Curdione (C418576) and borneol (B119291) were purchased from Aladdin (China). Baofukang suppositories (Z46020058) were purchased from Bikai Pharmaceutical (China). LPS (L2880) was purchased from Sigma-Aldrich (USA). CpG (GC68897), and HPV E6 protein (GP25441) were purchased from GLPBIO (USA).

Antibodies: iNOS (22226-1-AP), TLR4 (66350-1-Ig), NF-κB p65 (10745-1-AP), CD274 (66248-1-Ig), ERK1/2 (66192-1-Ig), AKT (60203-2-IG) and STAT3 (60199-1-IG) were purchased from Proteintech (China). Antibodies: IL-6 (DF6087), p-STAT3 (AF3293), p-ERK1/2 (AF1014) and p-AKT (AF0016) were purchased from Affinity (USA). TLR9 (13674) and CD8a (98941) antibodies were purchased from CST (USA). Flow cytometry antibody: FITC-CD3 (E-AB-F1013C), APC-CD8a (E-AB-F1104E), PE-CD8a (E-AB-F1104D), APC-CD4 (E-AB-F1097E) and PE-CD4 (E-AB-F1097D) were purchased from Elabscience (China).

A retrospective cohort analysis was conducted to assess the therapeutic efficacy of Baofukang suppository in patients with bacterial mixed HPV infection. With the informed consent of the participants and ensuring their privacy, we collected the medical information and test results of female patients with HPV vaginal infection who visited the Gynecological Clinic of Wuhu Maternity and Child Healthcare Hospital in Anhui Province from June 2022 to June 2023. Patients were grouped based on their medical information, and underwent follow-up visits, with reassessments of vaginosis and HPV infection status at 3 and 5 months after the consultation. Two authors independently assessed patients for inclusion and risk of bias, extracted data and checked them for accuracy. All patients received basic treatment according to the 2021 Sexually Transmitted Infections Treatment Guidelines and were subsequently administered Baofukang suppositories in combination therapy based on their individual preferences (21). As a traditional Chinese medicine preparation, Baofukang suppository, while clinically proven beneficial for patients with HPV or BV infections, is not considered a necessary treatment option. Some patients are unwilling to undergo the treatment due to its vaginal administration route.

Excluding invalid data, a comprehensive dataset of 205 cases was successfully amassed for analysis. The valid data were categorized into two groups: a control group (n = 104) and a Baofukang suppository-treated group (n = 101), based on the administration status of Baofukang suppositories. This clinical cohort study revealed no statistically significant differences between the two groups of participants in terms of baseline characteristics and vaginal pathogenic microorganism examination indicators, including mean age, pregnancy history, delivery history, age at first pregnancy, sexual intercourse frequency, microbial diversity, microbial density, Nugent score, and Amsel (AV) criteria level ( Supplementary Table 1 , P > 0.05). The study protocol was approved by the Ethics Committee of Wuhu Maternal and Child Health Center (Approval Number: 20230006).

Inclusion criteria: 1. Female individuals with a history of sexual intercourse; 2. Meeting the relevant diagnostic criteria for infection with high-risk HPV types, with positive results in high-risk HPV tests; 3. No sexual intercourse, vaginal lavage, or intravaginal medication use within 3 days before the examination; 4. No receipt of any anti-HPV treatment within 4 weeks before consultation; 5. Patients with high-risk HPV positivity, undergoing colposcopy and biopsy, with pathological findings indicating inflammation or low-grade lesions; 6. Fulfilling the medication indications of this study; 7. Having signed the informed consent form.

Exclusion Criteria: 1. Presence of malignant tumors or precancerous lesions confirmed by pathological examination; 2. Existence of immune system or hematological disorders; 3. Concurrent dysfunction of other vital organs; 4. Cognitive or communication impairments; 5. Failure to adhere to medication instructions or discontinuation of treatment; 6. Pregnancy or lactation; 7. Receipt of other traditional Chinese medicine treatments within 7 days before the intervention; 8. Loss to follow-up.

Patients should administer one Baofukang suppository into the posterior fornix of the vagina after cleansing the external genitalia before bedtime, during non-menstrual periods. The administration should be done every other day, with a total of 10 administrations comprising one treatment course. Continuous treatment for three courses is recommended. During the treatment period, patients are advised to avoid bathtub baths and sexual intercourse, refrain from overwork, and ensure proper scheduling of rest and moderate exercise to enhance their immune resistance.

The RNAseq data for vaginal swabs of BV infection were sourced from the GSE113771 project in the GEO database (https://www.ncbi.nlm.nih.gov/geo/), where samples from BV-related (n = 3) and healthy (n = 3) subjects were collected (22). The RNAseq data for vaginal swabs of HPV infection were sourced from the GSE75132 project in the GEO database, where samples from high-risk HPV-infected individuals without cervical cancer (n = 10) and healthy (n = 11) subjects were collected (23). The normalization of sequencing data was accomplished via GEO2R (https://www.ncbi.nlm.nih.gov/geo/geo2r/).

The main active compounds of natural borneol and zedoary oil were collected from related literature, utilizing databases such as the Traditional Chinese Medicine Network pharmacology database (24) (TCMSP, https://old.tcmsp-e.com/tcmsp.php), the Traditional Chinese Medicine Integrated Database (25) (TCMID, http://www.megabionet.org/tcmid/), the Herbal Ingredients Targets Database (26) (HIT, http://hit2.badd-cao.net/), and other relevant sources (20). The detailed molecular information is presented in Supplementary Table 2 . Given that Baofukang suppository is intended for intravaginal use, the oral bioavailability of its main active ingredients is not a determining factor. Subsequently, drug target prediction for the major active ingredients was conducted using Swiss target prediction ( Supplementary Table 3 ).

The target genes associated with BV and HPV vaginal infections were collected and organized from GeneCards database (27) (https://www.genecards.org/) and DisGeNET (28) (http://www.disgenet.org/). Detailed disease targets are presented in Supplementary Excel 1 .

To evaluate the correlation between patients’ baseline characteristics (including mean age, pregnancy history, delivery history, age at first pregnancy, and sexual intercourse frequency) and indicators related to vaginosis (including microbial diversity, microbial density, Nugent level, AV level, and bacterial infection type) with HPV clearance rate, a Cox analysis is conducted. The univariate and multivariate Cox analysis was conducted by the survival coxph function of the R package.

The Gene Ontology database (GO, https://www.geneontology.org) groups terms of gene and gene product function into three different dimensions of the ontology, Cellular Component (CC), Molecular Function (MF), and Biological Process (BP), to help researchers identify sets of genes that are significantly enriched in specific biological processes. The Kyoto Encyclopedia of Genes and Genomes (KEGG, https://www.genome.jp/kegg) is a database for systematic analysis of gene function to identify the relevance of candidate targets in the corresponding biological functions and signaling pathways. GO and KEGG pathway analysis was performed by linking targets to the Database for Annotation, Visualization and Integrated Discovery (DAVID, https://david.ncifcrf.gov) (29). The top 20 GO terms and KEGG pathways with the smallest p-values were selected and uploaded to the bioinformatics mapping website (https://www.bioinformatics.com.cn) for graph generation.

The selected gene sets were imported into the STRING database (https://cn.string-db.org/) for PPI network analysis and TISSUS enrichment analysis, with confidence scores set to ≥ 0.7 (30, 31). Cytoscape 3.7.1 was used for data visualization for PPI network analysis (32). The CytoHubba plug-in within Cytoscape was employed to identify hub genes and rank them based on the MCC score (33).

Least Absolute Shrinkage and Selection Operator (LASSO) is a regression analysis method that performs gene selection and classification. Separate LASSO regression analyses were performed on the set of disease targets for BV and HPV vaginal infections to identify signature genes, using the R package “glmnet” (34). Characterization genes are detailed in Supplementary Table 4 .

CIBERSORT can use deconvolution with a collection of 22 immune cell signature genes to derive the proportion of various immune cells from tissue RNAseq data (35). For identifying potential immune cell targets related to BV and HPV vaginal infections, immune infiltration analyses were performed on co-DEGs, utilizing the R package “CIBERSORT”.

Lung epithelial cells (TC-1) expressing mouse HPV E6/E7 proteins, macrophages (Raw 264.7) and DCs (DC 2.4) were cultured in RPMI 1640 medium (Gibco, USA) supplemented with 10% FBS (Gibco, USA) in a humidified incubator containing 5% CO2. The co-culture model was based on Transwell 12 well plates (3401, Corning, USA), where DCs were mixed with macrophages in a 1:3 ratio in the upper chamber, and TC-1 cells were cultured in the lower chamber and synchronized with splenic lymphocytes in the stimulated addition chamber. Splenic lymphocytes were added to the lower chamber at the time of administration of the stimulus. Under unspecified conditions, Baofukang suppository, curdione and Borneol were all stimulated at 100 μg/mL for 24 hours. LPS stimulation at 2 μg/mL was employed to establish a model of bacterial infection-associated inflammation. To mimic a viral infection-associated inflammatory response, cells were stimulated with 1 μg/mL CpG in combination with 10 μg/mL HPV E6 protein. In the co-culture model, 20 ng/mL of IL-6 and IL-1β were used to induce CD274 expression in TC-1 cells.

For the qRT-PCR assay, total RNA was extracted from treated cells using Trizol reagent (Invitrogen, USA). 3 μg of mRNA was reverse transcribed to cDNA and qRT-PCR was performed to detect mRNA expression of specific genes (Vazyme, China). The primer sequences are shown in Supplementary Table 5 .

For the WB assays, cells were treated with RAPI lysate (Biosharp, China) containing protease and phosphatase inhibitors. The protein concentration was determined by the BCA Protein Assay Reagent (Pierce, USA). 45 μg of protein from each sample was separated by electrophoresis (10%, w/v) and transferred to the NC membranes (Amersham, USA). Membranes were blocked with 5% BSA (Biosharp, China) for 1 hour and then incubated with the corresponding antibodies overnight. After washing with TBST, they were incubated with secondary antibodies (Proteintech, USA) for 2 hours. Following another wash, the membrane was visualized by Universal Hood II (BIO-RAD, USA). The gray value of the bands was quantified using ImageJ.

Immunofluorescence staining was performed to detect the expression level and spatial distribution of specific proteins. The treated cells were fixed with 4% paraformaldehyde for 15 minutes at room temperature. After permeabilization with 0.1% Triton X-100 for 10 minutes, cells were blocked with 3% BSA for 30 minutes. After 3 washes with PBS, the cells were incubated overnight at 4°C with the indicated primary antibody and then with the fluorescent secondary antibody (Proteintech, USA) for 1 hour at room temperature.

Gardnerella vaginalis (ATCC 14018) was obtained from the American Type Culture Collection (ATCC). Colonies were inoculated onto blood agar plates and cultured in a 5% CO₂ atmosphere. Gradient dilutions were performed to ensure that Gardnerella vaginalis grew as individual colonies in the antimicrobial assay. Bacterial solutions were mixed with 100 μg/mL of Baofukang suppositories or Curdione + borneol, respectively, and then inoculated. The colonies were counted after 24-48 hours of incubation.

Female C57BL/6 mice (aged 6-8 weeks, weighing 18-20 g) were purchased from the Liaoning Changsheng Laboratory Animal Center (China) and maintained in an SPF-grade environment. Mice were anesthetized by intraperitoneal injection of 0.4ml/20g of 0.3% pentobarbital sodium. BV/HPV co-infections were mimicked by injecting 20 μL of TC-1 cell suspension (5 × 10⁵ cells/mL) mixed with 5 μg of LPS into the cervical wall tissue of mice. After the LPS/TC-1 cervical in situ injection model had been established for 4 days, vaginal lavage treatment was administered according to the group assignments. The vaginal lavage treatment consisted of a single dose of 0.5 mL, administered once every two days, for a total of seven treatments. Animal procedures and care were guided by the Animal Research: Reporting In Vivo Experiments (ARRIVE) guidelines. All protocols related to animal experiments in this study were approved by the Ethics Committee of Wuhu Maternal and Child Health Center (Approval Number: 2024009).

To extract mesenchymal cells from the vagina and cervix, the corresponding tissues were collected and minced. The tissues were digested with 0.125% trypsin and 0.8 mg/mL collagenase I at 37°C for 1 hour. After digestion, the tissues were passed through a 200-mesh strainer to obtain a single-cell suspension, which was subsequently cultured in RPMI 1640 medium containing 10% FBS for biocompatibility experiments.

For lymphocyte extraction, the appropriate tissues were collected and crushed in a mortar. The mixture was passed through a 200-mesh strainer. Cells were then collected by centrifugation, and excess erythrocytes were removed using an erythrocyte lysate solution. The single-cell suspensions were washed with PBS to obtain a pure single-cell suspension for flow cytometry.

The treated cells were collected and stained with appropriate concentrations of fluorescently labeled antibodies for 30 minutes, protected from light, at 37°C. After washing with PBS, the stained cells were stored in 2% paraformaldehyde awaiting detection. Flow cytometry was performed by LSRFortessa X-20 (BD, USA) and the data were analyzed by FlowJo.

For H&E staining, paraffin tissue sections were first deparaffinized, and then stained with hematoxylin for 5 minutes. After washing with ddH₂O, they were stained with eosin for 1 minute. After transparentization with alcohol and xylene, neutral resin was used to seal the slides.

For immunohistochemical staining, antigen retrieval was performed in citrate buffer solution (0.01 M) after deparaffinization of paraffin tissue sections. After treatment with 3% H₂O₂, the sections were incubated with the indicated antibodies overnight at 4°C. The following day, the sections were washed with PBS and incubated with the secondary antibodies for 30 minutes at room temperature. Subsequently, 3,3’-Diaminobenzidine (DAB) and hematoxylin were used for staining.

One-way analysis of variance (ANOVA) was used to compare data between multiple groups, while Student’s t-test was used to compare data between two groups. GraphPad Prism 8.0.2 is utilized for graphing and statistical analysis. Non-quantitative data from the clinical cohort information were compared using the nonparametric Mann-Whitney U test by SPSS 25.0. The threshold for statistical significance was set at P < 0.05.

We collected the medical information from patients diagnosed with vaginal HPV infection who visited a single center within one year from June 2022 and followed up continuously, with a total of 205 complete cases collected. Patients were categorized into control and Baofukang groups based on whether they were treated with Baofukang suppositories or not, and all patients with concurrent vaginosis were routinely treated. There were no group differences in age, sexual frequency, and number of pregnancies between the control and treatment groups ( Supplementary Table 1 ).

We investigated the impact of HPV-negative conversion in patients receiving conventional treatment. Multivariate Cox analysis showed only the type of initial bacterial infection was an independent influencing factor in the control group ( Figure 1A ). Although HPV will gradually be cured over time, the type of initial bacterial infection will affect the difficulty of HPV curing ( Figure 1B ). Introducing the Baofukan group and conducting a multivariate Cox analysis, the results showed that both the treatment with Baofukang suppository and the type of initial bacterial infection were independent influencing factors ( Figure 1C ). The complexity of initial bacterial infection still affected HPV cure in patients, but Baofukang suppository was effective in promoting cure of bacterial mixed HPV infections ( Figure 1D ). Concurrently, Baofukang suppository also mitigated the complexity of bacterial infection ( Figure 1E, F ). Nevertheless, we found that at therapeutic concentrations, Baofukang suppositories exhibited no significant bactericidal impact against Gardnerella vaginalis, a common pathogen associated with bacterial vaginosis ( Figure 1G, H ). Baofukang suppository can effectively promote the cure of HPV/BV co-infections, but not by directly killing the mixed infection pathogens and the mechanisms remain to be further investigated.

Bacterial infections in this cohort of HPV patients were predominantly BV alone, or BV combined with aerobic vaginitis (AV) or vulvovaginal candidiasis (VVC) infections, with mixed BV infections being the most prevalent (106/161). Consequently, BV was selected as a representative case of additional pathogen infection for further investigation.

The BV-related vaginal sample RNA sequencing (RNAseq) database GSE113771 and the HPV-infected vaginal sample RNAseq database GSE75132 were screened from the GEO database (35, 36). GSE113771 analysis showed 1,975 genes up-regulated and 3,819 genes down-regulated in BV-related vaginal epithelium ( Figure 2A ). GSE75132 analysis showed 1,036 genes up-regulated and 545 genes down-regulated in epithelial cells after HPV infection ( Figure 2B ). Venn diagram represented a total of 758 co-differently expressed genes (co-DEGs) in both databases ( Figure 2C ). GO analysis showed that co-DEGs were mainly enriched in biological processes such as immune response, inflammatory cytokine regulation and receptor recognition ( Figure 2D ; Supplementary Figure 4A, B ), and KEGG analysis showed that co-DEGs were mainly enriched in signaling pathways such as chemokines, NF-κB and FcγR mediated phagocytosis ( Figure 2E ).

Lasso regression analysis showed NFKBIA, PLAUR, SYK, and TLR2 as characteristic genes during BV, and 10 genes including CASP1, CD274, and CCL18, as characteristic genes during HPV infection ( Figure 2F, G ; Supplementary Table 4 ). These genes may be the core targets for regulating the disease process. Validation using Protein-Protein Interaction Networks (PPI) revealed the top 30 proteins with the highest maximum clique centrality (MCC) score, with TLR2, CASP1, and CD274 still labeled as core proteins (37–39) ( Figure 2H ). Proteins such as TNF-α, TLR4, IL-1β and CXCR4 have also been confirmed as immune-related genes (40, 41).

The immune cell composition was further analyzed by CIBERSORT (42). It revealed a significant infiltration of CD8⁺ T cells (P < 0.0001), resting DCs (P = 0.0033) and activated DCs (P = 0.0096) in BV-related vaginal epithelia ( Figure 2I ). Meanwhile, a significant infiltration of CD8⁺ T cells (P = 0.0204), Treg cells (P = 0.0470), activated NK cells (P = 0.0361) and activated mast cells (P = 0.0405) were observed in HPV-infected vaginal epithelia ( Figure 2J ). These results suggest that both innate and adaptive immune responses play crucial roles in BV/HPV co-infection, and they may jointly regulate disease progression through co-DEGs enriched signaling pathways and physiological processes.

Utilizing TCMSP, TCMID, HIT databases, and related literature (43–45), we identified 35 potential active substances within natural borneol and zedoary oil (primary ingredients of Baofukang suppository) ( Supplementary Table 2 ). The primary active ingredients were selected based on relative abundance, and their targets were predicted using SwissTargetPrediction (46, 47) ( Supplementary Table 3 ). Subsequently, Targets of BV and HPV vaginal infections were collected by Genecards and DisGeNET (48, 49), and analyzed for their co-targets of action with the active compounds. The top five proteins in the disease relevance score among the co-targets and their probability are shown in Figure 3A–D . ESR1, ESR2 and AR are mainly associated with sex hormone action, but have also been shown to play a key role in immune regulation (50). Proteins such as TLR9, CCR5 and PTGS2 are also important components of innate immune activation (51, 52). Proteins directly interacting with co-targets (sub-targets) were screened from co-DEGs by PPI and the top 60 proteins with the highest MCC scores were demonstrated ( Figure 3E ). The results showed that proteins such as CD274, TLR4, and IL-1β still appeared as strongly associated molecules, suggesting that the drug efficacy involves immunomodulation. GO and KEGG analysis demonstrated the co-targets and sub-targets were enriched in signaling pathways such as NF-κB, IL-6 regulation and ERK1/2 ( Figure 3F ; Supplementary Figure 6A-C ). TISSUES analysis showed the targets were enriched in physiological activities related to macrophages, T cells and neutrophils (53) ( Figure 3G ).

In summary, the co-targets and sub-targets of the primary active ingredients focused on regulating innate and adaptive immunity, encompassing pathogen recognition (e.g., TLR2, TLR9, CCR5, etc.), the secretion of inflammatory factors (e.g., IL-1β, IL-1α, TNF, etc.), T cell activation (e.g., ICAM-1, CD83, PTPRC, etc.), and other physiological processes.

Published literature has reported that in natural borneol, the relative abundance of borneol can reach 59%, while Dl-isoborneol can reach 38% (31) ( Supplementary Table 2 ). Due to the high overlap in their disease-related target predictions, borneol was selected as the representative active component ( Supplementary Table 3 ). In contrast, the active components of zedoary oil are more complex, primarily consisting of epicurzerenone, curzerene, curdione, curzerenone, and 1,8-cineole (30). However, the relative abundances of these components vary across different batches of zedoary oil, potentially due to variations in the growth environments of the extracted raw materials (30, 54, 55). Analysis of the disease-related target predictions indicated that curdione had the most abundant potential targets and its relative abundance can reach 20%, thereby making it the representative active component in zedoary oil ( Supplementary Table 2 , 3 ). Although epicurzerenone and 1,8-cineole exhibited higher relative abundances, their predicted targets suggested significantly lower biological activities compared to those of curdione ( Supplementary Table 2 , 3 ).

Based on the results of relative abundance and target prediction, we selected curdione and borneol as the representative active ingredients of Baofukang suppository to investigate their specific effects on immunity. In the classical inflammation model induced by LPS stimulation of Raw 264.7 cells, curdione, borneol and Baofukang suppositories dose-dependently inhibited the levels of IL-6, IL-1β and TLR4 mRNA expression ( Figure 4A-C ), while promoting the mRNA expression of IL-10 and TGF-β ( Figure 4D, E ). This suggests the main ingredient of Baofukang suppository may inhibit the excessive immune activation in BV/HPV co-infection. Based on the cytotoxicity and anti-inflammatory effects ( Supplementary Figure 2 ), 100 μg/mL of curdione or borneol was selected for the subsequent experiments. Co-stimulation with CpG and HPV core protein E6 was used to simulate the inflammatory stimulus of HPV infection. Both curdione and borneol effectively inhibited the mRNA expression of pro-inflammatory cytokines while promoting the mRNA expression of anti-inflammatory cytokines in macrophages under this model ( Figure 4F ). There may exist a synergistic effect between curdione and borneol, further enhancing their anti-inflammatory effect.

NF-κB signaling pathway has been shown in shared disease characteristics ( Figure 2 ) and drug-disease targets ( Figure 3 ), and it is a key regulatory pathway of the inflammatory response (56). We have validated NF-κB signaling pathway in LPS-simulated and CpG + E6-simulated models, respectively. Curdione and borneol effectively inhibited the phosphorylation of p65 protein in both models, which in turn blocked the activation of NF-κB signaling pathway ( Figure 4G, H ). Meanwhile, the protein levels of upstream TLR4 (bacterial infection response) (57), TLR9 (viral infection response) (51) and downstream iNOS were also inhibited synchronously ( Figure 4G, H ). Upon activation, the p65 protein undergoes phosphorylation and nuclear translocation to bind to target molecules, thereby influencing the expression of inflammatory factors (58). The intracellular distribution of phosphorylated p65 (p-p65) was monitored by confocal microscopy, and the results showed curdione and borneol prevented the accumulation of p-p65 in the nucleus as expected ( Figure 4I ).

Overall, curdione and borneol, the primary active ingredients of Baofukang suppositories, can effectively modulate the expression of inflammatory factors such as IL-6 and IL-1β in macrophages by inhibiting the phosphorylation of NF-κB p65 protein and nuclear translocation, thus preventing the innate immune over-activation in BV/HPV co-infection.

Next, we investigated whether curdione and borneol could directly kill HPV-infected epithelial cells. TC-1 cells (mouse lung epithelial cells), which express HPV E6 and E7 proteins and exhibit similar biological behaviors to HPV-induced cervical cancer cells, were utilized to simulate HPV-infected vaginal epithelial cells in vitro. The qRT-PCR results showed that borneol and curdione significantly inhibited the mRNA expression of HPV E7 and E6, respectively ( Figure 5A, B ). The MTT cell viability assay confirmed that curdione and/or borneol, as well as Baofukang suppositories, could only inhibit the growth of TC-1 cells but did not directly kill them ( Figure 5C ).

While curdione and borneol inhibited inflammatory factors such as IL-6 and IL-1β expression in macrophages, TNF-α and CASP1 mRNA expression levels were up-regulated ( Figure 5D, E ), which are associated with anti-tumor and anti-viral responses (59, 60). Furthermore, in the LPS or CpG + E6 stimulated DC 2.4 cell model, both borneol and curdione up-regulated IL-12 and IL-23 mRNA expression ( Figure 5F ), which efficiently regulate T cell activation (61). A co-culture model was used to simulate the action of innate versus adaptive immunocyte crosstalk on TC-1 cells. In this model, mixed cultures of Raw 264.7 and DC 2.4 cells were placed in the upper chamber of the transwell system, and TC-1 cells were cultured in the lower chamber along with mouse splenic lymphocytes upon stimulation ( Figure 5G ). Western blot results indicated curdione effectively inhibited the phosphorylation levels of AKT and ERK1/2 proteins in TC-1 cells located in the lower chamber ( Figure 5H ). Activation of AKT and ERK1/2 has been reported to be associated with cell proliferation and precancerous lesions (62). Furthermore, flow cytometry analysis of splenic lymphocyte composition revealed that co-administration of curdione and borneol significantly increased the proportion of CD3⁺ CD8⁺ T cells ( Figure 5I ). MTT assays demonstrated curdione + borneol combined with splenic lymphocytes exhibited a significant killing effect on TC-1 cells in the co-culture model ( Figure 5J ). These results suggest that curdione and borneol may modulate the crosstalk between innate and adaptive immune cells, thereby achieving the killing of HPV-infected cells via T cell activation.

It has been reported that IL-6 and IL-1β can stimulate tumor cells to express CD274 (PD-L1) proteins via activating the STAT3 signaling pathway, thereby facilitating tumor cell immune escape (22, 63). Considering the observed phenotype of curdione and borneol inhibiting macrophage secretion of IL-6 and IL-1β, we further explored the CD274 expression levels of TC-1 cells in a co-culture model. The qRT-PCR results showed that the level of CD274 mRNA expression in TC-1 cells was inhibited by curdione and borneol ( Figure 5K ). Western blot results showed the inflammatory environment, induced by the combined stimulation of macrophages and DCs with LPS/CpG + E6, promoted the phosphorylation of STAT3 protein and subsequently induced the expression of CD274 in TC-1 cells. Notably, this effect was reversed by both curdione and borneol ( Figure 5L ). Confocal microscopy results revealed that both direct stimulation with IL-6 and IL-1β, as well as combined LPS/CpG + E6 stimulation in the co-culture model, induced CD274 protein expression in TC-1 cells. The high CD274 expression observed in the co-culture stimulation model was inhibited by curdione + borneol ( Figure 5M ). The above results suggest that curdione and borneol effectively blocked STAT3 phosphorylation-mediated CD274 upregulation in TC-1 cells by inhibiting macrophage secretion of IL-6 and IL-1β. This may enhance the recognition and killing of HPV-infected cells by CD3⁺ CD8⁺ T cells. These findings also explain the clinical phenomenon wherein HPV is difficult to eradicate in cases of BV/HPV co-infection.

Finally, we investigated whether curdione and borneol could reproduce their immunomodulatory effects in vivo and eliminate HPV-infected epithelial cells in HPV/BV co-infections. Given that HPV exclusively infects humans, we employed TC-1 cells in combination with LPS orthotopic injection into the cervix of C57/B6 mice to mimic HPV/BV co-infections. Four days after the TC-1/LPS orthotopic injection, mice received intravaginal lavages with 100 μg/mL of curdione + borneol (combined treatment group) or PBS (control group) every two days for a total of seven treatments ( Figure 6A ). If curdione + borneol effectively promotes the clearance of HPV-infected epithelial cells in HPV/BV co-infections, it could prevent the formation of in situ HPV-associated cervical tumors derived from TC-1 cells. After 14 days, mice were euthanized, and their uteruses were harvested. In the control group, all TC-1 cells formed orthotopic tumors (11/11), whereas only 22.22% of mice in the curdione + borneol treatment group (2/9) developed orthotopic tumors ( Figure 6B , Supplementary Figure S12A, D ). When treated with curdione (3/5) or borneol (4/6) alone, the therapeutic effect was inferior to that of combination therapy ( Supplementary Figure S12B, C ). The cervical and vaginal tissues of the combined treatment group exhibited lower mRNA expression levels of IL-6, IL-1β and CD274, alongside higher mRNA expression levels of IL-10 and TGF-β ( Figure 6D ; Supplementary Figure S13 ). Mice in the treatment group maintained their weight, while mice in the control group experienced rapid weight loss due to tumor formation and vaginal inflammation ( Figure 6C ). Moreover, the mRNA expression levels of E6 and E7 in the cervical and vaginal tissues were significantly reduced after curdione + borneol treatment, reflecting the elimination of TC-1 cells in the treatment group ( Figure 6E, F ). While curdione or borneol alone exhibited some anti-inflammatory and antiviral effects, the therapeutic outcomes were less superior to combination therapy, and the mRNA levels of CD274 were also higher.

Further research was conducted to investigate the therapeutic effects of combination therapy. Flow cytometry analysis of spleens from the combined treatment group showed a significant increase in both the number of CD3⁺ CD8⁺ T cells and the ratio of CD3⁺ CD8⁺ to CD3⁺ CD4⁺ T cells ( Figure 6G, H ; Supplementary Figure 14 , 15 ). Furthermore, an elevation in the mRNA expression of TNF-α, IL-12, and IL-23 was detected in cervical and vaginal tissues ( Figure 6D ; Supplementary Figure 13 ). These results are consistent with the conclusion that curdione + borneol modulates innate-adaptive immune crosstalk in vitro, enhancing cytotoxic T cell activation. Analogously, flow cytometry of cervical tissues mirrored splenic findings, revealing an increase in both the quantity and proportion of infiltrating cytotoxic T cells ( Figures 6I, J ; Supplementary Figure 16 , 17 ).

H&E staining revealed a more normalized tissue and glandular architecture in the cervixes of the curdione + borneol treatment group, in contrast to the disorganized structure observed in the control group ( Figure 6K ; Supplementary Figure 18 ). Immunohistochemical staining demonstrated lower IL-6 expression and a higher infiltration of CD8a⁺ cells in the combined treatment group ( Figure 6L, M ; Supplementary Figure 19 , 20 ). Notably, CD274 protein expression was significantly lower in the combined treatment group compared to the control ( Figure 6N ; Supplementary Figure 21 ), which explains the pronounced infiltration of CD3⁺ CD8⁺ T cells in the cervical tissues of the combined treatment group and correlates with the reduced inflammation in this group. The primary active components of Baofukang suppository, curdione + borneol, demonstrated their ability to modulate both innate and adaptive immunity in the mouse model of HPV/BV co-infections. While activating cytotoxic T cells, they suppressed the secretion of IL-6 and IL-1β, thereby blocking the STAT3-CD274 axis-mediated immune evasion and enhancing T cell recognition and infiltration. This effective clearance of HPV-infected epithelial cells under mixed infection conditions facilitated the cure of persistent HPV infections.