Mesenchymal stem cell effects on T-cell effector pathways

Mesenchymal stem (stromal) cells (MSCs) are rare, multipotent progenitor cells that can be isolated and expanded from bone marrow and other tissues. Strikingly, MSCs modulate the functions of immune cells, including T cells, B cells, natural killer cells, monocyte/macrophages, dendritic cells, and neutrophils. T cells, activated to perform a range of different effector functions, are the primary mediators of many autoimmune and inflammatory diseases as well as of transplant rejection and graft-versus-host disease. Well-defined T-cell effector phenotypes include the CD4+ (T helper cell) subsets Th1, Th2, and Th17 cells and cytotoxic T lymphocytes derived from antigen-specific activation of naïve CD8+ precursors. In addition, naturally occurring and induced regulatory T cells (Treg) represent CD4+ and CD8+ T-cell phenotypes that potently suppress effector T cells to prevent autoimmunity, maintain self-tolerance, and limit inflammatory tissue injury. Many immune-mediated diseases entail an imbalance between Treg and effector T cells of one or more phenotypes. MSCs broadly suppress T-cell activation and proliferation in vitro via a plethora of soluble and cell contact-dependent mediators. These mediators may act directly upon T cells or indirectly via modulation of antigen-presenting cells and other accessory cells. MSC administration has also been shown to be variably associated with beneficial effects in autoimmune and transplant models as well as in several human clinical trials. In a small number of studies, however, MSC administration has been found to aggravate T cell-mediated tissue injury. The multiple effects of MSCs on cellular immunity may reflect their diverse influences on the different T-cell effector subpopulations and their capacity to specifically protect or induce Treg populations. In this review, we focus on findings from the recent literature in which specific modulatory effects of MSCs on one or more individual effector T-cell subsets and Treg phenotypes have been examined in vitro, in relevant animal models of in vivo immunological disease, and in human subjects. We conclude that MSCs have the potential to directly or indirectly inhibit disease-associated Th1, Th2, and Th17 cells as well as cytotoxic T lymphocytes but that many key questions regarding the potency, specificity, mechanistic basis, and predictable therapeutic value of these modulatory effects remain unanswered.

allogeneic bone marrow (BM) and solid organ transplantation may be complicated by alloantigen-specifi c T-cell immune responses, resulting in graft-versus-host disease (GvHD) or transplant rejec tion [8].
Mesenchymal stem (or stromal) cells (MSCs) are a heterogeneous population of fi broblast-like progenitor cells that may be isolated and expanded from BM, umbilical cord, fat, gingiva, and other tissues [9]. Th ey have the capacity to self-renew and diff erentiate into various mesodermal cell lineages, including adipocytes, osteocytes, and chondrocytes under controlled culture conditions [9]. In the past two decades, MSCs have garnered considerable attention for their potential use as regenerative therapeutic agents in a range of acute and chronic diseases [8][9][10][11]. Mechanistically, the benefi cial eff ects of MSC therapies have been more frequently linked to their 'trophic' (paracrine) eff ects rather than their ability to transdiff erentiate [11]. Specifi cally, MSCs are now viewed as having potent anti-infl ammatory and immune-modulating properties that, in many studies, have been shown to be associated with inhibition of eff ector T-cell activation with or without a concomitant increase in regulatory T cell (T reg ) numbers [4,6,[10][11][12]. Th e T-cell suppressive eff ects of MSCs were initially described over a decade ago [13] and have since been reported consistently for both CD4 + T helper (Th ) cells and CD8 + cytotoxic T lymphocytes (CTLs) [8,11,14]. Suppression of T cells by MSCs may be direct or may occur indirectly via modulatory eff ects on antigenpresenting cells such as dendritic cells (DCs), resulting in altered cytokine expression and impaired antigen presenta tion [15][16][17]. MSCs themselves demonstrate a lack of stimu latory capacity toward T cells [18,19]. MSCs isolated from various sources (BM, adipose tissue, and Wharton's jelly) have been reported to equally suppress proliferation of CD4 + and CD8 + T-cell subsets in a dosedependent fashion [20]. Reported roles for both cell-cell contact and release of soluble factors in MSC-mediated T-cell suppression are evident throughout the literature, and numerous candidate mediators have been reported: prostaglandin E 2 (PGE 2 ), indoleamine-2,3-dioxygenase, nitric oxide, interleukin (IL)-27, transforming growth factor-beta (TGF-β), monocyte chemotactic protein 1 (MCP-1/CCL2), human leukocyte antigen G, and intracellular adhesion molecule 1 among others [8,10-12, 17, 21-24]. Th e abundance of mediators identifi ed to date suggests that MSCs exploit diff erent immunosuppressive mechanisms under diff erent disease conditions. MSC therapy has been successful in a range of disease models and some clinical conditions known to be associated with damaging eff ector T-cell responses or failure of T regmediated counter-regulation or both [4,6,8,11,22,[25][26][27][28]. Overall, it is now very well established that MSCs exert diverse and potent modulatory eff ects on the T-cell compartment of the immune system, most of which are suppressive in nature and of potential therapeutic value. Nevertheless, some signifi cant controversies and a basic lack of information regarding the range of eff ects that MSCs have on individual T-cell eff ector subsets remain. In the remaining sections of this review, we focus on the most recent data related to MSC modulation of individual well-defi ned Th cell and CTL eff ector phenotypes in vitro and in vivo. Where possible, we emphasize the relevance of current knowledge on this topic to diseases for which MSC therapy is perceived to be benefi cial. We also highlight key gaps in our understanding and important unanswered questions that may be the subject of future studies.

Mesenchymal stem cell modulation of T helper cell subsets
Th cells are cytokine-producing CD4 + cells that recognize peptides presented to them by major histocompatibility complex (MHC) class II molecules [1][2][3]. Diff erentiation of Th cells into eff ector cells depends largely on the cytokine milieu present at the time of antigen presentation and activation [1][2][3]. In the context of this article, we will focus on reviewing recent progress (summarized in Figure 1) in understanding MSC eff ects on the welldescribed subsets Th 1, Th 2, and Th 17 cells, T reg , and CTLs with emphasis, where possible, on mechanistic and disease-specifi c in vivo studies.

T helper type 1 cells
Th 1 cell induction occurs when CD4 + T cells are activated in the presence of IL-12, interferon-gamma (IFN-γ), and IL-27 [29]. IFN-γ is the characteristic cytokine produced by Th 1 cells in addition to tumor necrosis factor (TNF). Known eff ector functions of Th 1 cells include activation and recruitment of macrophages to sites of infl ammation and induction of immunoglobulin (Ig) G2a production by B cells [2]. Th 1 cells are responsible for the clearance of intracellular pathogens and delayed-type hypersensitivity (DTH) reactions by amplifying cellular immunity [29]. DTH reactions are mediated by both Th 1 cells and CTLs [25], and, through their role as coordinators of this form of immune response, Th 1 cells have the capacity to cause maladaptive tissue damage. Examples of Th 1 cell-mediated infl ammatory and autoimmune diseases are type 1 diabetes mellitus and Crohn's disease [7]. Th e literature to date indicates that MSCs exert primarily suppressive eff ects on Th 1 cell diff erentiation and eff ector function, and evidence favors predominantly indirect mechanisms. In vitro, the generation of Th 1 cells is reduced in mixed lymphocyte cultures containing MSCs or MSC-conditioned medium, likely due to inhibition of Th 1 cellstimulating properties of DCs [30]. In the in vivo setting, Lim and colleagues [25] recently demonstrated that infusion of MSCs attenuated cutaneous DTH in mice and that this eff ect was associated with reduced infi ltration of CD4 + and CD8 + T cells at the challenge site and increased apoptosis of activated T cells in the draining lymph nodes. MSCs were detected close to the germinal center and paracortical region in lymph nodes [25], suggesting that they modulate immune responses directly in the area where DCs are likely to activate T cells. In experimental colitis (an animal model of infl am matory bowel disease), dose-dependent xenogenic, allogenic, and autologous adipose-derived MSCs amelior ated disease activity and were specifi cally asso ciated with reduced IFN-γ-pro ducing Th 1 cells in association with increased numbers of forkhead box P3 (FOXP3)-expressing T cells (T reg ) [7]. Furthermore, when treated with either total CD4 + T cells or T reg -depleted CD4 + T cells from mesenteric lymph nodes of MSC-treated colitic mice, mice with induced colitis demonstrated attenuated and enhanced colitis, respec tively [7]. Th ese data suggest that the administration of MSCs in experimental colitis dampens Th 1 cell responses via induction of T reg but does not eliminate Th 1 cells entirely. MSCs have also been used experimentally to eff ectively prevent or treat Th 1 cell-mediated autoimmune diabetes mellitus in streptozo tocin-treated rats and in nonobese diabetic (NOD) mice [31,32]. In the rat model, the protective eff ects observed following administration of MSCs were shown to be associated with increased IL-10 and IL-13 expression by T cells and with increased frequencies of both CD4 + and CD8 + FOXP3 + T cells as opposed to a direct reduction of IFN-γproducing T cells [31]. In NOD mice, a single MSC injection minimized beta-cell destruction following Bidirectional arrows indicate reported inter-conversion (plasticity) between Th1/Th17 phenotypes and Th17/iT reg phenotypes that may be of relevance to MSC immune modulatory eff ects. Relevant references are indicated numerically for individual statements. CTL, cytotoxic T lymphocyte; DC, dendritic cell; DTH, delayed-type hypersensitivity; FOXP3, forkhead box P3 transcription factor; GvHD, graft-versus-host disease; IFN-γ, interferon-gamma; IL, interleukin; iT reg , induced regulatory T cell; nT reg , natural regulatory T cell; Th1, T helper type 1 cell; Th2, T helper type 2 cell; Th17, T helper type 17 cell; T reg , regulatory T cell. Suppressive effect regardless of timing of interaction 39,40 Enhanced Th17 with late interaction. 39,40,43 Multiple potential mediators of suppression 21,[36][37][38] allergic and autoimmune disease. 6,33 Th2 inhibition via T reg induction. 6 cultures. 5,48,49 Inhibited proliferation and target cell lysis. 19,48,49 Low susceptibility of allo-MSC to lysis by CTL 19

CD4 + Naïve
Increased IL-10-secreting, regulatory T-cells in multiple models of autoimmune diseases and allotransplantation. 7 Limited or absent suppression of lysis by viral antigen-specific, memory CTL in vitro and 5 49 Th17 responses in relevant culture and disease models. 6,7,30,31 Induction of CD8 + regulatory T-cells demonstrated in human allo-antigen-stimulated cultures. 50 Potential for protective effect on cancer cells via T

ASE ANCE
T-cells. 24 Inhibition of IL-17/IFNγ co-secretors. 40 allotransplant models. 4,18,21,34 in vivo. 5 transfer of diabetogenic T cells. Protection was shown to be associated with MSC migration to pancreatic lymph nodes and with induction of IL-10-producing FOXP3 + T reg [32]. Th ese examples from the recent literature indicate that, in clinically relevant disease settings, MSCs consistently suppress harmful autoimmune Th 1 cell responses by predominantly indirect mechanisms, includ ing modulation of antigen-presenting DCs and promotion of naturally occurring or induced FOXP3expressing T reg .

T helper type 2 cells
Th 2 cell diff erentiation occurs when CD4 + T cells are activated in the presence of IL-4, which itself is produced by Th 2 cells in addition to IL-5, IL-9, IL-10, and IL-13 [2,3,29]. Th e role of Th 2 cells in adaptive immunity is linked to host defense against extracellular parasites, to antibody class switching to IgG1 and IgE in B cells, and to recruitment of eosinophils [2,3,29,33]. Dysregulated Th 2 cell responses are associated with allergic diseases such as asthma [2]. Very few studies have examined MSC eff ects on immune-mediated diseases in which Th 2 cell responses are dominant. However, Kavanagh and Mahon [6] recently reported that allogeneic MSC administration reduced the number of infi ltrating eosinophils, suppressed IgE induction, and inhibited IL-13 and IL-4 production in a mouse model of ovalbumin-induced airway infl am mation. Additionally, increases in IL-10 and FOXP3 expression were observed in this study, suggesting that MSCs suppress allergen-specifi c Th 2 cell responses in allergic airway infl ammation in part via induction of T reg . As further evidence of this, depletion of T reg resulted in reversal of the protective eff ects of the MSCs [6]. In human subjects with chronic GvHD (which is also characterized by predominant Th 2 cell activity), MSC infusion has been reported to result in clinical improvement with a reduction in IL-4-and IL-10-producing T cells and a concomitant increase in IL-2-and IFN-γproducing cells [34].
In other circumstances, there is evidence that MSCs may favor the emergence of Th 2 phenotype T cells. Bai and colleagues [4] demonstrated, for example, that mice treated with human BM-derived MSCs recovered function from limb paralysis in relapsing-remitting and chronic experimental allergic encephalomyelitis (EAE), a model for multiple sclerosis, by induction of Th 2 cells. In this study, neurological improvement was associated with reduced CD45 + leukocytic infi ltration of the brain and spinal cord, with increased levels of the Th 2 cell-related cytokines IL-4 and IL-5, and with potent reduction in the Th 1/Th 17 cell-related cytokines IL-17, IFN-γ, TNF, and IL-12 [4]. Th e results suggested that MSC administration in EAE favorably altered the balance between proinfl ammatory Th 1/Th 17 cell and anti-infl ammatory Th 2 cell responses. Likewise, Fiorina and colleagues [35] reported a shift in Th 1/Th 2 cell balance toward Th 2 cells following allogeneic MSC administration to NOD mice. Further evidence supporting a shift toward Th 2 cell responses is provided by Batten and colleagues [18], who describe the use of human BM-derived MSCs for tissue engineering of a heart valve. CD4 + T cells co-cultured with MSCs expressed lower levels of IL-1-α and -β, TNF, and IFN-γ but higher levels of IL-5, IL-8, and IL-3 in response to allogeneic peripheral blood mononuclear cells. Consistent with the fi ndings of Kavanagh and Mahon [6], those of Batten and colleagues [18] additionally indicated increased expression of FOXP3 in CD4 + T cells cocultured with MSCs, suggesting the induction of a T reg phenotype. Th us, though relatively limited, the experimental evidence to date suggests that MSCs suppress eff ector function of Th 2 cells in Th 2 cell-predominant infl ammation. In other T cell-mediated immunological disorders, however, predominant MSC suppression of the Th 1 and Th 17 cell pathways may result in a relative skewing toward less damaging Th 2 and T reg phenotypes. Whether MSCs actively induce the diff erentiation and expansion of Th 2 cells during primary or secondary antigen-specifi c immune responses has not been well tested but appears less likely.

T helper type 17 cells
Th e Th 17 cell eff ector phenotype is defi ned by prefer ential secretion of IL-17A (IL-17) along with other cytokines, including IL-17F, IL-21, and IL-22. Th 17 cells are pro-infl ammatory and protect against extracellular pathogens, including fungi, mycobacteria, and Gram-negative bacteria, via recruitment of neutrophils [36]. Th 17 cells may also be pathogenic and have been shown to have an important role in immunological diseases, including rheumatoid arthritis, multiple sclerosis, and infl ammatory bowel disease [36]. TGF-β and IL-6, with or without IL-21, IL-23, and IL-1, are necessary for the induction and expansion of Th 17 cells from naïve CD4 + precursors [2,36]. Recently, MSC eff ects on the Th 17 cell diff erentiation pathway have been examined in mice and humans. Ghannam and colleagues [37] observed that human MSCs induce regulatory characteristics in Th 17 cells in an infl ammatory environment by downregulating the Th 17 cell-specifi c transcription factor RORγt (retinoidacid receptor-related orphan receptor gamma t) and upregulating FOXP3. Moreover, when re-purifi ed, these regulatory-phenotype Th 17 cells suppressed proliferation of newly initiated CD4 + T cells [37]. In vivo, MSC administration has been shown to suppress the development of EAE via a reduction in IL-17 production in the central nervous system along with reduced IFN-γ, TNF, and IL-23 and increased TGF-β and IL-4 [22]. reported by Zappia and colleagues [38] and Rafei and colleagues [39], although the studies to date have identifi ed diff erent mechanisms for the MSC anti-Th 17 cell eff ect, including IL-27 [22], alternatively cleaved MCP-1 [39], and induction of a state of T-cell anergy [38]. In our own hands, MSCs potently suppress the in vitro diff erentiation and re-activation of mouse Th 17 cells derived from naïve and memory precursors via cyclooxygenase 2 upregulation and PGE 2 production (MM Duff y, R Ceredig, and MD Griffi n, unpublished work).

Inhibition of Th 17 cell activity in EAE has also been
Although these studies indicate that MSCs have the potential to suppress Th 17 cell-mediated immunity and may do so by several mechanisms, some evidence for a Th 17 cell-promoting eff ect of MSCs also exists. For example, Carrión and colleagues [40] observed that MSCs promoted Th 17 cells while inhibiting Th 1 cells in vitro if their addition to mouse T-cell diff erentiation cultures was delayed by 3 days. Similarly, Darlington and colleagues [41] observed that MSC-conditioned medium suppressed human Th 1 cells in vitro while having an opposing eff ect on Th 17 cells. In the same study, MSCconditioned medium was found to reduce numbers of IL-17/IFN-γ double-expressing CD4 + T cells; this fi nding may have clinical implications for patients with multiple sclerosis as this subset was recently described in immune-mediated demyelinating disease [41]. Whether MSCs inhibit or enhance disease-associated Th 17 cells in vivo is less well understood, although Ghannam and colleagues [37] observed that MSCs suppressed the produc tion of IL-17 and IL-22 by established human Th 17 cell clones with a paradoxical increase in IL-10producing cells. Furthermore, Rafei and colleagues [39] demon strated amelioration of EAE and inhibition of Th 17 cell activity when MSCs were fi rst administered 1 week after the onset of neurological signs of disease, suggesting the inhibition of established T-cell eff ector responses. Inhibition of Th 17 cell-mediated infl ammation and autoimmunity by MSC administration has also been reported in models of type 1 diabetes mellitus, collageninduced arthritis, and experimental autoimmune myasthenia gravis in association with shifts toward increased Th 2 or T reg activity or both [23,42,43]. Overall, a significant amount of evidence for specifi c eff ects of MSCs on the Th 17 cell eff ector pathway is emerging. Th ese eff ects would appear to be suppressive under diverse conditions but with the potential to enhance Th 17 cell activity under some circumstances. In this regard, it is worth noting that MSCs may act as a source of IL-6, which is one of the primary mediators of Th 17 cell diff erentiation [44]. Th e role of MSC-produced IL-6 is likely to be more complex, however, as MSCs derived from IL-6-defi cient mice were less eff ective than wild-type MSCs in suppressing infl ammation associated with collagen-induced arthritis in a study by Bouffi and colleagues [42].

Regulatory T cells
A subset of CD4 + T cells have been identifi ed as having regulatory (suppressor) functions that are essential for the prevention of autoimmunity and the resolution of infl am matory processes. Th ese CD4 + T reg are best characterized by surface expression of the IL-2 receptor alpha chain (CD25) and, more specifi cally, by intracellular expression of the transcription factor FOXP3. Th ey can be further subdivided into naturally occurring T reg (nT reg ) that develop in the thymus or induced T reg (iT reg ) that diff erentiate from naïve peripheral CD4 + T cells in the presence of TGF-β [2]. T reg exert potent immuno suppressive eff ects via cell-cell contact and production of soluble factors and may negatively regulate the activation of each of the major Th cell subtypes as well as other immune and infl ammatory cells [2]. As indicated at several points in the preceding sec tions, there has been a consistent theme among many in vitro and in vivo studies in support of MSC enhancement of T reg number and activity [8,10]. English and colleagues [17] showed that human FOXP3 + CD25 high T reg were induced upon coculture of allogeneic MSCs and CD4 + T cells and exerted suppressive activity when re-purifi ed and added to a newly initiated mixed lymphocyte culture. Th is was corroborated in a study of human adipose tissue-derived MSCs that, in addition to reducing IL-17, TNF, and IFN-γ production, induced IL-10-producing, FOXP3 + T reg in vitro among collagen-specifi c peripheral blood T cells of patients with rheumatoid arthritis [45]. Upon reisolation, the T reg originally generated in the presence of MSCs had the capacity to inhibit IFN-γ production and proliferation of a subsequent collagen-stimulated T-cell culture [45]. In in vivo models of kidney, liver, and heart allo trans plantation, several laboratories have linked the protective eff ects observed with MSC therapy directly to the presence of T reg [26,27,46]. Th e importance of the induced T reg populations in such transplant models, as well as in some models of allergic and autoimmune disease, is well illustrated by the induction of graft rejection or loss of therapeutic benefi t following T reg depletion [6,8,26]. One concern that requires further investigation regarding MSC-induced T reg relates to the potential for phenotypic plasticity of pro-and antiinfl ammatory CD4 + T-cell subsets under varying in vivo conditions. For example, as highlighted in a review by Afzali and colleagues [47], T reg may be converted to a Th 17 cell phenotype when exposed to infl ammatory stimuli. In such circumstances, MSC-induced T reg may exacerbate the disease state. Further human studies will be essential to fully elucidate the clinical relevance and robustness of MSC-induced T reg in vivo, as evidenced by the study of Carrión and colleagues [48] in which the presence of MSC-induced T reg did not alter the disease course in two patients with systemic lupus erythematosus.

Mesenchymal stem cell modulation of cytotoxic T lymphocytes
CD8 + CTLs recognize cytosolic antigen-derived peptides presented by MHC class I and thus are essential for the destruction of virus-infected cells and tumor cells. Follow ing their primary activation by professional antigen-presenting cells such as DCs, CTLs induce cell death upon secondary encounter of antigen expressed by any cell type via pro-apoptotic surface receptors or targeted release of cytotoxic granules [3,19]. CTLs also release IFN-γ, TNF, and lymphotoxin-α to inhibit viral replication and to recruit macrophages to the site of infection [3]. Death receptor ligands such as TRAIL (TNF-related apoptosis-inducing ligand) and Fas ligand are upregulated on activated CTLs. In patients receiving allogeneic BM or hematopoietic stem cell transplants, activated CTLs mediate GvHD, causing damage to liver, intestine, skin, and other tissues [19]. Th erapeutically administered MSCs have the potential to reduce disease severity in GvHD and other immune-mediated diseases via direct eff ects on CTLs as well as through inhibition of Th cell responses, which are required for full activation of CTLs [3]. For allogeneic MSC administration, the benefi cial eff ects may also be limited by alloantigen-specifi c CTL-mediated MSC lysis. In a study by Rasmusson and colleagues [19], however, MSCs were resistant to CTL lysis despite the expression of MHC class I on their surface. MSCs were also unable to induce pro-infl ammatory cytokine production or CD25 upregulation by CTLs [19]. Th is same group and others have also demonstrated that MSCs inhibit the formation of CTLs in mixed lymphocyte cultures and prevent CTL-associated lysis of target cells if added during the primary stimulation phase [49,50]. In contrast, MSCs were unable to suppress activated CTLs at the cytotoxic eff ector phase [50]. Such fi ndings were corroborated by Karlsson and colleagues [5], who showed that, while MSCs potently suppressed primary alloantigen-induced proliferation and IFN-γ produc tion by human peripheral blood leukocytes, they had no eff ect on cytomegalovirus (CMV)-induced prolifera tion or IFN-γ production. Furthermore, MSCs were unable to suppress proliferation or cytolytic killing in established CMV-or Epstein-Barr virus-specifi c CTL lines [5]. Th ese observations are of particular clinical rele vance to the treatment of GvHD patients who are at high risk from the reactivation of viral infections. However, in this case, administration of MSCs was ineff ective toward CMV-mediated CD8 + T-cell eff ector functions while potently suppressing alloantigen-induced responses [5].
Interestingly, some of the protective eff ects of MSCs in GvHD may also result from the generation of CD8 + T reg , as demonstrated by Prevosto and colleagues [51]. It is proposed that CD8 + T reg may amplify the immune modulatory eff ects of MSCs because, when re-purifi ed from peripheral blood leukocytes/MSC co-cultures, these cells potently suppressed subsequent peripheral blood leukocyte proliferation in response to alloantigen and to the non-specifi c mitogen phytohaemagglutinin [51]. It is also important to note that MSC-associated immune modulation may have detrimental eff ects in the setting of cancer. In a recent study by Patel and colleagues [52], the addition of MSCs to co-cultures of breast cancer cells and peripheral blood leukocytes resulted in enhanced T reg numbers and Th 2 cell-related cytokines as well as inhibited proliferation and release of granzyme B by CTLs, all of which resulted in protection of cancer cells from immune-mediated lysis. In vivo studies to examine this phenomenon are essential to fully understand the complex interaction between MSCs, T cells, and cancer cells and to ensure that MSC administration is not associated with recurrence or rapid metastasis of cancer in some patient groups. Overall, the experimental and clinical evidence to date suggests that MSCs exert both direct and indirect suppressive eff ects on the generation of antigen-specifi c CTLs and may foster the emergence of CD8 + T reg but do not signifi cantly inhibit the immune surveillance functions of pre-existing CD8 + memory T cells.

Concluding remarks and future directions
Th e immune suppressive and anti-infl ammatory properties of MSCs are now very well established and clearly encompass potent modulatory infl uences on the generation and disease-associated activity of multiple T-cell eff ector phenotypes [8][9][10][11][12]. Preclinical models provide a strong impetus for translating MSC therapy to widespread clinical use for a range of common, T-cellmediated autoimmune diseases and for prevention or treatment of transplant complications such as rejection and GvHD [8,10]. Despite this, a critical review of our current understanding of these eff ects and of recent developments in MSC clinical trials [28] indicates that much remains to be learned at both the mechanistic and logistic levels. Th e dizzying array of potential MSCassociated mediators of T-cell suppression, the many diff erences between small-animal immunological models and human immune-mediated diseases, and the lack of uniformity in MSC culture and administration protocols suggest that a more focused experimental pipeline will be required for the therapeutic potential to be realized in the near future. Table 1 summarizes a number of key questions that, on the basis of the literature to date, we believe to be important for translational progress in this fi eld. In particular, we would highlight the need to better understand conditions in which MSC administration has been found to be ineff ective or even harmful during T cell-mediated disease. Zappia and colleagues [38] demonstrated, for example, that the time of administration of MSCs was a critical parameter for successful treatment of EAE since MSC administration prior to or during the early disease course was eff ective whereas MSC benefi t was lost once central nervous system infl ammation was fully established. Similarly, in experi mental arthritis, the timing of MSC administration, the relative eff ects of MSCs on diff erent Th cell subsets, and the local joint conditions have been reported to critically determine the balance between therapeutic effi cacy, lack of benefi t, and detrimental eff ects [40,42,44,53]. Th ese studies should in no way dampen enthusiasm for further preclinical and clinical applications of MSCs in disease conditions in which one or more T-cell eff ector pathways are known to be the primary cause for acute or chronic tissue damage. Rather, they highlight the complexity of the interactions that occur between stromal cells and cells of the immune system and the wealth of basic and therapeutic insights that can be gained from continued investigation of these interactions.