Vaccine adjuvant MF59 promotes the intranodal differentiation of antigen-loaded and activated monocyte-derived dendritic cells

MF59 is an oil-in-water emulsion adjuvant approved for human influenza vaccination in European Union. The mode of action of MF59 is not fully elucidated yet, but results from several years of investigation indicate that MF59 establishes an immunocompetent environment at injection site which promotes recruitment of immune cells, including antigen presenting cells (APCs), that are facilitated to engulf antigen and transport it to draining lymph node (dLN) where the antigen is accumulated. In vitro studies showed that MF59 promotes the differentiation of monocytes to dendritic cells (Mo-DCs). Since after immunization with MF59, monocytes are rapidly recruited both at the injection site and in dLN and appear to have a morphological change toward a DC-like phenotype, we asked whether MF59 could play a role in inducing differentiation of Mo-DC in vivo. To address this question we immunized mice with the auto-fluorescent protein Phycoerythrin (PE) as model antigen, in presence or absence of MF59. We measured the APC phenotype and their antigen uptake within dLNs, the antigen distribution within the dLN compartments and the humoral response to PE. In addition, using Ovalbumin as model antigen, we measured the capacity of dLN APCs to induce antigen-specific CD4 T cell proliferation. Here, we show, for the first time, that MF59 promotes differentiation of Mo-DCs within dLNs from intranodal recruited monocytes and we suggest that this differentiation could take place in the medullary compartment of the LN. In addition we show that the Mo-DC subset represents the major source of antigen-loaded and activated APCs within the dLN when immunizing with MF59. Interestingly, this finding correlates with the enhanced triggering of antigen-specific CD4 T cell response induced by LN APCs. This study therefore demonstrates that MF59 is able to promote an immunocompetent environment also directly within the dLN, offering a novel insight on the mechanism of action of vaccine adjuvants based on emulsions.

MF59 is an oil-in-water emulsion adjuvant approved for human influenza vaccination in European Union. The mode of action of MF59 is not fully elucidated yet, but results from several years of investigation indicate that MF59 establishes an immunocompetent environment at injection site which promotes recruitment of immune cells, including antigen presenting cells (APCs), that are facilitated to engulf antigen and transport it to draining lymph node (dLN) where the antigen is accumulated. In vitro studies showed that MF59 promotes the differentiation of monocytes to dendritic cells (Mo-DCs). Since after immunization with MF59, monocytes are rapidly recruited both at the injection site and in dLN and appear to have a morphological change toward a DC-like phenotype, we asked whether MF59 could play a role in inducing differentiation of Mo-DC in vivo. To address this question we immunized mice with the auto-fluorescent protein Phycoerythrin (PE) as model antigen, in presence or absence of MF59. We measured the APC phenotype and their antigen uptake within dLNs, the antigen distribution within the dLN compartments and the humoral response to PE. In addition, using Ovalbumin as model antigen, we measured the capacity of dLN APCs to induce antigen-specific CD4 T cell proliferation. Here, we show, for the first time, that MF59 promotes differentiation of Mo-DCs within dLNs from intranodal recruited monocytes and we suggest that this differentiation could take place in the medullary compartment of the LN. In addition we show that the Mo-DC subset represents the major source of antigen-loaded and activated APCs within the dLN when immunizing with MF59. Interestingly, this finding correlates with the enhanced triggering of antigen-specific CD4 T cell response induced by LN APCs. This study therefore demonstrates that MF59 is able to promote an immunocompetent environment also directly within the dLN, offering a novel insight on the mechanism of action of vaccine adjuvants based on emulsions. PLOS

MF59 promotes intranodal differentiation and transient accumulation of Mo-DCs
To detect Mo-DCs in vivo, we took advantage from the evidence that CD64 has been described as an unambiguous marker to identify these cells in the mouse immune system [20,22]. We immunized mice in both legs with fluorescent protein Phycoerythrin (PE) as model antigen, in presence or absence of MF59 and using PBS as negative control. The fluorescence of the antigen made it possible to track the antigen-loaded cells [15]. Treated mice were sacrificed 15 minutes (15 min), 8 hours (8 hrs), 18 hours (18 hrs) and 3 days (3 d) after primary immunization and the dLNs were collected. LNs from the right legs were pooled and analyzed by flow cytometry, whereas LNs from the left legs were analyzed individually by confocal microscopy. Imaging studies of dLNs were performed because we wondered if MF59 may have an impact on the antigen distribution within the dLNs, on the organization of LN compartments and what was the localization of this adjuvant among the different zones of the LN [5][6][7].
The APCs were phenotypically characterized according to the reported gating strategy (S1 Fig), which is based on the current knowledge [8,18,19,23,24]. We observed a transient accumulation of a CD8α_F4/80 double positive APC subset between 8 and 18 hrs after immunization with MF59, revealed by the significant increase in the cell number, compared to mice immunized with PE alone, at the two time points (8 and 18 hrs after treatment) taken in consideration ( Fig 1A). This cell accumulation is also observed injecting MF59 alone without any co-administered antigen. The accumulated CD8α_F4/80 double positive APC population (red cells) specifically expresses the Mo-DC marker CD64, compared to macrophages (MFs; blue cells) and LN DCs (green cells) that are CD64 negative (Fig 1B), but, very interestingly, it displays the DC phenotype, via the CD11c marker expression, only 18 hrs after immunization with MF59 (Fig 1B and 1C, S2 and S3A Figs), whereas, at 8 hrs, both in presence or absence of MF59, this APC subset has a monocyte phenotype [8,18,19,23,24], being CD11c low or negative ( Fig 1C, S2A  Summarizing, when immunizing with MF59, our phenotypic analysis discriminated the following four APC subsets: DCs (green) as CD11c high _F4/80 -/low _CD8α -to+ _CD64 -_Ly6-C + ; MFs (blue) as CD11c low _F4/80 + _CD8α -_CD64 -_Ly6-C -; Monocytes (red) as CD11c -/low _ F4/80 + _CD8α + _CD64 + _Ly6-C high (8 hrs after immunization); Mo-DCs (red) as CD11c high _ F4/80 + _CD8α + _CD64 + _Ly6-C high (18 hrs after immunization). We thus concluded that MF59 promotes accumulation within the dLN of monocytes and Mo-DCs respectively 8 and 18 hrs after primary immunization, suggesting that MF59 induces intranodal differentiation of Mo-DCs. To support this hypothesis we performed an ex-vivo experiment in which we isolated the dLN 8 hrs after immunization and followed Mo-DC differentiation after an in vitro culture of the whole LN. Mice were immunized in both legs with PE or PE adjuvanted with MF59. Animals were sacrificed after 8 hrs and the popliteal dLNs from both legs were collected. The LNs from the left legs were immediately processed and analyzed by flow cytometry as described in the previous experiment, whereas LNs from the right legs were cultured intact and untouched for an additional 10 hrs and then analyzed by flow cytometry (Fig 1D). As expected and already described (Fig 1B), we found accumulation of monocytes in the dLNs analyzed 8 hrs after immunization with MF59 (left leg LNs) (Fig 1D, left dot plots). Interestingly, in the dLNs of the same animals explanted at the same time point (8 hrs) after immunization and cultured for 10 hrs (right leg LNs), we found accumulation of Mo-DCs similarly to what observed in the dLNs explanted and analyzed 18 hrs after immunization ( Fig 1D,   Representative results of one experiment out of the three reported in panel (a) are shown. (d) Flow cytometry dot plots of the dLN cell suspensions to identify APCs derived from mice immunized with MF59, whose LNs were explanted from both legs 8 hrs after the immunization, but: the LNs from the left legs (left dot plots) were immediately processed (8 hrs), whereas the LNs from the right legs (right dot plots) were processed after 10 hrs of in vitro culture (8 hrs + 10 hrs (in culture)). Representative results of one experiment out of three are shown.  Fig and Fig 1). An increase in number of antigen-loaded APCs in MF59-treated animals 8 hrs after immunization (in presence of monocyte accumulation) is observed, but it is not statistically significant due to the high variability (S6B Fig). Thirdly, when immunizing both in presence or absence of MF59, the percentage and the number of antigen-loaded APCs are generally higher at early time points, peak at 8 hrs after immunization (S6B Fig), and then decrease afterwards, but this decrease is slower in presence of MF59. We therefore concluded that MF59 induces a persistence of antigen positive APCs during the first 18 hrs, which correlates with a concomitant accumulation of Mo-DCs (S6 Fig and Fig 1). Consequently, we investigated what was the contribution of Mo-DCs in the enhancement of antigen-loaded APCs at 18 hrs following immunization. We found that, both monocytes and Mo-DCs, as entire cell populations, display a superior ability to engulf the antigen among APCs, because the percentage of antigen-positive cells in these populations is significantly higher compared to DCs and MFs ( In addition, formulation with MF59 mainly enhances the number of antigen-loaded Mo-DCs compared to monocytes, because the increment in the number of antigen-loaded cells is higher in Mo-DCs versus monocytes than in DCs, when immunizing with MF59, whereas MFs are not affected (Fig 2B). We thus concluded that, 18 hrs after immunization with MF59, Mo-DCs represent the most abundant antigen-loaded cell population within the dLN (Fig 2A and 2B and S6 Fig). In addition, using fluorescently labeled MF59 (which does not display an affected adjuvant capacity) together with PE antigen, we measured the percentage of antigen and MF59 positive cells and we found that, although the adjuvant emulsion is engulfed together with the antigen by all three APC subsets, the Mo-DC population is mostly double loaded with both antigen and adjuvant (  contribute to the induction of an adaptive immune response. Since activated APCs with upregulated co-stimulatory molecules are required for an optimal T cell response [17], we finally checked cell activation by measuring the expression of the co-stimulatory molecule CD80 on APCs within the dLN (Fig 2C). We found that Mo-DCs display the highest CD80 expression among the three APC subsets and a significantly increased CD80 up-regulation (based on CD80 expression of monocytes) after immunization with MF59 as compared to DCs and MFs (that do not up-regulate CD80) ( Fig 2C). Consistently with this observation, the Mo-DC population is the only APC subset which up-regulate also the co-stimulatory molecule CD86 (S8 Fig). All together these results demonstrate that, during the first 18 hrs after an immunization, the presence of MF59 adjuvant promotes accumulation and differentiation within the dLNs of Mo-DCs that, among APCs, display a more activated phenotype, a higher ability to engulf the antigen as a unique cell population and consequently increase the persistence of antigen-bearing APCs within the dLNs. Therefore, when immunizing with MF59, the Mo-DC population is the major source of antigen-loaded and activated APCs within the dLN.

Antigen-loaded Mo-DCs could be localized in the medullary region of the dLN
Consistently with flow cytometry data, by confocal microscopy we detected antigen and MF59 in the dLN already 15 min after the injection, distributed along the subcapsular sinus and within the medullary area. We also used two-photon microscopy to provide, for the first time, a real time kinetic analysis of the arrival of an MF59-adjuvanted antigen in the dLN (S1 and S2 Videos). We found that both antigen and MF59 are rapidly (in less than a minute) translocated to the dLN (S1 and S2 Videos) and in this early phase they appear to transfer independently, confirming a previous observation [25]. By confocal microscopy, eight hours after immunization, both antigen and MF59 are prevalently detected within the medullary area. However, focusing our attention on the 18-hour time point after immunization, we observed that the antigen is still localized within the medullary compartment of the LN either in absence or in presence of MF59 ( Fig 3A). Nevertheless, through CD11c labeling, after immunization with MF59 we observed a remarkable accumulation of DCs within the medullary area of the LN ( Fig 3A). Furthermore, using labeled MF59, we observed that also the adjuvant is localized within the medullary compartment together with the antigen and DCs (Fig 3B). The co-localization of antigen and adjuvant within the medullary region was confirmed also by 3D reconstruction of the dLN analyzed by two-photon microscopy (S3 and S4 Videos). Thus, we demonstrated that, 18 hrs after immunization with MF59, both antigen and adjuvant accumulate within the medullary compartment of the dLN together with DCs.
Collectively, flow cytometry and confocal microscopy results of this study suggest that, when immunizing with MF59, the observed differentiation and accumulation of antigen/adjuvant loaded and activated Mo-DCs could take place within the medullary compartment of the dLN. APCs derived from dLNs of mice immunized with MF59 have an enhanced capacity to trigger antigen-specific CD4 T cell response In order to correlate our finding with the adjuvant function of MF59, we analyzed the ability of APCs from dLNs of immunized mice to trigger antigen-specific CD4 T cell response [17]. We immunized mice with Ovalbumin (OVA) in presence or absence of MF59 and then cultured dLN APCs with CFSE-loaded OVA-specific CD4 T cells, measuring T cell proliferation by flow cytometry (S9 Fig). We discovered that the percentage of CFSE halving of OVA-specific CD4 T cells is significantly enhanced when the cells are cultured with dLN APCs obtained from mice immunized with MF59 adjuvant, revealing a potentiated triggering of antigen-specific CD4 T cell response. This result is consistent with the enhancement effect of MF59 on antibody response to OVA which has been already described [26]. In addition, we also confirmed that MF59 display adjuvant activity with our model antigen because it induces a significant increment of the humoral response to PE after immunization (S10 Fig) ( We can conclude that accumulation and differentiation of antigen-loaded and activated Mo-DCs within the dLN induced by MF59 correlate with the ability of dLN APCs to induce an enhanced antigen-specific T cell activation and, ultimately, with the adjuvant activity of MF59.

Discussion
This study elucidates a new aspect of the mode of action of MF59 emulsion adjuvant, which until now was believed to promote the establishment of a transient "immunocompetent" environment at the injection site, resulting in the recruitment of APCs, which then take up antigen and adjuvant and transport them to dLNs, where they trigger the immune response [11-14, 16, 25, 27]. Although this proposed mechanism of action for MF59 is still valid, our study demonstrates that at least an additional mechanism of MF59 adjuvant activity exists, by which MF59 promotes an "immunocompetent" environment that induces differentiation of antigen-loaded and activated Mo-DCs, directly in the dLN.
A previous study demonstrated that monocytes are the first non-dispensable APCs for immune response to reach both injection site and dLN [13]. In fact, also neutrophils, that are more rapidly recruited into the injection site and dLN, are vehicles of antigen, but these cells have been demonstrated to be dispensable for the adjuvant activity of MF59 [13]. The number of antigen/adjuvant double positive monocytes peaks at 17 hours from the treatment, at the injection site, whereas it peaks at 7 hours within the dLN, consistently with our findings [13]. In addition, only after immunization with MF59 these monocytes display a critical change in morphology toward a DC-like phenotype [14]. Based on these published observations, our current results, and the well described mechanism of trafficking of cells and substances within the dLN [5][6][7], we suggest the following non-exclusive hypothesis for the mode of action of MF59: after immunization with MF59, the adjuvant rapidly reaches the dLN where it stimulates local cells to produce soluble factors that induce monocyte recruitment and drive the differentiation of Mo-DCs; thus monocytes migrate from the circulation directly into the medulla of the LN where they engulf antigen and adjuvant and differentiate into DCs. Ultimately, we propose that the vaccine adjuvant MF59 promotes a stimulatory environment within the medullary compartment of the dLN which drives the differentiation of activated and antigen/adjuvant-loaded Mo-DCs.
In addition to this finding, we observed that dLN APCs from mice immunized with MF59 have an enhanced capacity to trigger antigen-specific CD4 T cell response. In this respect, it is noteworthy the fact that it has been recently discovered that Mo-DCs drive the follicular helper T cell differentiation, which is a critical event for developing a long lasting immunity [21]. Considering these last two findings, it is tempting to speculate that Mo-DC may play an essential role in triggering an adaptive immune response associated with an effective immunization and that intranodal differentiation of antigen-loaded and activated Mo-DCs may be one of the major mechanism driving MF59 adjuvant function, but further investigations are required to support this hypothesis. For example, CCR2 plays a key role in the monocyte egress from bone marrow into the blood circulation [22,23]. In agreement with this, it has been found that, after immunization with MF59, the recruitment of mononuclear cells into the inflamed tissues is dramatically reduced in CCR2 KO mice [11]. Thus it would be important to evaluate if in these mice a significant reduction of Mo-DCs within the dLNs can be observed when immunizing with MF59 and if this observation is associated with a reduction in MF59 adjuvant capacity. These results would ultimately demonstrate that the intranodal differentiation of Mo-DCs is essential for MF59 to carry out its adjuvant function.
We think that if in vivo differentiated Mo-DCs represent the APC subpopulation which is necessary and sufficient to drive MF59 adjuvanticity, the elucidation of the molecular mechanisms triggering this differentiation could help to improve the efficacy of emulsion adjuvants and to identify new more effective adjuvants using Mo-DCs as target cells. For example, it has been described that MF59 increases the ATP release at injection site which modulates its adjuvant function [26]. It would be intriguing to investigate if MF59 induces the ATP release also within the dLN and if this release is involved at some stage of the Mo-DC differentiation and/ or activation. Since the adjuvant capacity of MF59 is partially dependent on the adaptor molecule MyD88, but MF59 is not a TLR agonist [28], it would be interesting to examine if MyD88 is critical for a signal transduction pathway downstream to a receptor that works as a sensor for ATP or other intermediates of the ATP metabolism.
Whether antigen-loaded Mo-DCs play a role in the mode of action of other adjuvants, such as ALUM, has not been described. However, also ALUM adjuvant seems to promote the transient appearance of Mo-DCs within the dLN, but this occurs 3 days after the immunization, delayed compared to MF59. At this time, few Mo-DCs are loaded with the antigen and it is unlikely that they play a role in the triggering of the T cell response.
Our study is focused on the mode of action of adjuvant MF59, but additional studies are required to understand if the findings described here are also valid for other emulsion adjuvants approved for human vaccination, such as AS03 [2][3][4]10]. However, consistent with our findings, an AS03 induced increase of antigen-loaded monocytes within the dLN has been described [29], suggesting that a similar process may be promoted by this other emulsion adjuvant.
Beyond what is the role of Mo-DCs in the adjuvanticity of MF59, we believe that our results are very relevant to shed a light on the mode of action of emulsion adjuvants. The data presented in this study contribute to create an emerging vision by which emulsion-adjuvants act through multiple mechanisms and have a multifactorial role in the overall picture of the immune response induced by vaccines containing this type of adjuvants. Emulsions-based adjuvants can play a critical role in improving the efficacy of human vaccines, however, only two squalene-based emulsion adjuvants, MF59 and AS03, are currently approved for human use in influenza vaccines [2][3][4]10]. Mechanistic insights into licensed emulsion adjuvants, especially on the role of DCs, can be fundamental to develop new improved emulsion-adjuvanted human vaccines [1,8,9,18,20]. Thus, our study offers a solid contribution to the challenging objective to produce the most efficacious vaccines possible, with an optimized risk/benefit balance, in order to have a positive impact on human global health through vaccination programs.

Study approval
The animal experiments were approved by local Novartis Animal Welfare Body and performed in compliance with the European directive and Italian law. The code of the approved mouse project are: AEC201111 (starting date November 25th 2011) and AWB2015-01 (starting date August 6th 2015). Mice were sacrificed by cervical dislocation. Surgery for two-photon microscopy studies was conducted as reported in the specific paragraph of Materials and Methods section.
Adjuvants. MF59 is Novartis proprietary oil-in-water emulsion consisting of 4.3% (vol/ vol) squalene, 0.5% Tween 80, and 0.5% Span 85, in citrate buffer (10 mM). MF59 was prepared by homogenization at 12,000 psi with a Microfluidizer, model 110Y (Microfluidics). The emulsion was sterilized by passage through a polysulfone filter with 0.22 μm pore size (Gelman Sciences) and stored at 4˚C. The mean particle size of the emulsion droplets determined with a Mastersizer X (Malvern Instruments) was 194±76 nm. Fluorescently labelled-MF59 was prepared formulating chloroform dissolved DiO fluorescent lipophilic tracer (Molecular Probes; Invitrogen-Life Technologies) in the squalene phase of MF59. The complete evaporation of chloroform was performed before homogenization of the emulsion as described above. The final concentration of DiO was 0.25 μg/ml.
Formulation with MF59 or DiO fluorescently labeled MF59 was prepared diluting the emulsion ½ in the volume dose with physiologic solution. The volume of the immunization dose was 10 μl.
Formulation with ALUM was prepared diluting, ALUM (2 mg/ml) in a solution (prepared with Water For Injection) with 10 mM histidine buffer pH 6.5 and 136 mM NaCl. The volume of the immunization dose was 20 μl.
Antibodies ; anti-CD45.1_APCCy7 (BD-Pharmingen; clone A20); anti-CD45.2_Alexa Fluor 700 (eBioscience; clone 104). Each antibody was titrated to determine the optimal labelling concentration for flow cytometry or confocal microscopy. Relative isotype controls of the same companies were used as negative controls. For flow cytometry, antibodies were firstly titrated individually against the relative isotype control, using LN cells of naïve mice. Then the optimal antibody dilution calculated for each antibody type was used to bind the antibodies to compensation beads (Anti-Rat and Anti-Hamster Ig k/Negative Control Compensation Particle Set, Beckton Dickinson). The same voltage set up for each fluorescence channel used for the antibody titration was applied to acquire compensation beads. Labeled compensation beads were acquired and the compensation was calculated automatically by the software. Then the experiments were run in the same instruments where the compensation was set up, labeling the LN cells with the antibody mix or the corresponding mix of isotype controls. Before to set the gating strategy, all the compensations for each couple of fluorochromes were carefully checked to verify that they were calculated correctly by the software and no over or under compensations were present.

Immunizations
Flow cytometry and confocal microscopy analysis: mice (3 mice per treatment) were immunized between toes (b.t.) of both legs with a dose of PBS (as negative control), PE or PE adjuvanted with plain or fluorescently labelled MF59 (PE + MF59) and popliteal dLNs were collected 15 minutes, 8 hours, 18 hours and 3 days after the treatment.
Alternatively, mice (3 per treatment) were immunized b.t. in both legs with a dose of PE or PE + MF59 and popliteal dLNs were collected 8 hours after the treatment. LNs from the right legs were immediately processed as described later (flow cytometry analysis indent), whereas each whole LN of the left legs was cultured intact 10 hours in 2 ml (24-well plate) of RPMI-1640 medium (Gibco, Invitrogen-Life Technologies) supplemented with 10% FCS (HyClone) and 1% PSG (EuroClone). After the culture these LNs were processed in the same way (flow cytometry analysis indent). In addition, mice (3 per treatment) were immunized b.t. or intramuscularly (i.m) in the calf muscle (right leg) and the popliteal dLNs were collected 18 hours after the treatment; otherwise mice (3 per treatment) were immunized b.t. (right leg) with PBS, PE, PE + MF59 or PE mixed with ALUM (ALUM-PE) and popliteal dLNs were collected 18 hours after the treatment.
Two-photon microscopy intravital imaging: mice (1 mouse per treatment) were injected b. t. in the right leg with a dose of PE or PE + fluorescently labelled MF59 and the intravital imaging of the popliteal dLN was performed for 15 minutes or 1 hour from the injection, respectively.
Two-photon microscopy 3D-organ imaging: mice (3 mice per treatment) were immunized b.t. in the right leg with a dose of PE or PE + fluorescently labelled MF59 and popliteal dLNs were collected 15 minutes, 8 hours and 18 hours after the treatment.
ELISA: mice (10 mice per treatment) were immunized b.t. or i.m. in the calf muscle (right leg) twice 4 weeks apart with PE and PE + MF59. Sera of mice were collected prior the first immunization (pre-immune), 2 weeks after the first immunization (Post I) and 2 weeks after the second immunization (Post II).

Flow cytometry analysis
LNs from the right legs were collected at the time points indicated and were immediately processed by enzymatic digestion. LNs of each group of mice were pooled in RPMI-1640 medium (Gibco, Invitrogen-Life Technologies) containing 250 αg/ml DNase I (Roche) and 500 μg/ml Liberase Research Grade (Roche) and incubated for 1h at 37˚C agitating by pipetting every 15 min. The obtained cellular suspension was collected by centrifugation at 300Xg for 10 min at room temperature. Then the cells were washed with RPMI-1640 medium (repeating the centrifugation) and were suspended in RPMI-1640 medium supplemented with 10% FCS (HyClone), 1% PSG (EuroClone). The obtained dLN cell suspension was filtered with a 70 μm Cell Strainer (Falcon, Becton Dickinson) and counted with a hemacytometer (Bright-Line). 2 million of LN cells were labelled for 20 min at 4˚C in dark conditions, in 50 μl of PBS containing Live/Dead Cell Stain Kit yellow (Molecular Probes; Invitrogen-Life Technologies) used according to the manufacturer's instructions and titrated fluorescent antibodies (as specified in Fig 1). Then labelled cells were washed with PBS by centrifugation at 300Xg for 10 min at room temperature. Washed cells were suspended in 200 μl of PBS and analyzed by flow cytometry using a LSRII instrument (Becton Dickinson). Roughly 500.000 live gated cells were acquired to make a flow cytometry analysis. Acquisition of the samples were performed using DIVA software (Becton Dickinson), whereas the data analysis were performed using either DIVA or FlowJo software (FlowJo LLC).

Confocal microscopy
LNs from the left legs were collected individually in dry conditions at the time points indicated, immediately frozen in O.C.T. (Tissue-Tek, Sakura) using liquid nitrogen and stored at -80˚C until processing. Cryosections of the LNs were obtained with the cryostat CM1950 (Leica) and stained using fluorescent antibodies. The cryosection (50 μm) were cut along the entire organ in order to analyze all the sections of the organs. The cryosection were fixed using PBS, 3% formaldehyde for 10 min at room temperature, washed twice with PBS and permeabilized with PBS, 3% BSA, 1% saponin (permeabilization buffer) for 30 min at room temperature. Tissue sections were then incubated with titrated fluorescent antibodies (as specified in the Fig 3) diluted in permeabilization buffer for 1h at room temperature in the dark. After washing 3 times with permeabilization buffer and once with PBS, stained tissue sections were sealed using Gold Antifade reagent (Invitrogen-life Technologies) and a coverslip. Images were acquired with Axio Observer LSM700 confocal microscope (Zeiss) at 20˚C using a Z stack tool.

Two-photon microscopy
Intravital imaging of real time translocation of PE and fluorescently labelled MF59 to popliteal dLN was performed as previously described [30,31].
3D-organ imaging of popliteal dLNs from mice immunized with PE and fluorescently labelled MF59 was performed as previously described [31,32].

Measurement of anti-PE antibodies
Serum PE-specific total IgG titers were measured by ELISA. Briefly, maxisorp 96-well plates (Nunc) were coated with a PE solution (2.5 μg/ml) in carbonate buffer (100 μl/well) overnight at 4˚C. Plates were then blocked by addition of PBS, 3% polyvinylpyrrolidone (PVP) (SERVA) (200 μl/well), incubated for 2 h at 37˚C and then washed once with PBS, 0.05% Tween20 (washing buffer). Serial dilutions (3-fold step) of standard and serum samples in PBS, 0.05% Tween20, 1% BSA were added to the wells and incubated for 2 h at 37˚C. Plates were then washed 3 times with washing buffer and incubated for 1 h at 37˚C with anti-mouse IgG-alkaline phosphatase (Sigma-Aldrich) solution (100 μl/well). After 3 washes the substrate p-nitrophenylphosphate (Sigma-Aldrich) (100 μl/well) was added for 30 min at room temperature. Absorbance at 405 nm was then measured by a plate spectrophotometer (BioTek-ASHI). Antibody titers were expressed as the reciprocal dilution corresponding to a cut-off at OD 405 = 0.5.

Sorting of APCs
Mice (Ly5) (5 mice per treatment) were immunized b.t. in both legs with a dose of PBS (as negative control), OVA or OVA adjuvanted with MF59 (OVA + MF59) and popliteal dLNs were collected 18 hours after the treatment. Popliteal dLN cell suspensions were prepared as previously described ("Flow cytometry analysis" paragraph). LN cells, at a cell concentration of 20 million/ml, were labelled, as previously described, to identify APCs, using Live/Dead Cell Stain Kit aqua (Molecular Probes; Invitrogen-Life Technologies) according to the manufacturer's instructions and titrated fluorescent antibodies (as specified in Fig 1). Labeled dLN cells were washed as previously described, suspended in PBS at a concentration of 20 million/ml and filtered through a 30 μm cell strainer (Becton Dickinson). APCs from dLN cells were sorted using a FACS Aria II instrument (Becton Dickinson) in 0-32-0 sort precision, applying the gating strategies specified in the text. Sorted APCs were collected in 500 μl of RPMI-1640 medium supplemented with 10% FCS (HyClone), 1% PSG (EuroClone).

T cell proliferation assay
CD4 T cells were isolated from the spleen of OT-II mice (one mouse per assay) preparing splenocytes by meshing the spleen through a 70 μm Cell Strainer (Falcon; Becton Dickinson) and afterwards by MACS negative selection using the CD4+ T cell isolation Kit (Milteny Biotec) according to the manufacturer's instruction. T cell purity, routinely around 98%, was determined by flow cytometry labeling the cells as previously described with fluorescent anti-CD4, -CD3, -CD8, -CD19 antibodies. One million of purified OVA-specific CD4 T cells were loaded with fluorescent CFSE (Molecular Probes) suspending them in 1 ml of PBS, 1 μM CFSE, for 10 min in dark condition at room temperature. CFSE loading reaction was stopped adding 1 volume of FCS (HyClone) and the cells were washed with 10 volumes of PBS, centrifuging at 300xg for 10 min at room temperature. Purified APCs (2,5x10 5 cells) were cultured with CFSE-loaded OVA-specific CD4 T cells (5x10 4 cells) (APC/T cell ratio: 5/1) in 200 μl (96-well plate, U bottom) of RPMI-1640 supplemented with 10% FCS (Hyclone) and 1% PSG (Gibco; Invitrogen-Life Technolgies). After 3 days of culture, CD4 T cell proliferation was assessed by flow cytometry measuring the halving of CFSE fluorescence. Cell cultures were washed with PBS by centrifugation at 300Xg for 10 min at room temperature, labeled with fluorescent anti-CD3, -CD45.1 and CD45.2 (to discriminate T cells from APCs) and washed as previously described. Then cells were suspended in 200 μl PBS and analyzed by flow cytometry using a LSRII instrument (Becton Dickinson).

Statistics
Statistical analyses are fully described in each figure and were chosen considering the assumption that MF59 enhances the effect we were measuring, based on published literature or previous data obtained in the study. imaging analysis of the popliteal dLN was taken 1 hour from the injection (procedure described in Materials and Methods section). The time-lapse of the antigen translocation to popliteal dLN is reported. Elapsed time is shown as hours:minutes:seconds. In red is shown the fluorescence signal of PE, in green is shown the fluorescence of MF59 whereas in blue is shown the fluorescence of the LN collagen capsule. The time-frame between each snap-shot is 30 seconds. After injection between fingers the translocation of antigen and MF59 to popliteal dLN is very rapid (less than a minute). Representative result of one experiment out of two is shown.