Dendritic cell maturation and survival are differentially regulated by TNF receptors 1 and 2

Abstract

The capacity of dendritic cells (DC) to regulate adaptive immunity is controlled by their maturation state and lifespan. Although TNF is a well-known maturation and survival factor for DC, the role of the two TNF receptors (TNFR), TNFR1 and TNFR2, in mediating these effects is poorly understood. By using unique TNF variants that selectively signal through TNFR1 and/or TNFR2, we demonstrate differential functions of TNFR in human monocyte-derived and blood CD1c⁺ DC. Activation of TNFR1, but not TNFR2, efficiently induced DC maturation, as defined by enhanced expression of cell surface maturation markers (CD83, CD86 and HLA-DR) as well as enhanced T-cell stimulatory capacity. In contrast, both TNFR1 and TNFR2 significantly protected DC against cell death indicating that innate signals can promote DC survival in the absence of DC maturation. We further show differential activation of NFκB signaling pathways by the TNFR: TNFR1 activated both the p65 and p52 pathways, whereas TNFR2 triggered p52, but not p65 activation. Accordingly, the p65 NFκB pathway only played a role in the pro-survival effect of TNFR1. However, cell death protection through both TNFR was mediated through the Bcl-2/Bcl-xL pathway. Together, our data show that TNFR1-, but not TNFR2-signaling induces DC maturation, whereas DC survival can be mediated independently through both TNFR. These data indicate differential but partly overlapping responses through TNFR1 and TNFR2 in both inflammatory and conventional DC, and demonstrate that DC maturation and DC survival can be regulated through independent signaling pathways.

Introduction

Dendritic cells (DC) play a key role in both instigating effective immunity against pathogens and maintaining tolerance to self-antigens. The process of DC maturation is critical in determining their immunomodulatory role (1). Under steady state conditions, DC typically remain in an immature state, unable to initiate effector T-cell responses and instead induce T-cell tolerance. Upon maturation in response to ‘danger signals’ (e.g. pathogen associated molecular patterns or tissue-derived factors), DC express enhanced levels of MHC II-peptide complexes, co-stimulatory molecules, and cytokines. They are now equipped to drive distinct T-cell responses, e.g. Th1, Th2 etc. versus regulatory T-cell responses, depending on the nature of the maturation signals they received (1–3). In addition to DC maturation, the survival of DC also determines their immunomodulatory role. Extending the lifespan of DC breaks immune tolerance, resulting in autoimmune manifestations (4), whereas shortening the DC lifespan inhibits autoimmune disease (5). Thus, both the maturation status and survival of DC play a key role in controlling DC function. Understanding the pathways that control these important aspects of DC biology will assist the development of novel immunotherapeutic strategies.

Tumor necrosis factor (TNF) is a pleiotropic pro-inflammatory cytokine that promotes both DC maturation and survival (6, 7). Mice deficient in TNF fail to induce full DC maturation in vivo in response to a viral challenge (8) and the ex vivo generation and/or maturation of DC from these mice is impaired, but can be restored by exogenous TNF (9–11). For the human system, we and others have shown that neutralization of autocrine TNF during DC maturation impairs their survival, enhancement of co-stimulatory molecule expression, and T-cell stimulatory capacity (12–14). Furthermore, DC derived from rheumatoid arthritis patients on anti-TNF therapy display an impaired CD80 and CD86 upregulation after LPS stimulation (13).

TNF is produced as a type II transmembrane protein (mTNF), which can be processed by various metalloproteases resulting in the release of the soluble ligand (sTNF; (15)). Both mTNF and sTNF exist as homotrimers and exert distinct but also overlapping functions. Studies with genetically modified mice have demonstrated that sTNF is required for the development of acute and chronic inflammation, whereas mTNF can protect against chronic inflammation and autoimmunity (16).

TNF signals through two transmembrane receptors, TNF receptor (TNFR)1 and TNFR2. TNFR1 is ubiquitously expressed at low levels, whereas the expression of TNFR2 is restricted and primarily found on subpopulations of immune cells, neuronal tissues and endothelial cells (15). Notably, TNFR2 is only efficiently activated by mTNF, but not sTNF, whereas TNFR1 is equally well activated by both TNF forms (17). We have shown that this differential responsiveness of the TNFR to their cognate ligands is controlled by TNFR’s stalk region (18).

Here, we made use of our previously developed TNFR-selective ligands and a TNF variant mimicking mTNF bioactivity (19, 20) to dissect the roles of both receptors, TNFR1 and TNFR2, as well as sTNF and mTNF in regulating human DC maturation and survival. We found that both TNFR were expressed by monocyte-derived DC (a model for inflammatory DC) and freshly isolated blood-derived myeloid CD1c⁺ DC (representing steady state conventional DC) and demonstrate that sTNF signals (primarily) through TNFR1 whereas mTNF triggers both TNFR. Phenotypical maturation was selectively induced by TNFR1-signalling, whereas the main action of TNFR2 was to prolong survival of DC. These findings suggest that human DC maturation and survival are controlled differentially by TNFR1 and TNFR2, and that regulation of these critical DC features can be segregated. As therapies directed at modifying DC function gain promise in the clinic (21, 22), an understanding of how TNF influences DC biology will help to develop new strategies for the design of novel DC modulating therapeutics.

Materials and Methods

Monocyte-derived DC (moDC) and blood CD1c⁺ DC

All DC cultures were carried out at 37°C, 5% CO₂, in RPMI-1640 (Sigma-Aldrich) supplemented with 10% (v/v) fetal bovine serum (FBS; Gibco BRL), 100 U/ml penicillin, 100 μg/ml streptomycin and 2 mM L-glutamine (all Sigma-Aldrich). The use of Leukocyte Reduction System (LRS) cones from platelet donations from healthy volunteers was approved by the Newcastle and North Tyneside Research Ethics Committee 2. Peripheral blood mononuclear cells (PBMC) were isolated from LRS cones by density gradient centrifugation on Lymphoprep (Axis-Shield Diagnostics). To generate immature monocyte-derived DC (moDC) as previously described (23, 24), CD14⁺ monocytes were isolated from PBMC by positive magnetic selection using anti-CD14 magnetic microbeads (Miltenyi Biotec) and cultured at 0.5 × 10⁶ cells/ml in a 24-well plate in the presence of IL-4 and GM-CSF (50 ng/ml each, Immunotools, Friesoythe, Germany) for 6 days. Medium supplemented with cytokines was refreshed at day 3. CD1c⁺ DC were separated from PBMC using an immunomagnetic negative selection kit (EasySep human myeloid DC enrichment kit; StemCell) and were cultured in a 96-well plate at 6.5 × 10⁴ cells/200 μl.

Stimulation of DC with TNF

Immature moDC were harvested on ice, washed three times in HBSS (Sigma-Aldrich) containing 1% FBS and re-plated in a 24-well plate at either 2.5 × 10⁵ cells/ml (for flow cytometric analyses) or 5 × 10⁵ cells/ml (for all other assays). MoDC or blood CD1c⁺ DC were left untreated or were stimulated with recombinant sTNF (50 ng/ml; 2 × 10⁷ units/mg; provided by Knoll AG, Ludwigshaven, Germany) or with recombinant CysTNF_wt and mutants derived thereof (TNFR1-selective CysTNF32W/86T, CysTNF_R1; TNFR2-selective CysTNF143N/145R, CysTNF_R2; all at 50 ng/ml). The cloning, expression and purification of these TNF variants were performed as described previously (19, 20). As a positive control, DC were treated with e-coli LPS (100 ng/ml; Sigma-Aldrich). For TNFR2-selective stimulation, DC were pre-treated with 80M2 MAb (2 μg/ml; a kind gift from Peter Scheurich, University of Stuttgart, Germany) for 5 min prior to the addition of CysTNF_R2. Where indicated DC were incubated for 30 min with or without TNFR1- and/or TNFR2-specific antagonistic antibodies H398 (Hbt Hycult) and 22221 (R&D Systems), respectively; 10 μg/ml each or 10 μM of signalling pathway inhibitors Abt-737 (Selleckchem) or BAY11-7082 (Cell Signaling) prior to the addition of TNF ligands.

Cell surface protein expression

DC were harvested 24 h after stimulation, and re-suspended in fluorescence-activated cell sorting (FACS) buffer (phosphate-buffered saline (PBS) containing 3% FBS, 0.1% sodium azide and 2 mM EDTA). Cells were incubated for 30 min on ice with human IgG (a kind gift of Sophie Hambleton, UK) and the following mAbs were used as indicated: HLA-DR-PerCP (L203; R&D systems), CD83-APC (HB15e) and CD86-FITC (2331; both from BD Biosciences). For blood CD1c⁺ DC, purity was checked with CD11c-FITC (Miltenyi) and CD1c-PE-Cy7 (Biolegend) mAbs. For the detection of cell surface TNFR1 and TNFR2 a 3-step amplification system was used. First, moDC or PBMC were re-suspended in PBA (PBS containing 0.5% w/v bovine serum albumin (BSA) and 0.01% w/v sodium azide) and incubated with anti-TNFR1 mAb (H398, 100-150 μg/ml) or anti-TNFR2 mAb (MR2-1; 50 μg/ml; Hbt Hycult for moDC; clone 2222.1; 150 μg/ml for blood mDC) in the presence of human IgG and incubated for 1 h on ice. Cells were blocked in 2% rat serum (Sigma) prior to incubation with biotinylated rat anti-mouse IgG antibodies (eBioscience) for 20 min and APC- (moDC; eBioscience) or PE- (blood DC; Beckton Dickinson) coupled streptavidin for another 20 min on ice. The following antibodies were used in order to identify various DC subsets in PBMC: CD1c-PE-Cy7 (L161) and CD141-PerCP-Cy5.5 (M80; both from Biolegend), CD3-FITC, CD11c-V450 (B-ly6), CD14-APC-H7 (MΦPg), CD16-APC (B73.1), CD19-FITC, CD20-FITC, CD45-V500 (H130), CD56-FITC (B159), HLA-DR-AF700 (G46-6) (all BD Biosciences) and CD123-PerCP-Cy5.5 (6H6, eBiosciences). Dead cells were excluded by DAPI positivity and live leukocytes were identified as CD45⁺ cells. B cells (CD19⁺ or CD20⁺), T cells (CD3⁺) and NK (CD56⁺) cells were also excluded. Following this, pDC were identified as HLA-DR⁺/CD1c^-/CD123⁺; CD16⁺ DC were identified as HLA-DR⁺/CD11c⁺/CD14^-/CD16⁺; CD141 DC were identified as HLA-DR⁺/CD11c⁺/CD16^-/CD14/CD141⁺ and CD1c⁺ DC were identified as HLA-DR⁺/CD11c⁺/CD16^-/CD14/CD1c⁺. Data were acquired on a FACSCanto II flow cytometer or LSRII (Becton Dickinson) and analyzed using FlowJo (TreeStar) software.

Cell death analysis

For analysis of active caspase-3 by flow cytometry, moDC or blood CD1c⁺ DC were harvested after 48 h and 24 h of stimulation, respectively, fixed with Cytofix/Cytoperm buffer for 20 min on ice, washed in Perm/Wash buffer (both from BD Biosciences) and incubated with 2% (v/v) rabbit serum (Sigma) for 15 min prior to the addition of mAb specific for active caspase-3-PE (C92-605; BD Biosciences). Cells were incubated at room temperature for 20 min, washed and re-suspended in FACS buffer. Data were acquired on a FACSCanto II flow cytometer and analysed using FlowJo software. The caspase-3 colorimetric assay method was performed on 100 μg of total protein from lysed moDC samples according to the manufacturer’s instructions (R&D Systems) using p-nitroaniline-coupled Caspase-3-specific tetrapeptides (Asp-Val-Glu-Asp (DVED)-pNA) as substrate. Absorbance was measured at 405 nm on a microplate reader (Sunrise; Tecan Ltd.).

Mixed lymphocyte reaction

CD4⁺ T-cells were isolated from LRS cones using a human CD4⁺ T-cell enrichment cocktail (RosetteSep; StemCell Technologies). MoDC were stimulated with sTNF, variants thereof or LPS for 24 h, harvested and thoroughly washed in HBSS containing 1% FBS. 1 × 10⁴ DC were co-cultured with 1 × 10⁵ freshly isolated allogeneic CD4⁺ T-cells in a 96-well plate (200 μl/well, in triplicates). After 3 days supernatants (100 μl/well) were harvested and IFN-γ levels were determined using a specific sandwich ELISA (BD Biosciences); 10 kBq ³H-thymidine (Perkin-Elmer, Shelton, CT) was added for 8 h to measure T-cell proliferation. Radioactivity was quantified using a beta-scintillation counter (Perkin-Elmer Microbeta Trilux).

Western blotting

MoDC were harvested, pelleted and lysed in high salt buffer (250 mM Tris HCl, pH 7.5, 375 mM NaCl, 2.5% (w/v) Deoxycholate (SOC), 1% (v/v) Triton X-100) supplemented with protease inhibitors (Roche Applied Science) and 0.5 mM PMSF. Cell lysates were sonicated and centrifuged at 14,500g for 10 min. Protein concentrations of supernatants were determined by Bradford assay. 80 μg of total protein was denatured at 96°C for 3 min in Laemmli buffer (62.5 mM Tris HCl pH 6.8, 10 % (v/v) glycerol, 4% (w/v) Sodiumdodecylsufate (SDS), 0.1% w/v bromophenol blue, 5% (v/v) β-mercaptoethanol) and resolved by SDS-PAGE followed by transfer to nitrocellulose membranes. Membranes were sequentially probed with Bcl-xl (#2762, Cell Signaling) and GAPDH (6C5, abcam) mAbs, followed by probing with either goat anti-rabbit IgG-HRP (Santa Cruz) or anti-mouse IgG-HRP (Dianova) antibodies. Chemiluminescence substrate SuperSignal West (Thermo Scientific/Pierce) was used, and chemiluminescence was captured on autoradiography film.

NFκB activation assay

The oligonucleotide binding capability of p65 and p52 NFκB was assessed using TransAM transcription factor assay kits specific for p65 or p52 (Active Motif). Nuclear extracts were prepared from moDC stimulated with TNF, variants thereof or LPS for 15 min (for p65) or 24 h (for p52). 5 μg of total nuclear protein per DC sample was used, as determined by Bradford assay and NFκB-binding to oligonucleotide-coated 96-well plates was determined at 450 nm in a colorimetric reaction using p65- or p52-specific and anti-IgG-HRP antibodies.

Statistics

Statistics were performed using Prism 5.0 (GraphPad Software) and a general linear model was performed using SPSS (Version 19) software. Differences between the treatment groups were tested after adjusting for subject-to-subject variability and the assumptions underlying the model confirmed by residual analysis. An F-test was performed to test for differences between the treatment groups and contrasts formed to compare the groups stimulated with TNFR-selective ligands with and without inhibitors.

Results

TNFR1, but not TNFR2 mediates TNF-induced maturation of moDC

To study the mechanism of TNF-mediated DC maturation and survival through the two TNFR, we treated moDC with sTNF and TNF variants with unique specificity: sTNF, which primarily activates TNFR1; CysTNF_wt, an oligomerised TNF-derivative featuring mTNF-like bioactivity, efficiently activating both TNFR; CysTNF_R1, which selectively activates TNFR1, and CysTNF_R2 showing TNFR2-selective stimulatory capacity (19, 20). CysTNF_R2 was used in combination with the non-agonistic TNFR2-specific MAb 80M2, which facilitates efficient stimulation by the CysTNF_R2 ligand through pre-clustering of TNFR2 (19). sTNF and all TNF variants had been demonstrated to possess similar bioactivities when used at saturating concentrations of 50 ng/ml, i.e. at least 50-fold in excess of their half maximal dose (ED50; ED50(sTNF) and ED50(CysTNF_R1) = 0.1-0.3 ng/ml, ED50(CysTNF_R2+MAb 80M2) and ED50(CysTNF_wt) = 0.3-1 ng/ml) as determined by our previously established cellular in vitro system ((18); Supplemental Data Fig. S1)

Interestingly, all TNFR1-stimulating ligands (sTNF, CysTNF_wt and the TNFR1-selective CysTNF_R1) but not the TNFR2-selective ligands (CysTNF_R2 plus MAb 80M2) induced high level cell surface expression of the maturation marker CD83, albeit to a lower degree than in LPS-stimulated DC (Fig. 1A top). The expression level of this marker could be further enhanced by combining any of the TNFR1-stimulating agents with IL-4, which by itself did not affect DC maturation (Fig. 1A; bottom), but had been previously described to augment TNF-induced DC maturation in murine bone marrow-derived DC (25). However, even in presence of IL-4, TNFR2-selective stimulation did not significantly enhance expression of CD83, CD86 or HLA-DR, whereas the TNFR1-stimulating ligands significantly enhanced expression of all these markers (Fig. 1A-D and data not shown). The less potent upregulation of CD83 by CysTNF_R1 as compared to sTNF or CysTNF_wt cannot be explained by the use of suboptimal concentrations of CysTNF_R1, as its bioactivity is very similar to that of sTNF and CysTNF_wt (18), and CD86 upregulation by all three TNFR1-stimulating ligands is comparable. Notably, TNFR2-signaling did not inhibit TNFR1-mediated maturation as co-activation of both TNFR by CysTNF_wt resulted in similar DC maturation marker expression compared to TNFR1-activation alone (Fig. 1). As observed for TNF (24) TNFR-signaling did not lead to the production of proinflammatory cytokines such as IL-12p70, IL-6, IL-23; indicating that TNFR1-signaling results in partial rather than full moDC maturation (data not shown).

Figure 1. TNFR1- but not TNFR2-mediated signaling results in enhanced expression of cell surface maturation markers in moDC.

Immature moDC were left untreated or were treated as indicated with 100 ng/ml LPS, the TNFR2-specific and signaling enhancing antibody MAb 80M2 alone or 50 ng/ml of the TNF variants: sTNF, CysTNF_wt, CysTNF_R1, CysTNF_R2 or CysTNF_R2 in combination with MAb 80M2 for 24 h in the presence (A, bottom, B-D) or absence (A, top) of IL-4 (10 ng/ml). (A) CD83 expression was determined by flow cytometry (black line). Also shown are immature moDC (grey shaded). Data are representative of three independent experiments with moDC from different donors. (B-D) Expression of cell surface maturation markers CD83 (B), CD86 (C) and HLA-DR (D) was determined by flow cytometry. The median and SEM (standard error of the mean) of three independent experiments are shown. Statistical differences were determined in relation to unstimulated (immature) moDC using a student's t test; *P≤0.05 **P≤0.01 ***P≤0.001.

Next, we determined the T-cell stimulatory capacity of TNF-stimulated moDC in a mixed lymphocyte reaction (MLR). Immature moDC stimulated with TNFR1-stimulating ligands displayed a significantly increased ability to induce T-cell proliferation but not IFN-γ production; LPS-activated DC induced both T-cell proliferation and IFN-γ production (Fig. 2). In contrast, TNFR2-stimulated DC did not have an enhanced T-cell stimulatory capacity (Fig. 2), which is in keeping with the lack of a significant increase in maturation marker expression by these moDC (Fig. 1). Thus, TNF-induced moDC maturation as defined by enhanced expression of maturation markers and T-cell proliferative capacity is mediated via TNFR1, but not TNFR2.

Figure 2. MoDC matured in the presence of TNFR1-stimulating ligands possess enhanced T cell stimulatory capacity.

Immature moDC were left untreated or were treated with sTNF, CysTNF_wt, CysTNF_R1, CysTNF_R2 (50 ng/ml each) in presence of MAb 80M2, MAb 80M2 alone or LPS (100 ng/ml) as indicated for 24 h, washed thoroughly and co-cultured with allogeneic CD4⁺ T cells at a ratio of 2 × 10⁴ DC to 1 × 10⁵ T cells for 3 days. Results are representative of at least 3 independent experiments with different donors. The mean and SEM of the replicates are shown. Statistical differences were determined in relation to unstimulated (immature) moDC using student's t test; *P≤0.05 **P≤0.01. (A) Assessment of T cell proliferation by quantification of incorporated thymidine ³H. (B) Supernatants were harvested and IFN-γ production was determined by ELISA.

DC express functional TNFR1 and TNFR2 on their surface which differentially activate the p65 and p52 NFκB pathways

The finding that only TNFR1-signaling induced DC maturation raised the question of whether signaling-competent TNFR2 is expressed by immature moDC. Flow cytometric analysis confirmed that these cells homogenously expressed both TNFR1 and TNFR2 (Fig. 3A). To verify the signaling competence of, in particular, TNFR2, we analyzed the potency of our TNFR-selective ligands to activate p65 and p52 NFκB pathways in colorimetric oligonucleotide binding assays. Surprisingly, p65 was only efficiently activated in immature moDC by ligands stimulating TNFR1 (sTNF and CysTNF_R1), but not by TNFR2-selective stimulation (CysTNF_R2 plus MAb 80M2; Fig. 3B). In contrast, stimulation of either TNFR1 or TNFR2 resulted in significant nuclear translocation and DNA-binding of p52 (Fig. 3C). As expected, stimulation by MAb 80M2 alone did not induce p65 or p52 activation (Fig. 3B, 3C). The differential capacity of the two TNFR to activate p65 could be confirmed by flow cytometry using phospho-p65-specific antibodies (data not shown). These data demonstrate that immature moDC express signaling-competent TNFR2 and indicate that NFκB-signaling mediated by the two TNFR in immature moDC differs in terms of both quality and quantity.

Figure 3. Immature moDC express signaling competent TNFR1 and TNFR2.

(A) Cell surface expression of TNFR1 and TNFR2 on immature moDC. Cells were stained using a biotin/streptavidin-based amplification protocol and TNFR-specific antibodies (TNFR1, black line in panel A, left; TNFR2, black line in panel A, right) and analyzed by flow cytometry. Cells stained only with biotin-conjugated anti-human IgG antibodies and PE-coupled streptavidin are also shown (grey shaded). Data is representative of three independent donors. (B and C) Assessment of TNF- induced p65 (C) and p52 (D) NFκB activation by oligonucleotide binding assays. Immature moDC were stimulated for 30 min (B) or 24 h (C) as indicated, with 50 ng/ml of the TNF ligands or MAb 80M2. Nuclear proteins were extracted and the translocation and DNA-binding capability of p65 (B) and p52 (C) was determined in a colorimetric oligonucleotide binding assay using anti-human p65- and p52-specific antibodies, respectively. The mean and SEM from 3-6 independent donors are shown. Significance is shown in relation to unstimulated (immature) DC and is determined using student's t test; *P≤0.05 **P≤0.01 ***P≤0.001.

MoDC can be rescued from cell death by both TNFR1 and TNFR2

We have previously shown that a large proportion of immature moDC die within 48 hours of cytokine withdrawal (i.e. after removing exogenous IL-4 and GM-CSF) and that LPS-induced autocrine TNF can rescue these cytokine-deprived immature moDC from cell death (13). We therefore investigated whether TNF increased immature moDC survival through TNFR1 and/or TNFR2 by treating cytokine-deprived moDC with our TNF variants, and measuring the level of caspase-3 activation as an indicator of cell death. Interestingly, both TNFR1- and TNFR2-signaling significantly promoted the survival of immature moDC as shown by a reduction in the levels of intracellular active caspase-3 (Fig. 4A, 4B), caspase-3 enzymatic activity (Fig. 4C) and AnnexinV/7AAD staining (data not shown). As expected, CysTNF_R2 or MAb 80M2 alone had no effect on moDC survival (Fig. 4). We also addressed the question of whether the TNF ligands could rescue moDC from cell death after they had been matured with LPS for 24h. We found that fully LPS-matured moDC had down-regulated their expression of TNFR1 and TNFR2, and were not protected from cell death by any of the TNF ligands (data not shown). Thus, TNF plays a role in promoting the survival of immature but not mature DC.

Figure 4. TNFR1- or TNFR2-mediated signaling protects moDC from cytokine deprivation induced cell death.

Immature moDC were left untreated or were treated as indicated with sTNF, CysTNF_wt, CysTNF_R1, CysTNF_R2 (50 ng/ml each) in presence of MAb 80M2 or MAb 80M2 alone for 48 h. Cell viability was assessed by the presence of intracellular active caspase-3 determined by flow cytometry (A and B) and a colorimetric capsase-3 activity assay (C). (A) Cells were stained using anti-human active caspase-3 antibodies. The percentages of cells gated positive for active caspase-3 are indicated. The data shown are representative of six independent experiments. (B) Data of at least five independent experiments with different donors are shown (mean ± SEM). (C) Colorimetric assessment of caspase-3 activity in whole cell lyates using p-nitroaniline-coupled DEVD as substrate. The mean and SEM of three independent experiments are shown. (B and C) Statistical differences were determined in relation to unstimulated (immature) moDC using student's t test; *P≤0.05 **P≤0.01 ***P≤0.001.

To investigate the possibility that TNFR2-selective stimulation prolonged moDC survival indirectly, through TNFR1, e.g. by inducing the production of sTNF or by co-operating with TNFR1, we used TNFR1-specific antagonistic antibodies. The CysTNF_R1-mediated rescue of moDC apoptosis was completely blocked by anti-TNFR1, whereas it did not affect the TNFR2-mediated response (Fig. 5), indicating that TNFR2 exerts its effect in a TNFR1-independent manner. Furthermore, we addressed the question of through which TNFR the ligands sTNF and CysTNF_wt mediate their cell death protective effect. The sTNF-mediated rescue of apoptosis was almost fully inhibited by blocking TNFR1 but not TNFR2, indicating that sTNF primarily acts through TNFR1 in this respect (Fig. 5), as previously seen for other cell types and cellular systems (17, 19). In contrast, the pro-survival effect of CysTNF_wt was only partially alleviated by blocking either TNFR1 or TNFR2, which is most likely due to the strong avidity effect of pre-oligomerised CysTNF_wt and the ability of this ligand to signal through both TNFR. Indeed, the combination of antagonistic TNFR1- and TNFR2-specific antibodies almost completely inhibited CysTNF_wt-mediated rescue of apoptosis, indicating that membrane-bound TNF enhances moDC survival through both TNFR1- and TNFR2-mediated signaling pathways (Fig. 5). These data not only indicate an important role for TNFR2 in DC survival, but also demonstrate that TNFR1- and TNFR2-mediated rescue is independent of each other and that sTNF is not capable of efficiently activating TNFR2 in moDC, whereas membrane-bound TNF potently activates both TNFR1 and TNFR2 in these cells.

Figure 5. TNFR1-antagonism blocks TNFR1-mediated, but not TNFR2-mediated protection from cell death.

Immature moDC were left untreated or were treated with antagonistic TNFR1- or TNFR2-specific antibodies or a combination of both prior to stimulation with sTNF, CysTNF_wt or TNFR-selective variants. Cell viability was assessed by the presence of intracellular active caspase-3 determined by flow cytometry. The percentages of cells gated positive for active caspase-3 are indicated. The data shown are representative of at least four independent experiments with different donors. Statistical differences were determined using student’s t test; *P≤0.05 **P≤0.01 ***P≤0.001.

The Bcl-2/Bcl-xL pathway mediates pro-survival effects of both TNFR1 and TNFR2, whereas p65 NFκB is only involved in the TNFR1-mediated anti-apoptotic effect

To address the question of whether TNFR1 and TNFR2 regulated the lifespan of DC through similar or different pathways, we used specific small molecule inhibitors to target either the p65 NFκB or the Bcl-2/Bcl-xL pathways (Fig. 6A,6B) – both pathways have been shown to be critical for DC survival (14, 26–29). The BH3 mimetic ABT-737, which inhibits the heterodimerisation of Bcl-xL and Bcl-2 with members of the pro-apoptotic machinery such as Bax and Bak (30), reversed the TNFR1- and TNFR2-mediated rescue of moDC cell death (Fig. 6A), indicating that both TNFR can act through the Bcl-2/Bcl-xL pathway. Indeed, ligation of either TNFR1 or TNFR2 efficiently induced Bcl-xL in moDC (Fig. 6C). Expression of Bcl-2 in moDC could not be consistently detected in response to the TNFR-selective ligands (data not shown). In contrast, inhibition of IKKβ, i.e. the classical NFκB signaling pathway, with the BAY-11-7082 compound impaired the cell death-rescue by TNFR1-, but not by TNFR2-mediated signaling (Fig. 6B). These data confirm a pro-survival role of the classical NFκB pathway in moDC and are in line with our observation that activation of the p65 NFκB pathway could only be detected after TNFR1-, but not TNFR2-selective stimulation (Fig. 3A). Together, our data suggest that TNFR1 and TNFR2 activate distinct, but overlapping survival pathways.

Figure 6. Cell death protection mediated via both TNFR is dependent on the Bcl-2/Bcl-xL signaling pathway, however p65 NFκB is only required by TNFR1.

(A and B) Inhibition of signaling pathways upstream of Bcl-2/Bcl-xL inhibits TNFR1- and TNFR2-mediated protection from cell death. Immature moDC were left untreated or treated with 50 ng/ml of CysTNF_R1 or CysTNF_R2 plus MAb 80M2 in presence or absence of the chemical compound ABT-373 (A) or Bay 11-7082 (B). Cell viability was assessed by the presence of intracellular active caspase-3 determined by flow cytometry. The percentages of cells gated positive for active caspase-3 are indicated. The mean and SEM of at least 4 independent experiments with different donors are shown. Statistics were performed as described under materials and methods; *P≤0.05 **P≤0.01 ***P≤0.001. (C) TNFR1- and TNFR2-mediated signaling up-regulates Bcl-xL. Immature moDC were harvested immediately (-, 0 h), after 48 h (-, 48 h) or were stimulated with 50 ng/ml of CysTNF_R1 or CysTNF_R2 plus MAb 80M2 for 48 h. Whole cell extracts were analyzed for the expression of Bcl-xL (top) and reprobed for GAPDH (bottom) by Western blot analysis. The blots shown are representative of two independent experiments.

Steady-state CD1c⁺ DC respond differently to TNFR1- and TNFR2-mediated signals

MoDC are regarded as a model for inflammatory DC, i.e. DC derived from monocytes under inflammatory conditions (reviewed in (31)). To investigate whether TNFR-signaling differs in inflammatory DC versus steady-state DC, we examined the TNF-response in myeloid CD11c⁺/CD1c⁺ DC from peripheral blood (hereafter referred to as CD1c⁺ DC) as an example of a steady-state, conventional, DC subset. We confirmed expression of both TNFR on CD1c⁺ DC as well as other blood DC subsets: myeloid CD141⁺ and CD16⁺ DC (the latter also referred to as ‘non-classical monocytes‘ by other investigators (32)) and plasmacytoid DC (Fig. 7A). Isolation of CD1c⁺ DC generally obtained purities of 75-90 % (Fig. 7B). Interestingly, when analyzing their response to TNFR1- and TNFR2-ligation, sTNF and CysTNF_wt induced phenotypical maturation and protected CD1c⁺ DC against cell death. This was very similar to the response of moDC, however, sTNF was less effective in promoting CD1c⁺ DC survival when compared to moDC (Fig. 1, 4 and 7C, 7D). The same tendency was observed for TNFR1-selective treatment: CysTNF_R1 induced phenotypical maturation of CD1c⁺ DC (Fig. 1) and moDC (Fig. 7C) to a similar degree, but did not significantly protect CD1c⁺ DC from cell death (Fig. 7D), whereas it enhanced survival in moDC (Fig. 4). Together, these data indicate that the TNFR1-mediated survival pathway in CD1c⁺ DC is less effective than in moDC. In contrast, TNFR2-signaling efficiently promoted survival of blood CD1c⁺ DC, similarly as observed for moDC (Fig. 1, 7D). However, in contrast to moDC, TNFR2-mediated signaling also induced low but significant expression of the maturation markers CD83 and CD86 in blood CD1c⁺ DC, albeit to a much lower level than observed in response to sTNF (Fig. 7C). Thus, although the TNF-response of moDC and CD1c⁺ DC is largely the same, there are subtle differences in the extent that these DC respond to TNFR1- and TNFR2-agonists; however the reason for these differences remains unclear at present. Moreover, experiments with antagonistic TNFR2-specific antibodies prevented the protection from cell death by CysTNF_R2, whereas antagonistic TNFR1-specific antibodies had no effect, demonstrating that this protection is mediated through TNFR2 and is independent of TNFR1 (Fig. 7D). Furthermore, as expected, sTNF-mediated survival could be blocked only by antagonistic TNFR1-specific antibodies, but not by TNFR2-antibodies, indicating that sTNF acts through TNFR1 and not through TNFR2 (Fig. 7D). In contrast, we experienced again that it is difficult to antagonize the activity of the strong mTNF mimick, CysTNF_wt, most likely due to avidity effects caused by its wild-type its pre-oligomerised configuration. However, CysTNF_wt-mediated survival was most efficiently blocked by a combination of antagonistic TNFR1- and TNFR2-specific antibodies, indicating that both TNFR contribute to mTNF-mediated survival (Fig. 7D).

Figure 7. Steady state blood CD1c⁺ DC respond differently to TNFR1- and TNFR2-mediated signals.

(A) Cell surface expression of TNFR1 and TNFR2 by peripheral blood myeloid CD1c⁺ DC (CD1c DC), myeloid CD16⁺ DC (CD16 DC), myeloid CD141⁺ DC (CD141 DC) and plasmacytoid DC (pDC), as determined by flow cytometry; TNFR expression (black line), isotype matched control (grey, shaded). (B) Purity of CD1c⁺ DC isolated from peripheral blood as determined by CD1c and CD11c expression. (C and D) CD1c⁺ DC were cultured for 24 h in the absence or presence of TNFR ligands, and in the absence or presence of neutralizing anti-TNFR antibodies as indicated. Expression of CD83 and CD86 (C) and active caspase-3 (D) was determined by flow cytometry. The median and SEM from 3-7 independent donors are shown. Statistical differences were determined using a student's t test; *P≤0.05 **P≤0.01 ***P≤0.001.

In summary, our data demonstrate that TNFR1-mediated signaling is capable of inducing phenotypical maturation and survival of both moDC and blood CD1c⁺ DC, whereas the main role of TNFR2-mediated signaling is to prolong DC survival. These data indicate differential but partly overlapping responses through TNFR1 and TNFR2 in DC, and demonstrate that DC maturation and survival can be regulated through independent signaling pathways.

Discussion

The immunogenicity of DC is determined both by their maturation state and their lifespan. Despite the central role of TNF in promoting DC maturation and survival, the functional role of the underlying signaling receptors exerting these effects remains poorly understood, especially for human DC. We have explored the mechanism by which TNF controls the maturation and survival of human DC (‘inflammatory’ moDC and steady-state blood CD1c⁺ DC) through its two receptors, TNFR1 and TNFR2. We have shown that TNF-mediated phenotypical maturation of human DC is, overall, most potently regulated by TNFR1, whereas DC survival can be significantly enhanced by both TNFR, most likely through upregulation of the anti-apoptotic factor Bcl-xL (Fig. 8). Thus, these data show that TNF promotes DC maturation and survival through distinct signaling pathways, demonstrating that innate signals enhancing DC survival do not necessarily result in enhanced DC maturation.

Figure 8. TNF promotes DC maturation and survival through distinct signaling pathways in human moDC and CD1c⁺ DC.

TNF-mediated phenotypical maturation of human DC is controlled by sTNF- or mTNF-activated TNFR1, but not TNFR2 and independent of TNFR crosstalk. In contrast, signalling by either TNFR augments DC survival, which can be enhanced by mTNF through either TNFR1 or TNFR2, and by sTNF through TNFR1, but not TNFR2. In contrast to TNFR1, TNFR2-mediated DC survival is independent on the activation of p65, however, both TNFR control DC survival through the Bcl-2/Bxl-xL survival pathway resulting in an upregulation of Bcl-xL.

Our study is the first that uses selective TNFR ligands to dissect the roles of TNFR1 and TNFR2 in regulating human DC maturation and survival. Previous studies demonstrated that maturation of DC is impaired in TNFR1 knock-out mice in response to pathogenic challenges (33–35). However, it remains unclear whether this is due to an insufficient expression of mTNF, which also efficiently activates TNFR2, a lack of crosstalk between the two TNFR, or the inability of TNFR2 to induce DC maturation. We have therefore chosen a different approach, i.e. adequate activators for TNFR1, TNFR2 or both TNFR, allowing us to directly investigate the functional roles of TNFR1 and TNFR2 as well as their crosstalk in DC biology.

One of our key results is that immature human DC express signaling-competent TNFR1 and TNFR2, but that TNFR1 is the major TNFR controlling maturation of both moDC and blood CD1c⁺ DC. Although TNFR2 activation also resulted in a weak, but significant upregulation of maturation markers in CD1c⁺ DC (Fig. 7), TNFR2 failed to enhance expression of co-stimulatory molecules and HLA-DR in moDC (Fig. 1). Importantly, TNFR2-signaling does not seem to suppress TNFR1-mediated DC maturation as CysTNF_wt, mimicking mTNF action and thereby triggering both TNFR, potently induced DC maturation to a similar extent as TNFR1 ligands, sTNF and CysTNF_R1. These data indicate that DC maturation is independent of TNFR crosstalk and/or mixed TNFR1/2 receptor complex formation indicating opposing functions for the two TNFR as described previously (16, 36).

DC activation through TNFR1 enhanced the expression of cell surface markers that are typically involved in T-cell activation (e.g. CD86). Although phenotypically mature moDC displayed enhanced ability to induce CD4⁺ T-cell proliferation, they lacked the ability to promote the differentiation of IFN-γ-producing T-cells (Fig. 2). In contrast, LPS-matured moDC potently induced IFN-γ responses in T-cells (Fig. 2). These data are in agreement with the notion that inflammatory mediators (including cytokines such as TNF) by themselves are not sufficient for the full functional maturation of DC, as defined by the DC’s ability to induce effector T-cell responses (37). Induction of effector T-cell responses by DC minimally requires signal 1 (MHCII-peptide complexes), signal 2 (co-stimulatory molecules, e.g. CD86) and signal 3 (pro-inflammatory cytokines, e.g. IL-12p70; (1)). However, activation of moDC by TNFR1 (or TNFR2) agonists alone did not lead to the production of pro-inflammatory cytokines (e.g. IL-12p70, IL-6, IL-23; data not shown), explaining their inability to induce a Th1 response and suggesting a more contributory role for TNF in generating an effective T-cell response to pathogenic challenge.

Another key finding is that the two TNFR differentially activate the canonical (p65) and non-canonical (p52; (38)) NFκB pathways in DC, which may explain their distinct roles in regulating maturation and survival (Fig. 8). Activation of the non-canonical pathway is required for DC maturation (39–42). However, we observed only very little DC maturation after triggering TNFR2 in CD1c⁺ DC (Fig. 7C) and no significant maturation in moDC (Fig. 1), despite significant activation of the p52 NFκB pathway following TNFR2 activation (Fig. 3C). Since TNFR1-mediated activation of the canonical NFκB pathway, initiated by the temporary depletion of IκBα, activates the non-canonical pathway (40), qualitative and quantitative differences between TNFR1- and TNFR2-mediated activation of the non-canonical pathway are expected. Thus, the lack of TNFR2-induced phenotypical maturation in moDC is most likely caused by the failure of this receptor to induce detectable activation of the canonical p65 NFκB pathway (Fig. 3) suggesting that DC maturation requires p65 activation (Fig. 8). In accordance, IκBα was found to restrain spontaneous DC maturation (40).

Both the canonical and non-canonical NFκB pathways trigger pro-survival signals in antigen presenting cells (41–43). Our data suggest that activation of the non-canonical p52 pathway, in the absence of apparent p65 activation, is sufficient to promote DC survival. Furthermore, our inhibitory studies and western blot analyses suggest Bcl-xL as one of the major anti-apoptotic molecules by which both TNFR promote DC survival (Fig. 6A, 6C). Both p65 and p52 can bind to the Bcl-xL promoter, have been implicated in the regulation of its expression and protect against intrinsic and extrinsic induced cell death (44–47). Thus, the differential and independent functions of TNFR1 and TNFR2 in DC may converge at the level of the Bcl-2/Bcl-xL, thus, underlining the central role of the Bcl-2/Bcl-xL pathway in DC function (14, 29).

The survival of DC is regulated by several TNF superfamily members that can either work independently, cross-react and/or cooperate with TNFR-signaling (24, 48). In particular 41BB, CD40 and RANK show functional similarities to TNFR1 in DC survival and maturation (49–51), which may be at least partly due to the secretion of TNF (52–54). However, qualitative and quantitative differences exist. Whereas signaling through 41BB, RANK and the two TNFR is particularly potent in promoting survival through upregulation of Bcl-2 and/or Bcl-xL (Fig. 6A and B; (49, 50, 54, 55), CD40-signaling results primarily in full DC maturation and to a lesser extent DC survival (24). In contrast, RANK-signaling, similarly to TNFR1-signaling, leads to functional, but not full maturation of human moDC, but has high capacity to prolong DC survival (Fig. 1, 2 and 4; (24). This seems to leave TNFR2 with an exceptional role among the TNFR superfamily members, as TNFR2-signalling significantly extends DC’s lifespan while DC maturation remains nearly unaffected (Fig. 1, 2, 4 and 7C and D).

What might be the role of TNFR2-induced extended lifespan of DC in the absence of DC maturation? Studies manipulating DC lifespan demonstrated that the regulation of DC survival is in itself an important factor determining DC function: prolonging the DC lifespan can break immune tolerance (4), whereas reducing DC survival can inhibit autoimmune disease (5), proposing the assistance of immune and/or pro-inflammatory responses by TNFR2. Thus, specific activation of TNFR2, proposed as a new immunotherapeutic approach to kill auto-reactive T-cells (56), may counteract tolerance induction through prolonging the lifespan of DC. Alternatively, development of a therapeutic TNFR2 agonist may be a useful and convenient tool for the ex vivo generation of tolerogenic DC-based immunotherapies, as it may prolong the survival of these tolerogenic DC without inducing their maturation. In addition and in contrast to above the non-canonical NFκB pathway, i.e. RelB:p52, has been implicated in the control of the immunoregulatory phenotype of DC (57–59). Since TNFR2 activates primarily the non-canonical p52 NFκB pathway, it is tempting to speculate that TNFR2 is involved in the induction/maintenance of self-tolerance by inducing tolerogenic DC. Accordingly, TNFR2 has been attributed an anti-inflammatory function in several models of inflammation (60, 61). However, our initial attempt to support this idea by investigating whether TNFR2-signalling in moDC promoted IL-10-producing T-cells revealed inconclusive results, and further work is needed to explore this hypothesis.

In conclusion, the results of this study demonstrate that DC maturation and survival are differentially and independently regulated through TNFR1 and TNFR2 and point to the canonical NFκB pathway controlling DC maturation and the non-canoncial NFκB pathway regulating DC survival (Fig. 8). Thus, this study contributes to a better understanding of TNFR-mediated immune regulation, which is particularly important for the development of novel immunotherapeutics designed to specifically target TNFR1 or TNFR2.