Immunosuppression by mesenchymal stem cells: mechanisms and clinical applications

Mesenchymal stem cells (MSCs) are multipotential nonhematopoietic progenitor cells that are isolated from many adult tissues, in particular from the bone marrow and adipose tissue. Along with their capacity for differentiating into cells of mesodermal lineage, such as adipocytes, osteoblasts and chondrocytes, these cells have also generated great interest for their ability to display immunomodulatory capacities. Indeed, a major breakthrough came with the finding that they are able to induce peripheral tolerance, suggesting they may be used as therapeutic tools in immune-mediated disorders. The present review aims at discussing the current knowledge on the targets and mechanisms of MSC-mediated immunosuppression as well as the potential use of MSCs as modulators of immune responses in a variety of diseases related to alloreactive immunity or autoimmunity


Introduction
Mesenchymal stem cells (MSCs), also named multipotent mesenchymal stromal cells, are largely studied as new therapeutic tools for a number of clinical applications. Indeed, these cells have been shown to have diff eren tiation capacities as well as paracrine eff ects via the secretion of growth factors, cytokines, antifi brotic or angiogenic mediators [1]. A large body of studies also indicates that MSCs possess an immunosuppressive function both in vitro and in vivo. We review the present knowledge on the mechanisms underlying the immunomodulatory characteristics of MSCs and their applications in animal models of immune suppression or in clinics.

Defi nition of mesenchymal stem cells
MSCs were initially isolated from bone marrow but are now shown to reside in almost every type of connective tissue [2]. MSCs are characterized as a heterogeneous popu lation of cells that proliferate in vitro as plasticadher ent cells able to develop as fi broblast colony forming-units [3]. MSCs are distinguished from hematopoietic cells by being negative for the cell surface markers CD11b, CD14, CD34, CD45 and human leukocyte antigen (HLA)-DR but expressing CD73, CD90 and CD105. Importantly, the capacity to diff erentiate into multiple mesenchymal lineages including bone, fat and cartilage is used as a functional criterion to defi ne MSCs [4].

Immunosuppressive function of mesenchymal stem cells requires preliminary activation
MSC-mediated immunosuppression requires preliminary activation of the MSCs by immune cells through the secretion of the proinfl ammatory cytokine IFNγ, alone or together with TNFα, IL-1α or IL-1β [5,6]. Th is activation step has also been shown in vivo in a model of graft versus host disease (GVHD) since recipients of IFNγ -/-T cells did not respond to MSC treatment and succumbed to GVHD [7]. Indeed, MSCs from mice defi cient for the IFNγ receptor 1 do not have immunosuppressive activity, highlighting the important role of IFNγ in this process [6].

Mesenchymal stem cell immunosuppression is mediated by soluble factors
Although target cell-MSC interactions may play a role, the MSC-mediated immunosuppression mainly acts through the secretion of soluble molecules that are induced or upregulated following cross-talk with target cells. Among these factors, indoleamine 2,3-dioxygenase (IDO) has consistently been reported [8,9]. On stimulation with IFNγ, this enzyme metabolizes tryptophan to kynurenin, causing depletion of local trypto phan and accumulation of toxic breakdown products. IDO, however, exerts its eff ects mainly through the local accumulation of tryptophan metabolites rather than through tryptophan depletion [10]. Whereas the majority of studies

Abstract
Mesenchymal stem cells (MSCs) are multipotential nonhematopoietic progenitor cells that are isolated from many adult tissues, in particular from the bone marrow and adipose tissue. Along with their capacity for diff erentiating into cells of mesodermal lineage, such as adipocytes, osteoblasts and chondrocytes, these cells have also generated great interest for their ability to display immunomodulatory capacities. Indeed, a major breakthrough came with the fi nding that they are able to induce peripheral tolerance, suggesting they may be used as therapeutic tools in immune-mediated disorders. The present review aims at discussing the current knowledge on the targets and mechanisms of MSC-mediated immunosuppression as well as the potential use of MSCs as modulators of immune responses in a variety of diseases related to alloreactive immunity or autoimmunity.
indicate a potentially important function for IDO, human MSCs lacking both IFNγ receptor 1 and IDO still exerted important immunomodulatory activity [11]. Th is observation may be explained at least in part by a recent study reporting that Toll-like receptors expressed on MSCs augment their immunosuppressive activity in the absence of IFNγ through an autocrine IFNβ signaling loop, which was dependent on protein kinase R and able to induce IDO [12]. Contrary to human MSCs, lack of IDO activity was constantly reported for murine MSCs [13,14].
Induction of inducible nitric-oxide synthase (iNOS) by murine MSCs and production of nitric oxide was suggest ed to play a major role in T-cell proliferation inhibition [15]. Nitric oxide is a gaseous bioactive compound aff ecting macrophage and T-cell functions. iNOS is induced in mouse MSCs after activation by IFNγ and TNFα, IL-1α or IL-1β, and MSCs from iNOS -/mice had a reduced ability to suppress T-cell proliferation [6] (Bouffi C, Bony C, Courties G, Jorgensen C, Noël D, submitted). Th e expression level of iNOS mRNA in human MSCs was minimal [14], however, and secretion of nitric oxide by human MSCs was undetectable (Bouffi C, Bony C, Courties G, Jorgensen C, Noël D, unpublished results). Indeed, diff erent mechanisms of immunosuppression exist in diff erent species since human MSCs employ IDO as a major eff ector molecule whereas nitric oxide plays a critical role in mouse MSCs [14].
Prostaglandin E 2 (PGE 2 ) has also been involved in the immunosuppressive activity of MSCs. PGE 2 is a product of arachidonic acid metabolism that acts as a powerful immune suppressant, inhibiting T-cell mitogenesis and IL-2 production, and is a cofactor for the induction of T-helper (Th ) type 2 lymphocyte activity. Production of PGE 2 by MSCs is enhanced following TNFα or IFNγ stimulation, and its inhibition using specifi c inhibitors resulted in restoration of T-lymphocyte proliferation [16]. MSC-derived PGE 2 was shown to act on macrophages by stimulating the production of IL-10 and on monocytes by blocking their diff erentiation toward dendritic cells (DCs) [17,18].
Another MSC-secreted factor, IL-6, has been reported to be involved in the inhibition of monocyte diff erentiation toward DCs, decreasing their stimulation ability on T cells [13,19]. In parallel, the secretion of IL-6 by MSCs has also been reported to delay apoptosis of lympho cytes and neutrophils [20,21].
Other mediators -such as transforming growth factor beta 1, hepatocyte growth factor, heme oxygenase 1 and leukemia inhibitory factor -were shown to be produced by MSCs upon activation [16,[22][23][24]. Th e production of HLA-G5 by MSCs has more recently been shown to suppress T-cell proliferation, as well as natural killer cell cytotoxicity and T-cell cytotoxicity, and to promote the generation of regulatory T (T REG ) cells [25,26]. Cell contact between MSCs and activated T cells induced IL-10 production, which was essential to stimulate the release of soluble HLA-G5.
Any of these molecules alone does not lead to a complete abrogation of T-cell proliferation, indicating their nonexclusive role. Instead, MSC-mediated immunoregulation is the result of the cumulative action displayed by several molecules.

Suppressive eff ects on immune cells
Both CD4 + and CD8 + T-lymphocyte proliferation stimulated with mitogens or specifi c antigens is suppressed by MSCs. Suppression occurred with MSCs from autologous or allogeneic sources, indicating that it was not restricted by major histocompatibility complex (MHC) [27,28]. Inhibition of proliferation depends on the arrest of T cells in the G 0 /G 1 phase of the cell cycle, independently of apoptosis, but instead MSCs promote the survival of stimulated T cells [29,30]. MSCs alter other T-cell functions, such as the decrease of production of IFNγ, IL-2, and TNFα and the increase of IL-4 secretion [16]. MSCs are not targets of CD8 + cytotoxic T cells but they can suppress the cytotoxic eff ects of cytotoxic T cells [31]. Finally, MSCs have been reported to promote, both in vitro and in vivo, the generation of CD4 + CD25 + or CD8 + T REG cells with functional properties [32]. In vivo data, however, are contradictory [33,34]. Recent studies suggest that MSCs may induce a cytokine profi le shift in the Th 1/Th 2 balance toward the antiinfl ammatory phenotype Th 2 [35,36] (Bouffi C, Bony C, Courties G, Jorgensen C, Noël D, personal communication). Indeed, MSCs can suppress antigen-specifi c Tcell proliferation and cytotoxicity as well as inducing anti-infl ammatory or T REG cells.
Most studies have reported that MSCs inhibit the proliferation of B cells that are activated with antiimmunoglobulin antibodies, soluble CD40 ligand or cytokines [37]. Nevertheless, activated B cells became susceptible to the suppressive activity of MSCs in the presence of exogenously added IFNγ [5]. Th e suppressive eff ect of IFNγ was possibly related to its ability to stimulate the production of IDO by MSCs, which in turn suppressed the proliferative response of eff ector T cells. MSCs exert their suppressive eff ect on B-cell terminal diff erentiation through the release of humoral factor(s); they also increase B-cell viability while inhibiting proliferation, arresting B lymphocytes in the G 0 /G 1 phase of the cell cycle [38,39]. Another study, however, reported that MSCs promoted proliferation and diff erentiation of transitional and naive B cells into immunoglobulinsecreting cells, and strongly enhanced proliferation and diff erentiation of memory B-cell populations into plasma cells [40]. Again, diff erences in cell purifi cation proce dures, experimental conditions and timing of analysis may explain discrepancies between studies.
Myeloid DCs are the most potent antigen-presenting cells, essential in the induction of immunity and tolerance. During maturation, immature DCs acquire the expression of co-stimulatory molecules and upregulate the expression of MHC class I and class II molecules together with other cell surface markers such as CD11c, CD80, CD83 and CD86. MSCs inhibit in vitro the maturation of monocytes and CD34 + hematopoietic progeni tor cells into DCs, as shown by a decreased cellsurface expression of MHC class II and co-stimulatory molecules, as well as a decreased production of IL-12 and TNFα [16,19,41]. Th is eff ect is at least partially mediated via the secretion of IL-6 by activated MSCs [13,19] or PGE 2 , which was directly responsible for blocking DC maturation [18]. Th ese studies suggest that MSCs might direct DC maturation toward an anti-infl ammatory or regulatory phenotype responsible for an attenuated T-cell response.
Natural killer cells exhibit cytolytic activity that mainly targets cells which lack expression of HLA class I molecules. Killing by natural killer cells is regulated by a balance of signals transmitted by activating and inhibitory receptors interacting with HLA molecules on target cells. MSCs have been shown to suppress IL-2-driven or IL-15-driven natural killer cell proliferation, IFNγ produc tion and cytotoxicity against HLA class I-expressing targets [42]. Some of these eff ects seem to depend on cell-to-cell contact and on the release of soluble factors, including transforming growth factor beta 1 and PGE 2 , suggesting the existence of diverse mechanisms for MSCmediated natural killer cell suppression.
Neutrophils are also important mediators of innate immunity responsible for microorganism killing via the production of reactive oxygen species. MSCs were shown to delay apoptosis of neutrophils through an IL-6mediated mechanism that was associated with the downregulation of reactive oxygen species [20]. Delayed apoptosis was thought to preserve the pool of neutrophils that will be rapidly available in response to infections. Recently, Nemeth and colleagues suggested that LPS and TNFα stimulated MSCs during sepsis to secrete high levels of PGE 2 , which in turn reprogrammed monocytes and macrophages to produce large amounts of IL-10. Th e released IL-10 seemed to prevent neutrophils from migrating into tissues and causing oxidative damage, thus mitigating multiorgan damage [17]. Th e results therefore suggest that MSCs may modulate the host innate response and improve survival by preventing sepsis.
Th e various studies indicate that MSCs suppress the function of several immune cells; notably the proliferation of T lymphocytes, the DC maturation and the induction of anti-infl ammatory or T REG cells. Some mecha nisms of immunomodulation have been reproduced by several groups, in particular the secretion of IDO, PGE 2 , nitric oxide and HLA-G5 ( Figure 1). Diff erences in the secretome profi le -particularly for IDO and nitric oxide -exist between humans and mice, however, suggesting that several mechanisms are likely to be responsible for the various eff ects reported to date. Our recent data suggest that MSCs exert two levels of action (Bouffi C, Bony C, Courties G, Jorgensen C, Noël D, personal communication). One level occurs locally via the secretion of mediators that inhibit the proliferation of immune cells at the vicinity of MSCs. Th e second induces a systemic response, either an anti-infl ammatory Th 2 immune profi le or, in some instances, the generation of T REG cells.

Homing capacities
Th e traffi cking and homing properties of MSCs are of particular interest for clinical applications aiming at using non-invasive systemic cell administration to treat infl ammation. MSCs have been shown to express a variety of chemokines and chemokine receptors and can home to sites of infl ammation by migrating towards infl ammatory chemokines and cytokines [43,44]. Depend ing on the studies, heterogeneity in surface receptor expression is observed -which is probably due to diff erences in culture conditions and limitations in detection techniques. Homing of cultured MSCs, however, is ineffi cient compared with leukocytes. Th is ineffi ciency has been attributed to a lack of cell adhesion and chemokine receptors but also to the size of MSCs that promote passive cell entrapment and reduce traffi cking [45]. Together with the evidence that host MSCs can mobilize in response to infl ammation or injury, systemically infused MSCs are also frequently observed within the bone marrow or in injured tissues. Indeed, although the understanding of the underlying mechanisms is still required, accumulating evidence suggests that systemic infusion of MSCs may be used for immunosuppressive treatments of various disorders.

Therapeutic applications of mesenchymal stem cells
Th e hypoimmunogenicity of MSCs supports their therapeutic interest in a variety of diseases related to alloreactive immunity or autoimmunity. Indeed, the poor immunogenicity of these cells demonstrated in vitro and in vivo favors the possible use of allogeneic MSCs in acute clinical conditions where the availability of suffi cient numbers of cells is rapidly needed. Th e use of autologous cells, however, may have therapeutic applications in autoimmune diseases or pathologies that allow enough time for isolation and in vitro expansion of MSCs. Th e few clinical applications performed to date confi rm safety with a lack of major adverse side eff ects. Indeed, serial magnetic resonance imaging performed in 226 patients who received MSCs for various ortho pedic conditions showed no evidence of malignant transformation for a mean follow-up of 10.6 ± 7.3 months [46]. Accordingly, although some studies described the capacity of human MSCs to accumulate chromosomal instability in vitro, it was recently reported that, even though some aneuploidy was detected, MSCs showed progressive growth arrest and entered senescence without evidence of transformation either in vitro or in vivo [47]. Further studies are needed, however, to address the in vivo survival of MSCs, ectopic tissue formation and malignant transformation on a larger number of cell preparations.

Transplantation
One of the fi rst in vivo studies showed that systemic infusion of MSCs isolated from bone marrow prolonged the survival of allogeneic skin grafts from 7 to 11 days in baboons receiving MSCs [48]. Using a semi-allogeneic heart transplant mouse model, infusion of donorderived MSCs prolonged cardiac allograft survival through tolerance induction, which was due to CD4 + CD25 + Foxp3 + T REG cell expansion and impaired anti-donor Th 1 activity [49].
In hematopoietic stem cell (HSC) transplantation, MSCs may help reconstitution of the bone marrow stroma after chemoradiotherapy and enhance HSC engraft ment. As early as 2000, autologous MSC infusion was shown to improve the outcome of HSC trans plantation in advanced breast cancer patients [50]. Infusion of allogeneic MSCs, contrary to syngeneic MSCs, has since been demonstrated to result in rejection of stem cell grafts in a murine model of allogeneic bone marrow transplantation [51]. Th e results in animal models on the potential use of MSCs to prevent rejection of allogeneic grafts are confl icting (for a review, see [52]). In a more recent study, however, co-transplantation of donor MSCs with HLA-disparate CD34 + HSCs resulted in sustained hemato poietic engraftment in 14 children without any adverse reaction, indicating that MSCs reduce the risk of graft failure in haplo-identical HSC transplant recipients [53].
MSC infusion may also be very helpful in cord blood transplantation where the limited dose of stem cells delays engraftment and favors graft failure. Th is cell therapy approach has also been used as GVHD prophylaxis in HSC transplantation.

Graft versus host disease
Th e most signifi cant results on the immunosuppressive eff ects of MSCs so far have been observed in the treatment of acute GVHD after allogeneic stem cell transplantation. GVHD occurring beyond 100 days after HSC transplantation is generally called chronic GVHD, which has to be distinguished from acute GVHD that includes persistent, recurrent, or late-onset acute GVHD.
Th e fi rst case of ex vivo expanded haplo-identical MSC injection in a patient with severe grade IV GVHD of the gut and liver resulted in a striking improvement of the disease [54]. A phase II study has since reported that 30 out of 55 patients had a complete response and nine patients showed improvement, indicating that, irrespective of the donor, MSC infusion might be an eff ective therapy for patients with steroid-resistant, acute GVHD [55]. Another report on patients with leukemia, however, showed eff ective prevention of acute GVHD but a higher incidence of relapses in patients who were co-transplanted with MSCs and MHC-identical allogeneic HSCs [56]. Co-transplantation of third-party donor HSCs with cord blood transplants has been shown to overcome the limitation posed by low cellularity of cord blood units for unrelated transplants in adults. For optimization of this therapeutic approach, the risk of GVHD still has to be reduced. Th e co-infusion of MSCs from the same HSC donors was therapeutically eff ective for severe acute GVHD but no signifi cant diff erences in cord blood engraftment and incidence of GVHD were observed [57]. Th e results indicate the therapeutic potential of MSCs for acute GVHD control, but underline the need for better control of safety issues.

Autoimmune diseases
Based on their ability to moderate T-cell proliferation and function, MSCs have also been proposed as a therapeutic option in the treatment of autoimmune diseases.
Th ey have therefore been tested in a variety of animal models of diabetes, experimental autoimmune encephalo myelitis, systemic lupus erythematosus or rheumatoid arthritis.
Contrasted results were reported in rheumatoid arthritis using the experimental collagen-induced arthritis model. We fi rst showed that injection of the allogeneic C3H10T1/2 MSC line did not reverse the disease score [58]. In the same model, however, a single injection of primary MSCs was shown to prevent the occurrence of severe arthritis, which was associated with a decrease in serum proinfl ammatory cytokines [59,60]. Th e use of human adipose-derived MSCs was eff ective in the xenogeneic collagen-induced arthritis model. Th e therapeutic effi cacy was associated with decreased antigen-specifi c Th 1/Th 17 cell expansion, enhanced secretion of IL-10 and generation of CD4 + CD25 + FoxP3 + T REG cells with the capacity to suppress self-reactive T-eff ector responses [61]. Another study reported no convincing increase of T REG cells in vivo despite in vitro evidence of T-cell inhibition by MSCs [62]. Our recent data with primary syngeneic and allogeneic MSCs indicate that MSCs may have a dual eff ect: locally, reducing the clinical signs of infl ammation in the joints, probably via the secretion of antiproliferative mediators; and systemically, by switching the polarization of the host response towards a Th 2 immune profi le (Bouffi C, Bony C, Courties G, Jorgensen C, Noël D, personal communication). Th e divergent mechanistic results obtained from the various studies underline the complexity of the MSC-mediated immunosuppressive process and the diff erences that may be attributed to the various MSC species used [14] and to the diff erent techniques of MSC isolation and culture [7,59].
In the experimental autoimmune encephalomyelitis murine model of multiple sclerosis, MSCs were shown to decrease the clinical signs associated with demyelini zation when injected before or at the onset of the disease, demonstrating the therapeutic effi cacy of MSCs [34]. Th is eff ect was associated with immune suppression of eff ector T cells leading to IL-2 reversible T-cell anergy. Subsequently, it was reported that MSCs inhibited T-cell activation with reduced IL-17 and TNFα levels via the secretion of CCL2 by MSCs [63].
MSC transplantation conferred signifi cant therapeutic eff ects in the systemic lupus erythematosus mouse model of lupus by reconstructing the osteoblastic niche and restoring immune homeostasis [64]. On the basis of these promising results, four treatment-refractory patients were treated with allogeneic MSCs. Th e patients presented a stable 12-month to 18-month disease remission, showing improvement in serologic markers and renal function [64]. Th eir data showed that MSC infusion restored the Foxp3 + cell levels in both mice and systemic lupus erythematosus patients.
Development of autoimmune diabetes results from immune cell dysfunction to maintain peripheral and central tolerance. MSCs may therefore be helpful in regulating T REG /autoreactive T-cell balance. Th e fi rst results were obtained in the NOD/SCID model of chemically induced diabetes using human MSCs. In the treated diabetic mice, an increase in pancreatic islets and beta cells producing insulin was detected with a few glomerular endothelial cells of human origin. Th ere was also a decrease in macrophage infi ltration, resulting in the prevention of pancreatic injury [65]. Th e role of MSCs was fi rst suggested to induce the regeneration of endogenous insulin-secreting cells, and, second, to inhibit the T-cell-mediated immune responses against newly formed beta cells [66]. A shift of the host immune response toward Th 2-like responses was proposed to occur in treated NOD mice [67].

Conclusion
Overall, the current data indicate that MSCs represent a promising alternative strategy in the treatment of various immune-mediated diseases. Encouraging results have already been obtained from the clinics. Many questions remain to be addressed, however, in order to provide better ways to control and optimize the immune response for the benefi t of the patient. Th is implies a better understanding of the underlying mechanisms of immuno suppression as well as satisfying safety concerns as regards the in vivo survival, formation of ectopic tissue and malignant transformation.

Competing interests
The authors declare that they have no competing interests.