Synergy of therapeutic heterologous prime-boost hepatitis B vaccination with CpG-application to improve immune control of persistent HBV infection

Therapeutic vaccination against chronic hepatitis B must overcome high viral antigen load and local regulatory mechanisms that promote immune-tolerance in the liver and curtail hepatitis B virus (HBV)-specific CD8 T cell immunity. Here, we report that therapeutic heterologous HBcore-protein-prime/Modified-Vaccinia-Virus-Ankara (MVA-HBcore) boost vaccination followed by CpG-application augmented vaccine-induced HBcAg-specific CD8 T cell-function in the liver. In HBV-transgenic as well as AAV-HBV-transduced mice with persistent high-level HBV-replication, the combination of therapeutic vaccination with subsequent CpG-application was synergistic to generate more potent HBV-specific CD8 T cell immunity that improved control of hepatocytes replicating HBV.


Results
In high-titer HBV transgenic mice, CpG-injection improves immune control after heterologous prime-boost vaccination. We used HBV-transgenic mice as a well-characterized preclinical model of HBV persistence 15 to evaluate whether the combination of therapeutic heterologous prime-boost vaccination directed against HBcAg synergizes with CpG-treatment to control HBV infection. We used HBV-transgenic mice with intermediate to high HBsAg levels (50 to 450 IU/ml), where prime-boost vaccination fails to elicit strong HBVspecific CD8 T cell immunity 10 . CpG-application always led to formation of iMATEs, in wildtype C57Bl/6 mice, in HBV-transgenic mice or in vaccinated HBV-transgenic mice with high-level HBV replication (Fig. 1a). First, we determined the optimal time point for CpG application after heterologous prime-boost vaccination. CpGapplication within three days after heterologous prime-boost vaccination, but not later, resulted in increased numbers of vaccination-induced HBcAg-specific CD8 T cells (Suppl. Fig. 1, for gating strategy see Suppl. Fig. 2), which led us to apply CpG within the first three days after vaccination in all further experiments (see scheme in Fig. 1b). Since HBcAg-specific CD8 T cells are key for control of HBV infection 10,16 , we determined numbers of vaccination-induced HBcAg-specific CD8 T cells in the presence or absence of CpG-application. We detected an increase in the numbers of HBcAg-specific CD8 T cells after heterologous prime-boost vaccination in HBVtransgenic mice and a further increase upon CpG-application (Fig. 1c). Of note, we also found HBcAg-specific CD8 T cells in the spleen, but these cells were not further increased in numbers after CpG-application (Fig. 1d). This finding suggested hepatic expansion of already activated HBcAg-specific CD8 T cells generated by vaccination, but not local priming of naïve CD8 T cells after CpG application 14 . We found expression of Ki67, a marker for cell proliferation, in hepatic mononuclear cells within iMATEs after CpG-injection by immunohistochemistry in vaccinated mice (Suppl. Fig. 3a). Furthermore, we detected Ki-67 expression in the majority of hepatic CD8 T cells at d3 after CpG injection by flow cytometry (Suppl. Fig. 3b). Consistent with previous reports 14, 17 , application Figure 1. Intravenous CpG-injection leads to iMATE-formation in HBV-transgenic mice and expansion of HBcAg-specific CD8 T cells in the liver. (a) H&E staining of liver tissue slices detecting iMATEs at d3 after CpG-injection (white circles). (b) Time scheme for therapeutic heterologous prime boost vaccination against HBcAg and CpG-injection. (c) HBcAg-specific CD8 T cells from liver detected by HBc-specific multimerstaining using flow cytometry at d3 after CpG-injection, numbers denote percentage of total liver CD8 T cells. (d) Absolute numbers of total and HBcAg-specific CD8 T cells in liver and spleen at d3 after CpG-injection. Bars in (d) indicate mean value of n ≥ 3 mice per group + SEM. Statistical analyses were performed using Kruskal-Wallis test with Dunn's multiple comparison correction. Asterisks mark statistically significant differences: *p < 0.05; ns -not significant; n.d. -not detectable www.nature.com/scientificreports www.nature.com/scientificreports/ of CpG resulted in increased expression of costimulatory molecules like CD80 and CD86 as well as increased MHC-II expression in infiltrating monocytes (Suppl. Fig. 4a).
Following heterologous prime-boost vaccination and subsequent CpG-injection in HBV-transgenic mice, we observed development of increased serum ALT levels compared to mice receiving vaccination alone, indicating increased HBcAg-specific CD8 T cell immunity in the liver (Fig. 2a). Consistent with previous results 10 , in the presence of high HBV expression levels in HBV-transgenic mice, heterologous prime-boost vaccination alone did not reduce serum HBeAg levels, neither did CpG application alone (Fig. 2b). After sequential application of heterologous prime-boost vaccination and CpG injection, however, we found a significant decline in serum HBeAg-levels (Fig. 2b). These findings suggested that HBcAg-specific CD8 T cells locally expanded in the liver within iMATEs after CpG-injection, and controlled HBV replication in infected hepatocytes. We next determined the functionality of these HBcAg-specific CD8 T cells from liver or spleen following ex vivo restimulation. HBcAg-specific CD8 T cells isolated from liver or spleen of HBV-transgenic mice after heterologous prime-boost vaccination expressed IFNγ following re-stimulation with an HBcore peptide (Fig. 2c). After www.nature.com/scientificreports www.nature.com/scientificreports/ sequential application of vaccination and CpG, numbers of IFNγ-producing HBcAg-specific CD8 T cells were 4-fold increased, as were numbers of GzmB-expressing CD8 T cells, now constituting almost 30 percent of all hepatic CD8 T cells (Fig. 2c,d). Of note, vaccine-induced CD8 T cells expressed higher levels of GzmB but not IFNγ per cell (Fig. 2e) suggesting enhanced cytotoxic effector function.
Consistent with enhanced effector function and release of anti-viral cytokines of HBcAg-specific CD8 T cells, liver immune-histochemistry revealed that sequential application of vaccination together with CpG but not vaccination alone led to a 6-fold reduction in numbers of HBcAg-positive hepatocytes (Fig. 3a,b). The relatively low sALT values (see Fig. 2a) support the notion that reduction of HBcore-positive hepatocytes may also have been achieved by non-cytolytic activity of cytokines. Together, these results indicated that CpG-application acted in a synergistic fashion with therapeutic vaccination to augment both, numbers and functionality of vaccine-induced HBcAg-specific CD8 T cells in HBV-transgenic mice.

Control of infection in high-titer AAV-HBV infected mice after heterologous prime-boost vaccination followed by CpG injection.
Since elimination of an HBV-transgene from hepatocytes is not possible, we employed AAV-HBV infection to deliver HBV genomes into murine hepatocytes in vivo 18 , which leads to persistent HBV replication in hepatocytes 18 . This preclinical model system of persistent HBV replication can be used to characterize whether vaccine-induced HBV-specific CD8 T cells successfully eliminated HBV-replicating hepatocytes in vivo. After AAV-HBV-transduction, high serum HBeAg and HBsAg levels were detected, that remained stable until the end of the observation period twelve weeks later indicating persistent HBV replication (Fig. 4a). These HBsAg levels (200-1200 IU/ml) were comparable to the intermediate to high antigen level groups previously reported 10 where heterologous prime-boost vaccination fails to elicit strong HBV-specific CD8 T cell immunity. Of note, no anti-HBs or anti-HBe antibodies were detected in these mice before start of vaccination (Suppl. Fig. 5a). In AAV-HBV-transduced mice, CpG-application triggered iMATE-formation independent of heterologous prime-boost vaccination (Fig. 4b). As in HBV-transgenic mice, therapeutic vaccination induced HBcAg-specific CD8 T cells in AAV-HBV transduced mice (Fig. 4c,d). Consistent with an hepatic expansion of vaccine-induced T cells, we found more HBcAg-specific CD8 T cells in the livers but not spleens in AAV-HBV transduced mice after sequential application of heterologous prime-boost vaccination and CpG injection, although this did not reach statistical significance (Fig. 4c,d). Expansion of CD8 T cells after CpG injection in AAV-HBV transduced mice was confirmed by detection of Ki67 expression in hepatic CD8 T cells (Suppl. Fig. 3c,d). Compared to prime-boost vaccination in HBV-transgenic mice, that yielded only very few HBcAg-specific CD8 T cells (see Fig. 1), we detected more HBcAg-specific CD8 T cells in AAV-HBV-transduced mice after vaccination (Fig. 4c,d). This may be the result of the more stringent deletion of HBV-specific T cells through central tolerance in HBV-transgenic mice 19 , rather than changes in expression of co-stimulatory molecules on monocytes infiltrating the liver and forming iMATEs after CpG injection (Suppl. Fig. 4b). Again, CpG-application without prior vaccination, did not lead to any measurable increase in HBcAg-specific CD8 T cells (Fig. 4c,d).
Only the sequential application of heterologous prime-boost vaccination and CpG injection but neither CpG-injection nor vaccination alone resulted in increased sALT levels nine weeks after AAV-HBV transduction (Fig. 5a). Numbers of IFNγ-producing and GzmB-expressing HBcAg-specific CD8 T cells were higher after heterologous prime-boost vaccination in AAV-HBV infected mice, were further increased after CpG-injection, www.nature.com/scientificreports www.nature.com/scientificreports/ and were higher in liver than in spleen (Fig. 5b,c). Furthermore, GzmB but not IFNγ levels per HBcAg-specific CD8 T cell were significantly higher when heterologous prime-boost vaccination was followed by CpG-injection (Fig. 5d). Thus, also in AAV-HBV-transduced mice with high-level HBV replication, CpG-application after vaccination increased both, numbers and of functionality of vaccine-induced HBcAg-specific CD8 T cells.
We then followed AAV-HBV infected mice after vaccination ± CpG-injection for five weeks. Heterologous prime-boost vaccination alone led to a transient decrease of serum HBeAg-levels that was not very much different from non-vaccinated mice (Fig. 6a). In contrast, when vaccination was followed by CpG-injection serum HBeAg levels remained low over the entire observation period (Fig. 6a). Similarly, HBsAg levels were controlled after the combination of vaccination and CpG-injection (Fig. 6a), even in the absence of seroconversion to anti-HBs or anti-HBe (Suppl. Fig. 5b,c). Liver immunohistochemistry revealed that shortly after CpG-injection in vaccinated AAV-HBV infected mice (d32 after initial start of vaccination) there was no difference in numbers of HBcAg-expressing hepatocytes compared to mice that did not receive CpG (Fig. 6b,c). However, more than four weeks later (i.e., d63 after start of vaccination) following sequential application of vaccination and CpG-injection, numbers of HBcAg-expressing hepatocytes were reduced by 80%, while no control of HBcAg-expressing hepatocytes was observed after vaccination alone (Fig. 6b,c). At this late time point, only GzmB-expressing HBcAg-reactive CD8 T-cells were still increased in AAV-HBV-infected mice that received a combinatorial treatment of therapeutic vaccination followed by CpG-injection (Fig. 6d), suggesting that detection of GzmB expression in virus-specific CD8 T cells may allow for prediction of immune control of HBV-replicating hepatocytes. www.nature.com/scientificreports www.nature.com/scientificreports/

Discussion
Here we report that therapeutic heterologous protein-prime/MVA-vector-boost vaccination against hepatitis B synergizes with CpG-application to enhance numbers and functionality of HBcAg-specific CD8 T-cells in presence of high HBV-antigen levels in two preclinical models of persistent HBV infection, i.e. the AAV-HBV model and HBV-transgenic mice. High viral antigen levels prevented formation and expansion of HBcAg-specific CD8 T cells, that are required to gain control of HBV infection 11 . Cytokines released from T cells like IFNγ/TNF control HBV-antigen expression through degradation of the HBV persistence form in infected hepatocytes 20,21 . Cytokines also exert non-cytolytic control of HBV antigen expression in preclinical models such as the HBV transgenic mouse 22 and may thereby account in part for the rapid decline of serum HBeAg-levels after vaccination in combination with CpG-injection, which leads to release of cytokines from T cells 14 .
Long-term control of persistent HBV-infection, however, relies on reduction of the numbers of HBV-infected hepatocytes, which may either be achieved through degradation of the HBV persistence form or elimination of infected and HBV-replicating hepatocytes. Here, we demonstrate that the combination of therapeutic heterologous prime-boost vaccination with CpG-injection led to strong reduction of the numbers of HBcAg-expressing hepatocytes together with local hepatic expansion of vaccination-induced HBcAg-specific CD8 T cells. This www.nature.com/scientificreports www.nature.com/scientificreports/ suggests that CpG-injection, which functions to induce formation of iMATEs and to locally expand already activated CD8 T cells in the liver by providing costimulatory signals 14 , can be employed to increase the efficacy of therapeutic vaccination against HBV even in the presence of high-level viral antigens, which otherwise prevent induction of protective immunity. However, it is also possible that cytokines, e.g. interferons or TNF, induced by CpG-injection, that are known to have anti-viral effects 20,21 , further support the therapeutic effect of the prime-boost vaccination. Since cytokines and in particular interferons are important for induction of CD8 T cells 23,24 , it is not possible to mechanistically dissect non-cytolytic cytokine-mediated from cytolytic control of HBV infection in this experimental setting of therapeutic vaccination in combination with CpG application. Induction of iMATEs through CpG-injection has also been shown to improve CD8 T cell immunity against high-level expression of a transgenic model antigen (ovalbumin) in the liver 25 , to increase anti-viral immunity during lymphocytic choriomeningitis virus infection 14 and even increase anti-tumor immunity against liver cancer 26 . Such induction of increased T cell immunity is linked to hepatic recruitment of inflammatory monocytes that are functionally distinct from liver-resident macrophages 14,17 .
The combination of heterologous prime-boost vaccination to generate HBcAg-specific CD8 T cells in the first place is therefore nicely complemented by the subsequent local hepatic increase of HBcAg-specific CD8 T cells by CpG-mediated iMATE induction. Thus, our approach to apply first therapeutic vaccination followed by CpG injection combines two synergistic principles of immune therapy, i.e. first increasing numbers of HBcAg-specific CD8 T cells through potent cross-priming in the setting of a therapeutic vaccination and second amplifying these vaccine-induced HBcAg-specific CD8 T cells locally in the liver by CpG-mediated iMATE induction. Thus, iMATE-induced expansion of vaccine-induced HBV-specific CD8 T cells a may provide a liver-specific mean to increase the efficiency of therapeutic vaccination beyond currently available strategies 27 .
Although we cannot provide formal evidence for the role of GzmB expressed in HBcAg-specific CD8 T cells for elimination of HBV-infected hepatocytes, our finding that GzmB levels are increased in HBcAg-specific CD8 T cells after successful control of HBV-infected hepatocytes supports the notion that GzmB could be employed as biomarker for immune control of HBV-infection in future immune monitoring strategies during therapeutic vaccination against chronic hepatitis B. GzmB has been identified as essential downstream effector molecule for T cells to control viral infection 28 . GzmB expression has been shown to be higher in CX 3 CR1 + CD8 T cells with strong effector function 29,30 . The identification of GzmB + HBV-specific CD8 T cells after therapeutic vaccination in combination with CpG injection therefore supports the notion that those T cells may have strong anti-HBV activity.
Taken together, our results demonstrate in two preclinical models of persistent HBV infection that the sequential combination of therapeutic vaccination followed by CpG-application was superior to vaccination alone in controlling persistent HBV infection by increasing numbers and functionality of HBcAg-specific CD8 T cells. Experimental animals and AAV-HBV transduction. HBV transgenic mice (strain HBV1.3.32) carrying 1.3 overlength HBV (genotype D, serotype ayw) genome were created on C57BL/6 background (haplotype H-2 b/b ). Fourteen to sixteen weeks old female and male HBV transgenic mice were bred at the AVM Animal Facility, Helmholtz Center Munich. Wildtype C57BL/6 mice (haplotype H-2 b/b ) were purchased from Charles River Laboratories, Schulzfeld, Germany. Persistent HBV replication in wildtype mice was established by intravenous injection of 1 × 10 10 genome equivalents (geq) of the AAV-HBV1.2 vector encoding 1.2-fold overlength HBV genome of genotype D, as reported prevsioulsy 18 . Mice were kept under pathogen-free (SPF) conditions at the Animal Facility, University Hospital Rechts der Isar, Technical University of Munich, or the Helmholtz Center Munich following institutional guidelines. Experiments were performed during the light phase of the day. Animals were bled one day before treatment and allocated into groups with comparable HBsAg and HBeAg levels.

Materials and Methods
Heterologous protein prime MVA boost vaccination. Mice were immunized with a particulate protein prime followed by recombinant Modified Vaccinia Ankara virus (MVA) vector boost vaccination scheme described previously 10 . Briefly, mice were immunized intramuscularly (i.m.) into the quadriceps muscles of both hind limbs in isoflurane anesthesia. Protein immunization with 10 µg HBcAg expressed in E. coli (APP Latvijas Biomedicinas, Riga, Latvia) adjuvanted with 10 µg cyclic di-adenylate monophosphate (c-di-AMP) (InvivoGen, San Diego, CA) was given twice at 2-week interval. Two weeks after the second protein immunization, mice received 1 × 10 7 particles of recombinant MVA expressing HBcAg intramuscularly (i.m.).

Isolation of lymphocytes from spleen and liver and non-parenchymal liver cells. Preparation
of single-cell suspensions of splenocytes was performed as described previously 31 . Liver-associated lymphocytes (LAL) were isolated by density gradient centrifugation as described 31 . Briefly, mouse liver was perfused with pre-warmed PBS (to flush blood from the hepatic vasculature) and forced through a 100 µm nylon cell strainer (BD Falcon, Franklin Lakes, NJ). After washing, cell pellets were suspended in 10 ml of prewarmed enzyme solution containing 1 mg/ml of collagenase type IV (Worthington, Lakewood, NJ) in RPMI 1640 medium supplemented with 10% fetal bovine serum (Gibco, Thermo Fischer Scientific, Darmstadt, Germany) and digested for www.nature.com/scientificreports www.nature.com/scientificreports/ 30 min at 37 °C. Cell pellets were then resuspended in 40% Percoll (GE Healthcare, Munich, Germany), layered on 80% Percoll solution and centrifuged at 1600 × g for 20 minutes without brakes for density separation. Non-parenchymal liver cells were obtained after liver perfusion, mechanical separation of liver tissue and gentle collagenase digestion for 10 minutes followed by percoll gradient centrifugation and washing steps as outlined in detail in 14 . Cell yield and viability were determined by trypan blue exclusion. Immunohistochemistry. Liver tissue samples were fixed in 4% buffered formalin for 48 h and paraffin-embedded. Then, 2-μm-thin liver sections were prepared with a rotary microtome (HM355S, ThermoFisher Scientific, Waltham, USA). Immunohistochemistry was performed using a Bond Max system (Leica, Wetzlar, Germany, all reagents from Leica) with primary antibodies against Ki67 (clone SP6, Abcam ab16667) or rabbit anti-HBcAg primary antibody (Diagnostic Biosystems, Pleasanton, CA; 1:50 dilution) and horseradish peroxide coupled secondary antibody. Briefly, slides were deparaffinized using deparaffinization solution, pretreated with Epitope retrieval solution 1 (corresponding to citrate buffer pH6) for 20 minutes. Antibody binding was detected with a polymer refine detection kit without post primary reagent and visualized with DAB as a dark brown precipitate. Counterstaining was done with hematoxyline. Slides were scanned using a SCN 400 slide scanner (Leica Biosystems) and HBcAg-positive hepatocytes were determined based on localization, intensity and distribution of the signal in random 10 view fields (40x magnification). The mean numbers of HBcAg-positive hepatocytes were quantified per mm 2 . statistical analyses. Statistical analyses were performed using GraphPad Prism version 5 (GraphPad Software Inc., San Diego, CA). Statistical differences were analyzed using Kruskal-Wallis test with Dunn's multiple comparison correction, 2-way ANOVA with Tukey's multiple comparison correction, Mann-Whitney test and unpaired or paired t test. P-values < 0.05 were considered significant.