Follicular dendritic cell
mesenchymal stem cell
systemic lupus erythematosus
Mesenchymal stem cells (MSC) are a rare subset of stem cells residing in the bone marrow where they closely interact with hematopoietic stem cells and support their growth and differentiation. MSC can differentiate into multiple mesenchymal and non-mesenchymal lineages, providing a promising tool for tissue repair. In addition, MSC suppress many T cell, B cell and NK cell functions and may affect also dendritic cell activities. Due to their limited immunogenicity, MSC are poorly recognized by HLA-incompatible hosts. Based on these unique properties, MSC are currently under investigation for their possible use to treat immuno-mediated diseases. However, both their condition of immunoprivilege and their immunosuppressive function have recently been challenged when analyzed under particular experimental conditions. Thus, it is likely that MSC effects on the immune system may be deeply influenced not only by cell-to-cell interactions, but also by environmental factors shaping their phenotype and functions.
Bone marrow (BM) contains rare cells of non-hematopoietic origin (representing 0.01–0.001% of total BM cells) 1 capable of differentiating into cells of mesodermal lineage including osteoblasts, adipocytes, and chondrocytes 1, 2. These cells constitute the stromal trabecular scaffold which, close to the endosteum, tightly interacts with hematopoietic stem cells (HSC) 3. The non-hematopoietic BM cells supporting hematopoiesis are currently defined as BM stromal cells or "mesenchymal stem cells" (MSC), although they represent a non-homogeneous population of multipotent cells whose true stem cell component is limited 4. There is a general consensus that MSC contribute to the formation of the HSC "niche", thus providing an appropriate microenvironment for control of the maturation, differentiation and survival of blood-born cells 5, 6.
MSC have been isolated also from many other human tissues 7 and can give rise, under specific experimental conditions, to several "non-mesodermal" cell types including neurons ("transdifferentiation") 8–10. As a consequence, MSC hold promise not only for improving HSC engraftment upon allogeneic transplantation 11, but also for repairing damaged tissues. However, while MSC are naturally prone to differentiate into cells of mesodermal origin, their capacity of transdifferentiating in vivo is still debated and it is likely to be of limited biological relevance. If this holds true, their possible application for neural repair would be marginal.
More recently, MSC have been shown to inhibit T cell proliferation upon antigen stimulation 12–14. Since these early reports, many studies have demonstrated that MSC affect the function of different immunocompetent cell types, lending support to their pioneering utilization in acute graft-versus-host disease (GVHD) 15 and providing the rationale for their possible use in the treatment of autoimmune disorders 16. However, recent reports have stressed that the outcome of the interaction between MSC and immune cells is strongly influenced by environmental factors, possibly leading to conflicting results 17–19. This review addresses the known effect of MSC on the immune system in light of their possible use in the treatment of immune-mediated diseases.
MSC and the HSC niche
Multipotent MSC are part of the highly specialized "microenvironment" participating in the regulation of HSC survival, quiescence and, upon specific triggers, differentiation into mature elements. The supportive interaction of mesenchymal stromal cells with hematopoietic progenitors in the BM is well exemplified by the fundamental influence of the former cells on early B cell lymphopoiesis. In this context, it has been shown that the close interaction with BM stromal cells is crucial for the normal development of progenitor B cells and is supported by cytokines including interleukin-7 (IL-7), stem cell factor (SCF), Flt3 ligand, thymic stromal lymphopoietin (TSL) and CXCL12 (also known as stromal cell-derived factor-1, SDF-1) 20. Responsiveness of B cell progenitors to cytokines requires the surface expression of the pre-B cell receptor (pre-BCR) 21 which is indispensable for pre-B cell survival, proliferation, differentiation and allelic exclusion. It has long been postulated that tickling of the pre-BCR is necessary for pre-B cell rescue from spontaneous apoptosis, but the mechanism(s) has remained obscure until recently. Gauthier and colleagues demonstrated that Galectin-1 (Gal-1) is a stromal cell ligand of the human pre-BCR implicated in synapse formation between pre-B and stromal cells and in pre-BCR triggering 22. Interestingly, Gal-1 is a member of a large family of calcium-independent S-type lectins which are expressed at highest level in MSC 23. Recently, we demonstrated that several molecules involved in the constitution of the HSC niche synapse 24 such as Gal-1, angiopoietin-1, osteopontin and thrombospondin-1 and -2 are highly expressed by MSC displaying immunomodulating capacity (Pedemonte et al., manuscript submitted).
These data support the notion that MSC may play a direct role in the support of hematopoietic cells inside the BM stem cell niche.
Immunological phenotype and functions of MSC
MSC can be obtained easily from a BM aspirate and can be isolated and expanded through passages in plastic plates where they grow as adherent cells in appropriately enriched media, reaching confluence at time intervals related to plating density. As anticipated, these progenitor cells are capable of differentiating towards different lineages and do not appear to represent a homogeneous population of stem cells 4. Despite the lack of MSC-specific markers, there is now agreement on the fact that human MSC do not express the hematopoietic markers CD34, CD14 and CD45, while they are positive for CD44, CD71, CD73, CD90, and CD105. Similarly, murine MSC do not express the hematopoietic markers CD45, CD34 and CD11b, while they are characterized by surface CD9, Sca-1, and CD44. On steady state, MSC constitutively express low surface densities of MHC class I molecules and are negative for MHC class II as well as for costimulatory molecules such as CD80, CD86 and CD40. Both MHC class I and II molecules are up-regulated by inflammatory stimuli 14, 25, 26. In addition, MSC express different adhesion molecules including several integrins 27. This feature is consistent with their ability to establish firm adhesive interactions with hematopoietic progenitors inside the HSC niche.
The MSC capacity of inducing proliferative responses by allogeneic PBMC represents a controversial issue 14, 25, 26, 28; notably, however, both human and murine MSC are generally considered to be poorly immunogenic cells. MSC can also produce a variety of growth factors, cytokines, chemokines and proteases that are likely to play a role either in their immunomodulatory or in their migratory function 29–31. In this respect, MSC have been shown to express a restricted pattern of chemokine receptors, including CXCR4, allowing them to migrate to tissues upon specific chemotactic triggers 32–35. These features are likely to represent the basis for MSC homing to multiple organs where they undergo a program of tissue-specific differentiation 36. Despite the current lack of information on Toll-like receptor (TLR) expression, MSC are profoundly influenced by microenvironmental factors and can respond to some inflammatory cytokines such as IL-1β 37, IL-17 38 and (more importantly) IFN-γ, all capable of significantly affecting their function. In this context, it is worth stressing that while under some circumstances IFN-γ appears to enhance the immunosuppressive activity of human MSC 39, in other cases it can induce MSC to act even as non-conventional antigen-presenting cells (APC) 17, 40. These data are of great importance in view of the potential utilization of MSC for therapeutic purposes. Thus, IFN-γ-conditioned murine MSC were shown to present ovalbumin upon in vivo infusion and to confer protection from ovalbumin-expressing tumor cells 40. Overall, as suggested by in vitro and in vivo experiments, the functional behavior of MSC is the result of the combined effect of soluble factors and of mechanisms mediated by cell-to-cell contact.
MSC and T lymphocytes
The possibility that incompatible BM-derived stem cells transplanted into a mismatched recipient could induce tolerance to allogeneic or xenogeneic grafts has been postulated since 1984 41. However, only in 2002 has it been clearly demonstrated that human MSC (i) could inhibit proliferation of T cells that had either been cultured in mixed lymphocyte reactions (MLR) or stimulated by polyclonal activators 13, and (ii) upon in vivo infusion, could prolong skin engraftment in non-human primates 12. Krampera et al. further demonstrated that the anti-proliferative activity of murine MSC was also exerted on antigen-specific responses 42. Suppression of T cell proliferation did not require MHC restriction, but could also be mediated by allogeneic MSC 42, 43. Results confirming the in vivo immunosuppressive properties of MSC have been reported by Djouad et al., who demonstrated that growth of an allogeneic melanoma was significantly enhanced when the latter cells were co-transplanted together with the murine C3H10T1/2 MSC line 44. Despite the general consensus concerning the capacity of MSC to inhibit proliferation of T lymphocytes, little is known on the molecular mechanism(s) responsible for this effect. Thus, it did not appear to be dependent on the induction of apoptosis of proliferating cells 13, 16, 39. Inhibition of cell division could be a possible explanation since accumulation of cells in the G0 phase of the cell cycle has been detected 45. In addition, suppression of T cell proliferation induced by human MSC appears to depend, at least in part, on the cross-talk between the two cell populations, leading to the production of inflammatory cytokines such as IFN-γ 39 and IL-1β 37 by activated immune cells. Inhibition of T cell proliferation resulted in the decreased production of effector Th1 cytokines 16, 42, 46; however, the expression of activation markers in T cells that have been cultured with MSC remains a controversial issue 39, 43, 45, 47. Inhibition of T cell proliferation by MSC appears to be subsequent both to cell-to-cell interaction and to the release of soluble factors, and it is conceivable that discrepant findings reported in the literature reflect differences in the experimental conditions used 13, 14, 16, 48, 49. TGF-β1 and HGF 13, indoleamine 2,3-dioxygenase (IDO) 50 and prostaglandin E2 (PGE-2) 46 represent MSC-derived molecules that have been proposed to exert immunomodulatory activity on T cell responses. However, most of these results need to be confirmed, and it is likely that key molecules will be identified through functional molecular profiling of the MSC transcriptome.
In vitro experiments have shown that human MSC can induce the generation of CD4+ T cell subsets displaying a regulatory phenotype (Treg) 46, 47. However, data confirming the generation of Treg cells upon MSC transplantation are still missing 16, 42.
It is interesting to note that both human and murine MSC, despite their suppressive effect on T cell proliferation, can support survival of T cells that, as a consequence of TCR overstimulation, would undergo activation-induced cell death or FAS/FASL-dependent apoptosis (Benvenuto et al., submitted for publication). An explanation for this apparent discrepancy could rely on the possibility that environmental factors (possibly related to cell division versus cell death) may affect MSC functions 51.
MSC and B cells
Few studies have addressed the effects of BM-derived MSC on B lymphocyte function. In mouse experiments, B lymphocytes were enriched from splenocyte suspensions and activated in vitro using T cell-dependent stimuli. In this context, Glennie et al. used an anti-CD40 mAb and IL-4 45, while Augello et al. stimulated splenic B cells with pokeweed mitogen 49. Both studies reached the same conclusion, i.e. that MSC inhibit B lymphocyte proliferation; a clue to the underlying molecular mechanism(s) was the observation that B cell inhibition was in part due to the physical contact between MSC and B cells and in part to soluble factors released by MSC in culture supernatants.
Deng et al. have investigated the effects of allogeneic murine MSC on T and B cells from BXSB mice, which is considered an experimental model for human systemic lupus erythematosus (SLE) 52. These authors showed that allogeneic MSC inhibited B cell activation, proliferation and IgG secretion; in addition, MSC enhanced CD40 expression and CD40 ligand ectopic hyper-expression on BXSB-derived B cells 52.
Studies on human B lymphocytes yielded different results that may, at least in part, be accounted for by different experimental conditions. In the frame of a wide study design addressing the influence of human BM-derived MSC on T, NK and B cell function, Krampera et al. showed that MSC per se did not inhibit the in vitro proliferation of B cells stimulated with a CpG-containing oligonucleotide that targets TLR9 in the presence of allogeneic, T cell-depleted mononuclear cells. However, addition of exogenous IFN-γ to the cultures significantly reduced B cell [3H]thymidine incorporation through the induction of immunosuppressive IDO in MSC 39.
Corcione et al. carried out a B cell-focused study in which B lymphocytes were stimulated with anti-Ig antibodies, soluble CD40 ligand and cytokines (IL-2 and IL-4). Proliferation of activated B cells (which produced IFN-γ under these culture conditions) was inhibited by human MSC. Additional B cell functions were dampened by MSC. These included B cell differentiation to antibody-secreting cells and chemotaxis in response to CXCL12 and CXCL13, two chemokines that play a crucial role in B cell positioning in secondary lymphoid organs. Inhibition of B cell function was found to depend on MSC-derived soluble factors that were released upon their cross-talk with B cells themselves 53.
Altogether, published evidence would indicate that mouse and human B lymphocytes are amenable to MSC inhibition following both T cell-dependent and T cell-independent stimulation. The nature of the mechanism(s) involved in such inhibition has not yet been completely elucidated, although it is possible to exclude the induction of B cell apoptosis.
Follicular dendritic cells (FDC) differentiate from stromal precursors of mesenchymal origin upon interaction with lymphotoxin α1β2 produced by activated B cells. In humans, the stromal origin of FDC is supported by the following data: (i) FDC share several markers with fibroblasts; (ii) FDC-containing ectopic lymphoid follicles are found in autoimmune, infectious or malignant disorders, consistent with the ubiquitous presence of stromal precursors of FDC; and (iii) some fibroblastic cell lines exert typical FDC-like functions 54. Preliminary results from our group (Corcione et al., unpublished results) suggest that MSC exert an anti-apoptotic effect on tonsil germinal center B cells similar to the one mediated by FDC. These findings, together with the data reported on MSC-mediated rescue of T cells from apoptosis, support the notion that modulation of B cell function by human MSC may reflect cell differentiation (e.g. into FDC) and/or may be the result of microenvironmental signals.
MSC and NK cells
NK cells are major effectors of the innate arm of the immune system that play a key role in the elimination of virus-infected and malignant cells 55–57. As for T cells and B cells, NK cell progenitors originate in the BM where they establish close interactions with mesenchymal stromal cells.
Interactions between human MSC and NK cells have been addressed in a few studies that have concordantly demonstrated that IL-2- or IL-15-driven NK proliferation is inhibited by MSC 39, 58, 59. The effect of MSC on NK cell-mediated cytotoxicity is more controversial, but this is likely related to differences in the experimental approaches. Thus, when freshly isolated NK cells were tested for their ability to kill allogeneic, HLA class I-negative or -positive targets in the presence of MSC, no inhibition of NK-mediated lysis could be detected 58. On the other hand, NK cells cultured for 4–5 days with IL-2 in the presence of MSC were less efficient in killing K562 cells than NK cells not exposed to MSC 39. Finally, lysis of HLA class I-negative targets by IL-2-activated NK cells was unaffected by short-term exposure of NK cells to MSC, whereas under the same conditions cytotoxicity against HLA class I-positive tumor cells was significantly decreased 58. However, these finding are likely to simply reflect differences in the sensitivity of target cells to NK-mediated lysis. Thus, a partial inhibitory effect exerted by MSC on the NK cell cytolytic potential could only be appreciated against target cells (HLA-class I positive) that are less susceptible to NK-mediated lysis than HLA-class I-negative ones.
The effect of MSC on NK cell cytokine production has been assessed in three independent studies. Sotiropoulou and coworkers reported that IL-2-activated NK cells produced significantly reduced amounts of IL-15-induced cytokines (IFN-γ and, to a lesser extent, IL-10 and TNF-α) following culture with MSC 58. Spaggiari and coworkers demonstrated that IL-2-activated NK cells produced IFN-γ upon short-term incubation with MSC 59. Poggi et al. showed that freshly isolated NK cells expressed the CD69 activation marker and released IFN-γ and TNF-α following incubation with MSC. Interaction between ICAM-1 on the latter cells and LFA-1 on NK cells was instrumental for cytokine production to take place. In addition, the NKp30 natural cytotoxicity receptor was found to be involved in NK cell triggering of cytolytic activity or cytokine production upon binding to MSC 59, 60.
Consistent with the low level of surface expression of HLA class I molecules, MSC were highly susceptible to lysis by IL-2-activated NK cells; this was true for both autologous and allogeneic MSC 58–60. The activating NK cell receptors NKp30, NKG2D and DNAM-1 played a major role in the NK cell-mediated cytotoxicity against MSC, which in turn expressed (known) ligands recognized by these receptors, including ULBP, PVR and nectin-2 61. Incubation of MSC with IFN-γ protected, in part, the latter cells from NK cell-mediated lysis due to HLA class I up-regulation on the former cells 59. Similarly to what was observed for T cells, soluble factors such as TGF-β1 and PGE-2 have been suggested to play a role in the MSC-mediated suppression of NK cell proliferation 58.
Taken together, the published studies on NK cell-MSC interactions reiterate a concept that has been discussed throughout this review, i.e. that the final outcome of the interplay between MSC and other cell types in vivo will be primarily dictated by the microenvironment. Thus, for example, while NK cells can lyse MSC under steady-state conditions, it is possible that the IFN-γ production by NK cells interacting with MSC might lead to in vivo protection of MSC from NK cell-mediated lysis.
Whatever the outcome of the physiological interactions between MSC and NK cells would be, the reciprocal effects exerted by the two cell types, in particular the ability of activated NK cells to kill MSC, should be carefully taken into consideration in novel strategies of adoptive immunotherapy associated with BM transplantation in which MSC may be infused to improve engraftment and/or to suppress GVHD while NK cells may be given to better eradicate leukemic cells. In order to avoid undesired killing of MSC, NK cells should not be infused shortly after or simultaneously with MSC, but after an appropriate time interval allowing MSC to express HLA class I antigens and undergo differentiation to BM stromal cells and other tissues.
MSC and dendritic cells
The modulatory effect of MSC on immune responses can also rely on the capability of altering dendritic cell (DC) function, thus resulting in the generation of tolerogenic APC. Indeed, MSC have been shown to inhibit the maturation of monocyte-derived myeloid DC by down-regulating both the surface expression of CD11c, CD83, MHC class II and costimulatory molecules, and the production of IL-12 upon TLR-mediated DC activation 47, 62–64. Following interaction with MSC, myeloid DC have been shown to produce a decreased amount of TNF-α while plasmacytoid DC switched to an increased production of IL-10 46. This effect, in turn, led to decreased IFN-γ production by Th1 cells, increased IL-4 secretion by Th2 cells, and an increased number of regulatory T cells 46, 47. The mechanism of MSC-induced inhibition of DC differentiation and function appears to be mediated by soluble factors, such as PGE-2, released upon cell-to-cell contact 46, 63. Similarly to what was observed in T cells, Dazzi and colleagues reported that the cell cycle in DC was arrested in the G0/G1 phase, upon interaction with MSC 65. Thus, the inhibition of T lymphocyte proliferation by MSC is not the exclusive result of a direct suppressive effect on T cells, but may also be related to an inhibitory effect on DC maturation, activation and antigen presentation.
MSC-mediated immunological effects in vivo
Despite the relatively large amounts of in vitro studies addressing the effect of MSC on T cells, limited information is available on in vivo studies. On the other hand, the MSC capability of facilitating the engraftment of transplanted HSC has been investigated in more detail 11, 66. In a non-human primate model of organ transplantation, MSC delayed the T cell-mediated rejection of skin allografts 12. In addition, as mentioned, co-injection of murine MSC with melanoma cells was found to favor tumor growth in an allogeneic mouse model 44. Maitra et al. reported that allogeneic human MSC support the engraftment of unrelated HSC upon transplantation and also suppress the in vivo T cell proliferation 67.
As recently reported, i.v. injection of MSC leads to a substantial attenuation of experimental autoimmune encephalomyelitis (EAE), a model of human multiple sclerosis, through the induction of peripheral T cell tolerance against the pathogenic antigen 16. MSC from BALB/c mice modulated many in vitro functions of T and B cells obtained from BXSB mice, a model of SLE 52. In contrast, in another model of autoimmunity, namely the collagen-induced arthritis, the administration of a mesenchymal stem cell line was not beneficial 68. Conflicting results have also been reported regarding the effect of MSC on GVHD. In particular, while in humans MSC i.v. infusion was found to be therapeutically effective in some patients with GVHD refractory to standard treatments 15, in mice MSC failed to prevent GVHD 69.
In general, allogeneic MSC, when infused in MHC-mismatched recipients (even with no immunosuppression), home to a wide range of tissues and persist in the host 36, 70. Notably, the i.v. infusion of allogeneic MSC in humans has led to encouraging results in different human diseases including GVHD 15, breast cancer 71, osteogenesis imperfecta 72, metachromatic leukodystrophy and Hurler syndrome 73, hematological malignancies 74 and stroke 75. However, recent in vivo experiments raised some doubts about the reported immunoprivilege of MSC, indicating that at least in some cases, administration of allogeneic MSC into an MHC-mismatched host may result in their rejection 19. Similarly, infusion of allogeneic MSC in mice receiving an allogeneic BM transplantation did not prevent BM rejection as efficiently as in mice infused with autologous MSC 18.
The current body of data suggests that MSC may have a wide range of clinical applications, including tissue repair, drug or gene delivery to diseased tissues or organs, improvement of BM engraftment and inhibition of GVHD. Data discussed in this review underline how MSC can also exert a profound effect on immune responses, primarily through the inhibition of effector functions (Fig. 1), thus offering a promising option for treating immune-mediated disorders including not only GVHD, but also autoimmune diseases and organ transplantation. However, available data also suggest that the immunological outcome resulting from the in vivo administration of MSC is not fully predictable, as it may depend on environmental factors shaping their functional properties. Therefore, in spite of a number of promising results, the clinical use of MSC in immune-mediated diseases requires thorough pre-clinical data and should take into consideration disease-specific immunological mechanisms as well as environmental factors that are likely to significantly affect the biological properties of MSC.
Some studies reported in this review article were supported by grants from the Italian Foundation for Multiple Sclerosis, Istituto Superiore di Sanità (National Program on Stem Cells), the FONDAZIONE CARIGE, Ministero della Salute (Ricerca Finalizzata Ministeriale 2005 "Caratterizzazione delle proprietà di immunomodulazione delle cellule staminali mesenchimali e possible applicazione nel trattamento delle malattie autoimmuni") and Associazione Italiana per la Ricerca sul Cancro (AIRC).