B16/F10-ovalbumin (OVA) cell was a gift from A/Prof. Yuli Lin of Fudan University. Hepa1-6 and HEK293T cells were obtained from Shanghai Cell Bank of Chinese Academy of Sciences (Shanghai, China). These cell lines were maintained with DMEM (Biological Industries, Israel) supplemented with 10% fetal bovine serum (FBS) (Biological Industries, Israel), and 1% penicillin streptomycin (Biological Industries, Israel) at a humidified incubator (Thermo Fisher Scientific, USA) with 5% CO₂ at 37°C. Bone marrow-derived DCs (BMDCs) were collected from the femurs of 8-week-old C57BL/6 mice according to an established protocol,20 and cultured in BMDC complete medium composed of RPMI 1640 medium (Biological Industries, Israel) supplemented with 10% FBS, 1% penicillin streptomycin, 20 ng/mL mGM-CSF (#415-ML-020, R & D Systems, USA) and 10 ng/mL mIL-4 (#214–14, Peprotech, USA). 6-week-old male C57BL/6 mice and nod-obese diabetes server combined immune deficiency (NOD SCID) mice were purchased from Hangzhou Ziyuan Experimental Animal Technology (Hangzhou, China) and tkept in a specific pathogen free (SPF) facility. We used the ARRIVE checklist when writing our report.21

The specific procedure of SVMAV preparation is described in detail in online supplemental material.

The procedure of characterization and cytotoxicity assessment of SVMAV are described in detail in online supplemental material.

OVA peptide conjugated with fluorescein isothiocyanate (FITC) at the side chain of lysine residue was used for the tracking of OVA peptide in vivo. C57BL/6 mice were vaccinated subcutaneously with phosphate buffer saline (PBS), the combination of FITC-OVA, R848 and stattic in their free forms, or FITC-OVA@SVMAV at the dose of 40 nmol/mouse for each compound at the tail base. The inguinal LNs were harvested at 6 hours, 24 hours, and 48 hours postvaccination, and the fluorescence signals of LNs were detected using a in vivo fluorescence imaging system (Shimadzu, Japan). For the analysis of antigen uptake by DCs, the inguinal LNs were resected and processed through gentle mechanical disruption 6 hours after vaccination. The LNs homogenates were then passed through nylon mesh filters to obtain single cell suspensions, and stained with anti-CD11c-PE for the analysis by flow cytometer (DxFLEX, Beckman Coulter, USA).

The procedure of BMDC preparation is described in detail in online supplemental material.

The RNA sequencing analysis is described in detail in online supplemental material.

FITC-labeled OVA and FITC-OVA@SVMAV were used for antigen uptake experiment. BMDCs were plated in 24-well plates at 2× 10⁵ per well with BMDC complete medium, and then treated with different combinations of drugs for 4 hours at the dose of 40 nmol/mL for each compound and maintained without any treatments for 20 hours. BMDCs were then collected and stained with anti-CD11c-PE (#117307, BioLegend, USA). The percentage of FITC-positive CD11c⁺ DCs was analyzed by a flow cytometer (DxFLEX, Beckman Coulter, USA).

Unlabeled OVA and OVA@SVMAV were used for antigen presentation experiment. BMDCs (2×10⁵ cells/well) were incubated with different combinations of drugs for 4 hours at the dose of 40 nmol/mL for each compound and maintained without any treatments for 20 hours. Cells were collected and stained with anti-CD11c-FITC (#117305, BioLegend, USA), anti-CD80-PE (#104707, BioLegend, USA), and anti-SIINFEKL/H-2Kb-APC (#141606, BioLegend, USA) for 30 min at room temperature and subject for flow cytometry analysis (DxFLEX, Beckman Coulter, USA).

To assess the cytolytic capacity of CD8⁺ T induced by stimulated DCs, B16/F10-OVA cells were used as target cells. B16/F10-OVA cells were seeded to 24-well plates (2×10⁵ per well) and pretreated overnight with 5 µg/mL of mitomycin C to prevent cell proliferation. BMDCs were incubated with different combinations of drugs for 4 hours at the dose of 40 nmol/mL for each compound and maintained without any treatments for 20 hours. Subsequently, spleens from mice vaccinated with OVA (40 nM/mice) within a week were harvested and processed to obtain splenocytes with single cell suspensions. Then, the splenocytes and BMDCs were mixed at the cell number ratio of 10: 1 (BMDCs=2 × 10⁵) and added to B16/F10-OVA cells for 48 hours. Eventually, the supernatant was collected and centrifuged to remove cell debris for the measurement of lactate dehydrogenase (LDH) release by CyQUANT LDH cytotoxicity assay kit (# C20300, Thermo Fisher Scientific, USA). The cell death can be represented by the percentage of LDH release that was calculated as follows: (sample LDH – spontaneous LDH)/(total LDH- spontaneous LDH)×100%. After removing all the non-adherent cells, the viability of B16/F10-OVA remained adhered to the culture surfaces were assessed by CCK8 assay. Additionally, remained adherent cells were fixed with 4% paraformaldehyde and stained with 0.1% crystal violet at room temperature for 10 min and dried overnight for photographic imaging.

Murine tumor model establishment and antitumor activity evaluation of SVMAV by both prevention and therapeutic settings are described in detail in online supplemental material.

The procedure of flow cytometry analysis for tumor tissues is described in detail in online supplemental material.

The whole-exome sequencing and data analysis are described in detail in online supplemental material.

Murine orthotopic hepatocellular carcinoma (HCC) model establishment and antitumor activity evaluation of SVMAV by therapeutic settings is described in detail in online supplemental material.

The procedure of multiplex fluorescent immunohistochemistry (IHC) for immune cell markers in HCC is described in detail in online supplemental material.

Data were shown as means±SD Statistical significance was analyzed by Student’s t-test. Statistical differences in survival were assessed by log-rank test. *P<0.05, **p<0.01, ***p<0.001, and ****p<0.0001 were taken as statistically significant. Statistical analyses were performed using GraphPad Prism V.8.0.

Previous studies demonstrated that TLR activation enhanced the antigen-presentation functions of DCs, and the inhibition of STAT3 signaling promoted DC maturation. However, the combination effects of TLR activation and STAT3 inhibition have not been studied in the context of cancer immune therapies. Therefore, we treated mouse BMDCs with R848, stattic and their combination, and analyzed their transcriptome profiles by RNA sequencing to evaluate the effects of these molecules on DC functions. As expected, the combination of R848 and stattic increased the expression of immune-activating proteins, such as CSF3 and VCAM1 in BMDCs, and decreased the expression of genes involved in the suppression of DC maturation, such as STAT3 and JAK1 (figure 1A). These results suggested that these two compounds would improve the therapeutic effects of cancer vaccines via the modulation of DC functions.

With the aim of improving cancer vaccines by such additive effects of R848 and stattic, nanovaccines were constructed to simultaneously encapsulate antigen peptides, R848 and stattic. OVA _(257-264) (SIINFEKL) peptide was used as the antigen peptide to target B16/F10-OVA cells exogenously expressing OVA in this study. Briefly, the synthetic process of OVA@SVMAV can be divided into three steps (figure 1B, online supplemental figure 1A,B). First, the hydrophilic OVA-Cys·Ser·Ser·Val·Val·Arg (CSSVVR) peptide containing the cathepsin S response element VVR was covalently coupled to hydrophobic DHA by N- (2-aminoethyl) maleimide to yielding a controllable degradable nanopolymer monomer OVA-CSSVVR-DHA that is controllably degraded in response to cathepsin S. Then, the R848-SS-DHA containing a glutathione response element was obtained by covalently coupling hydrophilic R848 to DHA through a disulfide-bond via a reactive −OH group on R848. Finally, amphipathic OVA-CSSVVR-DHA, R848-SS-DHA and hydrophobic stattic were mixed at a molar ratio of 1: 1: 1 and spontaneously assembled into nanoparticles under the action of hydrophobicity in water via nanoprecipitation.

Dynamic light scattering analysis of OVA@SVMAV exhibited a narrow size distribution with a median particle diameter of ~100 nm (figure 1C). Transmission electron microscopy imaging showed that OVA@SVMAV had a uniform and spherical-like structure (figure 1D). Zeta potential analysis indicated that OVA@SVMAV had a positive zeta potential value of 40.83±1.52 mV (online supplemental figure S1E). We also evaluated the stability of the various nanoformulations in PBS containing 10% FBS at 37°C for 72 hours. The results demonstrated that the mixture of OVA-CSSVVR-DHA and R848-SS-DHA exhibited a higher stability than OVA-CSSVVR-DHA alone in a simulated body fluid environment. (figure 1E)

The drug release properties of OVA@SVMAV were analyzed by high performance liquid chromatography. R848-SS-DHA was incubated with dithiothreitol (DTT) to simulate the reductant intracellular environment, and the substrate peak areas of released R848 were measured after different durations. The results showed that R848 was steadily and effectively released via the disruption of the disulfide bond in R848-SS-DHA in the presence of DTT (figure 1F). Similarly, OVA-CSSVVR-DHA was incubated with cathepsin S, and the change in product peak areas at different time points demonstrated that cathepsin S cleaved OVA-CSSVVR-DHA to release the OVA peptide (figure 1G). Collectively, these results suggested that OVA@SVMAV was stable in a simulated body fluid environment, and its structural components could be disrupted to release functional drugs intracellularly. The cytotoxicity of OVA@SVMAV was evaluated in HEK293T cells using a CCK8 assay, and the results showed that SVMAV had a much higher IC50 than the combination of drugs in their free forms, which suggests that such nanoformulation improved their safety profile for normal epithelial cells (online supplemental figure S1C,D).

LNs are the major peripheral lymphatic organs within which APCs activate cytotoxic T cells after the subcutaneous administration of vaccines. Therefore, the accumulation of OVA@SVMAV in LNs was investigated in vivo using an OVA peptide labeled with FITC. After the administration of FITC-OVA +R848+ stattic in their free forms (hereafter referred to as free drugs) or FITC- OVA@SVMAV, draining inguinal LNs were harvested for fluorescence imaging at 6, 24 and 48 hours postinjection. The LNs exhibited a marginal and transient increase in FITC signal in mice treated with free drugs. In contrast, the fluorescence intensity of LNs showed a 5.8-fold increase at 6 hours postinjection in mice injected with FITC-OVA@SVMAV compared with mice treated with free drugs, and the signal was sustained until 48 hours (figure 2A–C).

In addition to vaccine accumulation in LNs, the efficiency of antigen uptake by DCs is a prerequisite for T cell activation by cross-presentation. Therefore, the uptake of FITC-OVA@SVMAV in LN-residing DCs was analyzed in mice treated with free drugs or FITC-OVA@SVMAV. Flow cytometry analysis of FITC-positive cells in CD11c⁺ DCs showed that OVA@SVMAV was taken up by 6.51% of DCs, but only 0.24% of DCs took up free drugs (figure 2D,E).

Collectively, these results showed that SVMAV promoted vaccine accumulation and antigen capture efficiency of DCs in LNs in vivo.

Our previous experiments showed that the R848 and stattic exhibited comprehensive influences on the intracellular signaling of DCs. Therefore, we speculated that SVMAV, which simultaneously contained these active compounds, would improve the functions of DCs. Therefore, functional analyses were performed to investigate changes in DC functions, including antigen uptake, antigen presentation and the activation of cytotoxic T cells in vitro.

BMDCs were treated with various formulations of vaccine compounds for 4 hours and maintained without any treatments for 20 hours before flow cytometry analysis to simulate the transient antigen exposure pattern in vivo. Similar to the previous results of the in vivo DC uptake analysis, in vitro OVA uptake analysis showed that BMDCs exhibited a 1.5-fold higher uptake efficiency for SVMAV compared with free drugs (figure 3A, online supplemental figure S2). Further analysis of the costimulatory molecules CD80 (a membrane marker for DC maturation) and H2kb-SIINFEKL (the antigen presentation complex) showed that the percentage of CD11c⁺CD80⁺SIINFEKL⁺ cells was significantly higher in BMDCs treated with OVA@SVMAV compared with BMDCs treated with free drugs (figure 3B). The results of these assays showed that OVA@SVMAV significantly improved antigen uptake and presentation by DCs. However, we did not observe significant improvement in these functions of BMDCs in these assays, which was likely due to the unphysiological abundance of antigens during in vitro incubation.

Next, we investigated whether OVA@SVMAV-treated BMDCs elicited enhanced activation of the OVA-specific CTL response in vitro by adding splenocytes from mice pulsed with the OVA peptide and BMDCs treated with various formulations of vaccine components to the B16/F10-OVA monolayer (figure 3C). Analysis of the LDH activity of these culture supernatants showed that the combination of R848 and stattic significantly increased the release of intracellular LDH to supernatants due to cell lysis by CTLs, regardless of the use of free drugs or nanovaccine (figure 3D). This observation was further confirmed in CCK8 assays and crystal violet staining assays of the remaining melanoma cells that adhered to the culture surface, which showed the lowest cell numbers in B16/F10-OVA cells coincubated with splenocytes and BMDCs treated with the combination of antigen, R848 and stattic in the forms of free drugs and nanoparticles (figure 3E,F). These results suggested that SVMAV containing R848 and stattic improved DC functions and more actively primed antigen-specific T cells for tumor-specific killing.

To further evaluate the antitumor efficacy of OVA@SVMAV in vivo, a subcutaneous melanoma model of B16/F10-OVA was established in immunocompetent C57/B6 mice (figure 4A). The mice were treated with various formulations of vaccine components three times at intervals of 1 week. Treatment with the OVA peptide in its free form did not obviously reduce tumor growth, but the combination of OVA, R848 and stattic significantly suppressed tumor growth, which was further enhanced by OVA@SVMAV (figure 4B,C). The stability of mouse body weight also indicated a low systemic toxicity of OVA@SVMAV (online supplemental figure S3D). Interestingly, when we repeated the same OVA@SVMAV treatment scheme in NOD SCID mice with B16/F10-OVA melanoma, OVA@SVMAV failed to exert obvious antitumor activity (online supplemental figute S3A–C), which suggests the critical role of specific immune responses during this antitumor process.

CTLs (CD8⁺ T cells) are essential for efficient vaccine-induced antitumor immunity. Therefore, we analyzed the status of CTLs in both peripheral blood and TME. Consistent with the differences in the antitumor properties of different formulations, analyses of the peripheral blood cells demonstrated that mice treated with OVA@SVMAV had the highest percentage of CD69⁺ cells and OVA-specific cells among peripheral CD8⁺ T cells, which indicates robust CTL activation and clonal expansion targeting the tumor antigen with the induction of SVMAV in vivo (figure 4D,E and online supplemental figure S4A,B). We also analyzed the status of CD4⁺ T cells in peripheral blood and observed strongest activation in OVA@SVMAV group, similar to our observation on peripheral CD8⁺ T cells (online supplemental figure S4C,D). IHC analysis of melanoma tissues on CTL-related markers demonstrated that mice treated with OVA@SVMAV had the highest CTL infiltration level and protein levels of interferon-γ and granzyme B, which are two important proteins secreted by CTLs to eliminate tumor cells in the TME (figure 4F and online supplemental figure S3E–G).

Overall, these results demonstrated that OVA@SVMAV exerted potent antitumor effects by inducing an OVA-specific CTL response.

Cancer metastasis is the major cause of cancer-related mortality. Therefore, the effect of SVMAV on the prevention of cancer metastasis was evaluated with a melanoma lung metastasis model in vivo (figure 5A). C57BL/6 mice were subcutaneously treated with different formulations, including PBS, the combination of OVA, R848 and stattic as free drugs, and OVA@SVMAV once weekly for a total of three times before intravenous injection of B16/F10-OVA cells via the tail vein. Black metastatic lesions were observed on the surface of the lungs after 20 days (figure 5A). To our delight, the number of visible lung metastatic lesions was significantly decreased in mice vaccinated with OVA@SVMAV compared with the mice vaccinated with PBS or the combination of free drugs (figure 5B,C). HE and IHC for Ki-67 also showed significantly fewer micrometastatic lesions within lung tissues resected from mice vaccinated with OVA@SVMAV (figure 5D).

Since tumor vaccines function by eliciting tumor-specific T-cell responses, we reasonably speculated that the therapeutic efficacy of SVMAV would be further improved in combination with ICB therapies. Therefore, the therapeutic effects of OVA@SVMAV combined with anti-programmed cell death protein 1 antibody (aPD-1) were investigated in mice bearing melanoma exogenously expressing OVA following the treatment scheme shown in figure 6A. The tumor growth curves for each tumor and the mean tumor volumes of each group showed that the monotherapies of OVA@SVMAV and aPD-1 exhibited a moderate antitumor effect, and the combination of OVA@SVMAV and aPD-1 achieved the most remarkable antitumor effect (figure 6B,C). Although the tumor volume rapidly increased after the withdrawal of treatments, the median survival durations were longer for mice treated with combination therapy (32 days) compared with untreated mice (23 days), OVA@SVMAV-treated mice (28 days) and aPD-1-treated mice (27 days) (figure 6D). One out of eight mice showed complete regression after treatment with the combination therapy of OVA@SVMAV and aPD-1. The stability of mouse body weight also indicated a low systemic toxicity of this combination therapy (online supplemental figure S6G). These results demonstrated that the combination of SVMAV and ICB therapy exerted a synergistic effect in a melanoma model.

To analyze whether the synergistic effect of combination therapy was attributed to the improvement of CTL infiltration in tumor, we measured the levels of intratumoral CTL infiltration during the treatment. Notably, the combination treatment of OVA@SVMAV and aPD-1 induced a significant increase in tumor-infiltrating CTLs within the TME and effectively augmented antitumor efficacy, but monotherapies with aPD-1 alone failed to elicit such a sufficient CTL infiltration (figure 6E,F). Moreover, the function of CD8⁺ T cells was also remarkedly improved by the combination therapy of OVA@SVMAV and aPD-1, as indicated by the flow cytometry analysis of IFNγ⁺ CD8⁺ T cells in tumor (online supplemental figure S5). To further evaluated the immune status in TME, the frequencies of tumor-infltrating CD4⁺ T cells, CD4⁺ Foxp3⁺ regulatory T cells (Tregs) and CD45⁺ F4/80⁺ macrophages were measured (online supplemental figure S6A, B, C, D). CD4⁺ T cells and Tregs exhibited similar levels of infiltration in all groups, but PD-1 blockade treatment decreased the tumor CD45⁺ F4/80⁺ macrophage infiltration (online supplemental figure S6C,F).

Overall, these results suggested that SVMAV as an effective adjunct to improve the therapeutic effect of ICB therapies, such as aPD-1, by synergistically potentiating the antitumor immunity.

Taken a step further, we wanted to test whether SVMAV could be used as a universal platform for neoantigen-targeted personalized cancer vaccines in a clinical setting (figure 7A). First, we analyzed the somatic mutations of Hepa1-6, and identified potential neoantigens according to the following criteria: (1) a high binding affinity with MHC I, (2) good hydrophilicity, and (3) an exposed position for the mutated amino acid in the predicted antigen-MHC complex. After screening, we chose three eligible mutant peptides: Htt _(2375-2383) (ISLARLPLV → ISLPRLPLV), Lifr _(180-188) (VALVLLNTM → VALVSLNTM), and Smarcal1 _(171-179) (ISDSFYVLG → ISDSFYALG) (figure 7B). We chemically synthesized and modified these neoantigens following a previously described method and obtained HTT-CSSVVR-DHA, LIFR-CSSVVR-DHA, and SMARCAL1-CSSVVR-DHA. These modified neoantigens were mixed with R848-SS-DHA and stattic and assembled into the personalized SVMAV loaded with antigens form mutated Htt, Lifr, and Smarcal1, named HLS@SVMAV (figure 7C). To evaluate the efficacy of the designed HLS@SVMAV, an orthotopic HCC mouse model was established by inoculating Hepa1-6 cells into the left liver lobe. Mice carrying HCC cells were treated with HLS@SVMAV or aPD-1 following the treatment scheme in figure 7D. Photographic images of resected livers showed that this HCC model was highly resistant to aPD-1, which is consistent with the clinical observation that monotherapy with aPD-1 failed to improve the overall survival of HCC patients.22–26 To our delight, mice treated with HLS@SVMAV exhibited a significantly smaller tumor volume compared with untreated mice or aPD-1-treated mice (figure 7E,F), which suggests that neoantigen-targeted personalized SVMAV might be a promising method for HCC resistance to aPD-1 therapy.

Multiplex fluorescent IHC for CD8a, F4/80, CD86 and PD-L1 was used to evaluate the landscape of the HCC immune microenvironment to explore the mechanisms underlying the different response patterns for aPD-1 and HLS@SVMAV treatments (figure 7G). Multiplex fluorescent IHC analysis of harvested tumor tissues should be considered as images of the final event resulting from therapy-related immune response, other than immune responses during therapy. The results showed that Hepa1-6 orthotopic HCC tissues contained a high abundance of CTLs and M1 macrophages. However, aPD-1 treatment reduced the number of intratumoral F4/80⁺CD86⁺ M1 macrophages, but HLS@SVMAV did not exert a significant influence on M1 macrophages (online supplemental figure S7A,B), which is consistent with the observations in the melanoma model (online supplemental figure S6C,F). Impairment of M1 macrophages might explain the failure of aPD-1 therapy in the HCC model, and such shortcomings were overcome by HLS@SVMAV.

Taken together, the successful treatment of an orthotopic HCC model by neoantigen-targeted personalized SVMAV provides a proof of principle for the application of SVMAV as a universal platform for neoantigen-targeted personalized nanovaccines as an alternative strategy of cancer immune therapy.