A TLR9 agonist enhances the anti-tumor immunity of peptide and lipopeptide vaccines via different mechanisms

The toll-like receptor 9 (TLR9) agonists CpG oligodeoxynucleotides (CpG ODNs) have been recognized as promising adjuvants for vaccines against infectious diseases and cancer. However, the role of TLR9 signaling in the regulation of antigen uptake and presentation is not well understood. Therefore, to investigate the effects of TLR9 signaling, this study used synthetic peptides (IDG) and lipopeptides (lipoIDG), which are internalized by dendritic cells (DCs) via endocytosis-dependent and endocytosis-independent pathways, respectively. Our data demonstrated that the internalization of lipoIDG and IDG by bone marrow-derived dendritic cells (BMDCs) was not enhanced in the presence of CpG ODNs; however, CpG ODNs prolonged the co-localization of IDG with CpG ODNs in early endosomes. Surprisingly, CpG ODNs enhanced CD8+ T cell responses, and the anti-tumor effects of IDG immunization were stronger than those of lipoIDG immunization. LipoIDG admixed with CpG ODNs induced low levels of CD8+ T cells and partially inhibit tumor growth. Our findings suggest that CpG ODNs increase the retention of antigens in early endosomes, which is important for eliciting anti-tumor immunity. These results will facilitate the application of CpG adjuvants in the design of different vaccines.


CpG ODNs have different effects on the intracellular trafficking of peptides and lipopeptides.
Antigen capture is the first step required for cross-presentation, and antigen uptake mechanisms are closely linked with the intracellular trafficking and fate of the antigen 28 . To determine the internalization efficiency of the peptides and lipopeptides combined with or without CpG ODNs in BMDCs, FITC-conjugated IDG or lipoIDG was incubated with BMDCs in the absence or presence of CpG ODNs for 10, 30 and 120 min. The internalization of the FITC-conjugated peptides and lipopeptides was analyzed by flow cytometry after trypan blue was used to quench surface-associated fluorescence. As shown in Fig. 1, both IDG and lipoIDG were efficiently internalized by BMDCs in as early as 10 min. However, the internalization was not enhanced in the presence of CpG ODNs (Fig. 1). In addition, we found that FITC-conjugated RAH (specific CTL epitope derived from HPV16 E7 protein) could not be internalized in BMDCs even in the presence of CpG ODNs (data not shown). Subsequently, CpG ODNs are taken up and trafficked to endosomes via nonspecific endocytosis or receptor-mediated endocytosis [29][30][31] . To determine whether the internalization of IDG, lipoIDG and CpG ODNs by BMDCs similarly occurred, FITC-conjugated IDG and lipoIDG combined with Cy5-conjugated CpG ODNs were incubated with BMDCs for 10 and 30 min. We found that intracellular IDG, but not lipoIDG, efficiently colocalized with CpG ODNs in both wild-type (WT) and TLR9 knockout (KO) BMDCs ( Fig. 2A). Our previous study demonstrated that IDG, but not lipoIDG, was internalized by BMDCs and was detected in early endosomes 6 . To further determine whether the intracellular localization of the peptides and lipopeptides could be regulated by CpG ODNs, BMDCs were incubated with FITC-conjugated IDG or lipoIDG in the absence or presence of CpG ODNs. Then, an antibody (Ab) against early endosome antigen-1 (EEA1) was used to detect early endosomes. After the incubation of IDG alone with BMDCs, internalized IDG was detected in early endosomes at 10 min but decreased at 30 min. By contrast, most of internalized IDG could be detected at 30 min in the presence of CpG ODNs (Fig. 2B). However, lipoIDG was not localized in early endosomes, and the presence of CpG ODNs did not change the localization (Fig. 2B).
Scientific RepoRts | 5:12578 | DOi: 10.1038/srep12578 To confirm the retention of IDG in early endosomes in the presence of CpG ODNs, BMDCs that were derived from TLR9 KO mice were used. We found that the retention of IDG in early endosomes was reversed in TLR9 KO BMDCs (Fig. 2C). Taken together, these findings suggest that CpG ODNs did not facilitate the internalization of peptide and lipopeptide antigens in BMDCs but increased the retention of peptides in early endosomes. The divergent effects of CpG ODNs on peptides and lipopeptides may lead to different adjuvant functions.
The peptides and lipopeptides did not enhance the CpG ODN-induced activation of BMDCs. Our previous data indicated that mono-palmitoylated long peptides (lipoIDG) and non-lipidated long peptides (IDG) were taken up by DCs using endocytosis-independent and endocytosis-dependent pathways, respectively. Therefore, we asked whether the co-administration of peptides that are internalized via different mechanisms with CpG ODNs would affect TLR9 ligation-induced DC activation. In this study, BMDCs were incubated with peptides alone or with peptides mixed with CpG ODNs for 18 h. The data revealed that the peptides (RAH, IDG or lipoIDG) alone did not induce the up-regulation of the co-stimulatory molecule CD40. The co-administration of the peptides with CpG ODNs did not enhance the up-regulation of CD40 compared with CpG ODNs administered alone (Fig. 3A). Accordingly, cytokine (TNF-α , IL-6 and IL-12p70) production was induced by CpG ODNs when administered alone; however, no significant differences were observed in the presence of the peptides (Fig. 3B,C). Furthermore, CpG ODN-induced cytokine production was dependent on the TLR9 Figure 1. Internalization efficiency of the peptides and lipopeptides with or without CpG ODNs. The cellular internalization activities at different time points were determined by flow cytometry after DCs were treated with FITC-conjugated IDG or lipoIDG (1 μ g/ml) (green) combined with or without CpG ODNs (10 μ g/ml). FITC-conjugated IDG or lipoIDG combined with or without CpG ODNs was incubated with BMDCs for 10, 30 and 120 min. After trypan blue quenching, the internalization of the peptides and lipopeptides by CD11c + cells was analyzed using flow cytometry. Dead cells were gated out by propidium iodide staining. These data are representative of three independent experiments. and MyD88 pathway, because the activation effect was not observed when the BMDCs from TLR9 and MyD88 KO mice were treated with CpG ODNs (Fig. 3B,C). These observations suggest that the co-administration of CpG ODNs with different peptides that are internalized into APCs by different mechanisms did not affect TLR9-induced DC activation.
CpG ODNs dramatically enhance peptide-induced, but not lipopeptide-induced, CTL responses. Previous studies have found that mixtures of protein antigens and CpG ODNs have substantial effects, such as tumor rejection and heightened Th1-biased immunity 14,15 . Therefore, we assessed whether the prolonged retention time of IDG in endosomes, which was induced by CpG ODN, could affect the efficiency of cross-priming to CD8 + T cells. The mice were injected subcutaneously (s.c.) once with PBS, IDG or lipoIDG in the presence or absence of CpG ODNs. After 7 days, splenocytes were collected and stimulated with the RAH peptide or an irrelevant peptide to detect IFN-γ -secreting cells in an ELISPOT assay. Surprisingly, we found that the IFN-γ -secreting CD8 + T cells were significantly increased when IDG, but not lipoIDG, was combined with CpG ODNs (Fig. 4A). To further confirm the observation that CpG ODNs enhance antigen-specific CD8 + T cells in vivo, the splenocytes from the immunized mice were stained with a PE-conjugated RAH-tetramer. The data indicated that the co-administration of IDG and CpG ODNs resulted in increased numbers of antigen-specific CD8 + T cells compared with the co-administration of lipoIDG and CpG ODNs (Fig. 4B). These data suggest that CpG ODNs can enhance IDG vaccination-induced CD8 + T responses in contrast to lipoIDG vaccination. We further investigated whether the induced CD8 + T cells were capable of killing cells in vivo. The mice were immunized with PBS, RAH, IDG or lipoIDG in the presence or absence of CpG ODNs. After 7 days, 5-carboxyfluorescein diacetate succinimidyl ester (CFSE)-labeled target cells were adoptively transferred to the immunized mice for 18 h. A large amount of peptide-pulsed cells were killed in the mice that were immunized with IDG and CpG ODNs (IDG + CpG). However, a mild killing effect was observed in the mice that were immunized with lipoIDG and CpG ODNs (lipoIDG + CpG). The peptides alone (RAH, IDG, and lipoIDG) could not kill the target cells in the immunized mice. Notably, the mice that were immunized with RAH and CpG ODNs (RAH + CpG) did not exhibit any killing ability (Fig. 4C). The specific lysis percentages were 0%, 92.1 ± 5% and 10.5 ± 28.5% in the RAH + CpG, IDG + CpG and lipoIDG + CpG groups, respectively (Fig. 4D). Because antigen processing and presentation is necessary to prime CTL responses using IDG and lipoIDG, but not RAH, CpG ODNs may be involved in the regulation of antigen processing and presentation. These results indicate that CpG ODNs increased the processing and presentation of non-lipidated peptide antigen.  TC-1 tumor-bearing mice were used. Seven days after the TC-1 cell inoculations, the mice were s.c. injected with peptides in the presence or absence of CpG ODNs. As shown in Fig. 5A, none of the mice that were immunized with PBS or peptide alone exhibited tumor growth inhibition. However, the mice that were immunized with lipoIDG + CpG displayed significantly delayed tumor growth. Furthermore, the mice that were immunized with IDG + CpG displayed completely inhibited tumor growth (Fig. 5A).

CpG
To determine whether TLR9 engagement is necessary for tumor growth inhibition, tumor-bearing TLR9 KO mice were treated with IDG + CpG or lipoIDG + CpG. Figure 5B shows that the anti-tumor effects of IDG + CpG and lipoIDG + CpG were lost. In addition, the anti-tumor response was lost with IDG + CpG and lipoIDG + CpG vaccination in TC-1 tumor-bearing MyD88 knockout mice ( Supplementary Fig. 1). Furthermore, we detected the Ag-specific CD8 + T cells and myeloid-derived suppressive cells (MDSCs) at the tumor site after different treatments. We found that the Ag-specific CD8 + T cell numbers were slightly higher in the IDG + CpG-treated group than in the other groups ( Supplementary Fig. 2A). Interestingly, tumor-bearing mice immunized with IDG + CpG exhibited a significantly decreased number of MDSCs (Supplementary Fig. 2B). Therefore, the combination of IDG and CpG leads to the suppression of immunosuppressive cells in the local tumor environment and induces strong anti-tumor immunity. These results demonstrate that immunization with IDG combined with CpG ODNs provided robust and durable CTL responses and elicited superior anti-tumor activity compared with lipoIDG.
LipoIDG + CpG ODN induced strong anti-tumor effects; however, lipoIDG + CpG ODNs did not increase antigen-specific T cell responses and induced only low levels of CTL activity (Fig. 4). We speculated that lipoIDG + CpG may induce non-specific anti-tumor immunity. Therefore, we synthesized lipidated long peptides that were derived from OVA aa (lipoEQL), which contained an H-2K b -restricted CTL epitope (SII peptide), to treat TC-1 tumor-bearing mice. As shown in Fig. 5C, TC-1 tumor growth was not inhibited in the lipoEQL + CpG group. These data demonstrated that the IDG + CpG immunization induced dramatic antigen-specific CTL responses and anti-tumor immunity. However, the lipoIDG + CpG immunization induced mild CTL responses and unknown anti-tumor immunity.

Discussion
The cross-presentation of exogenous antigens by APCs on MHC class I molecules is crucial for the development of CD8 + CTL responses against tumors and infectious pathogens 32,33 . DCs are now known to be the most potent APCs and can cross-present exogenous antigens to stimulate T cells after being internalized during macropinocytosis, phagocytosis, or receptor-mediated endocytosis 19 . Evidence has indicated that TLR9 ligands increase the antigen cross-presentation activities of B cells, conventional DCs and plasmacytoid DCs [34][35][36][37] . However, the detailed regulatory mechanisms that underlie this effect are not well understood. In this study, we found that the TLR9 ligands CpG ODNs enhance endocytosis-dependent antigen cross-presentation (via a vacuolar pathway), but not endocytosis-independent antigen cross-presentation (via a cytosolic pathway), and increase endocytosis-dependent antigen retention in early endosomes. To confirm our observations in TLR9-rich plasmacytoid DCs (pDCs), we adoptively transferred IDG + CpG-pulsed BMDCs and IDG + CpG-pulsed Flt3L-cultured DCs (pDCs) into tumor-bearing mice. We found that the cross-presentation efficiency and the anti-tumor effects did not vary between the IDG + CpG-treated BMDCs and the pDCs (Supplementary Fig. 3).
In contrast to TLR2 and TLR4, which recognize invariant foreign bacterial and viral constituents at the cell membrane, TLR9 is endosomally expressed in innate immune cells and recognizes distinct patterns of nucleic acids at endosomal compartments 29 47 . According to the present study, the uptaken CpG ODNs are colocalized with IDG in both WT DCs and TLR9 KO DCs ( Fig. 2A). In addition, the long antimicrobial peptides LL-37 and KLKL 5 KLK have been demonstrated to facilitate the uptake of CpG ODNs and to enhance anti-tumor immunity 26,27 . Similar to the KLKL 5 KLK peptide, lipoIDG interacts with the membrane and is internalized by non-phagocytic cells 6 . However, neither lipoIDG nor IDG enhanced the uptake of CpG ODNs (Supplementary Fig. 4) or the activation of DCs (Fig. 3) in this study. Furthermore, early or recycling endosomes have been linked to signaling by TLR9 48 , and MHC class I was located in the recycling endosome compartment 49 . These studies suggested that MHC class I may access exogenous antigen and translocate to the cell surface after TLR9 activation. Therefore, we speculated that TLR9 signaling-induced endosome recycling may increase the interaction between the antigen and MHC class I molecules, resulting in IDG retention in the endosome, thereby eliciting strong anti-tumor immunity.
APCs internalize antigens and generate peptides via lysosomal proteases. In contrast to macrophages, which contain high levels of lysosomal proteases and rapidly degrade internalized antigens, DCs are protease-poor and therefore efficiently accumulate, process, and disseminate antigens 50 . In this study, we directly observed that intracellular CpG ODNs efficiently colocalize with IDG according to the confocal microscopy analysis (Fig. 2A). In addition, CpG ODNs prolonged the retention of IDG in early endosomes (Fig. 2B). The mechanisms of cellular trafficking may play an important role in CpG ODN-induced CD8 + T cell responses. In contrast, the different mechanisms of cellular trafficking may have resulted in the limited effects of lipoIDG immunization on CpG ODN-enhanced CTL responses that were observed in this study. Moreover, certain antigens led to the generation of peptides and to the cross-presentation of peptide-loading complexes by cathepsin S-dependent and TAP-independent pathways in endocytic compartments 39,51 . We speculated that this increased retention may increase the cross-presentation of antigens to MHC class I molecules, thereby eliciting strong anti-tumor immunity. Similar observations have been reported, which suggests that prolonged antigen survival and retention in early endosomes is important for cross-priming CTLs 52,53 . The detailed mechanisms by which CpG ODNs regulate antigen retention in early endosomes require further study. These results suggest that CpG ODNs may not be appropriate for all types of antigens to induce CTL responses and that the efficiency of CpG ODNs may be correlated with the mechanism of antigen internalization. In summary, we demonstrated that CpG ODNs induce different levels of CTL responses and anti-tumor responses when combined with different modified peptide immunizations. CpG ODNs significantly improved the anti-tumor responses of mice following immunization with a long peptide, which had a similar cellular trafficking pathway. However, the anti-tumor responses of mice were limited following immunization with a monolipopeptide, which had a different cellular trafficking pathway. These results are important for the future application of CpG ODN adjuvant in different vaccines.

Peptide synthesis. Peptides and lipopeptides (
Animals. Female C57BL/6 mice, 6-12 weeks of age, were obtained from the National Laboratory Animal Breeding and Research Center (Taipei, Taiwan). TLR9 knockout (KO) and MyD88 KO mice were purchased from Oriental Bioservice (Tokyo, Japan). All of the animals were housed at the Animal Center of the National Health Research Institutes (NHRI). All of the animal studies were conducted in accordance with the protocol approved by the Institutional Animal Care and Use Committee of the NHRI.

Culture of BMDCs.
BMDCs were harvested as previously described 54 . Briefly, BM cells from the C57BL/6 mice, the TLR9 KO mice and the MyD88 KO mice were cultured at a density of 2 × 10 5 cells/ml in petri dishes that contained 10 ml of complete RPMI-1640 medium with 200 U/ml (20 ng/ml) recombinant mouse granulocyte-macrophage colony-stimulating factor (GM-CSF; PeproTech Inc., New Jersey, USA). On day 3, another 10 ml of complete RPMI medium that contained 20 ng/ml GM-CSF was added. On day 6, cells from each dish were collected, washed and counted.

Internalization of peptides.
BMDCs were incubated at 37 °C in complete culture medium with FITC-conjugated peptides (IDG) or lipopeptides (lipoIDG) (1 μ g/ml) combined with or without 10 μ g/ ml CpG ODN 1826 (GeneDireX, Nevada, USA). At different time points, the cells were harvested for intracellular fluorescence analysis. The harvested DCs were stained with a PE-labeled anti-CD11c antibody (Ab) (eBioscience San Diego, CA, USA) for 1 h after blocking Fc receptors using an anti-mouse CD16/CD32 Ab (BD mouse Fc-block ™ ). The internalization of the peptides and lipopeptides by CD11c + cells was analyzed using a FACSCalibur flow cytometer (BD Biosciences, USA). In the cell samples, surface-associated fluorescence was quenched using a method that was adapted from Busetto et al. 55 , which involved adding an equal volume of 0.1 M citrate buffer (pH 4.0) that contained 250 g/ml trypan blue (Biological Industries, Israel) to each sample and incubating the samples for 1 min on ice before the flow cytometry analysis. Dead cells were gated out by staining with 1 μ g/ml propidium iodide. Detection of peptide-specific CD8 + T cells. C57BL/6 mice were immunized subcutaneously (s.c.) once with 1 μ g of peptide mixed with or without 10 μ g of CpG adjuvant. After one week, splenocytes were harvested, and the response of IFN-γ -secreting cells was determined by ELISPOT after 48 h of peptide stimulation. Briefly, 2 × 10 5 splenocytes were incubated with 1 μ g/ml irrelevant peptide or RAH peptide in an anti-IFN-γ -coated polyvinylidene fluoride (PVDF) plate for 48 h. After incubation, the cells were removed, and a biotinylated anti-IFN-γ Ab (eBioscience, San Diego, CA, USA) was added to each well. The plates were incubated at 37 °C for 2 h. Following the addition of the avidin-HRP reagent (eBioscience, CA, USA), the assay was developed using a 3-amine-9-ethyl carbazole (AEC; Sigma-Aldrich, MO, USA) staining solution. The reaction was stopped after 4-6 min by placing the plate under tap water. The spots were counted using an ELISPOT reader (Cellular Technology Ltd., Shaker Heights, OH, USA). For RAH-specific T cell staining, spleens were harvested seven days after the immunizations, and RAH-specific CD8 + T cells were detected by tetramer staining using a PE-labeled RAH tetramer (Beckman Coulter, CA, USA) and a FITC-labeled anti-CD8 monoclonal antibody (mAb) (eBioscience, CA, USA). The stained RAH-specific CD8 + T cells were analyzed by flow cytometry.
In vivo cytolysis assay. The C57BL/6 mice were single injected with different vaccines (peptides mixed with or without CpG ODNs). After 7 days, 5-carboxyfluorescein diacetate succinimidyl ester (CFSE)-labeled target cells were adoptively transferred to the mice. The target cells were prepared from syngeneic splenocytes that were treated with red blood cell (RBC) lysis buffer (BioLegend) for 2 minutes to remove RBCs, and the cells were subsequently divided into two populations. Each population was pulsed with either 5 μ g/ml irrelevant peptide or RAH peptide for 30 min at 37 °C. After the splenocytes were washed in PBS, irrelevant and RAH peptide-pulsed cells were labeled at a final concentration of 1 μ M or 10 μ M of CFSE (Molecular Probes, Eugene, OR, USA) for 15 minutes at 37 °C. The splenocytes were added to ice-cold complete RPMI medium to stop CFSE labeling. The irrelevant and RAH peptide-pulsed splenocytes were re-suspended in PBS and remixed at a ratio of 1:1. Then, 2 × 10 7 CFSE-labeled cells were adoptively transferred via tail vein injection into the immunized mice. All of the experimental cells were harvested 18 h after adoptive transfer and analyzed using the FACSCalibur flow cytometer (Becton Dickinson Immunocytometry Systems, San Jose, CA, USA).
Animal studies. To assess the therapeutic value of the vaccines, tumors were first generated by injecting 2 × 10 5 TC-1 cells into the abdominal region of the mice. Seven days later, the mice were s.c. vaccinated in a separate abdominal region with 1 μ g of non-lipidated peptide (IDG) or lipopeptide (lipoIDG/ lipoEQL) mixed with or without 10 μ g of CpG ODNs at 200 μ l/dose. The tumor volumes were monitored over a 50-day post-tumor implantation period and were calculated according to the following formula: (length × width 2 )/2.