Abstract
Trehalose-6,6-dimycolate (TDM), the mycobacterial cord factor, is an abundant cell wall glycolipid and major virulence factor of Mycobacterium tuberculosis. Its synthetic analog trehalose-6,6-dibehenate (TDB) is a new adjuvant currently in phase I clinical trials. In rodents, the C-type lectin receptors Mincle and Mcl bind TDB/TDM and activate macrophages and dendritic cells (DC) through the Syk–Card9 pathway. However, it is unknown whether these glycolipids activate human innate immune cells through the same mechanism. We performed in vitro analysis of TDB/TDM-stimulated primary human monocytes, macrophages, and DC; determined C-type lectin receptor expression; and tested the contribution of SYK, MINCLE, and MCL by small interfering RNA knockdown and genetic complementation. We observed a robust chemokine and cytokine release in response to TDB or TDM. MCSF-driven macrophages secreted higher levels of IL-8, IL-6, CCL3, CCL4, and CCL2 after stimulation with TDM, whereas DC responded more strongly to TDB and GM-CSF–driven macrophages were equally responsive to TDB and TDM. SYK kinase and the adaptor protein CARD9 were essential for glycolipid-induced IL-8 production. mRNA expression of MINCLE and MCL was high in monocytes and macrophages, with MINCLE and MCL proteins localized intracellularly under resting conditions. Small interfering RNA–mediated MINCLE or MCL knockdown caused on average reduced TDB- or TDM-induced IL-8 production. Conversely, retroviral expression in murine Mincle-deficient DC revealed that human MINCLE, but not MCL, was sufficient to confer responsiveness to TDB/TDM. Our study demonstrates that SYK–CARD9 signaling plays a key role in TDB/TDM-induced activation of innate immune cells in man as in mouse, likely by engagement of MINCLE.
Introduction
Trehalose-6,6-dimycolate (TDM) is an abundant glycolipid in the cell wall of pathogenic mycobacteria such as Mycobacterium tuberculosis. TDM has been known as the mycobacterial cord factor since the 1950s (1, 2) because its hydrophobicity contributes to the cording phenomenon (3). TDM alone elicits several key aspects of the host response to mycobacterial infection, most importantly the formation of granulomas when injected in vivo (4; reviewed in Ref. 5) and the inflammatory response of macrophages in vitro (6). In contrast, the cord factor also appears to contribute to mycobacterial immune evasion [e.g., by delaying phagosomal maturation (7)]. Thus, the cord factor is of dual importance in the host–pathogen interaction in tuberculosis [i.e., both as a immunostimulatory mycobacterial component and as a glycolipid virulence factor (reviewed in Ref. 8)].
Innate immune activation induced by killed mycobacteria is the basis for the strong Th1/Th17 induction by CFA in animal models (9, 10). The adjuvanticity of CFA can at least in part be assigned to the action of the cord factor (11). There is a lack of adjuvants inducing robust Th1 or Th17 immune responses to protein Ag in humans because CFA is too toxic for application in humans. The synthetic cord factor analog, trehalose-6,6-dibehenate (TDB), has been developed together with a liposomal carrier as the so-called CAF01 adjuvant system (12–15). Although TDM and TDB share the same trehalose disaccharide head group, the two ester-linked lipid tails differ, with mycolic acids of 60–90 carbons (C) for TDM and 22C behenic acids in case of TDB. Following robust Th1/Th17 adjuvanticity in preclinical mouse, guinea pig and nonhuman primate studies, the TDB-containing CAF01 adjuvant has entered phase I clinical studies (NCT00922363, NCT01141205, and NCT01009762) (16–18) as adjuvant for protein Ag vaccines against M. tuberculosis and HIV (19, 20). Indeed, immunization with the H1 fusion protein of the TB Ags Ag85B and ESAT-6 promoted long-lived T cell responses in the vaccines (17).
The receptors and pathways of innate immune cells activated by the cord factor and by TDB have recently been studied in the mouse system. Both glycolipids are bound by macrophage-inducible C-type lectin (Mincle), a C-type lectin receptor (CLR) encoded by the Clec4e gene in the Dectin-2 cluster (reviewed in Ref. 21). In this paper, we followed the nomenclature of the National Center for Biotechnology Information Gene database such as “Mincle” for murine Clec4e and capitalized “MINCLE” for the human gene and protein. Deletion of Mincle in mice completely abrogated macrophage responses to TDB/TDM in vitro, Th17 adjuvanticity of TDB and granuloma formation to TDM in vivo (22, 23). The macrophage C-type lectin (Mcl, Clecsf8), encoded by the Clec4d gene adjacent to Clec4e, has recently been identified as a second receptor for the cord factor and is involved in protection against tuberculosis in mice (24, 25). Mcl can also form a heterodimer with Mincle (26, 27) or Dectin-2 (Clec4n, Clec6a) (28). In contrast to the prototypic CLR Dectin-1 (Clec7a), Mincle and Mcl do not contain an intracellular ITAM for recruitment of the spleen tyrosine kinase (Syk). Instead, Mincle and most likely also Mcl associate with the adaptor protein Fc Rγ-chain (Fcer1g, in this paper referred to as FcRγ) (29, 30), which is required for initiation of signaling by binding Syk through its Src homology 2 domain (reviewed in Ref. 31). The Card9-Bcl10-Malt1 complex is then essential for activation of NF-κB and induction of gene expression in macrophages and dendritic cells (DC) (32, 33). Activation of TLR–Myd88 signaling by TDM has been described with a contribution of the scavenger receptor MARCO (34, 35). TLR–Myd88-dependent signals can facilitate the response to the cord factor via upregulation of Mincle expression (36, 37).
Although these studies in mice have advanced our understanding of TDB/TDM recognition and responses, there is a surprising lack of data with regard to the effects of these glycolipids on human innate immune cells. The question of how human macrophages and DC interact with the cord factor during mycobacterial infection remains largely unanswered. Likewise, little is known about the response of human innate immune cells, despite the entry of the synthetic TDB as adjuvant component of CAF01 into clinical studies. In this study, we investigated the cytokine and chemokine response of primary human monocytes, macrophages and DC to TDB and TDM. In a comprehensive analysis of cells from multiple healthy donors, we found robust induction of several chemokines and cytokines by both glycolipids, but also some stimulus- and cell type–specific differences. Using pharmacological inhibitors, genetic complementation and small interfering RNA (siRNA) knockdown, we provide evidence that the human MINCLE receptor is able to mediate the response to TDB and TDM, dependent on SYK and CARD9.
Materials and Methods
Human cell isolation and culture
The use of human leukocytes from healthy donors with written informed consent complies with the Declaration of Helsinki (Ethical Committee Erlangen approval numbers 4055 and 111_12 B). PBMC were obtained from 100 ml whole blood or from leukoreduction system chambers (38, 39) by density centrifugation. Monocytes were positively selected from PBMC using anti-CD14 microbeads (Miltenyi Biotec), and purity was ≥ 90%. For culture, RPMI 1640 medium was supplemented with 10% (v/v) FCS (Biochrom) and penicillin/streptomycin. A total of 50 U/ml GM-CSF (Genzyme) or 50 U/ml M-CSF (Peprotech) were added for differentiation of macrophages (40). 50 U/ml GMCSF and 250 U/ml IL-4 (PeproTech) were added for differentiation of DC (38, 41). Cells were cultured at a density of 0.8 × 106 cells (GM-CSF macrophages and DC) or 1.6 × 106 cells (M-CSF macrophages) for 6–7 d without change of media at 37°C with 5% CO2/95% humidified air. After harvesting, viability was controlled by trypan blue staining. Generally, the fraction of dead cells was <10% and never exceeded 20%. Macrophages and immature DC were used for stimulation with glycolipids without further maturation steps.
Cell stimulation and use of inhibitors
Human cells were stimulated for cytokine analysis at a density of 4 × 106 live cells/ml for 24 h (monocytes) or 1.5 × 106 live cells/ml for 48 h (macrophages and DC).
TDB (Avanti Polar Lipids) or TDM (Bioclot) were used plate-bound at 5 μg/ml as previously described (33), whereas coating with isopropanol only was carried out as negative control. As a positive control, 10 ng/ml LPS or 0.5 μM CpG ODN1826 (for stimulation of murine DC) were used. A total of 100 U/ml IFNG (PeproTech) were added for costimulation of human APC.
siRNA knockdown
Cells were transfected with siRNA on day 4 using the Stemfect RNA transfection kit (Stemgent), knockdown efficiency was analyzed after 72 h. siRNA targeted to SYK (number L-003176-00), CARD9 (number L-004400-00), MINCLE (number L-021374-01), and MCL (number L-021373-00) were purchased from Dharmacon onTARGETplus smart pool (Thermo Scientific).
Cytokine and chemokine analysis
Secretion of human IL-6, IL-8 (CXCL8), IL-12 (IL12p70), IL-12B (IL12p40), and TNF was analyzed by sandwich ELISA (BioLegend). Secretion of G-CSF (CSF3), IL-1A, IL-1B, CXCL9 (MIG), CXCL10 (IP10), CCL2 (MCP1), CCL3 (MIP1A), and CCL4 (MIP1B) were measured using a cytokine bead array kit (eBioscience) and analyzed using FlowCytomixPro software.
Retroviral transduction
DC were generated from bone marrow cells of Clec4e−/− or C57BL/6 wild-type control mice in DMEM supplemented with 10% (v/v) FCS (Biochrom), penicillin/streptomycin, 2-ME, and 10% (v/v) GM-CSF (Csf2) containing supernatant of X63 cell line (cDMEM). On day 2, cells were infected with supernatants of Phoenix cells (43). Phoenix cells were transfected by Ca3(PO4)2 precipitation with the packaging vector pCL-Eco and the MigR1 overexpression construct coding for GFP alone (empty vector) or for bicistronic expression of murine Mincle (mMincle), human MINCLE (hMINCLE), or human MCL (hMCL) to produce virus-like particles (44, 45). Identical total amounts of supernatant were used for retroviral cotransduction of CLR, and the supernatants were mixed 1:1 as indicated. Transduction efficiency reached 60–90% on day 6. Cells were stimulated for 24 h, to harvest RNA. Cells were incubated in presence of brefeldin A (10 ng/ml) for 20 h for intracellular cytokine staining.
Transfection of HEK293T
HEK293T were transfected by CaPO4 precipitation with pcDNA3.1-based vectors encoding mFcRγ, hMINCLE, hMINCLE-HA, hMCL, human DECTIN2, or human DCIR together with a GFP-coding plasmid for gating on GFP-positive cells for analysis.
Flow cytometry
Flow cytometric analysis was carried out on a FACSCanto II (BD Biosciences). Staining was done in PBS/2% FCS in presence of FcR blocking reagent (Miltenyi Biotec) for 30 min (4°C) for surface staining and 2 h (4°C) with 0.5% saponin for intracellular staining after fixation with 1% PFA/PBS for 20 min. Secondary Ab staining was done for 15 min (4°C) in the respective buffer. Intracellular staining in the presence of saponin gave comparable results to a commercial Foxp3 staining kit (eBioscience). Abs anti-CD11b, CD11c, CD14, CD86, and CD83 were purchased from eBioscience, anti-CD80 and MHC class II were purchased from BioLegend, anti-MCL was purchased from R&D Systems, and anti-HA Ab was purchased from Miltenyi Biotec. Anti-MINCLE Ab clone 13D10H11 was a gift of Dr. S. Yamasaki (Kyushu University, Fukuoka, Japan). Anti-mouse Tnf Ab for intracellular cytokine staining was purchased from BD Biosciences. Analysis was carried out with FlowJo software (GraphPad Software).
Quantitative RT-PCR
RNA was isolated with Trifast (Peqlab), cDNA was transcribed using a cDNA synthesis kit (Applied Biosystems). Expression levels of the housekeeping genes hypoxanthin-guanin-phosphoribosyltransferase (Hprt, mouse) or cyclophilin A (PPIA, human) as of the genes of interest were analyzed using primer/probe combinations selected from the Roche Universal Probe Library (Roche). For MINCLE, the primer/probe combination was selected using the software PrimerExpress (Applied Biosystems). All primers and probes used (listed in Supplemental Table I) were purchased from Metabion. ΔCT values were calculated as ΔCT = CT(housekeeping gene) − CT(gene of interest) (such that higher values indicate higher relative expression); ΔΔCT values referred to the calibrator as indicated, and fold change was calculated as 2ΔΔCT.
Western blot analysis
Cellular lysates were prepared in radioimmunoprecipitation buffer containing proteinase and phosphatase inhibitors (Roche complete, 0.5 M sodium fluoride, 1 M β-glycerophosphate, and 200 mM sodium orthovanadate). Western blot was performed by 10% SDS-PAGE and wet blotting. Abs used were anti-SYK (Cell Signaling Technology), anti-CARD9 (Santa Cruz Biotechnology), anti-GRB2 (BD Biosciences) as loading control and HRP-conjugated secondary Abs (Jackson ImmunoResearch Laboratories).
Binding assay
TDB, TDM, and behenic acid were coated onto 96-well cell culture plates for a final concentration of 5–50 μg/ml as indicated. After washing and blocking with 3% BSA/HBSS, supernatant of hMINCLE-Fc–transfected Chinese hamster ovary cells was added. hMINCLE-Fc was generated according to previously published procedures (46). Following incubation at 4°C overnight, plates were washed, bound protein detected by incubation with peroxidase-conjugated anti-hFc Ab, washed again, and then developed with ELISA substrate. Absorbance was measured at 450 nm.
Statistical analysis
Statistical analysis was carried out using GNU R and Prism5 (GraphPad Software). Gaussian distribution was controlled by D’Agostino Pearson and Shapiro–Wilk test. A Wilcoxon signed-rank test was carried out for paired testing, a Mann–Whitney U test was applied for non-paired testing. *p < 0.05, **p < 0.01, and ***p < 0.001.
Results
To study the effects of the synthetic glycolipid adjuvant TDB and the mycobacterial cord factor TDM on different types of human APCs, we used primary monocytes and monocyte-derived macrophages and DC. CD14+ monocytes were positively selected from PBMC using MACS beads. Macrophages and DC were differentiated from CD14+ monocytes in vitro for 6–7 d (Fig. 1A, Supplemental Fig. 1B). The donor group used for stimulation experiments shown in Figs. 1 and 2 comprised 32 different donors (Supplemental Fig. 1A). Cells of 20 of these donors were characterized by flow cytometry prior to and after differentiation for expression of costimulatory molecules. As is typical for resting macrophages, there was little CD80 and CD83 but intermediate CD86 and MHCII expression, whereas immature DC generally expressed low levels of costimulatory molecules (Fig. 1B). As expected, maturation of DC by LPS stimulation strongly increased costimulatory molecule expression (Supplemental Fig. 1D) and potently induced T cell proliferation in a MLR (Supplemental Fig. 1E). GM-CSF–derived and M-CSF–derived macrophages (further termed GM-CSF or MCSF macrophages, respectively) exhibited a distinct morphology and differential mRNA expression of the macrophage polarization marker CD163 (Supplemental Fig. 1B, 1C) (40).
Differentiation and characterization of human macrophages and DC. (A) Overview of experimental setup for generation and stimulation of human macrophages and DC. Monocytes were differentiated for 6 d. GM: GM-CSF; M: M-CSF; GM+4: GM-CSF plus IL-4. (B) Phenotype (forward light scatter-side scatter [of light] [FSC-SSC], gating on singlets) and expression of costimulatory molecules MHC class II (MHCII), CD80, CD83, and CD83 on day 0 monocytes and day 6 macrophages and DC. Representative donor (top) and average expression (bottom), n = 20, mean ± SD.
Differentiation and characterization of human macrophages and DC. (A) Overview of experimental setup for generation and stimulation of human macrophages and DC. Monocytes were differentiated for 6 d. GM: GM-CSF; M: M-CSF; GM+4: GM-CSF plus IL-4. (B) Phenotype (forward light scatter-side scatter [of light] [FSC-SSC], gating on singlets) and expression of costimulatory molecules MHC class II (MHCII), CD80, CD83, and CD83 on day 0 monocytes and day 6 macrophages and DC. Representative donor (top) and average expression (bottom), n = 20, mean ± SD.
Stimulation of human monocytes, macrophages, and DC with TDB/TDM. (A) Secretion of IL-6 (n = 17/32/30/32) and IL-8 (n = 17/31/28/27) shown as scatter plots. Stimulation of CD14+ monocytes (24 h), GM and M macrophages, and GM+4 DC (48 h) with 5 μg/ml plate-bound TDB or isopropanol control (without). Wilcoxon signed-rank test. (B) Secretion of IL-6 (n = 6/5/11) after 48-h stimulation of GM and M macrophages and GM+4 DC with 5 μg/ml TDB or control and costimulation with/without 100 U/ml IFNG as indicated. Wilcoxon signed-rank test. (C) Secretion of IL-8 after stimulation with control or 5, 1.25, and 0.156 μg/ml TDB or TDM as indicated of GM (n = 7) or M (n = 6) macrophages or GM+4 DC (n = 5); dots represent median. (D) Box plots (min-max) comparing secretion of IL-6 (n = 29/26/28), IL-8 (n = 27/20/28), TNF (n = 15/15/17), and IL-12B (n = 10/11/11) following stimulation as with plate-bound TDB or TDM (5 μg/ml each) or control. Wilcoxon signed-rank test.
Stimulation of human monocytes, macrophages, and DC with TDB/TDM. (A) Secretion of IL-6 (n = 17/32/30/32) and IL-8 (n = 17/31/28/27) shown as scatter plots. Stimulation of CD14+ monocytes (24 h), GM and M macrophages, and GM+4 DC (48 h) with 5 μg/ml plate-bound TDB or isopropanol control (without). Wilcoxon signed-rank test. (B) Secretion of IL-6 (n = 6/5/11) after 48-h stimulation of GM and M macrophages and GM+4 DC with 5 μg/ml TDB or control and costimulation with/without 100 U/ml IFNG as indicated. Wilcoxon signed-rank test. (C) Secretion of IL-8 after stimulation with control or 5, 1.25, and 0.156 μg/ml TDB or TDM as indicated of GM (n = 7) or M (n = 6) macrophages or GM+4 DC (n = 5); dots represent median. (D) Box plots (min-max) comparing secretion of IL-6 (n = 29/26/28), IL-8 (n = 27/20/28), TNF (n = 15/15/17), and IL-12B (n = 10/11/11) following stimulation as with plate-bound TDB or TDM (5 μg/ml each) or control. Wilcoxon signed-rank test.
To investigate the inflammatory response to TDB stimulation, cell culture supernatants were analyzed for chemokine and cytokine secretion. Supernatants from monocytes were analyzed after 24 h of stimulation when the cells still exhibited a monocyte rather than a macrophage-like phenotype. Supernatants from macrophages and DC were analyzed after 48-h stimulation. In all cell types, the secretion of IL-8 (CXCL8) was significantly increased following TDB stimulation compared with unstimulated controls. GM-CSF–derived macrophages and DC but not monocytes and M-CSF–derived macrophages also secreted significantly increased amounts of the cytokine IL-6 (Fig. 2A). IL-8 and IL-6 production correlated positively with each other (Supplemental Fig. 2; data not shown). Cotreatment of macrophages and DC with IFNG as additional proinflammatory stimulus caused increased overall cytokine amounts but did not lead to more pronounced differences between TDB-stimulated and unstimulated samples (Fig. 2B). We did not find a direct correlation of the height of the cytokine response and clinical parameters (e.g., leukocyte counts or basal expression of costimulatory molecules) (data not shown).
To compare TDB with TDM stimulation, we first performed dose-response measurements with both glycolipids in test experiments and observed the strongest IL-8 response at concentrations of 1.25 and 5 μg/ml (Fig. 2C). Therefore, in all following experiments, cells were stimulated with 5 μg/ml of the respective glycolipid in parallel. Although the response of GM-CSF macrophages to TDB and TDM was mostly comparable, M-CSF macrophages produced significantly more IL-6 in response to TDM than to TDB (Fig. 2D). In contrast, DC differentiated with GM-CSF + IL-4 showed a tendency toward stronger response to TDB stimulation. The same trend was observed analyzing IL-8 and TNF secretion. Significant induction of IL-12B (IL12p40) by TDM stimulation was only measured in DC (Fig. 2D). We could not detect significant secretion of IL12p70 in TDB and TDM treated APC (data not shown).
We next wanted to compare the effects of TDB and TDM stimulation for additional chemokines and cytokines. A cytokine bead array was used to measure IL-1A, IL-1B, and several chemokines (Fig. 3). Monocytes produced CCL2 (MCP1), CCL3 (MIP1A), CCL4 (MIP1B), and IL-1B in response to TDB and TDM (Fig. 3A). GM-CSF macrophages responded with a strong release of IL-1B, CCL3, and CCL4 but little IL-1A production in some donors (Fig. 3B). M-CSF macrophages produced high amounts of CCL4, CCL3, IL-1B, IL-1A, and G-CSF (CSF3) in at least half of the donors, with higher levels induced by TDM than by TDB (Fig. 3C). Regarding this observation of responding and nonresponding donors, we wondered whether low IL-1A and IL-1B secretion would correlate with a principally high or low chemokine and cytokine production. Indeed, this was the case (Supplemental Fig. 2). DC produced CCL3, CCL4, and some CCL2; the response to TDB here was found to be higher than to TDM (Fig. 3D). These findings confirmed the ability of human APC to respond to TDB as well as TDM stimulation, whereas the pattern of response was dependent on the stimulus and cell type.
Characterization of cytokine and chemokine response to TDB/TDM stimulation. (A–D) Production of G-CSF (CSF3), IL-1A, IL-1B, CXCL9 (MIG), CXCL10 (IP-10), CCL2 (MCP1), CCL3 (MIP1A), and CCL4 (MIP1B) of CD14+ monocytes (24 h, n = 15) (A), GM (B), M (C) macrophages, and GM+4 DC (48 h, n = 17) (D) following stimulation with 5 μg/ml TDB, TDM, or control (without). Cytokine-bead array, medians are depicted as polar plots (top); scatter plots show individual responses for those cytokines, which were induced significantly by TDB/TDM (bottom). Wilcoxon signed-rank test.
Characterization of cytokine and chemokine response to TDB/TDM stimulation. (A–D) Production of G-CSF (CSF3), IL-1A, IL-1B, CXCL9 (MIG), CXCL10 (IP-10), CCL2 (MCP1), CCL3 (MIP1A), and CCL4 (MIP1B) of CD14+ monocytes (24 h, n = 15) (A), GM (B), M (C) macrophages, and GM+4 DC (48 h, n = 17) (D) following stimulation with 5 μg/ml TDB, TDM, or control (without). Cytokine-bead array, medians are depicted as polar plots (top); scatter plots show individual responses for those cytokines, which were induced significantly by TDB/TDM (bottom). Wilcoxon signed-rank test.
In rodents, the Syk/Card9 pathway is activated by binding of TDB and TDM to the CLR Mincle and Mcl (23, 24, 29). We therefore asked whether the response in human APC is also dependent on the kinase SYK. We used the production of the neutrophil chemotactic IL-8 as readout as TDB and TDM stimulation induced secretion of high amounts in most donors and all cell types. IL-8 acts also as chemoattractant for monocytes and T cells, is found in the tuberculosis-infected lung, and contributes to granulocyte killing of mycobacteria (47–49). The pharmacological SYK inhibitor R406 led to significantly reduced IL-8 secretion following TDB and TDM stimulation, whereas the response to LPS remained largely intact (Fig. 4).
Role of SYK kinase activity for TDB/TDM response. (A–D) Box plots (min-max) showing IL-8 response following stimulation of CD14+ monocytes (24 h, n = 7) (A), GM (48 h, n = 7) (B), M (n = 6) (C), and macrophages and GM+4 DC (n = 7) (D) with 5 μg/ml TDB, TDM, or control (without) without/with presence of SYK-Inhibitor R406 (1 μM) as indicated. Wilcoxon signed-rank test.
Role of SYK kinase activity for TDB/TDM response. (A–D) Box plots (min-max) showing IL-8 response following stimulation of CD14+ monocytes (24 h, n = 7) (A), GM (48 h, n = 7) (B), M (n = 6) (C), and macrophages and GM+4 DC (n = 7) (D) with 5 μg/ml TDB, TDM, or control (without) without/with presence of SYK-Inhibitor R406 (1 μM) as indicated. Wilcoxon signed-rank test.
Therefore, we analyzed the mRNA expression of SYK-dependent CLR in human APC. In monocytes, MINCLE, MCL, DECTIN2, and DECTIN1 were the most highly expressed CLR (Fig. 5A). The levels of MINCLE, MCL, and DECTIN2 were reduced during differentiation of DC and less so of macrophages, whereas DECTIN1 expression remained unaffected (Fig. 5B). By flow cytometry, we could not detect MINCLE protein and only low levels of MCL on the surface of resting monocytes, macrophages, and DC. Intracellular staining in presence of the detergent saponin revealed the expression of both receptors in monocytes (Fig. 5C, Supplemental Fig. 3) and at varying levels in differentiated macrophages and DC (Fig. 5C). We confirmed that cotransfection of murine FcRγ and human MINCLE in HEK293T cells was sufficient to induce surface expression of MINCLE, as it has been previously described (26, 29), and that FcRγ is expressed in all types of APC (Supplemental Fig. 3C; data not shown). In contrast to mice, where Mincle is an inducible receptor hardly expressed in resting bone marrow–derived macrophages (24, 37), human MINCLE and MCL were expressed at similarly high levels in resting cells (Fig. 5B). Although we observed intermediate induction of MINCLE and MCL mRNA upon LPS stimulation, glycolipid stimulation was not sufficient to induce upregulation (Fig. 5D).
Expression of SYK-coupled C-type lectins. (A) mRNA levels of SYK-coupled CLRs in PBMC and CD14+ monocytes (n = 5–8), determined by quantitative RT-PCR (qRT-PCR_: ΔCT = CT(PPIA)-CT(CLR). (B) mRNA levels (ΔCT) of MINCLE (CLEC4E, n = 40/38/33/34), MCL (CLEC4D, n = 32/31/25/25), DECTIN2 (CLEC4N/CLEC6A, n = 22/21/16/17), DECTIN1 (CLEC7A, n = 24/23/18/19) in monocytes, macrophages, and DC. (C) MINCLE and MCL expression on CD14+ monocytes determined by surface and intracellular FACS staining, positive cells relative to isotype control. Representative donor (left). Surface and intracellular MINCLE and MCL expression on monocytes (n = 14/12), macrophages, and DC (n = 8/6). Percentage of positive cells relative to isotype control (right). (D) Inducibility of MINCLE (left, mean ± SD, n = 7–8) and MCL (right, mean ± SD, n = 5–8) mRNA following 24-h stimulation with 5 μg/ml TDB, TDM, or 10 ng/ml LPS.
Expression of SYK-coupled C-type lectins. (A) mRNA levels of SYK-coupled CLRs in PBMC and CD14+ monocytes (n = 5–8), determined by quantitative RT-PCR (qRT-PCR_: ΔCT = CT(PPIA)-CT(CLR). (B) mRNA levels (ΔCT) of MINCLE (CLEC4E, n = 40/38/33/34), MCL (CLEC4D, n = 32/31/25/25), DECTIN2 (CLEC4N/CLEC6A, n = 22/21/16/17), DECTIN1 (CLEC7A, n = 24/23/18/19) in monocytes, macrophages, and DC. (C) MINCLE and MCL expression on CD14+ monocytes determined by surface and intracellular FACS staining, positive cells relative to isotype control. Representative donor (left). Surface and intracellular MINCLE and MCL expression on monocytes (n = 14/12), macrophages, and DC (n = 8/6). Percentage of positive cells relative to isotype control (right). (D) Inducibility of MINCLE (left, mean ± SD, n = 7–8) and MCL (right, mean ± SD, n = 5–8) mRNA following 24-h stimulation with 5 μg/ml TDB, TDM, or 10 ng/ml LPS.
To investigate the functional relevance of SYK signaling and of the CLR MINCLE and MCL, we used siRNA knockdown in GM-CSF macrophages and measured IL-8 release in response to the glycolipids. Data analysis was restricted to those donors showing an at least 1.5-fold increase of secreted IL-8 amounts in response to TDB or TDM. To allow comparison of knockdown effects despite donor variability, IL-8 amounts were normalized to the unstimulated control in the nonsilencing condition for each donor. SYK mRNA knockdown efficiency was 89% (7 donors, 7 of 10 donors included) and was confirmed on the protein level (Fig. 6A). SYK knockdown significantly reduced IL-8 secretion induced by TDB and TDM (Fig. 6B). CARD9 is recruited downstream of SYK phosphorylation by protein kinase Cδ (32). siRNA knockdown of CARD9 was 71% efficient on mRNA level (six of nine donors included) (Fig. 6C), it significantly reduced the induction of IL-8 by TDB and TDM (Fig. 6D).
siRNA knockdown of SYK, CARD9, MINCLE, and MCL. (A) Efficiency for SYK siRNA knockdown in GM macrophages. Fold change of mRNA relative to nonsilencing (ns) siRNA determined by quantitative RT-PCR (qRT-PCR) (left, mean ± SD, n = 7) and representative Western blot (right). (B) IL-8 response of GM macrophages treated with ns or SYK siRNA as indicated. Induction of IL-8 was calculated for each donor relative to isopropanol control (without) treated with ns siRNA. Stimulation with TDB, TDM, or LPS (48 h). Mean ± SD, n = 7. Wilcoxon signed-rank test. (C) Efficiency of CARD9 siRNA knockdown in GM macrophages. Fold change of mRNA (left, mean ± SD, n = 6) and representative Western blot (right). (D) Induction of IL-8 was calculated for each donor relative to isopropanol control treated with ns siRNA. Stimulation with TDB, TDM, or LPS (48 h). Mean ± SD, n = 6. Wilcoxon signed-rank test. (E) Fold change of MINCLE and MCL mRNA determined by qRT-PCR following siRNA knockdown relative to ns siRNA. Mean ± SD, n = 6 (left, identical donors). Intracellular FACS staining of MINCLE and MCL following 18 h of LPS stimulation of cells previously treated with siRNA, representative experiment (right). (F) IL-8 response of GM macrophages treated with ns, MINCLE, or MCL siRNA as indicated and stimulated with TDB, TDM, or LPS (48 h). Fold change for each donor relative to isopropanol control treated with ns siRNA. Mean ± SD, n = 6, identical donors. Wilcoxon signed-rank test.
siRNA knockdown of SYK, CARD9, MINCLE, and MCL. (A) Efficiency for SYK siRNA knockdown in GM macrophages. Fold change of mRNA relative to nonsilencing (ns) siRNA determined by quantitative RT-PCR (qRT-PCR) (left, mean ± SD, n = 7) and representative Western blot (right). (B) IL-8 response of GM macrophages treated with ns or SYK siRNA as indicated. Induction of IL-8 was calculated for each donor relative to isopropanol control (without) treated with ns siRNA. Stimulation with TDB, TDM, or LPS (48 h). Mean ± SD, n = 7. Wilcoxon signed-rank test. (C) Efficiency of CARD9 siRNA knockdown in GM macrophages. Fold change of mRNA (left, mean ± SD, n = 6) and representative Western blot (right). (D) Induction of IL-8 was calculated for each donor relative to isopropanol control treated with ns siRNA. Stimulation with TDB, TDM, or LPS (48 h). Mean ± SD, n = 6. Wilcoxon signed-rank test. (E) Fold change of MINCLE and MCL mRNA determined by qRT-PCR following siRNA knockdown relative to ns siRNA. Mean ± SD, n = 6 (left, identical donors). Intracellular FACS staining of MINCLE and MCL following 18 h of LPS stimulation of cells previously treated with siRNA, representative experiment (right). (F) IL-8 response of GM macrophages treated with ns, MINCLE, or MCL siRNA as indicated and stimulated with TDB, TDM, or LPS (48 h). Fold change for each donor relative to isopropanol control treated with ns siRNA. Mean ± SD, n = 6, identical donors. Wilcoxon signed-rank test.
Knockdown efficiency was 65% for MINCLE and 68% for MCL on mRNA level (six donors, six of nine donors included) (Fig. 6E). Intracellular FACS staining of LPS-stimulated cells was conducted to confirm the knockdown on protein level. MINCLE and MCL signals were reduced when the macrophages were treated with the respective siRNA (Fig. 6E). Following TDB and TDM stimulation, mean IL-8 induction was reduced after MINCLE or MCL siRNA knockdown. However, the strength of reduction did not reach the level of significance (Fig. 6F). In conclusion, the siRNA data corroborated a role for SYK and CARD9 in macrophage activation by TDB and TDM but could not definitively prove whether MINCLE and MCL are essential.
For this reason, we tried to further investigate the functionality of the human receptors. We therefore used retroviral transduction of murine DC unresponsive to TDB and TDM with cDNA encoding human MINCLE and MCL. Mouse Mincle–deficient (i.e., Clec4e−/−) DC robustly upregulated G-CSF (Csf3) mRNA when complemented with murine Mincle receptor cDNA encoded by the bicistronic mRNA of mMincle-MigR1. Retroviral transduction of human MINCLE similarly restored GCSF expression in response to TDB/TDM (Fig. 7A). In addition, intracellular staining showed TNF expression in GFP-positive cells after TDB/TDM stimulation when transduced with human MINCLE but not with the empty control vector (Fig. 7B). In contrast, human MCL transduction alone did not restore the response of Clec4e−/− murine DC. Compared with transduction of human MINCLE alone, cotransduction of human MINCLE and human MCL led to significantly stronger G-CSF induction after TDB stimulation (Fig. 7C). We performed a binding assay with glycolipid-coated plates and hMINCLE-Fc to confirm a direct interaction of human MINCLE with TDB/TDM. This clearly showed dose-dependent binding of hMINCLE-Fc to TDB/TDM but not to wells coated with behenic acid, the C22 fatty acid component of TDB (Fig. 7D). Thus, our findings indicate a role of human MINCLE for cord factor recognition, whereas MCL alone was not sufficient.
Retroviral transduction of human MINCLE and MCL in murine Clec4e−/− DC. (A) G-CSF induction (quantitative RT-PCR [qRT-PCR], fold change relative to isopropanol control [without]) following stimulation with 5 μg/ml TDB or TDM of murine wild-type (n = 8) or Clec4e−/− DC retrovirally transduced with MigR1 control vector (n = 14), MigR1-mMincle (n = 7), or MigR1-hMINCLE (n = 12). Mann–Whitney U test. (B) Tnf production of Clec4e−/− DC transduced with MigR1 or MigR1-hMINCLE measured by intracellular cytokine staining after stimulation with 5 μg/ml TDB, 0.5 μM CpG, or control (20 h). Representative FACS plot (left) and quantification of Tnf+ cells (right, mean ± SD, n = 4–5). (C) G-CSF induction (qRT-PCR, fold change relative to unstimulated control) following stimulation with 5 μg/ml TDB or TDM of murine Clec4e−/− DC transduced with equal amounts of MigR1 control vector, MigR1-hMCL, MigR1-hMINCLE, and MigR1-hMINCLE + MigR1-hMCL (1:1 mix). Mean ± SD, n = 6. Mann–Whitney U test. (D) ELISA-based binding assay with hMINCLE-Fc. Wells were coated with isopropanol (Iso), TDB, TDM (50/10/5 μg/ml each), and behenic acid (BeAc, 50 μg/ml).
Retroviral transduction of human MINCLE and MCL in murine Clec4e−/− DC. (A) G-CSF induction (quantitative RT-PCR [qRT-PCR], fold change relative to isopropanol control [without]) following stimulation with 5 μg/ml TDB or TDM of murine wild-type (n = 8) or Clec4e−/− DC retrovirally transduced with MigR1 control vector (n = 14), MigR1-mMincle (n = 7), or MigR1-hMINCLE (n = 12). Mann–Whitney U test. (B) Tnf production of Clec4e−/− DC transduced with MigR1 or MigR1-hMINCLE measured by intracellular cytokine staining after stimulation with 5 μg/ml TDB, 0.5 μM CpG, or control (20 h). Representative FACS plot (left) and quantification of Tnf+ cells (right, mean ± SD, n = 4–5). (C) G-CSF induction (qRT-PCR, fold change relative to unstimulated control) following stimulation with 5 μg/ml TDB or TDM of murine Clec4e−/− DC transduced with equal amounts of MigR1 control vector, MigR1-hMCL, MigR1-hMINCLE, and MigR1-hMINCLE + MigR1-hMCL (1:1 mix). Mean ± SD, n = 6. Mann–Whitney U test. (D) ELISA-based binding assay with hMINCLE-Fc. Wells were coated with isopropanol (Iso), TDB, TDM (50/10/5 μg/ml each), and behenic acid (BeAc, 50 μg/ml).
Discussion
To our knowledge, this paper represents the first comprehensive analysis of the response to cord factor and TDB by primary human macrophages and DC from multiple donors. All human APC tested robustly produced cytokines and chemokines following stimulation with TDB or cord factor. SYK kinase activation was essential for this response and is likely triggered by glycolipid binding to the CLR MINCLE. These data extend the major role of SYK-dependent signaling established in the murine system to human cells and their response to mycobacterial cord factor and its synthetic glycolipid analog TDB to human cells. However, some of our findings appear to be specific for human APC and will require further investigation.
Although both glycolipids triggered comparable responses in monocytes and GM-CSF macrophages, MCSF macrophages responded more strongly to TDM than to TDB. In contrast, DC produced more chemokines and IL-6 when stimulated with TDB. Such a differential activity of TDB and TDM was not observed in the murine system (22, 23). A stronger response to TDM than to TDB may be expected because TDB has shown less toxicity than cord factor after injection in rabbits (50). Especially the ketomycolic acids contained in TDM are very immunogenic (51, 52). We used a constant mass/volume concentration of the respective glycolipid. Differences in the used molar concentrations because of the longer fatty acid chains of TDM are not a likely explanation for the observed effects because we still observed cell type–specific responsiveness to the stimuli using much lower concentrations (Fig. 2C). Instead, differential responses may be caused by differences in expression and composition of receptor(s) among cell types. Our observation of preferential activity of TDB and TDM in different innate immune cell types suggests that the lipid part of trehalose-based glycolipids shapes the strength and type of the response. The synthesis and testing of new cord factor analog structures is an appealing strategy to identify novel adjuvant molecules for human use (53) and may allow to skew the type of APC activation and hence the adjuvant characteristics.
The TDB-containing adjuvant CAF01 directs a long-lasting Th1/Th17 response to protein Ags in experimental animal models (54). Human APC stimulated with TDB or cord factor produced G-CSF and the cytokine IL-6 (Figs. 2, 3), which can both contribute to Th17 differentiation (55, 56). Human monocytes and macrophages also produced IL-1A and IL-1B (Fig. 3), indicating inflammasome activation by TDB and TDM as described for murine neutrophils and BMDC (57, 58). IL-1R signaling contributes to Th17 polarization by CAF01 in mice (59) and drives human Th17 differentiation (60). The Th1-inducing IL12p70 heterodimer was not induced to measurable levels, and IL-12B was only weakly produced by DC. This pattern of cytokines (high GCSF and IL1 but not IL12) was also observed after stimulation of mouse APC with TDB (33). In fact, inhibition of DECTIN1-induced IL-12 production by TDB through MINCLE–SYK signaling has been described in human macrophages (61). The question whether the cytokine profile of TDB-activated human APC is conducive to Th1 or Th17 differentiation is important for the potential use of CAF01 as adjuvant and will be determined in the future in cocultivation assays of APC with Th cells. Interestingly, the response of human APC to stimulation with whole M. tuberculosis is characterized by strong production of IL-1 and IL-6 from monocyte-derived macrophages and significant production of IL-12 only by monocyte-derived DC (62), which was associated with instruction of CD4+ T cells to produce IFNG and IL-17 (63). Although blockade of DECTIN1 partially blocked the production of cytokines triggered by M. tuberculosis (63), it is reasonable to speculate that the TDM from the mycobacterial cell wall contributes to activation of the pathway controlling expression of IL-1 and IL-6.
Similar to murine macrophages and DC, the kinase SYK was required for TDB/TDM responsiveness of human primary macrophages and DC based on pharmacological inhibition (Fig. 4). Because the SYK-inhibitor R406 can also block the activity of other kinases (e.g., FLT3) (42), it was important to demonstrate the role of SYK by siRNA knockdown experiments (Fig. 6). Despite the SYK dependency we observed in human macrophages and DC, it should be noted that we currently cannot rule out an involvement of a non-CLR such as the scavenger receptor MARCO described earlier in the murine system (34). Although MINCLE and MCL siRNA knockdown did not yield a statistically significant downregulation of IL-8 secretion following TDB or TDM stimulation, several lines of evidence suggest that the human CLR MINCLE and MCL act as cord factor receptors in human APC: 1) the response to TDB and TDM in human APC is dependent on SYK and CARD9 expression and function; 2) MINCLE and MCL are highly expressed in monocytes, macrophages and DC; 3) we observed direct binding of a human MINCLE-Fc fusion protein to TDB and TDM; and 4) human MINCLE complemented the response of Clec4e−/− mouse DC to TDB and TDM.
We found high mRNA levels encoding the C-type lectins MINCLE and MCL in monocytes, macrophages, and DC as previously described (30, 64) or suggested from expression in rodents (29, 36, 65). Detection of MINCLE and MCL on the cell surface by flow cytometry turned out to be more difficult, consistent with previous reports of weak cell surface expression of MINCLE on human myeloid (66) and B cells (67). Both receptors showed a mostly intracellular localization in human APC. However, overexpression in HEK293T cells together with the FcRγ-chain resulted in robust cell surface expression (Supplemental Fig. 3C), consistent with previous reports (26, 29). It is not known whether MINCLE binds its ligand(s) on the cell surface or after phagocytosis in the endosome. In our experimental system, the glycolipids were coated onto the cell culture plate, suggesting that the interaction with the CLR takes place at the cell surface. Recent structural data of MINCLE-trehalose cocrystals indicate that ligand binding is sensitive to acidic pH, making a phagolysosomal localization of the MINCLE-TDB/TDM interaction rather unlikely (68–70). Stabilization of expression of murine Mincle by Mcl has recently been suggested as potential mechanism for phagosomal ligand binding (71).
The successful complementation of TDB/TDM responsiveness in Clec4e−/− murine DC through retroviral expression of human MINCLE confirms the functionality of human MINCLE as receptor for TDB/TDM recognition, whereas transduction of human MCL alone was not sufficient to confer responsiveness (Fig. 7C). It is controversially discussed whether MINCLE/Mincle and MCL/Mcl are able to form a heterodimer or a heterotrimer with FcRγ. Heteromerization had potential implications for ligand specificity and affinity (26, 27, 71, 72). The coexpression of MINCLE and MCL in primary human APC constitutes one prerequisite for heterodimer formation as described in rodents. Retroviral cotransduction of MINCLE and MCL led to higher GCSF induction than transduction of an identical total amount of MINCLE alone (Fig. 7C), which may be due to increased MINCLE protein expression as recently described in the murine system or higher affinity of a MINCLE-MCL heterodimer (71, 72).
Taken together, we have shown in this study that mycobacterial cord factor and its synthetic analog TDB activate human monocyte-derived APC to produce chemokines and cytokines through SYK-dependent signaling, likely triggered by the CLR MINCLE. Given the abundance of the cord factor in the mycobacterial cell wall, these findings are relevant for the response of the human innate immune system during tuberculosis. With regard to the novel glycolipid adjuvant TDB, our results provide the first in-depth analysis of its effects on primary human APC from multiple donors and will be important in better understanding the results of ongoing and future clinical trials. Because our results show preferential activation of different macrophage and DC types depending on the lipid moiety, it may be feasible to induce distinct profiles of innate immune activation by developing structurally diverse synthetic MINCLE ligands in the future.
Acknowledgements
We thank Dr. S. Yamasaki for the gift of anti-MINCLE Ab, Dr. F. Nimmerjahn for providing the mFcRγ plasmid, Dr. D. Dudziak for providing the human DCIR and DECTIN2 plasmid, and Dr. C. Bogdan for his comments and critical reading of the manuscript.
Footnotes
This work was supported by the European Commission (Grant FP7 NEWTBVAC) and the German Research Foundation (RTG1660-TPA2 and SFB 796, TP B6).
The online version of this article contains supplemental material.
Abbreviations used in this article:
References
Disclosures
The authors have no financial conflicts of interest.