In the literature, several protocols have been described for the generation of human TolDC. In these studies, TolDC have been derived from monocytes (MoDC) using the cytokines granulocyte macrophage colony-stimulating factor (GM-CSF) and IL-4. However, as described for tolerogenic bone marrow-derived DC (BMDC) in animal models, different drugs or cytokines could be added to GM-CSF/IL-4 culture to manipulate human DC in vitro, to obtain TolDC with specific features 2 . Among those methods, TolDC can be generated with vitaminD3 (VitD3). VitD3-treated DC have the properties of tolerogenic DC; the cells are maturation-resistant, produce IL-10 after stimulation and induce a low proliferation of allogeneic T cells 3 4 5 . More recently, Raïch-Regué et al. showed that VitD3-DC generated from the monocytes of healthy volunteers or patients with relapsing remitting multiple sclerosis have similar properties; a semi-mature phenotype, an anti-inflammatory profile and a low capacity to induce allogeneic T cell proliferation 6 . Furthermore, these cells seem to show potential for clinical application since hyporesponsiveness of myelin-reactive T cells from patients with relapsing remitting multiple sclerosis was observed when these T cells were cultured with autologous TolDC loaded with myelin peptides 6 . In parallel, several studies have investigated the generation of VitD3-TolDC together with dexamethasone (Dex) in order to increase their tolerogenic potential 7 . Prior to the clinical application of these dexamethasone/VitD3 TolDC in rheumatoid arthritis, Harry and colleagues compared the generation of TolDC from healthy volunteers with those from patients with rheumatoid arthritis; a similar phenotype and function was observed between the two groups 8 . In order to favor their migration to the draining lymph nodes and their antigen presentation to T cells, VitD3-DC or Dex/VitD3-DC can be matured in vitro with lipopolysaccharide (LPS). Such cells are described as alternatively activated DC 9 10 and induce memory T cell hyporesponsiveness and naive T cell proliferation associated with low IFN-γ and high IL-10 production 9 . Other maturation stimuli such as a cytokine cocktail or monophosphoryl lipid A have also been analyzed 11 .

By contrast to Dex-DC and VitD3-DC, rapamycin-treated DC (Rapa-DC) express CD83 and CD86 markers and produce low amounts of IL-10 and high levels of IL-12p40/p70, characteristics of a mature DC phenotype 12 . However, Rapa-DC induce low-level proliferation of allogeneic T cells, similar to Dex-DC and VitD3-DC 13 . Furthermore, Rapa-DC secrete high levels of IL-12 after LPS stimulation, thereby promoting the induction of Treg Foxp3⁺ cells in mice 14 ^a.

Another important molecule used to generate tolerogenic DC is IL-10. Two protocols have been used and lead to the differentiation of different types of TolDC depending on whether IL-10 is present from the initiation of culture or added at the end. In fact, DC generated with IL-10 added at the end of culture have an immature phenotype and display resistance to maturation stimuli 15 16 . These DC induce a state of anergy in CD4⁺ T cells 16 and CD8⁺ T cells 17 in an antigen-specific manner 18 . More recently, DC derived from macaque monocytes in the presence of VitD3 and IL-10 were described as having tolerogenic properties, including resistance to maturation and low-level induction of T cell proliferation 19 . The authors demonstrated the safe intravenous injection of these DC to major histocompatibility complex (MHC)-mismatched recipient macaques treated with antihistamine drug and CTLA4Ig (Cytotoxic T lymphocyte Antigen-4 Ig). A transient increase in donor antigen-specific T cell proliferation was detected in these animals without any increase in anti-donor antibodies 19 . Another protocol to generate TolDC with IL-10 consists of culturing monocytes with IL-10 (in addition to GM-CSF and IL-4) from the initiation of culture. In this case, TolDC (called DC10) express CD83, CD80 and CD86, similar to activated/mature cells, but also Ig-like transcript (ILT)2,ILT3, ILT4 and human leukocyte antigen G, similar to Tol-DC. Furthermore, DC10 secrete high levels of IL-10 and induce hyporesponsiveness in allogeneic T cells 20 . A key characteristic of DC generated with IL-10 is their ability to induce the differentiation of Tr1 regulatory T cells 20 21 ^b. Unfortunately, another property of IL-10-producing DC is a decreased trafficking of these cells to the lymph nodes. The chemokine CCR7 participates in the migration of DC to the lymph nodes, and generating mouse DC with IL-10 down-regulates their expression of CCR7 and impairs their in vivo homing to lymph nodes 22 . In a model of mouse cardiac allotransplantation, Garrod et al. showed that injection of DC co-expressing IL-10 and CCR7 induced a significant prolongation of graft survival. However, DC expressing either IL-10 only or CCR7 only had no effect 23 .

Alternative protocols to generate tolerogenic antigen-presenting cells have been described by other teams, with the resulting cells being referred to as myeloid-derived suppressor cells 24 ^c, mesenchymal stem cells 25 26 ^d or regulatory macrophages 27 ^e. Clinical trials in transplantation using mesenchymal stem cells and regulatory macrophages have already been performed.

From the studies described in this section, some TolDC were generated using a clinical grade protocol prior to an application in the clinic 6 8 11 13 . In our center, we chose to generate human TolDC using a simple protocol in which monocytes are cultured with low-dose GM-CSF without any additional cytokines or drugs. This protocol is in accordance with our work performed in mice 28 and is compatible with a clinical approach.

Over the last few years, we have generated and characterized tolerogenic BMDC in rats, mice and non-human primates. In these different models, injection of tolerogenic BMDC leads to a reduced immune response in vivo or an induction of tolerance in transplant models 29 30 31 . Based on this expertise in TolDC generation in animals, we decided to derive TolDC in humans from monocytes in the presence of GM-CSF only. Indeed, the conventional cytokines used to derive dendritic cells from precursors are GM-CSF and IL-4. However, a study performed in mice in 2000 showed that DC generated with a low dose of GM-CSF in the absence of IL-4 have the properties of immature tolerogenic DC. These cells have a high capacity of antigen capture and presentation and induce a low proliferation of allogeneic T cells. Furthermore, they are maturation-resistant and lead to an increase of graft survival after in vivo injection 32 . In parallel, more recently, human MoDC generated in the presence of GM-CSF and without IL-4 were described to have tolerogenic properties in vitro 33 . Human GM-TolDC are derived from monocytes (0.5 million/mL) cultured with a low dose of GM-CSF (100 U/ml) for 6 days. Different doses of GM-CSF were tested and the best tolerogenic phenotype was obtained in the presence of a low dose. In our protocol, medium and cytokines do not have to be renewed as no difference in the phenotype or function of the cells was observed with or without medium and/or cytokines replacement. On day 6, cells are harvested and characterized on their phenotype and their function.

To set up this protocol, we tested methods of monocyte selection. The most common method to obtain monocytes is to positively select CD14-positive cells using microbeads. Another possibility is to enrich monocytes from peripheral blood by elutriation. This purification technique is based on the separation of cells according to their size and density 34 . The latter technique, which has been adapted to Good Manufacturing Practice facilities, is much cheaper and isolates less manipulated monocytes. We derived TolDC in AIMV medium (Gibco Life Technologies) and GM-CSF (CellGenix) using these two techniques from the same donor. Analysis of the phenotype, function and maturation resistance of the cells generated from both monocyte isolation protocols gave similar results. Thus, all subsequent experiments were performed with elutriated monocytes. We next tested different clinical-grade culture media. We compared GM-TolDC cultured with GM-CSF alone (100 U/mL) in Roswell Park Memorial Institute (RPMI)/human albumin medium and in AIMV medium for six days. In the literature, CellGroDC or X-VIVO 15 mediums have also been used to derive human TolDC in clinical grade conditions 8 13 . Control non-tolerogenic DC were generated in parallel in the presence of GM-CSF (100 U/mL) and IL-4 (200 U/mL) in both types of media. After 6 days of culture, the DC were all non-adherent in RPMI/albumin medium whereas half of the cells were adherent in AIMV medium. Analysis of the phenotype and function of the cells revealed major differences as the DC generated with RPMI/albumin did not have tolerogenic properties. As shown in Figure 1A, DC cultured with RPMI/albumin induced a strong proliferation of allogeneic T cells, similar to that observed when T cells were cultured with control DC (either in RPMI/albumin or in AIMV media). By contrast, GM-TolDC induce a very low stimulation of allogeneic T cells, a feature that we described previously in rat and macaque tolerogenic BMDC 29 30 . Furthermore, a higher expression of CD80 was detected at day 6 in RPMI/albumin-GM-DC compared with the AIMV-GM-TolDC. This difference in phenotype was much greater after maturation with LPS/IFNγ, as RPMI/albumin-GM-DC highly over-expressed CD80, CD86 and CD83 (Figure 1B).

In summary, GM-TolDC were more than 90% MHC-II^low CD80^low CD86^low CD40^low with less than 2% contamination with T cells, B cells or natural killer cells. This immature phenotype of human TolDC is in accordance with the findings of our previous studies in rats, mice and non-human primates 28 29 30 . Furthermore, as we showed previously in rats 35 , human GM-TolDC also expressed the tolerogenic marker Epstein-Barr virus induced gene 3 protein. As regards to their function, GM-TolDC induced a weak stimulation of allogeneic T cells compared with control DC. We also found these cells to be semi-resistant to maturation induced by LPS/IFNγ (as shown in Figure 1). In terms of cytokine release, GM-TolDC produced IL-10 but no IL-12 when stimulated with LPS/IFNγ. Like their phenotype, the in vitro function (that is, T cell proliferation induction, maturation-resistance and cytokine production) of human GM-TolDC was similar to that of tolerogenic BMDC described in animal models 28 29 30 . The similarities between the in vitro features of animal TolDC, whose efficacy has been proven in vivo, with those obtained in humans, is encouraging for their potential use in the clinic.

Whereas clinical trials using immunogenic DC to treat cancer have been developed over the last 15 years 1 36 , TolDC therapy is only just emerging in the clinical arena. This was initiated by the pioneer study published in 2001 demonstrating the safety of injecting autologous immature DC into healthy volunteers 37 . Injections of these DC by the subcutaneous route were well-tolerated without signs of toxicity or development of autoimmunity. Inhibition of antigen-specific effector T cell function and induction of antigen-specific CD8 Tregs in vivo were detected in DC-treated volunteers 37 38 . More recently, the first phase I clinical trial using TolDC and including 10 patients was reported in type 1 diabetes 39 . Control DC generated in the presence of GM-CSF and IL-4 were injected into three patients while seven patients received immunosuppressive DC generated in the presence of GM-CSF, IL-4 and antisense oligonucleotides targeting CD40, CD80 and C86 transcripts. In this trial, intra-dermal injections of both control and immunosuppressive autologous TolDC were well-tolerated and safe 39 . Furthermore, two clinical trials in rheumatoid arthritis are on-going, one by Thomas’s group in Australia (University of Queensland) and the other one by Hilkens’s and Isaacs’s group in the UK (University of Newcastle). These different studies highlight the emergence of tolerogenic DC therapy as a new approach to treat autoimmune diseases.

As part of a European project, we plan to test the safety of autologous monocyte-derived TolDC in patients who have had a kidney transplant. To avoid graft rejection, transplant patients receive life-long immunosuppressive drugs (IS). That means that in this clinical trial, TolDC will be injected into patients receiving three IS commonly used in transplantation: tacrolimus, mycophenolate mofetil (MMF) and prednisolone. However, as summarized in Table 1, several studies performed in mice and humans have shown that generation of DC in the presence of these IS modifies the DC phenotype and function. In particular, generation of mouse BMDC with tacrolimus reduces the ability of DC to process and/or present antigens 40 41 . All the drug-treated DC have been shown to induce hypoproliferation of allogeneic T cells. Interestingly, a study published in 2005 showed that neither migration nor survival of injected BMDC were affected by tacrolimus-treatment of recipient rats 42 . Furthermore, co-treatment with donor DC and tacrolimus increased the hyporesponsiveness of recipient T cells. In a model of heart allotransplantation, donor DC alone did not delay rejection while tacrolimus or tacrolimus + donor DC induced a prolongation of graft survival 42 . This work was performed using donor-derived mature BMDC. Prior to the clinical trial, we will test the absence of toxicity of our TolDC in IS-treated animals in a mouse skin graft model. In parallel, graft survival will be monitored in animals treated with IS without TolDC therapy. So far, we have noted that injection of MMF induces a prolongation of graft survival and injection of TolDC does not impair this effect. In fact, a slight increase in graft survival was actually detected (Segovia et al., manuscript in preparation). Similar experiments using the two other IS or the combination of both IS associated or not with DC therapy are on-going. These studies should identify any potential side effects of DC therapy on the IS treatment used in the clinic.

BMDC: bone marrow-derived dendritic cells; CD: cluster of differentiation; CXCL10: chemokine (C-X-C motif) ligand 10; IFN-γ: interferon gamma; IL: interleukin; LPS: lipopolysaccharide; MHC: major histocompatibility complex; MMF: mycophenolate mofetil; MoDC: monocyte-derived dendritic cells; TNF α: tumor necrosis factor alpha.

Others parameters related to DC injections should be considered before doing a clinical trial, such as the time of cell product injection, the numbers of DC injected, and also the number of injections as well as the route of administration. Regarding this last parameter, experiments performed in mice have shown that intravenous injection of Dex/LPS-treated BMDC prolongs cardiac transplant survival whereas subcutaneous injection of the same Dex/LPS-treated BMDC does not increase graft survival 58 . Our preliminary experiments in macaques show that intradermal injection of autologous TolDC prime an immune response while intravenous injection does not (unpublished results). A study also performed in monkeys confirmed the fact that intravenous injection of TolDC is well-tolerated 19 .

The clinical trials described in the previous section of this review either have already been performed or are on-going in autoimmune diseases. In transplantation, another parameter has to be taken into consideration; this is the question of whether TolDC should be derived from the donor or from the recipient. Most of the studies described in rodents have been performed using donor TolDC or recipient TolDC loaded with donor peptides and administrated one week or more before transplantation 2 . In this last part, we discuss the relevance of using autologous TolDC from a safety and efficacy point of view.

Firstly and very importantly, the risk of donor sensitization due to the presence of a slight contaminant cell product or the destruction of the injected cells by non-self recognition cannot be excluded using donor TolDC therapy 59 ; we hypothesize that this risk is minimized using autologous TolDC. Furthermore, to be efficient, donor TolDC (or donor pulsed recipient TolDC) need to be sufficiently activated using LPS or other cytokine cocktails in order to migrate to lymphoid organs and present the antigen to T cells 9 10 11 . To avoid activation-induced maturation, TolDC are also modified using Dex, VitD3 or IL-10 as described in the first section of this review. By contrast, in accordance with our previous work in rodents 30 , autologous TolDC do not require activation or pulsing to be efficient, leading to a reduced risk of cell maturation. In accordance with this hypothesis, the first clinical trial using human TolDC performed with MoDC generated with GM-CSF and IL-4 showed no toxicity and no adverse effects in the patients injected with these DC 39 , supporting the theory that autologous TolDC do not become immunogenic after injection, correlated with an absence of maturation of the cells.

Secondly, as regards the efficacy of the cells, some studies performed recently in mice by Morelli’s group demonstrated that injected donor DC die quickly after in vivo injection. Even if administration of donor TolDC induces tolerance to a transplant, donor TolDC are unable to directly regulate an immune response in vivo 60 . In this context, donor DC mediate their suppressive effects on T cells through endogenous conventional DC from the recipient mouse 61 . We believe the mechanisms of action to be different when using autologous TolDC. Indeed, we detected injected recipient rat TolDC in the spleen at least two weeks post-injection 30 . Studies performed in mice and humans have also shown that DC accumulate in the spleen after intravenous injection 62 63 . Furthermore, our experiments showed that donor-derived MHC ClassII⁺ cells from the graft are present in the spleen of the recipient 3 to 5 days after transplantation. Stainings performed in the spleens of these animals suggest that the donor cells interact with the injected TolDC. Moreover, depletion of graft passenger leukocytes from the donor organ before transplantation prevents any effect of the autologous TolDC injection (Segovia et al. submitted manuscript). These results validate the hypothesis that injected autologous un-pulsed TolDC are able to migrate to the spleen where they capture and process the donor antigen from graft passenger leukocytes (Segovia et al. submitted manuscript) leading to antigen-specific graft acceptance 31 .

Human leukapheresis samples were collected from healthy donors following institutional-approved protocols (Etablissement Français du Sang, Nantes, France). All animal experiments were performed under specific pathogen-free conditions in accordance with the European Union Guidelines and in compliance with the ethical rules of the INSERM.