY + TP group (TP, 1.0 g/kg).

The mice were intragastrically administered GE, TG and TP for 28 days in the GE, TG and TP groups (0.1 ml/10 g body weight) (7,28), whereas mice in the other groups were administered equivalent volumes of normal saline. Mice in the Model, GE, TG and TP groups were hypodermically injected with CY saline solution (50 mg/kg) on days 25, 26, 27 and 28. The dose of GE was calculated according to the human adult daily dose of ginseng (9 g/day, 60 kg/body weight) (29), whereas the doses of TG and TP (1.0 g/kg) were selected according to previous studies (6,7).

The fresh feces of mice were collected on day 28 and stored at -80°C until further analysis. All mice were then euthanized by cervical dislocation under anesthesia with 30 mg/kg pentobarbital sodium by intraperitoneal injection after fasting for 12 h. If an animal reached the predefined humane endpoints [loss of >15% of body weight in 1-2 days or an overall reduction of >20% in body weight; or displaying obvious signs of suffering (lethargy, squinted eyes, dehydration and hunched back)], they were humanely euthanized as aforementioned. Animal death was confirmed by cessation of respiration and heartbeat. The study protocol was approved by the Ethics Committee of Changchun University of Chinese Medicine (protocol no. 2023033; Changchun, China). No mice showed abnormal signs or reached humane endpoints throughout the experiment.

Anesthesia with 30 mg/kg pentobarbital sodium (concentration 5 mg/ml, dose 60 µl/10 mg body weight) was induced by intraperitoneal injection on day 29. Subsequently, ~0.5 ml blood was collected from the retro-orbital vein (30). The mice were then sacrificed after this blood collection. The blood cell parameters [WBC, neutrophils, lymphocytes (LYMPH), RBC, hemoglobin (HGB), reticulocytes (Ret) and PLT] were detected using the ProCyte DX hematology analyzer (IDEXX Laboratories Inc.).

A QIAamp DNA Stool Mini Kit (cat. no. 51504; Qiagen, Inc.) was used to extract and purify the DNA of the fecal samples (7 samples from 7 mice each group). The primers (341F: 5′-CCT AYG GGR BGC ASC AG-3′ and 806R: 5′-GGA CTA CNN GGG TAT CTA AT-3′) were used to amplify the 16S rRNA genes in the V3-V4 region. The library construction (NEBNext^® Ultra™ IIDNA Library Prep Kit; cat. no. E7645; New England BioLabs, Inc.) and RNA sequencing were performed by Novogene Bioinformatics Technology Co., Ltd. The detailed steps of this sequencing procedure have been previously reported (6,31). For the effective tags obtained previously, denoise was performed with DADA2 or deblurmodule in the QIIME2 software (version QIIME2-202006; https://library.qiime2.org) to obtain initial amplicon sequence variants (ASVs; default: DADA2), and then ASVs with abundance <5 were filtered out. the annotation database was the Silva Database (https://www.arb-silva.de/). The following analyses were performed: i) Principal co-ordinates analysis was performed according to the weighted UniFrac distance matrices; ii) α-diversity was calculated from 4 indexes in QIIME2, including Chao1, Dominance, Pielou_e, and Shannon; iii) cluster analysis was performed with principal component analysis (PCA), which was applied to reduce the dimension of the original variables using the ade4 package and ggplot2 package in R software (version 3.5.3); iv) the significantly different species (P<0.05) at each taxonomic level (phylum, class, order, family, genus and species) were analyzed using the R software (version 3.5.3), P<0.05 was considered to indicate a statistically significant difference; and v) the LEfSe software (version 1.0) was used to perform LEfSe analysis] linear discriminant analysis (LDA) score >3; Kruskal-Wallis test, false discovery rate P<0.05] so as to identify the biomarkers.

Bone marrow from the right femur of the mice in the Control, Model, GE, TG and TP groups (n=10/group) was washed with PBS into the EasySep™ buffer (cat. no. 20144; Stemcell Technologies, Inc.) at room temperature using a syringe equipped with a 23-gauge needle to maintain the viability of bone marrow cells. The remaining aggregates and debris were removed by passing the cell suspension through a 70-µm mesh nylon strainer, before the sample was centrifuged at 300 × g for 10 min at room temperature. The cells were then resuspended at 1×10⁸ nucleated cells/ml in DMEM/F12 (cat. no. 11-330-032; Gibco; Thermo Fisher Scientific, Inc.). The HSCs from bone marrow mononuclear cells were purified using the EasySep™ Mouse Hematopoietic Progenitor Cell Isolation Kit according to manufacturer's protocol (cat. no. 19856; Stemcell Technologies, Inc.). HSCs were adjusted to 1×10⁵ cells/ml in PBS and were then labeled using Annexin V-FITC/PI (cat. no. c1052) for 20 min at room temperature. A total of 10,000 events were acquired to analyze cell apoptosis using a DxFLEX flow cytometer and CytExpert Software version 2.4.0.28 (both Beckman Coulter, Inc.).

HSCs (1×10⁵) were incubated with the PE-conjugated anti-TER-119 antibody (1:1,000) in the dark for 40 min on ice. Subsequently, TER-119 expression was detected using a DxFLEX flow cytometer and CytExpert software version 2.4.0.28 (both Beckman Coulter, Inc.).

Total RNA was extracted from HSCs according to the instruction manual of the TRlzol reagent (cat. no. AM9738; Thermo Fisher Scientific, Inc.). RNA concentration and purity was measured using a NanoDrop 2000 (Thermo Fisher Scientific, Inc.). RNA integrity was assessed using the RNA Nano 6000 Assay Kit (cat. no. 5067-1511) of the Agilent Bioanalyzer 2100 system (both Agilent Technologies, Inc.). A total of 1 µg RNA per sample was used as input material for the RNA sample preparations. Sequencing libraries were generated using NEBNext UltraTM RNA Library Prep Kit (cat. no. E7530S/L; New England BioLabs, Inc.) for Illumina following the manufacturer's recommendations. To select cDNA fragments preferentially of 240 bp in length, the library fragments were purified with AMPure XP system (Beckman Coulter, Inc.). Then, 3 µl USER Enzyme (cat. no. E7530S/L; New England BioLabs, Inc.) was used with size-selected, adaptor-ligated cDNA at 37°C for 15 min followed by 5 min at 95°C before PCR. PCR was then performed using Phusion High-Fidelity DNA polymerase (cat. no. M0530S; New England BioLabs, Inc.) with the following reaction conditions: 98°C for 30 sec; 25 cycles of 98°C for 10 sec and 72°C for 15 sec; 72°C for 5 min, 4°C for hold. At last, the PCR products were purified (AMPure XP system) and the library quality was assessed on the Agilent Bioanalyzer 2100 system. The loading concentration of the final library was 2 nM. The clustering of the index-coded samples was performed on a cBot Cluster Generation System using TruSeq PE Cluster Kit v4-cBot-HS (Illumina) according to the manufacturer's instructions. After cluster generation, the library preparations were sequenced on an Illumina platform and paired-end reads were generated. The libraries were constructed and sequenced by Beijing Biomarker Technologies Co., Ltd. (www.biocloud.net).

After sequencing, bioinformatics analysis was performed using BMKCloud (www.biocloud.net) to identify the DEGs. Differential expression analysis of two conditions/groups was performed using DESeq2 (version 1.30.1; https://www.bioconductor.org/packages/release/bioc/html/DESeq2.html). An adjusted P<0.05 and a fold change (FC) of >1.5 were considered to indicate a significant difference. Gene Ontology (GO) enrichment analysis of DEGs was implemented by the GOseq R packages-based Wallenius non-central hyper-geometric distribution (32). KOBAS software (33) was used to test the statistical enrichment of differential expression genes in Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways (http://www.genome.jp/kegg/). Heatmap analysis of DEGs was conducted according to the FPKM value using the cluster Profiler package (version 3.0.3) in the R software.

Total RNA was extracted from HSCs using TRIzol^® reagent (Invitrogen; Thermo Fisher Scientific, Inc.) and cDNA was obtained using the FastKing RT Kit (Tiangen Biotech Co., Ltd.) according to the manufacturer's protocol. Subsequently, qPCR was performed using the CFX Connect Real-Time PCR Detection System (Bio-Rad Laboratories, Inc.) with SuperReal PreMix Plus (SYBR Green; Tiangen Biotech Co., Ltd.). The primer sequences were as follows: β-actin forward, 5′-CTG TCC CTG TAT GCC TCT G-3′ and reverse, 5′-ATG TCA CGC ACG ATT TCC-3′; Hhex forward, 5′-CCA CCC GAG AGA AAG CGT CTG-3′ and reverse, 5′-TGC GTT GGA CAG TTT GGA CAC T-3′; Klf4 forward, 5′-GCG GGA AGG GAG AAG ACA CTG CGT C-3′ and reverse, 5′-TAG GAG GGC CGG GTT GTT ACT GCT-3′; Pira2 forward, 5′-ACT ACT GGA CAC CCA GCC TT-3′ and reverse, 5′-TGA ACC TGT CAT AGC TCG GC-3′; and Pbx1 forward, 5′-TGA AGC CTG CCT TGT TTA ATG T-3′ and reverse, 5′-ATG TTG TCC AGT CGC ATG AGC-3′. The thermocycling conditions were as follows: 95°C for 15 min, followed by 40 cycles at 95°C for 10 sec, 55°C for 20 sec and 72°C for 30 sec. The transcript levels were quantified and normalized to the internal reference gene β-actin using the 2^−ΔΔCq method (34).

The bone marrow from the left femurs of mice in the Control, Model, GE, TG and TP groups (n=6/group) was fixed in 10% formalin solution for 5 days at room temperature and demineralized with 10% EDTA for 14 days at room temperature. The samples were then embedded in paraffin and sectioned into 4 µm slices. Bone marrow tissue sections were dewaxed and then treated with high-pressured 2% EDTA (pH 9.0) antigen retrieval buffer at 95°C for 20 min for antigen retrieval. The tissue was incubated in Endogenous Peroxidase Blocking Buffer (cat. no. P0100A; Beyotime Institute of Biotechnology) at room temperature for 10 min to block and eliminate the interference of endogenous peroxidase. Subsequently, the sections were blocked with 3% BSA (cat. no. B24726; Shanghai Angyi Biotechnology Co., Ltd.) for 1 h at room temperature, and then incubated with primary antibodies overnight at 4°C. The sections were then incubated with secondary antibody [goat anti-rabbit IgG (H+L); cat. no. SA00001-2; Proteintech Group, Inc.; diluted to 1:1,000 in 3% BSA] at room temperature for 30 min. Finally, the binding antibody was detected by DAB (cat. no. DAB-2031; Fuzhou Maixin Biotechnology Development Co., Ltd.) staining at room temperature for 5 min and hematoxylin counterstaining at room temperature for 2 min. Negative (no primary antibody) control staining of one mouse was used in each experiment. Images were captured using the Axioscan 7 fully automatic digital slide scanning system (Zeiss AG). Rabbit anti-mouse Klf4 (cat. no. 11880-1-AP; Proteintech Group, Inc.) and rabbit anti-mouse Pbx1 (cat. no. 18204-1-AP; Proteintech Group, Inc.) antibodies were used as primary antibodies to detect protein expression in PBS-Tween (1%) at a concentration of 1:200.

Western blot analysis was performed as previously reported (35). Briefly, the proteins were extracted from the bone marrow cells of mice tibia and femur, and then lysed in RIPA lysis buffer (cat. no. P0013B; Beyotime Institute of Biotechnology) containing protease inhibitors (cat. no. 05892970001; Roche Diagnostics) at 4°C for 5 min. Subsequently, 35 µg protein was separated by 12% SDS-PAGE and then transferred onto PVDF membranes. The membranes were blocked with 3% BSA for 1 h at room temperature, and then incubated with the following primary antibodies (all from Proteintech Group, Inc.) overnight at 4°C: β-actin (1:2,000; cat. no. BS6007M), Bcl2 (1:2,000; cat. no. 26593-1-AP) and Bax (1:2,000; cat. no. 50599-2-Ig). The membranes were then incubated with a secondary antibody [goat anti-rabbit IgG (H+L) 1:5,000; cat. no. SA00001-2; Proteintech Group, Inc.] for 2 h at room temperature. The bands were visualized using an ECL luminescence reagent (cat. no. P001-500; Hunan Hui Bai Shi Biological Technology Co., Ltd.) a ChemiDoc™ MP imaging system (Bio-Rad Laboratories, Inc.) at room temperature. The relative protein expression levels were normalized to β-actin.

The DEGs were filtered according to fold change ≥1.5 and P<0.05. Quantitative data are presented as the mean ± SD. The relative abundance of intestinal microflora was compared between groups using one-way ANOVA followed by Dunnett's test, whereas other data were analyzed with one-way ANOVA followed by Tukey's test using GraphPad Prism 8.0 software (Dotmatics). Correlation analysis was performed using Spearman's rank correlation coefficient test. P<0.05 was considered to indicate a statistically significant difference.

Fig. 1A shows the number of shared and unique ASVs among the five treatment groups in a Venn diagram. There were 343, 741, 469, 539 and 624 unique ASVs in the Control, Model, GE, TG and TP groups, respectively. The principal co-ordinates analysis showed that the microbial community had some differences between the Control and Model or TP groups (Fig. 1B). The α diversity reflects the richness and evenness of intestinal microflora. The Chao 1 index shows the community richness, which is positively associated with the number of ASVs. The dominance index and pielou_e index show the community evenness, whilst the Shannon index shows the richness and evenness of ASVs. As shown in Fig. 1C, CY markedly decreased the Chao 1 index compared with that in the Control group. In addition, GE, TG and TP increased the Chao 1, pielou_e and Shannon indices, whilst reducing the dominance index compared with the Model group; however, these differences were not statistically significant. These results indicated that GE, TG and TP markedly increased the community richness and evenness of ASVs in model mice. Notably, 7 samples from 7 mice in the TP group showed the best uniformity (Fig. 1B and C).

The present study next analyzed the composition and abundance of the top 10, 15 and 15 bacteria at the phylum, genus and species levels, respectively. CY increased the relative abundance of Bacteroidota whilst decreasing that of Firmicutes at the phylum level (Fig. 2A). GE was found to significantly elevate the abundance of Firmicutes compared with that in the Model group (Fig. 2B). TG and TP also increased the expression of Firmicutes, whereas GE and TG also reduced the abundance of Bacteroidota. However, the difference was not statistically significant (Fig. 2B). Fig. 2C and D show the top 15 genera at the genus level clustered according to the relative abundance among the five groups (n=35 samples). CY significantly reduced the relative abundance of Lactobacillus, whilst increasing that of Bacteroides, Alloprevotella and Muribaculaceae (Fig. 2E). GE significantly increased the relative abundance of Lactobacillus. By contrast, TG also increased the relative abundance of Lactobacillus compared with that in the Model group, but the difference was not significant (Fig. 2E). In addition, GE and TG were found to reduce the relative abundance of Clostridia_UCG-014 and Odoribacter compared with the Model group (Fig. 2E). These results suggest that GE at least partially reversed the abnormal changes in some intestinal microflora, followed by TG and TP.

At the species level, CY increased the relative abundance of Bacteroides_sartorii, Candidatus_arthromitus, Bacteroides_vulgatus and Mucisporollum_sp., whilst decreasing that of Lactobacillus_ intestinalis and Metamycoplasma_sualvi (Fig. 2F). GE, TG and TP could reverse these aforementioned changes in the intestinal microflora. Notably, GE and TG significantly increased the abundance of Lactobacillus_intestinalis compared with that in model group (Fig. 2G). In addition, Lactobacillus_intestinalis was significantly positively correlated with ginsenoside content (r=0.8705; Fig. 2H). Furthermore, TP increased the abundance of low-abundance microflora compared with the Control group, such as Mouse_gut, Bacteroides_dorei, Prevotella_melanimogenica, Anaerotignum_lactatifermentans, Bacteroides_plebeius and Parabacteroides_merdae. Although CY and TP increased the abundance of some low-abundance microflora, the bacterial species were different. These results showed that the intestinal bacteria of model mice treated with GE, TG and TP were markedly improved. In general, the relative abundance and species of intestinal bacteria recovered by GE were consistent with the Control group at the phylum, genus and species levels.

The key microflora (biomarkers) were next identified using LDA (Fig. 3A) and an evolutionary branch tree (Fig. 3B). A total of 4, 9 and 48 key microflora (LDA score >3) were detected in the GE, TG and TP groups, respectively (Fig. 3A). s_Vibrio_metschnikovii, s_Streptococcus_hyointestinalis, s_Clostridium_ fusiformis and s_Eubacterium_plexicaudatum were key species in the GE group. The key species in the TG group included f_Prevotellaceae, s_Lactobacillus_ intestinalis, g_Alloprevotella and o_Burkholderiales. By contrast, c_, p_ and o_Bacteroidales, f_Muribaculaceae and g_Muribaculaceae were higher in the TP group compared with that in the other groups.

A total of 27,684 genes were annotated and used for further analysis. There were 4,988 DEGs in all samples. PCA showed marked differences between the Model/TP group and GE/TG groups (Fig. 4A). Fig. 4B shows that a total of 1,746 DEGs were obtained between the Model and Control groups, with 716 upregulated and 1,030 downregulated. There were 2,250 DEGs between the Model and GE groups, including 1,258 upregulated and 992 downregulated DEGs. The highest number of DEGs was detected between the Model and TG groups, including 1,833 upregulated DEGs and 1,599 down-regulated DEGs. The lowest number of DEGs was detected between the Model and TP groups, including 75 upregulated DEGs and 186 downregulated DEGs. A Venn diagram shows the numbers of upregulated DEGs (Fig. 4C) and downregulated DEGs (Fig. 4D) among the groups. There were 4 upregulated DEGs and 1 downregulated DEG in common between the Model and Control/GE/TG/TP (Fig. 4C and D). Cluster analysis was performed for the five groups based on 4,988 DEGs, where the profiling of the DEGs was found to be similar between the TG and GE groups or between the Model and TP groups (Fig. 4E).

In the top 20 biological processes, upregulated genes in the Control group compared with the Model group (Fig. 5A) were mainly enriched in immune regulation (including 'immune response', 'neutrophil chemotaxis', 'myeloid leukocyte differentiation', 'acute inflammatory response' and 'B cell proliferation involved in immune response') and hemopoiesis regulation (including 'negative regulation of cell proliferation', 'response to hypoxia', 'hemopoiesis', 'positive regulation of nitric oxide biosynthetic process', 'cellular response to decreased oxygen levels', 'cellular response to oxygen levels' and 'cellular response to macrophage colony-stimulating factor stimulus'). In the top 20 biological processes, the upregulated genes in the GE group compared with the Model group (Fig. 5B) were enriched in immune regulation ('transforming growth factor beta receptor signaling pathway' and 'negative regulation of interleukin-6 production') and hemopoiesis regulation ('small GTPase mediated signal transduction', 'positive regulation of blood vessel endothelial cell migration', 'regulation of small GTPase mediated signal transduction', 'cellular response to vascular endothelial growth factor stimulus' and 'positive regulation of nitric-oxide synthase biosynthetic process'). Compared with the Model group, the upregulated genes in the TG group (Fig. 5C) were related to 'inflammatory response', 'small GTPase mediated-signal transduction' and 'positive regulation of nitric-oxide synthase biosynthetic process' in the top 20 biological processes. Compared with the Model group, the upregulated genes in the TP group (Fig. 5D) were enriched in immune regulation ('negative regulation of interleukin-10 biosynthetic process', 'cellular response to interleukin-8', 'regulation of natural killer cell differentiation' and 'regulation of extrathymic T cell differentiation') and 'negative regulation of parkin-mediated stimulation of mitophagy in response to mitochondrial depolarization' terms. These results suggest that CY mainly affected the immune regulation and hematopoietic function of HSCs. In addition, the upregulated genes in the GE and TG groups showed similar enrichment results regarding biological processes (Fig. 5B and C, pink rectangles).

In the cellular component analysis, compared with the Model group, the upregulated genes in the GE and TG groups were mainly enriched in the 'cytosol', 'nucleoplasm', 'membrane', 'Golgi apparatus' and 'cell surface' (Fig. 5B and C), where the upregulated genes in the GE group were also enriched in the 'nucleus'. By contrast, the upregulated genes in the TP group were primarily related to the 'centrosome', 'cytoplasmic vesicle' and 'postsynaptic membrane' compared with the Model group (Fig. 5D). The main cellular component enriched by TP were different from GE and TG.

In the molecular function analysis, the upregulated genes in the Control, GE, TG and TP groups were mainly enriched in 'ATP binding' (Fig. 5A-C). In particular, the upregulated DEGs in the GE and TG groups were also enriched in 'zinc ion binding,' 'GTPase activator activity', 'actin binding', 'protein kinase activity', 'guanyl-nucleotide exchange factor activity', 'transcription corepressor activity', 'integrin binding', 'SMAD binding', 'cytokine receptor activity' and 'Rho GTPase binding' (Fig. 5B and C). However, TP showed obvious differences compared with the Control, GE and TG groups regarding molecular function. The upregulated genes in the TP group were mainly enriched in 'protein serine/threonine kinase activity', 'promoter-specific chromatin binding' and 'phosphorylase kinase regulator activity' (Fig. 5D). The majority of the genes enriched in the GE and TG groups were consistent in the top 20 GO analysis.

A total of 4,988 DEGs were involved in 297, 282, 306 and 125 KEGG pathways in the Control, GE, TG and TP groups, respectively. The DEGs were divided mainly into four branches based on KEGG annotations: Cellular processes, environmental information processing, human diseases and organismal systems. In cellular processes, the upregulated DEGs were enriched in 'endocytosis' and 'apoptosis' pathways in the Control, GE, TG and TP groups (Fig. 6A-D). Furthermore, 20 and 37 DEGs were enriched in 'signaling pathways regulating pluripotency of stem cells' in the GE and TG groups (Fig. 6B and C). In environmental information processing, the DEGs in the GE and TG groups were enriched mainly in the 'MAPK signaling pathway', 'PI3K-Akt signaling pathway', 'Rap1 signaling pathway', 'cytokine-cytokine receptor interaction', 'Ras signaling pathway', 'Wnt signaling pathway', 'apelin signaling pathway', 'JAK-STAT signaling pathway' and 'NF-kappa B signaling pathway'. The majority of these pathways are involved in the regulation and differentiation of HSCs (Fig. 6B and C) (36,37). The DEGs in the TP group were also mainly enriched in the 'MAPK signaling pathway', 'NF-kappa B signaling pathway', 'Rap1 signaling pathway', 'Wnt signaling pathway', 'apelin signaling pathway' and 'cell adhesion molecules'. In organismal systems, the DEGs were involved in 'osteoclast differentiation', 'Chemokine signaling pathway', 'C-type lectin receptor signaling pathway' and 'NOD-like receptor signaling pathway' in the Control, GE, and TG groups (Fig. 6A-C). TP exhibited notable differences compared with the GE and TG groups regarding organismal systems, where the DEGs were enriched in 'parathyroid hormone synthesis, secretion and action' and 'aldosterone synthesis and secretion' (Fig. 6D). The DEGs from the GE and TG groups participated mainly in regulating HSCs-related signaling pathways, such as 'MAPK signaling pathway', 'PI3K-Akt signaling pathway', 'Rap1 signaling pathway', 'cytokine-cytokine receptor interaction', 'Ras signaling pathway', 'Wnt signaling pathway', 'JAK-STAT signaling pathway' and 'NF-kappa B signaling pathway'. In addition, TP modulated HSCs-related signaling pathways. However, there were clear differences between the GE/TG and TP groups in terms of the profile of DEGs.

According to the GO analysis, the hematopoietic-related term of the upregulated DEGs had 'hemopoiesis', 'embryonic hemopoiesis', 'post-embryonic hemopoiesis', 'definitive hemopoiesis', 'regulation of hemopoiesis', 'negative regulation of hemopoiesis' and 'positive regulation of hemopoiesis'. Compared with the Model group, 9, 19, 23 and 1 DEGs were enriched in hematopoietic-related term in the Control, GE, TG and TP groups, respectively (Table SI). A total of 35 upregulated hematopoietic-related genes were clustered according to FPKM (Fig. 7A; Table SI). As shown in Fig. 7B, CY markedly decreased the expression levels of Klf4, Hhex and Pira2, whilst increasing those of Flt3 and Klf13, compared with those in the Control group. These results indicated that CY affected HSCs. GE and TG significantly increased Klf4 and Hhex expression compared with that in the Model group, whereas the expression levels of Hhex and Klf4 were higher in the GE group compared with those in the TP group (Fig. 7B). The expression levels of Pbx1, Kmt2a, Mecom, Zc3h12a, Zbtb16, Lilr4b, Flt3 and Klf13 were higher in the GE and TG groups than those in the Model and TP groups (Fig. 7B). In addition, TG markedly increased the expression of Pira2 compared with that in the Model, GE and TP groups (Fig. 7B). Vpreb1 expression in the Control and TG groups was higher compared with that in the Model group, but the difference was not statistically significant (Fig. 7B). These results suggest that both TG and GE activated the expression of hematopoietic-related genes, though TP conferred almost no activating effect on the HSCs.

The mRNA expression levels of several genes were next detected in HSCs. As shown in Fig. 7C, CY markedly decreased the expression levels of Klf4, Hhex, Pira2 and Pbx1. In the drug administration groups, GE significantly upregulated the expression levels of Klf4, Hhex and Pbx1 compared with those in the Model and TP groups (Fig. 7C). TG also increased Klf4, Hhex and Pira2 expression compared with that in the Model and TP groups (Fig. 7C). The expression of Pbx1 in HSCs treated with TG was also significantly higher compared with that in the Model group (Fig. 7C). TP only significantly increased the expression of Hhex compared with the Model group (Fig. 7C). Other genes measured in the TP group exhibited slightly higher expression compared with those in the Model group, but no statistical significance could be found. In addition, the protein expression levels of Klf4 and Pbx1 were detected by immunohistochemistry. As shown in Fig. 8A and B, positive Klf4 and Pbx1 protein expression was markedly increased in the TG and GE groups compared with that in the Model group. The aforementioned results appeared to be consistent with the results of RNA-seq analysis.

CY was found to promote the apoptosis of HSCs at the early and late apoptosis stages, whereas GE, TG and TP could inhibit the early apoptosis, late apoptosis and total apoptosis of HSCs induced by CY (Fig. 9A). TG exhibited the strongest effect on inhibiting cell apoptosis, whereas the effect of TG and GE on inhibiting cell apoptosis was significantly superior to TP (Fig. 9A). Furthermore, the expression levels of anti-apoptosis (Bcl2 and Mcl1) and pro-apoptosis (Bax, Bid and Bad) DEGs were next analyzed. The results showed that CY significantly inhibited the expression of Mcl1 and increased the expression of Bax (Fig. 9B). TG significantly increased the expression levels of Bcl2 and Mcl1, whilst decreasing those of Bax compared with the Model group. GE significantly inhibited the expression levels of Bax and Bad compared with those in the Model and TP groups (Fig. 9B). In addition, the expression levels of Bid were significantly lower in the GE group compared with those in the TP group (Fig. 9B). Notably, TP had little effect on the expression levels of apoptotic genes compared with the Model group. According to the western blotting analysis, TG increased the protein expression levels of Bcl2, whereas GE and TG decreased the protein expression levels of Bax compared with the Model group (Fig. 9C). These results suggest that both GE and TG inhibited the apoptosis of HSCs. However, the inhibitory effect of TP on cell apoptosis was weaker compared with that of GE and TG.

TER-119 has been previously used as a marker of erythroid differentiation of HSCs (38). The present study therefore detected TER-119 expression to assess the differentiation ability of HSCs induced by CY. The results showed that CY significantly reduced the expression of TER-119 compared with that in the Control group (Fig. 10A). By contrast, GE markedly increased the expression of TER-119 compared with that in the Model and TP groups (Fig. 10A). The expression of TER-119 in the TG and TP groups was not statistically different compared with that in the Model group. In the blood cell parameter analysis, the numbers of all peripheral blood cells were found to be significantly reduced in the Model mice (Fig. 10B), which were all significantly reversed by GE and TG [namely WBC, neutrophils, LYMPH, RBC, HGB, Ret and PLT; Fig. 10B]. By contrast, TP could only restore the levels of LYMPH, RBC, HGB and PLT compared with those in the Model group (Fig. 10B). TER-119 was expressed in Ret from HSCs (39), and this increasing trend in TER-119, Ret and RGB is consistent among the GE, TG, and TP groups. Notably, the improvement effect of TP on blood cells was weaker compared with that of GE and TG.