Mechanism of Xiao-ai-fei Honey Ointment, a Traditional Uyghur Multi-Ingredient Medicinal Preparation, Against Cervical Cancer Based on Network Pharmacology and In Vitro Evaluation of Anti-Cancer Activity

Abstract

Background/Objectives: Cervical cancer, primarily driven by persistent high-risk HPV infection, remains a major global health issue. Xiao-ai-fei honey ointment, a traditional Uyghur multi-ingredient preparation, has shown clinical promise in cancer treatment, but its mechanisms against cervical cancer are not fully understood. This study aimed to investigate the potential molecular mechanisms of ethanolic extract of Xiao-ai-fei honey ointment (XAFHO) in cervical cancer using network pharmacology, single-cell RNA sequencing, and experimental validation. Methods: Differentially expressed genes (DEGs) in cervical cancer were identified from TCGA database. Active components and corresponding targets of XAFHO were retrieved from the TCMSP database, and disease targets were obtained from GeneCard, OMIM, and the TTD. Intersection targets were subjected to multivariate Cox and LASSO regression to construct a prognostic model. Immune infiltration, TMB, and MSI were compared between risk groups. Single-cell RNA-seq data were analyzed to determine cellular origins and inter-cellular communication. In vitro assays were performed on HeLa and SiHa cells to assess the anti-cancer activity of XAFHO. Molecular docking evaluated binding affinities between active compounds and core targets. The expression and functional roles of FASN and SPP1 were further validated by RT-qPCR, Western blotting, and siRNA transfection. Results: Sixty-three potential XAFHO targets were identified, and an 11-gene prognostic model was established, effectively stratifying patients into high- and low-risk groups with significantly different overall survival (AUC > 0.7). The high-risk group exhibited an immunosuppressive microenvironment and higher TMB. Single-cell analysis revealed that FASN and ACACA were predominantly expressed in tumor cells, while SPP1 was enriched in macrophages/monocytes. Tumor cells communicated with immune cells via the TGFB1–TGFβR1/R2 axis, promoting immune evasion. In vitro, XAFHO significantly inhibited proliferation, colony formation, migration, and invasion of cervical cancer cells. Molecular docking confirmed the strong binding of quercetin, kaempferol, and isorhamnetin to FASN and SPP1 (binding energy < –6.0 kcal/mol). Functional validation indicated that upregulated FASN and SPP1 contribute to malignant behaviors in cervical cancer cells. Conclusions: This study integrates network pharmacology with single-cell and experimental approaches to demonstrate that XAFHO exerts multi-target and multi-cell anti-cervical cancer effects, potentially by modulating lipid metabolism and immune-related pathways via FASN and SPP1. These findings provide a scientific basis for the therapeutic application of XAFHO in cervical cancer.

1. Introduction

Cervical cancer is the fourth most common malignancy among women worldwide, with its pathogenesis being closely associated with persistent human papillomavirus (HPV) infection [1,2]. High-risk HPV infection is widely recognized as the primary etiological factor for cervical cancer [3]. Early-stage cervical cancer is typically managed through surgical intervention, whereas advanced stages are commonly treated with a combination of chemotherapy and radiotherapy [4]. Although targeted therapies and immunotherapies have significantly improved patient survival, not all patients respond favorably to these treatments [5]. This underscores that tumor heterogeneity remains a critical obstacle in developing effective therapeutics to improve patient prognosis [6]. Consequently, there is an urgent need to develop novel agents to extend the survival of cervical cancer patients.

Xiao-ai-fei honey ointment is a traditional Uyghur medicinal formulation with a long history of use in China for cancer treatment and in clinical practice, and it is composed of five Uyghur medicinal components: Bungarus multicinctus, Biba (Piperis Longi Fructus), Gaoliangjiang (Rhizoma Alpiniae Officinarum), Hujiao (Piper nigrum L), and Ganjiang (Zingiber officinale Roseco) [7,8,9]. This formulation has been primarily utilized in the treatment of tumors, as well as inflammatory conditions, leakage syndromes, and refractory ulcers affecting the digestive and reproductive systems, including obstinate and malignant sores associated with qi stagnation [7].

Zaocys dhumnades, the dried body of the juvenile Chinese many-banded krait, is a traditional Chinese medicinal material. Chemical analyses have shown that it contains more than 20 elements [10] and has been clinically employed in the treatment of rheumatism, tetanus, and leprosy [11]. Piper longum L, which serves both as a traditional Chinese medicine and a food, is classified under the Piper genus of the Piperaceae family [12]. It possesses various pharmacological activities, including anti-inflammatory, antioxidant, and cardiovascular protective effects, primarily attributed to its rich content of amide alkaloids [13]. Alpinia officinarum Hance, derived from the dried roots and rhizomes of Alpinia officinarum, has been reported to exhibit antioxidant, antidiabetic, anti-inflammatory, and anticoagulant activities [14,15]. Piper nigrum L is derived from the flowering vines of the Piperaceae family [16]. This species has been extensively investigated for its diverse biological activities and the presence of bioactive phytochemical compounds. Current studies have revealed that Piper nigrum L exhibits varying degrees of inhibitory effects on a variety of cancers, including breast cancer and gastric cancer [17,18]. Zingiber officinale Roscoe is the dried rhizome of ginger family plants [19]. Its chemical constituents are primarily volatile and pungent compounds. Modern pharmacological studies have demonstrated that ginger possesses analgesic, anti-inflammatory, antibacterial, anti-tumor, and local circulation-improving properties [20]. Although these agents have demonstrated a certain degree of anti-tumor activity, their mechanisms of action remain incompletely understood.

Previous research has investigated the anti-tumor mechanism of ethanolic extract of Xiao-ai-fei honey ointment (XAFHO) in gastric cancer by using network pharmacology analysis [7], suggesting that XAFHO may exert therapeutic effects through multi-component and multi-target interactions. However, that study mainly focused on target–pathway prediction and did not address the cellular localization of key targets, inter-cellular communication, or the biological context of the tumor microenvironment. In addition, direct functional validation remained limited. Therefore, whether XAFHO exerts similar multi-target effects on cervical cancer and how these effects are related to tumor cells, immune cells, and patient prognosis remain unclear.

In the present study, we extended this line of investigation to cervical cancer and integrated network pharmacology with single-cell RNA sequencing, prognostic modeling, immune microenvironment analysis, molecular docking, and in vitro evaluation. By doing so, we aimed not only to identify potential targets of XAFHO but also to clarify their cellular distribution, functional relevance, and possible roles in cervical cancer progression. This is expected to provide a theoretical foundation for the development of novel therapeutic strategies for cervical cancer.

2. Results

2.1. Identification of Differentially Expressed Genes (DEGs)

In this study, the cervical cancer transcriptome dataset from TCGA database was acquired to elucidate the changes in gene expression profiles associated with cervical cancer onset and progression. Rigorous preprocessing steps, such as normalization and standardization, were applied to the raw transcriptome data to minimize technical variability, batch effects, and outliers, ensuring comparability of gene expression levels across samples. Subsequently, utilizing the high-quality preprocessed data, a differential expression analysis between tumor and normal tissues was performed using the DESeq2 package in R. By applying the criteria of |log2FC| ≥ 1 and padj < 0.05, 4350 DEGs were identified, with 2672 genes being upregulated and 1678 genes being downregulated significantly in cervical cancer tissues compared with normal tissues (Figure 1B). To visualize the most significantly altered genes, the top 50 upregulated and top 50 downregulated genes were selected for heatmap analysis (Figure 1A).

Subsequently, we performed enrichment analysis on the identified genes to elucidate their biological functions and potential regulatory pathways. The results of the enrichment analysis revealed significant enrichment of KEGG pathways related to the cell cycle and the p53 signaling pathway, indicating the involvement of the screened genes in the development and advancement of cervical cancer through the regulation of these pathways (Figure 1C,D). In terms of biological processes (BPs), notable enrichment was observed in functions like keratinocyte differentiation and epidermis development (Figure 1C,D), which play crucial roles in shaping the tumor microenvironment. The dysregulation of interactions among epithelial cells, immune cells, and stromal cells in the tumor microenvironment is a key factor driving tumor progression. This indicates that the identified differentially expressed genes may impact the malignant characteristics of cervical cancer by modulating inter-cellular signaling within the tumor microenvironment.

2.2. Acquisition of Active Components and Disease Targets of XAFHO and Intersection Analysis

To elucidate the material basis and potential targets of XAFHO in its anti-cervical cancer effects, all chemical components of XAFHO were initially retrieved from the TCMSP database. Oral bioavailability (OB) and druglikeness (DL) are critical indicators for assessing the druggability of compounds. Therefore, this study established OB ≥ 30% and DL ≥ 0.18 as the screening criteria for active components. Subsequently, 37 effective components of XAFHO meeting these criteria were identified, corresponding to 205 component-related targets. Disease targets related to cervical cancer were obtained by simultaneously searching three target databases using “cervical cancer” as the keyword, namely, GeneCard, OMIM, and TTD. The search outcomes revealed 9325, 194, and 26 targets from GeneCard, OMIM, and the TTD, respectively. After eliminating duplicates and overlapping these targets, a total of 9481 cervical cancer-associated targets were ultimately obtained. To pinpoint potential core targets responsible for the anti-cervical cancer effects of XAFHO, an intersection analysis was conducted among the DEGs retrieved from TCGA database, the targets of active ingredients in XAFHO, and the amalgamated cervical cancer disease targets. A total of 63 intersection targets were identified, corresponding to 26 active ingredients of XAFHO, indicating that XAFHO may exert its anti-cervical cancer effects by modulating these 63 targets (Figure 1F and Table S1). Subsequently, a “drug–ingredient–target” visualization network was constructed using Cytoscape software (version 3.10.4) (Figure 2). Topological analysis of the network revealed that the three components with the highest degree values were MOL000098 (quercetin), MOL000422 (kaempferol), and MOL000354 (isorhamnetin). Notably, quercetin and kaempferol are not only network-predicted key components but also experimentally supported constituents of XAFHO, whereas isorhamnetin was identified as a candidate bioactive compound based on network pharmacology screening [21].

2.3. Construction and Validation of Prognostic Models

To investigate the association between the mentioned intersecting genes and cervical cancer patient prognosis, we initially conducted multivariate Cox proportional hazards regression analysis on the 63 intersecting genes to identify independent prognostic factors for cervical cancer. The analysis revealed 15 genes as independent prognostic markers for cervical cancer (Figure 3A), comprising 9 high-risk genes (FASN, ACACA, CA2, MMP1, MMP3, IL1A, SPP1, and HK2) and 6 protective genes (IL12B, HTR3A, BCL2, SLC2A4, SULT1E1, and E2F2). High expression of risk-associated genes was linked to poorer patient outcomes, whereas high expression of protective genes was associated with better prognosis. To establish a robust prognostic gene signature for cervical cancer, we employed LASSO–Cox regression analysis for further variable selection and model development using the aforementioned 15 prognostic-related genes. Initially, a coefficient profile plot was generated to illustrate the evolving trend of the regression coefficient for each gene with the regularization parameter λ (Figure 3B). The results revealed a gradual transition of coefficients for selected crucial genes from zero to non-zero values with decreasing λ, underscoring the ability of λ to assess the contribution of each gene to prognostic predictions. Subsequently, the optimal λ value was determined through 10-fold cross-validation (Figure 3C), leading to the identification of the most predictive gene combination at the optimal λ value (λ.min). Ultimately, 11 genes were selected for the definitive prognostic risk model, with individual patient risk scores calculated as Risk Score = Σ (gene expression level × corresponding regression coefficient) (Table S2).

To assess the prognostic model’s predictive validity, patients were stratified into high-risk and low-risk categories using the median risk score, followed by a series of validation analyses. Survival analysis revealed significantly shorter overall survival in the high-risk group compared with the low-risk group (p < 0.05), indicating the model’s ability to differentiate between patients with distinct prognostic outcomes (Figure 3D). Time-dependent ROC curve analysis demonstrated that the model’s AUC for predicting 1-, 3-, and 5-year overall survival exceeded 0.7 (Figure 3E), affirming its accuracy in medium- to long-term prognostic forecasting. The risk score plot visually depicted the model’s predictive performance, highlighting that high-risk patients (depicted in red) experienced more deaths and shorter survival times (Figure 3F). External validation in an independent cohort showed poorer overall survival in the high-risk group (Figure S1A). ROC analysis was available only for 0.5, 1, and 2 years because of limited follow-up, and the AUC exceeded 0.7 at 2 years (Figure S1B), supporting the model’s value for relatively longer-term prognosis.

To operationalize this prognostic model into a personalized predictive tool for clinical use, we developed a nomogram incorporating the risk scores of 11 genes (Figure 3G) to estimate the 1-year, 3-year, and 5-year overall survival probabilities for individual patients. Evaluation through calibration curve analysis demonstrated the nomogram’s accurate calibration in predicting survival rates at these time points, aligning closely with observed patient outcomes (Figure 3H), affirming its reliability. In assessing the independence of the 11-gene risk score from conventional clinicopathological factors, a multivariate Cox proportional hazards regression model was employed, incorporating the risk score alongside factors such as age, tumor grade, and T stage. The analysis (Figure 3I) revealed that following adjustment for all variables, only T4 stage (p = 0.001) and high-risk score (p = 0.001) emerged as autonomous prognostic indicators for overall survival in cervical cancer patients, with other factors exhibiting no statistical significance (p > 0.05). In summary, the 11-gene prognostic risk score established in this study not only correlates significantly with cervical cancer patient prognosis but also, crucially, exhibits predictive autonomy from the traditional clinical staging system.

2.4. Distinct Tumor Immune Microenvironment Characteristics Between Risk Subgroups

To investigate the heterogeneity of the tumor immune microenvironment (TIME) across different risk groups, we employed the CIBERSORT algorithm to conduct a deconvolution analysis utilizing tumor transcriptome data. This approach allowed us to quantitatively assess the infiltration proportions of 22 immune cell types in samples from each group, thereby elucidating the differences in immune infiltration characteristics between high- and low-risk groups. The findings indicated that cervical cancer tissues in the high-risk group displayed a pronounced immunosuppressive microenvironment phenotype (Figure 4A,B). In comparison to the low-risk group, the high-risk group exhibited significantly reduced infiltration levels of key effector immune cells, including B cells, CD8⁺ T cells, and regulatory T cells (Treg). The reduction in these cell types may compromise the body’s capacity for anti-tumor immune responses [22,23]. Conversely, the proportion of activated dendritic cells was significantly elevated in the high-risk group, and this abnormal enrichment may further exacerbate the inhibitory state of the immune microenvironment [24]. Spearman correlation analysis was performed to elucidate the relationship between the 11-gene prognostic signature and the tumor immune microenvironment. The analysis determined the correlation coefficients among gene expression levels, risk scores, and the infiltration scores of 22 immune cells. Our results revealed a significant negative correlation between the risk score and the levels of infiltrating T cells and B cells (Figure 4C,D). Particularly, the lipid metabolism genes (FASN and ACACA) integrated into the model exhibited a noteworthy correlation with immunosuppressive cells (Figure 4C). These findings suggest a close association of this prognostic signature with the development of an immunosuppressive microenvironment in cervical cancer.

2.5. Characterization of Somatic Mutations and Genomic Instability

Genomic variances among different risk groups were explored through comprehensive whole-exome/targeted sequencing analysis of cohort samples, resulting in the generation of mutation landscape plots (Figure 5A,B). The plots depicted commonly mutated genes, distribution of mutation types, and TMB at the individual sample level. The analysis unveiled notably higher TMB in high-risk group patients compared with their low-risk counterparts (Figure 5C). Regarding the spectrum of mutated genes, TTN (mutation frequency 30%) emerged as the most frequently altered driver gene, followed by PIK3CA, MUC16, and KMT2C, among others, with missense mutations being the prevailing type. Furthermore, an assessment of MSI status across risk groups indicated that while the high-risk group exhibited significantly elevated TMB, its MSI status did not significantly differ from that of the low-risk group, with both primarily characterized as microsatellite-stable (Figure 5D). These results imply that the heightened TMB in high-risk tumors is not attributable to DNA mismatch repair deficiency but may stem from alternative mechanisms contributing to genomic instability.

2.6. Analysis of the Cervical Cancer Microenvironment at Single-Cell Resolution and Investigation into the Cellular Origin and Function of Prognostic Genes

To elucidate the tumor microenvironment at single-cell resolution and identify the cellular origins of the 11 prognostic genes, we obtained single-cell RNA sequencing data of cervical cancer from the GEO database. A total of 43,493 high-quality cells were acquired and clustered into 10 distinct subsets, which included epithelial cells/tumor cells, CD8+ T cells, B cells, macrophages, and fibroblasts (Figure 6A,B). Expression analysis revealed specific high expression of FASN and ACACA from the prognostic model in the epithelial cells/tumor cells cluster, while SPP1 was predominantly derived from macrophages and monocytes (Figure 6C). Cell communication analysis further demonstrated that epithelial cells/tumor cells engaged in more frequent interactions with macrophages, monocytes, and CD8+ T cells via the TGFB1–TGFβR1_R2 ligand–receptor pair, potentially establishing a self-reinforcing immunosuppressive network (Figure 6H). Pseudotime analysis was conducted to investigate the functional states and differentiation trajectories of these cell subpopulations. The results revealed distinct differentiation paths for both macrophages and monocytes (Figure 6D,E). Dynamic changes in the expression of FASN, ACACA, and SPP1 were observed during the differentiation process, with FASN and ACACA expression gradually decreasing, while SPP1 expression progressively increased (Figure 6F). These findings suggest that these genes not only act as markers for specific cell types but also play a role in regulating the functional differentiation process of the cells.

2.7. Effects of XAFHO on the Malignant Biological Behavior of Cervical Cancer Cells

The anti-cervical cancer activity of XAFHO was evaluated by determining its IC50 against HeLa and SiHa cells using the CCK-8 assay. The 24 h IC₅₀ values for SiHa cells were 30.0 ± 1.67 and 55.2 ± 2.45 μg/mL, respectively (Figure 7A,B). For subsequent experiments, three concentrations were selected, 0.5 IC₅₀, IC₅₀, and 1.5 IC₅₀, with DDP serving as a positive control. Treatment with XAFHO at IC₅₀ and 1.5 IC₅₀ concentrations significantly inhibited the colony-forming ability of both HeLa and SiHa cells compared with the control, as shown in colony formation assays (Figure 7D). EdU incorporation assays indicated a dose-dependent decrease in the percentage of EdU-positive cells upon XAFHO treatment, suggesting effective suppression of DNA replication and cell proliferation (Figure 7E,F). Transwell invasion and migration assays demonstrated a marked reduction in the number of cells penetrating Matrigel-coated or uncoated membranes following XAFHO treatment (Figure 7C and Figure S2). In vitro, XAFHO effectively suppressed malignant cellular behaviors, including proliferation, colony formation, migration, and invasion, in cervical cancer cells.

2.8. Molecular Docking Analysis

To further explore the potential interactions between XAFHO and key targets in cervical cancer, we selected the three compounds (Quercetin, Kaempferol, and Isorhamnetin) with the highest median XAFHO values for molecular docking analysis. FASN and SPP1 were selected as docking targets because both were prognostic risk genes with distinct biological relevance in cervical cancer. Single-cell analysis showed that FASN was enriched in tumor cells, whereas SPP1 was enriched in monocytes/macrophages. Additionally, SPP1 has been reported to be highly expressed in cervical cancer and related to immunotherapy response [25]. These three compounds exhibited strong binding affinities with both FASN and SPP1, which are prognostic genes. All binding energies were below -6.0 Kcal/mol, suggesting potential binding affinity between representative XAFHO components and these target proteins (Figure 8A–F).

2.9. The Impact of the Knockdown of FASN and SPP1 on the Biological Behavior of Cervical Cancer Cells

To evaluate FASN and SPP1 as key targets underlying the anti-cervical cancer effects of XAFHO, we initially assessed their expression levels in cervical cancer cells. Relative to normal cervical epithelial cells, both FASN and SPP1 exhibited significant upregulation in the HeLa and SiHa cervical cancer cell lines (Figure 9A). We subsequently treated cervical cancer cells with XAFHO to examine its effects on the expression of these targets. RT-qPCR analysis indicated that XAFHO treatment did not significantly change the mRNA levels of FASN or SPP1 in HeLa or SiHa cells (Figure S3). In contrast, Western blot analysis revealed a substantial reduction in their protein expression (Figure 9B), suggesting that XAFHO may regulate FASN and SPP1 at the post-translational level rather than at the transcriptional level. To further elucidate the functional roles of FASN and SPP1 in cervical cancer cells, we conducted siRNA-mediated knockdown of these genes in HeLa and SiHa cell lines (Figure 9B). Functional assays demonstrated that the knockdown of either FASN or SPP1 significantly inhibited cell proliferation (Figure 9D) and diminished cell migration and invasion capacities (Figure 9C). These findings suggest that XAFHO may exert its anti-cancer effects, at least in part, through the downregulation of FASN and SPP1 at the protein level.

3. Discussion

This study combined network pharmacology, single-cell RNA sequencing, and molecular docking to investigate the anti-cervical cancer mechanism of XAFHO, a traditional Uyghur medicinal formulation. Our analysis revealed 37 active compounds and 63 essential anti-cervical cancer targets of XAFHO. An independent 11-gene prognostic model was established, with an AUC > 0.7 for predicting 1–5-year survival. The high-risk group exhibited a more immunosuppressive tumor microenvironment and higher TMB. Single-cell analysis elucidated the cellular localization of core genes and inter-cellular communication networks within the tumor microenvironment. In vitro experiments also verified the effectiveness of the drug. Molecular docking showed the stable binding (binding energy < −6.0 Kcal/mol) between XAFHO’s active components and key targets. Functional validation showed that FASN and SPP1 were significantly upregulated in cervical cancer cells, and silencing either target markedly suppressed malignant cellular behaviors. In conclusion, XAFHO exerts its anti-cervical cancer effects through the regulation of multiple components, targets, and cells.

Previous research explored the anti-gastric cancer mechanism of XAFHO by using network pharmacology [7]. However, such analyses mainly predict target–pathway relationships and do not adequately address cellular localization, inter-cellular communication, or dynamic regulation within the tumor microenvironment, thereby limiting direct biological validation [26]. This study integrates network pharmacology with single-cell sequencing to not only identify core targets but also validate through single-cell analysis that FASN and ACACA exhibit high expression in epithelial/tumor cells, whereas SPP1 primarily originates from macrophages and monocytes. Additionally, epithelial/tumor cells interact extensively with macrophages, monocytes, and CD8+ T cells through the TGFB1–TGFβR1_R2 ligand–receptor pair, establishing an immunosuppressive network. Furthermore, dynamic alterations in gene expression during cell differentiation were observed. Compared with network pharmacology alone, this integrated approach provides a more comprehensive view of the multi-cellular anti-tumor mechanism of XAFHO by linking tumor cells, myeloid cells, and inter-cellular signaling networks.

Building on our integrated analysis, the 11-gene prognostic model and its associated immune and genomic features provide further insight into the molecular basis of XAFHO’s anti-cervical cancer effects. The immunosuppressive tumor immune microenvironment (TIME) plays a critical role in the progression of cervical cancer by impairing the infiltration of effector immune cells and promoting inhibitory cell subsets [27,28,29]. This “cold” TIME profile is associated with decreased B-cell and CD8+ T-cell infiltration, alongside an increase in Tregs, collectively impairing anti-tumor cytotoxicity and contributing to the immune evasion and poor prognosis observed in cervical cancer [30,31]. In conjunction with higher TMB in the high-risk group, this immunosuppressive environment may reduce the effectiveness of monotherapy with immune checkpoint inhibitors, although further validation is needed. Thus, this prognostic model serves as a useful risk stratification tool, identifying high-risk patients who may benefit from combined therapeutic strategies, though the effectiveness of specific combinations requires further investigation. Furthermore, considering that the key active constituents of XAFHO (Quercetin, Kaempferol, and Isorhamnetin) target FASN and SPP1, which are markedly upregulated in this high-risk context, XAFHO could be considered a promising candidate for combination therapy, potentially enhancing the effects of immunotherapies (e.g., PD-1/PD-L1 inhibitors) by targeting tumor metabolic pathways (via FASN/ACACA) and reversing immunosuppression (via SPP1), though preclinical validation is required to confirm these effects. Subsequent preclinical investigations are essential to validating this combined treatment strategy and to assess the utility of this 11-gene signature as a predictive biomarker for treatment response.

Our genomic analysis revealed the complex molecular landscape associated with high-risk tumors, including factors related to the immune microenvironment. TMB, an indicator of genomic instability, has been suggested to interact with the immune microenvironment in cervical cancer [32,33]. High TMB is associated with increased production of neoantigens in cervical cancer, which could theoretically enhance immune recognition of tumor cells [33,34,35]. However, within an immunosuppressive TIME, these neoantigens may not effectively trigger anti-tumor immune responses, potentially leading to immune evasion in cervical cancer [35]. Consistent with previous studies, our genomic analysis showed that high-risk group cervical cancer patients exhibited significantly elevated TMB compared with those in the low-risk group, suggesting that TMB may influence the prognosis and treatment response in cervical cancer patients. Interestingly, there was no notable distinction in MSI status between the two groups, implying that the high TMB was not attributable to DNA mismatch repair deficiency.

The core prognostic genes are functionally involved in pathways crucial to tumor progression. FASN and ACACA represent pivotal enzymes in the de novo fatty acid synthesis pathway [36]. Previous studies have identified FASN as a critical driver of metabolic reprogramming and treatment resistance in cervical cancer, and pharmacological inhibition of FASN has been reported to restore cisplatin sensitivity in resistant models [37,38,39,40]. Likewise, SPP1 has been implicated in cervical cancer invasiveness, immune evasion, and responsiveness to PD-1 blockade, suggesting that it plays an important role in shaping the tumor immune microenvironment [25,41]. These findings indicate that both FASN and SPP1 are not merely correlated with malignant progression but may function as actionable targets in cervical cancer. Consistent with these previous observations, our study further supports the pathological relevance of FASN and SPP1 in cervical cancer. We found that both genes were significantly upregulated in HeLa and SiHa cells compared with normal cervical epithelial cells, which is in agreement with the reported oncogenic roles of these targets. Functionally, siRNA-mediated knockdown of either FASN or SPP1 significantly suppressed cervical cancer cell proliferation, migration, and invasion, further indicating that both molecules contribute to the malignant phenotype. Molecular docking reveals the stable binding of quercetin, kaempferol, and isorhamnetin to FASN and SPP1, with binding energies below −6.0 kcal/mol, supporting the hypothesis that XAFHO’s active components modulate lipid metabolism pathways and the immune microenvironment by targeting FASN and SPP1 to exert anti-cervical cancer effects. Importantly, quercetin and kaempferol are supported by both experimental evidence and network analysis, enhancing consistency between chemical characterization and prediction, whereas others (e.g., isorhamnetin) remain putative candidates requiring further validation.

This study, through the combination of computational biology and single-cell omics techniques, provides a systematic theoretical framework for the multi-target and multi-cell synergistic mechanism of XAFHO against cervical cancer. However, some limitations still need to be addressed in future work. First, animal models are needed for further in vivo verification. Second, although our results support the potential involvement of FASN and SPP1 at the protein and functional levels, their precise signaling pathways and regulatory mechanisms have not yet been fully characterized and will require further pathway-level validation in future studies. In addition, due to the complex composition of traditional compound preparations and variable metabolic processes, they cannot be fully simulated through computational models. Therefore, the molecular docking results should be considered preliminary and require experimental validation. Future research should combine medicinal chemistry and pharmacokinetic methods to isolate and identify the active components in vivo and clarify their pharmacodynamic basis. Although this study mainly focuses on static network pharmacology to analyze drug–target interactions, the dynamic regulation of the disease network—including XAFHO’s impact on edge perturbation and module regulation—is a key area for future research. This dynamic analysis is crucial to fully understanding XAFHO’s therapeutic potential.

4. Materials and Methods

4.1. Data Acquisition

Gene expression data, clinical data, and somatic mutation data for cervical cancer were downloaded from The Cancer Genome Atlas (TCGA) database (https://portal.gdc.cancer.gov/, accessed on 15 May 2025). Single-cell RNA sequencing (scRNA-seq) dataset GSE208653 was retrieved from the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/, accessed on 15 May 2025), and five samples (GSM6360682, GSM6360683, GSM6360686, GSM6360687, and GSM6360688) were selected for subsequent single-cell analysis.

4.2. Acquisition of Drug Components and Targets

The active constituents of the compound drug XAFHO were sourced from the Traditional Chinese Medicine Systems Pharmacology Database and Analysis Platform (TCMSP). Screening criteria included oral bioavailability (OB) ≥ 30% and druglikeness (DL) ≥ 0.18 to identify potential active components of XAFHO. Subsequently, the potential therapeutic targets corresponding to the screened active components were retrieved from the TCMSP database. Disease targets associated with cervical cancer were collected from the GeneCard database, Online Mendelian Inheritance in Man (OMIM) database, and the Therapeutic Target Database (TTD) using the search term “cervical cancer.” A relevance score > 5 from the GeneCard database was used as the screening threshold. Ultimately, targets from the three databases were amalgamated to constitute the cervical cancer disease target set.

4.3. Differential Gene Identification and Enrichment Analysis

R software (version 4.5.1) was employed to preprocess and analyze the raw RNA-seq count data, aiming to remove batch effects and random noise for ensuring data reliability. Differential expression analysis was conducted using the DEseq2 package, with the criteria set at |log2 fold change (log2FC)| > 1 and adjusted p-value (Padj) < 0.05 to identify differentially expressed genes (DEGs) between cervical cancer tissues and normal tissues. Subsequently, gene functional enrichment analysis was carried out utilizing the clusterProfiler package, encompassing Gene Ontology (GO) functional enrichment analysis (encompassing biological process, cellular component, and molecular function) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis, with a significance threshold of Padj < 0.05. The visualization of the analysis results involved the use of the ggplot2 package and enrichplot package to generate volcano plots, bar plots, bubble plots, and enrichment networks, facilitating an intuitive display of the characteristics of DEGs and enrichment outcomes.

4.4. Identification of Prognostic Genes

The active components in XAFHO, disease targets related to cervical cancer, and differentially expressed genes from TCGA database were analyzed to identify a key gene set for drug intervention in cervical cancer. Prognostic correlation analysis was conducted on this gene set using the survival and survminer packages in R software. Patients were categorized into high-expression and low-expression groups based on median gene expression levels, with overall survival (OS) as the primary outcome. The survival disparity between the groups was assessed using the log-rank test, identifying genes significantly linked to patient prognosis (p < 0.05 denotes significance). Subsequently, Cox proportional hazards regression analysis was employed to confirm and pinpoint independent prognostic genes. Initially, univariate Cox regression analysis was utilized to screen genes associated with patient OS (p < 0.1 for inclusion). These genes were then subjected to multivariate Cox regression to adjust for variables like age and clinical stage, ultimately revealing the prognostic genes.

4.5. Construction of the Gene-Associated Prognostic Model

To refine the gene set, eliminate false-positive results, and mitigate overfitting in the subsequent model, Lasso regression analysis was conducted as a secondary screening of the prognostic-related genes identified through Cox regression analysis. The glmnet package in R was utilized for this analysis, employing 10-fold cross-validation to ascertain the optimal penalty parameter λ (lambda). A core gene set, strongly associated with cervical cancer prognosis, was selected based on the principle of minimum deviation. Core genes identified through Lasso regression were incorporated into a multivariate Cox model, adjusting for confounding factors such as age and stage, to pinpoint independently predictive genes (p < 0.05). mRNA expression data for these genes in cervical cancer and adjacent normal tissues were extracted from TCGA database to validate expression differences and confirm clinical relevance. A cervical cancer risk scoring model was developed using independent prognostic genes from Cox regression analysis and their coefficients, calculating each patient’s risk score through a linear combination of gene expression levels and β values. The calculation formula is as follows:

Prognostic model/Risk Score = β₁ × gene₁ + β₂ × gene₂ +… + β_n × gene_n

4.6. Validation of the Prognostic Model

TCGA cervical cancer cohort was stratified into high-risk and low-risk groups according to median risk scores. Kaplan–Meier survival analysis was performed to compare overall survival (OS) using log-rank tests. The pROC package was utilized to generate ROC curves and calculate AUC values for 1-year, 3-year, and 5-year survival, with an AUC greater than 0.7 indicating good predictive accuracy. Furthermore, pheatmap was employed to visualize risk scores and differences in gene expression. A prognostic nomogram, developed using the rms package, predicted survival rates based on risk scores and clinical features. Lastly, calibration curves were used to evaluate model consistency against actual survival durations.

4.7. Immune Infiltration and Mutation Analysis

CIBERSORT (https://github.com/, accessed on 1 June 2025) analyzed the ratios of immune cell infiltration, establishing correlations between signature gene expression and immune cell infiltration in cervical cancer. The tumor mutational burden (TMB) was determined using mutation data from The Cancer Genome Atlas (TCGA) to compare high-risk and low-risk groups while also examining the distribution of microsatellite instability (MSI) status to evaluate disparities in genomic instability.

4.8. Single-Cell RNA Sequencing Analysis

The GSE208653 dataset served as the basis for single-cell analysis in this study. The Seurat package facilitated routine single-cell data processing, encompassing several steps: data preprocessing and filtering, during which low-quality cells exhibiting excessively high mitochondrial gene ratios, abnormal ribosomal gene ratios, and irregular gene/UMI counts per cell were excluded; subsequent normalization and standardization of the data aimed to mitigate technical variations across samples; Principal Component Analysis (PCA) was employed for initial dimensionality reduction, with key principal components being selected for further investigation. Additionally, the UMAP algorithm was utilized for visual dimensionality reduction of high-dimensional data, followed by a graph-based clustering algorithm to categorize cells. Cell–cell communication analysis was conducted using the CellChat package. By constructing inter-cellular signaling pathway networks, this study quantified communication intensity and identified key signaling molecules among different cell subsets, thereby elucidating interaction patterns within the cervical cancer microenvironment. Pseudotime analysis was carried out with the Monocle package, which traced differentiation paths and dynamic changes of various cell subsets by constructing cell development trajectories, thereby clarifying the state transition rules of cells throughout disease progression. Ultimately, distinct cell subsets were identified, and cell type annotation was completed using singleR.

4.9. Preparation of XAFHO

Piper longum L (Batch No. 231140132), Alpinia officinarum Hance (Batch No. Q30025805), Piper nigrum L (Batch No. M30067608), Zingiber officinale Roscoe (Batch No. 240401) and Zaocys dhumnades (Batch No. 20250101) were kindly provided by Xinjiang Uyghur Autonomous Region Uyghur Medical Hospital (Urumqi, China). The herbs were identified by Dilixiati Siyiti, a professor of traditional Chinese medicine at Xinjiang Uyghur Autonomous Region Uyghur Medical Hospital. The composition of the formulation is shown in Table S3. Regarding the preparation of the medicinal product, specifically, the medicinal herbs were first ground into powder and then mixed with excipients (honey) to form the ointment base. We took 5 kg of the prepared ointment, which was soaked in 8 times the volume of 95% ethanol for 12 h. The mixture was heated in a water bath to 78 °C and reflux-extracted three times (3 h, 3 h, and 2 h). The mixture was then filtered while still hot through a gauze, and the filtration was combined and concentrated under reduced pressure. The concentrate was subsequently freeze-dried to remove any remaining solvent. The final extract obtained was the ethanol extract of Xiao-ai-fei honey ointment, weighing 2.25 kg (yield 45%). The known active constituents with therapeutic activity are piperine and galangin, and the genuine extract content ranges from 76% to 87%. Quercetin and kaempferol were also identified as confirmed constituents of XAFHO, and components of XAFHO were identified by GC-MS and provided by Uyghur Medical Hospital [21]. The extract was then dissolved in DMSO at a concentration of 4 mg/mL, followed by centrifugation at 12,000 rpm for 15 min to collect the supernatant. Subsequently, the supernatant was filtered using a 0.22 μm filter for future use. For application, the ointment should be diluted to the required concentration with a complete medium to make it ready for use. The material used in the in vitro experiments was the ethanolic extract of XAFHO; for simplicity, it is hereafter referred to as “XAFHO” unless otherwise specified.

4.10. Cell Proliferation Assays

Cell proliferation rates were assessed quantitatively using the CCK-8 assay kit by ZEN (Chengdu, China). For IC₅₀ determination, cells were seeded at a density of 5 × 10³ cells/well in 96-well plates and then treated with various concentrations of XAFHO (0–250 μg/mL) or cisplatin. For routine proliferation assays, cells were seeded at a density of 2 × 10³ cells/well in 96-well plates prior to treatment. The IC₅₀ value was calculated from the optical density readings at 450 nm. In the colony formation assay, cells were plated in 6-well dishes at a density of 1 × 10³ cells/well and incubated for 14 days. Colonies were fixed with 4% paraformaldehyde, stained with 0.1% crystal violet, and manually counted using a microscope.

4.11. Transwell Migration and Invasion Assays

HeLa and SiHa cells were seeded at a density of 3 × 10⁵ cells per well in the upper chamber of the transwell migration assay, with drugs introduced into the lower chamber. Subsequently, after 24 and 48 h, the migrated cells were fixed, stained with crystal violet, and quantified. In the matrigel invasion assay, the upper chambers were pre-coated with Matrigel (Corning, Corning, NY, USA) using procedures analogous to those of the migration assay.

4.12. EdU Assays

HeLa and SiHa cells were seeded at a density of 1 × 10⁵ cells per well in 24-well plates, cultured overnight, and then exposed to specified concentrations of test compounds for 24 or 48 h. Following treatment, the cells were subjected to a series of steps: incubation with 10 μM EdU (Elabscience, Wuhan, China) for 2 h at 37 °C, fixation with 4% paraformaldehyde for 15 min, permeabilization with 0.5% Triton X-100 for 20 min, and addition of the Click-iT reaction cocktail for 30 min in the dark. Subsequently, the cell nuclei were stained with Hoechst 33342 for 5 min. Proliferating cells, identified by EdU incorporation, were observed and captured using a fluorescence microscope.

4.13. Western Blot Assays

Total protein was extracted using RIPA buffer (Bioss, Beijing, China) supplemented with protease inhibitors. Protein concentrations were determined using the BCA assay kit (Biosharp, Hefei, China). Equal amounts of protein were separated by SDS-PAGE, transferred to PVDF membranes, and blocked with 6% non-fat milk. Membranes were incubated with primary antibodies overnight at 4 °C, followed by the corresponding secondary antibody (Abways, Shanghai, China, AB0103). Protein bands were visualized using the ChemiScope 6000 Touch imaging system (Clinx Science Instruments, Shanghai, China). This work utilized the following antibodies: b-actin (cat#AB0035, Abways), SPP1 (cat#0806-6, HUABIO, Woburn, MA, USA) and FASN (cat#66591-1-lg, Proteintech, Rosemont, IL, USA).

4.14. RT-qPCR

Total RNA was extracted from cervical cancer cells using an RNA Extraction Kit (Accurate Biotechnology, Guangzhou, China, AG21023). Reverse transcription of 1 μg of total RNA was performed with the SynScript III RT SupreMix for qPCR with gDNA Remover (Tsingke, Beijing, China, TSK314M). mRNA levels were quantified using 2 × TSINGKE Master qPCR Mix (Tsingke, TSE203) on a qTOWER3 G qPCR cycler (Analytik, Jena, Germany). The primer pairs of FASN were forward: 5′-TGAGAGATGGCTTGCTGGA-3′; reverse: 5′-CCGCTGTACTTGGGCTTG-3′. Those of SPP1 were forward: 5′-CACCTGTGCCATACCAGTTA-3′; reverse: 5′-TGTGTGCCTTTTTGTCCAAGC-3′. Those of Actin were forward: 5′-GCGAGAAGATGACCCAGATC-3′; reverse: 5′-CCAGTGGTACGGCCAGAGG-3′.

4.15. Cell Culture and Transfection

Cervical cancer cell lines including HeLa and SiHa and normal cervical epithelial cells (Ect1/E6E7) were acquired from West China Second University Hospital, Sichuan University (Chengdu, China). The cells were cultured in DMEM supplemented with 10% fetal bovine serum and 1% penicillin–streptomycin. Standard incubation conditions of 37 °C, 5% CO2, and humidity were consistently maintained.

For transfection experiments, we used two small interfering RNAs (siRNAs) targeting FASN and SPP1 obtained from GenePharma (Shanghai, China, si-SPP1: 5′-CCAUGAAGAUAUGCUGGUUTT-3′; si-FASN:5′-GGUAGUGAGUGGGAAGGUGUATT-3′; si-NC:5′-UUCUCCGAACGUGUCACGUTT-3′). Western blot analysis confirmed knockdown efficiency 48 h after transfection. Concurrent phenotypic experiments were carried out following the same protocol.

4.16. Molecular Docking

Molecular docking was performed using AutoDock Vina 1.5.7 to evaluate the binding interactions between XAFHO components and target proteins. The crystal structure of FASN (PDB ID: 3TJM) and the protein structure of SPP1 (UniProt ID: Q9BX95) were obtained and preprocessed using PyMOL 3.0.3. Ligands were energy-minimized using ChemDraw 20.0, converted into PDBQT format, and prepared for docking. The docking search space (grid box) was defined as follows: for FASN, size_x = 61.2, size_y = 54.7, and size_z = 68.8; for SPP1, size_x = 70.1, size_y = 45.3, and size_z = 71.7. Docking was performed with 20 independent runs, and the conformation with the lowest binding energy was selected for further analysis. The docking results were visualized using PyMOL 3.1 and analyzed with Discovery Studio 2021. All other parameters were set to default values.

4.17. Statistical Analysis

Statistical analyses were performed using R and GraphPad Prism 10.1.2. Categorical variables were assessed utilizing the chi-square test or Fisher’s exact test, selected based on expected cell frequencies. Survival outcomes were evaluated via Kaplan–Meier estimation, with group comparisons conducted using the log-rank test, and multivariate Cox proportional hazards regression was carried out. For all in vitro experiments involving group comparisons, the Kruskal–Wallis test was used regardless of whether two or more groups were compared. All experiments were conducted with at least three independent biological replicates, and data are expressed as means ± standard deviations. Statistical significance was set to p < 0.05, with significance being denoted by asterisks: n.s. (non-significant), * p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001, and **** p ≤ 0.0001. p-Values under 0.05 were deemed statistically significant.

5. Conclusions

In conclusion, this study integrated network pharmacology, single-cell analysis, and in vitro experiments to explore the potential anti-cervical cancer mechanism of XAFHO. By combining transcriptomic and network analyses, we identified 11 key genes and established a prognostic model associated with overall survival, mutation patterns, and immune microenvironment characteristics in cervical cancer. Single-cell analysis further suggested distinct cellular distributions of key targets, with FASN being mainly enriched in tumor cells and SPP1 being predominantly expressed in macrophages/monocytes. In vitro experiments supported the inhibitory effects of XAFHO on the malignant behaviors of cervical cancer cells, while molecular docking provided preliminary evidence that quercetin, kaempferol, and isorhamnetin may interact with FASN and SPP1. Collectively, these findings suggest that XAFHO may exert anti-cervical cancer effects through coordinated regulation of multiple components, targets, and cell types and provide a basis for further mechanistic and translational studies.