Quillaja brasiliensis saponin-based nanoparticulate adjuvants are capable of triggering early immune responses

Commercially available saponins are extracted from Quillaja saponaria barks, being Quil A® the most widely used. Nanoparticulate immunostimulating complexes (ISCOMs or ISCOMATRIX) formulated with these, are able to stimulate strong humoral and cellular immune responses. Recently, we formulated novel ISCOMs replacing QuilA® by QB-90 (IQB-90), a Quillaja brasiliensis leaf-extracted saponin fraction, and reported that IQB-90 improved antigen uptake, and induced systemic and mucosal antibody production, and T-cell responses. However, its mechanism of action remains unclear. In this study we provide a deeper insight into the immune stimulatory properties of QB-90 and ISCOMATRIX-like based on this fraction (IMXQB-90). We show herein that, when used as a viral vaccine adjuvant, QB-90 promotes an “immunocompetent environment”. In addition, QB-90 and IMXQB-90 induce immune-cells recruitment at draining-lymph nodes and spleen. Subsequently, we prove that QB-90 or IMXQB-90 stimulated dendritic cells secret IL-1β by mechanisms involving Caspase-1/11 and MyD88 pathways, implying canonical inflammasome activation. Finally, both formulations induce a change in the expression of cytokines and chemokines coding genes, many of which are up-regulated. Findings reported here provide important insights into the molecular and cellular mechanisms underlying the adjuvant activity of Q. brasiliensis leaf-saponins and its respective nanoparticles.

vaccines of veterinary use 10 . Although these compounds are able to trigger strong cellular and humoral immune responses, their use in human vaccines has been restricted due to undesirable side effects, such as local reactions, haemolytic activity and systemic toxicity 10,11 . A more purified saponin fraction, also extracted from the bark of the Q. saponaria, QS-21 is currently considered "the gold standard" molecule in terms of saponin-based adjuvants 12 , is being assessed in clinical trials due to its ability in tailoring Th1-biased immune responses, leading to protection against cancers and intracellular pathogens [13][14][15][16] . However, the mechanism of action by which QS-21 activates cells remains unknown, and the signaling pathways it induces are also still poorly understood [2][3][4]12 .
Throughout the last decade our group has been studying the adjuvant properties of saponins from the leaves of Quillaja brasiliensis (A. St.-Hil. etTul.) Mart., a native tree from southern Brazil and Uruguay. Several leaf saponin fractions from Q. brasiliensis were found to share structural and biological activities with Q. saponaria 17-20 , inducing lower levels of toxicity than Quil A ® in mice, and strongly potentiating humoral and cellular immune responses to viral antigens 18,[20][21][22][23] . Indeed, these molecules were shown to up-regulate both Th1-and Th2-like immune responses and to promote a robust DTH reaction. QB-90 and QB-80, two of the Q. brasiliensis saponin fractions, were shown to increase secretion of Th1-associated cytokines (IFN-γ and IL-2) and antigen-specific IFN-γ production by CD4 + and CD8 + T cells 20,24 when used as vaccine adjuvants.
In order to reduce the undesirable haemolytic activity that invariably accompanies the adjuvant effect of saponins, different colloidal preparations which can also act as antigen delivery systems have been formulated 9,[25][26][27] . One of such preparations, called immunostimulating complexes (ISCOMs) 28 are 40 nm cage-liked self-assembled structures combining Quil A ® , cholesterol, phospholipids and antigen. A similar preparation, named ISCOMATRIX TM , is a vaccine adjuvant formulation which does not include the antigen [29][30][31] . The physical properties of ISCOM adjuvants contribute to antigen stability, reduce the haemolytic effects associated with saponins, interact with dendritic cells (DCs) and enhance cross-presentation of the incorporated antigen, generating both antibody and CD4 + and CD8 + T cell responses 2,29,31,32 . In summary, ISCOM and ISCOMATRIX TM vaccines are known to induce long-lasting antibody responses, a balanced Th1/Th2 response, and generation of cytotoxic T lymphocytes in mice 33,34 and humans 13,15,35 .
Recently, we reported an alternative ISCOM formulation replacing Quil A ® by QB-90 (IQB-90). This formulation with reduced haemolytic activity was efficiently uptaken in vitro by murine bone marrow-derived dendritic cells (BMDCs). Moreover, subcutaneously inoculated IQB-90 induced strong serum antibody responses to ovoalbumin, robust DTH reactions, significant T cell proliferation and increased Th1 (IFN-γ and IL-2) cytokine responses. Similarly, intranasally delivered IQB-90 elicited serum IgG and IgG1, and mucosal IgA responses at distal systemic sites, even with low antigen doses 34 . Despite the fact that ISCOMs and ISCOMATRIX TM have been studied for nearly 30 years, the mechanisms of action of such nanoparticles are still not clearly understood 2,31,36 .
The understanding of saponin-based adjuvants mode of action, and in particular of those derived from Q. brasiliensis leaves, is highly relevant, as they constitute a more readily renewable alternative source of saponins and also present reduced toxicity compared to Q. saponaria bark-derived saponins. Therefore, the aim of the present work is to provide a deeper insight into the molecular mechanisms of action of QB-90 and its nanoparticle formulations. We will focus on the interaction with innate immunity receptors, as well as in the characterization of early immune events, such as cell recruitment and cytokine/chemokine profile, triggered by Q. brasiliensis saponin-based adjuvant.

QB-90 saponins promote a transient immunocompetent environment lasting 24 hours regardless the adjuvant-antigen mixture.
In order to assess if the mechanism of action of QB-90 was independent of the presence of antigen, BVDV antigen was injected in mice into proximal as well as distal sites, either at the same time or 24 and 48 hours before or after the administration of QB-90. As shown in Fig. 1, QB-90 was able to induce a strong immune response characterized by high antigen-specific antibody titers (IgG1 and IgG2a) and robust delayed type hypersensitivity (DTH) reaction, when administered at the same site 24 hours earlier or at the same time of the BVDV antigen. However, if the adjuvant was administered 48 hours before, or at distal injections sites, no adjuvant effect was observed. It is important to highlight, that no adjuvant effect was observed when QB-90 was administered 24 or 48 hours after the BVDV antigen administration (data not shown).
These findings suggest that the adjuvant QB-90 is able to trigger a transient (lasting 24 hours) and local immunocompetent environment that could be exploited by a subsequent administration of antigen. These data also suggest that QB-90 mechanism of action is independent of binding to BVDV antigen, but in order to trigger a systemic immune response, antigen should be administered up to 24 hours after the inoculation of QB-90, and in a proximal injection site.
Increased cell numbers in spleen and dLNs after QB-90 or IMXQB-90 administration without antigen. At 24 and 48 hours post subcutaneous inoculation (hpi) of either QB-90 or IMXQB-90, single-cell suspensions were prepared from spleen and dLNs. Spleens from QB-90 and IMXQB-90-treated mice ( Fig. 2A) showed a significantly increased number of cells when compared to mock controls at 24 and 48 hours (p < 0.001). However, when comparing 24 and 48 hpi among each group (QB-90 and IMXQB-90), statistically significant differences were observed only in the QB-90-inoculated group.
In dLNs from both QB-90 and IMXQB-90 treated mice (Fig. 2B), a significant increase in cell numbers was detected at 24 hpi (p < 0.001). However, at 48 hpi only the IMXQB-90-treated mice showed a significantly increased number of cells when compared to the control group (p < 0.01), although the number was significantly lower than that observed at 24 hpi.

QB-90 and IMXQB-90recruits and activates immune cells in spleen and dLNs.
In spleen, neutrophils (Gr1 high ), Natural Killers (NK;CD49 + CD3 − ), B cells (CD19 + ) and T cells (identified as CD3 + CD4 + and CD3 + CD8 + ) populations were higher in QB-90 and IMXQB-90-inoculated mice than in those of the mock (saline) inoculated group at 24 and 48 hpi (Fig. 3A-F left axis). Moreover, the dendritic cells (DCs) population, Figure 1. QB-90 adjuvant activity is independent of antigen-adjuvant mixture, but these should be administered in the same site. Female CF1 mice (n = 5) were subcutaneously immunized in the hind limb twice (0 and 14 day) at two weeks intervals with BVDV antigen and QB-90. Groups were inoculated with QB-90 and with BVDV antigen separately, either in the same or opposite sites, and at the same time or with different time points (BVDV antigen was inoculated simultaneously or 24 or 48 hours after the QB-90 inoculation). In addition, a group of mice was inoculated with the formulated vaccine (containing BVDV antigen and QB-90 as adjuvant). IgG1 (A) and IgG2a (B) anti-BVDV antibodies titers and DTH responses (C) at day 28 are shown in box-and-whiskers. Horizontal continuous lines represent median values in mice inoculated with BVDV alone, and dashed lines represent maximum and minimum values in the same group. Statistical analyses were performed using Kruskal-Wallis test (*** denotes p < 0.001, in relation to animals immunized with BVDV alone). In dLNs of mice that received either QB-90 or IMXQB-90, neutrophils, NKs, DCs, B cells and CD4 + and CD8 + T cells were significantly increased at 24 and 48 hpi (Fig. 3A-F right axis), with the exception of NK cells from the IMXQB-90 group at 48 hpi. Altogether, results showed that QB-90 and IMXQB-90-inoculations induce significant increase in cell numbers in both spleen and dLN.
Taken together, these results show that the early increased cell counts evidenced earlier is associated with an increase in neutrophils, NKs, DCs, B and T cell populations.

Differential modulation of gene expression at dLN by QB-90 and IMXQB-90 vaccine adjuvants
formulations. The expression of ninety two genes related to immune function in dLN was analyzed by qPCR (TaqMan ® Mouse Immune Array Applied Biosystems) 24 hpi with QB-90, IMXQB-90 or saline solution as mock control. In the QB-90-treated group 45 genes were differentially expressed whereas in IMXQB-90-treated group these amounted to 23, in comparison with the mock group, as shown in the Venn's diagram (Fig. 4A) and in the gene-expression heatmap (Fig. 4B).
The most prominent gene differential expression was detected in QB-90-treated mice, and involved the up-regulation of Ccr7, Cd28, Il-12b, Il-1β, Nos2, Prf1 and Lif and the down-regulation of Il-7and Fn1. In addition, and up-regulation of several genes coding for the costimulatory molecules, such as Cd40, Cd40lg, Cd80 and Cd86, was noticed.
In IMXQB-90 inoculated mice, only four genes showed differential expression in relation to QB-90 inoculated mice gene expression: Il-2 and Il-4 were down-regulated while Il-9 and Lta were up-regulated. Nineteen genes were found differentially expressed in both groups (Table 1). Regarding the chemokine gene expression, three genes were found to be up-regulated in both treated groups: Ccl2, Ccl3, and Cxcl10. In addition, Csf2, Ctla4, Ifn-γ, Il-10, Il-15, Il-5, Il-6 and Fasl expression was found to be increased (Table 1). Six other genes were up-regulated: Fas, Edn1, Ptgs2, Socs1, Stat1 and Gzmb. On the contrary, Lpr2 was down-regulated in both treated groups. In order to better underst and the molecular and cellular interactions between the differentially expressed genes in both groups we analyzed them as a query gene set using the web-based network tool STRING. Results indicate that nearly all genes differentially expressed in the context of QB-90 and IMXQB-90 inoculation are part of well-defined interaction networks (Fig. 4C). The analysis clearly shows that our proteins have more interactions among themselves than what would be expected for a random set of proteins (Protein-protein interactions enrichment p-value: <1.0e-16). The STRING analysis revealed the functional links between different proteins involving functions as "response to LPS" (GO:0032496), "immune response" (GO:0006955), "response to bacterium" (GO:0009617), "cellular response to LPS" (GO:0071222) and "positive regulation of cell proliferation" (GO:0008284). All cellular processes were identified with high confidence, as predictions showed a low false discovery rate (<1.47e-14).
Considering these results as a whole, QB-90 and IMXQB-90 were able to induce a differential expression pattern of cytokines, chemokines and costimulatory molecules which are particularly relevant in triggering early immune responses in dLNs, and thus, for creating a proinflamatory environment which is essential to orchestrate an effective and long-lasting immune response.
Interestingly, in the absence of LPS, QB-90 was not able to induce the IL-1β production at none of the assessed concentrations ( Fig. 6B) while IMXQB-90 induced a high production of the cytokine (both at 10 and 2.5 µg/mL; Fig. 6D). Cell viability after incubation remained unchanged (≥90%) in all cases (data not shown). Then, we sought to determine if the IL-1β production induced by both QB-90 and IMXQB-90 was a consequence of the canonical inflammasome pathways activation. To that end, BMDCs derived caspase-1/11 KO and WT C57BL/6 mice were stimulated with 10 µg/mL of QB-90 or IMXQB-90 after 2 hours priming with LPS (10 ng/mL), and IL-1β production was measured. In the presence of LPS both QB-90 and IMXQB-90 (10 µg/mL) induced the production of IL-1β in WT BMDCs, nonetheless, this production was not seen in caspase-1/11 KO BMDCs (Fig. 6E). No differences were observed among groups when cells were not primed with LPS (Fig. 6F).
These results indicate that the observed production of IL-1β is dependent on Caspase-1/11, and that both preparations would be able to activate canonical inflammasome pathways 37 .

QB-90 and IMXQB-90induce proinflammatory responses through MyD88 pathways in a TLR-independent manner.
In order to investigate the mechanisms by whichQB-90 and IMXQB-90 trigger proinflammatory responses, reporter cell lines Raw Blue ™ expressing many pattern recognition receptors (PRR), THP1-XBlue ™ , expressing functional myeloid differentiation primary response 88 (MyD88) adaptor molecule, , it becomes evident that these formulations were able to activate the reporter gene in THP-1 cells, but not in cells lacking the MyD88 adaptor molecule. Altogether, these results suggest that the both Q brasiliensis saponin based formulations interact with monocytes or macrophages in a TLR-independent but MyD88-dependent manner.

Discussion
Adjuvants are essential components of modern vaccines since they contribute to the initiation of the innate immune response to antigens 3,7 . Furthermore, they stimulate the production of cytokines by resident innate cells, which in turn leads to the rapid recruitment of neutrophils, monocytes and dendritic cells (DCs) that uptake and present antigens 35,38,39 . Adjuvant potential of plant-derived compounds has been largely explored. In particular, saponins (e.g., QS-21) or saponin-based formulations as ISCOMs or ISCOMATRIX TM have shown to be safe in animal and human clinical studies and to be well tolerated by humans 13,27,35,40 . When administered subcutaneously, such adjuvants have shown to trigger powerful immune responses, characterized by the induction of CD4 + and CD8 + T cells, balanced Th1/Th2 cytokine responses, as well as robust and persistent antibody responses 30,31,35 .
In addition, they have proven to be highly effective when delivered via mucosal surfaces 22,31,34 . However its mechanism of action is still poorly characterized 2,31,35 .
Herein we show that QB-90, a Quillaja brasiliensis leaf-extracted saponin fraction, is efficient in inducing adjuvant effect in mice even when administered up to 24 hours before the inoculation of the antigen, provided the antigen is administered at the same inoculation site. However, when the antigen is inoculated before the adjuvant, such effect is not observed. This interesting finding reveals that QB-90 is capable to promote a local and transient immunocompetent environment. This behavior is also observed in other adjuvants successfully used in human vaccines such as alum 41,42 and MF59 43 . In conclusion, the findings reported here support that QB-90 is a potent adjuvant with the ability to stimulate a systemic immune response and promote a local and transient immunocompetent environment.
In order to provide a deeper understanding of the mechanisms underlying the generation of the immunocompetent environment, the immune cell recruitment in spleen and dLNs was examined in mice 24 and 48 hpi of QB-90 or IMXQB-90 without antigen. Our results show that either QB-90 or IMXQB-90 inoculation significantly increase spleen cellularity, suggesting a possible rapid drainage from the injection site. Spleen cell influx Ccl3 chemokine (C-C motif) ligand 3 8,7 2,6 Cxcl10 chemokine (C-X-C motif) ligand 10 15,2 9,2 Cxcl11 chemokine (C-X-C motif) ligand 11 continues at least up to 48 hours after inoculation of QB-90. In the dLN, both formulations induce an early influx of cells 24 hpi that rapidly decay 24 hours later, being still increased at 48 hpi only in IMXQB-90 inoculated mice. These findings can be explained by the nature of each adjuvant-formulation. As ISCOMs-like structures act as an antigen-delivery system and also are particulate immunostimulatory compounds, they are able to induce a more durable cellular recruitment at the dLNs, than the soluble saponin adjuvant, which is essential to trigger an effective and long-lasting immune response 31 . The above results confirm and extend our previous reports, in which mice treated with ISCOMs-adjuvanted vaccines promoted higher antibodies titers than those inoculated with saponin-adjuvanted vaccines, both in Q. brasiliensis and commercially available saponins 34 . When the recruited cells populations were characterized by flow cytometry, an increase in neutrophils, DCs, NKs, B-and T-cells was found in QB-90 and IMXQB-90 treated mice in comparison with the mock control. Interestingly, dLN DCs (CD11b + MHCII + ) are transiently increased in both inoculated groups at 24 hpi, and their number rapidly decreases for the QB-90 inoculated group, but is still increased in the IMXQB-90 treated group at 48 hpi. On the other hand, splenic DCs behave differently: their number do not increase at 24 hpi, but is significantly augmented in both groups at 48 hpi. The observed behaviour in DCs populations might be explained by the fact that when activated by either QB-90 or IMXQB-90 adjuvants, dLNs DCs may rapidly migrate from dLN to spleen, in order to provide a further activation of immune effector cells. ISCOMs and ISCOMATRIX TM are known to enhance antigen uptake, and induce dLNs DCs activation, resulting in later strong antibody and T cell responses 2,34,37,44 . Our results are in line with previous reports which have proven that ISCOMATRIX TM particles rapidly move to the dLN within the first 2 h after injection in mice, where it can be loaded into dLN-resident DCs and other antigen presenting cells (APCs) 31 . Another work emphasized that DCs and APCs at the injection site also transfer captured ISCOMATRIX TM particles from the injection site into the dLN, thus, providing two waves of vaccine antigen presentation (direct trafficking of adjuvant and via peripheral APC-transport), which result in prolonged antigen presentation 31,35 .
Regarding splenic and dLN neutrophils, QB-90 is able to induce a transient increase 24 hpi, which rapidly decrease and was no longer detectable at 48 hpi. However, IMXQB-90 formulation was able to keep the increase Furthermore, our results show that QB-90 and IMXQB-90 are also able to recruit B and T cells (both CD4 + and CD8 + ). Taken together, our results are in good agreement with previous reports that showed that nanoparticulate formulations based in Q. saponaria saponins increase the number of DCs, NK-, B-, CD4 + and CD8 + T-cells, and granulocytes in the dLN within 6 h after inoculation, exhibiting a peak 24-48 hpi 4,30,35,44 . Additionally, it has been shown that the cellular influx is transient and reversible, returning to normal levels by 72 hours 35,37 . Regarding NK cells, we show herein that QB-90 and IMXQB-90 promote an increase in this cell population recruitment, in both spleen and dLNs. Interestingly, NK cells are a major source of IFN-γ during early immune responses and are known to enhance antigen-specific T cell responses 48 . In this sense, we detected similar levels of IFN-γ in sera from mice previously sensitized for 24 h with QB-90 or IMXQB-90. However, TNF-α was detected only in mice treated with QB-90, suggesting that IMXQB-90 induces a more controlled inflammatory environment than QB-90.
As mentioned before, our results show a significant increase in DCs recruitment in spleen as well as dLN. In this sense, it has been reported that QuilA ® -as well as particulate adjuvants-dramatically enhance IL-1β secretion by DCs 3,5 . Our results show that QB-90-stimulated BMDCs secrete IL-1β in the presence of a TLR agonist as a positive control for Raw Blue TM and both THP-1 cell lines, respectively. In B and C, data were analyzed by One-way ANOVA and Dunnet's post test, comparing groups against the negative control (C-). In (D) data were analyzed by One-way ANOVA and t-test. Differences were considered statistically significant when P value < 0.05 (**P < 0.01). such as LPS. Interestingly, BMDCs stimulated solely with IMXQB-90 -without the need of LPS -promote the secretion of IL-1β, and when primed with LPS, IL-1β production is higher than that observed with alum stimulation (a well-known inflammasome activator). In addition, IL-1β secretion observed in IMXQB-90-stimulated DCs is carried out in a caspase-1/11 dependent manner. These results suggest that the canonical-inflammasome pathway is being activated by this adjuvant formulation. When the interaction with innate immunity cellular receptors was assessed in vitro, our findings revealed that Q. brasiliensis saponins, as well as its nanoparticles, interact with macrophages or monocytes in a TLR-independent but in a MyD88-dependent manner. In summary, we suggest that both formulations from Q. brasiliensis saponins may activate canonical inflammasome pathways, as previously shown for other particulate adjuvants 5,49 and specifically for ISCOMATRIX TM 36 . Additionally, the adjuvant effects of QB-90 and IMXQB-90 might depend on the interaction of Caspase-1/11 and MyD88 pathways, consistent with previous reports on saponins from Q. saponaria 3,4,36 , which have shown to activate innate immunity in a TLR-independent manner and have demonstrated in vivo that Q. saponaria nanoparticles did interact with the adaptor molecule MyD88 44 . Considering these previous reports and our results obtained herein, it can be suggested that MyD88 adaptor molecule is involved in the mechanism of action of Q. brasiliensis leaf-derived saponins. In fact, it has been reported that ISCOMATRIX TM (formulated with commercially available saponins) induces lysosomal destabilization leading to IL-1β release in vitro by APCs through NLRP3 inflammasome 36 . Thus, it has been proposed that ISCOMATRIX TM nanoparticles migrate faster into the cytosol of multiple DCs subsets, promoting the induction of cytokines and chemokines, and bridge innate and adaptive immune responses in vivo in a Toll-like receptor-independent but MyD88-dependent manner 44 . To the best of our knowledge, until now, only commercial saponins derived from Q. saponaria have been described to activate inflammasome pathways 3,4,36,50 . It has been recently reported that soluble QS-21 delivery in vitro and in vivo induces inflammasome-mediated IL-1β/IL-18 secretion in mice, but the exact mechanism is not clear yet 2,3 . In addition, a novel mechanism of action has recently been proposed for saponins, based on inflammasome activation that enhances antigen cross-presentation by forming intracellular lipid bodies. In this mechanism, lipid bodies escape from endosomes and then translocate antigens for cross-presentation 2 .
Finally, expression profiles of ninety-two immune related genes were assessed by qPCR24 hours after QB-90 and IMXQB-90 sensitization in dLNs. Our results show that some genes encoding for cytokines/chemokines and/or their receptors are differentially regulated by QB-90 or IMXQB-90. QB-90 treatment induces changes in the expression of 41 genes in comparison to control group, while IMXQB-90 alters the expression of 23 genes, being 19 genes in common. These results may suggest that inflammation induced by QB-90 is stronger than that induced by IMXQB-90. For example, QB-90 -unlike IMXQB-90 -induces the up-regulation of genes involved in cellular activation (Cd28, Cd3e, Cd40, and Cd40lg), co-stimulation (Cd80, Cd86, Cd8a) and transcription factors (Stat3, Stat4, Tbx21). On the other hand, IMXQB-90 do not activate as many pro-inflammatory and activation genes as QB-90, suggesting it may promote less intense and perhaps more controlled inflammatory responses than saponins alone. These findings are in line with results mentioned earlier regarding the cytokine profile elicited by both formulations, in which we found that QB-90 promoted a more proinflammatory environment. In addition, these results correlate with the characteristics of low toxicity and hemolytic activity of these ISCOMs-like formulations 3,27,31,34 . Finally, both formulations yielded a change ≥10-fold in expression of a set of immune-related genes (Infg, Gzmb, Ccl2, Cxcl10, Cxcl11, Il5 and Il6). Previous reports on early activation of innate cells by ISCOMs or ISCOMATRIX TM showed induction of an array of cytokines and chemokines in serum of vaccinated animals, including pro-inflammatory cytokines (IL-1β, TNF-α and IL-6), as well as chemokines and cytokines involved in macrophage and NK cell activation (CCL3, CCL2, MCP-1, IFN-γ), and/or neutrophil activation and migration (CXCL1, KC, CXCL10, IP-10, G-CSF) 4,37,44 . Interestingly, studies performed with MF59 on human innate immune cells, such as DCs, monocytes, macrophages and granulocytes, showed induction of very similar immune-related genes 51,52 .
Network analysis using the web-based network tool STRING indicates that nearly all differentially expressed genes in the context of QB-90 and IMXQB-90 inoculation are part of well-defined interaction networks. The STRING analysis reveals a strong link with pathways related to immune responses, response to microorganisms and positive regulation of cell proliferation. It is important to highlight that our study was focused only on specific genes, and may have missed possible targets for both formulations. In order to broaden the scope of this assessment, full transcriptome analysis experiments may be conducted to further characterize the transcriptional activity, and enable to create a global picture of the immunocompetent environment created by these promising adjuvants.
In summary, our results show that Q. brasiliensis saponin-based adjuvants induce an antigen-independent and transient immunocompetent environment. Early cellular responses are observed for both QB-90 and IMXQB-90, at local (dLN) and systemic (spleen) levels. Interestingly, IMXQB-90 elicits a less intense systemic pro-inflammatory response, whilst still being able to activate by itself -at least in vitro -the inflammasome through Caspase-1/11 and MyD88 pathways. Moreover, in terms of number of immune related genes, IMXQB-90 also seems to induce a less intense inflammation in the dLN. Nonetheless, when the adaptive immune response was assessed, our previous reports show that both formulations induced similar Th1/Th2 immune profiles, but the nanoparticulate formulation displayed a stronger immune response than the soluble saponin fraction 34 .
Overall, our results reinforce the postulation of using leaf-extracted Q. brasiliensis saponins and in particular their nanoparticles as potent vaccine adjuvants. We have demonstrated here that they are capable of promoting early immune responses that, considering our previous data, correlate with adequate adaptive immune responses 18,[20][21][22][23]34 . In addition, Q. brasiliensis leaves constitute a sustainable and readily renewable alternative source of saponins in relation to Q. saponaria barks. Last but not least, the results obtained here are particularly relevant, as they contribute to confirm that the mechanism of action of these new adjuvants is similar to that described for commercially available saponins. Thus, these results imply that these novel vaccine formulations may be used to attend uncovered medical needs for animal and human health.

Materials and Methods
Ethics statement. Animal manipulation was performed in accordance with "Comisión Honoraria de Experimentación Animal" (CHEA) guidelines and was approved by the Uruguayan University Research Ethics Committee (approval number 070153-000531-13) and Ethics Commission on Animal Experimentation (CEUA) in the "Instituto de PesquisasVeterinárias Desidério Finamor". Animals were appropriately housed with controlled temperature ( 20,24 . The infectious titre of the virus suspension prior to inactivation was 10 7.5 tissue culture infectious doses per mL. The inactivated virus suspension (referred to as BVDV onwards) was used as antigen for adjuvant testing for all assays.
Conventional female Rockefeller mice (n = 5) of the CF-1 breed (5-6 weeks old) were purchased from Fundação Estadual de Produção e Pesquisa em Saúde (FEPPS, Porto Alegre, RS, Brazil). Mice were subcutaneously inoculated (100 µL) in the hind limb twice with a two week interval with either BVDV antigen alone or with adjuvant QB-90 (50 µg). Injections were applied either in the same or opposite member sites, performed simultaneously, or saponin adjuvant was injected 24 or 48 hours prior or post antigen injection. Additionally, at zero hour, the animals were inoculated with formulated (adjuvanted vaccine) or unformulated (two shots; one for antigen and after 5 minutes, one for adjuvant). Mice were bled prior to inoculations (on day 0) and 2 weeks after the second immunization (day 28); sera were kept frozen (at −20 °C) until processed.

Immunoassays for antibodies and delayed type hypersensitivity (DTH). Total anti-BVDV IgG1
and IgG2a were determined for each serum samples by ELISA, carried out essentially as previously described 20,24 using as antigen the BVDV suspension used for mouse immunization. Antibody titres were expressed in OD 492 nm. Sera were appropriately diluted in PBS-T (1:300 for IgG1 and 1:100 for to IgG2a).
The DTH responses were evaluated in immunized mice on day 28 post first immunization. The BVDV-specific response of each animal was calculated based on values of injected footpad minus the average of the basal swelling 21,24,34 . Cell isolation from spleen, draining lymph nodes (dLNs) and flow cytometry analysis. BALB/c mice (8-12 weeks old) were purchased from DILAVE (MGAP-Uruguay) and kept at the Instituto de Higiene (UdelaR). Mice were injected subcutaneously (s.c.) (n = 15) at the base of the tail with 25 µL of QB-90 (10 µg) or IMXQB-90 (2.5 µg). The concentration of IMXQB-90 used in this study is defined as the saponin concentration within the particles. Control mice received 25 µL of saline. Spleen and inguinal draining lymph nodes (dLNs) were collected 24 and 48 h post inoculation (p.i.) (n = 5). Five 5 animals were used for gene expression studies. From these, dLNs were collected 24 h post adjuvant administration and kept in RNAlater (Ambiom) solution at −80 °C.
Spleen and dLNs were collected in cold PBS and processed by mechanical disaggregation and then filtered through a 100 µm cell strainer (BD Falcon). Cells from the two symmetric dLNsof each mouse were pooled. Splenocytes were incubated in red blood cell lysis buffer, washed with PBS containing 2% FBS and passed through a 100 µm cell strainer. Cells were suspended in FACS buffer (PBS, pH 7.4, 0.5% BSA, 2 mM EDTA and 0.1% sodium azide) and counted with aid of a cell counter (Countess ® Automated Cell Counter, Life Technologies TM ). the presence of 20 ng/mL recombinant mouse GM-CSF (PeproTech) as described elsewhere 54 . The cells obtained were between 85 and 95% CD11c + . On day 10, cells were plated at 0.4 million per well in 96-well plates and stimuli were added in medium containing 5 ng/mL GM-CSF. BMDC were primed with 10 ng/mL lipopolysaccharide (LPS from Escherichia coli O127:B8, Sigma) for 2 hours and then stimulated with different doses (10, 2.5 and 0.5 µg/mL saponins concentration) of QB-90 or IMXQB-90. Alum (Alhydrogel, Sigma) was used as a control for inflammasome activation. After 3 hours, cell response was measured in terms of IL-1β production. Cell culture supernatants were assayed for IL-1β with ELISA kits from Biolegends according to the manufacturer's instructions. Cell viability was evaluated at the end of the assay by flow cytometry, using Topro-3 staining (Invitrogen).

Immunostimulation of pattern recognition receptors.
To determine if QB-90 or IMXQB-90 were able to activate innate immunity receptors activation, Raw Blue TM (murine macrophages-derived) THP-1 Blue TM (human monocytes-derived) and THP-1 Blue TM -defMyD (human monocytes-derived lacking MyD88 gene) reporter cell-lines (InvivoGen) were stimulated with QB-90 or IMXQB-90. These commercial cell-lines used constitutively over express pattern recognition receptors (PRRs), and stably express an NF-κB and AP-1-inducible secrete embryonic alkaline phosphatase (SEAP) reporter gene. Upon TLR stimulation this cells activate transcription factors and subsequently the secretion of SEAP, which is detectable by the addition of QUANTI-Blue TM .
Cell cytotoxicity assay. Cytotoxicity was determined using the MTT assay. Briefly, reporter cell lines RawXBlue TM and THP1-XBlue ™ (InvivoGen ® ) were seeded at a concentration of 4.0 × 10 4 per well on 96-well cell-culture plates (Greiner) and maintained at 37 °C under a humid atmosphere with 5% CO 2 . After 18 h, the medium was removed and 100 µL of culture medium containing different concentrations of either QB-90 or IMXQB-90 (6.250, 1.562, 0.390, 0.097, 0.0244, 0.006 µg/mL of saponins) were added to each well. Six replicates of each sample were performed. Plates were incubated overnight at 37 °C and 5% CO 2 . After incubation, 50 µL of 5 mg/mL of MTT (Sigma) were added to each well and the cells were incubated 4 hours. The plates were centrifuged (1500 × g for 5 min) and the supernatants containing untransformed MTT were carefully removed. 50 µL of 20% sodium dodecyl sulfate (SDS, Sigma) were added to each well in order to solubilize the formazan crystals, and the OD 540 nm was measured in an ELISA reader (Tecan Sunrise TM ). Results were expressed as the percentage of viability, considering OD of untreated control cells as 100% of viability.
RNA isolation and immune-related gene expression. Changes in mouse immune-related gene expression in dLNs from mice inoculated with either QB-90 or IMXQB-90 preparations were detected using a TaqMan ® Mouse Immune Array by qPCR with the 7500 Real-Time PCR System (Applied Biosystems). For these studies, RNA was isolated from dLNs at 24 h post adjuvant administration (or mock inoculated with saline solution). Each experimental sample represents two pooled dLN collected from mice (n = 5). Prior to RNA isolation, dLNs were homogenized in 1 mL of TRIzol (Ambiom). RNA was further purified using Pure Link RNA mini Kit (Ambiom). RNA integrity was checked on Bioanalyzer (Agilent) and quantified by fluorometry (Qubit, Invitrogen). cDNA was synthetized by High-Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific). Fold change was calculated using the Pfaffl method (2 −ΔΔCt ) for dLN from QB-90 and IMXQB-90 and being compared with the mock group 55 . The online software STRING (combined score >0.4) was carried on to perform protein interaction network analysis 56 .
Statistical analysis. The significance of differences was assessed using nonparametric Kruskal-Wallis test with Dunn's post test. Students T-test was used for evaluation BMDC experiment. GraphPad Prism version7.00 (GraphPad Software, Inc.) was used for data analysis. Results are expressed as the mean or median value from individuals in each group ± SEM or box-and-whiskers. A p value equal or less than 0.05 was considered statistically significant.