All mice were bred and housed under specific pathogen-free conditions at the Präklinische Experimentelle Tierzentrum of the Medical Faculty in Erlangen, Germany, or at the Statens Serum Institute, Denmark. Frozen sperm cells of TNFR1^fl/wt mice (41) were obtained from the European Mouse Mutant Archive in Athens, Greece (EMMA strain 11019, B6.129P2(Cg)-Tnfrsf1a^{tm3.3Gkl/Flmg}, archived on a C57BL/6 background), and used to re-derive the mice by in vitro fertilization at the Transgenic Facility of the Friedrich-Alexander-University Erlangen-Nürnberg. To generate cell type-specific conditional TNFR1 KO lines, TNFR1^fl/fl mice were crossed with LysM-Cre (B6.Lyz2^tm1(cre)Ifo (44);), Clec9a-Cre B6J.B6N(Cg)-Clec9atm2.1(icre)Crs/J (45);), CD11c-Cre (B6.Cg-Tg(Itgax-cre)1-1Reiz/J (46);), or Lck-Cre (Tg(Lck-cre)1Jtak (47)) mice. Mincle^tg mice (48) were obtained from Sho Yamasaki and crossed with Mincle KO mice (Clec4e^tm1.1Cfg) generated and provided by the Consortium for Functional Glycomics (49). Tnf^-/- mice (50) were provided by Dr. Ulrike Schleicher. C57BL/6N mice were purchased from Charles River Laboratories. At the end of the experiment the mice were euthanized by cervical dislocation. All mouse experiments were approved by the Regierung von Unterfranken (protocol number 55.2.2-2532-2-1641).

IL-17A fate reporter mice were made by crossing the Il17a^{tm1.1(icre)Stck}/J (IL-17cre) strain with the B6.129X1-Gt(ROSA)26Sor^tm1(EYFP)Cos/J (R26R-EYFP) strain (51) (from The Jackson Laboratory (Bar Harbor, USA)) and handled at the experimental animal facility at Statens Serum Institut. At the end of the experiment the mice were euthanized by cervical dislocation. Experimental work was conducted in accordance with the regulations of the Danish Ministry of Justice and the Danish National Experiment Inspectorate under permit 2017-15-0201–01363 and in compliance with the European Community Directive 2010/63 EU for the care and use of laboratory animals.

To study the cellular immune response, we used a 7-day immunization protocol, as previously described (26, 34, 52). Mice were immunized s.c. in both footpads with 50 µl CAF01 (23) mixed with 1 µg H1 (43) per foot, except in the IL-17A fate-reporter mice, where s.c. immunization was performed at the base of tail using a 100 μl volume. Unimmunized mice were injected with the same volume of PBS. All footpad immunizations were performed on unconscious animals after inhalation anesthesia with 4% Isoflurane. For the immunization in the base of tail (s.c.) no anesthesia was used. After footpad immunization, the footpad thickness was measured before the immunization (day 0) and every second day after immunization. The footpad swelling was calculated by subtracting the initial thickness (on day 0) from the later obtained values. On day 7 after immunization, the mice were killed, and the inguinal and popliteal lymph nodes as well as the spleen were isolated.

For the experiments requiring TNF blockade, Etanercept (3 mg/kg in 100 µl PBS) was injected s.c. in the flank (without anesthetization) on the day of immunization and every other day thereafter. The control group was injected in parallel with 100 µl PBS.

For the depletion of NK1.1-positive cells, an anti-NK1.1 antibody (clone PK136, Leinco Technologies) was used in addition to etanercept treatment in the 7-day immunization protocol. 250 µg/mouse of this antibody or an isotype control (clone C1.18.4, Leinco Technologies) was injected i.p. (without anesthetization) one day before and a second time two days after immunization with H1/CAF01.

To study the humoral immune response, a 5-week immunization protocol was used. Here, the mice received a second immunization at day 21 (2 µg H1 in 100 µl CAF01, s.c. base of tail). Blood was collected at day 0 (no H1-specific Ab detectable, data not displayed) and after the killing on day 36.

To assess the role of TNF on the Ab production, additionally etanercept (3 mg/kg in 100 µl PBS) was injected s.c. in the flank on the day of immunization and every other day until day 18. Then, to prevent hindrance of the TNF block by the formation of anti-etanercept Ab, we switched to injecting anti-murine TNF Ab (200 µg/mouse, clone TN3-19.12, Leinco Technologies) every second day starting from day 20. A control group was injected with a matching isotype control (clone PIP, Leinco Technologies). All s.c. and i.p. injections of antibodies and etanercept were performed without anesthetization. In this protocol, the mice were killed 1 week after the booster immunization and blood was collected.

Bone marrow cells were isolated from C57BL/6 mice and Mincle^-/-; Mincle^tg mice and differentiated as described before (34). TDM was dissolved in isopropanol and coated to the bottom of the well (final concentration of 2 µg/ml) by allowing the isopropanol to evaporate. For the control wells, the same volume of isopropanol was evaporated. The BMDM were seeded in a flat-bottom 96-well plate (2*10⁵ cells/well). Etanercept 100 µg/ml was added to the respective wells. The cells were stimulated for 24 h.

To isolate monocytes from the bone marrow, the monocyte Isolation Kit (Miltenyi Biotec, Order No.: 130-100-629) was used according to the instructions of the producer. BM monocytes were stimulated by seeding them into a flat-bottom 96-well plate (2*10⁵ cells/well). The wells were coded before with isopropanol, TDB (5 µg/ml), or TDM (2 µg/ml) as described above. Etanercept (final concentration 100 µg/ml) was added to the respective wells. The cells were stimulated for 24 h.

A single cell suspension of lymph node cells was obtained by mechanically dissociating the LN using a 70 µm cell strainer. The spleen was also mashed through a 70 µm strainer, and additionally, the erythrocytes were lysed using ammonium chloride. 5 x 10⁵ LN and spleen cells were restimulated in a 96-well U-bottom cell culture plate for 96h with H1 (1 µg/ml).

Cytokines were measured in the supernatants of restimulated cells. For this purpose, the following ELISA kits were used according to the manufacturer’s instructions: IFNγ (R&D Systems, DY485), IL-17 (R&D Systems, DY421), IL-10 (R&D Systems, DY417), and IL-6 (R&D Systems, DY406).

After blocking with anti-mouse CD16/32 antibody (eBioscience) the dead cells were stained with the Fixable Viability Dye eFluor 506 (Invitrogen 65-0866-14). Next, fluorophore-coupled antibodies were used to stain for the surface markers (see Supplementary Table S1 for a list of the used antibodies). For the staining of Mincle, an uncoupled primary antibody (MBL, clone 4A9) was used and detected by using a suitable fluorescent-labeled secondary antibody.

For intracellular cytokine staining, 1.5 x 10⁶ cells were restimulated in a 96-well plate using 5 µg/ml of each of the three H1 peptides Ag85B _p241-255, Ag85B _p261-280, ESAT-6 _p1-15 (obtained from Peptides and Elephants) or using the whole H1 protein. The cells were stimulated for one hour, followed by five hours in the presence of Brefeldin (10 µg/ml). After surface staining and fixation with 1% PFA overnight at 4 °C, the cells were permeabilized by saponin, and antibodies for the intracellular staining were added.

Flow cytometry measurements were performed using an LSRFortessa (BD Bioscience, 5-laser configuration, Model No. 647794E6). The flow cytometry data were analyzed using FlowJo (BD Live Sciences, v. 10.7.1).

We defined the cell types by the following marker (gating strategy in Supplementary Figure S3): B cells (CD19⁺, CD3-), neutrophils (CD3^-, CD19^-, NK1.1^-, Ly6C⁺, Ly6G⁺), classical monocytes (CD3^-, CD9^-, NK1.1^-, Ly6G^-, CD11b⁺, Ly6C^high), Ly6C^low monocytes (CD3^-, CD19^-, NK1.1^-, Ly6G^-, CD11b⁺, Ly6C^low), NK cells (CD3^-, CD19^-, NK1.1⁺), NK T cells (CD3⁺, CD8^-, CD4^-, NK1.1⁺), macrophages (CD3^-, CD9^-, NK1.1^-, Ly6G^-, CD11b⁺, Ly6C^low, F4/80⁺), cDC1 (CD3^-, CD19^-, NK1.1^-, CD88^-, CD26⁺, CD64^-, MHCII⁺, CD11c⁺, XCR1⁺), cDC2 (CD3^-, CD19^-, NK1.1^-, CD88^-, CD26⁺, CD64^-, MHCII⁺, CD11c⁺, XCR1^-) CD4 T cells (CD3⁺, CD19^-, γδ TCR^-, CD4⁺, CD8^-), CD8 T cells (CD3⁺, CD19^-, γδ TCR^-, CD8⁺, CD4^-), γδ T cells (CD3⁺, CD19^-, γδ TCR⁺) Th17 cell (CD3⁺, CD4⁺, CD8^-, IFNγ^-, IL-17⁺), Th1 cell (CD3⁺, CD4⁺, CD8^-, IFNγ⁺, IL-17^-).

Cells were lysed using Tri reagent (Bio&Sell, order No: BS67.211.0100). The RNA was isolated using the Direct-zol™ -96 RNA isolation Kit (ZymoResearch, cat. No.: R2056) according to the instructions of the producer. The RNA was transcribed to cDNA using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, cat No.: 4368814).

To perform the qPCR, primers and probes from the KiCqStart^® Probe Assay (Sigma-Aldrich) were used for Clec4e (Mincle) and Hprt (sequence see Supplementary Table S2). For Il6 and Il1b, the primers were used together with a probe from the Roche UPL universal probe library (sequence see Supplementary Table S2). Gene expression was normalized using the Hprt expression, and the fold change was calculated using the 2^-ΔΔct method (53), with isopropanol-treated samples serving as calibrators.

Blood was collected at the end of the experiment by retroorbital bleeding from unconscious animals after i.p. application of Ketamin (100 µg/g bodyweight) together with Xylazin (10 µg/g bodyweight). Afterwards, still under narcosis, the mice were killed by cervical dislocation. Sera were separated from the blood by centrifugation (900 g, 10 min). H1 (400 ng/ml) was adsorbed overnight onto an ELISA plate (Sarstedt, ref: 82.1581.200). The non-bound H1 was discarded, and after BSA block (1% BSA in PBS), the sera were applied at different dilutions. The sera were washed away and the bound H1-specific Abs were detected using the following HRP-coupled secondary Ab: goat anti-mouse IgG1 (Southern Biotech, 1071-05, 1:4000), goat anti-mouse IgG2c (Southern Biotech, 1078-05, 1:4000), goat anti-mouse IgG2b (Southern Biotech, 1091-05, 1:4000).

For statistical analysis, GraphPad Prism (GraphPad Software, LLC, version 9.5.1) was used. The used statistical test for every graph is indicated in the figure legends. In all graphs, each dot corresponds to one mouse, the height of the bar indicates the mean. For the two-way ANOVA tests the normal distribution of the data was tested using the Shapiro-Wilk test. If the requirement of the normal distribution was not meet, the ANOVA was performed on log transformed values, this is indicated in the figure legend.

The analysis of Ab titers was only done for immunized groups. To this end, a correlation was modeled between the measured OD values and the respective dilution step. For this, a three-parameter non-linear regression with the equation OD = Bottom + (Top-Bottom)/(1+(titer/mid-titer point)) was used. The mid-titer point was obtained together with its SEM, and to compare these values between different groups, a one-way ANOVA followed by Dunnett’s multiple comparison test was performed (or Mann-Whitney U test in case of two groups).

The p-values of all tests are displayed on the respective comparisons; to maintain clarity, only meaningful comparisons are shown in the graphs.

We previously demonstrated that TNF is required for antigen-specific IL-17 production in mice vaccinated with the TB antigen H1, combined with the Mincle-dependent adjuvant CAF01 (34). Here, we immunized mice using the same antigen/adjuvant system while blocking TNF with Etanercept (see Figure 1A), restimulated splenocytes with H1 antigen and analyzed production of IL-17 and IFNγ by CD4⁺ T cells using intracellular cytokine staining (Figure 1B). While less than 0.1% of all CD4+ T cells from naïve mice produced IFNg or IL-17 upon stimulation with H1, the frequency of H1-specific IFNg+ Th1 cells and IL-17+ Th17 cells was around 2% and 0.6%, respectively, seven days after a single s.c. injection of CAF01/H1 (Figure 1B; gating strategy shown in Supplementary Figure S1A). The frequency of IL-17-producing cells among activated CD44+ CD4+ cells is approximately 3% (Supplementary Figure S1B), which is similar to previous results in a comparable antigen/adjuvant system after repeated immunization (54). Consistent with our previous ELISA-based results, the frequency of antigen-specific Th17 cells was reduced by TNF blockade during immunization, whereas the H1-specific Th1 response remained unaffected (Figure 1B; gating strategy shown in Supplementary Figure S1).

We next sought to understand whether TNF is necessary to induce or to maintain antigen-specific Th17 cells. To address this question, we utilized Th17 fate reporter mice, in which IL-17A expression results in Cre recombinase-mediated constitutive YFP expression (51). Indeed, while only few YFP⁺ cells were detected in spleens of naive mice (Figure 1c), CAF01/H1 induced a significant increase of YFP⁺ cells, to a percentage comparable to the frequency of IL-17A⁺ cells (0.3 vs. 0.5%). In mice immunized under TNF blockade, the number of YFP-marked cells decreased to the level found in unimmunized mice (Figure 1c). In mice immunized without Etanercept, around 30% of YFP-marked cells produced IL-17 after H1 re-stimulation. The YFP-marked cells detected in mice treated with TNF blockade were unable to produce IL-17 after specific restimulation, comparable to unimmunized mice (Figure 1D). This small percentage of YFP+ cells that are non-reactive to H1 likely represents Th17 cells with other antigen specificities. Thus, we conclude that TNF is essential for inducing the differentiation of naïve T cells into Th17 cells in the H1/CAF01 vaccination model.

TNF affects multiple target cell types and exerts diverse biological effects. It signals through TNFR1 and TNFR2, with TNFR1 being more crucial for its pro-inflammatory function (1). We aimed to determine which immune cell type requires TNFR1 signaling for the induction of Th17 cells. Considering TNFR1’s vital role in upregulating the PRR Mincle on macrophages (34), we hypothesized that TNF facilitates the recognition of the adjuvant by antigen-presenting and other innate immune cells. Alternatively, or in addition, an effect on B or T cells could explain the requirement for TNF in the H1/CAF01 vaccination model.

We utilized cell type-specific deletion of TNFR1 to dissect the mechanistic cellular targets. We initially measured TNFR1 surface expression on immune cells from wild type (WT) mice 7 days after immunization (experimental scheme in Figure 2A) and without immunization (Figure 2B; Supplementary Figure S3). We observed no significant TNFR1 expression on B cells. Consequently, we did not further pursue the role of TNFR1 signaling in B cells. TNFR1 was most highly expressed on neutrophils and classical monocytes, while Ly6C^low monocytes expressed the receptor at a lower level and only on a subset (Supplementary Figure S3A). NK cells and NK T cells expressed TNFR1 at a level comparable to Ly6C^low monocytes. cDC1 (CD88^-, CD26⁺, CD64^-, MHCII⁺, CD11c⁺, XCR1⁺) and cDC2 (CD88^-, CD26⁺, CD64^-, MHCII⁺, CD11c⁺, XCR1^-) as well as T cells (CD4 T cells, CD8 T cells and γδ T cells) showed a lower expression level. On macrophages (Ly6G^-, CD11b⁺, Ly6C^low, F4/80⁺) only a small subpopulation showed a detectable TNFR1 expression.

First, we checked the importance of TNF signaling via TNFR1 on myeloid cells. To this end, we crossed LysM-Cre mice (44) with mice in which exons 2–5 of the TNFR1 are floxed (Tnfrsf1a^tm3.3Gkl (41),). We immunized the LysM-Cre^hom; TNFR1^fl/fl mice with H1/CAF01 and determined the TNFR1 expression 7 days after immunization via flow cytometry (Figure 2A). LysM-Cre-mediated deletion of the TNFR1 resulted in the complete loss of TNFR1 expression on neutrophils and a substantial reduction on Ly6C^high inflammatory monocytes (Figure 2B). On Ly6C^low monocytes and on macrophages, TNFR1 median fluorescence intensity (MFI) was not reduced significantly by LysM-Cre mediated deletion. Closer examination of the flow cytometry data showed that TNFR1 was expressed on subpopulations of Ly6C^low monocytes and F4/80⁺ macrophages (Supplementary Figure S3A). LysM-Cre strongly reduced TNFR1 staining in these subpopulations, decreasing TNFR1^high cells from 34% to 6% for Ly6C^low monocytes and from 15% to 5% for macrophages (Figure 2C). For cDC1 the percentage of TNFR1^high cells was significantly reduced by approximately 50% (Figure 2C). In all other tested cell types, the MFI of the TNFR1 expression was not significantly changed, except for a slight increase on NK cells and a reduction on NK T cells. We will address the potential importance of TNFR1 on cDC1, NK and NK T cells later.

We first studied the effect of TNFR1 deficiency in myeloid cells on the local inflammatory response to immunization with H1/CAF01 (Figure 2A). CAF01, a liposome-based adjuvant, forms a depot at the site of injection (55, 56). Innate immune cells migrate to this location and slowly transport the adjuvant and antigen to the draining lymph node (57). Upon s.c. immunization in the footpad, the measurement of its thickness over time can serve as a proxy for immune cell infiltration and inflammation at the site of injection. The robust footpad swelling after H1/CAF01 injection was reduced by approximately 50% in myeloid TNFR1 knockout (KO) mice compared to the control group (Figure 2D), which corresponds well to the reduction observed in Tnf^-/- mice (34), indicating that TNF signaling in myeloid cells plays a significant role in the local inflammatory response.

The absence of TNFR1 in myeloid cells completely prevented the induction of H1-specific production of IL-17 by LN cells, while the production of IFNγ and IL-10 remained unaltered (Figure 2E; similar results observed for spleen cells see Supplementary Figure S4A). Of note, neither the floxing of TNFR1 alone (Supplementary Figure S5A) nor the insertion of the Cre recombinase under the LysM promoter (Supplementary Figure S5B) influenced the IL-17 production after immunization. The reduced IL-17 production was accompanied by the vaccine’s inability to induce H1-specific Th17 cells in the myeloid TNFR1 KO mice, whereas the induction of Th1 cells was unaffected (data from LN in Figure 2F, comparable results were observed for spleen Supplementary Figures S4B, C). In summary, the absence of TNFR1 on myeloid cells abrogated the Th17 response induced by the Mincle-activating adjuvant CAF01, replicating the effect of genetic whole body deletion of the Tnf gene we showed before (34).

We also tested the consequence of genetic deletion or pharmacologic blockade of TNF versus the specific knockout of TNFR1 in myeloid cells for the induction of the humoral immune response. To do so, we measured H1-specific antibody (Ab) titers after two injections of H1/CAF01 (Experimental layout Supplementary Figure S2A). Knockout of TNF reduced titers of all studied isotypes by more than one order of magnitude (Supplementary Figure S2B). This reduction was not surprising, as it is well established that the formation of germinal centers as well as the production of isotype switched antibodies, is impaired in Tnf^-/- and TNFR1^-/- mice (50, 58). Transient pharmacological TNF blockade of the response to vaccination by sequential treatment with Etanercept and a neutralizing anti-TNF antibody caused a weaker but significant reduction in H1-specific Ab titers of the proinflammatory isotypes IgG2c and IgG2b, but not of IgG1 (Supplementary Figure S2B). In contrast to the effect of genetic deletion or pharmacological inhibition of TNF, the myeloid cell-specific TNFR1 KO did not affect the induction of H1-specific Ab after immunization (Supplementary Figure S2C).

After we found that TNFR1 on myeloid cells is essential for the Th17 response, we wanted to test the importance of the TNFR1 on DCs. For this, we employed two well-characterized DC-targeting Cre-deleter mouse lines. First, a Cre knock-in at the Clec9a locus specifically deletes loxP-flanked genes in DC progenitors (45). In Clec9a-Cre^het; TNFR1^fl/fl mice, we observed a complete absence of TNFR1 staining on cDC1 and a significant, but incomplete reduction (ca. 65%) of TNFR1 surface protein on cDC2 (Figure 3A, representative histograms Supplementary Figure S6A). Besides DCs, there was a reduction of TNFR1 expression on macrophages and on NK cells, whilst the remaining tested cell types had intact TNFR1 expression. We will address the potential role of NK cells in the induction of antigen-specific Th17 cells later in this manuscript.

To achieve efficient deletion of TNFR1 in both cDC subsets, we used CD11c-Cre-mediated deletion (46) of TNFR1 in addition. In this model, TNFR1 staining was completely absent on cDC1 and cDC2 (Figure 3B; representative histograms Supplementary Figure S6B). In contrast to the Clec9a-Cre model, there was also a reduction in TNFR1 surface protein across all T cell subsets. Additionally, a significant reduction of TNFR1 was observed in Ly6C^low monocytes and in macrophages.

In both models for deleting TNFR1 on DC, the inflammatory response at the site of injection, measured by the footpad swelling, was unaffected (Figures 3C, D). Clec9a-mediated TNFR1 deletion did not alter cytokine production after the restimulation of LN cells from immunized mice (Figure 3E). These findings were confirmed by the unaltered induction of Ag-specific Th17 and Th1 cells after immunization (Figure 3F). This suggests that TNFR1 expression on cDC1 cells is not required for H1/CAF01-induced Th17 responses. Therefore, the complete loss of IL-17 production and Th17 cell induction LysM-Cre-mediated deletion (Figures 2E, F) cannot be caused by the reduction of TNFR1^high cDC1 observed in Figure 2C.

The CD11c-Cre-mediated complete TNFR1 deletion on all DC subsets did not affect the production of IFNγ and IL-10 in LN cells (Figure 3G) nor the induction of Th1 cells (Figure 3H). In contrast, it led to a significant reduction in IL-17 production upon restimulation of LN cells with H1, with 75-80% reduction when measured by ELISA (Figure 3G), or around 50% reduction in LN CD4+ T cells producing IL-17A, as measured by intracellular cytokine staining (Figure 3H). The same pattern was observed after restimulation of spleen cells and measurement by ELISA (Supplementary Figures S8A, C) and after intracellular cytokine staining of spleen cells (Supplementary Figures S8B, D). No significant inhibition of IL-17 production was observed in CD11c-Cre⁺; TNFR1^wt/wt mice (Supplementary Figure S5C). In summary, TNFR1 on cDC2 seems to play a role in inducing a Th17 response after H1/CAF01 immunization, but the Th17 response does not entirely depend on it, as observed for the TNFR1 on myeloid cells. Of note, CD11c-Cre-mediated deletion of TNFR1 is not entirely specific for DC but also leads to reduced expression on monocytes, macrophages and T cells.

We also measured the H1-specific Ab titer in H1/CAF01-immunized CLec9a-Cre^het; TNFR1^fl/fl and CD11c-Cre⁺; TNFR1^fl/fl mice. In both models, the specific Ab titers were similar to those in the control group (Supplementary Figure S2C). This implies that TNFR1 on DC is dispensable for the H1/CAF01-induced Ab response.

The impaired Th17 response in CD11c-Cre⁺; TNFR1^fl/fl mice was associated with a reduction in TNFR1 surface protein on T cells (Figure 3B), raising the question whether TNFR1 on T cells is required for Th17 differentiation in our immunization model. To address this, we used the Lck-Cre mouse (47) to delete the floxed TNFR1 during thymic development of T cells.

In this model, TNFR1 was no longer present on the surface of CD4⁺ and CD8⁺ T cells as well as on γδ T cells and NK T cells (Figure 4A; representative example in Supplementary Figure S7). Besides this, TNFR1 expression on all other studied cell types was unaffected. Thus, Lck-Cre-mediated deletion was efficient and specific in the T cell compartment.

Of note, a lower frequency of CD4⁺ T cells was observed in the LN of naïve and immunized Lck-Cre⁺; TNFR1^fl/fl mice (Figure 4B; comparable results were observed for CD4+ spleen cells Supplementary Figure S9B). The local inflammatory response in the footpad remained unchanged (Figure 4C). The absence of TNFR1 did not significantly affect IL-17 production following the restimulation of LN cells from immunized mice (Figure 4D; comparable to spleen cells Supplementary Figure S9A). However, a significant reduction in IFNγ and IL-10 production was noted in the supernatants, whereas the frequency of induced specific Th17 and Th1 CD4+ cells in the LN remained unchanged (Figure 4E; spleen comparable see Supplementary Figure S9B). In conclusion, the TNFR1 on CD4+ T cells may contribute to their overall numbers in spleen, but was not essential for the induction of specific Th17 cells by the H1/CAF01 vaccine.

As TNFR1 deletion by LysM-Cre caused a complete loss of Th17 induction, we sought to dissect which myeloid cell type (monocytes, neutrophils or macrophages) needs to respond to TNF for successful vaccination. We previously found, using antibody-mediated or genetic depletion models, that CCR2⁺ monocytes are critically required for IL-17 induction by CAF01 in vivo, whereas an important contribution of neutrophils was excluded (59). Therefore, we now focused on the response of monocytes isolated from mouse bone marrow cells to the vaccine adjuvant and its regulation by TNF. Stimulation with the Mincle ligand TDB (contained in CAF01) as well as with the mycobacterial cord factor TDM caused a transcriptional upregulation of Mincle that was strongly reduced by the TNF blocker Etanercept (Figure 5A). Both Mincle ligands TDB and TDM induced the expression of Il6 and Il1b in monocytes, again in a TNF-dependent manner (Figures 5B, C). IL-6 protein in the supernatant of monocyte cultures paralleled mRNA data (Figure 5D), whereas for IL-1β basal levels were unexpectedly high but still TNF-dependent after stimulation (Figure 5E). Thus, the CAF01 constituent TDB promotes the production of the Th17 polarizing cytokines IL-6 and Il-1β by monocytes, probably via TNF-dependent upregulation of Mincle.

Next, we aimed to assess whether the impaired upregulation of the TDB receptor Mincle in mice treated with Etanercept or lacking TNFR1 in myeloid cells is causal in the abrogation of Th17 induction after vaccination. To achieve this, we used a transgenic mouse model in which Mincle is constitutively expressed at a high level (Mincle^tg (48)). Additionally, to rule out confounding factors due to the regulation of endogenous Mincle by TNF signaling, the Mincle^tg mice were crossed onto a Mincle-deficient background (Clec4e^-/-, further referred to as Mincle^-/- (49)).

To determine Mincle regulation in these mice, bone marrow-derived macrophages (BMDM) were generated and stimulated with the Mincle ligand TDM (Figure 6A; representative histograms Supplementary Figure S10A). Mincle was expressed at low levels on resting WT BMDM and as expected strongly upregulated in a TNF-dependent manner after stimulation. In Mincle^-/-; Mincle^tg BMDM, Mincle was highly expressed under resting conditions, and, importantly, its expression level was unaffected by the TNF blocker Etanercept.

We immunized the Mincle^-/-; Mincle^tg mice to test the effect of TNF block when Mincle expression is no longer regulated (experimental design Figure 6B). The local inflammatory response, measured by footpad swelling, was reduced by the TNF block in WT mice (Figure 6C). The magnitude of this reduction was comparable to that seen in mice lacking TNFR1 on myeloid cells (Figure 2D). In the immunized transgenic mice, footpad swelling was greater compared to WT mice. TNF blockade with Etanercept reduced it, but only to the level seen in immunized non-TNF-blocked WT mice (Figure 6C).

IL-17 production after restimulation of LN cells from immunized mice was, as expected, entirely prevented in TNF-blocked WT mice (Figure 6D; comparable results for spleen cells, see Supplementary Figure S10B). In contrast, IL-17 production was restored in TNF-blocked transgenic mice to the level observed in immunized WT mice. The production of IFNγ and IL-10 was not altered by the TNF block. Correspondingly, the frequency of specific Th17 cells was reduced in immunized TNF-blocked WT mice, whereas in Mincle-transgenic mice, the TNF blocker did not affect the frequency of Th17 cells compared to immunized WT mice (Figure 6E; comparable results for spleen cells, see Supplementary Figures S10C, D).

Under physiological conditions, Mincle is expressed on myeloid cells and some dendritic cells (DC) (28). Some reports have found it under special circumstances on B cells (29, 30) and T cells (31). We found that NK and NK T cells from Mincle^-/-; Mincle^tg, but not WT, mice expressed significant Mincle protein on the surface, (Supplementary Figure S11A). Whether NK and NK T cells contributed to the observed rescue of Th17 induction, was tested by depleting NK1.1-positive cells using the antibody PK136 (Supplementary Figures S11B, C). Immunization of WT mice in the absence of NK1.1-positive NK and NK T cells still led to robust IL-17 induction (Supplementary Figure S11D), and its complete prevention by TNF blockade. This lack of a role for NK1.1-positive cells also allows us to exclude an impact of the observed differences in TNFR1 expression on NK and NK T cells after Clec9a-Cre- or CD11c-Cre-mediated deletion (Figures 3A, B). Importantly, the rescue of IL-17 production in TNF-blocked Mincle^-/-; Mincle^tg mice was unaffected by the absence of NK1.1-positive cells (Supplementary Figure S11D). Together, the Th17 rescue effect observed in the transgenic mice most likely originates from the constitutive expression of Mincle on either monocytes or DC, since NK1.1-positive cells were dispensable and transgenic Mincle was not stably expressed on T and B cells (Supplementary Figure S11A).