Exogenous S100A4 is a strong mucosal adjuvant

Based on our previous findings of endogenous S100A4 being critical for the in vitro antigen-presenting capacity of DCs, and for the in vivo mucosal adjuvant effect of CT,5 we wished to examine whether exogenous S100A4 might also work as a mucosal adjuvant in its own right. We tested this by immunizing mice intranasally on three occasions with a model antigen, OVA, administered with or without S100A4, and for comparison in a separate group of mice also with CT, and by examining immune responses 3 days after the last dose (Supplementary Fig. 1a). Serum anti-OVA total IgG, IgG1, and IgG2c levels were all robustly augmented after immunization together with S100A4, to levels comparable to those achieved using CT as adjuvant (Supplementary Fig. 1b). S100A4 also potentiated mucosal antibody production in mucosal tissues, resulting in strongly enhanced OVA-specific IgA, total IgG, and IgG1 antibody levels in lung exudate (Supplementary Fig. 1c). Importantly, the intranasal co-administration with S100A4 further resulted in enhanced antigen-specific IgA and IgG responses at a remote site, the vaginal mucosal surface (Supplementary Fig. 1d), consistent with previous studies showing a nasal-genital tract mucosal immunological link.15 Similar to the serum antibody responses, the immunization with S100A4 also promoted mucosal antigen-specific antibody production to levels comparable to those achieved by using CT as adjuvant (Supplementary Fig. 1).

Next, we largely repeated this experiment but halved the amount of S100A4 (10 μg per dose) and extended the antibody measurement to as long as 196 days (roughly 6.5 months) after the last immunization (Fig. 1a). At this substantially late time point, serum anti-OVA IgG antibody responses, including total IgG, IgG1, and IgG2c, remained high in S100A4-adjuvanted mice, not much less compared to a much earlier time point, 10 days after the last dose of immunization (Fig. 1b). Furthermore, mucosal anti-OVA IgG and IgA levels, in lung exudate (Fig. 1c), broncho-alveolar lavage fluid (BALF) (Fig. 1d), vaginal lavage (Fig. 1e), and feces (Fig. 1f), remained substantially high at this late time point in mice which received OVA admixed to S100A4. The S100A4-driven antibody sustainability was comparable to CT-adjuvanted immunization (Fig. 1b‒f). These late time point mucosal antibody levels were also found to be sustained in comparison to the levels in previous experiments measured at an earlier time point (Supplementary Fig. 1c, d), despite a higher dose of S100A4 being used in that immunization (20 μg per dose).

Fig. 1: Intranasal immunization adjuvanted with S100A4 provokes greater humoral and cellular immune responses.

Mice were immunized three times intranasally (i.n.) with ovalbumin (OVA; 10 μg) alone, or admixed to S100A4 (10 μg) or cholera toxin (CT; 1 μg) at a 10-day interval. Unmanipulated naïve mice were included for baseline control. Mice were euthanized 10 days or 196 days after the last immunization for collection of various types of samples. a The immunization and tissue collection procedures are indicated using a flowchart. b–f Various classes of OVA-specific antibody levels in serum (b), lung exudates (c), broncho-alveolar lavage fluid (BALF) (d), vaginal lavage (e) and pooled feces (f) were analyzed by ELISA. g–i Single-cell suspensions were prepared from the lung tissue (g), lymph nodes (h), and spleen (i). Percentages of CD8 T cells that recognize H-2Kb-OVA (SIINFEKL) tetramer were revealed by flow cytometry. Contour plots indicating a representative mouse from each treatment (upper panel) and pooled data from all the mice (lower panel) are shown. Numbers adjacent to outlined areas indicate percent cells in each gate. Each dot represents data from an individual mouse and blue lines indicate the average values. #P = 0.0022 or the exact P values (italic numbers) are indicated by Mann–Whitney U test. Except for the long-term observation experiment, the other measurements represent at least three separate experiments.

We also determined the accumulation of OVA257-264 peptide (SIINFEKL)-specific CD8 T cells in various tissues using the H-2Kb-SIINFEKL tetramer analyzed by flow cytometry (Supplementary Fig. 2). Co-administration with S100A4 promoted the expansion of CD8 T cells that recognized the MHC class I-associated SIINFEKL peptide in the lung (Fig. 1g), lymph node (Fig. 1h), and spleen (Fig. 1i) at 10 days after the last immunization. A trend of enhanced expansion of these T cells in the lymph nodes and spleen was observed when comparing the effect of S100A4 with CT (Fig. 1h, i). For SIINFEKL peptide-specific T cell expansion in the lung, we also measured at 196 days after the last immunization, and S100A4-adjuvanted immunization maintained the response better than CT (Fig. 1g). These findings suggest that S100A4 could be helpful in driving cytotoxic T cell responses, in addition to stimulating humoral immunity, and may exert a critical role in maintaining protective immunity against viral infections.

To examine if the mucosal adjuvant activity of S100A4 on the immune response to the model antigen OVA would extend to a clinically relevant vaccine antigen, such as the spike protein of SARS-CoV-2 (“Spike”), we immunized mice with Spike alone or together with either S100A4 or CT as adjuvants using a similar intranasal immunization regimen as with OVA (Fig. 2a). Again, co-immunization with S100A4 resulted in substantially increased levels of Spike-specific total IgG and IgG1 antibodies in serum (Fig. 2b), as well as Spike-specific mucosal IgA responses in BALF (Fig. 2c), lung parenchyma (Fig. 2d), nasal mucosa (Fig. 2e), and eye mucosa (Fig. 2f). It has recently been reported that respiratory tract mucosal IgA exerts major neutralizing activity against SARS-CoV-2.16,17

Fig. 2: S100A4 promotes spike protein-specific antibody responses after intranasal immunization with the recombinant spike protein of SARS-CoV-2.

Mice received three intranasal immunizations (i.n.) at an interval of 10 days. For each immunization, mice were intranasally administered with 5 μg spike protein alone or admixed with 10 μg S100A4 or 1 μg cholera toxin (CT). Samples were collected 10 days after the last immunization (a). Levels of anti-spike protein IgG and IgG1 in serum (b) as well as anti-spike protein IgA in broncho-alveolar lavage fluid (BALF) (c), lung exudates (d), nasal lavage (e), and eye exudates (f) were measured by ELISA. Each dot represents measurement from an individual mouse and columns indicate the average values. #P = 0.0079 by Mann–Whitney U test. Shown is one of two similar experiments.

S100A4 extends nasal residence time and promotes tissue transport of antigens

Nasal mucociliary clearance is a mechanism that limits the nasal residence time of vaccine antigens delivered intranasally,18 thus compromising mucosal antigen uptake. Efficient cell-mediated antigen transport to tissues and secondary lymphoid organs is critical to achieving vaccine efficacy.19 We performed in vivo real-time fluorescence optical imaging to monitor the kinetic biodistribution of intranasally administered fluorescently labeled OVA, either in the absence or presence of S100A4. Ventral and dorsal images of live mice were captured at various time points up until 6 h after the administration of Alexa Fluor 647-conjugated OVA (OVA-AF647). Shortly after the intranasal administration of OVA-AF647, identical ventral view fluorescence signals were observed at the snouts of mice (5 min; Fig. 3a; ventral view). No fluorescence signals were detected from the dorsal side at this time point, probably due to blocking of the signal by the overlying turbinate bone. As can be clearly observed ventrally, the fluorescent signal persisted more robustly in the snouts of the mice that received the fluorescently labelled antigen together with S100A4, suggesting a longer nasal residence time of the antigen (Fig. 3a, b).

Fig. 3: S100A4 promotes nasal antigen retention and antigen transport.

a–f Mice received a single intranasal delivery of Alexa Fluor 647-conjugated ovalbumin (OVA-AF647; 10 μg) in the absence or presence of S100A4 (20 μg) or cholera toxin (CT; 1 μg). Unmanipulated naïve mice were included for background control. The ventral and dorsal images showing the biodistribution of OVA-AF647 in the mice were taken at various time points as indicated using an in vivo animal imaging system (a) and nasal retention of OVA-AF647 from the ventral view was quantified (b). Mice were euthanized at 6 h after antigen delivery and dorsal skin was cut open to expose the internal organs around the spleen site which was imaged (c, bottom; arrows indicate the spleen) and fluorescence accumulation in the spleen was quantified (c, top). Also at 6 h after OVA administration, mouse nasal tissues (d), cervical lymph nodes (e) and lungs (f) were harvested for flow cytometric analysis. Representative contour plots indicating the biodistribution of OVA-AF647 in one representative mouse out of four in each treatment group are shown. g Bone marrow-derived dendric cells (BMDCs) were treated with or without S100A4 (1 μg/ml) overnight and stained with CellTracker™ (CMFDA-green). Mice were then injected with 2 × 106 BMDCs into the footpad and popliteal lymph nodes were harvested 24 h later for flow cytometric analysis. Numbers in or adjacent to outlined areas indicate percent cells in each gate (d–f). Each dot represents measurement from an individual mouse (c). Data are expressed as mean + s.e.m. of four biological replicates (b–g). #P = 0.0286 by Mann–Whitney U test (b, c). ###P < 0.0001 by two-way ANOVA comparing the two groups for all the time points (b). ##P < 0.0001 or the exact P values (italic numbers) are indicated by unpaired t-test (d–g). Shown is one of two representative experiments.

Furthermore, co-administration with S100A4 enhanced the dorsal appearance of the fluorescence signals associated with tissues around the spleen, which started to appear at 30 min, peaked at around 4 h, and declined at 6 h post-antigen delivery (Fig. 3a; dorsal view). We then euthanized the mice at about 6 h after antigen delivery and removed the skin at the left side of the abdominal cavity in order to get clearer views of the fluorescence deposition in relevant organs. The co-administration with S100A4 clearly promoted the accumulation of fluorescent OVA in the spleen, stomach, and kidney (Fig. 3c). In particular, the spleen demonstrated the most remarkable S100A4-driven antigen accumulation, as the mouse that received only fluorescent OVA showed merely the background fluorescence (Fig. 3c). For both the nasal retention and tissue accumulation of the fluorescent OVA, S100A4 demonstrated stronger effects than the CT used for comparison (Fig. 3a‒c).

Next, we collected nasal tissues, cervical lymph nodes and lung tissues from each individual mouse at 6 h after the delivery of fluorescent OVA and prepared single-cell suspensions to investigate, using flow cytometry, the distribution of the fluorescent OVA at the cellular level including the CD103+ migratory DCs.20 Co-administration with S100A4 profoundly promoted accumulation of migrating DC (CD45+CD103+CD11c+MHC IIhigh/int)- and macrophage (CD45+F4/80+)-associated fluorescent OVA in the nasal tissue (Fig. 3d), cervical lymph nodes (Fig. 3e), and lung tissue (Fig. 3f). Of note, the antigen-associated DCs identified in the nasal tissue had a lower expression level of MHC class II. In contrast, MHC class IIhigh DCs did not carry fluorescent OVA (Supplementary Fig. 3a). CT only increased the cell-associated OVA accumulation in the lungs (Fig. 3f), but not nose (Fig. 3d) and lymph nodes (Fig. 3e) at 6 h after antigen delivery. At 12 h after OVA delivery, even more potent effects of S100A4 on the accumulation of antigen-loaded antigen-presenting cells in the lungs could be revealed, while CT showed at least a trend of weaker effects at this time point compared with S100A4 (Supplementary Fig. 3b). Efficient uptake and transport of vaccine antigens have been described as a critical quality for assessing the adjuvant activity.21

Using in vitro cultured bone marrow-derived DC (BMDC), we failed to observe any augmentation in the uptake of fluorescent OVA after treatment with either S100A4 or CT (Supplementary Fig. 4a). Similarly, uptake of fluorescent OVA by fresh ex vivo cells from the nasal tissue, cervical lymph nodes, and spleen was not enhanced by S100A4 or CT (Supplementary Fig. 4b). Thus, it is likely S100A4 promoted the migration of antigen-presenting DCs from the nasal mucosa but did not promote antigen uptake itself. Our data showing S100A4-promoted antigen accumulation in the nasal tissue, cervical lymph nodes and lung tissues after nasal delivery of the antigen (Fig. 3d‒f) support a stimulatory effect of S100A4 in promoting the migration of antigen-presenting cells including DCs and macrophages that had captured the antigen after mucosal delivery. It has been previously reported that S100A4 is critical for regulating the motility of macrophages.9 We also observed the distribution of S100A4 in the lung tissues and spleen 24 h after intranasal inoculation (Supplementary Fig. 5), which supports the long-distance function of S100A4.

To fundamentally confirm the effect of S100A4 on DC migration, we designed an adoptive transfer assay to unambiguously demonstrate more efficient DC migration driven by S100A4. Control and S100A4-primed BMDCs were labelled with a green fluorescent dye for easy tracking. Following injection of the fluorescent DCs into the mouse footpad, substantially increased numbers of S100A4-primed DCs arrived in the draining lymph nodes compared with the control DC (Fig. 3g).

S100A4 effectively induces germinal center responses after intranasal immunization

The development of germinal centers is a recognized mechanism for the generation of high-quality antibodies after immunization.22,23 We harvested mouse spleens and cervical lymph nodes after intranasal immunization for flow cytometric analysis of the impact of S100A4 on immune cell activation, including the germinal center response, using the gating strategy defined in Supplementary Fig. 6. Frequencies of activated non-germinal center B cells (CD69+CD38+B220+)24 were increased in both the spleen and cervical lymph nodes of mice that received the antigen together with S100A4 (Fig. 4a). In both compartments, S100A4 also expanded germinal center B cells (GL7+FAS+CD38–B220+) (Fig. 4b) and T follicular helper cells (PD-1+CXCR5+CD4+Foxp3–), a cell type which is critical in the formation of germinal centers (Fig. 4c). The immune enhancing effects of S100A4 were similar to those of CT (Fig. 4a‒c). These results were confirmed using immune staining and confocal microscopy to demonstrate the typical germinal center morphology. Mouse spleens were first cryosectioned, followed by immunofluorescent staining for the expression of B220, GL7, as well as the cell proliferation marker Ki-67, as previously reported.5 Immunization with S100A4 strongly promoted the expansion of GL7+ and Ki-67+ B cells and morphological changes associated with germinal center formation (Fig. 4d).

Fig. 4: Intranasal immunization adjuvanted with S100A4 activates secondary lymphoid organ cells and promotes germinal center formation.

Mice received three intranasal immunizations as explained in Supplementary Fig. 1a. Spleens and cervical lymph nodes (CLN) were collected 3 days after the last immunization. a–c Single cell preparation was analyzed by flow cytometry. Frequencies of CD69+CD38+ activated non-germinal center B cells (a), GL7+FAS+CD38‒ B cells that characterize germinal center formation (b), and PD-1+CXCR5+Foxp3‒ T follicular helper cells (c) were analyzed. Each contour plot indicates data from one representative mouse and numbers in or adjacent to outlined areas indicate percent cells in each gate (left panels). The bars show the pooled data (right panels). d Spleen sections were analyzed by immunohistochemistry to visualize germinal center morphology. Red, B220; green, GL7; blue, Ki-67. The top panels of the microscopic pictures show merged images. The bottom bar figure panels show the quantification of staining intensities. Scale bar, 25 μm. Each dot represents data from an individual mouse and columns indicate the average values. #P = 0.0022; ##P = 0.0043 or the exact P values (italic numbers) are indicated by Mann–Whitney U test. Shown is one of two representative experiments.

S100A4-adjuvanted intranasal immunization promotes lipid accumulation in the spleen

It has recently been reported that germinal center B cells require active oxidation of fatty acids instead of glycolysis to meet the energetic challenge of rapid cell proliferation.25 In order for efficient germinal center development to take place, hosts need to mobilize lipid transport to the germinal centers. Therefore, we examined the accumulation of lipids using MALDI-TOF MS after S100A4-adjuvanted immunization. A variety of lipids, with glycerolipids and glycerophospholipids being the dominant forms, were identified in the spleens (Supplementary Fig. 7). A shotgun lipidomics approach was used to analyze all lipid classes together. The abundance of total lipids was increased in the spleen after co-immunization with S100A4 (Fig. 5a). The lipid intensities were quantified and expressed using an intensity box plot which displayed a remarkable increase of lipids in spleens from S100A4-adjuvanted mice (Fig. 5b). Principal component analysis (PCA) revealed a complete separation of S100A4-adjuvanted immunization from both the OVA only immunization and the naïve mouse controls (Fig. 5c). A similar and more intense lipid accumulation response than that induced by S100A4 was seen in mice given CT together with OVA (Fig. 5).

Fig. 5: Intranasal immunization adjuvanted with S100A4 promotes lipid accumulation in the spleen.

Mice received three intranasal immunizations as explained in Supplementary Fig. 1a. Spleens were collected 3 days after the last immunization and sectioned for MALDI-TOF analysis. a The intensity and distribution of lipids within a range of mass-to-charge (m/z) ratio (754.8 ± 314.3 Da) are shown as MALDI MSI ion pseudo-color images. Scale bar, 6 mm. b Expression of overall levels of the identified lipids were quantified by intensity box plots. The intensity of a mass peak from each mass spectrum is depicted by a spot within the plot, and the outliers are colored red. c Principal component (PC) analysis of MALDI MSI data using the score plot of the three first principal components (PC1–PC3) displays various lipid mass spectra highlighting different lipid signatures between various treatment groups. Results show one representative mouse (a) or represent 3–6 mice (b, c) in each treatment group. ****P < 0.0001 by Kruskal–Wallis test (b). OVA, ovalbumin.

As our germinal center activity investigation using various approaches was based on the same cohort of mice, we compared the lipid accumulation and the expression of GL7, an established germinal center marker, in individual mice. Good correlations were established between the expression of GL7 using flow cytometry and fluorescence microscopy and the lipid abundance determined using MALDI-TOF (Supplementary Fig. 8), supporting the usefulness of lipid abundance as a surrogate marker for the germinal center response.

Extracellular S100A4 potentiates DC activation

DCs are critical antigen-presenting cells that bridge innate and adaptive immune responses and, as such, are rational targets for the choice of effective adjuvants.26 We have previously demonstrated that adoptive engraftment of wild-type DCs could restore the defective adaptive immune responses of S100A4-deficient mice.5 Here we examined the regulatory effect of exogenous S100A4 on cultured DCs. Our data clearly demonstrates that mRNA expression of a group of cytokines and molecules with immune regulatory activities was upregulated in BMDCs following incubation with S100A4 (Fig. 6a). Specifically, the transcript levels of a number of cytokines that are critical to adaptive immunity, including IL-1β, IL-2, IL-6, and IL-10, were augmented after treatment with S100A4, as were co-stimulatory molecules, including CD80, CD86, and CD40. Furthermore, at the protein level, S100A4 was found to potently activate BMDCs by augmenting the production of these co-stimulatory molecules, as well as IL-1β and IL-6 (Fig. 6b).

Fig. 6: S100A4 activates dendritic cells.

a, b Bone marrow-derived dendritic cells (BMDCs) were incubated with or without 1 μg/ml S100A4 for 3 h (a) or overnight (b). Some cells were stimulated with 1 μg/ml LPS as a positive control. Expression of a number of molecules important for DC activation as indicated was analyzed at the mRNA level by quantitative reverse transcription PCR (a) or at the protein level by flow cytometry (b; the lower panel shows the quantification). c BMDCs were cultured from wild-type or NFκB-deficient mice followed by incubation overnight with or without S100A4 or cholera toxin (CT; 0.1 μg/ml). Cell surface expression of CD80 and CD86, and intracellular production of IL-1β were measured by flow cytometry. d BMDCs were incubated with anti-RAGE (left panel) or anti-TLR4 (right panel), or with isotype control antibodies, followed by flow cytometric analysis. e BMDCs were incubated with or without S100A4 and their surface expression of RAGE was determined as explained in (d). f BMDCs were incubated overnight with S100A4 or S100A4 pre-mixed with FPS-ZM1 or TAK-242, inhibitors of RAGE and TLR4, respectively. Cell surface expression of CD86 (left panel) and intracellular production of IL-1β (right panel) were measured by flow cytometry as explained in (b). g BMDCs or a DC cell line (DC2.4) were treated with OVA (1 mg/ml) mixed with or without S100A4 (1 μg/ml) or CT (1 μg/ml) overnight before co-culturing with RF33.70 hybridoma (T cells with T cell receptor specific for the OVA257-264 peptide) at a 1:1 (BMDC to RF33.70) or 1:10 (DC2.4 to RF33.70) ratio overnight. T cell activation was quantified by measuring IL-2 release using ELISA, with a schematic diagram showing the experimental design. Expression was normalized using GAPDH as the calibrator gene (a). Representative data out of at least three separate BMDC cultures are shown (d, e), or data are expressed as mean + s.e.m. of 3–4 individual cell cultures (a, b, c, f, g). #P = 0.0022 or the exact P value (italic number) is indicated by Mann–Whitney U test (c, g). Alternatively, the exact P values (italic numbers) are indicated by unpaired t-test (b).

We have previously shown a dependence on NFκB signaling as a critical DC activation pathway used by CT to exert mucosal adjuvant activity.27 Here, we can also demonstrate that S100A4-mediated DC activation is at least partially dependent on NFκB signaling. Thus, S100A4-induced production of IL-1β, a critical molecule in the pathway of adjuvanticity, was substantially suppressed in NFκB–/– BMDCs, although the increased expression of CD80 and CD86 by S100A4 remained intact in NFκB–/– cells (Fig. 6c). The latter finding may suggest that S100A4 could have a broader spectrum of signaling pathways than CT, which failed to increase both co-stimulatory molecule expression and IL-1β production in NFκB–/– BMDCs (Fig. 6c). Extracellular S100A4 has been shown to activate NFκB in various types of cells.11,28,29 Therefore, it is not surprising that S100A4, as shown here, also stimulates DCs through the NFκB pathway.

It is reported that S100A4 triggers intracellular signaling pathways through RAGE and TLR4.30 Indeed, BMDCs expressed both RAGE and TLR4 (Fig. 6d), and we found that RAGE expression was upregulated by exogenous S100A4 (Fig. 6e). We then tested whether S100A4-mediated DC activation would require either or both of these two receptors. To this end, BMDCs were incubated with S100A4 either in the presence or absence of FPS-ZM1, an inhibitor of RAGE, or TAK-242, an inhibitor of TLR4, and the effects on CD86 expression as well as production of IL-1β were measured. The results showed that S100A4-induced upregulation of both of these activation markers was almost completely abrogated by blocking either RAGE or TLR4 (Fig. 6f), suggesting the requirement of both receptors for S100A4-induced DC activation.

Extracellular S100A4 enhances MHC class I antigenic peptide cross-presentation by DC and activation of cytotoxic T cells

We have demonstrated the potent effect of S100A4 in stimulating cytotoxic CD8 T cells against a co-administered antigen (Fig. 1g‒i). Next, we wished to also confirm that S100A4 could enhance MHC class I-associated antigenic peptide cross-presentation on DCs, a critical step in the activation of cytotoxic T cell responses after immunization with protein-based subunit vaccines.31 To this end, we incubated OVA-pulsed DCs either in the presence or absence of S100A4, followed by co-culturing the DCs with RF33.70 hybridoma T cells that had T cell receptors (TCRs) specific for OVA257-264 presented by H-2Kb. S100A4-pulsed DCs substantially augmented the activation of T cells as measured by IL-2 production (Fig. 6g). S100A4 demonstrated a more potent effect compared with CT (Fig. 6g).

The adjuvanticity of S100A4 is not affected by contaminating lipopolysaccharide (LPS)

We performed both in vivo immunization and in vitro DC activation assays to rule out the possible confounding effect of a residual low amount of endotoxin present in the recombinant S100A4 product (10 pg LPS per μg S100A4, as measured using a Limulus amebocyte lysate assay). In vivo, mice were given three intranasal immunizations at ten-day intervals with OVA in the presence of various doses of S100A4 or various amounts of LPS corresponding to the residual LPS amount in a given S100A4 dose. Ten days after the last immunization, blood was collected and OVA-specific total IgG antibody production in serum was measured. The tested range of LPS amounts failed to promote OVA-specific IgG production, supporting that the enhancement of this response by S100A4 is a true adjuvant effect of S100A4 but not the result of contaminating LPS (Supplementary Fig. 9a). On the other hand, as a positive control, LPS at 1 μg (5000-fold of the highest contamination level) effectively promoted the anti-OVA response (Supplementary Fig. 9a).

In the in vitro experiments, we incubated BMDCs with LPS at 10 pg/ml, i.e., the estimated contamination level of LPS in 1 μg/ml S100A4, which was tested in parallel, and observed no enhancement in the expression of the genes under study (Supplementary Fig. 9b). In contrast, a 100,000-fold higher level of LPS (1 μg/ml) demonstrated potent stimulatory effects (Supplementary Fig. 9b). Further support for the S100A4 protein, but not the residual endotoxin contamination, being the entity responsible for DC activation came from studies showing that the strong promotion of CD80, CD86 and IL-1β in BMDCs by the recombinant S100A4 was largely suppressed by a neutralizing antibody against S100A4 (Supplementary Fig. 9c).

S100A4 does not induce humoral immune responses against itself after intranasal immunization and does not induce inflammation of the olfactory bulb

In vaccination practice, an ideal adjuvant should be able to support different vaccine antigens in multiple administrations, so that the adjuvant can be formulated in different vaccine preparations for subsequent innoculation. Therefore, immune responses against the adjuvant itself should preferably be avoided. To assess the quality of S100A4 in this respect, we repeated our basic intranasal immunization model using OVA as an experimental antigen admixed with either S100A4 or CT as an adjuvant (Supplementary Fig. 10a). Consistent with the previous findings (e.g., Supplementary Fig. 1b and Fig. 1b), we again showed that both S100A4 and CT were able to remarkably augment anti-OVA IgG levels (Supplementary Fig. 10b, c). Interestingly, while CT, as a foreign protein, stimulated substantial levels of anti-CT IgG antibody responses (Supplementary Fig. 10c), S100A4 failed to stimulate anti-S100A4 IgG antibody production above the level seen in naïve mice (Supplementary Fig. 10b). This could be due to the fact that S100A4 is an internal protein subject to immune tolerance. The data suggests that S100A4 could lead to improved tolerability over non-host proteins (e.g., CT) in repeated use as a vaccine adjuvant.

It is generally believed that adjuvant with neurotoxicity may affect olfactory bulbs through retrograde axonal transport from the olfactory epithelium, resulting in the inflammatory responses in the olfactory bulb, and this may disrupt the normal functionality of the central nervous system.32 Thus, we stained for the leukocyte infiltration in the olfactory bulb, a sign of inflammation. Compared with CT which caused substantial leukocyte accumulation in the olfactory bulb, S100A4 did not induce such inflammatory responses (Supplementary Fig. 11a). Consistently, we could show that S100A4 did not trigger increases in flammatory cytokines including IL-6 and IL-12 in the olfactory bulb after intranasal immnunization (Supplementary Fig. 11b). Taken together, these data preliminarily support a good safety profile of S100A4 at least with respect to olfactory bulb inflammation.