Cutting Edge: TNF Is Essential for Mycobacteria-Induced MINCLE Expression, Macrophage Activation, and Th17 Adjuvanticity

Signaling C-type lectin receptors (CLR) expressed in innate immune cells have emerged as a major class of pattern recognition receptors (1). Binding of bacterial and fungal ligands by CLR such as DECTIN-1, MINCLE, and DECTIN-2 triggers Syk-dependent activation of myeloid cells (2–5) and expression of cytokines and chemokines, which in turn direct adaptive immunity toward a Th17 response (6, 7). The DECTIN-2 family of CLR mouse chromosome 6 comprises DCAR (encoded by Clec4b1/2), DECTIN-2 (Clec4n), MCL (Clec4d), and MINCLE (Clec4e) (8). The mycobacterial cord factor trehalose-6,6-dimycolate and its synthetic analogue trehalose-6,6-dibehenate (TDB) bind to MINCLE and are potent Th17 adjuvants (2, 5, 9). Expression of MINCLE is regulated, with low levels in resting macrophages but strong induction by TLR and CLR ligands (10–12). The cytokine IL-4 downregulates MINCLE expression in mouse and human APC (13), whereas TNF was shown to upregulate its expression (10).

TNF is essential for mounting protective immune responses against intracellular pathogens, such as Mycobacterium tuberculosis (14). In contrast, TNF is causative in several chronic inflammatory diseases (e.g., rheumatoid arthritis and inflammatory bowel disease). Blockade of TNF with neutralizing Abs or sTNFR2-Fc protein is an established treatment in rheumatoid arthritis and inflammatory bowel disease (15). While alleviating inflammatory disease activity, TNF-targeting therapies increase the risk for developing active tuberculosis (16) and impair the generation of protective Ab responses following immunization (17). Because TNFR are expressed by different immune and nonimmune cell types, the cellular and molecular mechanism(s) responsible for these unwanted effects of TNF blockade are not fully understood.

In this study, we have investigated the role of TNF in the inducible expression of DECTIN-2 family CLR and its functional consequence for the macrophage response to mycobacteria and for the generation of Th17-type Ag-specific immunity by a MINCLE-dependent adjuvant.

C57BL/6 wildtype and Tnf^−/− mice on a C57BL/6 background (18) were bred under specific pathogen-free conditions at the Präklinische Experimentelle Tierzentrum of the Medical Faculty in Erlangen. C57BL/6N mice were purchased from Charles River Laboratories. Tnfrsf1a^tm1Mak (designated in this study Tnfr1 p55^−/−) mice (19) were bred at the Institute for Medical Microbiology and Hospital Hygiene (Heinrich Heine University, Düsseldorf, Germany). All mouse experiments were approved by the Regierung von Unterfranken (protocol number 55.2.2-2532-543).

Mice were immunized s.c. with 50 μl of a mixture of 1 μg H1, a fusion protein of the M. tuberculosis Ags Ag85B and ESAT-6, and the respective liposomal adjuvant (CAF01 or G3D6A-10) in the footpads of the hind legs. CAF01 is composed of TDB and cationic dimethyldioctadecylammonium liposomes and has been described in detail before (20). G3D6A-10 is a liposomal adjuvant formulation consisting of the synthetic TLR4 agonist 3-O-de-acyl-hexaacyl monophosphoryl lipid A [3D-6A-SMPLA, 3D(6-acyl)-PHAD] purchased in cGMP quality from Avanti Polar Lipids (Alabaster, AL) that is embedded in a matrix of 1,2-dimyristoyl-sn-glycero-3-phosphocholine, 1,2-dimyristoyl-sn-glycero-3-phosphoglycerol, and cholesterol (9:1:7.5 M ratio) in a PBS-buffered aqueous suspension. Footpad swelling was monitored regularly and measured prior to immunization as well as every second day postimmunization. On day 7 postimmunization, inguinal and popliteal lymph nodes were analyzed.

Mice were treated s.c. in the flank with 3 mg/kg of etanercept (human TNFR2-Fc) or rituximab (anti-human CD20 Ab) in a volume of 100 μl in PBS prior to immunization as well as 2, 4, and 6 d after immunization.

Bone marrow cells were differentiated to bone marrow–derived macrophages (BMM) for 6–7 d in complete DMEM containing 10% FCS, 50 μM 2-ME, and penicillin/streptomycin supplemented with 10% L929-cell conditioned medium as a source of M-CSF.

M. bovis (bacillus Calmette–Guérin [BCG]) was grown in Middlebrook 7H9 broth supplemented with 10% OADC-enrichment medium and 0.05% Tween 80 in small cell culture flasks constantly shaking at 125 rpm at 37°C to an OD₆₀₀ ∼1–2. Prior to in vitro stimulation, BCG were washed with PBS and diluted in complete DMEM.

BMM were stimulated with plate-coated TDB (5 μg/ml; Avanti Polar Lipids), using isopropanol as a mock control, as described (7), LPS (Escherichia coli serotype O55:B5, 10 ng/ml; Sigma), CpG oligodeoxynucleotide 1826 (0.5 μM; TIB MOLBIOL), TNF (10 ng/ml; R&D Systems), IL-6 (10 ng/ml; PeproTech), or BCG at the indicated multiplicity of infection. Etanercept and rituximab were used at a concentration of 100 μg/ml. Stimuli were added to 3 × 10⁵ cells in 48-well plates and cultured for 24 h at 37°C.

Inguinal and popliteal lymph nodes were collected and meshed through a 70-μm nylon filter to get a single cell suspension. Then, 5 × 10⁵ cells were stimulated in 96-well U-bottom plates with H1 (1 μg/ml), soluble anti-CD3 (0.5 μg/ml), or left untreated (mock) for 96 h.

RNA samples were subjected to RNA sequencing (RNAseq) and quantified using Qubit 2.0, according to the manufacturer’s instructions. Libraries were prepared and sequenced on an Illumina NovaSeq 6000 platform. Raw mapped reads were processed in R to determine differentially expressed genes and generate normalized read counts. For quantitative real-time PCR (qRT-PCR) analysis, total RNA was transcribed to cDNA (High-Capacity cDNA Reverse Transcription Kit, Applied Biosystems). Expression of the housekeeping gene Hprt and of the genes of interest was analyzed by qRT-PCR. All primers and probes were selected from the Roche Universal Probe Library. Threshold cycle (Ct) values of the target genes were normalized to Hprt, calibrated to unstimulated cells, and depicted as fold change.

Secreted cytokines were analyzed by ELISA (R&D Systems) from cell culture supernatants of stimulated BMM or lymph node cells.

Cell surface expression of MINCLE was analyzed by flow cytometry. Cells were blocked with anti-mouse CD16/32 (eBioscience) and stained with the primary Ab anti-MINCLE (1 μg/ml, clone 4A9; MBL), followed by anti-rat IgG1-APC (eBioscience). FACS data were acquired on a LSRFortessa (BD Biosciences) and analyzed using the software FlowJo (v10.6.1).

GraphPad Prism software (version 8) was used for statistical analysis. Statistical significance was calculated using the Mann–Whitney U test to compare two nonpaired groups: *p < 0.05, **p < 0.01, ***p < 0.001, and NS, p > 0.05.

We were interested in the mechanism of inducible expression of MINCLE, MCL, and DECTIN-2 triggered by microbes or their pathogen-associated molecular patterns (PAMPs) (12, 21, 22). The proinflammatory cytokine TNF was abundantly produced by BMM in response to the MINCLE ligand TDB, the TLR ligands CpG oligodeoxynucleotide and LPS, and BCG but not after stimulation with IL-6 (Fig. 1A). RNAseq data from BMM stimulated with TNF revealed that Mincle was strongly upregulated (Fig. 1B). We validated induction of Mincle by TNF by qRT-PCR (Fig. 1C). In addition, we found a moderate increase in Dectin-2 and Mcl, whereas Dectin-1 expression was not affected by treatment with TNF (Supplemental Fig. 1). Flow cytometry confirmed upregulation of MINCLE protein on the BMM surface (Fig. 1D). Thus, TNF production is elicited from macrophages by microbial PAMPs and is sufficient to increase DECTIN-2 family CLR expression.

We next asked whether TNF is required for the upregulation of Dectin-2 family CLR. The synthetic cord factor analogue TDB increased Mincle, Dectin-2, and Mcl mRNAs, confirming earlier results (12), which was completely abrogated in Tnf^−/− BMM (Fig. 2A, Supplemental Fig. 2A). In contrast, the induction of Mincle and Mcl by LPS was unaffected by TNF deletion and only partially reduced for Dectin-2. Expression of Dectin-1 was neither upregulated by TDB or LPS (Supplemental Fig. 2A). BCG was a particularly strong, dose-dependent inducer of Mincle and Dectin-2 expression, which was significantly decreased in Tnf^−/− BMM (Fig. 2B, Supplemental Fig. 2B). MINCLE cell surface protein was upregulated by TDB and, more strongly, by BCG in a TNF-dependent manner (Supplemental Fig. 2C). Although Mcl expression at a low multiplicity of infection was TNF dependent, it was not reduced in Tnf^−/− BMM at the higher BCG doses (Supplemental Fig. 2B). Tnf^−/− BMM failed to produce G-CSF after stimulation with the MINCLE ligand TDB, whereas LPS-induced G-CSF and IL-6 were not strongly affected (Fig. 2D). BCG-induced G-CSF and IL-6 were reduced in the absence of TNF (Fig. 2D).

TNF signals through two main receptors. Whereas TNFR2 signaling promotes tissue homeostasis and immune regulation, TNFR1 p55 predominantly drives proinflammatory responses (23). Therefore, we tested the requirement for TNFR1 p55. TDB- and BCG-induced upregulation of Mincle was strongly reduced or abrogated in TNFR1 p55-deficient BMM (Fig. 2E), whereas for Dectin-2 and Mcl, a significant reduction was observed after stimulation with TDB but not BCG (Supplemental Fig. 2D). Functionally, G-CSF as well as TNF production was absent after stimulation with TDB and significantly reduced after BCG infection (Fig. 2F). Thus, TNF mainly functions through its TNFR1 p55 to increase CLR expression and function.

To investigate whether TNF deficiency impairs MINCLE-dependent adjuvant activity in vivo, we s.c. immunized Tnf^−/− mice with the mycobacterial fusion protein H1 adsorbed to the TDB-containing liposomal adjuvant CAF01. Footpad swelling and induced mRNA expression of TNF, MINCLE, and G-CSF after vaccination was significantly reduced in Tnf^−/− mice (Fig. 3A, 3B), indicating a dampened local inflammatory response. Draining lymph node cellularity was high after vaccination in wildtype mice but reduced in the absence of TNF (Fig. 3C). Upon restimulation of draining lymph node cells with H1 in vitro, robust production of IL-17, IFN-γ, and IL-10 was observed as expected in C57BL/6 mice. Strikingly, Tnf^−/− mice failed to produce Ag-specific IL-17, whereas IFN-γ production was not significantly impaired and IL-10 levels were enhanced (Fig. 3D). Because IL-17 levels from Tnf^−/− draining lymph node cells after stimulation with anti-CD3 Ab were also reduced (Fig. 3D), we wondered whether Tnf^−/− T cells may have an intrinsic defect in producing IL-17. We therefore tested the TNF dependence of MINCLE-independent Th17-inducing adjuvants. The synthetic TLR4 agonist G3D6A-10 was identified by us to induce robust IL-17 production upon restimulation and therefore tested in comparison with CAF01. Footpad swelling was more transient for G3D6A-10 compared with CAF01 and independent of TNF (Supplemental Fig. 3A). Draining lymph node cell numbers were not reduced in Tnf^−/− mice immunized with G3D6A-10 (Supplemental Fig. 3B). Analysis of the T cell response in these mice showed that G3D6A-10 robustly induced IL-17 production in wildtype mice, which was not significantly reduced in Tnf^−/− mice (Fig. 3E). Moreover, IFN-γ and IL-10 levels in wildtype and Tnf^−/− mice immunized with G3D6A-10 were comparable (Fig. 3E). We conclude that TNF deficiency specifically abrogates the Th17-inducing capacity of the MINCLE-dependent adjuvant CAF01, correlating with a lack of upregulation of Mincle expression in vivo, but spares the IL-17–promoting activity of the TLR4-dependent adjuvant G3D6A-10.

The clinically used TNFR2-Fc fusion protein etanercept binds and neutralizes murine TNF (24), which enabled us to test whether pharmacological blockade of TNF can have similar effects as genetic deletion. Etanercept, but not the control biological rituximab (anti-CD20 Ab, sharing the IgG1 isotype of etanercept), caused complete inhibition of TDB- and TNF-induced upregulation of Mincle in BMM in vitro and a partial reduction after BCG (Fig. 4A). Treatment of C57BL/6 mice with etanercept during immunization with CAF01/H1 reduced footpad swelling by 50% (Fig. 4B), attenuated Mincle expression at the injection site (Supplemental Fig. 4A), and significantly reduced draining lymph node cellularity (Supplemental Fig. 4B). Importantly, it strongly reduced IL-17 production upon restimulation with H1, whereas IFN-γ was not altered, and IL-10 production was enhanced (Fig. 4C). We excluded nonspecific effects of the Fc fusion protein by using rituximab as a control treatment in the immunization model (Supplemental Fig. 4C, 4D). Together, TNF blockade with etanercept closely phenocopied the effects of genetic TNF deficiency on induced Mincle expression and CAF01 adjuvant activity in vivo.

We show in this study a pivotal role of TNF signaling through TNFR1 p55 for transcriptional upregulation specifically of several DECTIN-2 family CLR in macrophages activated with the cord factor analogue TDB or with whole M. bovis BCG. TNF was also essential for the production of cytokines in response to mycobacteria and for Th17 biasing of adaptive immunity driven by a MINCLE-dependent adjuvant. These findings have implications for a potential involvement of regulated CLR in the TNF-dependent host response to infections and during vaccination. As patients receiving TNF blockers are at risk to develop infectious complications, downregulation of CLR expression may be an underlying mechanism of impaired responsiveness to latent or invading pathogens.

Intriguingly, the requirement for TNF was specific for the mycobacterial cord factor analogue TDB and for BCG because CLR expression induced by the TLR4 ligand LPS was conserved in Tnf^−/− BMM. This raises the question whether other secreted factors (e.g., cytokines) induced by LPS but not TDB may act redundantly with TNF. One cytokine produced at much higher levels in response to LPS than to TDB is IL-6, which indeed was found by Matsumoto et al. (10) to induce Mincle mRNA similar to TNF. However, in our hands, rIL-6 did not cause increased Mincle expression on its own. Another cytokine stimulus shown recently to promote expression of Mincle is GM-CSF (25). The differences in TNF dependence between Mincle, Dectin-2, and Mcl and the lack of inducibility of Dectin-1 point to a differential transcriptional regulation of these CLR that needs to be dissected in future studies.

The mycobacterial cell wall contains ligands for all CLR family members that require TNF for full expression (trehalose-6,6-dimycolate for MINCLE and MCL, lipoarabinomannan for DECTIN-2), and these likely act in concert to enhance mutually their expression. Triggering of other pattern recognition receptors (e.g., TLR2/4, TLR9, DECTIN-1, etc.) may also promote DECTIN-2 family CLR expression by mycobacterial PAMPs (e.g., 19 kDa lipopeptide and lipoarabinomannan) but is apparently not sufficient to fully overcome TNF dependence.

DECTIN-2 family CLR contribute to antimycobacterial protection; Mincle- and Mcl-deficient mice developed increased bacterial burden and mortality after M. tuberculosis challenge (26, 27), and Dectin-2–deficient mice showed increased lung pathology after mycobacterial infection (4). In humans, single-nucleotide polymorphisms in CLEC4D (MCL) and in CLEC4E (MINCLE) were associated with altered susceptibility to pulmonary tuberculosis (27, 28). Thus, it will be important to determine whether their impaired expression contributes to the severely increased susceptibility of mice deficient in the TNF–TNFR1 p55 pathway and in humans treated with biologicals blocking TNF.

To date, the cellular and molecular mechanisms of impaired immunological control of tuberculosis in conditions of TNF deficiency are not well understood. In mice deficient for TNF or the TNFR1 p55, there is a failure to form organized granulomas postinfection with M. tuberculosis (14), and administration of neutralizing Abs to TNF to mice during chronic infection lead to dissolution of granuloma structures (29). Although these results suggest that TNF is required for the organized interaction of different immune cells contributing to granuloma formation, there is also evidence for a macrophage-autonomous effect of TNF blockers on control of mycobacterial replication through the inhibition of phagosomal maturation (30). It is conceivable that impaired expression of several TNF-controlled CLR during acute or latent infection with M. tuberculosis may decrease sensing of mycobacteria, resulting in the diminished expression of cytokines and chemokines, defective granuloma formation, and alterations in phagosomal maturation.

Technical support by Barbara Bodendorfer and animal husbandry by Manfred Kirsch is gratefully acknowledged. We thank Dr. Gerhard Krönke for providing etanercept and rituximab and Dr. Daniel Degrandi for help with TNFR knockout mice.

The authors have no financial conflicts of interest.