Combination of TLR1/2 and TLR3 ligands enhances CD4+ T cell longevity and antibody responses by modulating type I IFN production

Despite the possibility of combining Toll-like receptor (TLR) ligands as adjuvants to improve vaccine efficacy, it remains unclear which combinations of TLR ligands are effective or what their underlying mechanisms may be. Here, we investigated the mechanism of action of L-pampo, a proprietary adjuvant composed of TLR1/2 and TLR3 ligands. L-pampo dramatically increased humoral immune responses against the tested target antigens, which was correlated with an increase in follicular helper T cells and the maintenance of antigen-specific CD4+ T cells. During the initial priming phase, in contrast to the induction of type I interferon (IFN) and pro-inflammatory cytokines stimulated by polyI:C, L-pampo showed a greatly diminished induction of type I IFN, but not of other cytokines, and remarkably attenuated IRF3 signaling, which appeared to be critical to L-pampo-mediated adjuvanticity. Collectively, our results demonstrate that the adjuvant L-pampo contributes to the promotion of antigen-specific antibodies and CD4+ T cell responses via a fine regulation of the TLR1/2 and TLR3 signaling pathways, which may be helpful in the design of improved vaccines.

L-pampo adjuvanticity modulates the antibody response against the HBV antigen toward a Th1-type response. To examine whether the adjuvant effect of L-pampo would also apply to a different protein antigen, L-HBsAg, which includes HBV surface antigens S and preS proteins, was used with the same series of adjuvants as described above. Similar to the increased OVA-specific antibody response by L-pampo, L-HBsAg-specific IgG production was the highest in the mice given the L-pampo adjuvant (Fig. 2a). In this system, Pam3 strongly induced a substantial production of HBs-specific or PreS-specific antibodies with a relatively low IgG2c/IgG1 ratio, whereas polyI:C did not induce antibody titers as well as Pam3 but still had a higher IgG2c/ IgG1 ratio than Pam3 (Fig. 2a-c). When L-HBsAg was combined with L-pampo, the ratio of IgG2c/IgG1 was as high as it was when combined with polyI:C (Fig. 2c). The induction of high titers of Th1-polarized antibodies against L-HBsAg would be beneficial because Th1-type isotypes are closely related to cellular immunity, which is critical to defense against cell-residing viruses.
Interestingly, a synergistic expansion of GC B cells was observed in the L-pampo-adjuvanted group, which implies that L-pampo adjuvanticity most likely induces a better functional antibody response, such as the production of antibodies with higher affinity, despite the similar antibody titers observed in Pam3 adjuvanticity (Fig. 2d).
Pam3 and L-pampo promote the expansion of Tfh cells in the periphery, but only L-pampo adjuvanticity prominently maintains OVA-specific CD4 + T cells during the memory phase. To dissect the underlying mechanism for the enhanced antibody responses by L-pampo, we analyzed OVA-specific CD4 + T cell responses over time, including the effector and memory phases. To monitor the responses, OT-II cells were adoptively transferred prior to OVA immunization. OT-II cells in the blood of each adjuvant group expanded well at the effector phase and gradually contracted until day 46 after immunization ( Fig. 3a). However, OT-II cells from the L-pampo group expanded the most and remained in the blood at significantly high levels even at day 46 and in the spleen at day 48 (Fig. 3a,e).
To examine antigen-specific CD4 + T cell and B cell responses more closely, the lymphocytes in the spleens were analyzed at day 8 post-immunization. Tfh cells are a specialized subtype of CD4 + T cells used for B cell help; therefore, they are critical for enhancing antibody responses 24 . CXCR5 expression is a signature that enables Tfh to localize in the B cell zone and form germinal centers by interacting with cognate B cells 25 . A considerable expansion of PD-1 + CXCR5 + Tfh cells and GL7 + CXCR5 + germinal center-Tfh (GC-Tfh) cells was observed in mice immunized with OVA plus Pam3 or L-pampo (Fig. 3b,c). With regard to Th1 differentiation, OT-II cells from the polyI:C group had the highest frequency of IFN-γ + cells, but the number of IFN-γ + OT-II cells was generally similar among each adjuvant group (Fig. 3d). Taken together, our results indicate that both L-pampo and Pam3 adjuvanticity expands and polarizes more Tfh cells than the other adjuvants at the effector phase, but L-pampo is the only adjuvant that significantly supports the longevity of OVA-specific CD4 + T cells. At day 29 post-immunization, the highest titers of OVA-specific antibodies and isotypes, particularly the Th1-type, were observed in the mouse serum from the L-pampo group (Fig. 3f,g). This could be related to the high number of OT-II cells in the L-pampo group. That is, OVA-specific CD4 + T cells from the L-pampo group were significantly maintained and thus likely supported antibody responses well.
L-pampo attenuates type I IFN production and IRF signaling induced by polyI:C but amplifies AP-1/NFκB signaling. To investigate the factors that support the differentiation and maintenance of CD4 + T cells, we analyzed the effect of L-pampo on the in vivo profiles of secreted cytokines and signaling molecules in the TLR pathway in the first 24 hours, which could be a critical time point for the onset of T cell differentiation. As expected, polyI:C produced a substantial amount of IFN-α and IFN-γ , whereas Pam3 produced little IFN-α and IFN-γ but a significant amount of pro-inflammatory cytokines such as TNF-α in both the serum and the spleen (Fig. 4a). Surprisingly, when polyI:C and Pam3 were combined (L-pampo), a significantly attenuated level Figure 1. L-pampo is a potent adjuvant that enhances the production of OVA-specific antibodies and expands germinal center B cells upon tertiary immunization with OVA. Naïve B6 mice were immunized with 100 μ g of OVA alone or in combination with alum, polyI:C, Pam3, or L-pampo as adjuvants, followed by the same immunization 3 and 6 weeks after the first immunization. (a) Individual mouse sera were isolated at the indicated time points after the first immunization, and OVA-specific IgG levels were measured with an ELISA. (b) At the 7 th week, the indicated isotypes were measured with an ELISA. (c) The relative ratios of OVA-specific IgG2c to IgG1 were calculated as an indirect measure of T helper 1 (Th1) and Th2 type immune responses. (d) Seven weeks after the first immunization, splenocytes from each mouse were analyzed by flow cytometry. Representative FACS plots and the frequencies and numbers of germinal center B cells (gated as live B220 + CD19 + GL-7 + PNA + lymphocytes) in the spleen of each mouse are shown. Data are represented as the mean ± SEM with each dot indicating one mouse (n = 5). The data are representative of two independent experiments. *p < 0.05; **p < 0.01; ***p < 0.001.

Figure 2. L-pampo is a potent adjuvant that modulates the mode of antibody responses and expands germinal center B cells upon tertiary immunization with the HBV surface protein.
Naïve B6 mice were immunized with 40 μ g of L-HBsAg, an HBV surface protein, alone or in combination with alum, polyI:C, Pam3, or L-pampo as adjuvants, followed by the same immunization 3 and 6 weeks after the first immunization. (a) Individual mouse sera were isolated at the indicated time points after the first immunization, and HBs-or PreSspecific IgG was measured by an ELISA. (b) At 7 weeks, the indicated isotypes were measured by an ELISA. (c) The relative ratios of HBs-or PreS-specific IgG2c to IgG1 were calculated. (d) Seven weeks after the first immunization, splenocytes were analyzed by flow cytometry. Representative FACS plots and the frequencies and numbers of germinal center B cells (gated as live B220 + CD19 + GL-7 + PNA + lymphocytes) in the spleen are shown. Data are represented as the mean ± SEM with each dot indicating one mouse (n = 4). *p < 0.05; **p < 0.01; ***p < 0.001. Scientific RepoRts | 6:32526 | DOI: 10.1038/srep32526 of IFN-α but a similar amount of IFN-γ and TNF-α were observed in the serum and spleen compared to the amounts in the polyI:C group (Fig. 4a).
To verify the L-pampo-mediated attenuation of type I IFN production in TLR signaling molecules, CD11c + conventional dendritic cells (cDCs) enriched in vivo were isolated and treated ex vivo with polyI:C, Pam3, or L-pampo for 0.5, 1, or 3 h. The up-regulation of phosphorylated TBK1 and IRF3 observed in polyI:C-treated cDCs was significantly attenuated after 3 h of stimulation with L-pampo, reaching levels observed in the attenuated IFN-α production above (Fig. 4a,b). Likewise, a similar effect was observed in a macrophage cell line, Raw264.7, confirming that the activation of the TBK1-IRF3-type I IFN axis by polyI:C is considerably attenuated by Pam3 in L-pampo-treated Raw264.7 (Fig. 4c). However, signaling molecules such as phosphorylated JNK, Iκ B, and p38 that convergent to activate NFκ B or AP-1 were generally amplified in both cDCs and Raw264.7 cells when polyI:C and Pam3 were combined as L-pampo (Fig. 4c). Thus, L-pampo modulates TLR signaling pathways triggered by polyI:C or Pam3 alone, thereby leading to a different innate cytokine environment in which the production of type I IFNs is selectively attenuated.
It has been well understood that adjuvants promote the maturation of APCs by up-regulating co-stimulatory molecules such as CD40, CD80, or CD86 together with major histocompatibility complex (MHC) molecules, leading to a better priming of T cells 3,26 . In this regard, we analyzed the subtype-specific maturation status of DCs

Figure 3. Pam3 and L-pampo promote the expansion of follicular helper T cells in the periphery, but only L-pampo adjuvanticity prominently maintains OVA-specific CD4 + T cells during the memory phase.
Naïve Ly5.2 + B6/J mice were injected with Ly5.1 + OT-II cells (5 × 10 5 cells each) before the immunization and immunized with 100 μ g of OVA alone or in combination with alum, polyI:C, Pam3, or L-pampo as adjuvants. (a) Mouse PBMCs were isolated from the blood, and the frequency of the transferred OT-II cells (CD4 + CD44 + Ly5.1 + ) among CD4 + T cells was analyzed kinetically by flow cytometry at the indicated time points after immunization. (b-d) Eight days after the immunization, splenocytes were isolated and analyzed by flow cytometry. (b) Representative FACS plots and the frequencies and numbers of the transferred OT-II cells among CD4 + T cells (gated as CD4 + CD44 + Ly5.1 + ). (c) Representative FACS plots and the frequencies of PD-1 + CXCR5 + follicular helper T cells (Tfh) or GL-7 + CXCR5 + germinal center Tfh (GC-Tfh). (d) Concatenated FACS plots of cytokine-producing OT-II cells and the frequencies and numbers of IFN-γ + cells among the OT-II cells. (e) Forty-eight days after immunization, splenocytes were isolated and analyzed by flow cytometry. Representative FACS plots and the frequencies and numbers of the transferred OT-II (CD4 + CD44 + Ly5.1 + ) cells among CD4 + T cells. (f,g) Mouse serum was isolated at the indicated time points, and OVA-specific IgG (f) and the indicated isotypes at day 29 (g) were determined by an ELISA. The data are represented as the mean ± SEM with each dot indicating one mouse (n = 3-4). The data are representative of two independent experiments. *p < 0.05; **p < 0.01; ***p < 0.001. (a) Naïve B6 mice were immunized with 100 μ g of OVA alone or in combination with alum, polyI:C, Pam3, or L-pampo as adjuvants. The spleen and serum of each immunized mouse were collected at the indicated time points, and the levels of IFN-γ , TNF-α , and IFN-α were determined using a multiplex analysis. The level of cytokines detected in the spleen supernatant was divided by the weight (g) of the spleen of that mouse. n.d., not detected. The data are represented as the mean ± SEM (n = 3); *p < 0.05; **p < 0.01; ***p < 0.001. (b,c) CD11c + conventional dendritic cells (cDCs) isolated from flt3L-expressing-B16F10-injected mice (left column) or cells from the mouse macrophage cell line Raw264.7 (right column) were treated with polyI:C, Pam3, or L-pampo for the indicated time; western blots were performed to quantify proteins in the TBK1-IRF3 signaling pathway (b) or the AP-1/NFκ B signaling pathway (c). The data are representative of two independent experiments. by profiling the expression of co-stimulatory molecules and MHC molecules during the first 24 h after immunization (Fig. S1a). Indeed, the co-injection of polyI:C, Pam3, or L-pampo along with OVA induced a more potent upregulation of CD86 in lymphoid and myeloid DCs than an OVA immunization alone. The expressions of CD80, CD40, and MHC class II molecules were also slightly enhanced in lymphoid DCs in adjuvant-treated groups (Fig. S1b). However, no enhanced expression of such molecules was observed in DCs in the L-pampo group compared to the polyI:C or Pam3 group (Fig. S1b).
L-pampo has the ability to generate a high quantity and quality of memory CD4 + T cells. To analyze the boosting efficacy of each adjuvant, the expansion and cytokine-producing abilities of OVA-specific CD4 + T cells were examined after the boost. Notably, the expansion of OT-II cells in the blood was the most prominent in the L-pampo group (Fig. 5a). Compared to the other groups, the L-pampo group had the highest frequency and the most abundant numbers of OT-II cells in the spleen after the boost (Fig. 5b). Moreover, although the levels of IFN-γ produced (% IFN-γ + among OT-II cells) in the L-pampo group were comparable to those produced in the polyI:C group, the percentage of multi-cytokine-producing OT-II cells (TNF-α + IFN-γ + among IFN-γ + cells) was significantly higher than in the other groups (Fig. 5c,d).
The boost efficacy was far more amplified by the second boost. The spleens of the recipient mice were analyzed one week after the third immunization. OT-II cells from the L-pampo group expanded synergistically compared to those from the polyI:C or Pam3 groups (Fig. 5e). Moreover, the IFN-γ -producing capability of OT-II cells was higher in the L-pampo group than in the other groups, and their multi-cytokine-producing ability (TNF-α + IFN-γ + among IFN-γ + cells) was also superior to that observed in the other groups (Fig. 5f,g). Considering the number of OT-II cells in the L-pampo group and their ability to produce multi-cytokines together, the overall production of the cytokines should be much higher. This was verified by a cytokine ELISA of IFN-γ , IL-6, and TNF secreted from splenocytes and by an ELISpot assay of IFN-γ -and IL-4-secreting cells among splenocytes, which were ex vivo stimulated with OVA 323-339 peptides (Fig. 5h) or OVA (Fig. S2a,b).
The level of expansion of the transferred OVA-specific CD8 + T cells (OT-I cells) from the L-pampo group was a little lower than that of the polyI:C group (Fig. S3a), but their functional capability to produce IFN-γ and multi-cytokines was superior to the other groups, as determined by the frequency of TNF-α + IFN-γ + among IFN-γ + cells and their intensity of IFN-γ expression after ex vivo stimulation (Fig. S3b). The amount of IFN-γ and TNF-α secreted from the splenocytes and the spot numbers of IFN-γ -secreting splenocytes, when stimulated with the OVA 257-264 peptide, appeared to be similar between the polyI:C and L-pampo groups, but IL-6 secretion was higher than in the L-pampo group than in the other groups (Fig. S3c,d). Therefore, L-pampo adjuvanticity appears to enhance the expansion of multi-functional antigen-specific T cells, particularly CD4 + T cells, compared to polyI:C or Pam3 administered alone.

Discussion
In this study, we report an efficacious combination of TLR ligands that synergistically enhances the antibody responses against two different protein antigens. As a mechanism of action, we propose that the combination of polyI:C and Pam3 results in a cytokine environment that is distinct from the environment resulting from the use of polyI:C or Pam3 alone, exhibiting a prominent maintenance of antigen-specific CD4 + T cells during the memory phase. The expansion of functional CD4 + T cells is likely amplified during each round of boosting, leading to the significant expansion of GC B cells and the production of isotype-switched antibodies in mice immunized with OVA and L-pampo.
First, we examined the changes in a signaling pathway resulting from the combination of Pam3 and polyI:C. PolyI:C and Pam3 both inhibit and synergize together to form a unique cytokine environment. When poly-I:C and Pam3 are combined, the robust phosphorylation of TBK1 and IRF3 by polyI:C was inhibited by Pam3, whereas that of NFκ B and AP-1 signaling molecules was synergistically enhanced, both in primary CD11c + cells and in a macrophage cell line. These findings are consistent with a previous report that showed that the levels of some interferon-stimulated genes (ISGs) and ifn-b mRNA were reduced when mouse bone marrow-derived dendritic cells (BMDCs) were treated with polyI:C and Pam3 27 . Mechanistically, the polyI:C-mediated induction of type I IFNs has been reported to be directly inhibited by TLR1/2 adaptor proteins such as MyD88 and MyD88 adaptor-like (Mal) or negatively regulated by the NFκ B subunits RelB and cRel via transcriptional repressor protein YY1 in bone marrow-derived macrophages (BMDMs) and BMDCs [28][29][30][31] .
At the effector phase, we showed that the expansion of Tfh was higher in L-pampo-and Pam3-treated groups. Recently, it was shown that type I IFN acts as a repressor of signal transducer and activator of transcription 3 (STAT3), thereby limiting Tfh differentiation in a virus infection model 32 . This report partially provides a link between the attenuated type I IFN production and expansion of Tfh. Because of the importance of Tfh in forming and supporting the GC response, the expansion of Tfh by L-pampo and Pam3 is thought to contribute to the enhancement of antibody titers and isotype-switching later.
Interestingly, what makes L-pampo different from Pam3 and other adjuvants is its ability to adequately maintain antigen-specific CD4 + T cells during the memory phase. Because factors affecting the maintenance of antigen-specific CD4 + T cells are relatively under-studied compared to those for CD8 + T cells, many questions remain as to whether the maintenance of antigen-specific CD4 + T cells is programmed when they are primed at the early phase, sustained by a putative continuous signal during the late phase, or both. If the former is true, the early cytokine environment might have a role in programming maintenance. In our experimental system, different profiles of type I IFN and IFN-γ secretion were observed among the groups during the early phase. In fact, the role of type I IFN and IFN-γ in CD4 + T cell homeostasis is controversial depending on the conditions. A recent review describes important roles of type I IFN for the survival of CD4 + T cells depending on the context of TCR stimulation and the mode of infection 33 . A direct requirement of type I IFN for the survival of CD4 + T cells is not observed in bacterial immunization 34 . Similarly, direct IFN-γ signaling is reported to enhance CD4 + T cell memory in a virus infection model 35 , whereas the opposite effect is reported for CD8 + T cells in a peptide immunization model 36 . Therefore, further studies are required to determine the complex effects of type I IFN and IFN-γ on the expansion, differentiation, and maintenance of CD4 + T cells.
Upon several rounds of boosting, multi-functional antigen-specific CD4 + T cells that were thought to be amplified and further matured from the remaining memory CD4 + T cells were more abundant in mice immunized with OVA and L-pampo compared to the other groups. The multi-cytokine-producing CD4 + T cells are critical for clearing or reducing low levels of viruses in response to infection with various types of viruses, including HIV, influenza virus, Epstein-Barr virus (EBV), varicella-zoster virus (VZV), or cytomegalovirus (CMV), and correlating protection against a second challenge with L. major 37 . Furthermore, a higher quality of CD8 + T cells producing multi-cytokines was observed in the L-pampo group compared to those in the other groups despite a similar number of antigen-specific CD8 + T cells in the polyI:C group. Therefore, L-pampo enhances the quality of antigen-specific T cells upon boosting.
When L-pampo is used for clinical application, the toxicity issue should be carefully considered as a high-dose of pathogen-associated molecular patterns (PAMPs) could induce cytokinemia. Indeed, it is reported that TLR3 mediates the acute inflammation of skin after exposure to ultraviolet B (UVB) radiation 38 . Recently, phosphorothioate ODN-guided dsRNA (sODN-dsRNA), which triggers selectively TLR3 with no ability to activate RIG-I/ MDA5, is reported to be less toxic because of its capability to induce less pro-inflammatory cytokines but mount comparable anti-tumor cellular immunity compared to polyI:C that triggers both TLR3 and RIG-I/MDA5 pathways 39 . If we need a higher dose of polyI:C which is efficacious but could be toxic, TLR3-selective ligand seems to be a good alternative to reduce the toxicity.
Similar to our observation on the synergy between TLR ligands as an adjuvant, some studies have also shown that combinations of TLR ligands can provide advantages as vaccine adjuvants. The combinations provided more Scientific RepoRts | 6:32526 | DOI: 10.1038/srep32526 protection against pathogens than the constituents given individually. When rhesus monkeys were vaccinated with inactivated pandemic influenza virus mixed with MPL (TLR4 ligand) and R837 (TLR7 ligand), a robust humoral immune response was synergistically induced 5 . In this system, adjuvants activating three TLR ligands via MPL and R848 (TLR7 and TLR8 ligands, respectively) induced a greater antibody response, even greater than what is required for protection after a boost 5 . Indeed, the triple combination of polyI:C, MALP2 (TLR2/6 ligand), and CpG (TLR9 ligand) vaccinated with the HIV envelope peptide showed a remarkable protection against HIV envelope protein-expressing VV compared to that observed for the dual combinations 6 . In this context, our study provides an efficacious TLR ligand combination, composed of the TLR1/2 and TLR3 ligands, that synergistically enhances the quality of antibodies and T cells as well as the quantity of antibodies and CD4 + T cells, which have implications for efficacious vaccine development. Adoptive cell transfers. Congenically marked (CD45.1) splenic CD4 + T cells specific to the OVA 323-339 epitope of OVA from naive OT-II transgenic mice were isolated using the Magnisort TM Mouse CD4 T Cell Enrichment Kit (eBioscience), and 5 × 10 5 of the isolated OT-II cells were transferred intravenously into naïve B6 (CD45.2) mice before immunization. In some experiments, splenic CD8 + T cells specific to the OVA 257-264 epitope of OVA from naïve OT-I transgenic mice were isolated separately alongside the OT-II cells, and 5 × 10 5 of the isolated OT-I/OT-II cells from each mouse (total 1 × 10 6 /mouse) were intravenously co-transferred into naïve B6 (CD45.2) mice before immunization.

Methods
Immunization. The preparation of L-HBsAg and the formulation of L-pampo (Korean patent 10-0900837), which is a mixture of pam3csk and polyI:C, was conducted as previously described 16 . Briefly, the entire coding region of the HBV envelope gene (pre-S1-pre-S2-S) was ligated into the pMSG vector (Korean patent 10-2000-0043996) and transfected into CHO cells to obtain high amounts of L-HBsAg, including L, M, and S proteins. L-HBsAg was purified from the culture medium of the recombinant CHO cell line that showed the highest expression level of L-HBsAg 16 .
In vivo cytokine secretion. After immunization, the supernatants from homogenized spleens and serum were collected at the indicated time points. A multiplex analysis of IFN-γ , IFN-α , TNF-α , and IL-6 was assayed using Flowcytomix (eBioscience) according to the manufacturer's instructions. The samples were read using a FACS Canto II (BD) and analyzed using Flowcytomix Pro software (eBioscience).
In vivo DC generation and cell isolation. Splenic dendritic cells were expanded in the mice as previously described 41 . Briefly, B6 mice were injected subcutaneously with 5 × 10 5 Flt3L-secreting B16F10 melanocytes in Hank's Balanced Saline Solution (HBSS). After 2 weeks, the spleens were collected, digested by collagenase, and filtered through a mesh before magnetic isolation. Single-cell suspensions of the prepared splenocytes were pre-enriched by magnetic depletion of B and T cells with microbead-conjugated antibodies to CD19 (Miltenyi Biotec) and CD90.2 (Miltenyi Biotec), respectively. The depleted cells were then subjected to magnetic selection of CD11c + conventional dendritic cells using CD11c microbeads (Miltenyi Biotec) to > 90% purity.
Cell culture. The mouse macrophage cells and RAW 264.7 cells were cultured in high glucose Dulbecco's Modified Eagle's medium (DMEM; Welgene) supplemented with 5% fetal bovine serum and 1% antibiotics. The cell line was maintained at 37 °C under an atmosphere of 5% CO 2 .