Single-cell landscape of T-cell responses in acute HBV patients

A thorough understanding of the T cell responses to HBV infection in patients is limited but crucial to explore potential intercellular mechanisms of HBsAg presentation. To address this question, we conducted single-cell RNA sequencing (scRNA-seq) on peripheral blood mononuclear cells (PBMC) from five patients with acute HBV infection (AHB) and five healthy donors (HD) (Fig. 1a and Supplementary Data 1). This allowed us to examine the transcriptomic profiles of these cells and their T cell receptor (TCR) repertoire (Fig. 1a). Concurrently, we analyzed immune cells from a liver biopsy of one patient (Fig. 1a). By employing anchoring integration and unsupervised clustering, we identified 36 unique cell types and states (subsets) in a pool of 49,305 high-quality single cells (Fig. 1b), based on their distinct signature markers of gene expression (Supplementary Fig. 1a and Supplementary Data 2). We also confirmed that these cell clusters were consistently present across all PBMC samples (Supplementary Fig. 1b).

a A schematic workflow of the study design (top) and the clinical time course of five acute HBV infection (AHB) patients (bottom) (created with BioRender.com). b Uniform manifold approximation and projection (UMAP) visualization of PBMCs and liver cells, colored by cluster. CD4 TEF GZMK: Granzyme K⁺ effector CD4⁺ T cells; CD8 TEX GZMA: Granzyme A⁺ exhausted CD8⁺ T cells. c Box plots showing the fold enrichment of cell type proportions between AHB patients (n = 5) and healthy donors (HDs, n = 5). In each plot, the box represents the interquartile range, with the median indicated by a horizontal line. The lower whisker extends to the minimum (lowest value within 1.5×IQR) and the upper whisker extends to the maximum (highest value within 1.5 × IQR). d Pathway enrichment analysis of T cells in AHB patients versus HDs. e Box plots illustrating the cytotoxicity scores of TEF and TEX CD8⁺ T cell subsets in AHB patients (n = 5) versus HDs (n = 5). Each box plot displays the minimum, lower quartile, median, upper quartile, and maximum values. f Clonal status percentages of T cells across AHB patients and HDs. g Proportion of single and expanded clones (left) and quantification of clonal expansion (right) in T cell clusters in AHB patients (n = 5) versus HDs (n = 5). The point estimates (odds ratios) are depicted with error bars representing the 95% confidence intervals. h Gene set enrichment analysis (GSEA) of myeloid cell clusters in AHB patients versus HDs. ISG: Interferon-stimulated genes. i GSEA of dendritic cells (DCs) in AHB patients versus HDs. j Circos plot illustrating the prioritized interactions mediated by ligand-receptor pairs between hepatic-enriched myeloid cell subsets and activated and cycling CD8⁺ T cells within the liver. Statistical analysis was performed using two-tailed Mann-Whitney tests (c), one-tailed Fisher’s exact test (d), two-tailed Student’s t-tests (e), two-tailed Fisher’s exact test (g) and Kolmogorov-Smirnov (K-S) test followed by Benjamini-Hochberg procedure (h, i). Statistically significant results are labeled (*p < 0.05, **p < 0.01, ***p < 0.001).

We classified the T cells identified among the PBMCs into 18 subsets, including Tfh, effector (TEF), exhausted (TEX) clusters, innate-like T cells, and others (Fig. 1b). By comparing the relative proportions of these subsets in patients and healthy donors, we found the Tfh, cycling CD8⁺, and CD8⁺ TEX granzyme A⁺ (GZMA) subsets were significantly enriched, whereas gamma-delta (γδ) T cells were diminished in patients (Fig. 1c). Gene ontology (GO) pathway enrichment analysis of the T cells further indicated more transcription from genes involved in T cell responses in the cells from patients than in those from healthy donors (Fig. 1d). Moreover, patient TEF and TEX CD8⁺ T cells scored more highly in cytotoxicity assays than did these cells from healthy donors (Fig. 1e). Discriminant regulon expression analysis (DoRothEA), which estimates transcription factor activity from the mRNA levels of their target genes, predicted that the observed enhancement of T cell responses is driven by certain signaling pathways and transcription factors (Supplementary Fig. 1c, d). This includes, but is not limited to, well-known innate immunity modulators such as JAK-STAT signaling pathways, and the transcription factors IRF2 and RELA (Supplementary Fig. 1c, d)^34,35, suggesting the presence of a complex regulatory network that influences T cell dynamics. To determine which T cell subsets are involved in the response to acute HBV infection, we analyzed TCR clonality and observed more clonal expansion in patients than in healthy donors (Fig. 1f), particularly expansion of differentiated CD4⁺ T cells, CD8⁺ TEF granzyme K⁺ (GZMK), and CD8⁺ TEF granulysin⁺ (GNLY) subsets (Fig. 1g). Additionally, we observed an increase in the Vγ9- subset within the γδ T cell population from the PBMCs of acute HBV patients (Supplementary Fig. 1e), consistent with a previous report describing dominant clonal expansions of human Vγ9- T cells during adaptive antiviral immunosurveillance³⁶. These findings indicate a robust T cell response during acute HBV infection.

Myeloid cells are crucial for antigen presentation in T cell immunity¹⁵. These cells comprised 7 subsets within PBMCs, including monocytes, macrophages, and DCs (Fig. 1b). Gene set enrichment (GSEA) and gene expression analyses revealed that, during acute HBV infection, specific subsets of myeloid cells exhibited enhanced activation and adhesion (Fig. 1h and Supplementary Fig. 1f), a response that can be primed by granulocyte-monocyte colony-stimulating factor (GM-CSF)³⁷. In contrast, genes associated with interferon (IFN) production and signaling, as well as pro-inflammatory cytokine secretion, did not show increased expression in these cells compared to those from healthy donors (Fig. 1h and Supplementary Fig. 1f). Additionally, DoRothEA analysis indicated downregulation of key innate immunity modulators, including NF-κB and JAK-STAT signaling pathways, as well as the transcription factor RELA, in these cells (Supplementary Fig. 1g, h). These findings are consistent with previous reports that patients with acute HBV infection do not mount a robust innate immune response to the virus^6,8. Further GSEA of DC subsets found that genes involved in lymphocyte activation and immune synapse formation were downregulated in cDC2s from patients when compared with healthy controls (Fig. 1i), suggesting that cDC2s may be ineffective in driving CD4⁺ T cell responses in these patients, contrary of what might be expected from our understanding of classical CD4⁺ T cell activation by cDC2s^13,21.

Comparative analysis of the immune cell subsets in PBMCs and matched liver tissue from the same patient revealed significantly greater enrichment of certain myeloid cell and CD8⁺ T cell subsets in the liver than in PBMCs (Supplementary Fig. 1i). By using NicheNet algorithm to predict ligand-receptor interactions between these enriched cell subsets, we found that cDC1s established most ligand-receptor interactions with activated and proliferating CD8⁺ T cells, surpassing even the interactions of Kupffer cells (specialized macrophages in the liver) and other myeloid cells with CD8⁺ T cells (Fig. 1j). This suggests that cDC1s, in addition to the previously reported Kupffer cells^38,39,40, may be also crucial for priming CD8⁺ T cell responses against HBV in the liver, though further in-depth studies involving more patients are needed to confirm these findings. Overall, our scRNA-seq analysis indicates that HBsAg presentation underlying T cell immunity in acute HBV patients may involve cDC1s, independent of cDC2s, suggesting the potential presence of novel intercellular mechanisms.

Intrahepatic Batf3⁺XCR1⁺CCR7^– cDC1s drive CD8⁺ T cell response against HBsAg

As the location of CD8⁺ T cell priming in HBV infection remains poorly understood⁵, we began investigating the sites of CD8⁺ T cell responses following HBsAg expression in hepatocytes. To do so, we created a recombinant HBV construct, pHBV1.3/ovalbumin (OVA) 257-264, which expresses HBsAg fused to an epitope of OVA, OVA257-264, in the context of HBV replication (Supplementary Fig. 2a). This epitope is recognized specifically by the CD8⁺ T cells of OT-I mice, which express a transgenic TCR specific for the OVA257-264 peptide bound to the MHC-I molecule H-2K^b41. The recombinant HBsAg, containing the inserted epitope, remains functional: it is secreted and supports HBV release (Supplementary Fig. 2b). Also, it has a buoyant density and particle size similar to wild-type (WT) HBsAg (Supplementary Fig. 2c, d). Following hydrodynamic injection into the tail vein of mice (Supplementary Fig. 2e), this construct instigated earlier activation, increased division, and amplified effector functions of OT-I T cells in the liver than in the spleen and liver-dLNs, as analyzed by flow cytometry (Supplementary Fig. 2f). Consistent with this conclusion, the CD8⁺ T cell response in the liver was unchanged by prior splenectomy combined with an anti-CD62L antibody treatment (Supplementary Fig. 2g, h), which inhibits the homing of naïve T cells to LNs, or treatment with sphingosine-1-phosphate receptor modulator FTY720 (Supplementary Fig. 2g, i), which blocks the egress of T cells from LNs. We further confirmed that the liver supports CD8⁺ T cell responses against HBsAg using HBV Env28 TCR transgenic T cells (Supplementary Fig. 2e, j), which specifically recognize an HBsAg epitope presented by MHC-I (H-2K^d)⁴². These results corroborate the essential role of the liver in priming CD8⁺ T cells against HBsAg, at least in mice^42,43.

The mechanism of CD8⁺ T cell priming in the liver is controversial due to the multiplicity and heterogeneity of antigen-presenting cells^38,39,40,44. To identify the subsets and the cellular interaction responsible for HBsAg-related CD8⁺ T cell responses in the liver, we performed scRNA-seq on liver cells sorted by fluorescence activated cell sorting (FACS) from mice at days 0 and 2.5 after the hydrodynamic injection of pHBV1.3/OVA257-264 (Supplementary Fig. 3a). This analysis identified various cell types, including diverse DC and CD8⁺ T cell subsets in the liver (Fig. 2a, b, Supplementary Fig. 3b and Supplementary Data 3), with clustering patterns similar to those observed in humans (Fig. 1b)⁴⁵. We analyzed the subsets for the presence of MHC-I (H-2K^b) molecules bound to the OVA257-264 peptide epitope SIINFEKL by flow cytometry after labeling with an antibody, 25-D1.16, which binds specifically to the H-2K^b-SIINFEKL complex. Intrahepatic resident cDC1s (CCR7^–) had the most H-2K^b-SIINFEKL complex, hepatocytes, LESCs, migratory cDCs (CCR7⁺, mig-cDCs), and monocyte-derived DCs (mono-DCs) had lesser amounts, and Kupffer cells, B cells and cDC2s had none (Fig. 2c). Co-culture of OT-I T cells with various liver cell subsets showed that only cDC1s could induce OT-I T cell proliferation (Fig. 2d). Consistent with this finding, 3.5 days after hydrodynamic injection, OT-I T cell proliferation and differentiation were nearly completely blocked in Batf3 KO mice (Fig. 2e, f), which lack cDC1s in their liver (Supplementary Fig. 3c). Furthermore, when the recruitment of cDC1 cells via the XCR1 receptor and its ligand XCL1 was blocked using an anti-XCL1 antibody, there was significantly less OT-I T cell proliferation and differentiation (Fig. 2g, h). By contrast, OT-I T cell proliferation and differentiation were unaffected in monocyte-deficient CCR2 KO, Kupffer cell-depleted, and B cell-deficient μMT mice (Supplementary Fig. 3d–i). We conclude from these experiments that the intrahepatic resident Batf3⁺XCR1⁺CCR7^– cDC1s are the subset of antigen-presenting cells crucial for HBsAg-specific CD8⁺ T cell responses in the liver, aligning with the established role of cDC1s in CD8⁺ T cell responses^13,46, as exemplified by lymphocytic choriomeningitis virus (LCMV) Armstrong infection (Supplementary Fig. 3j, k).

a UMAP plot of 4829 single liver cells as illustrated in Supplementary Fig. 3a. b Dot plot of selected canonical cell markers across populations detailed in (a). c Representative FACS plots and percentages of H-2K^b-SIINFEKL-expressing antigen-presenting cells (APCs) in WT mouse livers at days 0 and 2.5 post-pHBV1.3/OVA257-264 hydrodynamic injection (HDI) (n = 3). d Representative Violet tracer dilution and proliferation index of OT-I T cells after a 3-day culture alone or with liver-sorted APCs from WT mice post-1 day pHBV1.3/OVA257-264 HDI (n = 6 biologically independent replicates). e, f Violet tracer-labeled OT-I T cells were transferred into WT or Batf3 KO mice, followed by pHBV1.3/OVA257-264 HDI for 3.5 days. e Representative Violet tracer dilution and proliferation index (n = 6). f Representative FACS plots and percentages of IFN-γ⁺ or GzmB⁺ OT-I T cells (n = 6). g, h Violet tracer-labeled OT-I T cells were transferred into WT recipients treated with anti-XCL1 versus control antibody, followed by pHBV1.3/OVA257-264 HDI for 3.5 days. g Representative Violet tracer dilution and proliferation index (n = 6). h Representative FACS plots and percentages of IFN-γ⁺ or GzmB⁺ OT-I T cells (n = 6). i, j Same as (g, h) but using HBV Env28 T cells as donors and BALB/C mice as recipients receiving pHBV1.3 for HDI (n = 3). k, l Representative FACS plots and percentages of HBV ENV353 tetramer-specific CD8⁺ T cells in Batf3 KO versus WT mice at day 5 post-pHBV1.3 HDI (n = 6), and representative FACS plots and percentages of EdU⁺ cells among these tetramer-specific CD8⁺ T cells (n = 3). m Intrahepatic HBsAg levels in WT and Batf3 KO mice on day 14 post-pHBV1.3 HDI (n = 15 fields from 5 mice). Scale bars, 50 μm. At least two independent experiments were performed. The horizontal lines represent median values. Statistical significance was calculated using one-way ANOVA with Dunnett’s multiple comparisons test (d) and two-tailed Student’s t-tests (c, e-m). Significance levels are indicated as follows: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Source data and exact p-values are provided in the Source Data file.

We confirmed our conclusion that the cDC1s initiate CD8⁺ T cell responses against HBsAg by using HBV Env28 TCR transgenic T cells (Fig. 2i, j). Further validation was provided by employing an MHC-peptide tetramer and 5-ethynyl-2’-deoxyuridine (EdU) incorporation assay to detect the proliferation of endogenous HBsAg-specific CD8⁺ T cells (Fig. 2k, l), thereby ruling out the potential artifact associated with transgenic TCRs. The predominance of cDC1s in the CD8⁺ T cell response in the liver was also evident in two additional HBV mouse models: the prcccDNA model (Supplementary Fig. 3l, m) and the AdHBV model with a high infection dose of 3 × 10⁹ infectious units (i.u.) (Supplementary Fig. 3n, o)³². In addition, supporting a specific role of cDC1s in HBsAg presentation, we found that cDC1s (Fig. 2m, and Supplementary Fig. 4a, b), but not macrophages (Supplementary Fig. 4c), were essential for the clearance of HBsAg from liver tissues. Thus, consistent with the NicheNet analysis of our human scRNA-seq data (Fig. 1j), the intrahepatic resident Batf3⁺XCR1⁺CCR7^– cDC1s are the primary antigen-presenting cells in the CD8⁺ T cell response against HBsAg in the liver.

The cDC1s acquire and present HBsAg by MHC-I cross-dressing but not cross-presentation

cDC1s can present the antigens that they take up by endocytosis on MHC-I rather than on MHC-II molecules, a process called cross-presentation⁴⁶. To investigate whether cDC1s elicit CD8⁺ T cell responses to HBsAg by endocytosis and cross-presentation, we first tested whether a non-secreted variant of recombinant HBsAg expressed by the pHBV1.3/OVA257-264 construct would prime the OT-I T cell response. Despite the fact that this form of recombinant HBsAg was not secreted (Supplementary Fig. 5a), the amount of MHC-I-SIINFEKL on cDC1s and the OT-I T cell response was unaffected after hydrodynamic injection of this construct into WT mice that had previously received a transfer of OT-I T cells (Supplementary Fig. 5b–d), indicating that cDC1s do not take up recombinant HBsAg by endocytosis for cross-presentation. We investigated whether cDC1s take up recombinant HBsAg released from hepatocytes in extracellular vesicles (exosomes) by using Rab27a KO mice in which exosome release is impaired^47,48. Again, MHC-I-SIINFEKL presentation and OT-I T cell responses were unaffected (Supplementary Fig. 5e–g). Also, we examined whether cDC1s acquired HBsAg from dying hepatocytes by expressing HBsAg as a fluorescent fusion protein with ZsGreen from an HBV replicon, utilizing the virus’s own hepatocyte-specific promoter (Supplementary Fig. 5h). The ZsGreen signal in hepatocytes indicated intracellular HBsAg (Supplementary Fig. 5i). Although we found ZsGreen in Kupffer cells and liver sinusoidal endothelial cells (LSEC), we did not find it in cDC1s (Supplementary Fig. 5j). Furthermore, blocking DNGR-1, a receptor implicated in dead cell debris recognition and internalization by cDC1s⁴⁹, did not affect antigen presentation or CD8⁺ T cell priming by cDC1s (Supplementary Fig. 5k–m). Together, these data indicate that the Batf3⁺XCR1⁺CCR7^– cDC1s do not acquire recombinant HBsAg from the environment, thus they may not use classical cross-presentation to present on MHC-I-HBsAg peptide antigens to CD8⁺ T cells.

cDC1s can acquire antigenic peptides bound to MHC-I by non-canonical MHC-I cross-dressing under certain circumstances⁵⁰. If the Batf3⁺XCR1⁺CCR7^– cDC1s obtain intact HBsAg peptide-MHC-I complexes from neighboring cells by MHC-I cross-dressing, MHC-I expression by the cDC1s would not be necessary for a T cell response. To test this prediction, we transferred cDC1s from MHC-I-deficient β2M KO mice, or control WT mice, into cDC1-deficient Batf3 KO mice (Fig. 3a). Unlike in LCMV Armstrong infection, where MHC-I expression on cDC1s was required for effective CD8⁺ T cell priming by cross-presentation (Supplementary Fig. 6a, b), the β2M KO cDC1s primed CD8⁺ T cell responses against HBsAg as well as the WT cDC1s did, indicating that the MHC-I defective cDC1s acquired HBsAg peptide-MHC-I complexes from neighboring cells by cross-dressing (Fig. 3b, c). To consolidate this conclusion, we implanted bone marrow from Batf3 KO or WT BALB/c (H-2K^d) into lethally irradiated F1 (C57BL/6 X BALB/c) mice (Fig. 3d). Upon pHBV1.3/OVA257-264 injection, OT-I T cell expansion occurred in chimeras reconstituted with BALB/c bone marrow, but not in those with Batf3 KO bone marrow (Fig. 3e), underscoring H-2K^b-restricted OT-I response mediated by BALB/c cDC1s. The cDC1s in the mice that received the BALB/c bone marrow were exclusively BALB/c-derived, as expected (Supplementary Fig. 6c). When we measured the percentages of various classes of BALB/c DCs that had H-2K^b-SIINFEKL, the class with the highest percentage was the cDC1s (Fig. 3f), indicating that these cells must have acquired the H-2K^b-SIINFEKL by cross-dressing.

a A schematic of the cell reconstitution strategy and experimental timeline. b Representative Violet tracer dilution and proliferation index from the mice described in (a) (n = 6). c Representative FACS plots and percentages of IFN-γ⁺ or GzmB⁺ OT-I T cells from the mice described in (a) (n = 3). d Schematic of bone marrow chimeras generation and experimental timeline (created with BioRender.com). e Representative Violet tracer dilution and proliferation index of OT-I T cells from mice described in (d) at day 3.5 post-pHBV1.3/OVA257-264 HDI (n = 5). f Representative FACS plots and percentages of H-2K^b-SIINFEKL-expressing BALB/c APCs at day 0 and 2.5 in the livers of mice from (d) post-pHBV1.3/OVA257-264 HDI (n = 6). g Representative FACS plots and percentages of ENV353 tetramer-specific CD8⁺ T cells from mice described in (a) at day 5 post-pHBV1.3 HDI (n = 6). h Representative Violet tracer dilution and proliferation index of Env28 T cells in F1 chimeras reconstituted with C57BL/6 versus Batf3 KO bone marrow at day 3.5 post-pHBV1.3 HDI (n = 3). i Intrahepatic HBsAg levels in Batf3 KO mice reconstituted with β2M KO or WT cDC1s at day 14 post-pHBV1.3 HDI (n = 15 view fields from 5 mice). Scale bars, 50 μm. j Schematic of chimeras generation and experimental timeline (created with BioRender.com). k Representative FACS plots and percentages of H-2K^b-SIINFEKL-expressing cDC1s from mice described in (j) at day 2.5 post-pHBV1.3/OVA257-264 HDI (n = 6). l Representative FACS plots and percentages of H-2K^b-SIINFEKL-expressing cDC1s after a 1-day co-culture with unpulsed or SIINFEKL-pulsed hepatocytes or LSECs (n = 3 biologically independent replicates). At least two independent experiments were performed. Horizontal lines in (b, c, e–i, k) represent median values. Error bars in (l) represent mean ± SD. Statistical significance was calculated using two-tailed Student’s t-tests (b, c, e–i, k) and one-way ANOVA with Tukey’s multiple comparisons test (l). Significance levels are indicated as follows: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Source data and exact p-values are provided in the Source Data file.

To demonstrate that MHC-I cross-dressing was not confined to a specific experimental system, we first confirmed that the response of endogenous HBsAg-specific CD8⁺ T cells was independent of MHC-I expression on cDC1s in the pHBV1.3 mouse model (Fig. 3g). Further, by reconstituting F1 mice with either Batf3 KO or C57BL/6 (H-2K^b) bone marrow, we showed that C57BL/6-derived cDC1s could also mediate the H-2K^d-restricted response of HBV Env28 TCR transgenic T cells (Fig. 3h). Similar evidence of cDC1 cross-dressing was observed in both the prcccDNA and high-dose AdHBV models. Specifically, HBsAg-specific CD8⁺ T cell expansion occurred independently of MHC-I expression on cDC1s (Supplementary Fig. 6d, e), and C57BL/6-derived cDC1s mediated the H-2K^d-restricted response of HBV Env28 TCR transgenic T cells in these two models (Supplementary Fig. 6f, g). cDC1 cross-dressing was also observed after hydrodynamic injection of a low dose of pHBV1.3/OVA257-264 (Supplementary Fig. 7a-c), which resulted in low HBsAg expression (Supplementary Fig. 7d, e), and in AdHBV (low-dose infection: 1 × 10⁸ i.u.³²) and AAV8-HBV infections (Supplementary Fig. 7f–k), both of which caused mild yet observable liver inflammation (Supplementary Fig. 7l)^32,51. These findings suggest that cDC1 cross-dressing may not be exclusively dependent on the rapid expression of a large amount of HBsAg or on marked inflammation. Moreover, hydrodynamic injection of the pALB-Igk/HEL-OVA257-264 plasmid, expressing a secretable hen egg lysozyme (HEL)-OVA257-264 fusion protein under the hepatocyte-specific albumin (ALB) promoter (Supplementary Fig. 8a), demonstrated that cDC1s drove OT-I T cell proliferation in a manner dependent on the expression of MHC-I by cDC1s themselves (Supplementary Fig. 8b, c). This raises the possibility that cDC1 cross-dressing may not be a general feature of secreted proteins from hepatocytes, but rather a mechanism potentially specific to HBsAg. Finally, while cDC1s played a significant role in intrahepatic HBsAg clearance (Fig. 2m and Supplementary Fig. 4a, b), this process did not require MHC-I expression by cDC1s (Fig. 3i), indicating that MHC-I cross-dressing of cDC1s contributes to the clearance of intrahepatic HBsAg. In conclusion, intrahepatic resident Batf3⁺XCR1⁺CCR7^– cDC1s drive the CD8⁺ T cell response against HBsAg in the liver through MHC-I cross-dressing.

Irradiation-resistant cells provide HBsAg peptide-MHC-I complexes for cross-dressing

To identify the source of HBsAg peptide-MHC-I complexes for cross-dressing of cDC1s, we reconstituted irradiated WT and β2M KO mice with bone marrow from WT mice (Fig. 3j). Many fewer cDC1s presented recombinant HBsAg in the β2M KO mice when compared with the WT controls (Fig. 3k), suggesting that the HBsAg peptide-MHC-I complexes used to cross-dress cDC1s come from irradiation-resistant cells. LSECs, which can uptake dying hepatocytes (Supplementary Fig. 5j), present H-2K^b-SIINFEKL complexes at higher levels than hepatocytes (Fig. 2c). This led us to hypothesize that LSECs, in addition to HBsAg-producing hepatocytes, may contribute to cross-dressing of cDC1s. To test this, we loaded SIINFEKL onto both LSECs and hepatocytes and cocultured them with cDC1s to assess H-2K^b-SIINFEKL complex transfer. Flow cytometry analysis showed that both LSECs and hepatocytes transferred these complexes to cDC1s, with LSECs demonstrating greater efficiency (Fig. 3l). In conclusion, intrahepatic resident Batf3⁺XCR1⁺CCR7^– cDC1s likely utilize HBsAg peptide-MHC-I complexes derived from irradiation-resistant cells, particularly LSECs, to drive the CD8⁺ T cell response against HBsAg.

B cells but not cDCs initiate CD4⁺ T cell responses against HBsAg

To investigate the mechanism by which antigen-presenting cells acquire HBsAg to drive CD4⁺ T cell responses, we generated a recombinant pHBV1.3 replicon (pHBV1.3/OVA323-339) that expresses HBsAg fused to the OVA323-339 peptide (Supplementary Fig. 2a). The OVA323-339 peptide bound to MHC-II is recognized by CD4⁺ T cells in OT-II mice, analogous to the previous CD8⁺ T cell experiments with OT-I mice. To identify the antigen-presenting cell subsets that present OVA323-339 to CD4⁺ T cells, we transferred OT-II T cells into mice lacking one subset and then introduced pHBV1.3/OVA323-339 by hydrodynamic injection. Ablation of all cDCs in Zbtb46-diphtheria toxin (DT) receptor (DTR) mice by treatment with the toxin had no effect on early OT-II T cell activation (Supplementary Fig. 9a), proliferation (Supplementary Fig. 9b), or Tfh and effector differentiation (Supplementary Fig. 9c, d) in the spleen, dLNs, and liver after injection of pHBV1.3/OVA323-339, as determined by flow cytometry. This was true, likewise, when either the cDC1 or the cDC2 subset was constitutively deficient in Batf3 KO mice and Irf4^fl/flCD11cCre mice, respectively (Supplementary Fig. 9e, f). As a positive control, immunization with OVA-conjugated sheep red blood cells (SRBC-OVA) elicited the classical cDC-dependent OT-II T cell responses (Supplementary Fig. 9g, h)⁵². These findings challenge the conventional paradigm that cDCs, especially cDC2s, initiate CD4⁺ T cell responses^13,21, but are consistent with our human scRNA-seq data (Fig. 1i) and with the clearance of intact HBsAg from the circulation and liver tissues in Irf4^fl/flCD11cCre mice (Supplementary Fig. 10a, b).

Similarly, the absence of monocytes or macrophages had no effect on OT-II T cell proliferation (Supplementary Fig. 9i, j). In B cell-deficient μMT mice, by contrast, OT-II T cell activation was abrogated in the spleen and impaired in the dLNs but remained intact in the liver (Fig. 4a). Notably, there was almost no proliferation (Fig. 4b) or differentiation (Fig. 4c, d) of OT-II T cells in these three organs of μMT mice. These findings indicate that B cells are essential for initiating OT-II T cell responses, though other cells may provide weak activation signals that are insufficient for full responses, particularly in the liver. A similar phenomenon was observed in vesicular stomatitis virus (VSV)-specific BCR transgenic VI10YEN mice (Supplementary Fig. 11a-c), which lack a normal B cell repertoire. Importantly, a role of B cells was also observed in the response of endogenous CD4⁺ T cells, which differentiated less in μMT mice than in WT mice (Fig. 4e, f), whereas they differentiated normally in cDC-ablated mice (Supplementary Fig. 9k, l). The essential role of B cells in initiating endogenous CD4⁺ T cell responses to HBsAg was also demonstrated in both the prcccDNA and high-dose AdHBV mouse models (Supplementary Fig. 11d–g). Additionally, B cells initiated CD4⁺ T cell responses under low HBsAg expression induced by hydrodynamic injection of a low dose of pHBV1.3/OVA323-339 (Supplementary Fig. 12a), or during mild inflammation in low-dose AdHBV and in AAV8-HBV infections (Supplementary Fig. 12b–e). Moreover, the lack of MHC-II on B cells (Fig. 4g), but not on phagocytes (Supplementary Fig. 9m), significantly impeded OT-II T cell proliferation, suggesting that B cells prime CD4⁺ T cell responses against recombinant HBsAg by presenting antigens on MHC-II. Finally, B cells contributed to the clearance of both circulating and intrahepatic HBsAg (Fig. 4h, i), as did CD4⁺ T cells (Supplementary Fig. 13a, b). However, injection of the mouse HBsAg-specific monoclonal antibody 129G1⁵³ resulted in the clearance of only circulating HBsAg, without affecting intrahepatic HBsAg clearance (Supplementary Fig. 13c, d). These findings suggest that B cells may not only mediate an antibody response, with the assistance of Tfh cells, to clear circulating HBsAg, but also initiate effector CD4⁺ T cell responses to clear intrahepatic HBsAg. Thus, B cells are the principal antigen-presenting cells that initiate CD4⁺ T cell responses against HBsAg.

a–d OT-II T cells, labeled with Violet tracer where indicated, were transferred into WT or μMT mice, followed by pHBV1.3/OVA323-339 HDI. Spleens, draining LNs (dLNs), and livers were harvested for FACS analysis from days 2 to 4.5 post-HDI. a Representative histogram and geoMFI data of CD69 expression at day 2 (n = 3). b Representative Violet tracer dilution and proliferation index at day 3.5 (spleen and dLN: n = 5; liver: n = 4). c Representative FACS plots and percentages of CXCR5⁺Bcl6^hi OT-II T cells at day 3.5 (n = 5). d Representative FACS plots and percentages of T-bet⁺GzmB^hi OT-II T cells at day 4.5 (n = 6). e, f Representative FACS plots and percentages of endogenous CD25⁺OX40⁺ (e, n = 7) and T-bet⁺GzmB^hi (f, n = 6) CD4⁺ T cells in the spleens, dLNs, and livers from μMT versus WT mice at day 5 post-pHBV1.3 HDI. g Representative Violet tracer dilution and proliferation index of transferred OT-II T cells in spleens, dLNs, and livers from μMT mice reconstituted with MHC-II KO versus WT B cells at day 3.5 post-pHBV1.3/OVA323-339 HDI (n = 3). h, i Circulating HBsAg levels (h, n = 4) and intrahepatic HBsAg levels (i, n = 15 view fields from 5 mice) in WT and μMT mice, with serum collected at the indicated time point and liver tissues on day 14 after pHBV1.3 HDI. Scale bars, 50 μm. At least two independent experiments were performed. Horizontal lines in (a–g, i) represent median values. Error bars in (h) show mean ± SD. Statistical significance was calculated using two-tailed Student’s t-tests (a–g, i) and two-way ANOVA (h). Significance levels are indicated as follows: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Source data and exact p-values are provided in the Source Data file.

B cells acquire nonopsonized HBsAg in situ

To study where and how HBsAg-specific B cells acquire HBsAg for initiating CD4⁺ T cell responses, we developed a knock-in mouse model, termed S129G1_HL, that expresses a monoallelic immunoglobulin (Ig) heavy/light chain derived from the 129G1 monoclonal antibody⁵³, along with a human Igκ constant region (HuCκ) (Fig. 5a). B cells in these mice predominantly expressed the engineered B cell receptor, which recognizes HBsAg (Fig. 5b). We transferred S129G1_HL B cells into congenic WT mice and blocked the entry of patrolling B cells into the white pulp (lymphoid region) of the spleen by using anti-LFA-1 and anti-α4 antibodies⁵⁴, and into LNs by using an anti-CD62L antibody, respectively (Fig. 5c). Following the blockade and 18 hours after pHBV1.3 injection, we observed that the BCR signaling intensity, as indicated by Nur77 expression, remained intact in S129G1_HL B cells from both the spleen and dLNs (Fig. 5d, e), suggesting that S129G1_HL B cells can acquire HBsAg in the white pulp of the spleen and dLNs without the need for patrolling the periphery. Likewise, splenectomy and B cell egress inhibition from LNs using FTY720 also did not alter S129G1_HL B cell activation in the liver (Fig. 5f), suggesting that S129G1_HL B cell antigen engagement in the liver is independent of the spleen and dLNs. Furthermore, 18 hours after pHBV1.3 injection, the S129G1_HL B cells had not migrated to outer B cell follicles near antigen-rich areas like the marginal or subcapsular sinus (Fig. 5g)¹⁷, indicating that S129G1_HL B cells acquire HBsAg in situ.

a Schematic illustrating the targeting strategy for generating HBsAg-specific Ig knock-in mice, named S129G1_HL. b Representative FACS plots and percentages of HuCκ⁺, HBsAg-binding cells in the blood from S129G1_HL versus WT mice, gated on B220⁺ cells (n = 3). c Schematic showing the time course of the B cell response to HBsAg. d Representative histograms and geoMFI data of Nur77 expression in S129G1_HL B cells in the spleens from WT recipients treated with both anti-LFA-1 and anti-α4 versus control antibodies, at 21 hours post-pHBV1.3 HDI (n = 3). e Representative histograms and geoMFI data of Nur77 expression in S129G1_HL B cells in the dLNs from WT recipients treated with anti-CD62L versus control antibodies, at 21 hours post-pHBV1.3 HDI (n = 3). f Representative histograms and geoMFI data of Nur77 expression in the livers from mice with both splenectomy and FTY720 versus control treatments, at 21 hours post-pHBV1.3 HDI (n = 3). g Immunohistochemistry of spleens and dLNs from WT recipients at 18 hours post-pHBV1.3/OVA323-339 HDI, showing S129G1_HL B cells (HuCκ, blue) and endogenous B cells (IgD, brown). Scale bars, 50 μm. h–k Representative histograms and geoMFI data of Nur77 expression in transferred S129G1_HL B cells in the spleens, dLNs, and livers from C3 KO (h, n = 3), FcγRIIB KO (i, n = 3), Zbtb46-DTR chimeras treated with DT ( j, n = 3), and WT mice treated with CLD liposomes (k, n = 3), compared to their respective controls, at 21 hours post-pHBV1.3 HDI. At least two independent experiments were performed. The horizontal lines represent median values. Statistical significance was calculated using two-tailed Student’s t-tests. Significance levels are indicated as ****p < 0.0001. Source data and exact p-values are available in the Source Data file.

Once antigens are passively transported to the lymphoid tissues via blood or lymph, small, soluble antigens with a radius of less than approximately 5.5 nm (about 70 kDa) can filter through the conduit system to gain direct access to lymphoid B cell follicles¹⁴, while larger antigens require active transport by accessory or resident cells after opsonization to gain access to these follicles¹⁷. Complement component 3 (C3) and Fc-gamma receptor IIB (FcγRIIB) are essential for antigen opsonization¹⁷. Supporting this model, the small soluble HEL protein (approximately 14.4 kDa) expressed via hydrodynamic injection of the pALB-Igk/HEL plasmid, induced intact BCR signaling in HEL-specific Hy10 B cells after knockout of C3 or FcγRIIB expression (Supplementary Fig. 14a, b), or depletion of cDCs or macrophages (Supplementary Fig. 14c, d). Surprisingly, HBsAg, despite being a large subviral particle far exceeding 70 kDa in size, similarly did not require the expression of C3, FcγRIIB, or the involvement of cDCs or macrophages to elicit BCR signaling in S129G1_HL B cells following pHBV1.3 injection (Fig. 5h–k). Importantly, HBsAg’s ability to bypass these factors is not due to a lack of HBsAg-specific antibodies for opsonization. Natural IgG binding to HBsAg was detectable in WT but absent in B cell-deficient μMT mice (Supplementary Fig. 14e). Furthermore, HBsAg-specific IgG was induced as early as 18 hours after hydrodynamic injection of pHBV1.3 in WT mice (Supplementary Fig. 14f). These findings indicate that HBsAg can reach lymphoid B cell follicles without undergoing the opsonization process or being constrained by the conduit system’s size limitations, thereby enabling S129G1_HL B cells to acquire HBsAg in situ, challenging established assumptions^14,17.

B cells initiate CD4⁺ T cell responses sequentially in follicular and interfollicular areas

To examine where HBsAg-specific B cells present antigen to their cognate CD4⁺ T cells, we followed the migration of S129G1_HL B cells and OT-II T cells in WT recipients by immunofluorescence microscopy after injection of pHBV1.3/OVA323-339. After 24 hours, around half of the OT-II T cells had migrated from the T cell zone to follicular regions of the spleen and dLNs (Fig. 6a, b), some of which co-localized with S129G1_HL B cells. When non-cognate VI10YEN B cells and OT-II T cells were transferred into HEL-specific BCR transgenic MD4 mice, by contrast, the T cells neither migrated into follicles (Supplementary Fig. 15a, b) nor were activated (Supplementary Fig. 15c). Thus, T cell migration into follicles appears to depend on cognate B cells. After 1.5 days, both S129G1_HL B and OT-II T cells began to accumulate in the interfollicular regions between the follicles and remained enriched in these regions on day 2.5 (Fig. 6a, b). We observed a similar migratory pathway when we used SMARTA TCR transgenic T cells, specific for LCMV glycoprotein (GP) 66-80 epitope and injected the pHBV1.3/GP66-80 replicon (Supplementary Fig. 15d, e). By contrast, consistent with our previous report⁵², in the SRBC-OVA immunization model, splenic cDCs marked by DCIR2 re-localized from marginal zone bridging channels to the outer T cell zone, and nearly the entire population of activated OT-II T cells co-localized with the activated DCIR2⁺ cDCs (Supplementary Fig. 15f, g). Together, these findings reveal that B cells present antigens to CD4⁺ T cells through a series of interactions first in the follicles and then in the interfollicular regions of the spleen and dLNs, suggesting an alternative to the traditional view that CD4⁺ T cells are primed in T cell zones by cDCs^13,21.

a Immunofluorescence of S129G1_HL B cells (HuCκ, green), OT-II T cells (CD45.1, red), and endogenous B cells (B220, blue) in the spleens and dLNs of WT recipients at the indicated time points post-pHBV1.3/OVA323-339 HDI. The boxed area indicates regions shown at higher magnification. Scale bars, 50 μm. b Schematic illustration of the spatial organization of B and T cell zones and their interfaces in spleen and dLN (top), and the relative proportions of S129G1_HL B cells (bottom, left) and OT-II T cells (bottom, right) localized in each anatomical region at the indicated time points. (middle: n = 6 white pulp areas from 3 mice; bottom: n = 6 LN slices from 6 mice). The data represent the relative proportions of each group. Three independent experiments were performed. Source data are available in the Source Data file.

CCR5 and EBI2 guide the follicular and interfollicular priming of CD4⁺ T cells by B cells, respectively

To identify the cues that might guide B cell and T cell migration and mediate their interactions, we conducted RNA-seq analysis on the transferred S129G1_HL B cells in the spleen 18 hours after pHBV1.3/OVA323-339 injection. This revealed that the genes encoding CCL3 and CCL4 were most significantly induced (Fig. 7a and Supplementary Data 4), both of which are ligands for the chemokine receptor CCR5 on naive CD4⁺ T cells⁵⁵. These were the only upregulated chemokine genes in S129G1_HL B cells (Fig. 7b). Treatment of the mice with maraviroc, a commonly used CCR5 antagonist^56,57, halted OT-II T cell migration towards S129G1_HL B cells in follicles and hindered their activation after injection of pHBV1.3/OVA323-339 (Fig. 7c–e). Our previous work identified a key role for the oxysterol receptor EBI2 and its ligand, 7α,25-dihydroxycholesterol, in positioning activated T cells and cDC2 cells in the outer T cell zone⁵². Similarly, both EBI2 KO S129G1_HL B cells and EBI2 KO OT-II T cells failed to concentrate in interfollicular regions, instead localizing to the central T-B border (Fig. 7f, g) or the T cell zone (Fig. 7h, i), respectively. EBI2-deficient OT-II T cells proliferated less rapidly than WT OT-II T cells (Fig. 7j), which might be due to impaired interactions with their cognate WT B cells in the interfollicular regions. In summary, these findings establish that B cells present the antigen to CD4⁺ T cells first in follicular areas guided by the chemokine receptor CCR5, likely binding its ligands CCL3 and CCL4, and subsequently in interfollicular areas, guided by the oxysterol receptor EBI2 and its ligand 7α,25-dihydroxycholesterol.

a Volcano plot of RNA-seq showing differentially expressed genes in sorted S129G1_HL B cells at 18 versus 0 hours post-pHBV1.3 HDI. b RNA-seq heat map of chemokine gene expression in S129G1_HL B cells at 0 and 18 hours post-pHBV1.3 HDI. c Immunofluorescence of S129G1_HL B cells (HuCκ, green), OT-II T cells (CD45.1, red), and endogenous B cells (B220, blue) in the spleens and dLNs of WT recipients treated with Maraviroc versus DMSO at day 1 post-pHBV1.3/OVA323-339 HDI. d Percentages of OT-II T cells within B cell follicles in (c) (n = 40 follicles from 3 mice). e Representative histograms and percentages of CD69⁺ OT-II T cells in spleens, dLNs, and livers from (c) (n = 7). f Immunohistochemistry of spleens and dLNs from WT recipients at day 2.5 post-pHBV1.3/OVA323-339 HDI, showing WT or EBI2 KO S129G1_HL B cells (HuCκ, blue) and endogenous B cells (IgD, brown). g Quantification of WT and EBI2 KO S129G1_HL B cells in (f) (top: n = 6 white pulp areas from 3 mice; bottom: n = 6 LN slices from 6 mice). h Immunohistochemistry of spleens and dLNs from WT recipients at day 2.5 post-pHBV1.3/OVA323-339 HDI, showing WT or EBI2 KO OT-II T cells (CD45.2, blue) and endogenous B cells (IgD, brown). i Quantification of WT and EBI2 KO OT-II T cells in (h) (top: n = 6 white pulp areas from 3 mice; bottom: n = 6 LN slices from 6 mice). j Representative FACS plots and proliferation index of EBI2 KO versus WT OT-II T cells in spleens, dLNs, and livers from WT recipients at day 3.5 post-pHBV1.3/OVA323-339 HDI (n = 3). At least two independent experiments were performed. Horizontal lines of (d, e–j) represent median values. Data in g, i represent the relative proportions of each group. Statistical significance for d, e, j was calculated using two-tailed Student’s t-tests. Significance levels are indicated as: *p < 0.05, ***p < 0.001, ****p < 0.0001. Scale bars (c, f, h), 50 μm. Source data and exact p-values are available in the Source Data file.

Deciphering the cellular crosstalk landscape associated with T cell priming

The above results demonstrated that cDC1s and B cells drove CD8⁺ T and CD4⁺ T cell responses against HBsAg, respectively. We next set out to identify the signals provided by antigen-presenting cells to initiate T cell responses in addition to recognition of MHC-peptide complex by TCR. To do so, we used the NicheNet algorithm to prioritize ligand-receptor interactions between interacting antigen-presenting cells-T cells based on their bulk or single-cell transcriptome data. We first compared the single-cell transcriptome of CD11c^highMHCII⁺ DCs, hepatocytes and OT-I T cells in liver between day 0 and day 2.5 after the hydrodynamic injection of pHBV1.3/OVA257-264 (Supplementary Fig. 3a). Then, we built up a cDC1-proliferating CD8⁺ T cell ligand-receptor interaction map with NicheNet algorithm (Fig. 8a). We blocked those potential integrins and co-stimulator crosstalk pathways including CD86-CD28 and ALCAM-CD6 in in vitro co-culture system of cDC1s and CD8⁺ T cells (Fig. 8b). Intriguingly, CD8⁺ T cell proliferation induced by mature cDC1s was not affected by the CTLA4-Ig fusion protein (Fig. 8c), suggesting a weaker role for CD28 co-stimulation in priming CD8⁺ T cells than CD4⁺ T cells as has been previously reported^58,59. Consistent with previous observations that CD6 was a T cell co-stimulatory receptor^60,61, the presence of the recombinant soluble CD6-Fc molecules greatly inhibited CD8⁺ T cell proliferation (Fig. 8c). Notably, ALCAM mRNA was highly enriched in cDC1s, and moderately expressed in cDC2s and virus-infected hepatocytes, but not in other DC subsets, Kupffer cells and LSECs (Fig. 8a). Moreover, cDC1s constitutively expressed ALCAM protein at homeostasis (Supplementary Fig. 16a), with expression levels unaffected by pHBV1.3 hydrodynamic injection (Supplementary Fig. 16b). While LSECs also constitutively expressed ALCAM (Supplementary Fig. 16c), they co-expressed the immuno-inhibitory checkpoint PD-L1 (Supplementary Fig. 16d), consistent with previous reports^62,63, and their ALCAM expression decreased after pHBV1.3 hydrodynamic injection (Supplementary Fig. 16e). However, hepatocytes did not express ALCAM under either condition (Supplementary Fig. 16f, g). These findings may partially explain why only cDC1s were capable of priming CD8⁺ T cell responses. Additionally, scRNA-seq data from liver biopsy of a patient with acute HBV infection revealed the presence of ALCAM mRNA in cDC1s (Supplementary Fig. 16h), suggesting a potential role for ALCAM in co-stimulating CD8⁺ T cell responses by human hepatic cDC1s. Together, these results demonstrated a CD6-dependent but CD28-independent co-stimulation mechanism for CD8⁺ T cell priming uniquely mediated by cDC1s.

a Ligand activities (left, brown), potential receptors (middle, red), and predicted target genes (right, blue) identified by NicheNet analysis based on single-cell transcriptome data described in Supplementary Fig. 3a. Dot plots showing the average expression levels of ligands, receptors, and target genes are also presented. b Schematic showing the cell co-culture strategy. c Representative Violet tracer dilution and proliferation index of OT-I T cells after 3 days of culture alone or co-culture with liver-sorted cDC1s from WT mice hydrodynamically injected with pHBV1.3/OVA257-264 for 24 hours, in the presence of indicated proteins (n = 3 biologically independent replicates). Three independent experiments were performed. Statistical significance was calculated using one-way ANOVA with Tukey’s multiple comparisons test for c. Error bars represent mean ± SD, with significance levels indicated as: *p < 0.05. Source data and exact p-values are available in the Source Data file.

Next, we sorted the S129G1_HL B and OT-II T cells interacting at follicular and interfollicular regions of spleens, corresponding to days 1 and 2.5 time points (Fig. 6a), respectively, and performed RNA-seq analysis (Supplementary Fig. 17a). According to the expression of ligands, receptors, associated pathway components and genomic targets of these pathways, we identified the top predicted the ligand-receptor pairs between B cells and CD4⁺ T cells (Fig. 9a and Supplementary Fig. 17b). Nearly half of these ligand-receptor pairs were shared by cells interacting at follicular and interfollicular regions (Supplementary Fig. 17c). Furthermore, we focused on blocking those shared integrin and co-stimulatory signals individually by using commercially available neutralizing antibodies or proteins in mice, considering their well-appreciated roles in T cell priming⁶⁴. The proliferation and survival of CD4⁺ T cells were strongly inhibited by an anti-LFA-1 (encoded by Itgb2) blocking antibody (Fig. 9b, c). The differentiated CD4⁺ T cells showed a compensatory increase in frequency but still decreased in total number (Fig. 9d). The competitive blockade of CD28-CD80 costimulatory pathway by a CTLA4-Ig fusion protein potently suppressed CD4⁺ T cell proliferation (Fig. 9e) and subsequently decreased the total number of differentiated T cells without any significant effect on survival and differentiation itself (Fig. 9f, g). However, the CD4⁺ T cell priming remained intact in mice treated with a widely used anti-ICOS antibody to block ICOSL engagement of ICOS (Supplementary Fig. 17d–f), which was believed to be an important co-stimulatory molecule for optimal CD4⁺ T cell responses⁶⁵. Additionally, we sorted HBsAg-specific B cells from PBMCs of four patients and performed scRNA-seq analysis, revealing increased expression of Itgb2 (encoding LFA-1) and CD80 mRNA in activated B cells, along with upregulation of HLA-II genes (Supplementary Fig. 17g), suggesting a potential role for these molecules in co-stimulating CD4⁺ T cell responses by B cells in humans. These results indicated an LFA-1- and CD80-dependent but ICOSL-independent priming of CD4⁺ T cells by B cells. In sum, we mapped the ligand-receptor signaling landscape between antigen-presenting cells and T cells, identifying fundamental co-stimulatory signals required for optimal T cell responses to HBsAg.

a Ligand activities (left, brown), potential receptors (middle, red), and predicted target genes (right, blue) identified by NicheNet analysis based on the key signature genes of transferred OT-II T cells in spleens from WT mice, at day 2.5 post-HDI of pHBV1.3/OVA323-339. Heatmaps showing the absolute expression levels of ligands, receptors, and target genes are also shown. b Representative Violet tracer dilution and proliferation index of OT-II T cells at day 3.5 in the spleens, dLNs, and livers from WT mice treated with anti-Itgb2/LFA1 antibody versus control antibodies, at day 3.5 post-HDI of pHBV1.3/OVA323-339 (n = 6). c Percentages of dead cells within transferred OT-II T cells in spleens, dLNs, and livers from the mice described in (b) using the viability dye eFluor 780 (n = 6). d Percentages and number of CXCR5⁺Bcl6^hi cells within transferred OT-II T cells in spleens, dLNs, and livers from the mice described in (b, n = 6). e Same as (b), but mice were treated with control or CTLA4-Ig protein (n = 3). f Same as c, but mice were treated with control or CTLA4-Ig protein (n = 3). g Same as (d), but mice were treated with control or CTLA4-Ig protein (n = 3). At least two independent experiments were performed. Horizontal lines indicate median values. Statistical significance was calculated using two-tailed Student’s t-tests. Significance levels are indicated as: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Source data and exact p-values are available in the Source Data file.

Together, these findings elucidate how HBsAg initiates T cell responses in HBV infection, revealing novel intercellular antigen presentation mechanisms distinct from classical viral paradigms (Supplementary Fig. 18).