Objectives Tumour necrosis factor alpha (TNFα) blockade is an effective therapy for rheumatoid arthritis (RA). The immunomodulatory effects of TNFα antagonists are thought to contribute to their therapeutic action. This study investigated whether anti-TNFα therapeutics exerted their immunoregulatory effects through modulation of dendritic cell (DC) function.
Methods Two complementary approaches were taken: in the first ‘in vitro’ approach monocyte-derived DC from healthy donors were matured with lipopolysaccharide and treated with TNFα antagonists in vitro for 48 h. In the second ‘ex vivo’ approach monocyte-derived DC were generated from RA patients before and 8–12 weeks into anti-TNFα treatment. DC were analysed for survival, phenotype, cytokine production and T-cell stimulatory capacity.
Results TNFα blockade during DC maturation in vitro induced approximately 40% of DC to undergo apoptosis. Importantly, the surviving DC displayed a semimature phenotype with reduced levels of HLA-DR, CD80, CD83, CD86 and CCR7, and their production of IL-10 was enhanced compared with DC matured without TNFα antagonists. Furthermore, anti-TNFα-treated DC were poor stimulators of T-cell proliferation and polarised T-cell development towards a higher IL-10/lower IFNγ cytokine profile. Similarly, DC derived from RA patients after anti-TNFα treatment showed impaired upregulation of CD80 and CD86 upon lipopolysaccharide activation and displayed poor T-cell stimulatory activity.
Conclusions The data show that TNFα blockade has profound effects on DC function with downstream, potentially immunoregulatory, effects on T cells. These data provide an interesting new insight into the potential mechanism by which anti-TNFα drugs contribute to the restoration of immunoregulation in RA patients.
Statistics from Altmetric.com
Rheumatoid arthritis (RA) is an autoimmune disease characterised by chronic synovitis, causing cartilage and bone destruction.1 Although the precise aetiology of RA is unknown, recent genetic data clearly implicate T cells in its pathogenesis.2 3 In addition, tumour necrosis factor alpha (TNFα) has been identified as a key mediator of synovial inflammation.4 Therapeutic blockade of TNFα with neutralising antibodies (eg, infliximab) or soluble TNFα receptor immunoglobulin constructs (eg, etanercept) provides significant beneficial clinical effects in approximately 60% of RA patients.5
Anti-TNFα therapeutics target inflammation at multiple levels.6. For instance, they inhibit the induction of pro-inflammatory IL-1 and IL-6,7 8 downregulate cartilage-destroying matrix metalloproteinases,9 reduce the recruitment of inflammatory cells to the synovium10,–,12 and inhibit angiogenesis.13 The immunomodulatory effects of TNFα antagonists have also been reported but remain largely unexplained. For instance, anti-TNFα therapeutics can reverse T-cell anergy,14 restore regulatory T-cell function in RA patients15 16 and reduce Th17 responses in the skin of psoriasis patients.17
Dendritic cells (DC) are professional antigen-presenting cells that regulate T-cell responses. DC have been implicated in RA pathogenesis. Myeloid DC with high T-cell stimulatory capacity are enriched in the synovial tissue of RA patients.18,–,20 It is thought that the presentation of arthritogenic antigens to T cells by these DC contributes to the perpetuation of the inflammatory autoimmune response.21 22 TNFα plays a central role in DC biology. It is a classic DC maturation factor, enhancing the expression of MHC II and co-stimulatory molecules (eg, CD80, CD86).23 TNFα also promotes DC survival.24 25 The maturation status and survival of DC are both important factors that determine the balance between immunity and tolerance. Whereas immature DC induce T-cell tolerance, mature DC activate T cells and induce immunity. The enhancement of DC survival can break immune tolerance, resulting in autoimmune disease.26 27 Conversely, the ablation of DC abrogates T-cell priming to antigens and inhibits autoimmunity.28 29
Considering the key role of TNFα in DC biology, we hypothesised that anti-TNFα therapy exerts its immunoregulatory effects in RA through the modulation of DC function. Recent small studies in RA patients reported no effect of TNFα blockade on DC cytokine production,30 but a reduction in the DC maturation marker CD83.31 In contrast, an in-vitro study of TNFα blockade during DC differentiation and maturation demonstrated no surface phenotype modulation but reduced production of IL-1β, IL-6 and several chemokines by mature DC.32 In the current study we have investigated the effects of TNFα blockade during DC maturation with the Toll-like receptor 4 ligand lipopolysaccharide. Toll-like receptor 4 ligands are enhanced in the serum and synovial fluid of RA patients and are thought to contribute to DC activation and the breakdown of tolerance in RA.33 We have also studied the ex-vivo generation of monocyte-derived DC from RA patients receiving anti-TNFα therapies. Our data suggest that, under both circumstances, TNFα blockade endows reduced T-cell stimulatory properties on the DC, potentially contributing to the immunoregulatory effects of these drugs.
Materials and methods
Peripheral blood samples were obtained with informed consent and following approval by the Newcastle and North Tyneside Research Ethics Committee 2.
Generation of monocyte-derived DC
DC were generated from peripheral blood as described previously.34 Briefly, peripheral blood mononuclear cells were isolated by density gradient centrifugation on Lymphoprep (Axis-Shield Diagnostics, Dundee, UK). CD14 monocytes were isolated by positive magnetic selection using anti-CD14 microbeads (Miltenyi Biotec, Bergisch Gladbach, Germany). To generate immature DC, monocytes were cultured at 0.5×106 cells/ml in the presence of IL-4 and granulocyte macrophage colony-stimulating factor (50 ng/ml each; Immunotools, Friesoythe, Germany) for 6 days. All cells were cultured in RPMI-1640 supplemented with 10% fetal calf serum, 2mM glutamine, 100 U/ml penicillin and 100 μg/ml streptomycin at 37°C with 5% carbon dioxide. Medium supplemented with cytokines was refreshed on day 3.
Treatment of dendritic cells with TNFα antagonists
Immature DC were washed and re-plated in fresh medium without cytokines at 0.25×106 cells/ml. DC were left untreated or were matured with lipopolysaccharides (from Escherichia coli; Sigma, 100 ng/ml) for 48 h in the absence or presence of a soluble TNF receptor p55 immunoglobulin fusion protein (10 μg/ ml),35 the therapeutic anti-TNFα monoclonal antibody infliximab (10 μg/ml; Schering Plough, Kenilworth, New Jersey, USA), or as a control human IgG (10 μg/ml; Grifols, Los Angeles, California, USA). In some experiments dead DC were removed by the annexin V dead cell removal kit (Miltenyi).
Apoptosis and cell phenotype analysis by flow cytometry
DC were washed and incubated with annexin V–APC and PI in annexin V binding buffer (BD Pharmingen, San Jose, California, USA) for 15 min at room temperature. For active caspase-3 intracellular staining, DC were washed and incubated with human IgG for 15 min on ice, then fixed and permeabilised using Fix/Perm buffer for 20 min on ice, washed twice with BD Perm/Wash buffer, and blocked for 15 min on ice in 2% rabbit serum. After blocking, anti-active caspase-3 antibody (BD Biosciences) was added for 20 min at room temperature. For cell surface phenotype analysis of DC and T cells the following monoclonal antibodies were used: CD25 (clone M-A251), CD80 (clone L307.4), CD86 (clone 2331), CD83 (clone HB15e), HLA-DR (clone L243) (all from BD Pharmingen) and CCR7 (clone 150503; R&D Systems, Minneapolis, Minnesota, USA). Cells were washed and incubated with monoclonal antibodies in phosphate-buffered saline with 3% fetal calf serum, 2 mM EDTA and 0.01% sodium azide for 20 min on ice in the presence of human IgG. Cell viability was assessed using Via-Probe (BD Pharmingen). Intracellular FoxP3 was detected using a FoxP3-APC staining kit (PCH101; eBioscience, San Diego, California, USA). Data were collected on a Becton Dickinson FACScan and analysed using FlowJo (Treestar, USA).
DC cytokine production
DC were stimulated at 2×105 DC/ml with CD40 ligand-transfected J558L mouse cells (kindly provided by Peter Lane, Birmingham University, UK) at a 1:1 ratio. After 24 h, IL-10 and TNFα in supernatants were quantified using sandwich ELISA (BD Pharmingen).
T cells were isolated from peripheral blood using CD3-negative selection with the RosetteSep T-cell enrichment cocktail (Stem Cell Technologies, Vancouver, British Columbia, Canada). DC were co-cultured with allogeneic CD3 T cells at a 1:10 ratio, unless otherwise stated. Cytokines in supernatants were analysed by ELISA and T-cell proliferation was assessed by measuring 3H-thymidine incorporation in the last 16 h of culture. In re-stimulation experiments T cells primed by DC for 6 days were rested for 5 days with 0.1 ng/ml rIL-2 (gift from Professor John Robinson), washed and re-stimulated with CD3/CD28 expander beads (1:1; Dynal, Carlsbad, California, USA). No residual DC were observed in the T-cell cultures before re-stimulation. After 72 h, supernatants were harvested. Cytokines in supernatants (IL-10, IL-4, IFNγ, IL-17) were quantified using sandwich ELISA (BD Pharmingen).
Mann–Whitney, two-way analysis of variance (ANOVA) with Bonferroni posttests and t tests were performed using GraphPad software (San Diego, California, USA). A p value less than 0.05 was considered statistically significant with a 95% confidence interval.
TNFα blockade during DC maturation results in apoptosis
Because DC produce TNFα and other survival factors in response to lipopolysaccharide, the effect of TNFα blockade on DC apoptosis after lipopolysaccharide activation was investigated. Two TNFα antagonists were used: A soluble TNF receptor (TNFR) p55 Ig fusion protein and the chimaeric anti-TNFα monoclonal antibody infliximab. Both TNF antagonists were used at 10 µg/ml because (1) this concentration is within the range of anti-TNF serum concentrations found in patients undergoing treatment6 and (2) in-vitro experiments with a TNFR1-Fas transfected cell line showed that this concentration effectively blocked cell death induced by lipopolysaccharide-activated DC supernatants (data not shown).
Removal of granulocyte macrophage colony-stimulating factor/IL-4 from immature DC cultures led to high levels of apoptosis and cell death within 48 h, as visualised by the high percentage of annexin Vpos/PIneg (early apoptotic), annexin Vpos/PIpos (late apoptotic and dead cells) and active caspase-3pos cells (apoptotic cells, figure 1A,B). Lipopolysaccharide activation of immature DC rescued a large number of cells from apoptosis, but the addition of TNFR or infliximab abrogated this ‘rescue’. Immature DC could also be rescued from apoptosis by the addition of soluble rTNFα at concentrations produced by DC upon lipopolysaccharide activation (approximately 10 ng/ml, data not shown). Control IgG did not affect the level of apoptosis, indicating that the effect of the TNFα antagonists was not mediated through Fc-gamma receptor binding. Apoptosis of lipopolysaccharide-activated DC treated with TNFR (LPS-TNFR DC) or infliximab (LPS-inflix DC) increased during the 72 h culture period (figure 1C). These data show that TNFα blockade abrogates the enhanced survival of DC following maturation in response to lipopolysaccharide.
TNFα blockade affects phenotype and cytokine production of lipopolysaccharide-activated DC
To assess the modulation of DC maturation by TNFα antagonists, the expression of CD80 and CD86 (co-stimulatory molecules), CD83 (DC maturation marker), HLA-DR (antigen-presenting molecule) and CCR7 (chemokine receptor) was measured. Dead cells were excluded from the analysis. CD80, CD86 and CD83 expression was significantly reduced by treatment of DC with TNFα antagonists, whereas HLA-DR expression was not significantly inhibited (figure 2A). There was a trend towards reduced CCR7 expression after anti-TNF treatment (p=0.057). DC clustering, normally observed after lipopolysaccharide activation of DC in vitro, did not occur in the presence of TNFα antagonists (figure 2B).
Interestingly, upon CD40 ligation LPS-TNFR DC and LPS-inflix DC produced significantly higher levels of IL-10 than control DC (LPS-IgG DC), while TNFα levels were similar (figure 2C). These data show that TNFα blockade during lipopolysaccharide maturation results in a DC population with a semi-mature phenotype and substantially enhanced IL-10 production.
TNFα antagonists reduce T-cell stimulatory capacity of DC
We next determined whether the partial maturation of DC treated with TNFα antagonists affected their ability to stimulate allogeneic CD3+ T cells. DC treated with TNFR or infliximab had a significantly reduced capacity to induce T-cell proliferation and IFNγ production (figure 3A,B). The reduced T-cell stimulatory capacity of LPS-TNFR and LPS-inflix DC was not simply caused by lower levels of live DC in these populations, because removal of apoptotic DC did not enhance T-cell proliferation (figure 3C). However, from these experiments it cannot be excluded that apoptotic DC may have influenced the maturation of the surviving DC. Nevertheless, together these data show that blocking TNF during maturation diminishes the immunostimulatory capacity of DC.
Anti-TNFα-treated DC modulate cytokine production by T cells
To assess whether anti-TNFα treatment of DC affects the ‘quality’ of T-cell responses, T cells expanded by the various DC populations were analysed further. First, the expansion of regulatory T cells was determined by measuring the percentages of CD25hi/Foxp3pos T cells. FoxP3 was measured 11 days after T-cell activation to ensure that transient activation-induced FoxP3 expression in non-T-regulatory (Treg) cells had disappeared.36 Although T cells primed by anti-TNFα-treated DC (figure 4A) contained a higher percentage of Tregs than T cells primed by immature DC, the percentages of Tregs were similar to the percentages found in T-cell populations primed by mature DC (LPS-DC or LPS-IgG DC). The data indicate that the neutralisation of TNFα during DC maturation does not obviously alter the ability of these DC to expand Tregs.
Second, the polarising activity of anti-TNFα-treated DC on T-cell cytokine production was investigated. Resting T cells were recovered from primary DC–T-cell cultures and re-stimulated with CD3/CD28 beads. T cells primed by anti-TNFα DC produced significantly higher levels of IL-10, IL-4 and IL-17 but lower levels of IFNγ than T cells primed by mature lipopolysaccharide DC (figure 4B). These data indicate that anti-TNFα treatment of DC affects the ‘quality’ of T-cell responses by skewing their cytokine profile, overall favouring anti-inflammatory cytokines.
Phenotype and function of DC derived from RA patients before and after anti-TNFα therapy
To investigate the effects of TNFα blockade on DC from RA patients treated with anti-TNF drugs, monocyte-derived DC were generated from RA patients ex vivo, before and after 8–12 weeks of anti-TNFα therapy (adulimumab or etanercept; see table 1). No anti-TNFα was added to in-vitro cultures. The lipopolysaccharide-induced upregulation of CD80 and CD86 was lower during anti-TNFα therapy in most patients (figure 5A,B): CD80 was lower in three out of five cases and CD86 was lower in four out of four cases. Furthermore, in four out of five cases the T-cell stimulatory capacity of DC during anti-TNFα therapy was reduced (figure 5C) shown by a decreased ability to initiate T-cell proliferation. Comparing the cytokine profiles of primed T cells revealed considerable heterogeneity. However, there was a trend towards enhanced IL-10 and reduced IL-4, IFNγ and IL-17 production by T cells primed by mature DC derived during anti-TNFα therapy (figure 5D). These data suggest a reduced T-cell stimulatory capacity of DC derived from RA patients on anti-TNFα drugs. However, no association was found between the clinical response to anti-TNFα therapy and the functional modulation of DC.
Our experiments demonstrate a profound effect of TNFα blockade on the generation and maturation of DC from their myeloid precursors. Our in-vitro data clearly indicate that blockade of TNFα during final DC maturation results in enhanced DC apoptosis and the induction of semi-mature DC, with reduced co-stimulatory molecule expression and T-cell stimulatory capacity. Importantly, T cells activated by these semi-mature DC produced enhanced levels of anti-inflammatory IL-10 and IL-4 and lower levels of pro-inflammatory IFNγ. It is perhaps surprising that IL-17 levels are modestly but significantly increased in these experiments, but this could be a downstream consequence of reduced IFNγ levels.37 38 When we compared monocytes derived from RA patients receiving TNFα blockade with cells harvested pretreatment from the same patients we found that, even without adding anti-TNF drugs in vitro, they also generated DC with reduced co-stimulatory molecule expression and T-cell stimulatory capacity. These effects on DC generation were most likely caused by anti-TNFα treatment per se, as RA patients receiving and responding to methotrexate alone did not show impaired DC generation or function (data not shown). The question of whether anti-TNF-modulated DC function through the neutralisation of autocrine TNFα and/or by reverse signalling was not addressed here, but both mechanisms may be important.6 39
The difference in IL-17 and IL-4 production between T cells in our in-vitro and ex-vivo experiments was subtle. Furthermore, these are fundamentally different experiments from the perspective of anti-TNFα exposure. In the in-vitro experiments monocytes develop normally and are only exposed to TNFα blockade during the final stages of DC maturation. The ex-vivo experiments are complementary because monocytes develop in the presence of TNFα blockade, which is absent during DC development and maturation. Despite this we can detect an ‘imprinted’ effect of in-vivo anti-TNF exposure in the ex-vivo cultures, with the overall finding of reduced T-cell stimulatory capacity. Our data thus suggest an important effect of TNFα on both DC precursors and DC.
Although TNFα blockade was originally viewed as a simple but potent anti-inflammatory therapy, studies on patients receiving treatment have shown enhancement of suboptimal regulatory T-cell function15 and the generation of a novel subset of regulatory T cells in association with therapy.16 Our findings suggest an additional mechanism by which anti-TNFα therapy enhances immunoregulation, by the generation of ‘semi-mature’ DC that secrete high levels of IL-10 and with the capacity to generate T cells with enhanced anti-inflammatory cytokine secretion. Furthermore, by inducing apoptosis in DC during maturation, anti-TNFα therapeutics may also act by reducing the strength of the autoimmune response.
Differences between our data and limited previous reports also attest to the critical importance of experimental conditions (eg, type of anti-TNFα used and time point of anti-TNFα exposure) when interpreting similar phenomena. Van Lieshout et al32 cultured monocytes in the presence of a monovalent PEGylated form of p55 TNFr (essentially a monomeric form of our sTNFr). In contrast to our data, even after maturation with lipopolysaccharide they could not demonstrate a difference in surface phenotype compared with conventionally matured DC, although chemokine, IL-1β and IL-6 production were reduced when DC from RA patients were matured in the presence of this agent. Another study showed that the addition of infliximab at the start of the monocyte-derived DC culture period resulted in immature DC with reduced CD1a and CD86 expression as well as reduced T-cell stimulatory capacity. These effects were less marked following DC maturation with lipopolysaccharide.40 Zaba et al17 used immunohistochemistry and real-time PCR of psoriatic skin to demonstrate reduced IL-23, IL-17 and IFNγ levels following etanercept therapy, as well as reduced CD83 and DC-lysosomal-associated membrane protein (LAMP) expression by skin DC—data more consistent with our ex-vivo findings, although it is unclear whether these effects were caused by the reduced infiltration of inflammatory cells and/or immunomodulation. They also generated immature DC in vitro in the presence of etanercept, showing reduced CD86 and HLA-DR expression and reduced T-cell stimulatory capacity. A further study showed the pro-apoptotic effects of etanercept on dermal DC in psoriatic plaques,41 again consistent with our observations on DC survival.
TNFα is a pleiotropic cytokine with acute and chronic effects on many cell types. Consequently TNFα blockade, which is widely used to treat active RA, psoriasis and inflammatory bowel disease, can no longer be viewed as a purely anti-inflammatory therapy.42 It has pro-apoptotic effects on various cells and has also been shown to have immunoregulatory consequences in vivo. To these effects can now be added an important modulation of DC survival, phenotype and function. RA is accepted by many as an autoimmune disease initiated by inappropriate T-cell responses against self-antigens and several novel approaches are being developed to target T cells.43 However, T cells are instructed in their activities by DC, which therefore have an even more pivotal role in the development and control of autoimmunity and thus provide a more tempting target. Clinical studies are starting to hint at the potential of TNFα blockade to induce remission in the early stages of RA44 and evolving data, including those presented here, suggest mechanism(s) for these therapeutically revolutionary observations.
In conclusion, we have adopted two complementary approaches to study the effect of TNFα blockade on DC survival, phenotype and function. Whereas DC are more likely to die when TNFα is blocked during their maturation, those that survive exhibit a semi-mature phenotype and poor T-cell stimulatory capacity. In line with this, DC derived from RA patients during anti-TNFα therapy have a similar phenotype. Incubation of T cells with anti-TNFα-treated DC skews the T-cell cytokine profile towards higher IL-10/lower IFNγ production, potentially contributing to the previously demonstrated immunoregulatory effects of TNFα blockade.
The authors would like to thank Tom Wooldridge for the collection of blood samples from RA patients. Mr Wooldridge is funded by Newcastle's UK NIHR Biomedical Research Centre for Ageing and Age-Related Disease.
JDI and CMUH contributed equally.
Funding This work was supported by the Oliver Bird programme from the Nuffield Foundation, the Arthritis Research Campaign (grant no 16361) and the Society of Japanese Pharmacopoeia for Dr Kazuo Suzuki's project of the Ministry of Health, Labour and Welfare in Japan.
Competing interests None.
Ethics approval Peripheral blood samples were obtained with informed consent and following approval by the Newcastle and North Tyneside Research Ethics Committee 2.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.