Targeted antigen delivery to dendritic cells elicits robust antiviral T cell-mediated immunity in the liver

Hepatotropic viruses such as hepatitis C virus cause life-threatening chronic liver infections in millions of people worldwide. Targeted in vivo antigen-delivery to cross-presenting dendritic cells (DCs) has proven to be extraordinarily efficient in stimulating antigen-specific T cell responses. To determine whether this approach would as well be suitable to induce local antiviral effector T cells in the liver we compared different vaccine formulations based on either the targeting of DEC-205 or TLR2/6 on cross-presenting DCs or formulations not involving in vivo DC targeting. As read-outs we used in vivo hepatotropic adenovirus challenge, histology and automated multidimensional fluorescence microscopy (MELC). We show that targeted in vivo antigen delivery to cross-presenting DCs is highly effective in inducing antiviral CTLs capable of eliminating virus-infected hepatocytes, while control vaccine formulation not involving DC targeting failed to induce immunity against hepatotropic virus. Moreover, we observed distinct patterns of CD8+ T cell interaction with virus-infected and apoptotic hepatocytes in the two DC-targeting groups suggesting that the different vaccine formulations may stimulate distinct types of effector functions. Our findings represent an important step toward the future development of vaccines against hepatotropic viruses and the treatment of patients with hepatic virus infection after liver transplantation to avoid reinfection.


Results
Targeting antigen to DCs induces humoral immunity. We first compared the OVA-specific humoral immune response after immunization with either α DEC-205/OVA adjuvanted with Poly(I:C) and CpG (α DEC-205/OVA + Poly(I:C)/CpG; for simplification termed α DEC-205/OVA), BPPcysOVAMPEG or, in addition to the two DC targeting approaches, BPPcysMPEG co-administered together with soluble OVA (OVA + BPPcysMPEG). As controls we included α DEC-205 and OVA alone, both adjuvanted with Poly(I:C) and CpG as well as BPPcysMPEG alone, OVA peptides alone, and OVA peptides adjuvanted with BPPcysMPEG.
Already after two vaccinations with α DEC-205/OVA, we observed a strong OVA-specific IgG response ( Fig. 1A) that was significantly increased in comparison to both the OVA + BPPcysMPEG and the OVA + Poly(I:C)/CpG groups at this time (Fig. 1B). Mice vaccinated with the novel antigen targeting system BPPcysOVAMPEG initially exhibited a weaker IgG response, which was however strongly enhanced following the third boost ( Fig. 1A,C).
We next compared OVA-specific IgG 1 and IgG 2c titres which represent the main IgG isotypes stimulated by Th2 and Th1 cells, respectively 26 . Vaccination with either α DEC-205/OVA, OVA + Poly(I:C)/CpG or BPPcysOVAMPEG induced a rather balanced IgG 1 /IgG 2c ratio, whereas OVA + BPPcysMPEG induced substantially higher IgG 1 than IgG 2c titres, pointing towards a more Th2-dominated immune response (Fig. 1D). As expected, neither the injection of the adjuvant BPPcysMPEG alone, nor the vaccination with the immunodominant CD4 + and CD8 + OVA peptides without adjuvant did induce humoral immunity. Moreover, peptide targeting to DCs by means of vaccination with BPPcysOVAMPEG was superior in inducing antibody responses than vaccination with soluble peptides adjuvanted with BPPcysMPEG ( Supplementary Fig. S1A-C).
Induction of CTLs following αDEC-205/OVA and BPPcysOVAMPEG immunization. To further characterize the type of T cell response induced with the different vaccine formulations, OVA-specific IL-4-and IFNγ -secreting T cells were evaluated by ex vivo stimulation of splenocytes from immunized mice with the dominant OVA MHC-I and -II peptides. As expected, due to the exclusive selection towards these epitopes when immunizing with BPPcysOVAMPEG, the highest number of cytokine producing T cells was observed in this group ( Fig. 2A-C). BPPcysOVAMPEG vaccination induced increased numbers of IL-4-producing Th2 cells compared to the α DEC-205/OVA and OVA + Poly(I:C)/CpG immunized groups ( Fig. 2A). Compared to the vaccine formulations lacking direct antigen targeting to DCs, BPPcysOVAMPEG, and to a lesser extent α DEC-205/OVA immunization, at the same time showed the strongest potential to induce IFNγ -producing T cells (Fig. 2B,C). Of note, the adjuvant BPPcysMPEG alone or vaccination with non-adjuvanted immunodominant OVA peptides did not induce IFNγ -producing T cells. Furthermore, peptide targeting to DCs using the BPPcyOVAMPEG formulation induced significantly more IFNγ -producing T cells than vaccination of mice using peptides adjuvanted with BPPcysMPEG ( Supplementary Fig. S1D-G).
Since IFNγ -producing CTLs are important for virus clearance from the liver 27 , we evaluated OVA-specific CTL activity in vaccinated mice. Immunization with α DEC-205/OVA or BPPcysOVAMPEG induced a robust Scientific RepoRts | 7:43985 | DOI: 10.1038/srep43985 OVA-specific CTL response already after the first immunization (Fig. 2D). Despite the fact that IFNγ -producing CD8 + T cells were also induced in OVA + BPPcysMPEG immunized mice (Fig. 2C) no lysis of target cells was detectable in this group (Fig. 2D). Thus, while antigen delivery to DCs by targeting DEC-205 resulted in a rather weak and balanced Th1 (IFNγ ) and Th2 (IL-4) cell response, it was highly effective in generating antigen-specific CTLs. Similarly, BPPcysOVAMPEG-mediated antigen targeting to TLR2/6 induced balanced Th1/Th2 T helper and robust OVA-specific CTL responses.
The release of IFNγ by CTLs has been shown to mediate non-cytolytic clearance of infected hepatocytes and IFNγ per se directly inhibits viral replication in the liver 29,30 . While hardly any IFNγ production was detectable in the hepatic CD44 -CD8 + T cells (Fig. 3C), the AdOVA-GFP-luc challenge resulted in a nearly two-fold increase of the proportion of IFNγ -producing CD44 high CD8 + T cells in the OVA + BPPcysMPEG (11.3%) and Presence of IFNγ + CD44 high CD8 + effector T cells in the liver correlates with hepatitis development following AdOVA-GFP-luc challenge. We next asked whether the IFNγ -producing effector memory CD8 + T cells in the liver would respond to virus-expressed antigen on hepatocytes. While Ad-GFP-luc control infection induced only marginal hepatocellular cell death or cellular infiltrations (Fig. 4A), infection of all OVA-immunized groups with AdOVA-GFP-luc led to drastic histological lesions in the liver, hepatocellular cell death and marked infiltration of lymphocytes and macrophages into the tissue (Fig. 4A-C). Since histological scoring revealed a trend towards an enhanced immunopathology in the liver of α DEC-205/OVA immunized and AdOVA-GFP-luc infected mice compared to the BPPcysOVAMPEG group, we further elucidated whether α DEC-205/OVA and BPPcysOVAMPEG vaccination would result in distinct immune cell recruitment following viral infection. However, automated multidimensional fluorescent microscopy (MELC) on liver sections revealed no significant differences in the hepatic composition of leukocytes, macrophages, neutrophils, NK cells, B cells and T cell subsets among the groups (Fig. 5A,B and Supplementary Fig. S2). Altogether, the observed type of immunopathology in α DEC-205/OVA and BPPcysOVAMPEG immunized mice was indicative for acute hepatitis as a consequence of the immune-mediated clearance of virus infected hepatocytes suggesting that antigen-specific CTLs present in the liver following vaccination accounted for the hepatocyte damage.
Vaccination with αDEC-205/OVA and BPPcysOVAMPEG results in effective antigen-specific virus clearance in the liver. To further substantiate the cytolytic function of CTLs, serum concentrations of alanine transaminase (ALT) were determined in vaccinated and adenovirus-challenged animals. As expected, infection with the Ad-GFP-luc control virus did not affect serum ALT levels in any of the experimental groups (Fig. 6A). Strikingly, in α DEC-205/OVA as well as in BPPcysOVAMPEG immunized mice we observed a dramatic and comparably high increase of the serum ALT level (~950 U/l and ~770 U/I, respectively) indicating massive CTL-mediated killing of AdOVA-GFP-luc infected hepatocytes (Fig. 6A). Significantly lower ALT concentrations (~330 U/l) were measured in OVA + BPPcysMPEG immunized mice which is in line with lower abundance of IFNγ -producing memory CD8 + T cells (Fig. 3).
To further confirm specific killing of AdOVA-GFP-luc infected hepatocytes, virus clearance was examined by assessing luciferase expression in liver tissue. As expected, strong luciferase activity was detectable in the liver of all mice infected with the control Ad-GFP-luc virus (Fig. 6B). In contrast, luciferase activity in the livers of immunized and AdOVA-GFP-luc infected mice was significantly reduced in α DEC-205/OVA and BPPcysOVAMPEG vaccinated mice in comparison to both OVA + BPPcysMPEG and α DEC-205 + Poly(I:C)/CpG mock-immunized animals (Fig. 6B). Taken together, our results clearly demonstrated that antigen targeting to cross-presenting DCs by either α DEC-205/OVA or BPPcysOVAMPEG is highly effective at inducing effector T cells capable of recognizing and eliminating virus infected hepatocytes.

Vaccination with αDEC-205/OVA and BPPcysOVAMPEG induces distinct patterns of CD8 + T cell interaction with virus infected and apoptotic hepatocytes.
In order to gain first indication regarding the underlying mechanisms for the elimination of infected hepatocytes we analysed liver sections of α DEC-205/ OVA and BPPcysOVAMPEG immunized and AdOVA-GFP-luc infected mice for potential differences in the distance of CD8 + T cells to virus infected hepatocytes. While we did neither observe significant differences in the number of virus infected hepatocytes, nor in the number of hepatocytes staining positive for active caspase-3 (Fig. 5C,D) or in the frequency of CD8 + T cells being in loose contact with infected hepatocytes (> 22 μ m distance, Fig. 5E), we indeed observed significantly fewer CD8 + T cells being in direct contact (< 9 μ m) to AdOVA-GFP-luc infected hepatocytes in α DEC-205/OVA immunized mice compared to the BPPcysOVAMPEG immunized group (Fig. 5E). Strikingly, despite lower abundance of CD8 + T cells being in intimate contact with virus infected cells, substantially more of them were found to be closely associated with active caspase-3 and thus apoptotic cells in the liver of α DEC-205/OVA immunized mice (Fig. 5E). Together with the observation that compared to BPPcysOVAMPEG vaccination α DEC-205/OVA immunization resulted in the accumulation of significantly higher frequency of IFNγ + CD44 high CD8 + effector T cells in virus infected livers (Fig. 3C) these data may indicate that the two different DC-targeting formulations may indeed stimulate distinct types of effector functions being active during virus elimination from the liver.

Discussion
We show here that in vivo antigen delivery to CD8α + cross-presenting DCs is highly effective in inducing antiviral immunity in the liver. BPPcysOVAMPEG vaccination, which was included as a novel antigen targeting system, exhibited outstanding properties in inducing high antibody titres (Fig. 1), Th1 and Th2 cells, as well as CTLs (Fig. 2) that effectively cleared virus infected hepatocytes (Fig. 6). Although not reaching statistical significance compared to α DEC-205/OVA vaccination, this approach tended to promote a more robust viral clearance, as well as slightly less pronounced hepatocellular cell death and pro-inflammatory effect. The value of TLR agonist-antigen conjugates for inducing Th1 and CD8 + T cell responses are clearly underscored by our new data and previous studies 31,32 . However, for clinical application such conjugates will need to be tailored to match different HLA alleles and rapidly mutating viruses (e.g. HCV) 33 . It could be possible to exploit chemical groups of BPPcysMPEG to generate conjugates with more complex antigens or to engineer universal docking systems enabling non-covalent binding, such as those used in the past for antibodies 34 .
Prajeeth et al. have shown that BPPcysMPEG-mediated activation of CD8 + DCs led to effective cross-priming against co-administered antigen resulting in CTL responses in the spleen 12 . While we detected small numbers of CD44 + CD8 + memory T cells in the liver of OVA + BPPcysMPEG immunized mice following AdOVA-GFP-luc infection (Fig. 3), these cells did not exert effector functions (Figs 2 and 6). These discrepancies may be due to differences in the antigen concentration used (3 mg vs. 7 μ g) 12 . However, we have no explanation yet why the memory T cells present in the liver of OVA + BPPcysMPEG immunized mice did not exhibit cytotoxic function. Strikingly, despite the lack of viral eradication in the OVA + BPPcysMPEG vaccinated group, these mice displayed marked immunopathology and elevated ALT levels following AdOVA-GFP-luc infection. We may speculate that, although being functionally impaired in terms of cytotoxicity, antigen-specific recognition of the virus by IFNγ + CD44 + CD8 + T cells results in cytokine release followed by innate immune activation and immune cell recruitment that finally result in hepatic cell death and inflammation independent of CTL mediated killing.
Immunization with α DEC-205/OVA induced a robust CTL response (Fig. 2), which resulted in effective viral clearance in the liver (Fig. 6). Our data are well in line with a previous study demonstrating that vaccination with α DEC-205/OVA induces CD8 + T cells capable to protect against tumour or virus challenge 14 . Similar results were obtained by targeting tumour 24,25 or viral antigens 21,35 to DEC-205 + DCs. Next to the CTL response induced in α DEC-205/OVA immunized mice, we detected antigen-specific CD4 + T cells. While previous studies described a preferential induction of Th1 responses by α DEC-205/antigen immunization 16,23 , we observed similar levels of Th1 (IFNγ + ) and Th2 (IL-4 + ) CD4 + T cells ( Fig. 2A and B), consistent with minor differences in the IgG subclasses (Fig. 1C). Irrespectively of the antibody subclasses, we demonstrate that in comparison to the other formulations the two DC targeting approaches were most effective at inducing a rapid and robust humoral immune response (Fig. 1).
While in our hands both DC targeting approaches next to CTL responses induced a rather balanced Th1/Th2 profile, immunization with OVA + BPPcysMPEG that does not involve DC targeting failed to induce CTLs and resulted in a Th2 dominated cytokine profile (Figs 1C and 2). Interestingly, BPPcysMPEG has been shown to induce Th1-type immune responses in the context of allergens and parasite antigens [36][37][38] . However, intramuscular immunization with hepatitis B surface antigen virus-like particles co-administered with BPPcysMPEG resulted in a Th2 dominated response 39 . Therefore, the adjuvant dose, the route of administration and the intrinsic features of the antigen may account for the observed differences in the T helper cell phenotype. While Schulze and colleagues proved BPPcysMPEG to be a potent mucosal adjuvant for vaccination against severe acute respiratory syndrome coronavirus, the authors focused on the cellular and humoral immune responses, but did not test for antiviral activity 40 . Thus, BPPcysMPEG has not been investigated before regarding its potency to induce antiviral immunity. In the absence of DC targeting, i.e. by using the OVA + BPPcysMPEG formulation, we clearly show the induction of a Th2-biased CD4 + T cell phenotype, the lack of CTLs and in consequence no antigen-specific clearance of virus infected hepatocytes. While previous studies have demonstrated the exceptional potential of DEC-205-mediated antigen targeting to elicit adaptive immunity [14][15][16][17][18]24 to our knowledge this is the first study showing the induction of liver-specific immune reactions. We demonstrate that targeting DEC-205 + DCs resulted in a high frequency of local IFNγ + CD44 + CD8 + memory T cells (Fig. 3C), which correlate with protection against adenovirus challenge in the liver (Fig. 6). One critical point in fighting viral infections is to keep the balance between protective immunity and immunopathology, which are both mainly driven by CTLs and often decisive for the fate of the infected host 41,42 . For hepatotropic viruses such as HCV, which is not believed to be directly cytopathic, liver damage is attributed to T cell-mediated immunity 29 and immunopathology is inevitable for sterilizing immunity. On the other hand, a less vigorous CTL response allows for viral persistence, ultimately leading to progressive tissue injury 43 . Only in the α DEC-205/OVA and BPPcysOVAMPEG group could immunopathology be correlated with antigen-specific viral clearance in the liver (Fig. 6). Regarding the severity of hepatitis that occurs as a side effect of pathogen elimination, it is highly conceivable that α DEC-205/OVA and BPPcysOVAMPEG immunized mice would have recovered from liver damage. In a published model of fulminant hepatic failure, ALT levels peaked at day 4-5 post infection, which was the end-point of the experiments in our study, and had almost normalized by day 6 post infection 44 . However, further studies are needed to evaluate the consequences of immune-mediated hepatitis.
In terms of clinical application the use of well tolerated adjuvants is of foremost importance. While a number of clinical trials indicated that CpG is sufficiently well tolerated, Poly(I:C) was shown to exhibit severe side effects in humans 45 . Of note, its synthetic derivative, PolyICLC, which has already been successfully used in DEC-205 targeting trials 46-48 exhibits a better resistance to hydrolysis as well as a greater potency to induce IFNγ -secreting T cells 49 than Poly(I:C) and, most importantly, was proven a safe adjuvant in healthy human volunteers 50 and in cancer patients 51,52 . Additional studies are needed to clarify whether α DEC-205/OVA vaccination using PolyICLC/CpG as an adjuvant will be as efficient as (or even more efficient than) α DEC-205/OVA + Poly(I:C)/ CpG vaccination in inducing antiviral immunity in the liver. Regarding the TLR2/6 agonist BPPcysMPEG which is known to activate cross-priming CD8α + DCs in mice, so far no published data are available for clinical applications. It has recently been shown that human CD141 + DCs which are thought to represent the functional equivalent of mouse CD8α + DCs do express TLR2 and 6 53 . Thus, an interaction of BPPcysOVAMPEG with DCs via the TLR2/6 receptors is likely to take place also in humans. Nevertheless, future in vitro studies are needed to clarify this issue in more detail.
We observed distinct patterns of CD8 + T cell interaction with virus infected and apoptotic hepatocytes, dependent on the vaccine formulation we used (Fig. 5C-E). While significantly fewer CD8 + T cells were in intimate contact with virus infected hepatocytes in the α DEC-205/OVA-immunized group, these CD8 + T cells appear to be more effective in inducing apoptosis which is in line with the higher expression of IFNγ as a marker for their cytotoxic potential. It is well established that CTLs eliminate infected hepatocytes after direct antigen recognition on target cells via perforin/granzyme or Fas-mediated killing. However, cytotoxicity in virus infected hepatocytes can also be exerted via the non-canonical CTL effector function where CTLs are stimulated by cross-presenting liver sinusoidal endothelial cells and secrete TNF after stimulation 54 . TNF induces cell death specifically in infected hepatocytes involving caspase-3 activation following TNFR stimulation. Of note, well in line with the observed elevated frequencies of IFNγ + CD44 + CD8 + CTL in AdOVA-GFP-luc infected livers of α DEC-205/OVA compared to BPPcysOVAMPEG immunized mice, the overall IFNγ and as well TNF expression levels were by far higher in the α DEC-205/OVA treated group (Supplementary Fig. S3). IFNγ secreted by highly abundant IFNγ + CD44 + CD8 + CTL may account for elevated TNF expression in hepatic macrophages in AdOVA-GFP-luc infected α DEC-205/OVA immunized mice. Alternatively, as shown before by Wohlleber et al. 54 , virus-specific CTLs induced during α DEC-205/OVA may be the source of TNF. While further investigation is needed to decipher the specific mechanisms underlying the elimination of virus infected hepatocytes in α DEC-205/OVA and BPPcysOVAMPEG vaccinated mice, we may speculate that immune surveillance in α DEC-205/ OVA immunized mice in addition to cell-contact dependent CTL-mediated killing to a higher degree involves cytokine-mediated effector functions. At first line TNF may be the most important cytokine involved in elimination of virus infected hepatocytes as it can induce apoptosis in adenovirus infected hepatocytes. Although IFNγ has been shown to exert non-cytolytic antiviral effector function in HBV infected hepatocytes 55 , previous reports have shown that IFNγ ko mice behave exactly the same as wildtype mice with regards to antiviral immune response against adenovirus infected hepatocytes and following liver damage 28,54 . This does not exclude that high levels of IFNγ may enhance secondary antiviral effects for example by the upregulation of antigen-presentation by hepatocytes enhancing recognition of infected cells by CTLs.
The induction of antiviral immunity in the liver is per se a challenging issue. A hallmark of e.g. HCV persistence is the appearance of functionally impaired and exhausted T cells that are unable to secrete antiviral effector molecules, show impaired proliferation 29,56,57 and dysregulate expression of activating/inhibitory receptors [58][59][60] . Next to this, in patients chronically infected with HCV the higher frequency of suppressive CD4 + CD25 + T regulatory cells 61 most likely contribute to T cell dysfunction. Since blocking of PD-1, CTLA-4 and Tim-3 hold some therapeutic promise for the functional recovery of exhausted T cells 59,60,62 , a combined strategy encompassing both antibody-mediated blocking of immunosuppressive pathways together with boosting antiviral immunity by means of DEC-205-or TLR-mediated antigen delivery to cross-presenting DCs could represent a promising scenario for future developments. This may also imply the use of DC targeting as a therapeutic vaccine, which has been shown to be clearly effective in several other studies 14,21,25 , and might also contribute to the treatment of patients with hepatic virus infection after liver transplantation to avoid reinfection.

Methods
Mice. Female C57BL/6 mice were obtained from Harlan Winkelmann (Borchen, Germany) and housed under specific pathogen-free conditions according to the national and institutional guidelines. All experiments were approved by the local government agency (Niedersächsisches Landesamt für Verbraucherschutz und Lebensmittelsicherheit; file number 33.12-42502-04-10/0108) and have been performed in accordance to these guidelines.
Antibodies, antigens and cell lines. EndoGrade  Conjugation of OVA to αDEC-205. Chemical conjugation of purified α DEC-205 to OVA was performed as previously published 64 with minor modifications. In brief, α DEC-205 was first activated by sulfo-SMCC (sulfosuccinimidyl 4-[N-maleimidomethyl] cyclohexane-1-carboxylate) (Thermo Fisher Scientific, USA), according to the manufacturer's protocol, and in parallel the sulfhydryl-groups of the OVA protein were exposed by using 30 mM tris(2-carboxyethyl) phosphine hydrochloride (TCEP-HCl) (Thermo Fisher Scientific, USA) (1.5 hours at room temperature). Excess of TCEP and sulfo-SMCC in the respective samples was removed using Zeba Desalt Spin Columns, according to the manufacturer's recommendations (Thermo Fisher Scientific, USA). The reduced OVA was immediately mixed with the activated antibody and incubated overnight at 4 °C. The resulting α DEC-205/OVA conjugate was concentrated and separated from unbound OVA using 150 K MWCO Pierce ® concentrators (Thermo Fisher Scientific, USA). Quantification of the OVA content within the α DEC-205/OVA conjugate was achieved by comparison of OVA signal intensities on Western blot with graded quantities of OVA on the same blot ( Supplementary Fig. S4) 14  Immunization. Mice were subcutaneously immunized on days 0, 14 and 28 with either 30 μ g α DEC-205/ OVA, 30 μ g α DEC-205, 10 μ g BPPcysOVAMPEG 12 or 7 μ g OVA co-administered either with 50 μ g Poly(I:C) and 50 μ g CpG, as previously published 15,16,24,25 or 5-10 μ g BPPcysMPEG 39 in a total volume of 50 μ l PBS per animal.
Determination of serum ALT. For ALT quantification, 75 μ l peripheral blood were mixed with 25 μ l 1% heparin (Ratiopharm, Germany), centrifuged (10,600× g, 10 minutes, room temperature) and 32 μ l of the supernatant were used for detecting ALT activity using the scil Reflovet ® Plus reflection-photometer (scil animal care, Germany).

Detection of antigen-specific serum IgG.
To monitor the humoral immune response, 75 μ l of blood were collected from the retro-orbital sinus and serum was obtained through incubation of the samples for 45 minutes at 37 °C, followed by 45 minutes incubation at 4 °C and subsequent centrifugation (10 minutes at 420× g). Sera were assayed for the presence of antigen-specific IgG and IgG subclasses (IgG 1 , IgG 2c ) by enzyme-linked immunosorbent assay (ELISA) using 96-well Nunc-Immuno MaxiSorp plates (Nunc, Germany) coated with 2 μ g/ml OVA protein in 0.05 M carbonate buffer (pH 9.6). After overnight incubation at 4 °C, the plates were washed with PBS supplemented with 0.1% Tween 20 and blocked with 3% BSA in PBS for 1 hour at 37 °C. Serial 2-fold dilutions of sera in 3% BSA/PBS were added and plates were further incubated for 2 hours at 37 °C. After washing, antibody binding was detected using biotin-conjugated goat α -mouse IgG (Sigma, Germany), goat α -mouse IgG 1 (Southern Biotech, USA) and goat α -mouse IgG 2c (Southern Biotech, USA) antibodies (1 hour, 37 °C), respectively, and streptavidin-HRPO (BD Biosciences, Germany) (30 minutes, 37 °C). ABTS in 0.1 M citrate-phosphate buffer (pH 4.35) containing 0.03% H 2 O 2 was added to each well and the absorbance at 405 nm was recorded after 30 minutes of incubation. Endpoint titres were expressed as the reciprocal value of the last serum dilution which yielded an absorbance two times above the values of negative controls.

Enzyme-linked immunosorbent spot (ELIsPOT) assay. ELISPOT kits for the detection of murine
IFNγ (eBioscience, Germany) and IL-4 (BD Biosciences, Germany) were used according to the manufacturer's instructions. Spots were counted with an ELISPOT reader (C.T.L. Europe GmbH, Germany) and analysed using the ImmunoSpot image analyser software v3.2 (C.T.L. Europe GmbH, Germany). The results are presented as spot forming units per 10 6 cells.
In vivo cytotoxicity assay. Splenocytes were isolated from naive syngeneic donor mice and equal cell numbers were pulsed with 1 μ g/ml SIINFEKL (OVA 257-264 ) peptide for 30 minutes at room temperature or left untreated followed by labelling with 2.5 μ M (CFSE high ) or 0.25 μ M (CFSE low ) CFSE, respectively, and mixing in a one to one ratio. A total of 2 × 10 7 target cells was intravenously (i.v.) injected into recipient mice. After one day, splenocytes were re-isolated and the ratio of CFSE high to CFSE low labelled cells was determined by flow cytometry using a FACS Canto instrument (BD Biosciences, Germany). Adenovirus challenge assay. On day 42 after the first immunization, mice were infected intravenously with 2 × 10 8 plaque forming units (PFU) of a recombinant adenovirus expressing fusion proteins either of GFP and luciferase (Ad-GFP-luc) or GFP, luciferase and the OVA aa257-264 SIINFEKL (AdOVA-GFP-luc). Control animals received PBS and were either mock-infected (PBS) or infected with AdOVA-GFP-luc 28 . For luciferase activity assays, the lower liver lobe was separated into two parts, weighed and tissue fragments were homogenized in proportional volumes of reporter lysis buffer (Promega, Germany) using Lysing Matrix D (MP Biomedicals, Germany) and a FastPrep-24 instrument (3 × 10 sec, 5.5 m/sec). After centrifugation (3 minutes, 10,000× g, 4 °C), lysates were mixed with Luciferase Assay Reagent II (Promega, Germany) and measured in a luminometer (Berthold Technologies, Germany).

MELC analyses.
Liver tissues of α DEC-205/OVA + Poly(I:C)/CpG and BPPcysOVAMPEG immunized mice were snap frozen and embedded into O.C.T. (Sakura Finetek). Liver samples from α DEC-205/OVA and BPPcysOVAMPEG mice displaying comparable ALT levels were in each case analysed by the MELC robot in parallel. Cryosections of 10 μ m thickness, which adhere to silan-coated cover slides, were prepared on Leica Cryostat CM3050, fixed with 2% paraformaldehyde (Santa Cruz) and permeabilized with 0.2% Triton-X-100 before blocking with 1% BSA/PBS (Sigma) for 1 h. Detailed information on image acquisition and analysis are described in the Supplementary material. Optimal antibody dilutions, incubation times, and positions within the MELC experiment for all antibodies used (see Supplementary Table S1) were validated systematically using conditions suitable to MELC 65 (see Supplementary Figure S5).

Histological analyses.
The left upper liver lobe was fixed in 4% formaldehyde solution, followed by paraffin embedding, preparation of 2-4 μ m sections and staining with haematoxylin and eosin (H&E). Histological evaluation was performed by an animal pathologist certified by the American College of Veterinary Pathologists in a blinded fashion.
Statistical analyses. Results were statistically analysed by one-way Anova followed by the Dunnett's test, two-way RM Anova, non-parametric Mann Whitney test or the paired, two-tailed t-test using the Graph Pad Prism 5 software (Graph Pad software, La Jolla). Data are presented as mean ± SEM or mean ± SD and a p-value below 0.05 was considered significant.