Engineering a dual-antigen mRNA vaccine to restore immune control in chronic hepatitis B

Abstract

Therapeutic vaccination for chronic hepatitis B (CHB) remains challenging, as persistent immune tolerance to hepatitis B surface antigen (HBsAg) impedes anti-HBsAg seroconversion. Here we show that screening antigen combinations incorporating HBsAg identifies preS1-HBsAg as the optimal pairing. The resulting dual-antigen mRNA vaccine elicits robust hepatitis B virus (HBV)-specific immune responses in chronic HBV mouse models, leading to near-complete viral genome clearance, marked reduction of multiple HBV antigens, and serological conversion. Mechanistic analyses reveal that preS1 serves as the primary driver of HBV-specific T cell responses, while HBsAg contributes to both anti-HBsAg seroconversion and intrinsic adjuvant activity. Specifically, when delivered in mRNA form, HBsAg promotes antigen-presenting cell (APC) activation, enhances antigen presentation, and amplifies preS1-specific CD8⁺ T cell responses. Furthermore, combining the preS1-HBsAg vaccine with interferon-α (IFN-α) enhances antiviral efficacy and immune memory while maintaining a favorable safety profile. These findings establish preS1-HBsAg mRNA vaccination as a promising and translatable therapeutic strategy for functional cure of CHB.

Here, the authors develop a dual-antigen (preS1-HBsAg) mRNA vaccine that elicits specific immune responses in a mouse model of chronic Hepatitis B virus (HBV) infection, that when combined with IFN-α has enhanced antiviral efficacy and immune memory.

Introduction

Chronic hepatitis B virus (HBV) infection remains a major global health burden, affecting over 250 million people worldwide and significantly increasing the risk of liver cirrhosis and hepatocellular carcinoma^1–3. While prophylactic vaccines based on hepatitis B surface antigen (HBsAg) provide effective protection against new infections, they do not offer therapeutic benefit for individuals with chronic hepatitis B (CHB)⁴. This therapeutic gap stems from the failure to restore virus-specific immune responses in CHB, where persistently high levels of circulating HBsAg induce immune tolerance and T cell dysfunction, key immunological hallmarks preventing functional cure^5–8. Notably, HBsAg-based vaccination has repeatedly failed to elicit robust immune responses in CHB patients due to this immune tolerance⁹. These challenges underscore the need for optimized antigen combinations that retain the essential role of HBsAg in inducing anti-HBsAg seroconversion while overcoming its associated immune tolerance, thereby restoring comprehensive antiviral immunity^10–12.

HBV-infected cells express multiple viral antigens, including HBsAg, hepatitis B core antigen (HBcAg), and hepatitis B e antigen (HBeAg), along with regulatory proteins, such as hepatitis B X antigen (HBxAg) and polymerase¹³. HBsAg exists in three isoforms: large (L-HBsAg), middle (M-HBsAg), and small (S-HBsAg): L-HBsAg contains the preS1, preS2, and S domains; M-HBsAg contains the preS2 and S domains; and S-HBsAg contains only the S domain^14,15. In CHB, persistently high circulating HBsAg levels are associated with immune dysfunction, including impaired HBV-specific T cell function and reduced neutralizing anti-HBsAg responses¹⁶. By contrast, preS1 appears less involved in tolerance due to its much lower expression, making it a promising target for HBV-specific immune responses¹⁷. HBcAg and HBxAg mainly localize intracellularly and elicit strong T cell responses in acute infection but show reduced immunogenicity in CHB owing to T cell exhaustion^18,19. HBeAg, a secreted variant of HBcAg, promotes tolerance by inducing regulatory T cells (Tregs) and inhibiting dendritic cell (DC) maturation, while polymerase aids immune evasion by blocking innate signaling via DDX3 interaction^20–22. These previous findings on the distinct immunological properties of HBV antigens collectively provide a rational framework for designing multi-antigen therapeutic vaccines against chronic HBV infection²³.

In this study, we conducted systematic antigen screening using an mRNA vaccine platform with HBsAg as the baseline component to identify combinations capable of overcoming immune tolerance and inducing comprehensive antiviral immunity. We selected HBsAg as the core antigen due to its indispensable role in inducing anti-HBsAg antibodies required for seroconversion. We combined HBsAg with preS1, HBcAg, or HBxAg, given their potential to enhance antiviral efficacy. The optimized preS1-HBsAg mRNA vaccine induced near-complete viral genome clearance, marked reduction of multiple HBV antigens, and serological conversion in 80% of CHB model mice. Mechanistically, preS1 served as the primary functional antigen driving potent HBV-specific T cell responses, while HBsAg provided essential B cell epitopes for anti-HBsAg seroconversion and exerted a potent adjuvant-like effect that enhanced immunogenicity. Specifically, when delivered as lipid nanoparticle (LNP)-formulated mRNA, HBsAg promoted antigen-presenting cell (APC) activation, enhanced antigen presentation, particularly in macrophages, and amplified preS1-specific CD8⁺ T cell responses (Supplementary Fig. 1). Furthermore, combining the preS1-HBsAg vaccine with interferon-α (IFN-α) improved antiviral efficacy and immune memory while maintaining a favorable safety profile. These findings establish preS1-HBsAg vaccination as a promising and translatable therapeutic strategy for achieving functional cure in CHB.

Results

PreS1-HBsAg mRNA vaccine elicits enhanced antiviral effects in chronic HBV models

First, we designed four mRNA constructs: (1) mHBsAg, encoding S-HBsAg only; (2) mPreS1-HBsAg, encoding the preS1 domain (excluding preS2) fused to S-HBsAg; (3) mHBxAg-HBsAg, encoding HBxAg fused to S-HBsAg; and (4) mHBcAg-HBsAg, encoding HBcAg fused to S-HBsAg (Fig. 1a). Unless otherwise specified, “HBsAg” in this study refers to S-HBsAg. All mRNAs demonstrated efficient HBsAg expression following in vitro transfection (Supplementary Fig. 2). For in vivo delivery, these mRNAs were encapsulated into LNPs, generating four vaccine formulations. For clarity, each vaccine formulation is referred to throughout the text and figures by the antigen it encodes (e.g., “preS1-HBsAg” denotes the mRNA-LNP formulation encoding preS1 fused to HBsAg).

Fig. 1. PreS1-HBsAg mRNA vaccine elicits potent antiviral efficacy in the chronic HBV infection model.

a Schematic of vaccine mRNA constructs. b Experimental timeline for therapeutic vaccination in AAV-HBV1.3 carrier mice. c HBV DNA in serum (left) and liver (right) quantified by qPCR 13 weeks after virus injection. d HBV total (left) and 3.5 kb RNA (right) in the liver quantified by qPCR 13 weeks after virus injection. Relative HBV RNA levels were calculated using the 2^−ΔΔCt method, with Gapdh used as the internal reference and the mean ΔCt of the eLNP group used as the calibrator. n = 5 biologically independent mice per group for data in (c, d). Representative HBcAg immunofluorescence (e) and quantification (f) of HBcAg⁺ hepatocytes per 0.3 × 0.3 mm² liver tissue. n = 8 fields of view from 3 biologically independent mice per group. In the box plots, the center line indicates the median, the box limits indicate the 25th and 75th percentiles, and the whiskers indicate the minimum and maximum values; dots represent individual fields of view. g Immune tolerance challenge protocol boost with rHBsAg and incomplete Freund’s adjuvant (IFA). HBsAg (h) and anti-HBsAg (i) levels in serum pre- (7 weeks after virus injection) and post- (10 weeks after virus injection) rHBsAg injection. Ratios in (i) show seroconverted mice/total mice per group. n = 5 biologically independent mice per group, except for the post-rHBsAg-injection eLNP group, for which n = 3. eLNP, empty LNP. All data are shown as the mean ± s.e.m. Statistical analysis was performed by a two-tailed unpaired t-test. n.s. is P ≥ 0.05, and significant P values are shown. The schematics in (b, g) were created in BioRender.com with attribution line Zhang, B.W. (2026) https://BioRender.com/ypsq4zb. Source data are provided as a Source data file.

A chronic HBV infection mouse model was established by intravenous injection of AAV-HBV1.3, characterized by sustained expression of HBsAg and HBeAg, the presence of HBV DNA and RNA in both serum and liver, and positive immunofluorescence for HBcAg in liver tissue sections (Supplementary Fig. 3)^24,25. Before antigen-combination screening, we compared weekly versus every-2-week HBsAg vaccination while keeping the total dose number constant (Supplementary Fig. 4a, c). The two regimens elicited comparable T cell phenotypes/function (including effector-memory (TEM, CD44⁺ CD62L⁻) and granzyme B⁺ readouts), whereas the weekly regimen yielded higher serum anti-HBsAg levels (Supplementary Figs. 4b, d–g, and 5). Therefore, weekly dosing was selected for subsequent studies.

For antigen-combination screening, HBV-carrier mice were assigned to five groups with comparable baseline serum HBsAg and HBeAg levels at 4 weeks after virus injection and received equimolar mRNA vaccines weekly for three doses (Fig. 1b and Supplementary Fig. 6a). At 13 weeks after virus injection, serum and intrahepatic HBV DNA levels were reduced in the preS1-HBsAg group, whereas HBxAg-HBsAg and HBcAg-HBsAg showed weaker effects (Fig. 1c). HBV total RNA and 3.5 kb RNA levels in the liver were reduced in the dual-antigen vaccine groups compared with empty LNP (eLNP) controls (Fig. 1d). Consistently, immunofluorescence staining showed fewer HBcAg⁺ hepatocytes after dual-antigen vaccination, particularly in the preS1-HBsAg group (Fig. 1e, f). Serum HBsAg analysis further showed lower HBsAg levels in the preS1-HBsAg group, whereas HBxAg-HBsAg and HBcAg-HBsAg produced more limited HBsAg decline (Supplementary Fig. 6b). Consistent with these virologic outcomes, preS1-HBsAg significantly increased CD8⁺ TEM frequency versus the eLNP group, whereas HBxAg-HBsAg and HBcAg-HBsAg did not (Supplementary Fig. 7). Collectively, these data suggest that the incorporation of preS1 antigen significantly enhances the therapeutic efficacy of the HBsAg-based mRNA vaccine.

In chronic HBV infection, high circulating levels of HBsAg contribute to immune tolerance, which impedes effective therapeutic responses⁶. To directly assess whether our preS1-HBsAg vaccine could overcome this tolerance, we boosted vaccinated mice with recombinant HBsAg (rHBsAg) to evaluate the recall response of HBsAg-specific immunity (Fig. 1g). While mice receiving HBsAg vaccine showed only limited responsiveness to rHBsAg boost, mirroring the immune paralysis seen in chronic infection, preS1-HBsAg-vaccinated animals achieved anti-HBsAg seroconversion in 80% of cases (4/5 mice, Fig. 1h, i). These results indicate that preS1 incorporation reshapes the host response to HBsAg and helps overcome the tolerogenic barrier that limits therapeutic vaccine efficacy.

PreS1-HBsAg vaccination elicits HBV-specific immunity in peripheral lymphoid tissues

We next characterized the peripheral immune responses induced by preS1-HBsAg vaccination in HBV-carrier mice (Supplementary Fig. 8a). Comprehensive immunophenotyping of lymphoid organs revealed that immunization with preS1-HBsAg, but not HBsAg alone, significantly expanded antigen-experienced T cell populations, increasing both CD8⁺ TEM and TCM subsets (Fig. 2a, b). Compared with the eLNP group, the preS1-HBsAg vaccine also induced stronger T cell activation, as evidenced by increased expression of CD69 and PD1 on splenic CD8⁺ T cells (Fig. 2c, d, Supplementary Figs. 8b, c, and 9), while simultaneously reducing immunosuppressive Treg frequencies (Fig. 2e).

Fig. 2. PreS1-HBsAg mRNA vaccine-mediated reconstitution of HBV-specific immunity.

Representative dot plot image (a) and frequencies (b) of TEM (CD44⁺ CD62L⁻) and TCM (CD44⁺ CD62L⁺) subsets among CD8⁺ T cells in lymph node and spleen. Exact P values are shown for comparisons versus eLNP. Representative histograms and frequencies of CD69 (c) and PD1 (d) expression on CD8⁺ T cells. e Splenic Treg (Foxp3⁺ CD4⁺) frequencies. Representative flow cytometry plots (f) and quantification (g) of granzyme B expression by splenic T cells after ex vivo stimulation with a mixed peptide set comprising two HBsAg peptides (HBsAg_190–197, VWLSVIWM; and HBsAg_208–215, ILSPFLPL) and an overlapping preS1 peptide pool spanning the full preS1 domain (15-mers with 10 aa overlap). eLNP, empty LNP. Data are shown as the mean ± s.e.m. n = 5 biologically independent mice per group for data in (b–e, g). Statistical analysis was performed by a two-tailed unpaired t-test. n.s. is P ≥ 0.05, and significant P values are shown. Source data are provided as a Source data file.

To evaluate antigen-specific immunity, we first quantified HBsAg tetramer⁺ CD8⁺ T cells in the spleen. However, frequencies were low and not significantly different across groups, limiting detailed phenotypic analysis (Supplementary Fig. 8d). We therefore performed ex vivo peptide-restimulation assays. Under mixed stimulation (full preS1 overlapping peptide pool plus two HBsAg CD8⁺ epitopes, HBsAg_190–197 (VWLSVIWM) and HBsAg_208–215 (ILSPFLPL)), splenocytes from preS1-HBsAg-vaccinated mice displayed a cytotoxic-skewed response, including increased granzyme B in both CD8⁺ and CD4⁺ T cells, modestly increased interferon-γ (IFN-γ) in CD8⁺ T cells, and reduced transforming growth factor-β (TGF-β) in CD4⁺ T cells; these responses were stronger than those in the HBsAg-only group (Fig. 2f, g, Supplementary Figs. 10 and 11a). Separate stimulation with either HBsAg peptides or the preS1 peptide pool confirmed enhanced granzyme B responses in CD8⁺ T cells after preS1-HBsAg vaccination (Supplementary Fig. 11b, c). Consistent with these findings, preS1-HBsAg vaccination increased cytotoxic killing activity against HBV-infected hepatocytes (Supplementary Fig. 12) and raised the frequency of germinal-center B cells in HBV-carrier mice, a hallmark of humoral immunity (Supplementary Fig. 8e). Taken together, these results demonstrate that the inclusion of preS1 in the HBsAg-based vaccine significantly enhances both cellular and humoral immune responses, essential features for overcoming chronic HBV infection.

PreS1-HBsAg mRNA vaccine reprograms the liver immune microenvironment

The immunosuppressive liver microenvironment represents a fundamental obstacle to achieving functional cure in chronic HBV infection²⁶. To determine whether our preS1-HBsAg vaccine could overcome this barrier, we performed a comprehensive analysis of intrahepatic immune responses following vaccination. Flow cytometric evaluation revealed that preS1-HBsAg treatment increased total CD45⁺ leukocyte infiltration in the liver, with particular expansion of CD8⁺ T cells (Fig. 3a, b, and Supplementary Fig. 13).

Fig. 3. PreS1-HBsAg mRNA vaccine reprograms the hepatic immune microenvironment in chronic HBV infection.

a Frequency of CD45⁺ immune cells in the liver by flow cytometry 7 weeks after virus injection (i.e., 1 week after the third vaccination). b t-SNE visualization and quantification of major CD45⁺ immune cell populations in the liver. P values shown in the stacked bar plot indicate comparisons of CD8⁺ T cell proportions between each vaccinated group and the eLNP group. c Heatmap of specific gene expression. d GO and KEGG analysis of immune-related pathways. e Fragments per kilobase of transcript per million mapped reads (FPKM) of genes from the antigen processing and presentation pathway. f Frequency of HBsAg tetramer⁺ CD8⁺ T cells in total liver cells. Granzyme A and granzyme B MFI in intrahepatic CD8⁺ T cells (g) and CD4⁺ T cells (h) after ex vivo stimulation with a mixed peptide set comprising two HBsAg peptides (HBsAg_190–197, VWLSVIWM; and HBsAg_208–215, ILSPFLPL) and an overlapping preS1 peptide pool spanning the full preS1 domain (15-mers with 10 aa overlap). eLNP, empty LNP. Data are shown as the mean ± s.e.m. n = 5 biologically independent mice per group for data in (a, b, f–h); and n = 3 biologically independent mice per group for RNAseq data. Statistical analysis was performed by a two-tailed unpaired t-test. n.s. is P ≥ 0.05, and significant P values are shown. Source data are provided as a Source data file.

Next, we performed transcriptome sequencing on sorted CD45⁺ cells from the liver to explore functional changes in hepatic immune cells. The preS1-HBsAg group exhibited distinct transcriptional profiles compared to eLNP controls, while the HBsAg group showed minimal differences from eLNP-treated mice (Supplementary Fig. 14a–c). Transcriptomic analysis showed that genes upregulated by preS1-HBsAg vaccination were enriched for immune response, metabolism, cell cycle regulation, cell adhesion, and immune exhaustion (Fig. 3c). This enrichment indicated enhanced immune cell proliferation, activity, cytotoxicity, chemotaxis, and residency within the liver. Importantly, genes associated with effector function and cytotoxicity, such as the chemokine Ccl5 and granzyme family members (Gzma and Gzmb), showed an upward trend in expression (Fig. 3c). This direction was consistent with the earlier flow cytometry results (Fig. 2g), and subsequent qPCR analysis showed a similar trend but did not reach statistical significance (Supplementary Fig. 14d). Concurrently, immune checkpoint genes (Havcr2, Nr4a2, Tigit, Ctla4, and Rgs16) were also upregulated, likely reflecting a physiological feedback mechanism to modulate immune activation (Fig. 3c)²⁷. Gene Ontology (GO) enrichment analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses showed that the upregulated genes in the preS1-HBsAg-immunized mice were enriched in several immune response-related pathways, including innate/adaptive immune response, signaling receptor activity, chemotaxis, positive regulation of IFN-γ production, and T cell receptor signaling pathway, indicating robust reprogramming of liver immune function following vaccination (Fig. 3d and Supplementary Fig. 15)²⁸. Moreover, abundant genes in the antigen processing and presentation pathway, such as Tap1 and Tap2, were significantly upregulated (Fig. 3e), suggesting enhanced APC maturation and T cell priming capacity.

We next assessed antigen-specific T cell responses at the site of infection by analyzing intrahepatic lymphocytes. HBsAg tetramer staining detected HBsAg-specific CD8⁺ T cells in both vaccine groups, and the preS1-HBsAg group showed a numerically higher frequency than the HBsAg-only group, although the difference did not reach statistical significance (Fig. 3f). In addition, functional assays showed enhanced intrahepatic effector activity in the preS1-HBsAg group: after ex vivo peptide restimulation, intrahepatic T cells exhibited higher granzyme A and granzyme B expression than those from HBsAg-only controls (Fig. 3g, h). Vaccination did not significantly increase intrahepatic IFN-γ⁺ T cells (Supplementary Fig. 16). These data suggest that preS1-HBsAg vaccination enhances intrahepatic HBV-specific antiviral immunity.

Collectively, these data indicate that the preS1-HBsAg mRNA vaccine effectively reconditions the liver immune microenvironment by boosting immune infiltration, enhancing effector activity, and improving antigen presentation, likely contributing to the observed therapeutic efficacy.

Both preS1 and HBsAg are essential for the functional cure of CHB

Having established that the preS1-HBsAg mRNA vaccine elicited robust HBV-specific immune responses and achieved effective therapeutic outcomes in CHB models, we next sought to delineate the respective contributions of preS1 and HBsAg within this dual-antigen system. To this end, we directly compared mRNA vaccines encoding preS1-HBsAg, preS1 alone, or HBsAg alone. Consistent with our initial screening results, vaccination with HBsAg alone exhibited minimal therapeutic effect, whereas preS1 alone reduced multiple virological parameters, including serum HBV DNA and RNA levels, intrahepatic HBV DNA and RNA loads, serum HBsAg levels, and hepatic HBcAg expression, to levels approaching those achieved by preS1-HBsAg vaccination (Fig. 4a–c and Supplementary Fig. 17).

Fig. 4. PreS1 and HBsAg synergistically act to achieve HBsAg seroconversion.

a Relative HBV DNA levels in serum and liver 10 weeks after virus injection were quantified by qPCR. b Serum HBsAg content at 10 weeks after virus injection was quantified using ELISA. c The number of HBcAg⁺ cells per 0.3 × 0.3 mm² liver tissue. n = 8 fields of view from 3 biologically independent mice per group. In the box plots, the center line indicates the median, the box limits indicate the 25th and 75th percentiles, and the whiskers indicate the minimum and maximum values; dots represent individual fields of view. d Quantification of granzyme A expression by CD8⁺ T cells and granzyme B expression by NK cells in spleen after ex vivo stimulation with a mixed peptide set comprising two HBsAg peptides (HBsAg_190–197, VWLSVIWM; and HBsAg_208–215, ILSPFLPL) and an overlapping preS1 peptide pool spanning the full preS1 domain. e Quantification of granzyme B ELISpot results. The color scale indicates the number of spots per well. Splenocytes (5 × 10⁵ cells per well) from mice immunized three times with the preS1-HBsAg mRNA vaccine were stimulated for 24 h with individual overlapping 15-mer peptides spanning preS1, the full preS1 peptide pool, or two HBsAg epitopes. PMA (phorbol 12-myristate 13-acetate) served as a positive control. f HBV antigen content in serum (left) and liver (right) 9 weeks after virus injection was quantified using ELISA. g Anti-HBsAg antibody content in serum 10 weeks after virus injection was quantified using ELISA. eLNP, empty LNP. All data are shown as the mean ± s.e.m. n = 5 biologically independent mice per group for data in (a, b, d, g); and n = 4 biologically independent mice per group for data in (f). Data in (e) are shown as the mean of 3 biologically independent mice per group. Statistical analysis was performed by a two-tailed unpaired t-test. n.s. is P ≥ 0.05, and significant P values are shown. Source data are provided as a Source data file.

At the immunological level, preS1-containing vaccination induced cytotoxic effector responses upon ex vivo stimulation with a mixed peptide stimulus consisting of the full preS1 peptide pool and two HBsAg peptides. PreS1 alone increased the frequency of granzyme A-producing CD8⁺ T cells, whereas the dual-antigen preS1-HBsAg vaccine further increased this response, with significantly higher frequencies than those observed in the preS1 or HBsAg groups (Fig. 4d and Supplementary Fig. 18a). Both preS1 alone and preS1-HBsAg vaccination also increased granzyme B-producing NK cells compared with eLNP controls (Fig. 4d and Supplementary Fig. 18a). Under the same stimulation conditions, IFN-γ production by splenic CD8⁺ T cells did not significantly differ among groups, while IFN-γ-producing CD4⁺ T cells were increased in the preS1-HBsAg group (Supplementary Fig. 18b). As shown in Supplementary Fig. 11c, restimulation with the preS1 peptide pool induced robust granzyme B production in CD8⁺ and CD4⁺ T cells from preS1-HBsAg-vaccinated mice, supporting the induction of preS1-specific T cell immunity. We next mapped preS1-specific T cell epitopes by stimulating splenocytes ex vivo with individual preS1 peptides and quantifying responses using granzyme B ELISpot. This analysis revealed broad reactivity across multiple preS1 peptides, with a clear immunodominant hotspot: peptides #4-#6 (covering the sequence HQLDPAFRANTANPDWDFNPNKDTW) elicited the strongest responses, comparable to those induced by the full preS1 peptide pool. Under the same assay conditions, the three preS1 hotspot peptides elicited stronger responses than the two previously reported HBsAg-derived CD8⁺ T cell epitope peptides (HBsAg_190–197, VWLSVIWM; HBsAg_208–215, ILSPFLPL)²⁹ (Fig. 4e and Supplementary Fig. 19). These results establish preS1 antigen as the primary functional antigen responsible for driving antiviral T cell immunity in this vaccine platform.

To further understand why preS1 elicited such strong immune responses, we examined its antigen abundance in CHB mice. High circulating levels of viral antigens, such as HBsAg, are known to induce host immune tolerance in chronic infection settings¹⁷. Consistent with this, our analysis showed that preS1 antigen levels were over 100-fold lower than HBsAg levels in both serum and liver (Fig. 4f). This relatively low antigen abundance likely prevents the establishment of immune tolerance toward preS1, allowing it to remain immunologically accessible.

However, a key difference was observed in antibody responses. While the preS1-HBsAg vaccine induced robust anti-HBsAg seroconversion, anti-HBsAg antibodies were undetectable in the preS1-alone group (Fig. 4g). This finding suggests that although preS1 alone is sufficient to drive T cell-mediated viral control, incorporation of HBsAg is indispensable for eliciting antibody responses, thereby ensuring comprehensive immune control and long-term functional cure. Together, these findings demonstrate that the synergistic action of preS1 and HBsAg is required to elicit both antiviral T cell responses and protective antibody production, ultimately enabling functional cure in CHB.

mRNA-encoded HBsAg provides adjuvant-like support for preS1-specific immune responses

Given that preS1 alone showed antiviral effects and that the preS1-HBsAg vaccine showed greater efficacy, we hypothesized that HBsAg exerts adjuvant-like effects that enhance preS1-driven immune responses, especially considering the relatively weak intrinsic immunogenicity of preS1^17,24. To directly test this hypothesis while avoiding confounding from HBV-specific immunity, we employed an ovalbumin (OVA) model system to evaluate the adjuvanticity of HBsAg, using HBcAg, preS1, and HBxAg as comparative controls (Fig. 5a). Naive mice were immunized with mRNA vaccines encoding OVA alone or OVA fused with each HBV-derived protein. Notably, fusing HBsAg or HBxAg with OVA markedly enhanced T cell responses, as reflected by increased frequencies of OVA-specific T cells, along with greater proliferative capacity and elevated production of effector molecules, such as IFN-γ and granzyme B upon OVA peptide stimulation (Fig. 5b, c, and Supplementary Fig. 20a–c). Radar chart further confirmed that OVA coupled with HBsAg or HBxAg protein elicited a much stronger T cell immune response, with the OVA-HBsAg group exhibiting a preferentially higher CD8⁺ T cell response (Fig. 5d). Given the critical role of CD8⁺ T cell dysfunction in HBV persistence and immune tolerance, the preferential enhancement of CD8⁺ T cell immunity by HBsAg holds particular relevance for therapeutic vaccine strategies in CHB^30,31.

Fig. 5. HBsAg shows adjuvant-like effects in the mRNA vaccine context to enhance T cell immunity against fused antigens.

a Schematic of mRNA constructs expressing OVA or OVA fused to HBV antigens. b Flow cytometry analysis of OVA-specific CD8⁺ T cells in the spleen 5 days after the second vaccination. c Flow cytometry analysis of the proportion of IFN-γ⁺ among CD8⁺ T cells in the spleen after ex vivo stimulation with two OVA peptides (OVA_257–264, SIINFEKL; and OVA _323–339, ISQAVHAAHAEINEAGR). d Radar plot of T cell immune profiles and area under the curve (AUC)-based quantification of the CD8⁺ T cell response 5 days after the second vaccination. e Representative flow cytometry plots and quantification of APCs presenting the OVA peptide SIINFEKL by H-2K^b in the lymph node 24 h after a single vaccination. The right-hand tables show P values for comparisons of OVA or OVA-HBsAg versus eLNP within each APC population. M., macrophage; B, B cell; Mo., monocyte. f Representative flow cytometry plots and analysis of the proportion of APCs expressing CD86 in the lymph node 24 h after a single vaccination. g Serum IL6 content of each vaccine-treated group 6 h after the second vaccination. Flow cytometry analysis of the proportion of granzyme B⁺ among CD8⁺ T cells and CD4⁺ T cells in the spleen of healthy (h) or CHB (i) mice after ex vivo stimulation with an overlapping preS1 peptide pool (15-mers overlapping by 10 aa) spanning the full preS1 domain. eLNP, empty LNP. Data are shown as the mean ± s.e.m. n = 5 biologically independent mice per group for data in (b–d, g–i), and n = 4 biologically independent mice per group for data in (e, f). Statistical analysis was performed by a two-tailed unpaired t-test. n.s. is P ≥ 0.05, and significant P values are shown. Source data are provided as a Source data file.

To further elucidate the mechanism through which HBsAg enhances immune responses against co-delivered antigens, we analyzed APCs in draining lymph nodes 24 h post-vaccination. Incorporation of HBsAg enhanced OVA presentation and upregulated activation and maturation markers on APCs, highlighting HBsAg’s potential to function as an adjuvant by promoting innate immune activation and antigen presentation (Fig. 5e, f, Supplementary Figs. 20d, and 21). Notably, macrophages exhibited higher SIINFEKL-H-2K^b and maturation markers than DCs. To investigate this, we tracked vaccine uptake using dual-labeled LNP/mRNA (DiO-labeled LNP and Cy5-labeled mRNA) and found efficient uptake in both cell types, with higher uptake in macrophages in the OVA-HBsAg group (Supplementary Fig. 22a–d). In parallel, Tap1, Tap2, and B2m transcripts were increased in macrophages after OVA-HBsAg vaccination (Supplementary Fig. 22e), consistent with enhanced MHC-I processing/loading capacity.

To distinguish endogenous MHC-I presentation from cross-presentation of secreted antigen, we performed in vitro studies in RAW264.7 macrophages. Direct mRNA transfection (mOVA or mOVA-HBsAg) produced higher SIINFEKL-H-2K^b than incubation with matched conditioned medium; mOVA-HBsAg induced the strongest signal (Supplementary Fig. 22f). Although OVA protein was detectable in conditioned medium, this route induced only weak SIINFEKL-H-2K^b presentation; OVA-HBsAg was minimal in conditioned medium (Supplementary Fig. 22f, g). Together, these data support that lymph node macrophages predominantly present antigen via direct mRNA uptake, intracellular translation, and endogenous MHC-I loading, rather than dominant cross-presentation of secreted antigen.

Because circulating soluble HBsAg in CHB is not established as a systemic APC activator, we next tested whether the observed effect is platform-dependent. In bone marrow-derived cells (BMDCs) and peripheral blood dendritic cells (PBDCs) generated from healthy and CHB mice, mHBsAg-LNP robustly increased CD86 expression, whereas recombinant HBsAg showed weak or minimal effects (Supplementary Figs. 23 and 24). These data indicate that the potential adjuvant-like effect is context-dependent and is most evident when HBsAg is delivered and expressed via mRNA in APCs, rather than implying that circulating HBsAg in CHB acts as a systemic adjuvant.

We then verified this adjuvant-like effect within the HBV vaccine context. Compared with preS1 mRNA alone, incorporation of HBsAg (preS1-HBsAg) increased BMDC activation (CD86 upregulation) in both healthy and CHB mice and elevated serum interleukin-6 (IL-6) after vaccination (Fig. 5g and Supplementary Fig. 25a, b), supporting enhanced innate immune activation. Given the role of IL-6 in promoting T and B cell proliferation, activation, and differentiation, we next assessed whether HBsAg could further augment preS1-specific T cell responses^32,33. Consistently, splenocytes from healthy or CHB mice vaccinated with preS1-HBsAg exhibited higher frequencies of granzyme B⁺ CD8⁺ T cells upon preS1 peptide stimulation than those from mice vaccinated with preS1 alone (Fig. 5h, i), confirming that HBsAg contributes to stronger preS1-specific T cell immunity. In addition, adding HBsAg to HBxAg- or HBcAg-containing constructs similarly increased antigen-specific responses, suggesting a broader immunomodulatory effect of mRNA-encoded HBsAg across co-delivered HBV antigens (Supplementary Fig. 25c, d).

Together, these results demonstrate that mRNA-encoded HBsAg confers context-dependent, adjuvant-like effects within the preS1-HBsAg mRNA vaccine, enhancing innate immune activation and amplifying antigen-specific T cell responses against co-delivered antigens.

Combination with IFN-α amplifies antiviral efficacy and immune responses

Because IFN-α is clinically used for CHB yet has limited benefit as monotherapy, we next evaluated whether co-administration with the preS1-HBsAg mRNA vaccine could provide additional benefit (Fig. 6a)^16,34. Compared with eLNP controls, preS1-HBsAg vaccine-containing regimens reduced serum HBsAg levels and showed a trend toward lower serum HBV DNA levels at this endpoint (Fig. 6b, c). However, direct comparisons between the combination group and preS1-HBsAg monotherapy were not statistically significant for serum HBsAg or HBV DNA (Fig. 6b, c). Anti-HBsAg and anti-preS1 antibodies were induced in both vaccine-containing groups, with no significant difference between vaccine alone and combination therapy (Fig. 6d and Supplementary Fig. 26a). Immunofluorescence analysis showed significantly fewer HBcAg⁺ hepatocytes in combination-treated mice (Fig. 6e), indicating a reduction of infected cells.

Fig. 6. Enhanced antiviral effects of preS1-HBsAg mRNA vaccine and IFN-α in chronic HBV infection.

a Experimental timeline. b Serum HBsAg levels at 10 weeks after virus injection. c Serum HBV DNA levels at 10 weeks after virus injection were quantified by qPCR. d Serum anti-HBsAg levels at 10 weeks after virus injection. e Intrahepatic HBcAg detection by immunofluorescence (left) with quantitative analysis of the number of HBcAg⁺ cells per 0.3 × 0.3 mm² liver tissue (right). n = 8 fields of view from 3 biologically independent mice per group. In the box plots, the center line indicates the median, the box limits indicate the 25th and 75th percentiles, and the whiskers indicate the minimum and maximum values; dots represent individual fields of view. The proportion of splenic CD8⁺ T (f) and CD4⁺ T (g) cells with TEM and TCM phenotypes. Frequencies of CD8⁺ T and/or CD4⁺ T cells producing granzyme B (h) or IFN-γ (i) in the spleen after ex vivo stimulation with a mixed peptide set comprising two HBsAg peptides (HBsAg_190–197, VWLSVIWM; and HBsAg_208–215, ILSPFLPL) and an overlapping preS1 peptide pool spanning the full preS1 domain (15-mers with 10 aa overlap). j Immunofluorescence staining of granzyme A and granzyme B in liver tissue. eLNP, empty LNP. Data are shown as the mean ± s.e.m. n = 5 biologically independent mice per group for data in (b–d, f–i). Statistical analysis was performed by a two-tailed unpaired t-test. n.s. is P ≥ 0.05, and significant P values are shown. The schematic in (a) was created in BioRender.com with attribution line Zhang, B.W. (2026) https://BioRender.com/ypsq4zb. Source data are provided as a Source data file.

To characterize immune correlates of the combination regimen, we next performed comprehensive immune profiling. Relative to vaccine monotherapy, the combination regimen significantly increased CD4⁺ T cell and NK cell activation, while CD8⁺ T cell activation showed nonsignificant upward trends (Supplementary Fig. 26b). Combination therapy selectively altered T cell memory phenotypes, increasing CD8⁺ TCM and CD4⁺ TEM frequencies (Fig. 6f, g). Upon ex vivo mixed-peptide stimulation, the combination therapy showed a trend toward increased granzyme B production in CD8⁺ T cells and induced higher IFN-γ production in CD8⁺ and CD4⁺ T cells (Fig. 6h, i). Granzyme A-producing CD8⁺ T cells were also increased in the vaccine-containing groups compared with eLNP controls (Supplementary Fig. 26c). In liver tissue, immunofluorescence analysis revealed increased expression of granzyme A/B in the combination group (Fig. 6j). Together, these data indicate that IFN-α co-administration strengthens vaccine-elicited anti-HBV immunity and improves intrahepatic viral control.

PreS1-HBsAg vaccine shows a favorable safety profile as monotherapy and in combination therapy

Given the increased cytotoxic T cell activity observed during treatment (including higher granzyme⁺ T cell frequencies in spleen and liver), we asked whether this immune activation was associated with liver injury. Although alanine aminotransferase (ALT) showed a mild upward trend in the preS1-HBsAg group, values largely remained within the normal range (Supplementary Fig. 27a). Serum aspartate aminotransferase (AST) was unchanged across groups (Supplementary Fig. 27a). Correlation analysis showed no significant association between serum ALT and granzyme A/B responses (Supplementary Fig. 27b, c).

We then performed a comprehensive safety assessment of preS1-HBsAg monotherapy and its combination with IFN-α. Both regimens were well tolerated, with stable body weight throughout the study period (Fig. 7a). Longitudinal serum biochemistry showed no sustained elevation of ALT/AST, while other blood indices remained within normal ranges, without significant differences from baseline (Fig. 7b and Supplementary Fig. 28). Histopathological examination showed that while control mice receiving eLNP exhibited marked inflammatory cell infiltration in hematoxylin and eosin (H&E)-stained liver sections, tissues from both preS1-HBsAg vaccine and preS1-HBsAg vaccine + IFN-α treated animals showed preserved hepatic architecture without evidence of treatment-related pathological changes (Fig. 7c). These results collectively demonstrate that the preS1-HBsAg vaccine maintains a favorable safety profile both as monotherapy and when combined with IFN-α, with no detectable adverse effects on systemic physiology or liver histology at the therapeutic doses employed.

Fig. 7. Safety profile of preS1-HBsAg vaccine and IFN-α combination therapy.

a Longitudinal monitoring of body weight in AAV-HBV1.3 carrier mice receiving indicated treatments. n = 2 biologically independent mice in the NT group and n = 6 biologically independent mice in other groups. b Longitudinal monitoring of ALT/AST in the serum of AAV-HBV1.3 carrier mice receiving indicated treatments. n = 3 biologically independent mice in the NT group and n = 5 biologically independent mice in other groups. c H&E staining of liver tissue at 10 weeks after virus injection. NT, healthy mice. eLNP, empty LNP. Data are shown as the mean ± s.e.m. Source data are provided as a Source data file.

Discussion

In the present study, building on the established safety profile and epitope diversity of HBsAg, we developed a dual-antigen vaccine co-expressing preS1 and HBsAg, which showed therapeutic efficacy in a murine CHB model. The vaccine achieved a pronounced reduction in HBsAg, near-complete viral genome clearance, and an 80% seroconversion rate, meeting key functional cure criteria. Our findings highlight three major advances: first, the combination of preS1 and HBsAg breaks immune tolerance by leveraging preS1 as a dominant immunogen that drives robust HBV-specific CD8⁺ T cell responses, while HBsAg exerts a potent adjuvant-like effect to enhance immunogenicity. Second, this antigen combination uniquely induces coordinated humoral (anti-HBsAg) and cellular (cytotoxic T cell) immunity, addressing a longstanding challenge in CHB therapy. Third, the clinical potential of this approach is supported by additional benefits observed with IFN-α co-therapy and a favorable safety profile.

The success of therapeutic vaccines for chronic HBV infection hinges on the selection of optimal antigens capable of overcoming immune tolerance. While HBsAg-based vaccines have been widely used for prophylaxis, their therapeutic utility is limited due to T cell exhaustion from prolonged antigen exposure^8,35. Our systematic evaluation of antigen combinations revealed that preS1, when paired with HBsAg in an mRNA format, outperformed other candidates (HBcAg and HBxAg) by inducing robust immune responses in both peripheral organs and the liver. Notably, preS1 functioned not as a mere adjuvant but as an independent immunogen, circumventing the immune tolerance typically observed with HBsAg. This can be attributed to preS1’s low endogenous expression, which minimizes tolerance induction, though its weak antigenicity historically required potent adjuvants like IFA¹⁷. Our mRNA platform overcame this limitation by enhancing preS1’s immunogenicity while unexpectedly uncovering HBsAg’s intrinsic adjuvant-like properties, which broadened the vaccine’s therapeutic potential. It should be noted that the adjuvant-like effect of HBsAg is platform-dependent and is primarily mediated by intracellular expression in APCs following LNP-mRNA delivery, rather than in the form of rHBsAg.

Extensive prior work has established the advantages of incorporating the preS region (including preS1) into HBV vaccine designs. Early studies reported that the preS region is more immunogenic than the S domain^36,37. Subsequent work further showed that defined preS1 T cell determinants can provide helper activity that broadens antibody responses to additional envelope epitopes (including preS2 and S) and can help overcome non-responsiveness to the preS2/S regions in certain settings^38,39. In addition, multivalent display of preS1 B-cell epitopes on core-based carriers has been explored to elicit neutralizing anti-preS1 antibodies and to mitigate tolerance-related constraints^40,41. Collectively, these studies provide important context for antigen selection and support the rationale for leveraging preS1 to broaden immune recognition beyond the S region.

Prior clinical studies of preS-containing protein vaccines (e.g., BRII-179 with or without low-dose IFN-α in virally suppressed CHB patients) reported favorable safety and measurable immunogenicity but no notable reduction in circulating HBsAg⁴². Notably, earlier clinical experience with the related Pre-S1/Pre-S2/S vaccine Sci-B-Vac combined with lamivudine suggested short-term antiviral and anti-HBsAg signals, although HBsAg loss was rare and the benefit was not sustained after vaccination was discontinued⁴³. Together, these findings suggest that, in CHB, vaccine-elicited immunity may be limited by response magnitude/kinetics and baseline antigen burden, and that optimized dosing schedules, lower-HBsAg patient populations, and/or combinations with antigen-lowering or immune-modulatory therapies may be required. Our CHB mouse data showing clear cellular and humoral responses together with HBsAg reduction, support further translational evaluation of mRNA-based multi-antigen vaccination in such optimized contexts. However, translation to patients will likely require careful selection of clinical context and rational combination regimens to address CHB-associated immune dysfunction.

Multi-antigen therapeutic vaccination represents a rational strategy to broaden HBV-specific immunity and potentially improve antiviral efficacy in CHB by expanding epitope coverage beyond a single antigen^44–48. For example, a recent work using heterologous prime-boost vaccination illustrates this principle: Kosinska et al. reported the TherVacB regimen, which uses an adjuvanted protein prime with HBsAg/HBcoreAg virus-like particles followed by a boost with MVA-HBVac encoding five HBV proteins (S-HBsAg of genotype A, L-HBsAg of genotype C, HBcAg_1–183 of genotype C, HBcAg_1–149 of genotype D, and the consensus reverse transcriptase domain of the polymerase) designed to expand multi-specific T cell responses against multiple HBV proteins with broad genotype coverage, and achieved sustained antiviral effects in HBV carrier models⁴⁸. This emerging literature supports the potential value of extending preS1-HBsAg vaccination with additional HBV antigens; however, such multi-antigen optimization (format, dosing, and immunodominance management) will require dedicated investigation.

A key mechanistic insight from our study is that granzyme-expressing CD8⁺ and CD4⁺ T cells may play a central role in viral control, shifting attention from the conventional IFN-γ-centric framework. Granzyme-mediated cytotoxicity can directly eliminate infected hepatocytes, thereby disrupting viral persistence and indirectly stimulating type I IFN production⁴⁹. Consistent with this mechanism, we observed a mild post-immunization increase in serum ALT and direct ex vivo killing of infected hepatocytes by vaccine-primed immune cells. In our model, preS1-HBsAg vaccination induced only modest IFN-γ⁺ T cell responses, and antiviral efficacy was preserved in IFN-γ-deficient mice⁵⁰. Together, these findings support an IFN-γ-independent, granzyme-associated cytotoxic antiviral response. Future studies should explore whether this pattern represents a broader feature of mRNA vaccine immunobiology and test the causal role of granzymes using targeted loss-of-function models. In parallel, we observed significant activation of the innate immune system, particularly NK cells, which are often functionally impaired in chronic HBV infection. Both transcriptomic and flow cytometry analyses showed enhanced NK cell activity, supporting coordinated innate-adaptive remodeling of antiviral immunity²⁸.

The path to functional cure will likely require rational combination therapies⁵¹. While our preliminary data show enhanced antiviral efficacy between the mRNA vaccine and IFN-α, this pairing may not be the most optimal. The vaccine’s strong bias toward cellular immunity, evidenced by modest anti-HBsAg titers despite robust viral clearance, suggests that combining it with HBsAg-neutralizing antibodies or immune checkpoint inhibitors (to counter T cell exhaustion markers observed in our study) could be more effective⁵². Prior “sandwich” strategies (anti-HBsAg antibodies + antiviral drugs + vaccines) and siRNA-based HBsAg knockdown approaches provide a framework for such combinations^53,54.

However, the translational potential of these strategies depends on addressing key model limitations. Our current AAV-based system, while enabling stable HBV persistence, represents a simplified setting. First, the use of a single genotype (ayw) does not capture the global diversity of HBV strains, which differ in prevalence and pathogenicity⁵⁵. Future studies should therefore include genotype-diverse settings (e.g., B and C, which are prevalent in Asia) to assess breadth across viral variants. Moreover, AAV vectors do not fully recapitulate the recurrent hepatocyte infection dynamics seen in natural HBV disease: HBV infection is initiated through sodium taurocholate co-transporting polypeptide (NTCP)-dependent entry and sustained by reinfection cycles¹⁰, whereas AAV-mediated transduction bypasses this receptor-dependent viral life cycle. Consequently, this model primarily evaluates immune control of HBV-antigen-expressing hepatocytes by the vaccine, but does not directly resolve effects on entry blockade, de novo cccDNA reservoir dynamics, or durability under continuous reinfection pressure. To strengthen translational inference, transgenic models with humanized livers and immune systems can provide complementary insight into therapeutic vaccine efficacy by more closely modeling natural HBV infection, persistence, and reinfection dynamics, as well as host immune responses^56,57.

To accelerate clinical translation, further optimization of the mRNA platform is essential. Artificial intelligence-driven sequence design can improve mRNA stability and ribosomal engagement, while advanced LNPs may enhance delivery and immunogenicity^58,59. Additionally, the interplay between vaccine-induced immunity and host factors (e.g., genetic background, immune status) warrants investigation through collaborative studies with clinical cohorts. These refinements, coupled with expanded preclinical validation across models, will solidify the foundation for human trials. Despite remaining challenges, our study provides a compelling proof of concept for mRNA-based dual-antigen vaccines as a transformative therapeutic paradigm for chronic HBV infection.

Methods

Ethics statement

All animal experiments were performed in compliance with relevant ethical regulations and were approved by the Animal Care and Use Committee of the University of Science and Technology of China (approval ID: USTCACUC26080125092).

Cell lines and mice

293T cells and RAW264.7 cells were kindly provided by Jun Wang, South China University of Technology. All cells were cultured in DMEM medium (Gibco), supplemented with 10% fetal bovine serum (FBS) (Excell Bio).

Male C57BL/6JGpt mice (6–7 weeks) were purchased from GemPharmatech and used in the study, reflecting the male-predominant epidemiology of CHB. All mice used in this study were housed in a specific pathogen-free facility and received humane care. Animals were maintained at 21 ± 1 °C, 40–70% humidity, and a 12 h light/dark cycle from 8 a.m. to 8 p.m. Sex-based comparisons were not applicable because only male mice were included, and the conclusions are therefore limited to male mice.

Animal experiment

To establish the HBV-carrier mouse model, the AAV-HBV1.3 virus (Beijing GenePeutics Biotechnologies, Inc.) was used. This virus contains 1.3 copies of the HBV genome (genotype D, serotype ayw) and is packaged in AAV serotype 8 capsids. C57BL/6J mice were intravenously injected with the AAV-HBV1.3 virus at a dose of 2 × 10¹⁰ viral genome copies per mouse. After 4 weeks, mice with stable HBV carriage were selected for subsequent experiments. Blood samples were collected from the retro­orbital sinuses at the indicated time points to monitor the serum levels of HBsAg, HBeAg, HBV DNA, and HBV RNA.

For vaccine immunization, 6-week-old male C57BL/6J mice or HBV-carrier mice were intramuscularly injected in a total volume of 50 μl per dose. In HBV antigen experiments, the HBsAg vaccine was administered at 10 μg mRNA per dose, and all other HBV antigen vaccine formulations were dosed at molar equivalents to the 10 μg HBsAg mRNA dose. In OVA antigen experiments, the OVA vaccine was administered at 2.5 μg mRNA per dose, and all OVA-related vaccine formulations were dosed at molar equivalents to the 2.5 μg OVA mRNA dose. eLNP was used as the control, where indicated. The immunization schedule is shown in the corresponding experimental timelines.

For IFN-α treatment, HBV-carrier mice received intravenous IFN-α (SinoBiological) at 5 μg per injection, administered once weekly for a total of three injections. The treatment schedule is shown in the corresponding experimental timeline.

RNA constructs and in vitro transcription

Plasmid templates for in vitro transcription of antigen-encoding RNAs were based on the pUC57-HSP70SBD-Kan vector. We obtained the antigen sequences of preS1, HBsAg, HBxAg, HBcAg, and OVA from NCBI, optimized them using the peptide 2.0 website (https://www.peptide2.com/), and then inserted the optimized antigen sequences into the pUC57-HSP70SBD-Kan vector. mRNA sequences are listed in Supplementary Data 1. DNA templates for in vitro transcription of mRNA were synthesized by Sangon Biotech. All in vitro transcription of mRNA and capillary electrophoresis in this study was performed by RNAlfa Biotech.

LNP preparation

LNPs were formulated using a previously established method. In brief, an ethanolic lipid mixture containing SM-102, cholesterol, DSPC (2,2-distearoyl-sn-glycero-3-phosphocholine), and DMG-PEG₂₀₀₀ (1,2-dimyristoyl-sn-glycero-3-methoxypolyethylene glycol 2000) was rapidly mixed with an aqueous mRNA solution at predetermined molar ratios (50%:38.5%:10%:1.5%) in a microfluidic chip device, maintaining a 3:1 ratio. The resulting LNPs were then subjected to dialysis against H₂O using a 100,000 molecular weight cut-off ultrafiltration device at 4 °C. The average particle size and PDI of vaccines were determined using a Zetasizer Nano ZS90 (Malvern Instruments). The encapsulation efficiency and mRNA concentration were detected by Quant-iT RiboGreen RNA assay kit (Invitrogen).

Quantitative real-time PCR

For quantification of total HBV RNA and 3.5 kb HBV RNA, liver tissues were homogenized using a bead-based homogenizer, and tissue homogenates were stored at −80 °C until RNA extraction. Blood serum samples were likewise stored at −80 °C prior to analysis. Viral RNA was extracted from liver homogenates and serum samples using a Viral Genomic DNA/RNA Extraction Kit (Tiangen). RT-qPCR was performed using the HiScript II One Step RT-qPCR SYBR Green kit (Vazyme) according to the manufacturer’s instructions. All primers were synthesized by Sangon Biotech, and the primer sequences are listed in Supplementary Table 1. The thermal cycling conditions consisted of an initial denaturation at 95 °C for 30 s, followed by 40 cycles of 95 °C for 10 s and 60 °C for 30 s, and a melt curve analysis with incubation at 95 °C for 15 s, 60 °C for 60 s, and 95 °C for 15 s.

For the detection of Tap1, Tap2, and B2m expression in macrophages, single-cell suspensions were prepared from mouse lymph nodes, and macrophages were isolated by BD FACSAria Fusion Cell Sorter.

For analysis of Gzma and Gzmb expression in the liver, immune cells were isolated from liver tissue suspensions by Percoll (Cytiva) density gradient centrifugation.

Total cellular RNA was isolated from sorted macrophages and liver-derived immune cells using an Animal Tissue/Cell Total RNA Extraction Kit (Tiangen). RT-qPCR was performed using the HiScript II One Step RT-qPCR SYBR Green kit (Vazyme) according to the manufacturer’s instructions. All primers were synthesized by Sangon Biotech, and the primer sequences are listed in Supplementary Table 1. The thermal cycling conditions consisted of an initial denaturation at 95 °C for 30 s, followed by 40 cycles of 95 °C for 10 s and 60 °C for 30 s, and a melt curve analysis with incubation at 95 °C for 15 s, 60 °C for 60 s, and 95 °C for 15 s.

ELISA

ELISA kits for the detection of HBsAg and HBeAg were kindly provided by Shanghai Kehua Bio-engineering Co., Ltd. The assays were performed in accordance with the manufacturer’s instructions. Briefly, diluted serum samples were added to the wells, followed by incubation with horseradish peroxidase (HRP)-conjugated anti-HBsAg or anti-HBeAg antibodies. Colorimetric development was initiated using a substrate solution containing 0.5% hydrogen peroxide and 0.26% TMB, and the reaction was terminated accordingly. Absorbance was measured at 450 nm with 630 nm as the reference wavelength using a fully automated microplate reader (ELX800, BioTek Instruments). HBsAg and HBeAg concentrations were calculated based on OD₄₅₀–OD₆₃₀ values.

For treatment experiments, quantitative measurements of HBsAg and anti-HBsAg were additionally performed on the Alinity I system with the corresponding kits (Abbott). For Fig. 4f, HBxAg, HBsAg, HBeAg, HBcAg, and preS1 were quantified using the corresponding kits from Mlbio, Ruixin, and Win-Win Biotech, as listed in Supplementary Data 2.

Isolation of leukocytes from murine lymph nodes, spleens, liver, and peripheral blood

Lymph nodes, spleens, livers, and peripheral blood were collected from mice. Lymph node leukocytes were isolated by mechanical dissociation, followed by centrifugation at 1800 rpm for 5 min. Splenocytes and liver leukocytes were prepared by mechanical dissociation, red blood cell (RBC) lysis, and centrifugation at 1800 rpm for 5 min. Peripheral blood leukocytes were obtained by direct RBC lysis of anticoagulated whole blood, followed by washing and centrifugation at 1800 rpm for 5 min.

Flow cytometry

To evaluate T, B, and NK cell responses, single-cell suspensions prepared from tissues were incubated with fluorochrome-conjugated antibodies (anti-mouse CD44, anti-mouse CD62L, anti-mouse CD69, anti-mouse CD95, anti-mouse GL7) for 30 min in the dark. For OVA-specific CD8⁺ T cell or HBsAg-specific CD8⁺ T cell assays, cell samples were first incubated with the H-2K^b-SIINFEKL tetramer or H-2K^b-VWLSVIWM tetramer for 10 min, followed by staining with surface marker antibodies.

To evaluate the antigen-specific CD4⁺ T and CD8⁺ T cell responses, single-cell suspensions (5 × 10⁶ cells per well) were seeded into 24-well plates (NEST Biotechnology Co., Ltd.) and stimulated for 6 h with either HBsAg peptide (10 μg/ml) or the preS1 peptide pool (10 μg/ml) (synthesized by GL Biochem Ltd.; sequences are provided in Supplementary Data 3) in the presence of the protein transport inhibitor brefeldin A (BFA, BioLegend). After stimulation, cells were first stained with indicated antibodies for surface antigens, then fixed and permeabilized using the Foxp3/Transcription Factor Staining Buffer Set (Invitrogen) according to the manufacturer’s instructions. Intracellular staining was performed using antibodies against IFN-γ, granzyme B, granzyme A, and TGF-β.

To evaluate the innate immune activation, single-cell suspensions of lymph nodes were incubated with antibodies against CD40, CD86, and H-2K^b bound to SIINFEKL for 30 min and washed thoroughly.

Data acquisition was carried out on a CytoFLEX flow cytometer (Beckman Coulter), and analysis was performed using CytExpert 2.4 and FlowJo v. 10.0.6 software. Detailed antibody information is provided in Supplementary Data 2. H-2K^b-SIINFEKL tetramer-APC and H-2K^b-VWLSVIWM tetramer-APC were kindly provided by the NIH Tetramer Core Facility.

Cytotoxicity assay and flow cytometry analysis

Livers were excised from CHB mice, minced into small pieces, mechanically dissociated through a copper mesh, and centrifuged at 900 rpm for 3 min at 4 °C. Primary hepatocytes were then prepared as a single-cell suspension and labeled with CellTracker™ Green CMFDA (Beyotime Biotechnology, 1:1000). Labeled hepatocytes were seeded in complete RPMI 1640 medium and incubated overnight at 37 °C prior to the assay.

Splenocytes isolated from immunized HBV model mice were used as effector cells and co-cultured with hepatocytes at an effector-to-target (E:T) ratio of 50:1 for 24 h at 37 °C. Cytotoxicity was quantified by measuring lactate dehydrogenase (LDH) release in culture supernatants using the CyQUANT™ LDH Cytotoxicity Assay Kit (Invitrogen), according to the manufacturer’s instructions. Briefly, supernatants were incubated with the reaction mixture in the dark, and absorbance was measured at 490 nm with 680 nm as the reference wavelength using a multifunctional microplate reader (SpectraMax iD3). Background-corrected absorbance values were calculated as OD₄₉₀–OD₆₈₀ and used to determine LDH release and cytotoxicity. Maximum LDH release was determined by treating target cells with 10% Triton X-100.

For flow cytometric analysis, cells were collected after co-culture, stained with propidium iodide (PI, Yeasen Biotechnology, 1:500) and DAPI, and immediately analyzed by flow cytometry.

Immunofluorescence staining

Liver tissues from mice were harvested, fixed in 4% paraformaldehyde (Biosharp), and cryosectioned at 10 μm thickness using a Leica CM1950 cryostat. Sections were fixed in 4% paraformaldehyde for 15 min, then permeabilized with 0.3% Triton X-100 in 1 × PBS, and blocked with 1% BSA in PBS for 30 min at room temperature. For HBcAg detection, sections were incubated with a primary antibody against HBcAg (1:200 dilution in 1% BSA; Immunoway) for 2 h at 37 °C. Following three washes with PBS, a secondary antibody (Goat anti-mouse IgG H&L, Alexa Fluor® 488, 1:500 in 1% BSA) (abcam) was applied and incubated for 1 h at 37 °C. For the detection of granzyme A and B expression, sections were incubated with fluorescently labeled antibodies against granzyme A (BioLegend, 1:100 dilution in 1% BSA) or granzyme B (BioLegend, 1:100 dilution in 1% BSA). Nuclei were counterstained with DAPI for 5 min at room temperature (protected from light). Finally, stained sections were imaged using Olympus SLIDEVIEW VS200 (Olympus) and analyzed with OLYMPUS OlyVIA software.

Western blot

RAW264.7 cells were transfected with OVA mRNA (0.5 μg) or an equimolar dose of OVA-HBsAg mRNA using an mRNA-specific transfection reagent (GoldenTran) according to the manufacturer’s instructions. For intracellular protein analysis, cells were lysed in RIPA buffer (Beyotime Biotechnology) supplemented with protease inhibitor (PMSF, Beyotime Biotechnology). For secreted protein analysis, culture supernatants were collected and clarified by centrifugation. Protein concentrations were determined using NanoAccu S (Yooning). Equal amounts of protein (25 μg) were separated on 12% SDS-PAGE gels (Biosharp) and transferred onto PVDF membranes. Membranes were blocked with 5% non-fat milk in TBST for 2 h at room temperature and incubated overnight at 4 °C with anti-OVA primary antibody (Sigma-Aldrich, 1:5000 dilution in TBST). After washing with TBST, membranes were incubated with HRP-conjugated Goat anti-rabbit antibody (CST, 1:2000 dilution in TBST) for 45 min at room temperature. Protein bands were visualized using ImageQuant 800 (Cytiva).

Endogenous presentation versus cross-presentation assay in RAW264.7 cells

To distinguish endogenous MHC-I antigen presentation from cross-presentation of secreted antigen, RAW264.7 cells were analyzed under two matched conditions: direct mRNA transfection (endogenous presentation) and conditioned-medium (CM) transfer (cross-presentation). For endogenous presentation, RAW264.7 cells were transfected with OVA or OVA-HBsAg mRNA as described above. For cross-presentation, donor RAW264.7 cells were transfected in parallel with the corresponding mRNAs; after 24 h incubation, CM was collected, clarified by centrifugation to remove debris, and applied to naïve RAW264.7 cells for matched incubation. After 24 h of treatment, surface SIINFEKL-H-2K^b presentation was measured by flow cytometry and quantified as mean fluorescence intensity (MFI).

Isolation and in vitro differentiation of BMDCs and PBDCs

BMDCs and PBDCs were generated from both healthy mice and CHB model mice. Bone marrow cells were isolated from the femurs and tibias, and peripheral blood mononuclear cells (PBMCs) were obtained from anticoagulated whole blood by Ficoll density-gradient centrifugation. Cells were seeded at 1 × 10⁶ cells/ml in complete RPMI 1640 medium supplemented with GM-CSF (Sino Biological) and IL-4 (Sino Biological), at 20 and 10 ng/ml for BMDCs and 100 and 50 ng/ml for PBDCs, respectively. On day 2, non-adherent cells were completely removed, and half of the medium was replaced with prewarmed complete medium containing GM-CSF and IL-4. On day 4, adherent cells were discarded and non-adherent cells were retained; half of the medium was then replaced with fresh prewarmed medium containing GM-CSF and IL-4. On day 6, non-adherent and loosely adherent cells were collected by gentle washing with 1 × PBS.

Detection of anti-preS1 antibodies by ELISA

96-well plates were coated overnight at 4 °C with preS1 protein (Beyotime Biotechnology) at 0.2 μg per well in PBS (100 μl/well). Plates were washed three times with PBS and blocked with 2% BSA in PBS for 2 h at room temperature. Serum samples were heat-inactivated at 56 °C for 30 min, equilibrated to room temperature, and serially diluted in PBS-T (PBS with 0.1% Tween 20) containing 2% BSA (starting at 1:10 followed by 1:3 serial dilutions; 7 dilutions, 100 μl/well). Diluted sera were added to the coated plates and incubated for 1 h at room temperature, followed by four washes with PBS-T. Peroxidase AffiniPure® Goat Anti-Mouse IgG (H + L) (Jackson Immuno Research, 1:10,000) was added and incubated for 1 h at room temperature. Plates were washed four times with PBS-T, developed with 3,3′,5,5′-tetramethylbenzidine (TMB) substrate (100 μl/well) for 8 min in the dark, and the reaction was stopped with TMB stop solution (50 μl/well). Absorbance at 450 nm was measured using a SpectraMax iD5 plate reader. Each sample was assayed in triplicate.

RNA sequencing

Liver tissues (n = 3 biologically independent mice per group) from AAV-HBV1.3-infected mice were collected, and leukocytes were isolated using Percoll density gradient centrifugation (Cytiva). Total RNA was extracted with TRIzol Reagent (Invitrogen) following standard protocols. RNA sequencing was performed by GENEWIZ utilizing the Illumina platform, with subsequent bioinformatics analysis including comprehensive functional annotation through GO analysis and pathway enrichment analysis via the KEGG database.

Granzyme B ELISpot

Granzyme B-secreting cells were quantified using a PVDF-based ELISpot assay. Briefly, ELISpot plates (Merck) were coated overnight at 4 °C with capture antibody (from the Mouse GZMB ELISA Kit, R&D Systems; 1:50 in 1 × PBS). Plates were washed with 1 × PBS and blocked with 1 × PBS containing 1% BSA and 5% sucrose for 2 h at room temperature. Splenocytes were isolated from mice that had received 3 doses of preS1-HBsAg vaccine and resuspended in complete RPMI 1640 medium. Cells were seeded at 5 × 10⁵ cells per well and stimulated with individual preS1 peptides (20 peptides in total) at a final concentration of 20 µg/ml per well. Plates were incubated at 37 °C with 5% CO₂ for 24 h. After incubation, plates were washed with PBST (1 × PBS containing 0.05% Tween-20) and incubated overnight at 4 °C with biotinylated detection antibody (from the Mouse GZMB ELISA Kit; 1:50 in 1 × PBS containing 1% BSA). Plates were then washed and incubated with streptavidin-alkaline phosphatase (Mabtech; 1:1000 in 1 × PBS containing 0.5% FBS) for 1 h at room temperature. Spot development was performed using ALP substrate (Mabtech) at 37 °C until distinct spots appeared. The reaction was stopped by rinsing with tap water, and plates were air-dried prior to analysis.

Safety evaluation

C57BL/6J mice received three vaccine immunizations (10 μg per dose) at weekly intervals starting at week 4. Throughout the study period, we monitored serum ALT/AST levels, as well as body weight. At the experimental endpoint (week 10), blood samples and livers were collected for comprehensive safety assessment, including hematological analysis and liver histopathology by H&E staining. ALT and AST were measured using commercial kits (Nanjing Jiancheng Bioengineering Institute) according to the manufacturer’s instructions. Hematological analysis was performed using a Sysmex XT-1800i automated analyzer to quantify key parameters, including white blood cell (WBC) and platelet counts. Livers were fixed and processed for histopathological examination via H&E staining, and all slides were digitally imaged using an Olympus SLIDEVIEW VS200 system (Olympus).

Statistical analysis

All statistical analyses were performed using GraphPad Prism 9.5 software. The statistical tests used for each analysis are specified in the respective figure legends. Intergroup comparisons were evaluated by a two-tailed unpaired Student’s t-test. For correlation analyses, Pearson correlation coefficients and two-sided Pearson’s correlation P values were calculated. Results are presented as mean ± s.e.m., with the sample size (n) indicated in respective figure legends. n.s. is P ≥ 0.05, and significant P values are shown.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

Y.C.W., M.L., and B.W.Z. conceptualized and designed the research. B.W.Z. and Y.Y.W. performed the experiments and contributed to data analysis. D.C., M.Y.L., J.X., and X.M. provided help in experiments. J.C.Z. and Y.W. contributed to discussions and provided critical suggestions. B.W.Z. and Y.Y.W. prepared the figures. B.W.Z., M.L., Y.C.W., and Y.Y.W. wrote the manuscript. Y.C.W. and M.L. supervised the project.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

Funding

This work was supported by the National Key R&D Program of China (2024YFB3814600 to M.L.), the National Natural Science Foundation of China (32571604 and 82173390 to M.L., 52495014 and U25A20146 to Y.C.W., and 32401192 to Y.W.), the Strategic Priority Research Program of the Chinese Academy of Sciences (XDB0490000 and XDB0940303 to Y.C.W.), Anhui Provincial Key Research and Development Project (2023s07020019 to Y.C.W.), Anhui Provincial Major Science and Technology Project (202303a07020010 to Y.C.W.), Anhui Provincial Natural Science Foundation (2408085J042 to M.L. and 2408085QC112 to Y.W.), USTC Research Funds of the Double First-Class Initiative (YD9100002054 to Y.C.W. and YD9110002021 to M.L.), and China Postdoctoral Science Foundation (2025T180790 and 2025M772932 to Y.W.).

Data availability

The data supporting the results in this study are available within the paper and its Supplementary Information. The transcriptomic data generated in this study are available under GSA accession CRA030981. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains Supplementary material available at 10.1038/s41467-026-74006-y.