Expression of Viral Antigen by the Liver Leads to Chronic Infection Through the Generation of Regulatory T Cells

Background & Aims The constant exposure of the liver to food and bacterial antigens through the mesenteric circulation requires it to maintain tolerance while preserving the ability to mount an effective immune response against pathogens. We investigated the contribution of the liver’s tolerogenic nature on the establishment of chronic viral infections. Methods TTR-NP mice, which express the nucleoprotein (NP) of lymphocytic choriomeningitis virus (LCMV) specifically in hepatocytes under control of a modified transthyretin (TTR) promoter, were infected with the Armstrong (Arm) or WE acute strains of LCMV. Results The infection persisted for at least 147 days in TTR-NP mice. Expression of NP by the liver induced a strong peripheral tolerance against NP that was mediated by interleukin-10-secreting CD4+ regulatory T cells, leading to high PD-1 (programmed death-1) expression and reduced effector function of virus-specific T cells. Despite an active immune response against LCMV, peripheral tolerance against a single viral protein was sufficient to induce T-cell exhaustion and chronic LCMV Armstrong (Arm) or WE infection by limiting the antiviral T-cell response in an otherwise immunocompetent host. Regulatory T-cell depletion of chronically infected TTR-NP mice led to functional restoration of LCMV-specific CD4+ and CD8+ T cell responses and viral clearance. Conclusions Expression of a viral antigen by hepatocytes can induce a state of peripheral tolerance mediated by regulatory T cells that can lead to the establishment of a chronic viral infection. Strategies targeting regulatory T cells in patients chronically infected with hepatotropic viruses could represent a promising approach to restore functional antiviral immunity and clear infection.


SUMMARY
Expression by the liver of a viral protein induced an immunologic tolerance mediated by interleukin-10secreting regulatory T cells that impaired the antiviral T-cell response, leading to a chronic infection. This mechanism could be involved in the establishment of persistent infection by hepatotropic viruses.

BACKGROUND & AIMS:
The constant exposure of the liver to food and bacterial antigens through the mesenteric circulation requires it to maintain tolerance while preserving the ability to mount an effective immune response against pathogens. We investigated the contribution of the liver's tolerogenic nature on the establishment of chronic viral infections.
METHODS: TTR-NP mice, which express the nucleoprotein (NP) of lymphocytic choriomeningitis virus (LCMV) specifically in hepatocytes under control of a modified transthyretin (TTR) promoter, were infected with the Armstrong (Arm) or WE acute strains of LCMV.

RESULTS:
The infection persisted for at least 147 days in TTR-NP mice. Expression of NP by the liver induced a strong peripheral tolerance against NP that was mediated by interleukin-10-secreting CD4 þ regulatory T cells, leading to high PD-1 (programmed death-1) expression and reduced effector function of virus-specific T cells. Despite an active immune response against LCMV, peripheral tolerance against a single viral protein was sufficient to induce T-cell exhaustion and chronic LCMV Armstrong (Arm) or WE infection by limiting the antiviral T-cell response in an otherwise immunocompetent host. Regulatory T-cell depletion of chronically infected TTR-NP mice led to functional restoration of LCMV-specific CD4 þ and CD8 þ T cell responses and viral clearance. T he liver is an immunoprivileged site prone to tolerance induction. For example, liver grafts are accepted without immunosuppression in several mammals, 1 and oral tolerance is abrogated when a portacaval shunt is performed. 2 The liver also has the unique ability among solid organs to activate naive CD8 þ T lymphocytes in an antigen-specific manner, a process that can be inefficient and lead to apoptosis through a Bim-dependant pathway. 3 The liver is also host to several chronic infections, but infection of the liver does not inevitably lead to viral persistence. Strong innate immune responses combined with specific T-cell responses can overcome viral escape mechanisms-as in hepatitis A infection or resolved acute hepatitis B (HBV) or C virus (HCV) infections-and achieve viral clearance. However, as observed in 50% to 80% of HCV infections, pathogens can evade early innate and adaptive immune responses through high antigen loads and increased coinhibitory signaling locally in the inflamed liver, leading to T-cell exhaustion and viral persistence. 4 After exposure, HCV reaches maximal titers several weeks before the induction of detectable humoral or cellular immune responses and the onset of liver disease; in cases where HCV titers remain relatively low, T cell responses may remain undetectable even during chronic infection. 5 Therefore, the liver may be exposed to HCV antigens in absence of strong immune responses. It has recently been shown in a model of chronic viral infection that CD4 þ regulatory T-cell depletion in combination with programmeddeath-ligand-1 (PD-L1) blockade can significantly reduce viral titers, highlighting the importance of immune inhibitory signals in the outcome of viral infections. 6 Thus, the liver's ability to induce tolerance to locally expressed antigens could contribute to the development of chronic liver infections by altering the immunologic response against liver-expressed viral antigens. 7 Therefore, we directly assessed whether expression of a viral antigen by hepatocytes can induce a state of peripheral tolerance able to contribute to viral persistence.
We show that expression of the nucleoprotein (NP) from lymphocytic choriomeningitis virus (LCMV) specifically in hepatocytes leads to strong peripheral tolerance mediated by interleukin-10 (IL-10)-secreting CD4 þ forkhead box P3 þ (FoxP3 þ ) regulatory T cells. This allows the establishment of chronic infection by acute strains of LCMV associated with the loss of CD4 þ and CD8 þ T-cell effector function, leading to high viral titers in the liver and spleen. Depletion/ silencing of CD4 þ regulatory T cells resulted in the progressive restoration of T-cell function, loss of PD-1 (programmed death-1) expression, and gradual viral clearance. This study demonstrates that expression of a viral antigen in the liver leads to the development of CD4 þ regulatory T cells able to interfere with the antiviral T-cell response, allowing the establishment of a chronic infection.

Mice and Viruses
Transthyretin-nucleoprotein (TTR-NP) transgenic mice 8 (8-to 12-week-old females) expressing LCMV NP specifically in hepatocytes (kindly provided by F. Alvarez, CHU Sainte-Justine, Montreal, Canada) or control C57BL/6 (B6) mice (8-to 12-week-old females) (Charles River, Montreal, Canada) were infected with either acute strains LCMV-Arm (200 plaque-forming units [pfu] intraperitoneally [IP]) or LCMV-WE (200 pfu intravenously [IV]). Hemizygous TTR-NP1/0 mice were generated as a F1 cross between homozygous TTR-NP and C57BL/6 mice. The NP 396-404 -specific T-cell receptor (TCR) transgenic TNP4 mice 9 were kindly provided by F. Alvarez (CHU Sainte-Justine). The glycoprotein GP 33-41 -specific TCR transgenic P14 mice were kindly provided by P. Ohashi (Princess Margaret Cancer Centre, Toronto, Canada). The recombination-activating gene-nucleoprotein (RAG-NP) mice were obtained by crossing TTR-NP and RAG-1 mice (kindly provided by C Daniel, INRS-Institut Armand-Frappier, Laval, Quebec, Canada). The RAG-1 phenotype was assessed by flow cytometry, and NP expression was monitored by polymerase chain reaction, as previously described elsewhere. 8 LCMV-WE and Armstrong strains were obtained from R.M. Zinkernagel at the Institute of Experimental Immunology (Zurich, Switzerland). LCMV titration was performed via a standard focus-forming assay. All experiments were performed under protocols approved by the INRS Institutional Committee for Animal Care and following guidelines published by the Canadian Council on Animal Care.

Lymphocyte Isolation From Liver and Spleen
Livers were perfused via the portal vein with RPMI 1640 (Thermo Fisher Scientific, Mississauga, ON, Canada) and removed. The livers and spleens were finely minced in RPMI 1640, passed through a 100-gauge steel mesh, and centrifuged at 400g for 5 minutes at 4 C. Cells contained in the supernatant were washed 3 times with RPMI 1640/5% fetal calf serum before being centrifuged on a Percoll (GE Healthcare Canada, Mississauga, ON, Canada) gradient to purify lymphocytes. 10

Type I Interferon and Interleukin-10 Measurement
The IFN-a levels were measured in sera, liver, and spleen homogenates in LCMV-infected B6 and TTR-NP mice with the Verikine mouse IFN-a enzyme-linked immunoassay (ELISA) kit (PBL Interferon Source, Piscataway, NJ) according to the manufacturer's instructions. IFN-b levels were measured in sera of infected B6 and TTR-NP mice with the Verikine mouse IFN ELISA kit (PBL Interferon Source). The IL-10 serum levels were measured using the multiplex bead immunoassay (Life Technologies/Gibco, Grand Island, NY).
Alternatively, 1.5 Â 10 6 CD8 þ T cells from P14 and TNP4 mice (Stemcell Technologies) were adoptively transferred IV into RAG-NP and RAG control mice. The mice were killed 7 days later. Isolated lymphocytes from the spleen and liver were stained (CD3, CD4, CD8, and 7-AAD), and the proliferation of CFSE and e450-labeled cells was analyzed by flow cytometry on a BD LSRFortessa (BD Biosciences) and using the FlowJo software (Tree Star).
In Vivo Regulatory CD4 þ CD25 þ T-Cell Depletion/Silencing LCMV-infected TTR-NP mice (45 days after infection) were injected IP with 200 mg of anti-CD25 antibodies (PC61.5 clone) or 200 mg of IgG1 isotypic control (BioXcell, West Lebanon, NH) every 7 days. The CD4 þ CD25 þ regulatory T cell depletion was confirmed by flow cytometry. The mice were killed 28 days after the start of regulatory T cell (Treg) depletion.

Quantification of LCMV NP Expression Levels by Flow Cytometry
To measure the expression levels of LCMV NP on liver cells, purified anti-NP antibodies from the monoclonal IgGsecreting hybridoma VL4 (kindly given by R.M. Zinkernagel, Institute of Experimental Immunology, Switzerland) were coupled to Alexa Fluor 647 using the Alexa Fluor Protein Labeling Kit (Life Technologies, Rockville, MD). Mechanically disrupted liver cells were separated on a discontinuous 40%/80% Percoll (GE Healthcare Canada) gradient to separate parenchymal from nonparenchymal cells. Parenchymal cells were stained with fluorochrome labeled anti-CD45, 7-AAD, and Alexa647-anti-NP and were analyzed by flow cytometry on a BD LSRFortessa (BD Biosciences) and using FlowJo (Tree Star).

Serum Alanine Aminotransferase Levels
Serum aminotransferase (ALT) levels were measured using a TRILOGY Multipurpose Analyzer system (DREW Scientific, Dallas, TX).

Statistical Analysis
Differences between groups were tested using one-way analysis of variance (ANOVA) with Tukey's post hoc test. The Student t test was used when two groups were compared. In all graphs, the error bars represent the standard error of the mean. All statistical analyses were performed using GraphPad Prism version 5 (GraphPad Software, La Jolla, CA).

Expression of LCMV Nucleoprotein by Hepatocytes in TTR-NP Mice Leads to Chronic Infection With Acute Strains of LCMV
TTR-NP mice express LCMV NP specifically in hepatocytes under the control of a modified transthyretin (TTR) promoter. 8 This liver-specific transthyretin minimal enhancer promoter sequence limits expression to hepatocytes without any detectable expression in the spleen, thymus, muscle, or kidney. 8 TTR-NP mice were infected with low doses (200 pfu) of LCMV-Arm or WE. Infected mice showed a gradual increase in serum alanine aminotransferase (ALT) levels 55 days after infection indicative of chronic hepatitis ( Figure 1A). Infected TTR-NP mice showed 20% to 30% mortality rates compared with 0% for infected B6 mice (see Figure 1B). The LCMV titers in infected TTR-NP mice were higher on day 8 than in B6 mice and remained high in the spleen, liver, and kidney for over 147 days after infection (see Figure 1C). Within each infected mouse, the majority of virus was found in liver (see Figure 1C).
Active Antiviral Immune Response in Infected TTR-NP Mice With High PD-1 Levels on LCMV-Specific CD8 þ and CD4 þ T Cells Fifty-five days after LCMV infection, TTR-NP mice showed higher proportions of effector CD8 þ T cells in the spleen and among liver-infiltrating lymphocytes (LIL) compared with infected B6 mice (Figure 2A). Beginning on day 8 and increasing until day 55 after infection, TTR-NP mice showed high PD-1 expression on CD8 þ and CD4 þ T cells (see Figure 2B). The PD-1 expression was higher on effector/memory CD8 þ T cells than on naive CD8 þ T cells in infected TTR-NP mice (see Figure 2C). The levels of PD-1 on epitope-specific CD8 þ and CD4 þ T cells were measured using class I and II tetramers (gating strategy described in Supplementary Figure 1). The PD-1 levels were elevated on NP 396-404 and GP 33-41 CD8 þ T cells from TTR-NP mice compared with the B6 mice (see Figure 2D). The levels of PD-1 on NP 396-404 -specific CD8 þ T cells from TTR-NP mice were also significantly higher than those of GP 33-41 CD8 þ T cells from TTR-NP mice (see Figure 2D). The levels of PD-1 on GP 31-45 , GP 66-77 , and NP 309-328 CD4 þ T cells from TTR-NP mice were significantly higher than on those from the infected B6 mice (see Figure 2E).

Normal Type I Interferon Response in LCMV-Infected TTR-NP Mice
Type I IFN is critical for the development of a proper T-cell response against LCMV. 12 In vitro, LCMV NP has been shown to inhibit the type I IFN response. 13 The IFN-a and -b levels in the serum, spleen, and liver of infected TTR-NP and B6 mice were compared, and no significant differences were observed (see Figure 2F).
These observations are consistent with the fact that although NP396-404-specific T cells are present in fewer numbers in LCMV-infected TTR-NP mice, they are not deleted (see Figure 3). Cytokine-producing LCMV-specific CD8 þ T cells were reduced in the spleen of TTR-NP mice compared with B6 mice 55 days after infection. Fewer polyfunctional TNF-a þ IFN-g þ CD8 þ T cells were detected after stimulation with GP 33-41 and very few after  stimulation with NP 396-404 ( Figure 5A, upper left panel). The CD4 þ T-cell responses were similarly reduced in TTR-NP mice compared with B6 mice, with very few cells responding to NP 311-325 peptide stimulation (see Figure 5A, upper right panel). Therefore, although TTR-NP mice mount CD4 þ and CD8 þ T-cell responses against both NP and GP-derived epitopes after the establishment of chronic LCMV infection (see Figure 3), these cells secrete fewer cytokines 8 days after LCMV infection (see Figure 5A, lower left panel), and their levels remained low throughout chronic infection (see Figure 5A, lower right panel), especially those targeting NPderived epitopes.
This reduced number of NP-specific polyfunctional (TNF-a þ IFN-g þ ) T cells could limit the ability of TTR-NP mice to eliminate infected cells and control viral burden. To test this, an in vivo cytotoxicity assay was performed in B6 and TTR-NP mice 30 days after infection. The LCMVinfected B6 mice developed strong CTL responses against both GP [33][34][35][36][37][38][39][40][41] and NP 396-404 pulsed cells (see Figure 5B), resulting in their rapid elimination. In TTR-NP mice an efficient CTL response against GP 33-41 -labeled target cells was detected whereas the response against NP 396-404pulsed target cells was undetectable (see Figure 5B). The in vitro proliferative capacity of these cells was also compromised, as NP 396-404 -specific T cells in TTR-NP mice did not proliferate in response to peptide stimulation whereas GP 33-41 -specific T cells did (see Figure 5C).

Efficient Presentation of Hepatocyte-Expressed NP to CD8 þ T Cells
To assess the efficiency of NP presentation early during LCMV infection in TTR-NP mice, GP [33][34][35][36][37][38][39][40][41] and NP 396-404specific CD8 þ T cells isolated respectively from P14 and TNP4 TCR transgenic mice were labeled with CFSE or efluor450, respectively, and were adoptively transferred into TTR-NP and B6 mice before LCMV infection. Proliferation of GP 33-41 -and NP 396-404 -specific T cells in the spleen was analyzed 5 days later. The GP 33-41 -specific CD8 þ T cells proliferated more than the NP 396-404 -specific T cells in B6 mice, highlighting the previously reported immunodominance of this epitope at this early stage of LCMV infection 15 ( Figure 6A). However, although the GP 33-41 -specific proliferation was similar between TTR-NP and B6 mice, NP 396-404 -specific proliferation was increased in TTR-NP mice, suggesting that NP 396-404 -T cells are not anergic at early stages of infection (see Figure 6A). However, the impact of endogenous versus viral LCMV-NP expression on NP 396-404 -T cell proliferation was indistinguishable in this setting.
To test the hypothesis that endogenous NP in absence of LCMV-expressed NP could lead to the observed NP 396-404specific T-cell proliferation, GP 33-41 -and NP 396-404 -specific T cells were adoptively transferred into uninfected RAGdeficient and RAG-deficient TTR-NP mice (RAG-NP). Naive T cells expand in RAG-deficient mice, mostly through an IL-7 driven homeostatic T-cell proliferation process, 16 thereby facilitating the determination of the proliferation of NPspecific T cells in presence of survival factors (IL-7) without the interference of LCMV-expressed NP. RAGdeficient animals also lack any prior Treg-mediated tolerance to NP that could interfere with NP presentation. Seven days after adoptive transfer, NP 396-404 -specific T-cell proliferation was increased in RAG-NP mice compared with RAG mice whereas GP 33-41 -specific T-cell proliferation was similar (see Figure 6B). Therefore, endogenous NP can lead to the proliferation of NP 396-404 -specific T cells in RAG-NP mice.
Taken together these results indicate that although TTR-NP mice are hyporesponsive toward NP after LCMV infection, this does not result from liver-mediated NP 396-404 CD8 þ T-cell deletion (see Figure 4) or ineffective T-cell activation through tolerizing endogenous NP presentation (see Figure 6A and B).

CD4 þ FoxP3 þ Regulatory T Cell Conversion/ Expansion in TTR-NP Mice
To determine whether Tregs could specifically convert/ expand in response to liver-expressed NP, 1 Â 10 8 splenocytes from B6 mice were adoptively transferred into RAG and RAG-NP mice, and the T-cell proliferation was monitored by Ki-67 staining. This number of adoptively transferred cells in a RAG-deficient mouse, while minimizing homeostatic proliferation, facilitated tracking of proliferation based solely on NP expression and was not influenced by preestablished tolerance to NP. The CD4 þ FoxP3 þ cells, but not the CD4 þ FoxP3 À T cells, showed an increased Ki-67 expression in RAG-NP mice in comparison with RAG mice ( Figure 7A).

CD4 þ Regulatory T Cell Expansion in Infected TTR-NP Mice
The number of CD4 þ FoxP3 þ regulatory T cells before LCMV infection was not statistically significantly different between the TTR-NP and B6 mice in the spleen and liver (see Figure 7B). Eight days after LCMV-WE infection, the number of splenic Tregs in B6 mice decreased slightly while their numbers remained elevated in infected TTR-NP mice (see Figure 7B). The number of liver Tregs was low and did not significantly vary after infection ( Figure 7B).
Similar results were obtained after LCMV-Arm infection. The TTR-NP mice showed a significant increase in the proportion of CD4 þ FoxP3 þ T cells among the CD4 þ T cells after infection with LCMV-WE or LCMV-Arm compared with the B6 mice in both the spleen and liver (see Figure 7C), and they remained elevated throughout the chronic infection. The FoxP3 þ regulatory T cells from infected TTR-NP mice expressed statistically significantly higher levels of Helios than those from infected B6 mice (see Figure 7D). Fifty-five days after infection, the serum levels of IL-10 were statistically significantly higher in the infected TTR-NP mice than in the B6 mice and remained elevated up to 147 days after infection (see Figure 7E). In these mice, CD4 þ FoxP3 þ regulatory T cells were responsible for the majority of IL-10 production among splenocytes (see Figure 7F).

Level of NP Expression in Hepatocytes Is Critical for the Establishment of LCMV Chronicity
The expression levels of NP in the liver of TTR-NP mice could have a direct impact on the fate of CD8 þ T cells 17 and the chronicity of infection. To assess this, hemizygous TTR-NP 1/0 mice expressing NP at half the levels of that of TTR-NP mice were used ( Figure 8A). The number of CD4 þ regulatory T cells before infection was similar between TTR-NP 1/0 and TTR-NP mice in both the spleen and liver (see- Figure 8B). However, the significant expansion of CD4 þ FoxP3 þ Tregs observed in the TTR-NP mice after LCMV infection was not seen in the hemizygous TTR-NP 1/0 mice (see Figure 8C). LCMV infection of hemizygous TTR-NP 1/0 mice led to increased IFN-g and TNF-a CD8 þ T cell responses against both GP [33][34][35][36][37][38][39][40][41] and NP 396-404 epitopes compared with the TTR-NP mice. However, the NP 396-404 T cell responses did not reach that of infected B6 mice (see Figure 8D).
T cells from hemizygous TTR-NP 1/0 mice showed increased in vivo cytotoxicity against NP-expressing target cells compared with LCMV-infected TTR-NP mice while remaining lower than that of B6 mice (see Figure 8E). Despite the impaired T-cell response against NP-derived epitopes in hemizygous TTR-NP 1/0 mice, infection with acute strains of LCMV did not persist (see Figure 8F).

Depletion/Silencing of CD4 þ Regulatory T Cells Leads to Viral Clearance in Chronically Infected TTR-NP Mice
To ascertain the role of CD4 þ regulatory T cells in viral persistence of infected TTR-NP mice, these cells were depleted/silenced using CD25-specific antibodies. Forty-five days after LCMV infection, the TTR-NP mice were injected weekly with anti-CD25 antibodies (PC61.5) or isotypic control. Treatment reduced the levels of CD4 þ CD25 þ FoxP3 þ T cells by 96.8% ± 2.1% in the spleen and 87.6% ± 1.6% in the liver. Over a 28-day period, the PD-1 expression by CD4 þ and CD8 þ T cells gradually returned to control levels ( Figure 9A). CD8 T-cell functionality was also improved, with increased numbers of IFN-g þ TNF-a þ CD8 þ T cells after stimulation with GP 33-41 and NP 396-404 peptides (see Figure 9B).
The serum levels of IL-10 were significantly reduced in infected Treg-depleted/silenced TTR-NP mice compared with infected TTR-NP mice (see Figure 9C). LCMV viremia gradually declined, and the mice achieved viral clearance 28 days after beginning the anti-CD25 treatment (see Figure 9D).

Discussion
We found that expression of a viral protein by the liver in the absence of inflammation can establish a state of immunologic tolerance against the virus, leading to the establishment of a persistent infection. This expression led to the generation of regulatory T cells able to impede the antiviral immune response, allowing acute LCMV strains to persist with characteristics normally only observed during chronic LCMV Clone 13 infection, such as T-cell exhaustion and PD-1 up-regulation. 18 Chronic infection with acute LCMV-Arm or LCMV-WE strains has only been shown to occur in either neonates 19 or severely immunocompromised animal models such as perforin-invalidated mice. 20 However, TTR-NP mice are not immunodeficient and mount efficient antiviral T-and B-cell responses against vesicular stomatitis virus and clear infection (data not shown).
TTR-NP mice also develop an early type I interferon response to LCMV similar to that of B6 mice. In vitro experiments have shown that NP expression could inhibit the type I interferon response. 13 However, NP expression is restricted to the hepatocytes in TTR-NP mice, and these cells are not major producers of type I interferon during LCMV infection. 21 TTR-NP transgenic mice express NP specifically in hepatocytes under the control of a modified TTR promoter and do not express NP in the thymus. 8 As evidenced by the presence of NP-specific T cells after infection, extensive central deletion of NP-reactive T lymphocytes does not occur in these mice. This contrasts with a previous transgenic model tolerant to NP due to clonal deletion. 22 The active peripheral tolerance against NP in TTR-NP mice likely stems from an expansion of or conversion to regulatory CD4 þ T cells.
FoxP3 þ regulatory T cells are central to peripheral tolerance both by their direct action on T cells but also by keeping dendritic cells in an immature state that can then efficiently induce peripheral tolerance. 23 This regulatory T cell-mediated active tolerance curbed the antiviral response against NP-derived epitopes but also interfered with CD4 þ and CD8 þ T-cell responses against GP-derived epitopes. CD4 þ regulatory T cells initially require TCR stimulation to expand and initiate their suppressive effects, but once this condition has been met, their ensuing suppression can act in a non-antigen-specific way, in part through IL-2 sequestration. 24 Therefore, expansion of CD4 þ regulatory T cells at high levels after LCMV infection of TTR-NP mice resulted in both antigen-specific and antigenunspecific suppression of anti-LCMV responses, as seen by the lower numbers of polyfunctional GP-and NP-specific CD4 þ and CD8 þ T cells in chronically infected TTR-NP mice.
Interestingly, there is an increased level of GP 33-41 -specific CD8 þ T cells in the liver compared with the spleen at 55 days after LCMV infection. This suggests a greater retention of these cells in the liver as opposed to the spleen where GP 33-41 -specific CD8 þ T cell levels decline significantly compared with the levels found at day 8. This increased retention likely stems from a combination of high local viral burden and an inability of NP 396-404 T cells to exert an efficient cytotoxic response, leading to a shift of T-cell immunodominance in latter stages of infection in these animals.
The expression levels of NP in TTR-NP mice also contributed to the observed LCMV persistence because hemizygous TTR-NP 1/0 mice did not become chronically infected after LCMV infection. Recently, Tay et al 17 suggested that low expression levels of a liver-antigen led to the activation of CD8 þ T cells with full effector functions whereas high levels induced functional exhaustion. This suggests that high NP expression in TTR-NP mice may have led to exhaustion of NP-specific CD8 þ T cells and could explain why the CD4 þ regulatory T-cell silencing/depletion of infected TTR-NP, though leading to improved CD8 þ T cell responses and viral clearance, did not restore CD8 þ T-cell functionality up to B6 levels.
In addition to their classic roles of maintaining immunologic tolerance to self and preventing autoimmunity, CD4 þ regulatory T cells can also inhibit immune responses against pathogens. In HCV infection, the increased Treg numbers correlate with higher viral burdens and increased disease activity, 25 and progression to persistent infection correlates with an expansion of Gal-9-expressing regulatory T cells. 26 In nonhuman primates, subinfectious exposure to HCV suppressed the T-cell responses against subsequent HCV challenge through the generation of regulatory T cells. 27 Therefore, although the NP expression in transgenic TTR-NP mice is artificial, expression of HCV viral proteins in presence of low inflammation levels occurs and can lead to the generation of regulatory T cells able to hinder the antiviral immune response.
Regulatory T cells in TTR-NP mice express the Helios transcription factor at higher levels than in B6 mice.
Although initially believed to distinguish natural Tregs from peripherally induced Tregs, the expression of the Helios transcription factor has been recently reported in both subsets 28 and is now believed to reflect the context of stimulation during FoxP3 induction. 29 These observations preclude using Helios as a marker of thymic Tregs, but its expression is indicative of recent antigenic stimulation, 29 suggesting that regulatory T cells of infected TTR-NP mice might be more responsive to viral and/or endogenous NP than those of B6 mice.
CD4 þ FoxP3 þ regulatory T cells actively secreted IL-10, and its levels dramatically decreased after Treg depletion, leading to viral clearance. Interestingly, IL-10 promoter polymorphisms can predict the initial response to IFN-a treatment of HCV patients, with high producers of IL-10 responding poorly to treatment. 30 Moreover, early IL-10 predominant responses have been associated with an increased progression to chronic HCV infection in injecting drug users. 31 Infection with acute LCMV strains is not associated with T-cell exhaustion. 32 In TTR-NP mice, expression of the T-cell exhaustion marker PD-1 on CD4 þ and CD8 þ T cells correlated directly with LCMV titers after Treg depletion and viral clearance (Pearson's correlation, R 2 ¼ 0.86, P ¼ .008, and R 2 ¼ 0.88, P ¼ .0006, respectively), suggesting that PD-1 expression by T cells is likely secondary to chronic infection and high antigen levels. Thus, the PD-1 expression is triggered by TCR signaling; so long as LCMV persists, PD-1 expression on LCMV-specific T cells will be sustained. 33 As a result, we speculate that Treg-mediated tolerance against one viral antigen (NP) significantly impaired antiviral immune responses, leading to sustained viral loads and high PD-1 expression levels on T cells. This would explain why PD-1 expression is high on both GP and NP-specific T cells after establishment of chronic LCMV infection and decreases upon viral clearance after Treg depletion. This is in agreement with recent findings by Penaloza-MacMaster et al 6 that showed significantly reduced viral titers upon CD4 þ regulatory T-cell depletion and PD-L1 blockade in mice infected with the chronic LCMV strain Clone 13.
Anti-CD25 antibody-mediated silencing/depletion of Tregs in infected TTR-NP mice led to improved T-cell functionality and viral clearance. Anti-CD25 antibodymediated silencing/depletion of Tregs was chosen because LCMV chronicity requires high NP expression, which would be difficult to achieve using an alternative method such as depletion-inducible Foxp3DTR on a TTR-NP background.
In addition, the use of anti-CD25 has the benefit of potentially being translated more rapidly to clinical application because daclizumab, an anti-CD25 humanized antibody, is already approved for human use. 34 Daclizumab is used to treat acute-graft rejection by targeting CD25expressing effector T cells, 34 but it has also been shown to effectively deplete CD4 þ regulatory T cells in humans, improving immune responses after tumor-antigen vaccination in patients with metastatic breast cancer. 35 In conclusion, our results show that expression of a viral protein by hepatocytes can induce a state of active peripheral tolerance, mediated by IL-10-expressing Tregs, sufficient to limit an antiviral T-cell response and lead to the establishment of a chronic viral infection, which suggests that interventions aimed at depleting or silencing Tregs could be useful for the treatment of patients chronically infected with hepatotropic viruses.