Innovative peptide-loaded hybrid membrane nanovaccine ...

Introduction

Globally, cervical cancer represents a major health challenge as the fourth most prevalent malignancy in the female population.1,2 Recent data from 2022 highlight this burden, reporting over 661,000 new diagnoses and approximately 348,000 deaths worldwide.2 The development of cervical cancer is fundamentally driven by persistent infection with high-risk human papillomavirus (HPV). While epidemiological evidence indicates that the majority of HPV infections undergo spontaneous immune clearance within 12 to 24 months of exposure, approximately 10% of cases may persist and eventually progress to squamous intraepithelial lesions (SILs) or invasive cervical cancer.3,4 Although prophylactic HPV vaccine and cervical cancer screening programs have reduced cancer incidence, immunization provides negligible efficacy against established HPV infection, and no targeted antiviral therapies are approved for eradicating persistent HPV infection.5 Current treatments (e.g., loop electrosurgical excision/ablation therapy) achieve short-term pathological remission in high-grade SIL (HSIL) patients, while the posttreatment recurrence rate exceeds 10%, accompanied by risks of cervical injury.6 This clinical predicament necessitates long-term surveillance, even repeated colposcopy/biopsy, creating significant psychological distress for patients and clinical management dilemmas for physicians. Additionally, current strategies, such as surgery, radiation, chemotherapy, and immunotherapy, show limited efficacy against advanced cervical cancer, which is characterized by a 5-year survival rate of less than 20% and varying degrees of toxicity to adjacent healthy tissues.7,8 Confronted with these challenges, the 10-15-year progression window from HPV infection to invasive carcinoma provides a critical opportunity for developing stage-specific therapeutic strategies targeting HPV-related cervical diseases.8,9,10

Recent advancements in immunotherapy, particularly those involving therapeutic vaccines and immune checkpoint blockade, have shown great potential for inducing precise antitumor immunity with reduced systemic toxicity compared to conventional interventions.11,12,13 The antigenic component of a vaccine determines its ability to elicit an immune response.14 HPV infection presents unique therapeutic opportunities, as the persistent expression of viral oncogenes establishes defined tumor-specific antigens (TSAs) for targeted intervention.10 The process of malignant transformation is classically triggered by the persistence of high-risk HPV, primarily mediated by the expression of the early proteins E6 and E7. These oncoproteins function by neutralizing the tumor suppressors p53 and Rb, thereby serving as the primary targets for the development of therapeutic vaccines.15 In phase III clinical trials (NCT03721978), the E6/E7-targeted DNA vaccine VGX-3100 achieved a 27.6% rate of lesion regression and viral clearance among patients with HSILs (compared to 8.7% in the placebo group), yet the response remains insufficient to meet clinical demands. Beyond these canonical targets, growing evidence highlights E5 as a pivotal oncogenic driver that synergizes with E6/E7 during carcinogenesis. Our prior work demonstrated that E5 facilitates cell proliferation and cell cycle progression through the modulation of spindle checkpoint proteins, specifically Bub1 and Mad2, alongside the downregulation of p21.16 Notably, E5 expression manifests during initial infection phases and benign/premalignant stages, offering a critical biological window not addressed by existing E6/E7-focused vaccines.6 Building upon our group’s earlier validation that a multivalent strategy targeting E5, E6, and E7 can elicit more robust and durable T-lymphocyte responses, the current study incorporates E5 into a novel platform to extend therapeutic coverage to early HPV infection phases and potentially enable intervention in SIL.17

The tumor heterogeneity in cervical cancer manifests through differential mutational profiles across patients and even among subpopulations within individual tumors, with limited shared mutations.18 This biological complexity necessitates innovative vaccination strategies beyond single-antigen approaches. Whole-cell antigens have emerged as a promising alternative, leveraging tumor cell membranes, which preserve diverse antigenic information, including potential tumor-associated antigens (TAAs) and TSAs, while eliminating genetic material to enhance safety and clinical translatability.19,20 However, emerging evidence has revealed limitations in autologous tumor vaccines: suboptimal immunogenicity of tumor cell membranes, the presence of unquantified immunosuppressive factors, and indeterminate ratios between immunogenic epitopes and nonreactive antigens, introducing therapeutic uncertainties when relying solely on tumor membrane-derived antigens.21,22,23 Therefore, this study proposes a novel therapeutic strategy integrating tumor cell membranes with HPV16-specific E5/E6/E7 long peptides as dual-antigen sources. Tumor cell membranes preserve individualized broad-spectrum antigenic profiles, while the E5/E6/E7 long peptides enable synergistic and stage-adaptable immune targeting. This combinatorial approach aims to achieve comprehensive immunotherapeutic efficacy across the cervical carcinogenesis continuum through multiantigen cooperative activation of cytotoxic T-cell responses.

The low immunogenicity of tumor cell membranes or peptides often hinders effective immune recognition and activation. Incorporating potent adjuvants can substantially enhance immunogenicity.24 Bacterial-derived components function as effective natural adjuvants, leveraging the recognition of pathogen-associated molecular patterns (PAMPs) to trigger innate immune responses. In cancer vaccines, bacterial constituents enhance adaptive immunity to improve antitumor responses through Toll-like receptor (TLR) signaling pathways.22 Recent studies have shown that nanovaccines combining tumor cell membranes with bacterial outer membrane vesicles (OMVs) can effectively activate immune responses, leading to significant tumor growth inhibition and suppression of metastasis in murine models.25 While genetically attenuated bacteria can generate OMVs with reduced endotoxin content, optimizing the balance between adjuvant potency and biosafety remains a challenge.26 Notably, bacterial cytoplasmic membranes, structurally distinct from the cell wall and devoid of lipopolysaccharides (LPS) or other wall components, present lower acute toxicity risks, making them promising adjuvant candidates.27 Furthermore, biological membranes exhibit inherent stability and fluidity, allowing fusion between heterogeneous membrane structures to form thermodynamically stable hybrid membranes (HMs). This fusion capability provides a novel platform for engineering multifunctional membrane vaccines in cancer immunotherapy.28

The continuous evolution of nanotechnology has facilitated the engineering of hybrid vaccines that integrate tumor-specific antigens with immunostimulatory components. This approach leverages membrane fusion technology to create hybrid membranes (HMs) by combining tumor cell membranes and bacterial cytoplasmic membranes through a codelivery strategy.28 Such vaccines have demonstrated promising results, stimulating specific cellular immune responses against cervical cancer models while maintaining a biosafety profile. Taking advantage of this technology, this study developed a multifunctional nanovaccine (HM-NPs@LP) that synergistically combines the tumor cell membrane (encompassing TAAs), bacterial cytoplasmic membrane (effective immune adjuvants), and tumor-specific antigen HPV16 E5/E6/E7 long peptides through a codelivery strategy. This design establishes a novel strategy that targets both HPV16 E5/E6/E7 oncoproteins and whole-cell antigens, effectively eliminating malignant cells. Our approach aims to address key clinical challenges, such as precancerous lesions and cancer recurrence, while also serving as a potential paradigm for personalized tumor vaccine design. By considering the complexities associated with tumor heterogeneity and refractory disease, this platform may provide a valuable strategic framework for developing interventions against aggressive malignancies. In this work, we performed comprehensive in vitro and in vivo evaluations to investigate the immune-activating capacity of HM-NPs@LP, assessed its therapeutic performance in both immunocompetent tumor models and humanized HLA-A*02:01 transgenic models, and explored its potential for clinical translation.

Results

Construction and characterization of the HM-NPs@LP nanovaccine

To develop the novel nanovaccine HM-NPs@LP, we integrated the tumor cell membrane (TM), bacterial cytoplasmic membrane (EM), and HPV16 E5/E6/E7 long peptides (LP) into poly(lactide-co-glycolic acid) (PLGA) nanoparticles. As illustrated in Fig. 1a, the engineering of HM-NPs@LP was executed through three primary stages: the initial isolation and preparation of TM, EM, and the hybrid membrane (HM); the formulation of PLGA nanoparticles (NPs@LP) loaded with HPV16 long peptides, following the in vitro validation of the immunostimulatory potential of these peptides; and the surface coating of NPs or NPs@LP with different biological membranes (TM, EM, or HM), yielding diverse formulations including TM-NPs, EM-NPs, Mix NPs+LP, HM-NPs, and HM-NPs@LP. These nanoparticles were subsequently characterized in vitro for stability and biocompatibility evaluations. The isolation of TM from tumor cells and EM derived from E. coli strain DH5α followed previously established protocols.29,30 To further confirm the reproducibility of membrane preparations, proteomic profiling was conducted on six independent batches of EM and TM (Supplementary Fig. 1; Supplementary Table 1). The Venn diagrams demonstrated that a large majority of the identified proteins were consistently shared across all replicates, indicating high batch-to-batch uniformity of our membrane isolation procedures. Minor variations were observed among batches, which reflect inherent biological variability and do not affect the overall consistency or functional integrity of the membrane preparations.30

Fig. 1

Construction and physicochemical characterization of HM-NPs@LP. a Schematic representation of the HM-NPs@LP fabrication process. TMs were harvested from murine tumor tissues via surgical excision and isolation. EMs were obtained from E. coli DH5α through lysozyme-mediated cell wall removal and extraction buffer treatment. EM and TM were subsequently coextruded at a 3:1 protein mass ratio to generate HM vesicles. HM-NPs@LP were formed by coextruding these HM vesicles with PLGA nanoparticles (NPs) previously loaded with HPV16 long peptides (LP). The illustration was created using BioRender. b Morphological visualization of PLGA NPs, EM-NPs, TM-NPs, and HM-NPs@LP via TEM. Scale bars represent 100 nm. c–f Size distribution (c) and surface charge distributions (e) of various nanoparticle formulations. Corresponding quantitative analysis of particle size (d) and ζ-potential (f) is shown based on five independent replicates (n = 5). The results are presented as the mean ± SD. g Ultraviolet–visible absorbance spectra of PLGA NPs, LP, and NPs@LP. a.u. absorbance unit. h Analysis of protein signatures in TM-NPs, EM-NPs, and HM-NPs@LP using SDS‒PAGE with Coomassie Brilliant Blue staining. Yellow rectangle: EM-derived protein bands; red rectangle: TM-derived protein bands. All lanes were loaded with a consistent protein mass (15 μg). i Western blot verification of protein components in EM-NPs, TM-NPs, and HM-NPs@LP, with each sample containing 10 μg of total protein. Statistical significance was determined via one-way ANOVA followed by Bonferroni post hoc correction. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; ns denotes no significant difference

To synthesize HMs, EM and TM were fused via repeated passes through a 400-nm-cutoff extruder at protein mass ratios of 3:1, 1:1, and 1:3. The optimal ratio for DC activation was subsequently identified by quantifying the release of proinflammatory cytokines from BMDCs following a 24-hour incubation. Among these formulations, the 3:1 (EM:TM) protein mass ratio was found to elicit the most robust cytokine production (Supplementary Fig. 2a–c). HM-coated PLGA nanoparticles prepared with the same ratios were compared for BMDC uptake efficiency. Notably, the 3:1 (EM:TM) ratio showed the greatest internalization by BMDCs (Supplementary Fig. 2d–f). Integrating the results from both cytokine induction and cellular uptake profiles, the 3:1 (EM:TM) protein mass ratio was established for subsequent experiments. Based on our prior work and established studies,17,31 this study selected H-2Db binding epitopes with the most effective cytotoxic T lymphocyte (CTL) peptides: HPV16 E5 (LSVSTYTSL), E6 (NKPLCDLLI), and E7 (RAHYNIVTF). Thus, we synthesized a chimeric long peptide (LP) consisting of HPV16 E5, E6 and E7 peptides linked by AAY spacers (LSVSTYTSLAAYNKPLCDLLIAAYRAHYNIVTF). This linker is known to be cleaved by proteasomes in antigen-presenting cells, acting as a bridge between MHC class I-restricted CTL epitopes. The synthesis of PLGA NPs and NPs@LP was carried out via the double-emulsion method, in accordance with previously reported protocols.32 The ultraviolet–visible absorbance spectra of PLGA NPs, LP, and NPs@LP were evaluated to verify their successful conjugation (Fig. 1g). PLGA NPs exhibited a prominent absorbance peak at approximately 220 nm, whereas the LP displayed two absorbance peaks at approximately 200 nm and 275 nm. Notably, the spectrum of NPs@LP retained these signature peaks, demonstrating the successful incorporation of peptides into the nanoparticle system.

To generate the HM-NPs@LP, NPs@LP were coextruded with HM vesicles through a 200-nm-cutoff extruder; a similar procedure was applied to produce EM-NPs and TM-NPs by fusing PLGA NPs with their respective membrane vesicles. The assembly of the HM-NPs@LP was visualized via transmission electron microscopy (TEM), which revealed a well-defined spherical morphology (Fig. 1b). Surface charge measurements provided further evidence of successful membrane coating. At a 5:1 nanoparticle-to-membrane mass ratio, the ζ-potential of HM-NPs@LP shifted to -20.1 mV, representing a significant decrease compared to the -9.0 mV observed for bare PLGA NPs (Fig. 1e, f).33 Dynamic light scattering (DLS) measurements across five independent batches yielded an average hydrodynamic diameter of 177.33 ± 4.76 nm (Fig. 1c, d).

The stability profile of HM-NPs@LP was evaluated through both physicochemical and functional assays. Following lyophilization and subsequent resuspension in phosphate-buffered saline (PBS), the nanovaccines maintained their original particle size and ζ-potential (Supplementary Fig. 3a, b). Additionally, these properties remained stable during storage in PBS at 4°C for at least 8 days (Supplementary Fig. 3h, i). To further validate biological stability, BMDC uptake and activation assays were conducted to compare freshly prepared HM-NPs@LP with lyophilized/reconstituted formulations. No significant differences were observed in BMDC internalization or the secretion of key cytokines (IL-6, IL-1β, and TNF-α) when comparing fresh versus reconstituted formulations (Supplementary Fig. 3c–g). These findings indicate that lyophilization preserves both the physicochemical integrity and functional activity of HM-NPs@LP, supporting their potential suitability for long-term storage and translational applications.

The protein composition of these nanoparticles was evaluated using sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE), followed by visualization via Coomassie Brilliant Blue staining. The results showed that the overall protein profiles of HM-NPs@LP displayed a combined pattern of bands from both TM and EM, indicating that the vast majority of membrane proteins were successfully retained during the hybridization process (Fig. 1h).30 Specifically, the successful integration of both membrane sources was confirmed by the detection of filamentous temperature-sensitive protein Z (FtsZ), which serves as a bacterial cell division marker, and Na⁺/K⁺-ATPase, a representative integral membrane protein of mammalian cells (Fig. 1i). Collectively, these complementary findings demonstrate that the HM-coated nanoparticles successfully incorporated essential antigenic and structural components from both parental membranes.

In vitro immune cell activation by HM-NPs@LP

Professional antigen-presenting cells (APCs), represented by BMDCs, are essential to the induction of antigen-specific T-cell immunity through the internalization and processing of TAAs for presentation.34 As shown in Fig. 2a, we investigated the efficiency of various nanovaccines being taken up and processed by mouse BMDCs. Following an 8-hour incubation with rhodamine B-labeled nanovaccines, the resulting cell-associated fluorescence was quantified via flow cytometry to monitor internalization (Fig. 2b, c). Flow cytometry analysis revealed that the TM-NPs were less efficiently internalized by BMDCs, with an uptake efficiency of approximately 10%. Conversely, nanovaccine formulations incorporating components derived from bacterial cytoplasmic membranes (EM-NPs, Mix NPs+LP, HM-NPs, and HM-NPs@LP; Mix NPs+LP refers to a physically mixed formulation composed of EM-NPs and TM-NPs at a protein mass ratio of 3:1 together with long peptides, without coassembly into hybrid nanoparticles) exhibited significantly higher uptake efficiencies than the TM-NPs group, all exceeding 60% (Fig. 2b, c). This discrepancy likely arises from the relative lack of intrinsic adjuvant properties in tumor cell membrane components, which limits their recognition and efficient internalization by BMDCs.35,36 The bacterial cytoplasmic membrane possesses an abundance of pathogen-associated molecular patterns (PAMPs) that facilitate detection by BMDC pattern-recognition receptors, thereby driving the efficient internalization of the nanovaccine.23 To corroborate this hypothesis, we further examined the activation of the Toll-like receptor (TLR) signaling pathway in BMDCs. Western blot analysis showed that, compared with the control group, HM-NPs@LP upregulated the expression of TLR, accompanied by increased levels of MyD88 and phosphorylated NF-κB/p65 (p-p65, Fig. 2d). To determine the functional roles of TLR, we performed lentiviral-mediated shRNA knockdown in BMDCs. The results demonstrated that p-p65 was attenuated following the knockdown of TLR2. Modest decreases in p-p65 were also observed upon TLR1 or TLR6 silencing, although to a lesser extent than those in the TLR2 group. In contrast, TLR4 silencing showed minimal impact, suggesting that this pathway is not involved in activation. Collectively, these results demonstrate that the presence of bacterial membrane–derived PAMPs promotes BMDC recognition and activation primarily by engaging the canonical TLR2–MyD88–NF-κB signaling pathway (Supplementary Fig. 4). Additionally, immunofluorescence experiments confirmed the internalization and uptake of various nanovaccines by BMDCs (Fig. 2e). Our results demonstrated that combining TM and EM within a single nanoparticle (as a hybrid membrane system) did not significantly alter the uptake efficiency of HM-NPs and HM-NPs@LP compared to other nanovaccine groups. However, both groups showed a notable increase in BMDC uptake efficiency compared to single-component vaccines, aligning with prior research that emphasized the superior immunogenicity achieved by integrating antigens and adjuvants into a unified delivery platform.30,37

Fig. 2

HM-NPs@LP promoted BMDC activation through synchronized antigen-adjuvant delivery. a Experimental design for investigating in vitro immune cell activation. b, c Flow cytometric assessment of BMDC internalization after an 8-hour incubation with rhodamine B-labeled nanovaccines (gating strategy in Supplementary Fig. 5, n = 5). d Western blot analysis of TLR1, TLR2, TLR4, TLR6, MyD88, NF-κB/p65, and p-NF-κB/p65 in BMDCs after 24 hours of coculture with various nanoparticle formulations. β-Actin was used as the internal loading control. e Colocalization analysis of TM-NPs, EM-NPs, Mix NPs+LP (a formulation comprising EM-NPs and TM-NPs physically combined at a 3:1 protein mass ratio with long peptides but lacking the coassembled hybrid nanoparticle architecture), HM-NPs, and HM-NPs@LP (rhodamine B; red) with the nucleus (Hoechst 33342; blue) in BMDCs by immunofluorescence. White scale bars, 20 μm. f–h Proinflammatory cytokine levels in BMDC supernatants following a 24-hour treatment with various nanovaccines (n = 5). i–k Flow cytometry quantification of CD80+, CD86+, and MHC-II+ BMDCs after 24 hours of incubation with various nanovaccines (n = 5). l Representative confocal microscopy images showing BMDCs after 24 hours of exposure to Mix NPs+LP or HM-NPs@LP. Nuclei are visualized in blue (DAPI), while TM and EM membranes are identified by green (DiO) and red (DiD), respectively. Scale bars represent 20 μm. m Quantitative assessment of colocalization between TM (green) and EM (red) signals based on Pearson’s correlation coefficient as described in (l) (n = 15 cells). Statistical analysis was performed using one-way ANOVA followed by Bonferroni post hoc correction or unpaired Student’s t test, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; ns denotes no significant difference

Next, we investigated the ability of HM-NPs@LP to drive the maturation and activation of DCs. The transition of BMDCs toward an active and mature state was characterized by the upregulated expression of surface markers, specifically CD80 and CD86, along with the elevated release of proinflammatory cytokines.34,38 Nanovaccine formulations were incubated with BMDCs for a 24-hour period, after which the concentrations of secreted proinflammatory cytokines were determined (Fig. 2f–h). ELISA analysis revealed that HM-NPs@LP significantly increased the secretion of proinflammatory cytokines, including IL-6, IL-1β, and TNF-α, compared to other single-membrane or physically mixed formulations. These findings were further supported by flow cytometric analysis (Fig. 2i–k), which demonstrated the upregulation of the costimulatory molecules CD80 and CD86, consistent with the cytokine secretion results (Fig. 2f–h). Compared to the other groups (TM-NPs, EM-NPs, and Mix NPs+LP), the proportions of CD11c+ cells expressing CD80, CD86, or MHC-II were significantly higher in the HM-NPs@LP group (P < 0.0001), indicating a potent capacity for driving BMDC maturation. Thus, this study demonstrated that multicomponent nanovaccines (HM-NPs or HM-NPs@LP) exhibited enhanced antigen uptake by BMDCs compared to single-component formulations (TM-NPs and EM-NPs) and physical mixtures (Mix NPs+LP). Both HM-NPs and HM-NPs@LP promoted the uptake and activation of DCs, likely due to the immunostimulatory molecular patterns associated with the bacterial membrane.36 Importantly, the inclusion of HPV16 long peptides in HM-NPs@LP further provides defined antigens that enable specific T-cell responses. We reasoned that the observed superiority over Mix NPs+LP may arise from the coordinated codelivery of tumor antigens and bacterial components within the same nanoparticle, which could help promote more synchronized internalization and processing by DCs, whereas physically mixed components may be taken up separately and thus exhibit reduced synergy.28,36 To directly verify this mechanism, we examined the intracellular colocalization of TM and EM components within Mix NPs+LP and HM-NPs@LP after incubation with BMDCs for 24 hours. As shown in Fig. 2l, TM and EM were labeled with DiO (green fluorescence) and DiD (red fluorescence), respectively. A subsequent quantitative analysis based on Pearson’s correlation coefficient indicated that HM-NPs@LP displayed a significantly higher degree of colocalization than Mix NPs+LP (P < 0.0001), indicating that hybrid membrane fusion enables more efficient integration of both components within the same cell (Fig. 2l, m). These findings confirm the hypothesis that the hybrid membrane platform facilitates the synchronized delivery of tumor antigens and bacterial adjuvants within a single nanoparticle, which leads to superior immune activation compared to that observed with simple physical mixtures.

HM-NPs@LP accumulated in inguinal LNs and elicited immune responses in the LNs and spleen

To evaluate the in vivo maturation of BMDCs and activation of splenic T cells, female C57BL/6 mice were first established with murine TC-1 HPV16+ tumor cells via subcutaneous inoculation. Once the tumors attained a volume of 300 to 500 mm3, they were surgically excised to prepare the vaccine formulations, as illustrated in Fig. 3a. To examine whether HM-NPs@LP can achieve accumulation in inguinal LNs, nanovaccines were labeled with the fluorescent dye DiR, followed by subcutaneous administration into the dorsal region of C57BL/6 mice (Fig. 3a; Supplementary Fig. 6). In vivo imaging of the fluorescence distribution at various time points postinjection revealed a progressive accumulation of DiR signals within the bilateral inguinal LNs (Fig. 3b). Quantitative analysis confirmed that the peak fluorescence intensity in the LNs was achieved at 48 hours (Fig. 3c). To further validate these targeting properties, mice received subcutaneous injections of IR-780-labeled nanovaccines. Ex vivo imaging of the excised inguinal LNs at 24 hours using an in vivo imaging system (IVIS) revealed that the fluorescence intensity in the HM-NPs@LP group was 3.63-fold and 10.29-fold higher than that observed in the EM-NPs and TM-NPs groups, respectively (Fig. 3d, e). Furthermore, to achieve a more intuitive observation of the accumulation of HM-NPs@LP within LNs, 3,3′-dioctadecyloxacarbocyanine perchlorate (DiO)-labeled HM-NPs@LP was subcutaneously administered to mice, and intravital microscopy imaging was performed 48 hours later on the left inguinal LN (Fig. 3f). Intravital microscopy fluorescence imaging and three-dimensional (3D) reconstruction revealed significant accumulation of HM-NPs@LP around the nuclei of cells within the lymph nodes (Fig. 3g; Supplementary Movies 1, 2), demonstrating that HM-NPs@LP efficiently migrated to LNs to facilitate the subsequent uptake of tumor antigens.

Fig. 3

HM-NPs@LP accumulated in inguinal lymph nodes and promoted DC maturation and splenic T-cell activation in vivo. a Experimental design for the in vivo studies. b Representative in vivo fluorescence images captured at the indicated intervals following subcutaneous administration of 0.9% saline or DiR-labeled HM-NPs@LP into the dorsal region of mice. Representative in vivo fluorescence images of subcutaneous injections into the dorsal region of mice are presented in Supplementary Fig. 6. c Longitudinal quantification of mean fluorescence intensity (MFI) for DiR-labeled HM-NPs@LP at the inguinal lymph node sites (n = 3). d Ex vivo fluorescence images of excised bilateral draining lymph nodes 24 hours post-injection. IR-780-labeled EM-NPs, TM-NPs, or HM-NPs@LP were subcutaneously injected into mice (n = 5). e Fluorescence quantification of (d) (n = 5). f Schematic illustration of intravital microscopy imaging. Mice received a subcutaneous dorsal injection of DiO-labeled HM-NPs@LP. Intravital microscopy imaging of the left inguinal lymph node was performed 48 hours after the injection. g Representative intravital microscopy images obtained 48 hours following subcutaneous administration of DiO-labeled HM-NPs@LP. Flow cytometric assessment of h CD80+ and i CD86 + DC populations in inguinal LNs following subcutaneous immunization of mice (n = 5). j–l Serum levels of proinflammatory cytokines measured after subcutaneous vaccination of mice (n = 5). Representative images of the IFN-γ ELISPOT assay (m) along with the corresponding spot counts (n, o) (n = 5). The results are presented as the mean ± SD. Statistical analysis was performed using one-way ANOVA followed by Bonferroni post hoc correction. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; ns denotes no significant difference

The in vivo maturation of LN-resident BMDCs was evaluated by subcutaneously administering various nanovaccine formulations to female C57BL/6 mice on days 0, 2, and 6. At 24 hours post-final immunization, the expression of costimulatory markers on DCs within the inguinal LNs was quantified. The results showed that CD80 and CD86 levels on DCs were significantly upregulated following treatment with nanovaccines. Among all formulations, the HM-NPs and HM-NPs@LP groups exhibited the most pronounced marker expression (Fig. 3h, i), consistent with the maturation profiles observed in vitro (Fig. 2i, j). Furthermore, cytokine profiling revealed that HM-NPs@LP elicited significant increases in serum concentrations of IL-6, IL-1β, and TNF-α (Fig. 3j–l), confirming its potent capacity for systemic immunostimulation.

To evaluate HPV16-specific T-cell immunity, splenocytes were incubated with the long peptides or TC-1 cell membrane vesicles for 18 hours, and ELISPOT assays were conducted (Fig. 3m). Under peptide stimulation (Fig. 3m, n), HM-NPs@LP and EM-NPs@LP induced comparably high frequencies of IFN-γ-secreting T cells, whereas the Peptides, TM-NPs@LP, and Mix NPs+LP groups showed only minimal activity, and peptide-free formulations remained in the background. When stimulated with TM vesicles, HM-NPs and HM-NPs@LP elicited strong responses, while TM-NPs and TM-NPs@LP produced only weak activation, highlighting the essential role of EM adjuvanticity in facilitating T-cell immunity (Fig. 3m, o). To further validate functional cytotoxicity, splenocytes from vaccinated mice were tested in cytotoxicity assays using HPV16⁺ TC-1 cells and HPV⁻ controls. T cells from HM-NPs@LP–immunized mice exhibited the most potent and antigen-specific cytotoxic activity, selectively eliminating HPV16⁺ TC-1 cells while sparing HPV⁻ U14 and MC38 targets (Supplementary Fig. 7). Collectively, these findings demonstrate that HM-NPs@LP can enhance DC maturation through synchronized codelivery of tumor antigens, long peptides, and EM adjuvants, thereby eliciting robust production of HPV16-specific T-cell responses.

HM-NPs@LP exhibited potent antitumor efficacy in a murine TC-1 tumor model

Once the capacity of HM-NPs@LP to stimulate DC maturation and T lymphocyte activation in vivo was established, its potential to inhibit tumor progression was evaluated (Fig. 4a). Female C57BL/6 mice were subcutaneously inoculated with murine TC-1-luciferase HPV16+ cells, and the resulting tumor tissues were harvested to prepare the nanovaccine formulations as previously described (Fig. 3a). When the subcutaneous tumor volume reached approximately 50 mm³, tumor-bearing mice were randomly allocated into different treatment groups. The vaccination regimen was performed subcutaneously according to a three-dose schedule on days 10, 12, and 16, a regimen selected based on previous studies showing suppression of tumor growth using bacterial membrane-derived vaccine platforms (Fig. 4a).30 IVIS imaging revealed a gradual expansion of tumor volumes in all groups except for the HM-NPs@LP group (Fig. 4b). As demonstrated in Fig. 4c, d, tumor growth remained unaffected in mice receiving peptides, while the Mix NPs+LP treatment provided only a modest delay in progression with limited antitumor activity. In contrast, HM-NPs@LP immunization achieved superior therapeutic efficacy, yielding complete tumor regression in 20% (1/5) of mice by day 31. Importantly, the HM-NPs@LP group exhibited a downward trend in tumor volume from approximately day 24 compared to the peptide-free HM-NPs group, although the difference did not reach statistical significance at the study endpoint (Fig. 4c–f). These findings establish that codelivery of TM, EM adjuvant, and HPV16 long peptides within a single nanoparticle generates synergistic antitumor immunity, highlighting the potency of this multicomponent nanoplatform in achieving therapeutic outcomes.

Fig. 4

HM-NPs@LP induced tumor regression and modulated the immunosuppressive TME in a murine TC-1 model. a Schematic of the experimental timeline for the therapeutic studies. Female C57BL/6 mice (n = 5) received a subcutaneous inoculation of murine TC-1-luciferase HPV16+ cells at a dose of 2 × 105 per mouse. Upon tumors reaching approximately 50 mm3 in volume, the vaccination regimen was administered on days 10, 12, and 16, followed by weekly monitoring using IVIS imaging. b Longitudinal in vivo bioluminescence imaging documenting tumor progression in mice across various treatment groups at the indicated time intervals. Individual (d) and average (c) tumor growth curves. e Representative images displaying tumors harvested at the study endpoint (n = 5). f Tumor weights of TC-1-luciferase (n = 5). g Representative images showing terminal deoxynucleotidyl transferase (TdT) dUTP nick-end labeling (TUNEL)-positive cells and Ki67-labeled proliferating cells within TC-1 tumor tissues. Scale bars, 20 μm (TUNEL assay) and 100 μm (Ki67 immunostaining). Analysis of the quantification of TUNEL-positive regions (h) and Ki67+ cells (i). j Multiplex immunohistochemistry imaging of TC-1 tumors immunolabeled for CD4 (red), CD8 (pink), PD-1 (green), and Foxp3 (yellow). Tumor cell nuclei were stained with DAPI (blue). The top panels for each treatment group provide broad field-of-view results with a scale bar of 50 μm, where white dashed boxes highlight the magnified regions. The white solid arrows indicate exhausted CD8 + T cells (Tex, CD8+ PD-1+), and the white dashed arrows indicate regulatory T cells (Treg, CD4+ Foxp3+). All scale bars are set at 20 μm. Statistical analysis was performed using two-way ANOVA followed by Bonferroni post hoc correction for analyzing tumor growth kinetics. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; ns denotes no significant difference

To elucidate the immune-mediated mechanisms contributing to the antitumor activity of HM-NPs@LP, we performed a comparative profiling of tumor immune microenvironments across the experimental groups. The composition of tumor-infiltrating lymphocytes (TILs) was characterized via flow cytometry, revealing that HM-NPs@LP immunization promoted an increase in CD8 + T-cell infiltration and elevated CTL frequencies (Supplementary Fig. 8), establishing a mechanistic correlation between lymphocyte recruitment and therapeutic efficacy. The Peptides and Mix NPs+LP groups demonstrated diminished cytotoxic lymphocyte infiltration, aligning with the tumor growth inhibition observed in Fig. 4c. Moreover, TUNEL apoptosis assays revealed significantly increased DNA damage rates in the HM-NPs@LP group (Fig. 4g, h). Immunohistochemical analysis of Ki67, a proliferation marker linked to tumor aggressiveness and poor prognosis,39 confirmed relatively low Ki67 expression in HM-NPs@LP-treated tumors (Fig. 4g, i), indicative of suppressed malignant proliferation.

Regulatory T cells (Tregs), defined by the CD4 + CD25+ Foxp3+ phenotype and acting as central mediators of immune tolerance, contribute to tumor immune evasion by releasing immunosuppressive factors, such as IL-10 and TGF-β, which in turn foster tumor progression.8,40 Emerging evidence suggests that the increased presence of both Tregs and exhausted CD8 + T cells (Tex, characterized by CD8 + PD-1 + ) within the tumor microenvironment, along with reduced helper T cells and CTLs, are hallmarks of cervical cancer progression and metastasis.41 To visually demonstrate the local infiltration of various lymphocytes in the tumor microenvironment, multiplex immunohistochemistry (mIHC) was implemented to visualize the distribution of various immune subsets, including CD4+ helper T cells, CD8+ cytotoxic T lymphocytes, Tex, and Tregs. As shown in Fig. 4j, vaccine formulations with suboptimal tumor suppression (Peptides and Mix NPs+LP groups) and moderate therapeutic efficacy (HM-NPs group) exhibited limited CD8 + T-cell recruitment accompanied by the upregulated expression of PD-1 and Foxp3. Notably, these groups demonstrated significant expansion of immunosuppressive Tregs (white-dashed arrows) and exhausted T cells (white-solid arrows), both of which are associated with immunotherapy resistance and T-cell dysfunction. Remarkably, the HM-NPs@LP group displayed substantially enhanced CD8+ T-cell infiltration with minimal PD-1 or Foxp3 expression, accompanied by a marked reduction in both Treg and Tex populations.

Collectively, these findings indicated that the enhanced therapeutic efficiency of HM-NPs@LP likely stemmed from its integrated triple-component delivery platform, which codelivered TM antigens, bacterial adjuvants, and HPV-specific long peptides. This strategic formulation not only enhanced immunocyte infiltration but also effectively counteracted the formation of an immunosuppressive tumor microenvironment, thereby achieving potent and sustained antitumor effects. To further assess whether these benefits extend to a more clinically relevant anatomical setting, we established an orthotopic cervical model. Consistent with our subcutaneous results (Fig. 4), HM-NPs@LP markedly restricted orthotopic tumor progression compared with saline or Mix NPs+LP, as monitored by longitudinal bioluminescence imaging (Supplementary Fig. 9), providing evidence of therapeutic potential in a cervical niche.

HM-NPs@LP induced a long-lasting antitumor response

To evaluate the durability of the immune protection, a TC-1 tumor rechallenge model was established using surviving mice from the HM-NPs@LP-treated group beyond day 60 (Supplementary Fig. 10a). These mice were assigned to receive a secondary challenge with either PBS or TC-1 HPV16+ cells to assess the potential for a long-term antitumor response. Remarkably, mice subjected to TC-1 tumor rechallenge exhibited protection against tumor development, achieving a 100% inhibition rate (Supplementary Fig. 10b, c). No evidence of tumor recurrence was observed in the rechallenged subjects throughout a 90-day observation period. Furthermore, serum analysis revealed that the concentrations of proinflammatory cytokines were elevated in the rechallenged mice relative to the control group (Supplementary Fig. 10d–f). This systemic response was accompanied by an expansion of effector memory T cells (Tem) compared with the naïve T-cell population (Supplementary Fig. 10g, h). Collectively, these findings demonstrated that the HM-NPs@LP nanovaccine not only prevented tumor recurrence but also established a state of long-lasting immunological protection.

HM-NPs@LP elicited antigen-specific T cells and tumor regression in a humanized HLA-A*02:01 transgenic mouse model

To assess the translational potential of this nanovaccine within a humanized immunological context, a specialized tumor model, TC-1-A2-E5, was established via lentiviral-mediated coexpression of HLA-A*02:01 (human leukocyte antigen) and the HPV16 E5 oncogene into TC-1 cells. The expression of HLA-A*02:01 and HPV16 E5 in TC-1-A2-E5 cells was verified through flow cytometry and qRT‒PCR (Supplementary Fig. 11a, b). Building on our previous studies and earlier reports,42,43,44 CTL-specific epitopes from HPV16 E5 (YIIFVYIPL), E6 (TIHDIILECV), and E7 (YMLDLQPET) were selected for vaccine construction based on their high binding affinity for HLA-A*02:01. We then synthesized a chimeric long peptide by integrating these sequences via AAY spacers (YIIFVYIPLAAYTIHDIILECVAAYYMLDLQPET) to ensure efficient antigen processing and constructed HLA-A*02:01 HPV16 E5, E6, and E7 tetramers (Supplementary Fig. 12). To assess the in vivo efficacy of humanized HM-NPs@LP in activating naïve T cells, we utilized humanized HLA-A*02:01 transgenic mice inoculated with subcutaneous TC-1-A2-E5 tumors. The humanized HLA-A*02:01 transgenic mice utilized in this study feature a chimeric HLA class I molecule, comprising human HLA-A2.1 α1 and α2 domains fused to the murine H-2Dd α3 transmembrane and cytoplasmic regions.45 The expression of HLA in these transgenic mice was validated with an anti-HLA-ABC antibody followed by flow cytometric analysis (Supplementary Fig. 13). To prepare the vaccine formulations, tumors were surgically removed for preparation of the vaccine formulations when they reached a volume of 300-500 mm3. Subsequently, another batch of female HLA-A*02:01 transgenic mice was immunized subcutaneously on a three-dose schedule (days 0, 2, and 6) (Fig. 5a). One week following the final injection, splenocytes were collected and subjected to an 18-hour coculture with either HPV16 long peptides or TC-1-A2-E5 membrane-derived vesicles. ELISPOT assays revealed that HM-NPs@LP induced significantly higher IFN-γ–producing T-cell responses than Mix NPs+LP upon stimulation with HPV16 long peptides or TC-1-A2-E5 membranes (Fig. 5b–d), which is in line with the potent immunogenicity observed in Fig. 3m–o and related reports.20,30,36 Moreover, splenocytes from HM-NPs@LP–immunized mice exhibited strong ex vivo cytotoxicity against HPV16⁺ TC-1-A2-E5 cells but not HPV⁻ controls, confirming antigen-specific killing (Supplementary Fig. 14), consistent with the antigen-specific immune activation observed in C57BL/6 mice (Fig. 3). Antigen-specific T-cell responses against HPV16 E5, E6, or E7 were quantified via flow cytometry using HLA-A*02:01-restricted tetramers. As shown in Fig. 5e–h, the HM-NPs@LP formulation elicited a higher proportion of these specific T-cell populations than the Mix NPs+LP group, a finding that was in alignment with the results obtained from ELISPOT assays (Fig. 5b). Additionally, cytokine analysis of the culture supernatants indicated that HM-NPs@LP treatment significantly enhanced the production of IL-6, IL-1β, and TNF-α, further demonstrating its potent immunostimulatory properties (Fig. 5i–k).

Fig. 5

HM-NPs@LP vaccination induced antigen-specific T-cell immunity and inhibited tumor growth in humanized HLA-A*02:01 transgenic mice. a Overview of the in vivo experimental procedure. Representative images of the IFN-γ ELISPOT assay (b) and the associated statistical analysis of spot counts (c, d) in HLA-A*02:01 transgenic mice (n = 4). e–h Flow cytometric scatterplots and frequencies of HPV16 E5/E6/E7-specific T-cell subsets within the CD3+ population (%) from the control, Mix NPs+LP, or HM-NPs@LP groups (n = 4). The Y-axis denotes labeling with the specified tetramers, whereas the X-axis identifies CD3+ cells. i–k Levels of proinflammatory cytokines in splenocyte culture supernatants following a 24-hour stimulation with HPV16 long peptides (n = 4). l Experimental design. Female HLA-A*02:01 transgenic mice (n = 4) were established with TC-1-A2-E5 cells via subcutaneous injection of 1 × 106 cells per mouse. Mice were immunized on days 11, 13 and 17 posttumor inoculation (tumor reached a volume of approximately 50 mm3). Representative tumor images (n), mean tumor growth kinetics (m), and final tumor weights (o) for the TC-1-A2-E5 model (n = 4). p TUNEL assay and Ki67 immunostaining images of TC-1-A2-E5 tumors. Scale bars, 20 μm (TUNEL assay) and 50 μm (Ki67 immunostaining). Quantification of TUNEL-positive regions (q) and Ki67+ cells (r) (n = 4 for Control and Mix NPs+LP groups; n = 2 for HM-NPs@LP group). Statistical analysis involving the HM-NPs@LP group was precluded due to the limited number of residual tumors available, as 2 mice in this group achieved complete tumor regression by the study endpoint. Data are presented as the mean ± SD. Statistical analysis was performed using either one-way or two-way ANOVA followed by Bonferroni post hoc correction for analyzing tumor progression. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; ns denotes no significant difference

To evaluate the therapeutic potential of HM-NPs@LP within the HLA-humanized context, female HLA-A*02:01 transgenic mice were subcutaneously inoculated with TC-1-A2-E5 cells (Fig. 5l). As shown in Fig. 5m–o, HM-NPs@LP immunization yielded antitumor efficacy and a reduction in tumor burden, paralleling the observations previously recorded in the murine TC-1 tumor model (Fig. 4b–f). Furthermore, TUNEL and Ki67 analyses revealed enhanced DNA fragmentation alongside suppressed malignant proliferation in the HM-NPs@LP-treated group relative to the control and Mix NPs+LP groups (Fig. 5p–r), consistent with findings in murine TC-1 tumor models (Fig. 4g–i). Taken together, these data indicate that HM-NPs@LP could generate potent systemic immunity and antigen-specific T-cell responses targeting HPV16 E5, E6, and E7 within HLA-A*02:01-humanized tumor environments.

HM-NPs@LP enhanced therapeutic responses to TP and anti-PD-1 therapy in a humanized HLA-A*02:01 transgenic mouse model

Considering the modest response rates of immune checkpoint blockade monotherapy in cervical cancer,8,46 we explored whether the integration of HM-NPs@LP with taxane-platinum (TP) chemotherapy and anti-PD-1 could optimize therapeutic benefit. A combination of anti-PD-1 and TP was utilized as the foundational treatment regimen. Given that tumor-derived TC-1 cells exhibit significant PD-L1 expression (Supplementary Fig. 15), thereby offering a clear target for checkpoint blockade, the synergistic potential of HM-NPs@LP in conjunction with anti-PD-1 and TP was investigated in vivo (Fig. 6a). Despite the limited antitumor activity observed in groups receiving anti-PD-1, TP, or anti-PD-1 + TP, formulations containing HM-NPs@LP induced a profound suppression of tumor growth (Fig. 6b, c). These findings were corroborated by flow cytometric analysis, which identified high levels of CD8 + T-cell recruitment and functional activation in the HM-NPs@LP-containing groups (Supplementary Fig. 16). Collectively, these data indicate that HM-NPs@LP acts synergistically with chemotherapy and PD-1 blockade to amplify antitumor immunity. Consequently, HM‑NPs@LP represents a promising strategy for treating advanced cervical cancer, as well as HPV infection and cervical intraepithelial neoplasia.

Fig. 6

HM-NPs@LP vaccination enhanced therapeutic responses to TP and anti-PD-1 therapy in a humanized HLA-A*02:01 transgenic mouse model. a Flowchart of the TP and anti-PD-1 combination regimen within the TC-1-A2-E5 subcutaneous tumor model. Mice were systematically treated with monotherapies (αPD-1, TP, or HM-NPs@LP), dual-combination therapies (TP + αPD-1, HM-NPs@LP + TP, or HM-NPs@LP + αPD-1), or a triple-combination regimen (HM-NPs@LP + TP + αPD-1) to evaluate synergistic antitumor efficacy (n = 5 for each group). Mean (b) and individual (c) tumor growth curves. Data are presented as the mean ± SD. Statistical analysis was performed using two-way ANOVA followed by Bonferroni post hoc correction for analyzing tumor progression. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; ns denotes no significant difference

HM-NPs@LP exhibited a favorable biosafety profile

Given that EM incorporated into HM-NPs@LP might harbor residual impurities such as lipopolysaccharides (LPS) from bacterial cell walls during the preparation process, we conducted a biosafety evaluation.47 Quantitative analysis of LPS across different vaccine formulations showed that endotoxin levels in HM-NPs@LP were comparable to those of the LPS-free control group (Supplementary Fig. 17a), suggesting that the majority of bacterial LPS had been effectively removed. Furthermore, we conducted hemolysis analysis by incubating erythrocytes with TM-NPs, EM-NPs, HM-NPs, and HM-NPs@LP. The hemolytic percentage recorded for the HM-NPs@LP group remained below 3%, which is well within the accepted safety threshold for biomedical materials (Supplementary Fig. 17b, c).30 Additional hemolysis tests were conducted using erythrocytes treated with various concentrations of HM-NPs@LP. No obvious hemolysis was observed across the tested concentration range of HM-NPs@LP (Supplementary Fig. 17d, e). As components derived from bacterial pathogens could theoretically provoke hepatic and renal toxicity through diverse pathogenic mechanisms,48 hepatic and renal functions were monitored during the in vivo studies. Histological examination revealed no apparent structural changes in any of the examined organs (Supplementary Fig. 18a). Specifically, no significant differences were observed in the serum activities of liver enzymes, including alanine aminotransferase (ALT) and aspartate aminotransferase (AST), in mice following treatment with HM-NPs@LP (Supplementary Fig. 18c, d). Similarly, the levels of renal function markers, such as blood urea nitrogen (BUN) and creatinine, were maintained within physiological ranges across all treatment groups (Supplementary Fig. 18e, f). Furthermore, body weight was continuously tracked during the therapeutic window to assess the systemic tolerance of the nanovaccines. The absence of notable weight fluctuations throughout the study period further underscores the biocompatibility of this platform (Supplementary Fig. 18b).25 Collectively, these results indicate that HM-NPs@LP possesses a favorable biosafety profile.

Discussion

Despite advancements in prophylactic HPV vaccination and systematic screening initiatives, cervical cancer persists as a leading gynecological malignancy worldwide.49,50 Although multiple prophylactic HPV vaccines are commercially available, they have not shown therapeutic efficacy against established HPV infections or related cervical pathologies. Current interventions for high-grade squamous intraepithelial lesions, including surgical resection and ablation therapies, are associated with significant risks of disease recurrence, cervical incompetence, and preterm delivery.6 For individuals diagnosed with advanced, recurrent, or metastatic cervical cancer, the standard of care generally involves combined chemoradiation and systemic therapies, yet these patients still face poor 5-year survival rates (10-20%) due to limited treatment efficacy.10 In this study, we developed and characterized HM-NPs@LP, an innovative nanovaccine that employs PLGA NPs for the codelivery of tumor cell membranes, bacterial cytoplasmic membranes, and HPV16 E5/E6/E7 long peptides to antigen-presenting cells. This tri-component system exhibited potent activation of tumor-specific cellular immunity in HPV-related tumor models. Furthermore, evaluation in humanized HLA-A*02:01 transgenic mice revealed comparable antitumor efficacy with conventional murine models, demonstrating translational potential for human applications. Additionally, to strengthen the clinical applicability of our observations, we expanded our research to encompass an orthotopic cervical model. HM-NPs@LP vaccination markedly inhibited tumor growth in this orthotopic setting (Supplementary Fig. 9). These findings indicate that the antitumor potency of HM-NPs@LP is not confined to subcutaneous environments but is effectively preserved within an anatomically relevant cervicovaginal environment. While further mechanistic validation remains to be explored, this experiment provides evidence supporting the translational potential of HM-NPs@LP for HPV-related cervical diseases in a clinically relevant tumor niche. In addition to the direct cytotoxic effects of TP chemotherapy observed in our study, prior evidence suggests that taxanes can also modulate the tumor immune microenvironment. Recent studies indicate that taxanes may directly trigger T cells to exert noncanonical cytotoxic activity through vesicle-mediated mechanisms, independent of classical T-cell receptor signaling, while selectively targeting tumor cells.51 Collectively, these effects could create a more permissive environment for vaccine-induced T-cell responses, potentially contributing to the enhanced antitumor efficacy observed with the TP plus nanovaccine combination. While these mechanisms were not directly assessed here, they represent a plausible explanation for the synergy and will be further explored in future work.

Cancer immunotherapy focuses on harnessing and amplifying endogenous antitumor immunity through the targeted activation of immune effector mechanisms, potentially overcoming the nonspecific toxicity associated with conventional radiotherapy and chemotherapy.11 The limited clinical efficacy of existing immunotherapies in cervical cancer may be attributed to multiple factors, including high tumor heterogeneity in advanced stages, inadequate immunogenicity of conventional tumor vaccines, physical barriers imposed by dense fibrotic stroma limiting immune cell infiltration and drug penetration, and immunosuppressive TME impairing therapeutic effects, among others.12,52 To address these challenges, biomimetic strategies employing cell membrane-coated nanoparticles demonstrate particular promise through enhanced drug delivery and immunomodulatory capabilities.36,53,54 Notably, tumor cell membrane-encapsulated nanoparticles retain comprehensive tumor-associated antigen profiles, including naturally presented neoantigens, while the absence of genetic material enhances both safety profiles and feasibility of clinical translation.20 Nevertheless, preliminary clinical findings have shown that autologous tumor vaccines lack the capacity to fully prevent the recurrence of tumors.21 This stems from the fact that the inherently poor immunogenicity of tumor cell membranes poses substantial challenges for the immune system to achieve effective recognition and activation, leading to eukaryotic membrane-coated nanoparticles being recognized as self-components rather than foreign substances. However, the combination with potent vaccine adjuvants can significantly enhance vaccine immunogenicity.24 Hybrid nanovaccines integrating tumor membranes with bacterial OMVs have demonstrated enhanced dendritic cell activation and cytotoxic T lymphocyte responses, achieving tumor growth inhibition in murine models.25 Although genetically engineered bacteria can produce OMVs with lower endotoxins, further investigation is required to balance toxicity reduction and adjuvant efficacy maintenance.26

Cervical cancer has unique advantages for immunotherapy, as HPV infection and integration provide clear therapeutic targets, such as E5, E6, and E7.10 Peptide vaccines have attracted attention due to their high specificity, versatility, ease of production and transportation, and biosafety. However, their low immunogenicity when used alone necessitates the use of adjuvants or delivery systems.15,55 A study developed CoPoP liposomes combined with HPV E7 short peptides, which promoted the accumulation of CD8 + T cells within the TME and resulted in tumor eradication across both cervical cancer and lung metastasis mouse models.56 Short peptide vaccines (8-10 aa) exhibit weak immunogenicity and fail to elicit CD4 + T-cell-mediated assistance while potentially triggering T-cell tolerance.57 In contrast, synthetic long peptides (SLPs) necessitate internalization and processing by DCs before MHC binding, which facilitates sustained epitope presentation and immune activation.58 Our in vitro findings indicated that DC activation was largely attributable to bacterial membrane-derived PAMPs. The distinctive advantage of HM-NPs@LP lies in incorporating HPV16 epitopes, which endow the nanovaccine with the capacity to elicit antigen-specific T-cell responses, a finding corroborated by ex vivo ELISPOT assay results (Fig. 3m–o). Beyond our findings, it is important to compare EM with conventional adjuvants such as MPLA, Poly(I:C), and CpG, which function as agonists for individual pattern recognition receptors (PRRs). In contrast, EM carries multiple PAMPs that simultaneously engage diverse innate receptors, driving broader and more sustained dendritic cell activation. This multimodal signaling likely accounts for the superior immunostimulatory effects observed in our study. Prior studies have also shown enhanced tumor control with EM-based nanovaccines compared to MPLA-containing systems, underscoring its translational potential.30 Nevertheless, we acknowledge that direct benchmarking against canonical adjuvants remains an important next step, and future studies will systematically evaluate EM performance relative to established standards. The HM-NPs@LP developed in this study demonstrates an innovative platform for personalized immunotherapy, potentially addressing two persistent clinical challenges: (1) therapeutic limitations in managing persistent HPV infections and associated precancerous cervical lesions and (2) the need for treatment strategies against advanced and metastatic malignancies. Three functionally optimized components synergistically characterize this approach. First, by removing most bacterial LPS from EM, the nanovaccine exhibits better biosafety than conventional OMV-based vaccines. Second, the incorporation of patient-derived tumor membranes enables autologous antigen presentation while bypassing complex neoantigen identification processes. Notably, the PLGA-based nanoparticles allow customization: (a) antigenic epitope selection guided by individual HPV genotyping and HLA polymorphism analysis and (b) dynamic modification of modulators (e.g., TP or checkpoint inhibitors) based on longitudinal immune profiling (Fig. 6).59 This adaptable platform demonstrates significant potential for addressing critical clinical needs across disease progression stages. For HPV-associated cervical diseases, the platform reveals stage-specific adaptability, showing the capacity to eliminate lesions in early-stage patients, remodel the immunosuppressive microenvironment and suppress tumor growth in advanced cases. While further validation is needed, these findings indicate that the nanovaccine platform could advance precision medicine by synchronizing therapeutic components with individual pathophysiological characteristics.

A primary limitation of the current study involves the incomplete characterization of immunosuppressive constituents within the tumor cell membranes and bacterial cytoplasmic fractions. Such elements, specifically the protein components situated on the tumor cell surface, fulfill a dual role by presenting self-antigens while concomitantly participating in immune evasion. The targeted regulation of these inhibitory factors represents a potential avenue to amplify the therapeutic potency of the nanovaccine.60 However, the absence of a detailed analysis of these components and their mechanisms restricts our full understanding of the efficacy of nanovaccines and their potential for clinical application.36,61 Additionally, although EM preparations were predominantly enriched for inner-membrane proteins, outer-membrane and cytosolic contaminants were identified. Complete removal of such residues remains technically challenging, but biosafety assays confirmed low endotoxin levels and negligible toxicity. Future refinements such as density-gradient purification will be explored to further reduce contamination. Finally, the study does not explore the integration of presurgical therapies that could influence tumor cell membrane proteins, which might otherwise complement and enhance the effects of neoadjuvant therapies.

In summary, this study successfully established an innovative nanovaccine platform for the immunological management of HPV-related cervical diseases. By merging TM with EM to construct HM, which was subsequently integrated with immunogenic HPV16 long peptides, the HM-NPs@LP formulation was engineered. HM-NPs@LP offers a potent therapeutic modality for treating advanced cervical malignancies along with persistent HPV infections and cervical intraepithelial lesions, taking advantage of the clinical accessibility of cervical tissue samples. The biosafety profile, streamlined and economical production process, and suitability for personalized treatment regimens collectively provide novel therapeutic strategies for HPV-related cervical diseases while offering critical perspectives for the development of immunotherapies against other malignancies.

Materials and methods

Study design

This study aimed to explore the antitumor potential of an autologous cancer vaccine utilizing HPV-positive TC-1 murine tumor models. This study utilized membrane fusion technology to combine the mouse TC-1 tumor cell membrane (TM) with the Escherichia coli cytoplasmic membrane (EM) to prepare a hybrid membrane (HM) with enhanced immunogenicity. PLGA nanoparticles loaded with HPV16 cytotoxic T-lymphocyte epitopes served as the delivery vehicle for constructing a new type of cervical cancer nanovaccine, HM-NPs@LP. Composition optimization and in vitro characterization of HM-NPs@LP were performed, including assessments of particle size, morphology, ζ-potential, stability, and safety. Importantly, to ensure consistent and well-defined formulation states, all in vitro and in vivo experiments in this study were conducted using freshly prepared nanovaccine formulations. The in vitro experiments of the study focused on characterizing the cellular uptake of HM-NPs@LP by bone marrow-derived dendritic cells (BMDCs), the stimulation of BMDCs and T lymphocytes, and the assessment of tumor-killing capacity. Furthermore, lymph node accumulation of HM-NPs@LP and the induction of systemic immune activation were investigated in vivo. For nanovaccine administration, the subcutaneous route was employed because it enables efficient drainage to regional lymph nodes, which are the principal sites of dendritic cell uptake, antigen presentation, and T-cell priming, while also minimizing nonspecific systemic exposure.30 This methodology is consistent with the strategic design of therapeutic vaccines. Utilizing the subcutaneous TC-1 tumor model in C57BL/6 mice, the study assessed antitumor efficacy along with systemic antitumor immune responses and the specific profiles of immune infiltration within the tumor microenvironment. The study evaluated the long-term immune memory and biosafety of HM-NPs@LP. Additionally, a humanized TC-1-A2-E5 tumor model expressing HLA-A*02:01 (human leukocyte antigen) and HPV16 E5 was established. Following the selection of HLA-A*02:01 binding epitopes and the synthesis of chimeric long peptides, the capacity of the vaccine to activate T cells and its antitumor activity were examined within this humanized transgenic system. Finally, the therapeutic efficacy of HM-NPs@LP when administered in conjunction with anti-PD-1 and taxane-platinum (TP) chemotherapy was determined using the HLA-A*02:01 transgenic murine model.

Animals

C57BL/6 mice (female, 6-8 weeks of age, weighing 18-20 g) were procured from GemPharmatech and housed under specific-pathogen-free (SPF) conditions within the animal facilities of Tongji Hospital, Tongji Medical College of Huazhong University of Science and Technology. To establish the TC-1 tumor-bearing model, C57BL/6 mice received subcutaneous injections of tumor cells into the right flank. All protocols involving mice were formally approved by the Ethics Committee of Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology, and every procedure was carried out in strict accordance with the prevailing ethical regulations for animal experimentation. Transgenic C57BL/6 mice expressing HLA-A*02:01 were supplied by Jackson Laboratory and kept in SPF environments at the Institute of Zoology, Chinese Academy of Sciences. These animal studies received ethical clearance from the Committee on the Ethics of Animal Experiments at the Institute of Microbiology, Chinese Academy of Sciences (IMCAS). All procedures adhered to the institutional recommendations set forth in the Guide for the Care and Use of Laboratory Animals (Ethics Reference: IOZ-IACUC-2022-240).

Cell lines, murine splenocytes, peptides

The murine TC-1 cell line (ATCC CRL-2785TM) originated from C57BL/6 lung epithelial cells that were immortalized using full-length human HPV16 E6/E7 and an activated ras oncogene. For this study, TC-1 cells were further engineered via lentiviral transduction to stably express the HPV16 E5 gene. Simultaneously, the humanized TC-1-A2-E5 cell line was established by cotransducing TC-1 cells with the HPV16 E5 gene and the human leukocyte antigen HLA-A*02:01. These models enabled the evaluation of HPV16-derived long peptides under murine H-2Db and human HLA-A*02:01 restriction contexts. TC-1 and TC-1-A2-E5 cells were cultured in DMEM (Gibco, C11995500BT) containing 10% FBS (Gibco, 16000044), 100 IU/mL penicillin and 100 μg/mL streptomycin (Invitrogen, 15140122). Murine splenocytes were cultured in RPMI 1640 medium (Gibco, C11875500BT) supplemented with 10% FBS (Gibco, 16000044), 100 IU/mL penicillin and 100 μg/mL streptomycin (Invitrogen, 15140122), 1× mycoplasma elimination reagent (Yeasen, 40607ES03), and 50 IU/mL rIL-2 (PeproTech, 212-12). All cells were incubated at 37 °C in a 5% CO2 humidified incubator.

All peptides utilized in this study were commercially synthesized by GenScript (Nanjing). These peptides were dissolved in DMSO (Amresco, 0231-500 ML) and then diluted to a final concentration of 5 mg/mL using DMEM.

Isolation and in vitro differentiation of mouse bone marrow-derived dendritic cells (BMDCs)

Female C57BL/6 mice (6-8 weeks old) were utilized to generate BMDCs. Following euthanasia, the mice were disinfected in 75% ethanol for 5-10 minutes. The femurs and tibias were harvested by clearing the surrounding tissues. After trimming the bone ends near the joints, the marrow was extracted by flushing the cavities with RPMI 1640 medium containing 5% FBS, 1% penicillin‒streptomycin, and 1× mycoplasma elimination reagent. The resulting marrow suspension was collected, washed, and mechanically dispersed by pipetting. This suspension was passed through a 70 µm cell strainer and centrifuged at 300 × g for 5 minutes (4°C), followed by the elimination of erythrocytes using red blood cell lysis buffer. After performing cell counts and viability assessment, the bone marrow cells were resuspended at a concentration of 1 × 10⁶ cells/mL in 10 cm dishes using complete BMDC medium (RPMI 1640 supplemented with 10% FBS, 1% penicillin‒streptomycin, 20 ng/mL GM-CSF, 10 ng/mL IL-4, and β-mercaptoethanol). The cultures were incubated at 37°C with 5% CO₂, with a 50% medium exchange performed on days 2 and 4. On day 6, nonadherent and loosely attached cells were harvested by collecting the supernatant and gently rinsing the dishes with PBS. The identification of BMDCs was conducted through flow cytometric analysis of CD11c expression, with purity established by the percentage of CD11c⁺ cells.

Isolation of cytoplasmic membranes derived from E. coli

The cytoplasmic membranes of the E. coli strain DH5α were extracted based on a modified version of a previously reported protocol.62 Briefly, initial cultures were prepared by growing freeze-preserved DH5α cells in liquid LB medium at 37°C with overnight shaking at 200 rpm. Once the optical density (OD600) reached 1.2, the bacterial cells were harvested via centrifugation at 3000 × g for 20 min at 4°C and subjected to triple-washing with PBS. The resulting bacterial pellet was resuspended in 10 mL of Buffer A (1 M sucrose, 0.2 M Tris-HCl, pH 8.0). Following the addition of lysozyme to a final concentration of 2 mg/mL, the cell suspension was incubated for 1 hour at 37°C with shaking at 120 rpm. To induce cell lysis, 90 mL of sterile water containing DNase (10 μg/mL) was added, followed by gentle mixing 20 times. Centrifugation at 3000 × g for 20 min (4°C) was utilized to collect the spheroplasts, which were then resuspended in 10 mL of ice-cold Buffer B (20 mM Tris-HCl, pH 7.2, 50 mM NaCl, 5 mM EDTA) supplemented with 20% w/v sucrose. After clarifying the lysate through centrifugation at 10,000 × g for 30 min at 4°C, 5 mL aliquots of the supernatant were layered onto a discontinuous sucrose gradient. This gradient was prepared from bottom to top as follows: 10 mL at 50%, 5 mL at 46%, 10 mL at 42%, 10 mL at 36%, 5 mL at 32%, and 10 mL at 27%. Ultracentrifugation was then performed at 120,000 × g for 2 hours at 4°C using a 70Ti rotor in a Beckman Optima XPN-100 Ultracentrifuge (Brea, CA, USA). From this gradient, a 1.5 mL fraction containing the cytoplasmic membranes was isolated and diluted with ice-cold Buffer B to achieve a final sucrose concentration of 10% (w/v). The membranes were recovered by centrifugation at 120,000 × g for 1 hour at 4°C, resuspended in PBS, and preserved at -80°C. Finally, the protein content of the membrane preparation was measured using a BCA protein assay kit.

Isolation of cell membranes from resected tumor tissue

Tumor cell membranes were sourced from female C57BL/6 mice inoculated subcutaneously with TC-1 or TC-1-Luc cells, as well as from HLA-A*02:01 transgenic females receiving TC-1-A2-E5 cells. Following tumor growth, the resulting masses (300-500 mm3) were resected and fragmented. Tissue digestion was subsequently conducted at 37°C using 2 mL of Hank’s balanced salt solution (HBSS) containing collagenase IV (1.0 mg/mL), DNase (0.1 mg/mL), and hyaluronidase (0.1 mg/mL). The resulting cell suspension was collected by scraping and subsequent centrifugation at 1000 × g for 5 minutes. After homogenization of the pellet in 2 mL of isolation buffer (225 mM mannitol, 75 mM sucrose, 0.5% BSA, 0.5 mM EGTA, 30 mM Tris, and a phosphatase/protease inhibitor cocktail), cell disruption was achieved through ultrasonication (30 W) for 3 minutes in an ice bath. The homogenate underwent initial centrifugation at 3000 × g for 5 minutes (4°C) to remove debris. The collected supernatant was further processed at 10,000 × g for 10 minutes (4°C), followed by ultracentrifugation of the resulting supernatant at 100,000 × g for 2 hours (4°C). The isolated tumor tissue-derived membranes were finally resuspended in 5 mM Bis-Tris buffer (pH 6.0) and preserved at -80°C. BCA protein assay kits were utilized to determine the protein concentration of the membrane extracts. To verify batch-to-batch consistency, proteomic profiling via liquid chromatography–mass spectrometry (LC–MS) was conducted on six independent preparations of EM and TM. The protein expression profiles among these membrane samples were then established and compared.

Generation and physicochemical characterization of HM-NPs@LP

For the preparation of HM vesicles, the combination of EM and TM was subjected to gentle agitation at 37°C for 15 min using a dry bath incubator. Membrane fusion was subsequently facilitated by repeated physical extrusion of the mixture through a 400 nm cutoff extruder (Hamilton Company, Reno, Nevada, USA). The double-emulsion method was employed to fabricate poly(lactide-co-glycolic acid)-OH (PLGA) nanoparticles. PLGA was utilized as the core material due to its status as an FDA-approved, biodegradable polyester that undergoes hydrolysis into naturally metabolizable lactic and glycolic acids, thereby ensuring high biocompatibility. This double-emulsion approach allows for consistent control over particle size and surface characteristics, facilitating effective peptide loading and providing a stable foundation for subsequent membrane encapsulation. In brief, PLGA was dissolved in dichloromethane to reach a concentration of 10 mg/mL. A 1 mL aliquot of this solution was combined with 0.2 mL of sterile water containing the target peptides. Primary emulsification was achieved via ultrasonication at 25 W for 3 min on ice, followed by the introduction of 2 mL of 1% sodium cholate and secondary emulsification (30 W) for 5 min. This resulting emulsion was then added dropwise to 10 mL of 0.5% sodium cholate solution and stirred for 30 min at ambient temperature. Following a 15-min rotary evaporation, PLGA NPs were recovered through centrifugation at 10,000 x g for 15 min and purified by two washes in sterile water. To generate the final HM-NPs@LP, the PLGA NPs were coextruded with HM vesicles through a 200 nm cutoff extruder (Hamilton Company). This process was repeated while maintaining a polymer-to-membrane protein mass ratio of 5:1.

The coating efficiency was evaluated by centrifuging the suspension at 12,000 × g for 30 min, allowing the separation of unencapsulated peptides and unbound membrane fragments from the HM-NPs@LP pellet. The resulting pellet (HM-NPs@LP) was retained for downstream use, and the supernatant was collected for quantification. Membrane coating efficiency was estimated using total membrane protein as a proxy for membrane material. The input protein of the HM fraction was recorded prior to coating, and unbound protein remaining in the supernatant after the 12,000 × g centrifugation step was measured by BCA assay. The coating efficiency was calculated as:

$${Membrane}\,{EE}\,\left( \% \right)=\frac{{Protei}{n}_{{input}}-{Protei}{n}_{{supernatant}}}{{Protei}{n}_{{input}}}\times 100$$

Across independent replicates, the mean membrane coating efficiency was 90.70 ± 5.53%.

To determine peptide encapsulation efficiency (EE) and drug loading (DL), the concentration of residual peptide in the postfabrication supernatant was measured via UV–vis spectrophotometry at 275 nm. This quantification process relied on a standard curve established using the same HPV16 long peptide. Subsequently, the values for EE and DL were determined according to the following equations:

$${EE}\left( \% \right)=\frac{{W}_{{{input}}_{{\_}}\mathrm{peptide}}-{W}_{{{supernatant}}_{\_}\mathrm{peptide}}}{{W}_{{{input}}_{\_}\mathrm{peptide}}}\times 100$$

$${DL}\left( \% \right)=\frac{{W}_{{{encapsulated}}_{{\rm{\_}}}\mathrm{peptide}}}{{W}_{{{total}}_{\_}\mathrm{nanoparticles}}}\times 100$$

The mean EE and DL values were 70.73 ± 3.15% and 12.25 ± 2.04%, respectively, based on replicate measurements.

To further examine formulation stability, lyophilization was performed. Importantly, this procedure was designed to assess the physical and biological stability of HM-NPs@LP for potential future applications rather than for preparing the formulations used in the biological studies reported here. For morphological characterization of the nanoparticles, 10 μL of each sample was applied to carbon-coated copper grids followed by a 15-min incubation period. Excess liquid was subsequently removed via filter paper absorption. The samples on the grids were then subjected to negative staining with 1% (v/v) uranyl acetate for 8 min. After air drying, the grids were analyzed using a transmission electron microscope (HT7700, HITACHI, Tokyo, Japan). A Zetasizer Nano ZS dynamic light scattering instrument (Malvern, UK) was utilized to determine the hydrodynamic size distribution and ζ-potential.

In vitro innate immunity experiments

To assess the potential of various nanoparticles in triggering innate immune activation, BMDCs were seeded into 96-well tissue culture plates at a density of 4 × 104 cells per well. The total protein concentration for each formulation was initially measured via BCA assay, followed by dilution in complete RPMI-1640 medium to achieve a 0.5 mg/mL working concentration. Specifically, for HM-NPs@LP, TM and EM proteins were integrated at a 3:1 mass ratio (EM:TM) to reach the 0.5 mg/mL total protein level, with long peptides coencapsulated into the PLGA nanoparticles at a standardized loading of 10 µg per 100 µg total protein. The BMDCs were then treated with these formulations (final concentration: 0.5 mg/mL) and maintained at 37°C for 24 hours. After the incubation period, the culture supernatants were harvested to determine the secretion levels of IL-1β, IL-6, and TNF-α using specific enzyme-linked immunosorbent assay (ELISA) kits in accordance with the manufacturer’s protocols.

Confocal imaging and colocalization analysis

To evaluate the intracellular colocalization of the hybrid membrane constituents, TM and EM were fluorescently tagged with DiO (green) and DiD (red), respectively. BMDCs were subsequently treated with these labeled nanoparticles and incubated for a 24-hour period. Following this incubation, the cells were subjected to triple-washing with PBS and fixation in 4% paraformaldehyde for 15 min at room temperature. DAPI (blue) was utilized for nuclear counterstaining. Fluorescence imaging was conducted using a Zeiss LSM 900 confocal microscope, with all experimental groups being imaged under strictly identical acquisition parameters to maintain consistency. For the quantification of signal overlap, the ImageJ Coloc2 plugin was employed to perform colocalization analysis. The degree of spatial correlation between the TM and EM fluorescence signals was determined by calculating the Pearson’s correlation coefficient (PCC) for individual cells, with 15 cells analyzed per group.

Ex vivo cytotoxicity assay

To evaluate antigen-restricted cytotoxicity, a CCK-8-based colorimetric assay was performed. Splenocytes collected from immunized mice served as effector cells and were coincubated with target cells at an effector-to-target (E:T) ratio of 20:1. These cultures were maintained in RPMI-1640 medium containing 10% FBS for 24 h at 37°C (Supplementary Fig. 7). The target cell panel comprised HPV16-positive TC-1 cells, while U14 (murine cervical carcinoma) and MC38 (murine colon adenocarcinoma) were utilized as HPV-negative controls. Following the 24-h coculture period, 10 μL of CCK-8 reagent (Vazyme) was introduced into each well, followed by an additional 2-h incubation under light-shielded conditions. The absorbance was recorded at 450 nm using a microplate reader.

In vivo and ex vivo fluorescence imaging

Following random allocation into experimental groups, mice were administered subcutaneous (s.c.) injections into the dorsal flank to evaluate nanoparticle biodistribution and lymph node (LN) accumulation. Unless otherwise specified, each mouse was administered 100 µg of total nanoparticle-associated protein per dose in 0.9% saline on days 0, 2, and 6 (for HM-NPs@LP, each dose contained 10 µg of encapsulated long peptide). 1) Fluorescent labeling of nanoparticles: The lipophilic tracers DiR (1,1′-dioctadecyl-3,3,3′,3′-tetramethylindotricarbocyanine iodide; Invitrogen) or DiO (3,3′-dioctadecyloxacarbocyanine perchlorate; Invitrogen) were employed for nanoparticle labeling. For labeling, 1 mg of protein-equivalent nanoparticles from each formulation (TM-NPs, EM-NPs, HM-NPs, and HM-NPs@LP) was incubated with 50 µL of 5 µM DiR or DiO at 4°C overnight. Removal of excess dye was achieved through triple-washing with PBS using a 50 kDa ultrafiltration centrifugal unit (Millipore). The final nanoparticles were resuspended in PBS at equivalent protein concentrations.

2) In Vivo Fluorescence Imaging: For whole-body near-infrared imaging, DiR-loaded HM-NPs@LP or saline was administered via s.c. injection into the murine back. An in vivo imaging system (IVIS Spectrum, PerkinElmer) was utilized to monitor fluorescence signals at defined time intervals. For quantification, Living Image software was used to determine the mean fluorescence intensity (MFI) within the inguinal LN region of interest.

3) Ex Vivo Fluorescence Imaging: For comparative biodistribution analysis, IR-780–labeled EM-NPs, TM-NPs, or HM-NPs@LP were delivered to mice through s.c. injection. Following a 24-h postinjection period, the inguinal LNs were surgically harvested and subjected to ex vivo imaging using IVIS under identical acquisition parameters. Fluorescence intensity was quantified as total radiant efficiency normalized to background.

4) Intravital Microscopy: To visualize nanoparticle accumulation in situ, DiO-loaded HM-NPs@LP were injected subcutaneously into the dorsal flank. Once 48 h had elapsed following the injection, intravital confocal microscopy of the left inguinal LN was performed using intravital confocal and two-photon microscopy (IVIM Technology Inc.). Z-stack images were acquired and reconstructed into three-dimensional (3D) projections using IVM Studio. Representative still images and movies (Supplementary Movies 1, 2) were generated to illustrate nanoparticle localization within LN-resident cells.

Tetramer preparation

The production of HLA-A*02:01-restricted tetramers specific for HPV16 E5, E6, and E7 peptides was conducted according to established protocols.63 Briefly, the extracellular segment of HLA-A*02:01 (GenBank: AZL48402.1, 1-276) was modified through the insertion of a BirA biotinylation substrate sequence at the α3 domain C-terminus. The modified HLA-A*02:01 and β2-microglobulin were expressed in E. coli (BL21), followed by in vitro refolding in the presence of the corresponding peptides. Once renatured, the peptide/HLA-A02:01 complexes were purified and subjected to biotinylation at 4°C for 12 h using D-biotin, ATP, and the BirA biotin protein ligase (GeneCopoeia, BI001). Further purification was performed using a Superdex 200 10/300 GL gel filtration column (GE Healthcare). Tetramer assembly was subsequently achieved by conjugation with PE-labeled streptavidin (BD, 554061). The resulting tetramers were maintained at 4°C in storage buffer (PBS) consisting of 10 mM Tris-HCl (pH 8.0) (Novon, ZZ02531), 150 mM NaCl (SCR, 10019318), 0.5 mM EDTA (Amresco, 0105-500 G), 0.2% BSA (Sigma‒Aldrich, 9048-46-8), and 0.09% NaN3 (Sigma‒Aldrich, 26628-22-8).

Flow cytometry

The following conjugated antibodies were used in this study: anti-mCD3 (BD, 562600), anti-mCD4 (BD, 563151), anti-mCD8 (BD, 564422), anti-mCD45 (BD, 557659), anti-mIFN-γ (BD, 557649), anti-mCD80 (BD, 560016), anti-mCD86 (BD, 560582), anti-mCD11c (BD, 563057), anti-mCD25 (BD, 562606), anti-mCD44 (BD, 563970), anti-mCD62L (BD, 553151), and anti-mFoxp3 (BD, 563101). For surface marker analysis, T cells (1 × 106 per sample) were incubated with 0.05 μg/μL of tetramer for 30 min at room temperature in FACS buffer (PBS supplemented with 2% FBS). Subsequently, additional fluorescently conjugated antibodies were added at a 1:100 dilution for a 20-min staining period. Data were acquired via flow cytometry, and the resulting files were processed and analyzed using FlowJo v10 software.

The intracellular cytokine staining (ICS) procedure followed established protocols. In brief, effector T cells and target cells were coincubated in 96-well plates for 2 h. Control groups included cells maintained in medium alone (negative control) or those stimulated with PMA/ionomycin (Dakewe Biotech Co., 2030421; positive control). To inhibit protein transport, cells were treated with GolgiStop (BD, 550583) for an additional 6 h at 37°C. After surface labeling, the cells were fixed and permeabilized using a dedicated buffer system (BD, 562574). Subsequently, intracellular staining was performed. Upon final washing and resuspension, the processed samples were analyzed by flow cytometry.

Western blot analysis

Protein samples were resolved via SDS–PAGE and subsequently electroblotted onto PVDF membranes (Millipore). After a 30-min blocking step at room temperature, the membranes were incubated with specific primary antibodies at 4°C overnight. For the validation of membrane protein isolation, antibodies against FtsZ (Agrisera, AS10715) and Na⁺/K⁺-ATPase (Abcam, ab283318) were employed. To assess dendritic cell activation, BMDCs were treated with various nanovaccine formulations (control, TM-NPs, EM-NPs, Mix NPs+LP, HM-NPs, and HM-NPs@LP) for 24 h. The resulting cell lysates were analyzed for the expression of TLR1 (Proteintech, 19816-1-AP), TLR2 (Proteintech, 66645-1-Ig), TLR4 (Proteintech, 66350-1-Ig), TLR6 (Proteintech, 22240-1-AP), MyD88 (Proteintech, 67969-1-Ig), NF-κB/p65 (Proteintech, 80979-1-RR), and phosphorylated NF-κB/p65 (82335-1-RR), all obtained from Proteintech. Following triple-washing with TBST (Tris-buffered saline with 0.1% Tween-20), the membranes were incubated for 1 h at room temperature with HRP-conjugated secondary antibodies (goat anti-rabbit or anti-mouse IgG, 1:10,000). After six additional washes with TBST, protein bands were visualized using a Monad chemiluminescence imaging system.

Lentiviral-mediated shRNA knockdown

To silence specific TLR genes, shRNA sequences targeting mouse TLR1, TLR2, TLR4, and TLR6 were synthesized and cloned and inserted into lentiviral vectors. A nontargeting shRNA (NC-shRNA) was used as a negative control. On day 6 of culture, BMDCs were harvested and infected with the respective lentiviruses at a multiplicity of infection (MOI) of 200. To enhance transduction efficiency, spinoculation was performed at 200 × g for 1 h at room temperature. After 6 h of incubation, the medium was replaced with fresh complete medium, and the cells were cultured for an additional 48 h to achieve stable knockdown. The modified BMDCs were then treated with HM-NPs@LP for 24 h, and downstream signaling activation was evaluated by Western blot analysis as described above on day 9. The target shRNA sequences are provided in Supplementary Table 2.

IFN-γ ELISPOT assay

HPV16-specific T-cell responses were evaluated in female C57BL/6 mice (6–8 weeks, n = 5 per group) following subcutaneous immunization on days 0, 2, and 6. The experimental groups received the various vaccine formulations previously detailed. On day 14 postinitial immunization, the mice were euthanized for spleen harvest. Splenocytes were subsequently isolated and maintained in complete RPMI-1640 medium (containing 10% FBS) for further analysis. Antigen-specific activation was quantified using a precoated IFN-γ-specific ELISpot kit (ELISPOT, Mabtech, 3321-4APT-2) following the manufacturer’s instructions. In brief, 2.5 × 105 splenocytes were seeded into each well and subjected to an 18 h stimulation period at 37°C. Stimulation was conducted using HPV16 long peptides and TM. Cell Activation Cocktail (BioLegend, 423301) and culture medium served as positive and negative controls, respectively. After incubation, plates were developed following the kit instructions, and IFN-γ–secreting T cells were visualized as spots using an automated ELISpot reader (Cellular Technology Limited).

Tumor regression experiments

To evaluate the therapeutic efficacy of the nanovaccines, two different tumor models were employed. For the subcutaneous model, female mice were subcutaneously challenged with tumor cells, followed by the administration of the vaccination regimens as previously described. To track tumor progression, bioluminescent signals from TC-1-luciferase cells (n = 5 per group) were monitored using an IVIS system (PerkinElmer). Additionally, tumor dimensions were recorded every two days with digital calipers, and volumes (V) were determined using the following formula:

$$V=\frac{L\times {W}^{2}}{2}$$

where L and W represent the longest and shortest diameters of the tumor, respectively. For the orthotopic cervical model, anesthetized mice were intravaginally inoculated with 5 × 104 tumor cells suspended in a 1:1 mixture of culture medium and Matrigel. Immunization was initiated on study day 11, when the tumors were established. The vaccination was administered via subcutaneous injection following a three-dose regimen on study days 11, 13, and 17. For both models, an IVIS (PerkinElmer) was used to monitor the bioluminescence signal. In orthotopic studies, the bioluminescence index serves as the primary metric for tumor burden, as internal tumors are inaccessible for caliper measurement. Mice were euthanized when the tumor volume reached approximately 2000 mm³ or when established animal license termination criteria were met for orthotopic models.

To ensure appropriate experimental design, two stages of control selection were implemented. In the C57BL/6 tumor model (Figs. 2–4), a broad panel of vaccine formulations (TM-NPs, EM-NPs, Mix NPs+LP, HM-NPs, and HM-NPs@LP) was included to comprehensively evaluate immunogenicity and therapeutic efficacy. As TM-NPs and EM-NPs consistently exhibited minimal activity, only formulations with superior performance (Mix NPs+LP and HM-NPs@LP) were advanced into the orthotopic cervical model (Supplementary Fig. 9) and the HLA-A*02:01 transgenic mouse experiments (Figs. 5, 6). This streamlined design minimized unnecessary animal use in accordance with the 3Rs principles (replacement, reduction and refinement) while maintaining valid comparisons across experiments.

Statistical analysis

Data are presented as the mean ± standard deviation (SD). For comparisons between two independent groups, Student’s t test was employed, while multiple group comparisons were performed using either one-way or two-way analysis of variance (ANOVA) followed by Bonferroni post hoc correction. Statistical significance was defined as follows: *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; ns denotes no significant difference. All statistical analyses were performed using GraphPad Prism software version 9.5.