Serum HBsAg clearance has minimal impact on CD8⁺ T cell responses in mouse models of HBV infection

Hepatitis B virus (HBV) is a noncytopathic, hepatotropic virus containing a 3.2-kb partially double-stranded DNA genome packaged together with the polymerase in viral nucleocapsids that are wrapped by envelope proteins (Guidotti and Chisari, 2006). All three in-frame envelope proteins, the large, medium, and small proteins, include the immunodominant determinant region that is detected by the diagnostic tests for hepatitis B surface antigen (HBsAg; Liaw, 2019). Besides infectious virions, HBsAg is found in the circulation in the form of noninfectious subviral spheres and filaments that are present in a 10²–10⁵-fold excess over virions, reaching exceptional quantities of up to >300 µg/ml (Kim and Tilles, 1973). The reason for evolving this extraordinary level of biosynthetic effort is not well understood, but absorption of circulating anti-HBsAg neutralizing antibodies (Abs) that would otherwise limit the spread of infectious virions within the liver is a likely explanation. The high level of circulating HBsAg has long been thought to reflect the robust transcriptional activity of an episomal, replication-competent template known as covalently closed circular DNA (Seeger and Mason, 2015). However, a recent study in chronically infected patients and chimpanzees treated with an RNA interference–based antiviral strategy has suggested that most circulating HBsAg actually derives from subgenomic, replication-incompetent HBV DNA integrants (Wooddell et al., 2017). Whatever the origin, detectable HBsAg is the serological hallmark of persistent HBV infection and, accordingly, sustained serum HBsAg loss and detection of anti-HBsAg Abs (HBsAb; seroconversion) is a widely accepted marker of therapeutic success (Fanning et al., 2019). Circulating HBsAg is thought not only to have a negative impact on antigen-specific B cell responses, but also to delete or functionally impair antigen-specific T cells (Zhu et al., 2016). This notion is supported by the observation that HBsAg seroconversion induced by nucleos(t)ide analogues is associated with an increase in the quality and vigor of HBV-specific T cell responses (Boni et al., 2012; Bazinet et al., 2020). Whether these T cell responses are actively unleashed by HBsAg loss (for instance, through cross-presentation of HBsAg–HBsAb immune complexes by professional antigen-presenting cells) or whether HBsAg loss is simply a consequence of such responses is unknown because no studies have yet systematically investigated the impact of extracellular, circulating HBsAg levels on HBV-specific CD8⁺ T cell responses at the single-cell level.

Here, we took advantage of HBV replication-competent transgenic mice that produce and secrete high levels of HBsAg from integrated viral DNA to show that Ab-mediated clearance of circulating HBsAg has minimal impact on the expansion of HBV-specific CD8⁺ T cells. It does not alter their differentiation, nor does it enhance their functional restoration by immunotherapeutic strategies.

We began this study by longitudinally measuring HBsAg concentrations in the serum of HBV replication-competent transgenic (HBV Tg) mice (Guidotti et al., 1995) that are profoundly tolerant to HBV-encoded antigens at the T cell level (Shimizu et al., 1998) and maintain high levels of hepatocellular HBV gene expression and replication (displaying viremias of ∼10⁷–10⁸ viral genomes per milliliter) without developing signs of spontaneous liver immunopathology (Guidotti et al., 1995). Surprisingly, we found that, beginning at around 7 wk of age, serum HBsAg levels in some HBV Tg mice gradually decrease over time, so that, at the end of the >20-wk observation period, ∼60% of HBV Tg mice on a C57BL/6 background and ∼70% of HBV Tg mice on a C57BL/6 × Balb/c (H-2^bxd) F1 hybrid background showed undetectable serum HBsAg (Fig. 1, A and B). Interestingly, there was no difference in the early (<7 wk) concentrations of serum HBsAg between mice that eventually lost serum HBsAg and mice that did not (Fig. 1 B). Also, serum HBsAg concentrations in mice that did not become HBsAg-negative remained unchanged for the entire observation period, thus allowing us to stratify mice into two distinct groups (referred to hereafter as HBsAg⁺ and HBsAg⁻ mice; Fig. 1 B). Moreover, we found no correlation between early serum hepatitis e antigen (HBeAg) concentrations (an indirect marker of hepatic HBV gene expression), serum transaminases, or viremia and subsequent serum HBsAg loss (Fig. 1 C and Fig. S1, A and B); conversely, the absence of serum HBsAg (even for a period longer than 10 wk) had no effect on serum HBeAg levels (Fig. 1 C). Serum HBsAg clearance associated with a significant reduction in viremia and serum HBV RNA (Fig. 1, D and E; and Fig. S1 B). The hepatic content of HBV transcripts and HBV replicative intermediates as well as the lobular distribution of HBV core antigen (HBcAg) appeared comparable in HBsAg⁺ and HBsAg⁻ mice (Fig. S1, C and D; and Fig. 1 F), indicating that clearance of serum HBsAg is not due to a decreased expression of HBV gene products in the liver. Hepatocellular retention is also not likely the cause of HBsAg clearance from the serum, as this rapidly secreted Ag remained undetectable in the liver of both HBsAg⁺ and HBsAg⁻ mice (Fig. S1 E). Finally, we saw no evidence of spontaneous immunopathology in HBsAg⁻ mice, as assessed by serum transaminases and intrahepatic leukocyte recruitment (Fig. 1, G and H; and Fig. S1 A), suggesting that HBsAg clearance does not cause overt reactivation of the HBV-specific tolerant T cells present in HBV Tg mice. Together, these results show that the majority of HBV Tg mice undergo spontaneous serum HBsAg loss and raise the question of what the underlying mechanism is.

We reasoned that the spontaneous serum HBsAg loss observed in HBV Tg mice might be due to the production of HBsAb. Indeed, HBsAb were detected in the plasma of >50% of HBsAg⁻ mice beginning at 13 wk of age, while they were always undetectable in HBsAg⁺ mice (Fig. 2, A and B). Note that the slight delay between HBsAg loss and HBsAb detection (Fig. 1 B and Fig. 2 A), as well as the observation that not all HBsAg⁻ mice showed detectable level of HBsAb in their sera (Fig. 2 B), might be due to the presence of HBsAg–HBsAb complexes that cannot be picked up by our ELISA-based detection assay. The observation that a PCR-based assay still detects serum HBV DNA and RNA in HBsAg⁻ mice (albeit at very low levels; Fig. 1, D and E) further supports the presence of circulating immune complexes in these animals. To assess whether such HBsAg loss observed in HBsAg⁻ mice is indeed due to HBsAb, we crossed HBV Tg mice to D_HLMP2a mice that lack both surface and circulating immunoglobulins (Casola et al., 2004). All of the resulting Ab-deficient HBV Tg × D_HLMP2a mice showed detectable levels of circulating HBsAg (Fig. 2, C and D). HBV Tg mice × D_HLMP2a mice showed serum HBsAg concentrations, serum HBeAg concentrations at weaning and throughout the observation period, as well as serum HBV DNA levels, that are comparable to those of Ab-competent HBV Tg mice (Fig. 2, D–F; and data not shown). Finally, HBV Tg mice × D_HLMP2a mice showed no evidence of liver immunopathology (Fig. 2, G and H; and data not shown) and had hepatic HBcAg amount and staining pattern that are comparable to those of Ab-competent HBV Tg mice (Fig. 2 I). Together, the results indicate that the spontaneous HBsAg loss observed in HBV Tg mice is indeed due to Ab-mediated clearance (i.e., seroconversion). While the reason behind the seroconversion in a significant fraction of HBV Tg mice is unclear and might be different from that occurring in patients, the results are in line with the notion that most chronically infected patients keep producing low levels of HBsAbs in spite of retaining dysfunctional T cell responses (de Niet et al., 2014). Of note, these Abs are not quantitatively sufficient to appreciably reduce the large excess of serum HBsAg and circulate as immune complexes that cannot be detected by currently available diagnostic tests (de Niet et al., 2014).

We next took advantage of HBV Tg mice undergoing spontaneous seroconversion to study the role of circulating HBsAg in modulating HBV-specific CD8⁺ T cell responses. First, we adoptively transferred into HBsAg⁺ and HBsAg⁻ HBV Tg mice HBV envelope-specific CD8⁺ TCR transgenic T cells that had been differentiated in vitro into effector cells (Isogawa et al., 2013; Guidotti et al., 2015; HBV envelope-specific T cells [Env28 T_E]; Fig. 3 A). Serum HBsAg concentrations in recipient mice before the adoptive transfer of HBV-specific CD8 T_E are shown in Fig. 3 B. The presence of circulating HBsAg had no impact on either the magnitude or the duration of T cell–induced liver disease, as assessed by serum transaminase elevation (Fig. 3 C). In line with this, the number and the phenotype of Env28 T_E recovered from the liver was equivalent in HBsAg⁺ and HBsAg⁻ HBV Tg mice (Fig. 3, D–F; and Fig. S2 A). Finally, in keeping with the known antiviral activity of HBV-specific effector CD8⁺ T cells (Guidotti et al., 1996), hepatic HBV replication was suppressed and cytoplasmic HBcAg became undetectable in the vast majority of hepatocytes in both groups of mice (Fig. S2 B and Fig. 3 G). Similar results were obtained when HBV core-specific effector CD8⁺ T cells (Cor93 T_E) were used in place of Env28 T_E (Fig. S2, C–J). Thus, the capacity of HBV-specific effector CD8⁺ T cells to induce transient liver immunopathology is independent of circulating extracellular HBsAg.

We next investigated the influence of circulating, extracellular HBsAg on the de novo priming of naive HBV-specific CD8⁺ T cells upon adoptive transfer into HBV Tg mice. Previous studies have shown that, upon transfer into HBV Tg mice, HBV-specific naive CD8⁺ TCR transgenic T cells undergo hepatocellular antigen recognition, which leads to local activation, vigorous proliferation, but failure to develop antiviral effector functions (Isogawa et al., 2013; Bénéchet et al., 2019). Thus, to address the role of HBsAg on the intrahepatic priming of naive CD8⁺ T cells, Cor93-specific naive CD8⁺ TCR transgenic T cells (Cor93 T_N) were adoptively transferred into HBsAg⁺ and HBsAg⁻ HBV Tg mice (Fig. 4 A). Serum HBsAg concentrations in recipient mice before the adoptive transfer of Cor93 T_N are shown in Fig. 4 B. When we looked at the number of Cor93 T cells that we could recover from the liver 5 d after adoptive transfer, we noticed a minimal, although statistically significant, increase in intrahepatic Cor93 T cell numbers in the HBsAg⁻ HBV Tg mice (Fig. 4 C). The reason for this minor difference is unclear, but the absence of circulating HBsAg had absolutely no impact on the activation state (Fig. 4, D–G) or the phenotype (Fig. S3 A) of the intrahepatic naive HBcAg-specific CD8⁺ T cells following antigen recognition in the liver. These dysfunctional cells accumulated in the characteristic periportal clusters (Bénéchet et al., 2019) and did not have any effect on hepatic HBcAg expression, regardless of serum HBsAg concentrations (Fig. 4 H). Similar results were obtained when HBV envelope-specific naive CD8⁺ T cells (Env28 T_N) were used in place of Cor93 T_N (Fig. S3, B–I).

To confirm these findings in a potential therapeutic setting, we treated HBsAg⁺ HBV Tg mice with an anti-HBsAg monoclonal Ab that completely depletes circulating extracellular HBsAg (Fig. 4, I and J). 15 d after the first HBsAb (or isotype control) injection, we adoptively transferred both Cor93 and Env28 naive CD8⁺ T cells and measured their intrahepatic number and function 5 d later. Consistent with the data reported in Fig. 4, A–H, HBsAg clearance caused a minimal increase in intrahepatic HBV-specific CD8⁺ T cell number, but it had no impact on their functional response or phenotypic differentiation upon recognition of their cognate antigens when presented intrahepatically by the hepatocytes (Fig. 4, K–O). Together, the results indicate that Ab-mediated clearance of HBsAg has minimal impact on the expansion of HBV-specific CD8⁺ T cells undergoing intrahepatic priming, and it does not alter their propensity to evolve into dysfunctional cells lacking antiviral effector functions as a consequence of the inhibitory effects of hepatocellular antigen presentation (Isogawa et al., 2013; Bénéchet et al., 2019).

We have recently shown that IL-2–based immunotherapeutic strategies can prevent or reverse the dysfunctional CD8⁺ T cell response to intrahepatic antigen presentation (Bénéchet et al., 2019). We therefore set out to address whether the therapeutic effect of IL-2–based treatment was affected by circulating, extracellular HBsAg. To this end, we injected HBV Tg × D_HLMP2a mice with HBsAb (or isotype control) before the adoptive transfer of Cor93 T_N (Fig. 5 A). Selected mice received IL-2/anti–IL-2 immune complexes (IL-2c; Boyman et al., 2006; Bénéchet et al., 2019) 1 d after T cell transfer (Fig. 5 A). Serum HBsAg concentrations in recipient mice before the adoptive transfer of Cor93 T_N are shown in Fig. 5 B. Of note, HBsAg clearance did not alter the capacity of IL-2c to increase Cor93 T cell expansion and intraparenchymal accumulation, and to induce their cytolytic, IFN-γ–producing and antiviral functions (Fig. 5, C–G). Thus, exposure to circulating HBsAg does not affect the functional improvement of intrahepatically primed CD8⁺ T cells by IL-2–based immunotherapeutic strategies.

Together, the data reported here indicate that Ab-mediated clearance of circulating HBsAg has minimal effect on the expansion of HBV-specific CD8⁺ T cells. It does not affect their differentiation, nor does it interfere with their functional improvement by immunotherapeutic strategies. These results, obtained in immunocompetent mouse models of HBV pathogenesis in which T cell tolerance and production of high levels of HBsAg from integrated viral DNA mimic some aspects of chronic HBV infection in humans, seems at a first glance to be at odds with the observation that HBsAg seroconversion in patients under nucleos(t)ide analogue therapy is often associated with an increase in the quality and vigor of HBV-specific T cell responses (Boni et al., 2012). The systematic investigation of the impact of seroconversion on HBV-specific CD8⁺ T cell responses at the single-cell level reported here suggests instead that improved virus-specific T cell responses in chronically infected patients might precede, and possibly induce, HBsAg loss. Mechanistically, this may include the capacity of HBV-specific CD8⁺ T cells to reduce the levels of circulating HBsAg by killing or curing hepatocytes that produce HBsAg from episomal and/or integrated transcriptional templates. Whether circulating HBs clearance improves the response to subsequent therapeutic vaccination, as suggested by other investigators (Zhu et al., 2016), was not addressed in this study.

Whereas the current study evaluated the impact of circulating HBsAg on HBV-specific CD8⁺ T cell responses, previous work found that reducing the number of antigen-expressing hepatocytes (to <5%) or the amount of antigen per hepatocyte (by >15-fold) lowered HBV-specific CD8⁺ T cell expansion but was not sufficient per se to alter their propensity to evolve into dysfunctional cells lacking antiviral effector function (Bénéchet et al., 2019). Other studies have suggested that reducing hepatocellular antigen expression might increase the efficacy of subsequent therapeutic vaccination (Cobleigh et al., 2013; Michler et al., 2020).

A potential limitation of our study is that it relied on the adoptive transfer of a relatively large number of TCR transgenic CD8⁺ T cells. Since this is different from the polyclonal T cell response observed in chronically infected patients, it may raise issues related to the relevance of our findings. However, it is worth noting that endogenous HBV-specific CD8⁺ T cells are not deleted in the thymus of HBV transgenic mice; rather they are present in the periphery of adult mice; they are functionally silent and can be expanded and activated by, e.g., dendritic cell immunization (Shimizu et al., 1998), possibly reflecting the situation observed in chronically infected patients. Thus, the observation that HBsAg loss did not induce spontaneous immunopathology argues for HBsAg clearance per se not causing overt reactivation of HBV-specific tolerant T cells.

We believe that our results have significant clinical implications as monoclonal Abs against HBsAg are being evaluated for the treatment of chronic hepatitis B virus infection (Li et al., 2017; Zhang et al., 2016; Neumann et al., 2010; Galun et al., 2002). While treatment of chronically infected patients with monoclonal Abs targeting HBsAg might have the benefit of reducing the covalently closed circular DNA pool maintained by infection of new hepatocytes (Fanning et al., 2019), a potentially important aspect that we could not assess as mouse hepatocytes are refractory to HBV infection, the data presented here suggest that, in contrast to other chronic viral infections such as the one caused by HIV (Niessl et al., 2020), anti-HBV Ab therapy alone is unlikely to increase virus-specific CD8⁺ T cell immunity.

C57BL/6, CD45.1 (inbred C57BL/6), Balb/c, Thy1.1 (CBy.PL(B6)-Thy^a/ScrJ) mice were purchased from Charles River or The Jackson Laboratory. D_HLMP2A mice (Casola et al., 2004), which are devoid of surface-expressed and secreted Abs, were originally provided by K. Rajewsky (Harvard Medical School, Boston, MA) and bred >10 generations against C57BL/6 mice. HBV replication-competent transgenic mice (HBV Tg, lineage 1.3.32, inbred C57BL/6, H-2^b), which express all of the HBV antigens and replicate HBV in the liver at high levels without any evidence of cytopathology, have been previously described (Guidotti et al., 1995). In indicated experiments, HBV Tg mice were used as C57BL/6 × Balb/c (H-2^bxd) F1 hybrids or bred with D_HLMP2a mice. Cor93 TCR transgenic mice (lineage BC10.3, inbred CD45.1), in which >98% of the splenic CD8⁺ T cells recognize a K^b-restricted epitope located between residues 93 and 100 in the HBV core protein (MGLKFRQL), have been previously described (Isogawa et al., 2013). Env28 TCR transgenic mice (lineage 6C2.36, inbred Balb/c), in which ∼83% of the splenic CD8⁺ T cells recognize a L^d-restricted epitope located between residues 28 and 39 of HBsAg (IPQSLDSWWTSL), have been previously described (Isogawa et al., 2013). They were mated once with Thy1.1 mice (inbred Balb/c), so that the TCR transgenic T cells could be easily followed upon adoptive transfer by anti-Thy1.1 Ab staining. Mice were housed under specific pathogen-free conditions and used at 8 wk of age, unless otherwise indicated. In all experiments, mice were matched for age, sex, and (for the HBV Tg mice) serum HBeAg levels before experimental manipulations. All experimental animal procedures were approved by the Institutional Animal Committee of the San Raffaele Scientific Institute.

Mice were adoptively transferred with 2.5 × 10⁶ HBV-specific naive CD8⁺ TCR transgenic T cells isolated from the spleens of Cor93 and/or Env28 TCR transgenic mice, as previously described (Bénéchet et al., 2019). In indicated experiments, naive CD8⁺ T cells from the spleens of Cor93 or Env28 TCR transgenic mice were differentiated in vitro for 7–9 d into effector cells before adoptive transfer (10⁷ effector cells per mouse), as described (Guidotti et al., 2015; Manjunath et al., 2001).

HBV Tg mice were injected intraperitoneally with 25 mg/kg of HBsAb (mouse IgG2a, GS650225 XTL-17; Eren et al., 2000) or isotype control Abs (mouse IgG2a, GS-652025 CRL-2185) 15, 12, 8, 5, and 1 d before HBV-specific naive CD8⁺ T cell transfer (in the experiments described in Fig. 4 I) or 2 d and 4 h before T cell transfer (in the experiments described in Fig. 5 A). In selected experiments, mice were injected intraperitoneally with IL-2c 1 d after HBV-specific naive CD8⁺ T cell transfer, as described (Bénéchet et al., 2019). IL-2c were prepared by mixing 1.5 µg of recombinant IL-2 (clone 402 ML/CF, R&D) with 50 µg anti–IL-2 monoclonal Ab (clone S4B6-1, BioXcell) per mouse, as previously described (Bénéchet et al., 2019; Boyman et al., 2006).

Single-cell suspensions of livers were generated as described (Bénéchet et al., 2019). For analysis of ex vivo intracellular cytokine production, single-cell suspensions of livers were obtained as described above except that 1 µg/ml of brefeldin A (Sigma-Aldrich) was included in the digestion buffer. All flow cytometry stainings of surface-expressed and intracellular molecules were performed as described (Bénéchet et al., 2019). Cell viability was assessed by staining with Viobility 405/520 fixable dye (130-109-814, Miltenyi). Abs used included anti-CD8 (clone: 53–6.7, 558106, BD Biosciences), anti-CD45.1 (clone: A20, 110708, 110720, 110718, BD Biosciences), anti-CD90.1 (clone: OX-7, 561404, BD Biosciences), anti-CD25 (clone: PC61, 102017), anti-CD69 (clone: H1.2F3, 552879, BD Biosciences; 104525), anti-LAG-3 (clone: C9B7W, 125212), anti-PD-1 (clone: J43, 17-9985, 17-9985, eBioscience), anti-CD62-L (clone: MEL-14, 104428, 104420), anti-CD44 (clone: IM7, 103032, 103027), anti-IFN-γ (clone: XMG1.2, 557735 BD Biosciences; 505813), and anti-Granzyme B (clone: GB11, 515406; clone: QA16A02, 372207). All Abs were purchased from BioLegend, unless otherwise indicated. Recombinant dimeric H-2L^d:Ig and H-2K^b:Ig fusion proteins (BD Biosciences) complexed with peptides derived from HBsAg (Env28-39) or from HBcAg (Cor93-100), respectively, were prepared according to the manufacturer’s instructions. Dimer staining was performed as described (Iannacone et al., 2005).

All flow cytometry analyses were performed in FACS buffer containing PBS with 2 mM EDTA and 2% FBS on a FACS CANTO II (Becton Dickinson) and analyzed with FlowJo software (Treestar).

To isolate viral nucleic acids, 20 µl of serum were incubated for 2 h at 37°C with 180 µl IsoHi Buffer (150 mM NaCl, 0.5% NP-40, 10 mM Tris, pH 7.4), 5 mM CaCl₂, 5 mM MgCl₂, 1 U DNaseI (Life Technologies, AM1907), and 5 U Micrococcal Nuclease (New England Biolabs, MO247S). The digestion was stopped by the addition of 20 mM EDTA, pH 8.0, and viral nucleic acid purification performed with the QIAmp MiniElute Virus Spin Kit (Qiagen, 57704), according to the manufacturer’s instructions.

Serum nucleic acids were diluted 100 times, and qPCR was performed using TaqMan Fast Universal PCR Master Mix (Life Technologies, 4352042).

DNA was removed from purified serum nucleic acid using Ambion 10 TURBO DNA-freeTM DNase (AM1907). Total RNA was reverse-transcribed and amplified using TaqMan Fast Virus 1-Step PCR Master Mix (Life Technologies, 4444464).

Viremia quantification was performed for HBV envelope (forward: 5′-CCC​GTT​TGT​CCT​CTA​ATT​CC-3′; reverse: 5′-GTC​CGA​AGG​TTT​GGT​ACA​GC-3′; probe: 5′-CTC​AAC​AAC​CAG​CAC​GGG​ACC​A-3′) by absolute qPCR using a standard curve drawn with plasmid DNA. Reactions were run and analyzed on Quant Studio 5 instrument (Life Technologies). All experiments were performed in duplicate.

RNA was extracted from frozen livers (left lobe) using ReliaPrep RNA Tissue Miniprep System (Promega, Z6111), according to the manufacturer’s instructions, and genomic DNA was removed using Ambion TURBO DNA-freeTM DNase (AM1907). DNA lysis was performed with 5 µg of extracted RNA. 200 ng of total RNA was reverse-transcribed with Superscript IV Vilo (Life Technologies, 11766050) before qPCR analysis for HBV envelope. Reactions were run and analyzed on Quant Studio 5 instrument (Life Technologies). Analysis were normalized to the reference gene Gapdh (Life Technologies, Mm99999915_g1). All experiments were performed in duplicate.

Total liver DNA was analyzed for HBV replicative intermediates by Southern blot analysis, as described (Guidotti et al., 1996). Nylon transfer membranes (Roche, 11417240001) were hybridized with a digoxigenin-labeled HBV DNA probe (DIG High Prime DNA Labeling and Detection Starter kit II, Roche, 115856149109), according to the manufacturer’s instructions. Image acquisition was obtained after 24 h membrane exposure on a Syngene Imaging System.

For H&E staining and HBcAg immunohistochemistry, livers were perfused with PBS, harvested in Zn-formalin, and transferred into 70% ethanol 24 h later. Tissues were then processed, embedded in paraffin, and stained as previously described (Guidotti et al., 2015). Bright-field images were acquired through an Aperio Scanscope System CS2 microscope and an ImageScope program (Leica Biosystem) following the manufacturer’s instructions.

Serum HBsAg, HBeAg, and HBsAb were measured by ELISA, as previously described (Bénéchet et al., 2019; Tonti et al., 2013). The extent of hepatocellular injury was monitored by measuring plasma alanine aminotransferase (ALT) activity at multiple time points after treatment, as previously described (Guidotti et al., 2015).

Results are expressed as mean ± SEM. All statistical analyses were performed in Prism (GraphPad Software, version 8). Means between two groups were compared with the Mann–Whitney U test. Means among three or more groups were compared with two-way ANOVA followed by Sidak’s multiple comparison test. Kaplan–Meier curves were compared with the log-rank (Mantel–Cox) test. Comparisons are not statistically significant unless indicated.

Fig. S1 shows no correlation between early serum transaminases or viremia and subsequent serum HBsAg loss; moreover, it shows that HBsAg loss in HBV Tg mice is not due to decreased hepatic expression or to increased hepatocellular retention. Fig. S2 describes the phenotype and antiviral activity of adoptively transferred HBV-specific effector CD8⁺ T cells in the liver of HBsAg⁺ and HBsAg⁻ HBV Tg mice. Fig. S3 depicts the phenotype and function of adoptively transferred HBV-specific naive CD8⁺ T cells in the liver of HBsAg⁺ and HBsAg⁻ HBV Tg mice.

We thank M. Mainetti, M. Freschi, M. Raso, and A. Fiocchi for technical support; M. Silva for secretarial assistance; and the members of the Iannacone laboratory for helpful discussions. Flow cytometry was carried out at FRACTAL, a flow cytometry resource and advanced cytometry technical applications laboratory established by the San Raffaele Scientific Institute. We would like to acknowledge the PhD program in Basic and Applied Immunology and Oncology at Vita-Salute San Raffaele University, as V. Fumagalli and V. Venzin conducted this study as partial fulfillment of their PhDs in Molecular Medicine within that program.

M. Iannacone is supported by European Research Council consolidator grants 725038 and 957502, Italian Association for Cancer Research grants 19891 and 22737, Italian Ministry of Health grant RF-2018-12365801, Lombardy Foundation for Biomedical Research grant 2015-0010, the European Molecular Biology Organization Young Investigator Program, and a Funded Research Agreement from Gilead Sciences. L.G. Guidotti is supported by Italian Association for Cancer Research grant 22737, Lombardy Open Innovation grant 229452, Italian Ministry of Education, University and Research Progetto di ricerca di Rilevante Interesse Nazionale (PRIN) grant 2017MPCWPY, and a Funded Research Agreement from Gilead Sciences.

Author contributions: V. Fumagalli designed and performed experiments, analyzed data, performed the statistical analyses, prepared the figures, and drafted the manuscript; P. Di Lucia, V. Venzin, and E.B. Bono performed experiments and analyzed data; R. Jordan, C.R. Frey, and W. Delaney provided and characterized the monoclonal Ab targeting HBsAg; F.V. Chisari provided conceptual advice and edited the manuscript; L.G. Guidotti provided funding and conceptual advice and edited the manuscript; M. Iannacone designed and coordinated the study, provided funding, and wrote the paper.