HBsAg-specific CD8+ T cells as an indispensable trigger to induce murine hepatocellular carcinoma

Hepatitis B virus (HBV)-associated hepatocellular carcinoma (HCC) is mediated by an inappropriate attack by HBV-specific T cells in patients. However, this immunopathogenic process has not been clarified because of the lack of a suitable animal model. Here, we used immunocompetent Fah−/− mice as the recipients in the adoptive transfer of HBsAg+ hepatocytes from HBs-Tg mice to replace the recipient hepatocytes (HBs-HepR). HBs-HepR mice exhibited persistent HBsAg expression with chronic hepatitis and eventually developed HCC with a prevalence of 100%. HBsAg-specific CD8+ T cells were generated and specifically and continuously induced hepatocyte apoptosis with progressive chronic inflammation, and the depletion of CD8+ T cells or their deficiency prevented HCC, which could then be reproduced by the transfer of HBsAg-specific CD8+ T cells. In summary, our results demonstrated that CD8+ T cells plays a critical role in HBsAg-driven inflammtion and HCC tumorigenesis.


INTRODUCTION
Liver cancer is the fourth leading cause of cancer-related death worldwide and led to an estimated 782,000 deaths in 2018. 1 Hepatocellular carcinoma (HCC) accounts for 75-85% of liver cancer cases, among which over 50% are attributed to hepatitis B virus (HBV) infection. 2,3 The underlying mechanisms in hepatocyte transformation include direct gene activation and transactivation by HBV DNA integration 4 or HBx protein regulatory activity, 5 and indirect hepatocyte damage and regeneration caused by immunity against HBV. [6][7][8][9] The humoral immune response with anti-HBV surface antigen (anti-HBs) antibody production is thought to be protective during HBV infection. 10 Furthermore, HBV-specific CD8 + T cells exhibit antiviral activity either indirectly by producing cytokines, such as interferon-γ (IFN-γ) and tumor necrosis factor-α (TNF-α), 11,12 or directly by killing infected hepatocytes. 7,8,13 Clinical observation has shown that a vigorous HBV-specific CD8 + T cell response results in HBV clearance in acute infected patients 10 and that an inappropriate cytotoxic T lymphocyte (CTL) response mediates cumulative hepatocyte damage sufficiently strong to induce chronic inflammation and ultimately predispose chronic HBV-infected (CHB) patients to HCC development. [14][15][16][17][18] HBVspecific CD8 + T cell functions seem to be decreased in CHB patients with low frequency and high levels of inhibitory molecules, [19][20][21] which are typical features of chronic inflammation. However, recent elaborate experiments detected numerous HBVspecific T cell populations in the circulation that were associated with long-term memory and polyfunctionality in CHB patients. [22][23][24][25] In addition, even the residual antigen-specific CD8 + T cell response appears to be related to HCC development in chronic immune-tolerant HBV patients with a high rate of HBV replication. 15,26,27 Importantly, HBV surface antigen (HBsAg)specific T cells, which are absent from the circulation, could also be detected after in vitro expansion. 24 The presence of HBVspecific CD8 + T cells resulted in continuous hepatocyte injury and regeneration, which promoted inflammatory mutagens and HCC initiation. 8,28,29 However, the precise immune mechanisms of CD8 + T cells in HBV-related HCC have not been studied in depth due to the absence of a suitable mouse model.
Due to host limitation, HBV cannot naturally infect mouse hepatocytes, which is the major obstacle in HBV-host interaction studies on topics including viral infectious processes and immunological responses. HBV transgenic (Tg) mouse models are widely used to study the anti-HBV immune response, but these models exhibit central tolerance to HBV, in which an HBV-specific immune response cannot be or is not fully induced. 30,31 Studies on adoptive transfer have provided evidence that the presence of HBV-specific T cells can result in acute hepatitis through different mechanisms; however, the direct adoptive transfer of HBV-specific effector T cells into HBV-Tg mice triggered T cell proliferation but not functional activation and therefore did not induce chronic hepatitis or HCC; 32 thus, adoptive transfer studies have their own defects. Nakamoto, Y. et al. improved the HBV-Tg mouse model via thymectomy, bone marrow reconstruction, and the adoptive transfer of HBsAg-specific CD8 + T cells from HBsAg-immunized mouse spleens. They observed that immune cells constantly attacked HBsAg-expressing hepatocytes and that the mice eventually developed HCC. 9,33 Though this model is the first to provide strong evidence that anti-HBV adaptive immunity induces HCC, reconstituted anti-HBV immunity in HBV-Tg mice involves the transfer of an excess amount of CD8 + T cells, which usually cannot be detected in CHB patients, and the manipulation process is relatively technically complicated. Hepatocyte-specific, plasmiddelivered HBV mouse models were created to study anti-HBV immune responses. However, these models exhibited only acute liver inflammation with HBV elimination or did not (i) induce liver inflammation (similar to healthy HBV carriers) or (ii) sustain HBV long enough for HCC development. [34][35][36] Several HCC mouse models (e.g., genetically engineered, chemical carcinogeninduced, and implantation models) 37 have been developed, but these models are not suitable for immunological studies of HBVrelated HCC.
In this study, we generated an HBV mouse model by transferring HBsAg + hepatocytes from HBs-Tg mice into immunocompetent recipient mice with the same genetic background. In this mouse model, HBsAg-specific CD8 + T cells were generated naturally and mediated chronic hepatitis and eventual HCC. Our single-cell RNA sequence analysis and other dynamic observations revealed that effector CD8 + T cells were generated 18 weeks post-transfer and that effector CD8 + T cells played a critical role in chronic hepatitis and HCC by specifically and continuously inducing hepatocyte apoptosis. HBV-related HCC could be prevented with the deficiency or depletion of CD8 + T cells but was reproduced by transferring HBsAg-specific CD8 + T cells.

RESULTS
A mouse model of HBV-related HCC To study naturally occurring anti-HBV adaptive immunity and immunopathology, we generated a mouse model by transferring hepatocytes via the spleen (Fig. 1a). Hepatocytes from C57BL/6J or HBs-Tg mice were reconstituted successfully in recipient immunocompetent Fah −/− mice (which have been reported to be "ideal" hepatocyte recipients after the withdrawal of 2-(2-nitro-4trifluoromethylbenzoyl)-cyclohexane-1,3-dione (NTBC) from drinking water 38 ) and were named "B6-HepR" and "HBs-HepR" mice, respectively. Beginning 14 weeks after the transfer of HBsAg + hepatocytes, chronic hepatitis was observed in HBs-HepR mice, as shown by the increased serum levels of ALT and AST (Fig. 1b, c). 9 months after hepatocyte reconstruction, we confirmed that there was no significant difference in the fah-positive area in the livers of B6-HepR and HBs-HepR mice; almost all areas were fahpositive (Fig. S1A). HBsAg expression could be detected in fahpositive hepatocytes from HBs-HepR mice but not in those from control mice, and the proportion of HBsAg + hepatocytes in the liver reached~60% (Fig. S1B). In addition, lymphocyte infiltration and inflammatory foci were observed in the livers of HBs-HepR mice by pathological studies (Fig. 1d). Sirius Red staining showed obvious collagen deposition at 9 months after hepatocyte transfer into the livers of HBs-HepR mice, indicating the development of chronic hepatitis and spontaneous liver fibrosis (Fig. 1d). Consequently, tumor nodules were observed in the livers of HBs-HepR mice but not in those of B6-HepR mice (Fig. 1e), and tumorigenesis was observed with a prevalence of 6/6 at 9 months posttransfer (Fig. 1e). Therefore, these results indicated that we had created an HBV-associated HCC mouse model.
Humoral immune response could not be triggered in HBs-HepR mice To determine the major cause of hepatitis, we determined whether anti-HBs antibody could be triggered. After 10 weeks of hepatocyte reconstitution, we immunized the mice with 1 μg of an HBsAg vaccine twice every other week. As shown in Fig. S2A, 21 days after vaccination, serum HBsAg was detected in HBs-HepR mice regardless of their vaccination status; in contrast, anti-HBs antibodies could not be detected (Fig. S2B). However, in HBsAgvaccinated B6-HepR mice, anti-HBs antibodies could be induced (Fig. S2B). We extended our observation until 35 days after vaccination and found that the serum levels of HBsAg in HBs-HepR mice remained high (Fig. S2C) and that anti-HBs antibodies could not be generated, indicating that the HBsAg-specific humoral immune response cannot be triggered in HBs-HepR mice, similar to what has been observed in human CHB.
Generation of HBsAg-specific CD8 + T cells in HBs-HepR mice As hepatocyte injury, shown by increased serum levels of ALT, was induced in HBs-HepR mice (Fig. 1b), single-cell RNA analysis was undertaken to examine intrahepatic mononuclear cells (MNCs) in HBs-HepR mice 12 and 18 weeks post-transfer (Fig. 2a). Nine cell populations were identified depending on their expression of specific cell markers (Figs. 2b and S3A, B). Three stable candidate cell populations that appeared at 12 weeks but were dramatically increased in number at 18 weeks were selected for further analysis: effector CD8 + T cells, NK cells and γδT cells (Fig. 2c, d).
HBsAg-specific CD8 + T cells mediated hepatocyte apoptosis in HBs-HepR mice Next, intrahepatic CD8 + T cells were gated to identify their features by flow cytometry (Fig. S4). The livers of HBs-HepR mice had a higher percentage and increased number of CD8 + T cells 18 weeks post-transfer compared to control mice that did not receive transferred hepatocytes (Fig. 4a, b). Compared to control mice, more activated effector cells with CD62L lo CD44 hi expression and fewer naïve cells with CD62L hi CD44 lo were present in HBs-HepR mice (Fig. 4c, d), suggesting that CD8 + T cells were functionally activated. To evaluate HBV-specific T cell responses, we analyzed CD11a and CD49d expressions on CD8 + T cells and found more antigen-experienced CD8 + T cells with CD11a hi CD49d hi expression compared to those in control mice (Fig. 4e, f). Similar results were observed in the spleens of HBs-HepR mice (Fig. S5). Importantly, HBsAg-specific CD8 + T cells recognizing the HBsAg 190-197 VWLSVIWM peptide were detected in the livers of HBs-HepR mice (Fig. 4g, h). The presence of CD8 + T cells in the livers of HBs-HepR mice was further confirmed by histological immunofluorescence staining. There were more HBsAg + hepatocyte-adjacent CD8 + cells 18 weeks post-transfer in HBs-HepR mice than in control mice (Fig. 4I, j). To determine the functions of antigen-specific CD8 + T cells, we performed a TUNEL assay in hepatocytes from HBs-HepR mice. More TUNEL + hepatocytes were observed in HBs-HepR mice than in B6-HepR mice (Fig. 4k, l). Collectively, these results indicated that HBsAg-HBsAg-specific CD8 + T cells as an. . . X Hao et al. specific CD8 + T cells were generated and mediated hepatocyte apoptosis in HBs-HepR mice.
Depletion of CD8 + T cells prevented HBV-related HCC development in HBs-HepR mice To further confirm their role in HCC development, CD8 + T cells were depleted from HBs-HepR mice by treatment with anti-CD8 monoclonal antibody (Fig. 5a). A total of 125 μg of anti-CD8 or rat IgG antibody was injected weekly into HBs-HepR mice. Anti-CD8 antibodies led to the complete depletion of CD8 + T cells in the livers of mice in the treatment group (Fig. 5b). The number of tumor nodules and prevalence of tumorigenesis were significantly decreased in CD8 + T cell-depleted HBs-HepR mice at 9 months post-transfer (Fig. 5c-e). To exclude the role of CD4 + T cells and NK cells in HCC development, we also depleted CD4 + T cells and NK cells by the injection of their specific antibodies in our mouse model (Fig. S6A). Antibodies or rat IgG were injected into B6-HepR or HBs-HepR mice weekly, and 9 months after hepatocyte reconstruction, CD4 + T cells and NK cells were also barely detectable in the livers of mice in the indicated antibody from C57BL6/J mice or HBsAg transgenic (HBs-Tg) mice via spleen injection, and NTBC was withdrawn from the drinking water immediately. Then, 3.5 weeks after hepatocyte transfer, NTBC was added to the drinking water for 3 days and then withdrawn. Nine months after reconstruction, HCC was generated in HBs-HepR mice. b, c Serum levels of ALT b and AST c were measured at the indicated time points after hepatocyte transfer (B6-HepR, n = 11; HBs-HepR, n = 13). d Nine months after hepatocyte transfer, liver samples were harvested, and 5-μm paraffin-embedded liver sections were used for pathologic analysis by hematoxylin and eosin (H&E) staining (top panel) and Sirius Red staining (bottom panel), scale bar = 50 μm. e Representative images of the liver morphology showed tumor nodules and the presence of tumorigenesis at 9 months after hepatocyte transfer. Two-way ANOVA and Fisher's exact test were used to compare experimental groups, and the data are shown as the mean ± SEM.
HBsAg-specific CD8 + T cells as an. . . X Hao et al.
treatment groups (Fig. S6B). The depletion of CD4 + T cells or NK cells did not affect HCC development in HBs-HepR mice (Fig. S6C).
Overall, these results demonstrated that HCC development in HBs-HepR mice is caused mainly by CD8 + T cells.
CD8 + T cell deficiency prevented HCC, but the adaptive transfer of CD8 + T cells induced HCC Four of the 9 antibody-treated mice still developed HCC 9 months after hepatocyte reconstruction (Fig. 5e). We suspected that a small number of CD8 + T cells were still present in the antibodydepleted mice due to leakage. To further confirm the role of CD8 + T cells in HCC development, we bred Fah −/− mice with CD8 −/− mice to obtain Fah −/− CD8 −/− double knockout mice, which resulted in CD8 + T cell deficiency at the genetic level. We transferred HBsAg + hepatocytes or control hepatocytes into Fah −/− CD8 −/− recipient mice (Fig. 6a). Nine months after the transfer, CD8 + T cells were still completely undetectable in the livers of HBs-HepR-CD8KO and B6-HepR-CD8KO mice (Fig. 6b, c), and tumor nodules were not observed (Fig. 6d, e). However, HCC recurred (with many more tumor nodules) upon the transfer of splenic HBsAg-specific CD8 + T cells isolated from HBsAgvaccinated wild-type mice 10 weeks after the transfer of HBsAg + hepatocytes (Fig. 6d, e). These transferred HBsAg-primed CD8 + T cells were still present in the livers of HBs-HepR-CD8 KO mice HBsAg-specific CD8 + T cells as an. . . X Hao et al. 9 months after hepatocyte transfer (Fig. 6b, c). These data further confirm that adaptive CD8 + T cells played a key role in HCC development in HBs-HepR mice.

DISCUSSION
Efficacious vaccines against HBV will greatly reduce the risk of new HBV infection. However, methods to reduce the high risk of CHB (which affects 260 million people worldwide) development into HCC are lacking. [39][40][41] Hence, we found that HBsAg-specific CTLs are the major player in inducing HCC, helping to design a strategy targeting these adaptive immune cells. Previous works have reported the single-cell transcriptional profiles of hepatocytes, endothelial cells, and KCs from the mouse liver, 42 but have ignored other immune cells in the liver. We applied single-cell transcriptional sequencing technology to identify the type of mononuclear cell responsible for hepatitis and HCC in HBs-HepR mice. According to their expression of marker genes, we identified 9 clusters of immune cells in the mouse liver (Fig. S2A). By comparing the single-cell transcriptional profiles of these clusters at different time points, we determined that CD8 + T cells play a key role in the development of hepatitis and might lead to HCC development (Fig. 2). The percentage of effector CD8 + T cells was significantly higher at 18 weeks post-transfer (Fig. 2b), and the serum level began to increase at that time (Fig. 1b). Functional trajectory analysis of CD8 + T cells revealed more functional activated CD8 + T cells at 18 weeks posttransfer than before transfer (Fig. 3), suggesting the important roles of CD8 + T cells in the HBs-HepR mouse model.
We further confirmed these single-cell transcriptional profiles by flow cytometry and immunofluorescence analysis (Fig. 4). There were more HBsAg-specific CD8 + T cells and apoptotic hepatocytes in the livers of HBs-HepR mice than in the livers of control mice. Importantly, when CD8 + T cells, but not CD4 + T cells or NK cells, were depleted and in mice genetically deficient in CD8 + T cells, the tumor incidence decreased significantly (Figs. 5 and 6 and S5). These results suggest that HBsAg-specific CD8 + T cells are sufficient to induce HCC development.
HBV-specific CD8 + T cells may kill HBV-expressing hepatocytes directly or indirectly by secreting cytotoxic cytokines. 7,32,43,44 A strong correlation was noted between the HBV-specific CD8 + T cell response and HCC development in patients with CHB or HCC. 26 However, there are relatively few data demonstrating that anti-HBV CTLs directly induce HCC, which greatly limits study of the HBV-HCC axis. In 1998, Nakamoto, Y. et al. were the first team to reconstitute donor HBsAg-specific CD8 + T cells in recipient HBs-Tg mice via thymectomy, bone marrow reconstruction, and the adoptive transfer of splenocytes from HBsAg-vaccinated mice, which showed that HBsAg-specific CTLs mediated HCC. 9,33 Recently, we illustrated that the breakdown of adaptive immune tolerance by the TIGIT blockade combined with HBsAg vaccination might recover the anti-HBV function of autologous HBsAg-specific CTLs and induce HCC in HBs-Tg mice, 45 indicating that the central tolerance to HBV in HBs-Tg mice is reversible. The two HBV-related HCC mouse models developed from HBs-Tg mice mentioned above are based on the reconstitution of donor immune-primed or autologous tolerance-recovered Ag-specific CTLs, so they still do not completely mimic naturally occurring anti-HBV immunity and immunopathology. In the HBs-HepR mouse model, the recipient Fah −/− mice readily harbored donor HBsAg + Fig. 4 Generation of HBsAg-specific CD8 + T cells and hepatocyte apoptosis in HBs-HepR mice. Eighteen weeks after hepatocyte transfer, liver samples were harvested from HBs-HepR and Fah −/− mice. Intrahepatic mononuclear cells were isolated for flow cytometry, and liver tissue was isolated for immunofluorescence analysis. a Representative plots show CD4 vs. CD8 expression on T cells. Numbers next to the gate line show the percentages of the indicated cells. b The percentages and absolute numbers of CD8 + T cells in a. c Representative plots show the expression of CD62L and CD44 on CD8 + T cells. d Percentage analysis of the data in c. e Representative plots show CD49d and CD11a expression on CD8 + T cells. f Percentages and absolute numbers of CD11a hi CD49d hi CD8 + T cells in e. g Representative plots show the percentage of HBsAgtetramer + CD8 + T cells. h Percentages and absolute/numbers of HBsAg-tetramer + CD8 + T cells in g (n = 5 mice per group). i Ten-micrometerthick sections of frozen liver samples were used for immunofluorescence analyses. Representative images of anti-HBsAg (green), anti-mouse-CD8 (red) and 4′,6-diamidino-2-phenylindole (DAPI; blue) triple-staining show the location and number of CD8 + cells in liver tissue. j Statistical analyses of CD8 + cell numbers in i (n = 9 random areas from three mice per group). k TUNEL staining of liver tissue sections from mice 18 weeks after hepatocyte reconstruction. Representative images of TUNEL (pink) and DAPI (blue) double staining show apoptotic cells in the liver. l Statistical analyses of the numbers of TUNEL-positive cells in k (n = 6 random areas from three mice per group). Scale bar = 50 μm. A two-tailed unpaired t-test was used to compare the experimental groups, and the data are presented as the mean ± SEM.
hepatocytes and were immunocompetent without showing central tolerance to HBV. These features characterize this interesting platform in which to produce, observe and investigate the dynamics of HBsAg-specific CD8 + T cells in HBV-related HCC. Hepatocyte transfer Mouse hepatocytes were isolated as described previously. 46 Briefly, the liver was perfused with a solution of EGTA and digested with a 0.075% collagenase solution. Isolated mouse hepatocytes were resuspended in serum-free Dulbecco's modified Eagle's medium (DMEM). Hepatocytes (1.0 × 10 6 ) from C57BL/6J mice or HBs-Tg mice were transferred into Fah −/− mice via intrasplenic injection, and NTBC-treated water was withdrawn immediately. 25 days after hepatocyte reconstruction, NTBC was added to the drinking water for 3 days. 12 weeks after hepatocyte reconstruction, the mouse model was deemed to be successful. Body weight was measured every week.

MATERIALS AND METHODS
HBsAg vaccination Mice were immunized with 1 μg of HBsAg vaccine (HBsAg/alum, Kangtai, China) per mouse in a total volume of 50 μL twice every other week.
Immunohistochemistry and immunofluorescence analysis Liver samples were fixed in 10% neutral-buffered formalin and embedded in paraffin. Sections with a thickness of 5 μm were stained with hematoxylin and eosin using standard methods. Collagen analysis was performed by Sirius Red staining. Slides were viewed with a microscope (Imager A2; Zeiss, Wetzlar, Germany). Immunofluorescence analyses were carried out using liver tissue sections with a thickness of 10 μm. Anti-HBsAg (ab68518; Abcam) and anti-CD8 (ab217344; Abcam) antibodies were used to stain the indicated cells. Then, fluorescent secondary antibodies from the corresponding species were applied for analyses. A Click-iT™ Plus TUNEL Assay kit for in situ apoptosis detection using Alexa Fluor™ 647 dye (C10619; Thermo Scientific, Waltham, MA, USA) was employed according to the manufacturer's instructions. Fluorescent slides were viewed with a confocal microscope (LSM880; Zeiss).
Serum transaminase activity assays Serum ALT (lot#360100) and AST (lot#360200) activities were measured by standard photometric methods using a kit (Shanghai Rongsheng Biotech, China). Cell preparation Liver MNCs were isolated from experimental mice by forcing the dissected liver through 200-G stainless steel mesh. Then, MNCs were isolated by gradient centrifugation with 40% and 70% Percoll™ solutions. Splenocytes were harvested by forcing dissected splenic tissue through stainless steel mesh and the subsequent lysing of erythrocytes (RBC lysis buffer; 420301, Biolegend, San Diego, CA, USA).
Cell depletion PK136 (anti-NK1.1; American Type Culture Collection (ATCC), Manassas, VA, USA), TIB-207 (anti-CD4, ATCC), and TIB-210 (anti-CD8, ATCC) hybridoma cells were used to obtain the indicated monoclonal antibodies used to deplete NK, CD4 + T, and CD8 + T cells, respectively, in vivo. Mice were injected (125 μg, i.p.) with monoclonal antibodies against NK1.1, CD4, or CD8 1 week Fig. 6 Transfer of HBV-immunized CD8 + T cells leads to HCC. a Workflow for the transfer of CD8 + T cells (schematic). Briefly, 1 × 10 6 hepatocytes were transferred from C57BL6/J or HBs-Tg mice to Fah-CD8 double knockout mice, as described in Fig. 1a. Ten weeks after hepatocyte transfer, HBs-HepR mice were sublethally irradiated with 6.5 Gy and then received 5 × 10 6 HBsAg-primed CD8 + T cells via intravenous injection the next day. All mice were harvested 9 months after hepatocyte transfer. b The successful transfer of CD8 + T cells into Fah-CD8 double knockout mice was confirmed by flow cytometry analysis of liver cells. c Statistical analyses of the percentages of CD8 + T cells in each group are shown. d Representative images of the liver morphology showing tumor nodules (rat IgG, n = 13 mice; anti-CD8, n = 9 mice). e The numbers of tumor nodules in d were calculated (n = 4-13 mice per group). A two-tailed unpaired t-test and Fisher's exact test were used to compare experimental groups; the data are presented as the mean ± SEM. before intrasplenic hepatocyte transfer once a week for 9 months.
Detection of serum HBsAg and anti-HBs antibodies Serum HBsAg was measured using enzyme-linked immunosorbent assay kits according to the manufacturer's instructions (Zhongshan Bio-Tech, Shanghai, China). Anti-HBs levels were measured with the corresponding radioimmunoassay kits according to the manufacturers' instructions (North Institute of Biological Technology, Beijing, China).

Statistical analysis
The significance of differences between two independent groups was determined by Student's t-test. One-way analysis of variance (ANOVA) was applied to compare three conditions. For timeline analysis, two-way ANOVA was carried out. Fisher's exact test was applied to assess the prevalence of HCC. P < 0.05 indicated a significant difference.
Study approval All mice were maintained under specific pathogen-free conditions according to the guidelines for experimental animals set by the University of Science and Technology of China.

DATA AVAILABILITY
Raw sequencing data have been deposited in GEO Datasets with the GEO accession number GSE130880.