Ziyuglycoside II (purity > 98%) was obtained from the National Pharmaceutical Engineering Center for Solid Preparation in Chinese Herbal Medicine (Nanchang, China). A stock solution (100 mg/mL) was prepared in sterile 0.9% (w/v) sodium chloride (NaCl). Cyclophosphamide (Baxter Oncology GmbH, Frankfurt, Germany) was diluted in 0.9% NaCl for in vivo administration. Recombinant human granulocyte colony-stimulating factor (rhG-CSF) was purchased from Amoytop Biotech Co., Ltd. (Xiamen, China). The antibodies used for flow cytometer were purchased from eBioscience (Thermo Fisher, USA). The primary antibodies used in Western blot were purchased from Proteintech (Wuhan, China).

Specific pathogen-free (SPF) C57Bl/6J mice (male; 6–8 weeks old; 18 ± 2 g) were obtained from Beijing Kerisi Company Biotechnology Co., Ltd. (License: SCXK(Jing) 2019-0010). Mice were housed under controlled conditions (22 ± 1 °C, 55 ± 5% humidity, 12 hr light/dark cycle). All animal experiments were conducted in the AAALAC-accredited facility of Jiangxi Science and Technology Normal University (IACUC Approval No. Y202340).

After one week of acclimatization, the mice were randomly assigned into five groups: Vehicle control group (Control, 0.9% saline), cyclophosphamide model group (CY): single CY dose; positive control group (G-CSF): CY + rhG-CSF (40 μg/kg/day); ZGSII group (ZGSII): CY + ZGSII (12 mg/kg/day), as our previous research (21). All mice were injected intraperitoneally with cyclophosphamide (CY; 150 mg/kg) in 0.9% NaCl on Day 0, except for the Control group, which received the same volume of saline solution. The administration groups were injected with subcutaneous rhG-CSF for days 1–4 or ZGSII for days 1-5, as described in our previous study. The body weights of mice were recorded daily (± 0.1 g).

At the designated time point for the experiment, the peripheral blood of mice was collected from the retro-orbital plexus under isoflurane anesthesia. Hematological analysis was performed using BC-5000Vet veterinary hematology analyzer (Mindray, Shenzhen, China), which quantified leukocyte differentials (neutrophils, lymphocytes, monocytes) and erythrocyte parameters.

Blood samples were collected from the retro-orbital plexus and centrifuged at 3,000 g for 15 min; the supernatant serum was aliquoted and stored at -80 °C until analysis. The serum levels of alanine aminotransferase (ALT), aspartate aminotransferase (AST), total protein (TP), albumin (ALB), urea (UREA), and creatinine (CREA) were determined using an automated clinical chemistry analyzer (Chemray240, Shenzhen) according to the manufacturer’s instructions. Commercially available reagent kits with calibrated standards were used for each parameter to ensure accuracy. The serum concentrations of GM-CSF, IL-6, and TNF-α were quantified using commercial enzyme-linked immunosorbent assay (ELISA) kits (Baipeng Biotechnology Co., Ltd., China) following the provided instructions. A standard curve was generated for each assay, and cytokine concentrations for unknown samples were calculated by interpolation. All samples and standards were assayed in duplicate.

Femurs were dissected, decalcified in EDTA, dehydrated through an ethanol series, cleared in xylene, and paraffin-embedded. Longitudinal sections (4 μm) were stained using hematoxylin and eosin (H&E). Representative images were acquired at 100× and 400× magnification using bright-field microscopy.

The thymus and spleen were excised, blotted dry, and weighed. Organ indices were calculated as (Organ index (mg/g) = Organ weight (mg)/Body weight (g).

The human acute promyelocytic NB4 cell line was acquired from Baidi Biotech Co., Ltd. (Baidi Biotech, Shanghai), and authenticated using short tandem repeat (STR) profiling. Cells were cultured in RPMI-1640 medium (Servicebio, Wuhan) supplemented with 10% FBS and 1% penicillin-streptomycin at 37 °C/5% CO₂.

Cell viability was assessed using the Cell Counting Kit-8 kit (CCK-8; Biosharp Life Sciences, Hefei). Cells were seeded in 96-well plates at a density of 5 × l0³ cells/well and allowed to adapt for 4–6 hr under standard culture conditions. After treatment with serially diluted concentrations of ZGSII (24–120 hr, 24-hour intervals), 20 μL of CCK-8 reagent was added to each well and incubated for 4 hours at 37 °C. The optical density (OD) was measured at 450 nm using a multimode microplate reader (SpectraMax Id3, California, USA).

In vivo neutrophil maturation: Bone marrow cells from CIN mice were lysed using ACK lysing buffer with 10 min incubation at 25 °C. Cells were stained with CD11b-FITC (1:100; 11-0112-82), and Ly-6G/Ly6C-Alexa Fluor 700 (1:100; 56-5931-82) (eBioscience). Antibody-cell mixtures were incubated for 30 min at 4 °C in the dark, followed by two PBS washing cycles (400 ×g, 5 min). Cells were resuspended in 300 μL PBS and analyzed using a Gallios flow cytometer (Beckman Coulter). Neutrophil subpopulations were quantified through Ly-6G⁺CD11b⁺ dual-positive gating strategies (22), and mature neutrophils (CD11b⁺ Ly6G^+high), immature neutrophils (CD11b⁺ Ly6G^+low) (22, 23) were quantified using Flowjo.

In vitro differentiation: NB4 cells in the logarithmic growth phase were seeded into 6-well plates at 5 × 10⁴ cells/mL (2 mL/well) and were treated with a serial concentration of ZGSII. After 120 hours of incubation, cells were harvested by centrifugation (2000 rpm, for 5 min, 4 °C) and washed twice with ice-cold PBS. Surface marker staining was conducted with fluorochrome-conjugated antibodies CD11b-PE (1:100; 12-0118-42).

For Peripheral blood staining: Peripheral blood (5 μL) was smeared, air-dried, and stained with Swiss-Giemsa solution. The morphology of neutrophils was examined and calculated.

In vitro cell staining: NB4 cells (1×10⁶/mL) were smeared, air-dried, fixed in 4% paraformaldehyde (PFA), and stained (Solution A/B, Servicebio). After drying, morphology was captured using a BDS400 inverted microscope (Leica Microsystems, Wetzlar, Germany), and digital images were acquired at 100 × magnification for quantitative analysis.

NB4 cells were seeded into 6-well plates at 1 × 10⁵ cells/mL and were treated with a serial concentration of ZGSII. After 120 hours of incubation, cells were harvested and the cell cycle distribution and the apoptosis rate was detected by the cell cycle kit (KGA512, KeyGen Biotech, China) and the Annexin V-APC/PI Apoptosis Detection Kit (KGA1030, KeyGen Biotech, China). For cell cycle analysis, harvested cells were washed in PBS and fixed with ice-cold 70% ethanol solution for overnight. After washed, the PI/RNase A (9:1) working solution was added in accordance with the proportion of cells. The cell cycle distribution was determined by a BD flow cytometer and analyzed with the Flowjo (version 10) software. For apoptosis analysis, harvested cells were washed with PBS, resuspended in binding buffer, and incubated with Annexin V-FITC/PI fluorescent dyes. The apoptosis rate of each sample was detected by flow cytometer and analyzed with the Flowjo (version 10) software.

Gram-positive Staphylococcus aureus (Staphylococcus aureus, S. aureus, ATCC6538) was used for in vitro phagocytosis and survival rate assay.

Phagocytosis: Peripheral blood (10 μL) was incubated with an equal volume of log-phase S. aureus at 37 °C for 30 minutes. Following incubation, blood-bacterial smears were stained with the Swiss-Giemsa solution. A total of one hundred neutrophils were randomly counted, and the percentage of phagocytosing neutrophils was enumerated microscopically.

Survival rate: CIN mice were challenged i.p. with approximately 5 × 10⁷ CFU of S. aureus on Day 4. Mouse mortality was recorded every 12–24 h or 10 days to assess short-term and long-term survival rates. The Log-rank test was used to compare the median survival rate or actual survival rate of mice.

Peripheral blood samples were incubated with PBS containing PMA (1 μg/mL) and NBT (1 mg/mL) for 45 min. Following incubation, blood smears were Giemsa-stained. A total of one hundred neutrophils were randomly counted, and the proportion of NBT-positive neutrophils (those containing formazan particles) was calculated.

Molecular docking simulations between ZGSII and key TFs were performed using AutoDock vina 1.2.3 (31). The crystal structure of SPI1 (PDB ID: 8e3k) was acquired from the PDB, while the C/EBPϵ model was sourced from AlphaFold. The 3D structure of ZGSII was obtained from PubChem and energy-minimized under the MMFF94 force field. Before docking, protein receptors were processed with PyMOL 2.5.2 to remove heteroatoms. The conformation yielding the highest binding score was designated as the putative binding mode and visualized in PyMOL 2.5.2.

Molecular dynamics simulations were performed in AMBER 24 (32), using the highest-ranked docking pose as the initial structure. The protein-ligand complex was solvated in a TIP3P water box (10 Å cutoff), neutralized with ions, and the necessary topology files were generated. The simulation protocol began with energy minimization, a stepwise equilibration phase, solvent relaxation, and system density equilibration for 500 ps at 298.15 K and 1 atm (NPT). Calculations used a 10 Å cutoff, PME for electrostatics, the SHAKE algorithm, a Langevin thermostat (γ = 2 ps^-¹), and a 2 fs timestep, with coordinates recorded every 20 ps. After the simulation was completed, trajectories were analyzed for root mean square deviation (RMSD), root mean square fluctuation (RMSF), radius of gyration (Rg), solvent accessible surface area (SASA), and hydrogen bond formation (HBond). The binding free energy (MM/GBSA) was calculated (33).

Gene expression in bone marrow cells was quantified via mRNA analysis. Total RNA was extracted using TRIzol reagent, and then reverse-transcribed into cDNA with HiScript II Q RT SuperMix (Vazyme Biotech). Quantitative PCR analysis was performed using ChamQ Universal SYBR qPCR Master Mix on a CFX Duet Real-Time System (BIO-RAD, USA), with gene expression levels normalized to 18S rRNA and calculated with the 2^-ΔΔCT method. Primer sequences are provided in Supplementary Table 3.

Bone marrow cells were collected and lysed with RIPA buffer. The protein concentration was quantified using a BSA assay kit (Solarbio, China). The protein samples were separated by 10% SDS-PAGE gels and transferred onto PVDF membranes. The membranes were blocked with 5% skim milk in TBST for 1.5 h at room temperature. The primary antibodies of LTF (10933-1, 1:5000), CXCR2 (85144-5,1:2000), SPI1 (66618-2,1:2000), C/EBPϵ (14271-1, 1:800), and β-actin (1:100000) were incubated overnight at 4 °C. The HRP-conjugated secondary antibodies were incubated at room temperature for 1 h at room temperature. Bands were visualized using an imaging system (Biorad, USA) and quantified with ImageJ software.

All experimental data were presented as the mean ± standard deviation (SD). Each independent experiment was replicated a minimum of three times to determine the mean, and the number of experiments in each group satisfied the statistical requirements. Statistical comparisons between experimental groups were performed using one-way ANOVA or student’s t-tests (GraphPad Prism 8.0). A pre-study sample size calculation was performed to ensure adequate statistical power. A p-value of less than 0.05 (P < 0.05) was considered statistically significant, indicating a meaningful difference between groups.

To evaluate the therapeutic efficacy of ZGSII in treating chemotherapy-induced neutropenia (CIN), mice were administered ZGSII (12 mg/kg/day for 5 days) or rhG-CSF (40 μg/kg/d for 4 days) as a positive control following cyclophosphamide (CY) injection. Peripheral blood cell counts, including white blood cells (WBCs), neutrophils, monocytes, and lymphocytes, and red blood cells (RBCs) were measured on days 0, 2, 4, and 6 post-CY treatment using an automated hematology analyzer. As expected, CY administration resulted in a marked reduction in leukocyte counts by days 2 and 4, confirming successful establishment of leukopenia and neutropenia (Figures 1A–D). Body weight significantly decreased following CY administration (Figure 1F).

G-CSF treatment significantly increased WBC and neutrophil levels from day 4 onward. In contrast, ZGSII induced a more gradual recovery, ultimately elevating WBC (↑1.6-fold, P < 0.001) and neutrophil counts (↑1.7-fold, P < 0.001) by day 6 compared to the CY group (Figures 1A, B). G-CSF produced a more remarkable increase in WBC (↑2.2-fold) and neutrophils (↑2.4-fold). ZGSII also promoted a progressive recovery in monocyte counts (Figure 1C), whereas neither treatment significantly affected lymphocyte or RBC levels (Figures 1D, E). Concurrent with the development of neutropenia, a significant decrease in body weight was observed following CY challenge (Figure 1F).

Histological analysis of H&E-stained bone marrow sections revealed severe hypocellularity in CY-treated mice at day 4, with partial recovery observed by day 6. Both ZGSII and G-CSF significantly improved marrow cellularity at these time points (Figure 1G). While ZGSII did not affect thymic or splenic indices, G-CSF induced marked splenomegaly (Figures 1H, I).

To evaluate potential hepatorenal toxicity, serum biochemical parameters were analyzed (Figure 1J). CY treatment significantly reduced ALT and total protein (TP) levels. ZGSII restored ALT levels, and TP level returned to baseline by day 6. No significant changes were observed in AST, ALB, urea, and creatinine (CREA) levels.

To investigate the impact of ZGSII on neutrophil development, bone marrow cells were collected from CIN mice on days 2, 4, and 6 post-treatment. Flow cytometric analysis was performed to quantify neutrophil subpopulations: total (CD11b⁺Ly6G⁺), mature (CD11b⁺Ly6G^{+ high}), and immature (CD11b⁺Ly6G^{+ low}) neutrophils. Within two days post-induction, CY treatment significantly decreased the proportions of immature neutrophils, while increasing that of mature neutrophils (P < 0.001 vs. control), suggesting that CY-induced myelosuppression may concurrently stimulate emergency granulopoiesis (Figures 2A, B). From day 2 to day 4, G-CSF significantly expanded the immature neutrophil pool, whereas ZGSII induced a more moderate response.

By day 6, total neutrophil levels in both treatment groups surpassed baseline values (1.5-fold increase, P < 0.01). Although CY-treated mice maintained an elevated immature neutrophil fraction (38.3% vs. 18.7% in controls), G-CSF was more effective than ZGSII in increasing mature neutrophils (1.58-fold vs. 1.3-fold increase relative to CY, P < 0.001) (Figures 2A, B). The neutrophil maturation index (mature/immature ratio) decreased sharply in CY-treated mice by day 4 (11.4-fold reduction vs. control, P < 0.001). By day 6, this ratio was restored to near-normal levels by ZGSII (97% of control) and above-normal by G-CSF (144% of control) (Figure 2C). These results demonstrate that while both agents promote neutrophil production, G-CSF may excessively drive bone marrow mobilization, whereas ZGSII supports a more balanced reconstitution without overt depletion of hematopoietic reserves.

The aforementioned findings that the white blood cell and neutrophil counts reached their nadir on day 4, suggesting the phase of cyclophosphamide-induced bone marrow suppression. Subsequently, a recovery in white blood cells and neutrophils was observed, with a statistically significant difference emerging between the ZGSII-treated group and the model group on day 6. To elucidate the mechanisms by which ZGSII promotes neutrophil differentiation and ameliorates neutropenia, we performed RNA sequencing (RNA-seq) on bone marrow samples collected at days 4 and 6 CY post-induction. Comparative transcriptomic analysis between CY-treated and control mice revealed 5,289 differentially expressed genes (DEGs), including 3,371 downregulated and 1,918 upregulated genes (Figure 3A). In ZGSII-treated mice compared to CY controls, 1,256 DEGs were identified at day 4 (604 downregulated, 652 upregulated). By day 6, ZGSII treatment resulted in 1,875 DEGs relative to CY controls (1,385 downregulated; 490 upregulated; Figure 3A).

We then profiled immune cell infiltration to evaluate changes in the bone marrow microenvironment. The relative abundance of 25 murine immune cell subsets was determined using the CIBERSORT algorithm. CY-treatment also resulted in the robust depletion of bone marrow neutrophils on day 4 that was reversed by ZGSII. By day 6, the neutrophil levels were once again increased (Figure 3B).

Subsequently, the transcriptomic data were subjected to time-series clustering, which revealed four distinct expression patterns. In cluster 1 (C1) and cluster 3 (C3), ZGSII was observed to cause significantly reverse the transcriptional alterations induced by CY, whether upregulated or downregulated (Figure 3C). Gene Ontology (GO) enrichment analysis indicated that the genes within clusters C1 and C3 were highly enriched in biological processes (BP) such as leukocyte migration, myeloid leukocyte activation, and regulation of inflammatory response, as well as molecular functions (MF) including immune receptor activity and integrin binding (adjusted p < 0.05; Figure 3D). Furthermore, KEGG pathway analysis revealed significant enrichment in pathways related to leukocyte transendothelial migration, cytokine-cytotokine receptor interaction, and hematopoietic cell lineage pathway (Figure 3E).

To further delineate the granulocytic developmental stages modulated by ZGSII, we analyzed the public dataset GSE137539 from GEO, focusing on four normal mouse samples. Following feature extraction, dimensionality reduction, and clustering, we annotated neutrophil-related populations (myeloid progenitors and neutrophils) based on established markers (30) (Figure 4A). Neutrophils were subsequently extracted and reclustered into nine distinct clusters and subpopulations (Figure 4B).

Cellular stemness was assessed using the CytoTRACE2 algorithm, which identified that cluster 6 exhibited the highest stemness score and was thus designated as the root for pseudotime trajectory analysis. Genes showing dynamic expression along the inferred trajectory were identified through Moran’s I statistic (Moran’s I > 0.3, p < 0.05, q < 0.05) (Figure 4C). Integration of cellular stemness levels with neutrophil developmental stages enabled annotation of the nine clusters into eight distinct neutrophil subpopulations G1-G4, GM, G5a-G5c, except for G0, which no significant enrichment was observed in the 9 clusters (30). Stage-specific enrichment analysis demonstrated that G1 and G2 stages were predominantly represented in cluster 6; G3 in clusters 4 and 7; GM in clusters 8; G4 in clusters 0; G5a-G5c in clusters 2, 1 and 5, respectively (Figure 4D).

By intersecting the gene sets of neutrophil subpopulations with the previously defined DEGs in clusters C1 and C3, we identified 37 key genes (Figure 4E). These genes were predominantly upregulated following ZGSII treatment and enriched in the G3 to G5 subpopulations, which displayed higher differentiation and maturity scores (Figures 4F, G). GO enrichment analysis revealed that these key genes were significantly enriched in biological processes such as myeloid leukocyte migration, leukocyte chemotaxis, and neutrophil activation (Figure 4H).

To further elucidate the molecular mechanisms by which ZGSII alleviates CIN, we sought to identify its target genes and associated transcriptional regulators. A protein-protein interaction (PPI) network was constructed based on the 37 key genes using the STRING database (confidence threshold > 0.4), which revealed several hub genes, including Ltf, Lcn2, Cxcr2, Mmp8/9, and S100a8/9 (Figure 5A). Pseudotime trajectory analysis demonstrated stage specific expression patterns during neutrophil maturation: secondary granule genes such as Ltf, Lcn2, Ngp, and Camp were predominantly expressed in G1-G3 populations, Cxcr2 and Mmp8 were enriched in the highly mature G4-G5 subsets, and S100a8 was broadly expressed across all stages (G1-G5c) (Figure 5B). The mRNA expression levels of selected genes, including Ltf, Cxcr2, Fpr2, Itgb2l, Anxa1, and S100a8 were validated by RT-qPCR (Figure 5C). Protein expression of LTF and CXCR2 was further confirmed by Western blot analysis (Figures 5D, E). The results demonstrate that ZGSII treatment reversed the dysregulation of LTF and CXCR2 at both the transcriptional and translational levels.

Transcription factors (TFs) play a crucial role in regulating the differentiation and maturation of granulocytes. To identify upstream transcription regulators, we analyzed the set of 37 key genes using the CHEA3 platform. Based on the top ten ranked TFs, a TF-gene regulatory network was constructed by integrating the PPI data, which identified SPI1 and C/EBPϵ as the most central regulators (Figure 5F). Collectively, these findings indicate that ZGSII promotes neutrophil development and maturation, likely through modulation of SPI1 and C/EBPϵ activity.

To further elucidate the mechanism of action of ZGSII specifically, the potential interaction of ZGSII with the two transcription factors SPI1 and C/EBPϵ, molecular docking experiments were conducted in this study. The docking results demonstrated that ZGSII exhibits high binding affinity toward both SPI1 and C/EBPϵ, with binding free energies of -7.073 kcal/mol and -7.235 kcal/mol, respectively. In the SPI1-ZGSII complex, ZGSII forms extensive hydrophobic interactions with TRP-190, LYS-204 and LYS-186, contributing to complex stabilization in a nonpolar environment. Additionally, hydrogen bonds are formed between ZGSII and SER-202 and SER-203, which enhance both binding affinity and specificity (Figure 6A). The ligand formed hydrophobic interactions within the C/EBPϵ pocket, specifically with residues ALA-271, ARG-261, and PHE-264. Moreover, hydrogen bonds with ARG-265 and a salt bridge with ARG-261 enhanced the structural stability of the complex composition and the improved electrostatic complementarity (Figure 6B). These molecular docking results suggest that ZGSII can form stable interactions with both transcription factors. To further validate complex stability, 100-ns molecular dynamics simulations were performed in this study, confirming the robustness of the predicted binding modes over time.

The dynamic behavior of the complexes was evaluated using standard biophysical metrics, including RMSD, RMSF, Rg, hydrogen bonds, and SASA. In the C/EBPϵ-ZGSII complex, RMSD exhibited significant initial fluctuation followed by stabilization in the later stages of the simulation, indicative of conformational plasticity and adaptive binding characteristics. Conversely, the SPI1-ZGSII complex exhibited constrained fluctuations, indicating a rigid binding mode (Figure 6C, D). Although the binding modes of two ligands differed, their binding states remained stable throughout the simulations. Another key molecular dynamics metric, RMSF, revealed that while the flexible loop region of C/EBPϵ displayed moderate fluctuations, its ligand-binding site remained structurally stable. In contrast, SPI1 showed reduced residue mobility across the entire protein, consistent with its rigid scaffold structure (Figure 6E). Rg analysis revealed distinct structural dynamics: the C/EBPϵ-ZGSII complex exhibited gradual compaction throughout the simulation, indicating structural tightening around the ligand. In contrast, the Rg value of SPI1-ZGSII complex remained largely unchanged, suggesting pre-organized structural features that support rigid binding (Figure 6F). Hydrogen bond analysis revealed that the C/EBPϵ complex maintained more stable and persistent interactions, indicating a dynamically stable binding interface (Figure 6G). SASA calculations showed a slight decrease for C/EBPϵ, consistent with reduced solvent exposure and shielding of the binding site, whereas the SASA of SPI1 remained constant, in agreement with the structural rigidity of its binding pocket and unaltered solvent accessibility. (Figure 6H).

To better quantify the binding affinity of ZGSII for the two transcription factors, molecular mechanics-generalized born surface area (MM-GBSA) calculations were performed based on the MD simulation trajectories. The calculated binding free energies of -29.94 ± 5.40 kcal/mol for the C/EBPϵ-ZGSII complex and -14.12 ± 1.16 kcal/mol for the SPI1-ZGSII complex indicate strong binding interactions, with ZGSII exhibiting higher binding affinity toward C/EBPϵ (Table 1). Free energy landscape (FEL) analysis revealed a single deep energy basin of the C/EBPϵ complex, indicative of high conformational stability. In contrast, the SPI1-ZGSII complex exhibited a more dispersed FEL distribution with several shallow minima, indicative of a greater conformational variability and reduced stability (Figure 6I). Energy decomposition analysis revealed that the binding interactions are predominantly driven by electrostatic and van der Waals forces, as well as the nonpolar solvation contributions. In the C/EBPϵ-ZGSII complex, ARG261 and ARG265 were the primary contributions to binding energy, whereas hydrophobic residues such as isoleucine and leucine made moderate contributions to complex stabilization. In SPI1-ZGSII complex, LYS247, LEU248, and SER203 were the most significant energy contributors, suggesting that the complex is stabilized through a combination of hydrogen bonding and hydrophobic interactions (Figure 6J).

The above computational modeling suggests potential interaction of ZGSII with SPI1 and C/EBPϵ. To further validate the cellular level effects of ZGSII on the expression of these transcription factors, this study measured SPI1 and C/EBPϵ protein levels. Results showed that, compared with the control group, the expression of SPI1 and C/EBPϵ was significantly reduced in the CY model group; however, ZGSII intervention effectively upregulated the protein expression levels of both transcription factors (Figures 6K, L).

The human promyelocytic leukemia cell line NB4 is a well-established model for studying neutrophil differentiation, in which growth arrest serves as a key phenotypic marker (34, 35). To investigate the effects of ZGSII, NB4 cells were treated with increasing concentrations (0-100 μM) for 24–120 hours. CCK-8 assays revealed a dose-dependent suppression of cell growth (Figures 7A, B), with IC₅₀ values decreasing from 33.95 μM at 24 h to 27.01 μM at 72 h, and further to 18.42 μM by 120 h. After 120 h of treatment, 10 μM ZGSII resulted in 23.11% growth inhibition. Flow cytometric analysis showed no significant alterations in cell cycle profile or apoptosis at concentrations of 5-10 μM after 120 h treatment (Figures 7C–F). In contrast, both flow cytometric and Wright-Giemsa staining confirmed concentration-dependent induced differentiation at these concentrations, accompanied by characteristic morphological maturation (Figures 7G–J).

Furthermore, serum levels of the cytokines GM-CSF, IL-6, and TNF-α were measured using ELISA. The results showed that CY treatment significantly reduced serum GM-CSF levels, while markedly increasing the levels of the pro-inflammatory cytokines IL-6 and TNF-α. These alterations were notably reversed by G-CSF or ZGSII administration (Figure 8A). Functional assessments demonstrated enhanced neutrophil effector functions following ZGSII treatment: phagocytic capacity approached normal levels compared to the CY controls (P < 0.01; Figure 8B), while reactive oxygen species (ROS) production increased by 1.8-fold, as measured by the nitroblue tetrazolium (NBT) reduction assay (P < 0.05; Figure 8C).

This functional restoration correlated with improved survival in Staphylococcus aureus-challenged CIN mice. To evaluate protective efficacy, animals were infected intraperitoneally with high-dose (5×10⁷ CFU) or low-dose (5×10⁶ CFU) S. aureus on day 4 post-induction. In the high-dose infection group, median survival was reduced to 12 h in CY controls, whereas ZGSII significantly extended survival beyond this threshold (P < 0.001, log-rank test; Figure 8D). Following low-dose S. aureus challenge, only 70% of untreated mice in the CY group survived to day 10, compared to 90% survival in both G-CSF- and ZGSII-treated groups over the entire observation period (Figure 8E). These findings indicate that ZGSII confers significant survival benefits in bacteremic neutropenic hosts.