In summary, our results demonstrated that chronic HBV infection with high levels of viral antigens resulted in upregulation of PD-1 expression and great levels of
dysfunction of HBV-specific CD8 T cells. Serum antigen levels in the early phase of infection was reduced in the absence of PD-1:PD-L1 signaling, along with the functional restoration of HBV-specific CD8 T cells. However, impeding the
PD-1:PD-L1 inhibitory signal was not sufficient to achieve long-term suppression of HBV in mice infected with high viral loads and the partially restored function of HBV-specific CD8 T cells was unable to eliminate the persistent virus.
4.1 Model advantage
More than 350 millions of people are suffering from persistent HBV infection and in high risks of viral hepatitis-derived liver cirrhosis and hepatocellular carcinoma (73, 74).
Although effective prophylaxis vaccinations have immensely reduced the rate of new infection, therapeutic approaches that facilitate the elimination of the cunning virus are still anticipated by chronic carriers, whose immune responses against HBV are found impaired in comparison to acute patients. Because HBV is unable to infect mouse hepatocytes, the lack of suitable mouse models of chronic infection impedes
experimental researches for understanding how HBV prevents the induction of antiviral immune responses and establishes persistent infection. HBV-transgenic mice were firstly introduced as a mouse model of chronic carrier for the studies of anti-HBV immunity in chronic infection (48, 49). HBV-specific CD8 T cells are absent in these mice due to central tolerance and have to be generated by other donors and adoptively transferred into the transgenic recipients. Hydrodynamic injection of plasmids or
infection of adenoviral vectors carrying the full-length HBV genome were reported to bypass the viral entry step and achieve long-term HBV replication and protein
expression in mouse hepatocytes (50, 51, 75). These approaches made successful models of chronic HBV infection in immunocompetent mice, however, it is difficult to manipulate viral doses in the infected hosts because of the experimental limitations of these models. Using adeno-associated viral vectors as a delivery system, we showed in this report that we are able to control the viral loads in immunocompetent hosts via different sizes of viral inoculum and all of the doses can establish persistent expression of HBV proteins (Figure 1). This allows us to address the effects of different viral loads on immune tolerance to HBV in a system with homogenous host genetics, viral
genotypes, and infection durations. For a comprehensive study of HBV-specific cellular immune responses, we screened HBV epitopes in the context of H2-Kb by overlapping peptide and identified 3 dominant epitopes of HBsAg and HBcAg that were previously reported (Figure 11) (67-69). We therefore focused on the functions and phenotypes of CD8 T cells recognizing these three HBV epitopes during persistent HBV infection.
4.2 Tolerance
Our AAV/HBV infection model recapitulates several immunological features of chronic viral infections. In chronic viral infections of humans with HIV, HBV, and HCV and of mice with LCMV, virus-specific CD8 T cells are characterized with exhausted phenotypes (1, 3, 76, 77). Virus-specific CD8 T cells show transient function upon stimulation of cognate antigens but acquire functional loss during the progression of chronic infection, including impairment of cytokine production, cytolytic activity, and proliferation. Consistently, in our AAV/HBV infection mouse model we observed a peak of HBs-specific CD8 T cells in the liver and the spleen at week 2 p.i. (early phase
of infection), which became undetectable in later time points. The transient responses against HBV are seldom observed in patients at an early stage of chronic infection. The exhaustion of CD8 T cells was specific to HBV, as the immunization of OVA in AAV/HBV-infected mice generated the same quantity and quality of OVA-specific IFN-γ production from CD8 T cells as compared with that in the control mice.
Interestingly, we observed a higher frequency of HBs-specific CD8 T cells in the liver than in the spleen, which is in line with clinical observations, indicating HBV-specific CD8 T cells are more enriched locally than peripherally (78).
4.3 Dose effect
Clinical studies showed an inverse correlation of viral loads and the antiviral functions of virus-specific CD8 T cells in chronic infections of HIV and HBV (22, 23).
However, due to the complexity of clinical samples it is still elusive that whether high viral loads of HBV are the cause or the consequence of CD8 T cell exhaustion. By manipulating the viral doses used for infection, we demonstrated clearly that higher viral loads led to more severe status of immune dysfunction of HBV-specific CD8 T cells. Therapeutic vaccination restored the production of IFN-γ by HBV-specific CD8 T cells to a greater extent in mice with low viral loads, whose serum antigen levels
reduced approximately 50% (Figure 2). In contrast, the level of HBsAg in mice with high viral load was slightly reduced (15% reduction). It is noteworthy that there is no elevation of ALT detected after vaccination, suggesting a non-cytolytic control could have contributed to the reduction of vial antigen through production of antiviral cytokines, such as IFN-γ, by HBV-specific CD8 T cells (74). Interestingly, Asabe and colleagues demonstrate that HBV persistency and impaired CD8 T cell functions were observed in chimpanzees inoculated with very low viral titers, attributed to the late
priming of CD4 T cells after 100% of hepatocytes are infected (79). Both high-dose and low-dose inocula of HBV cause high viral burdens in infected chimpanzees, as well as poor responses of intrahepatic CD8 T cells. In contrast, the viral loads are lower in chimpanzees infected with intermediate sizes of viral inocula, in which more robust CD8 T cell responses are found in the liver.
In clinical studies of chronic carriers receiving lamivudine treatments for 1 year, restoration of HBV-specific T cell functions can be extensively observed after dramatic decrease of HBV loads, as a result of suppression of viral replication (80, 81). This phenomenon indicates an effect of viral loads on the inhibition of T cell responses.
However, the restoration was only transiently induced and declined 4 months after the starting of treatments, regardless of the sustained viral suppression, suggesting the requirement of other immune stimulations to reinforce and extend the antiviral immunity (82).
4.4 PD-1 expression and other regulatory factors
During persistent infection, exhausted virus-specific CD8 T cells express several inhibitory receptors which generate suppressive signals when ligation with their ligands (4). PD-1, one of the first implicated inhibitory receptors in T cell exhaustion, is highly expressed on virus-specific T cells and recruits SH2-domain containing tyrosine phosphatase 1 (SHP-1) and SHP-2 which inhibit TCR signaling (7). Upregulation of PD-1 is observed on HBV-specific CD8 T cells of chronic carriers and inversely correlated with viral burdens (23, 42, 83). As shown in Figure 3, infection of higher doses of HBV led to not only higher frequencies of PD-1+ CD8 T cells but greater degree of PD-1 expression on each CD8 T cells, suggesting the tolerance of CD8 T cells in the liver was reinforced in the context of higher viral loads (84). The results were
only found in the liver, while the dose-dependent effect of PD-1 expression was absent in the spleen. This is in line with clinical findings that PD-1 expression on
HBV-specific CD8 T cells was higher in the liver than in the periphery (78, 85). The induction of PD-1 on intrahepatic CD8 T cells was not observed in AAV/Empty infected mice, indicating a HBV-dependent mechanism of PD-1 induction rather than an AAV-derived effect. Increased PD-1 expression on CD4 T cells was also observed in a dose-dependent manner (Figure 5), implicating a lack of CD4 T cell help which is necessary for the effector functions of HBV-specific CD8 T cells (40, 79, 86). It is controversial whether frequencies of CD4+CD25+Foxp3+ regulatory T cells (Treg cells) is increased in chronic HBV patients and whether the frequencies of Treg cells is correlated with outcome of HBV infection (87). Accumulation of Treg cells was not detected in this model of infection (data not shown), which is different from another model using AAV vector as chronic model of HBV infection (88).
4.5 Blocking PD-1
Beginning from the first evidence of the therapeutic effect by blocking PD-1 pathway in LCMV model (5), several independent studies demonstrated that the antiviral
functions of virus-specific CD8 T cells from chronic carriers can be restored through the blockade with anti-PD-1 or anti-PD-L1 monoclonal antibodies (6, 23, 89).
Immunotherapy targeting PD-1:PD-L1 interaction has become one of the most promising strategies for rescuing functionally impaired immunity against chronic infections and cancers (15, 90, 91).
The effects of blocking PD-1 pathway on the restoration of HBV-specific T cell functions have been shown in clinical studies (23), as well as in animal models of chronic HBV infection. The proliferation and degranulation of peripheral T cells were
partially enhanced after in vitro blockade with Abs against wPD-L1 and wPD-L2 (92).
In an in vivo study in HBV immune tolerance, blockade of PD-L1 resulted in a delay of the suppression of the antiviral functions of HBV-specific CD8 T cells adoptive
transferred into HBV transgenic mice (93). Administration of anti-PD-1 antibody reversed the functional impairment of HBV-specific CD8 T cells and viral persistence in a hydrodynamic injection model, in which the viral burdens are around 106 copies per milliliter in the carrier hosts (94). However, large quantities of antigens presented by hepatocytes can lead to a more exhausted state of specific CD8 T cells (95), which renders it unclear whether therapeutic blockade of PD-1:PD-L1 signaling could rescue anti-HBV immunity in the context of high viral loads. To confirm the role of PD-1 signaling in HBV tolerance, we used PD-L1 KO mice in which the major ligand of PD-1 is genetically deprived. Figure 6 showed that lack of PD-1 reinforced the specific functions of HBV-specific CD8 T cells and the levels of HBsAg were reduced in acute phase of infection, a result in line with clinical findings (96, 97). Thereafter, the effect of HBsAg reduction subsided in mice with prolonged exposure to high levels of Ag, while mice with low viral loads maintained the suppression of HBsAg levels (Figure 9A to C). The difference in HBsAg control among the different groups could be attributed to the different degrees of HBV-specific CD8 T cells, which were inversely correlated with the viral load (Figure 9D to G). These results are in keeping with the findings by Shata et al. that high viral loads of HBV cause a more severe status of dysfunction which is more difficult to be rescued (98). It should be noted that although
HBV-specific CD8 T cells were more potent in PD-L1 KO mice as compared with those in WT mice (Figure 9F, G vs. Figure 2B, C), these virus-specific T cells were not sufficient to eliminate the persistent virus, particularly in mice with high viral loads, suggesting the existence of regulating factors other than PD-1:PD-L1 pathway. Indeed,
we found several other inhibitory receptors that are upregulated on CD8 T cells in this model of chronic infection (Figure 4). Other suppressive factors have been reported that have negative effects on anti-HBV T cell responses while blocking these pathways restores T cell functions (99-102). Experimental evidences confirm the truth that combinatorial blockades of PD-1 and another inhibitory pathway are more effective to reinforce immunities that eliminate chronic viral infections (14, 17, 103). Besides, combining activation signals with blocking inhibitory pathways increases the functions of intrahepatic T cells of chronic hepatitis B patients (85). Depriving both PD-1 and LAG-3 pathways starting before the infection resulted in accelerated reduction of serum HBsAg level, but was not sufficient to clear the virus (Figure 10). Lack of PD-1 alone had more impact on the decrease of serum antigen level than blocking LAG-3 alone, suggesting a PD-1 pathway is more dominant in regulating T cell tolerance in this model of chronic HBV infection.
4.6 HBV-specificity of PD-1
+CD8 T cells
We could not detect comparable amounts of S190-pentamer+ CD8 T cells in AAV-HBV-infected mice after 4 months p.i. It could result from an altered binding reactivity of these T cells to MHC/peptide complex, which was reported in study of chronic HBV patients (104). Another plausible explanation is that the dominant clone of HBV-specific CD8 T cells has been deleted through apoptosis when exposed to high levels of antigens during chronic viral infection (1, 3, 70, 105).
The alternative ways to demonstrate the antigen-specificity is through functional assays. However, since exhausted CD8 T cells usually lose the antiviral functions, it is a great challenge to use functional approaches to prove the specificity of these T cells to HBV. As a result, we were unable to detect positive responses when stimulating PD-1+
CD8 T cells with overlapping peptides and even BMDCs presenting HBV epitopes (Figure 12 and 13). A surprising data shown in Figure 14 was acquired that PD-1+ CD8 T cells produced TNF-α in response to overlapping peptides S51-60 after in vitro expansion, whereas PD-1- CD8 T cells did not. There was no epitope found in pool S51-60 during epitope screening (Figure 11), indicating a subdominant clone might arise during chronic infection. However, as shown in Figure 15, some
antigen-experienced CD8 T cells did not overexpress PD-1, suggesting there are HBV-specific CD8 T cells in PD-1- population. And if PD-1- CD8 T cells were less exhausted, why didn’t they react to overlapping peptides? Since the experiment is only done once, it merits further reproducible evidences to confirm this finding.
Finally, indirect evidences that suggest the antigen specificity of intrahepatic PD-1+ CD8 T cells were obtained. By gating with surrogate activation marker (CD11a), we illustrated that all intrahepatic PD-1+ CD8 T cells possessed the phenotype of
antigen-experienced CD8 T cells (Figure 15). In addition, TCRβ spectratype analysis showed that there was a higher degree of clonal expansion in PD-1+ CD8 T cells than PD-1- ones (Figure 16).
4.7 Conclusion
Our results provide new insights for immunotherapies against chronic HBV infection. First, reducing viral antigens by antiviral therapy, such as nucleot(s)ide analogues or siRNA, could ameliorate the exhaustion of virus-specific T cells and gain benefits to the effect of immunotherapies, as indicated in the clinical reports of
lamivudine treatments by Boni and colleagues (80, 81). Second, blocking the inhibitory signals of PD-1:PD-L1 alone may not serve as a sufficient therapeutic approach for activating HBV-specific CD8 T cells, particularly in chronic carriers whose viral
burdens are high. Combinatorial treatments targeting multiple inhibitory pathways and/or activating stimulatory signaling may be required to achieve a better response to clear HBV.