Article Text

Download PDFPDF

Developing tolerogenic dendritic cell therapy for rheumatoid arthritis: what can we learn from mouse models?
  1. Jeroen N Stoop,
  2. John H Robinson,
  3. Catharien M U Hilkens
  1. Institute of Cellular Medicine, Musculoskeletal Research Group, Newcastle University, Newcastle upon Tyne, UK
  1. Correspondence to Dr Catharien M U Hilkens, Institute of Cellular Medicine,Musculoskeletal Research Group, Faculty of Medical Sciences, Newcastle University, Framlington Place, Newcastle upon Tyne NE2 4HH, UK; catharien.hilkens{at}


One of the therapeutic strategies under development for the treatment of rheumatoid arthritis is based on reinstating immune tolerance by vaccination with autologous dendritic cells with potent tolerogenic function. These tolerogenic dendritic cells (TolDC) can be generated ex vivo and have beneficial therapeutic effects in animal models of arthritis. Although experimental animal models have been instrumental in the development of this novel immunotherapeutic tool, several outstanding questions regarding the application of TolDC remain to be addressed. This paper reviews what has been learnt to date from studying the therapeutic potential of TolDC in animal models of arthritis and discusses issues relating to preventive versus curative effects of TolDC, the antigen specificity of TolDC therapy, the route, dose and frequency of TolDC administration and the safety of TolDC treatment. Lessons learnt from animal models will aid the design of clinical trials with TolDC.

View Full Text

Statistics from


Rheumatoid arthritis (RA) is a debilitating autoimmune disease affecting approximately 1% of adults. It is characterised by chronic inflammation of the synovial joints causing cartilage breakdown and bone destruction.1 The precise aetiology of RA has not been established, but there is evidence to support a role for autoreactive T cells in disease pathogenesis.2 3 T cells specific for joint autoantigens or citrullinated peptides of joint autoantigens can be detected in a proportion of patients with RA (table 1). A role for T cells activated in the presence of high cytokine levels and independent of antigen-specific stimulation has also been proposed in animal studies.4 The synovial membrane of patients with RA is heavily infiltrated with T cells, and T cells contribute to the destruction of cartilage and bone through the production of proinflammatory cytokines such as interleukin 17 (IL-17).5,,9 Although T cell depletion therapies have had limited success in RA, blockade of T cell co-stimulation with cytotoxic T lymphocyte antigen 4 immunoglobulin (CTLA-4 Ig) significantly reduces disease activity.10

Table 1

Overview of candidate autoantigens and the type of T cell response studied in patients with RA and controls

The treatment of RA involves (long-term) use of immunosuppressive drugs such as cytokine antagonists, B cell-depleting antibodies and CTLA-4 blockade. These agents can relieve symptoms in some patients, but treatment seldom results in a drug-free long-lasting remission.11 12 Moreover, these drugs can cause general immunosuppression, potentially compromising protective immunity. For instance, anti-tumour necrosis factor (anti-TNF) treatment can result in an increased risk of infections and malignancy13,,15 and, in rare cases, rituximab treatment can lead to progressive multifocal leucoencephalopathy.16 It is therefore desirable to develop therapeutic strategies to specifically target pathological autoreactive responses. One such new strategy is the adoptive transfer of dendritic cells (DC) with tolerogenic function (tolerogenic DC, TolDC).

DC are antigen-presenting cells that play a central role in the regulation of immunity by initiating immunity to invading pathogens while maintaining tolerance to self-antigens.17 18 TolDC can be generated in vitro by genetic or pharmacological modification. Generally, TolDC are characterised by low expression of T cell co-stimulatory molecules, low production of proinflammatory cytokines and high production of immunoregulatory cytokines compared with immunogenic DC (figure 1). The mechanisms by which TolDC induce tolerance include (1) deletion of T cells; (2) induction of T cell hyporesponsiveness; (3) deviation of the T cell cytokine profile (eg, from T helper type 1 (Th1) to T helper type 2 (Th2)); and (4) induction of regulatory T cells (Tregs). Furthermore, TolDC may also directly dampen immune responses by the secretion of immunosuppressive molecules.17,,20

Figure 1

Tolerogenic dendritic cells (DC) express lower levels of major histocompatibility complex (MHC) class II, CD40 and the co-stimulatory molecules CD80 and CD86 and produce lower levels of inflammatory cytokines (eg, interleukin 1β (IL-1β), IL-12, IL-23, tumour necrosis factor α) but higher levels of regulatory/suppressive cytokines (eg, transforming growth factor β, IL-10).

TolDC have been shown to attenuate pathogenic T cells in animal models of autoimmunity including arthritis models.21,,25 The main benefit of TolDC therapy over currently available immunosuppressive therapies is the potential to specifically target autoreactive T cells. Animal models have been instrumental in the development of TolDC therapy. In this review we discuss what has been learnt so far from investigating the effects of TolDC in mouse models of arthritis.

Experimental arthritis models used to study TolDC therapy

The most frequently used model for RA is collagen-induced arthritis (CIA). Like RA, CIA is characterised by cartilage degradation, fibrin deposition, mononuclear infiltration, synovial cell hyperplasia, pannus formation, periosteal bone formation and eventual ankylosis of one or more joints.26 Immunisation of mice with type II collagen (CII) in complete Freund's adjuvant leads to the activation and expansion of CII-specific CD4 T cells that help B cell production of CII-specific IgG antibodies. These antibodies enter the joint and bind complement, facilitating the entry of activated CD4 T cells.26 Initial studies implicated Th1 cells in the pathogenesis of CIA, however recent evidence suggests that Th17 cells play a predominant role.27 28

Another model of RA is antigen-induced arthritis (AIA). Mice are immunised with an exogenous non-joint antigen (eg, bovine serum albumin) in complete Freund's adjuvant and subsequently intra-articular injection of the same antigen. This protocol results in a T cell-dependent and immune complex-mediated destructive arthritis in the injected joint,25 accompanied by the induction of joint antigen (eg, CII)-specific T and B cells after disease onset.29

Immunotherapy with TolDC in arthritis models

There are several strategies to generate TolDC. A schematic overview of the different types of TolDC and their possible mechanisms of action is depicted in figure 2. Modification of DC in vitro with cytokines including IL-10, TNF or transforming growth factor β is a relatively simple way to generate TolDC. TolDC can also be created by in vitro treatment with drugs that inhibit nuclear factor κB (NF-κB) signalling—for example, the BAY 11-7085 compound, dexamethasone (Dex) and/or vitamin D3.25 30 DC treated with these cytokines or compounds characteristically display an immature or semimature phenotype and a reduced T cell stimulatory capacity.21 22 30,,34

Figure 2

Schematic overview of the potential mechanisms of action for tolerogenic dendritic cells (TolDC). The reduced T cell stimulatory capacity and lower cytokine production of the TolDC can have a direct effect on the pathogenic Th1 and Th17 cells or the TolDC can induce/activate T cells with a regulatory phenotype that will inhibit the function of the pathogenic T cells. CTLA-4, cytotoxic T lymphocyte antigen 4; Dex, dexamethasone; FasL, Fas ligand; GM-CSF, granulocyte macrophage colony stimulating factor; IFN-γ, interferon γ; LPS, lipopolysaccharide; TGF-β, transforming growth factor β; TNF, tumour necrosis factor; TRAIL, TNF-related apoptosis-inducing ligand; VitD3, vitamin D3.

CIA can be prevented by vaccination with DC treated with TNF, IL-10 or Dex.21 22 34 TNF-treated DC (TNF-DC) have been used in several different studies. However, each of these studies proposed a different mechanism by which TNF-DC ‘tolerise’ the immune system in vivo: TNF-DC have been shown to polarise Th2 cell responses,22 trigger the expansion of CD49b+ regulatory T cells (Tregs)34 or induce FoxP3+ Tregs.35

Established arthritis can be inhibited by BAY 11-7085-DC in the AIA model.25 This effect was dependent on increased IL-10 production and was associated with isotype switching from IgG2b to IgG1 and IgA.25 We have recently shown that DC treated with a combination of Dex and vitamin D3 (DexD3-DC) reduced established arthritis in the CIA model.36 We also found an increase in the proportion of IL-10-producing T cells but did not find any changes in CII-specific antibody production. Interestingly, clinical improvement after treatment with DexD3-DC was associated with reduced CII-specific T cell proliferation and a reduction in the proportion of Th17 cells.36 Thus, depending on the type of TolDC used (and perhaps also on the RA model used), TolDC may exert different immunoregulatory actions in vivo.

The function of DC can be genetically modified by their transduction with immunoregulatory cytokines or cell death-inducing molecules. For instance, IL-4-transduced DC skew T cell responses towards a Th2 phenotype with high IL-4 and low interferon γ (IFN-γ) production23 37 and reduce IL-17 production by T cells in vitro.38 These IL-4-transduced DC prevent the onset of CIA37 and inhibit the severity of established CIA.23 In contrast, IL-10-transduced DC were not capable of reducing established arthritis.23 An added benefit of treatment with IL-4-transduced DC may be that IL-4 directly inhibits cartilage and bone destruction.39 40 ‘Killer’ DC that are genetically modified to express Fas ligand (FasL) or TRAIL (TNF-related apoptosis inducing ligand) inhibit pathogenic T cell responses through the deletion of autoreactive T cells.24 41 This approach has been successful in suppressing development of CIA in CII-immunised mice41 and inhibiting established CIA.24

Indoleamine 2,3-dioxygenase (IDO) is a tryptophan-degrading enzyme that inhibits T cell function by depletion of essential tryptophan and/or by producing toxic metabolites.42 IDO also plays a role in the generation of FoxP3+ Treg.43 Both DC transduced with IDO and DC transduced with CTLA-4 Ig suppress established CIA.42 Rapamycin can also be used to generate FoxP3+ Treg-inducing TolDC. Treatment with these TolDC promotes graft survival in transplantation models, but these cells have not yet been tested in arthritis models.44

The constituents of culture media may influence the immunomodulatory action of TolDC. For instance, exposure to fetal bovine serum (FBS) primes DC to induce enhanced levels of Th2 cytokines to FBS-derived antigens45 and it has been suggested that this contributes to the observed immunosuppressive effects of TolDC generated in FBS46. Clinical grade TolDC will be cultured in specialised media that do not contain FBS or any other animal-derived proteins. It is therefore important to verify that the immunomodulatory effect of the TolDC is not due to FBS.

Timing of TolDC administration: prophylactic versus therapeutic effects

Experimental arthritis models provide the opportunity to investigate the effectiveness of TolDC treatment at various stages of disease (figure 3). One option is prophylactic treatment, in which mice are ‘vaccinated’ with TolDC before induction of arthritis. A second option is to treat with TolDC after immunisation but before the onset of clinical symptoms. This could be considered to model the ‘preclinical’ stage of RA. A third option is therapeutic treatment of established arthritis. TolDC treatment has been investigated during all these three disease stages (table 2) and it is promising that TolDC not only have a prophylactic effect but also have a therapeutic impact during established disease.23,,25 36 42 47 48

Figure 3

Schematic overview of the windows for therapeutic intervention in animal models. Prophylactic dendritic cell (DC) treatment is defined as administration of tolerogenic DC (TolDC) before the mice are immunised; preclinical treatment is defined as treatment with TolDC after the first immunisation (before the second immunisation) but before disease onset; and therapeutic treatment is defined as administration of TolDC after disease onset. t, time in days.

Table 2

Overview of studies describing beneficial effects of TolDC in arthritis models

Because of the bespoke nature of TolDC therapy and the high costs involved in applying this therapy, it is unlikely that TolDC will be used prophylactically on a large scale in the near future. The most desirable scenario is that TolDC will be used to cure or reduce disease in established RA. However, it is also feasible to ‘vaccinate’ individuals that are at risk of developing arthritis, such as patients in the long preclinical period of RA during which anticyclic citrullinated protein can be detected without symptoms of arthritis.49 Studies in mice showing that TolDC administration in the ‘preclinical’ phase of autoimmune arthritis prevents development of disease suggest that prophylactic vaccination may be a successful strategy.35 41 50 51

The required phenotype and function of TolDC is likely to depend on the timing of TolDC treatment. Van Duivenvoorde et al21 compared TolDC vaccination at different stages of arthritis development. They showed that TNF-DC, Dex-treated DC and IL-10-treated DC all prevented arthritis when injected before disease induction, but that Dex-treated DC had no beneficial effect on arthritis development when administered during the preclinical stage. The therapeutic effect of FoxP3+ Treg-inducing TolDC in non-obese diabetic mice depended on the timing of TolDC administration; TolDC treatment exacerbated disease in older mice because they produce more IL-1β.52 This cytokine-dependent abrogation of the therapeutic effect of Treg-inducing TolDC may not occur in CIA, since exosomes from IDO- and CTLA-4-transduced TolDC that act through induction of Tregs do have a therapeutic effect in established CIA.42 Nevertheless, these studies underscore the importance of considering the mechanism by which TolDC inhibit the pathogenic autoimmune response in relation to timing of TolDC administration. For instance, killer DC expressing FasL can probably not be used prophylactically since there may not be any pathogenic T cells to delete at that point. Another potential problem with killer DC is that they will only delete those T cells that are specific for the antigen delivered by the killer DC, but they are unlikely to prevent the spreading of immune responses to other autoantigens.

Another outstanding question that needs to be addressed in more detail in mice is the duration of the effect of TolDC treatment on arthritis severity. Prophylactic treatment with TNF-DC was protective for over 100 days. Therapeutic effects of IL-4-treated DC and DexD3-DC were only monitored up to 28 days, at which time the beneficial effect was still apparent.23 36 However, whether these beneficial effects extend beyond 28 days is not known. It can be argued that long-lasting therapeutic effects are most likely achieved with TolDC that induce antigen-specific Tregs, as these Tregs can provide ongoing immune regulation by, for instance, inhibition of DC maturation, thus creating a regulatory feedback loop. The benefit of such a feedback loop has been shown in a cardiac transplantation model.53 Long-term follow-up of mice in therapeutic arthritis studies would be informative. If the effect of TolDC treatment turns out to be transient, the number and timing of additional TolDC treatments will need to be investigated.

Antigen specificity of TolDC therapy

One of the main advantages of TolDC therapy over global immunosuppressive therapies is that TolDC are thought to specifically target pathogenic autoreactive T cells. The question of whether TolDC need to be ‘loaded’ with relevant autoantigen(s) is therefore important. In the CIA model, CII is used as the arthritis-inducing antigen and TolDC have therefore been pulsed with CII in several studies.21 22 35 47 48 Antigen pulsing was also required for the BAY 11-7085-DC treatment in the AIA model.25 According to van Duivenvoorde et al, prophylactic treatment with TNF-DC required CII pulsing whereas vaccination with Dex-treated DC did not.21 22 The requirement for CII pulsing of TNF-DC to delay arthritis onset has recently been confirmed by one group35 but has been contradicted by another group.34 Interestingly, contrary to DC treated with Dex alone,21 DexD3-DC did require pulsing with CII to have a therapeutic effect.36 Jaen et al successfully used non-pulsed DC for inhibiting arthritis severity, indicating that their TolDC could either act in a non-antigen-specific manner or that TolDC can take up the autoantigen in vivo.51 Another possibility is that TolDC were generated from DC precursors that already expressed relevant self-antigen/major histocompatibility complex II complexes.

Combining all these studies, an interesting scenario emerges. In the studies where TolDC were treated with a maturation/activating signal (eg, lipopolysaccharide, LPS), CII pulsing was required.36 41 47 50 51 In contrast, TolDC did not receive an in vitro maturation stimulus in studies showing that CII pulsing was not required.23 24 37 42 A possible explanation for this phenomenon is that immature DC that have not been (fully) activated/matured in vitro will take up antigen in vivo more efficiently.54 However, if in vivo antigen uptake does occur, it raises an important question: can TolDC therapy be targeted to the autoantigen(s) or is there a considerable risk that the immune response to other non-arthritis-related antigens is modulated by the TolDC as well?

Another consideration relating to the translation of TolDC therapy from mouse models to RA is the choice of autoantigen. The arthritogenic antigen in mouse models is known whereas the autoantigens involved in RA are not. There are several potential autoantigens that can be used to pulse DC (table 1). However, responses to these antigens can only be detected in subgroups of patients. A pragmatic approach could be to load TolDC with a selection of peptides derived from different candidate autoantigens (table 1) or to pulse TolDC with autologous synovial fluid. The rationale for the latter approach is that synovial fluid from patients with RA has been shown to contain soluble fragments of candidate autoantigens (eg, CII, aggrecan) that can be efficiently presented by antigen-presenting cells to T cells55 (J H Robinson, unpublished data).

TolDC administration: route, number and frequency

The commonest routes for TolDC administration are intraperitoneal, intravenous or subcutaneous (table 2). All three routes of administration have been shown to be effective for prevention of arthritis with TNF-DC.21 22 34 35 In contrast, the therapeutic effect of IL-4-transduced DC was superior when injected intravenously or intraperitoneally compared with subcutaneous administration.37 DexD3-DC have a therapeutic effect after intravenous administration but not after intraperitoneal administration.36 The reason for these discrepancies is unclear, but it is of interest to note that migration of IL-4-transduced DC depended on the route of injection: After intravenous and intraperitoneal administration most TolDC accumulated in the spleen, whereas subcutaneous vaccination directed TolDC migration to the lymph nodes.37

TolDC generated in different ways will have different phenotypical and functional properties and it is therefore possible that the ideal route of administration for each type of TolDC will be different. For instance, the capacity of different types of TolDC to migrate is likely to vary, especially if the expression of chemokine receptors such as CCR7 is altered by the tolerisation treatment. Another important consideration is that the ideal route(s) for prophylactic versus therapeutic treatment may be different because TolDC trafficking is likely to be highly affected by inflammatory processes in vivo.56 Further studies addressing TolDC migration in vivo will be useful in determining the optimal route of administration for each type of TolDC and disease.

A route of TolDC administration that could be appropriate for the treatment of arthritis but that has not been investigated so far is intra-articular injection. The rationale for this route is that TolDC would be directly ‘delivered’ to the diseased joint where potentially they could dampen ongoing immune responses through the production of anti-inflammatory molecules. Another potential route is intralymphatic administration of TolDC. This seems to result in superior T cell sensitisation compared with intradermal or intravenous administration.57 However, these injection routes will be difficult to test in mice and would require the use of larger animals.

The dose and frequency of TolDC administration are also likely to have an impact on the efficacy of this therapy, but most studies do not address this and inject a single dose either once or three times at 3–4-day intervals. Only few studies have addressed the effects of dose on TolDC action in vivo. Lim et al showed that administration of a low dose of TNF-DC (2×105/mouse) before disease onset reduced CIA symptoms whereas a 10-fold higher dose accelerated arthritis symptoms.35 In contrast, three other CIA studies injecting as many as 2.5×106 TNF-DC per mouse for three consecutive vaccinations showed a preventive effect.21 22 34 These different results may be explained by the different routes of administration used. In studies with the higher TolDC dose the cells were administered by the intraperitoneal or intravenous route whereas, in the study in which the lower dose proved to be more effective, the cells were administered subcutaneously.21 22 34 35 We have recently shown that three intravenous injections with 1×106 cells was the optimal dose for therapeutic treatment of established CIA with DexD3-DC; increasing the dose to 2.5×106 DC did not improve the therapeutic effect and decreasing the dose below 1×106 cells abrogated the therapeutic effect.36

The mechanism of action is likely to be an important factor in determining the required dose and frequency of TolDC administration. For instance, if TolDC act (at least in part) through the secretion of anti-inflammatory cytokines dampening down ongoing immune responses, multiple injections with large doses may be required. In contrast, if the main or sole mechanism of action of TolDC is the induction of antigen-specific Tregs, a single small dose may suffice.

It is also important to take into consideration the difficulty of translating the high doses of TolDC used in mice (up to 2.5×106) to an equivalent dose for treating patients. However, it may be possible to decrease the number of TolDC required by using a more targeted injection route such as intralymphatic or intra-articular administration.

Safety of TolDC therapy

A major problem relating to the use of certain types of TolDC (eg, immature DC or semimature TNF-DC) as a therapeutic tool is that these cells may be unstable—that is, they may remain responsive to further maturation signals. These signals can be provided by pathogens or damaged tissues and can convert TNF-DC into immunogenic DC.58 It has been shown that immunogenic DC can exacerbate CIA.59 Such a conversion may not readily occur in mice that are maintained in pathogen-free environments but could pose a serious threat in ‘real life’—for example, patients with RA with chronic inflammation and running a larger risk of infection. It should be noted, however, that some types of TolDC may benefit from proinflammatory signals. For instance, the immunoregulatory molecule IDO can be induced in DC by interferons or LPS.60

In some studies the TolDC are treated with LPS before administration, usually in the presence of the tolerising agent(s) (table 2). This will probably reduce the risk of further in vivo maturation and conversion into immunogenic DC because LPS-activated DC become refractory to further stimulation, a status referred to as DC ‘paralysis’ or ‘exhaustion’.61 62 For current good manufacturing practice-related regulations, LPS will have to be replaced by a non-toxic synthetic toll-like receptor 4 ligand. Our group has recently shown that human TolDC generated by Dex and VitD3 treatment and activated by a synthetic toll-like receptor 4 ligand (monophosphoryl lipid A) are resistant to further maturation stimuli in vitro.63 The possibility of in vivo maturation needs to be considered when developing TolDC therapy and therefore TolDC should be tested to determine whether they are maturation resistant before in vivo use.

Can combination therapy increase the efficacy of TolDC treatment?

An important consideration for designing TolDC therapy for RA is the mechanism by which TolDC act in vivo. Eradication of pathogenic T cells by killer DC is an attractive and powerful approach but may only be beneficial in the short-term; expansion of pathogenic T cell clones that have escaped deletion is likely to re-occur after killer DC have disappeared. However, depletion of most of the pathogenic T cells might generate a window of opportunity in which patients can temporarily be treated with immunomodulatory drugs such as rapamycin or Dex and VitD3 that prevent the expansion of pathogenic T cells and favour the expansion of FoxP3+ Treg or Tr-1 cells.32 64 65

Modulation of T cell responses (eg, the induction/expansion of Tregs) by TolDC has the potential to reinstate immune tolerance in the longer term. However, the proinflammatory environment in RA may interfere with the downstream immune regulatory effects initiated by TolDC. For instance, TNF, which is found at high levels in patients with RA, is capable of inhibiting the suppressive capacity of human FoxP3+ Tregs.66,,69 Adoptive transfer of Tregs is effective at early stages of arthritis70,,72 but is less successful after development of full-blown disease.72 However, the combination of TolDC therapy and (short-term) anti-TNF treatment might improve the efficacy of the treatment by synergistically enhancing the number and/or function of Tregs. Mouse models of RA will be useful for testing such combination therapies.


TolDC are a promising novel immunotherapeutic tool for the treatment of autoimmune diseases including RA. Animal models of RA have been instrumental in the development of TolDC therapy and have provided important proof-of-concept data showing that established arthritis can be significantly reduced by TolDC treatment. However, important issues remain to be resolved. Whether TolDC require pulsing with a relevant (joint) antigen (eg, CII) needs further clarification, as the main advantage of TolDC therapy over currently available treatments could be its targeting of autoantigen-specific responses. Furthermore, the issue of whether TolDC have to be activated/matured in vitro needs resolving as this issue is potentially of great importance for in vivo stability and safety of TolDC therapy. It is also important that the optimal dose, route and frequency of administration are determined separately for each type of TolDC. Further studies in animal models, addressing these issues, will continue to facilitate the design of future TolDC trials in patients with RA.


The authors thank Professor John Isaacs for discussions and for critical reading of the manuscript.


View Abstract


  • Funding The authors were funded by MRC grant G0601211.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

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.