Regulatory immune cells in transplantation

Immune regulation is fundamental to any immune response to ensure that it is appropriate for the perceived threat to the host. Following cell and organ transplantation, it is essential to control both the innate immune response triggered by the injured tissue and the adaptive immune response stimulated by mismatched donor and recipient histocompatibility antigens to enable the transplant to survive and function. Here, we discuss the leukocyte populations that can promote immune tolerance after cell or solid-organ transplantation. Such populations include regulatory T cells, B cells and macrophages, as well as myeloid-derived suppressor cells, dendritic cells and mesenchymal stromal cells. We consider the potential of these regulatory immune cells to develop and function in transplant recipients and their potential use as cellular therapies to promote long-term graft function.

Although leukocyte populations, including macro phages, dendritic cells (DCs), B cells and T cells, can participate in the destruction of transplanted cells or organs, they can also promote tolerogenic immune responses and the longterm survival of the graft 1 . Regulatory immune cells that prevent allograft rejec tion or graft-versus-host disease (GVHD) are special ized leukocyte populations that either are selected to have regulatory function during their development or acquire immunosuppressive properties in the local microenvironment of the allograft or in the graft draining lymphoid tissues. The presence of regulatory immune cells in a transplant recipient can shift the balance towards graft survival and away from graft rejection or GVHD. Thus, regulatory immune cells have important roles in determining the longterm outcome following transplantation.
Studies of transplantation provided some of the earliest evidence for immune regulation in vivo. Data from neonatal tolerance experiments performed in the 1950s by Billingham, Brent and Medawar suggested that additional mechanisms beyond the deletion of donorreactive effector immune cells were involved in allograft acceptance 2 . Later studies showed that rats that maintained longterm allograft survival in the absence of continuous drugbased immunosuppres sion -a status referred to as operational transplanta tion tolerance -had T cells that could actively prevent graft rejection 3 . This form of cellular regulation in transplantation was later shown to be associated with CD4 + T cells that express CD25, the αchain of the interleukin2 (IL2) receptor 4-7 , consistent with data examining immune regulation and the development of autoimmunity 8 .
In clinical transplantation, lifelong immunosuppres sive drug therapy is essential to prevent rejection for most patients. Individuals with a functioning transplant in the absence of immunosuppression are very rare but have been described in four clinical situations. First, there are patients whose immunosuppressive drugs are discontinued for medical reasons, including persistent viral infections and cancer, which are both sideeffects of nonspecific immunosuppression 9 . Second, there are recipients who are noncompliant, that is patients who choose to stop taking the immuno suppressive drugs 10 . Third, some liver transplant recipients have participated in immunosuppression weaning studies 11 . Fourth, some kidney transplant recipients have been treated with tolerance induction protocols, including mixed chimerism strategies [12][13][14][15] . Data characterizing the immune status of these recipients suggest that regulatory immune cells -including regulatory T (T Reg ) cells 13,16,17 , regula tory B cells [18][19][20] and DCs -contribute to operational tolerance.
Whether T cells and other immune cell types con tribute to graft rejection and GVHD, or promote graft acceptance, depends on the origin of the cells and, importantly, the conditions that prevail in the transplant recipient. For example, although B cells can promote transplant rejection by functioning as antigenpresenting Graft-versus-host disease (GVHD). Tissue damage that occurs in a recipient of allogeneic transplanted tissue (usually a bone marrow transplant) as a result of the activity of donor cytotoxic T lymphocytes that recognize the tissue of the recipient as foreign. GVHD varies markedly in severity, but can be life threatening in severe cases. Typically, damage to the skin and gut mucosa leads to clinical manifestations.

Mixed chimerism
A state of coexistence of host and allogeneic donor haematopoietic cells.
Natural killer T cells (NKT cells). A subset of T cells expressing both NK cell and T cell markers. In mice, NKT cells were first identified by their expression of the alloantigen NK1.1 (NK cellassociated antigen 1.1) in addition to CD3. Some mouse NKT cells express an invariant T cell receptor (TCR) containing the Vα14 variable region of the TCR α-chain and recognize antigens presented on CD1d. Similarly, human NKT cells express an invariant Vα24-containing TCR. NKT cells are characterized functionally by cytolytic activity and the rapid production of cytokines, including IFNγ and IL-4. cells (APCs) or by generating alloantibodies, regula tory B cells have been identified in immunosuppressive drugfree transplant recipients and may promote graft acceptance 18,19,21 .
In this Review, we summarize our current under standing of the regulatory immune cell populations that can promote tolerance during transplantation. Furthermore, we consider the clinical potential of target ing these populations to facilitate the longterm survival and function of allografts.

CD4 + regulatory T cells. T Reg cells that can control
allograft rejection and GVHD in vivo can arise via two distinct pathways. Thymusderived or naturally occur ring T Reg cells are selected in the thymus and function in the periphery primarily to suppress responses to self antigens 23 . They are present in every healthy individual and constitutively express high levels of the transcrip tion factor forkhead box P3 (FOXP3). In addition, CD4 + T cells that encounter antigens in a tolerogenic micro environment in the periphery can differentiate into 'adaptive' or 'induced' T Reg cells that also express high levels of FOXP3 (REF. 34). In this article, we use the term 'T Reg cell' to refer to thymusderived T Reg cells and 'induced T Reg cell' to refer to the T Reg cells that develop extrathymically. The cellsurface and intra cellular markers that are expressed by these populations in humans and mice are shown in TABLE 1.
In recipients of solidorgan transplants, there are usually insufficient T Reg cells at the time of transplanta tion to prevent the rejection of a fully allogeneic graft (that is, when both MHC and minor histo compatibility antigens are mismatched), especially when memory T cells specific for donorderived allo antigens are present 35 . The high frequency of such allo antigen reactive T cells in the immune repertoire of the recipient compared with the relatively small number of T Reg cells present means that the balance is shifted towards allo graft destruction. This does not mean that T Reg cells do not function, but instead suggests that they are over whelmed by the high frequency of effector immune cells. Interestingly, the allograft itself can induce or expand T Reg cell populations that can protect it from rejection 36 , and even when the primary allograft has been rejected, cells with the characteristics of T Reg cells are found in the recipients 37,38 . This confirms that prolonged expo sure to alloantigens can lead to the generation of induced T Reg cells and/or the expansion of thymusderived T Reg cell populations, even amidst a destructive immune response. Thus, both thymusderived and induced T Reg cells contribute to the overall pool of T Reg cells that can recognize and respond to donor alloantigens 39 . However, after transplantation it is likely that induced T Reg cells generated in response to donorderived allo antigens may be more important owing to the persistent presence of donor antigens 40 .
This crucial balance between graft destruction and regulation can be shifted in several ways, either before or after transplantation, notably by using strate gies that inhibit the activity of effector T cells and/or increase the relative frequency or functional activity of alloantigenreactive T Reg cells 39,[41][42][43] . The use of immuno suppressive drugs that permit or promote T Reg cell gen eration and function is obviously one approach that can be effective 44 . Other strategies include the infusion of

Ischaemia-reperfusion injury
An injury in which the tissue first suffers from hypoxia as a result of severely decreased, or completely arrested, blood flow. The restoration of normal blood flow then triggers inflammation, which exacerbates the tissue damage.

Induction therapy
A therapy given before transplantation to promote graft acceptance.

Trogocytosis
A process whereby lymphocytes that are interacting with an antigen-presenting cell (APC) can extract cell-surface molecules expressed by the APC and display them on their own surface.
alloantigens before transplantation (to expand or induce T Reg cell populations specific for donor allo antigens) 45 and the delivery of T Reg cells as a cellular therapy 46 (discussed later). Thymusderived T Reg and induced T Reg cells use several different mechanisms to inhibit the activity of the effec tor immune cells that, in the absence of T Reg cells, trigger graft rejection or GVHD 47,48 . For example, the expres sion of cytotoxic T lymphocyte antigen 4 (CTLA4; also known as CD152) by T Reg cells can inhibit APC activity, thereby preventing the development of effector T cells. Furthermore, the binding of CTLA4 to the costimulatory molecules CD80 and CD86 on APCs is implicated in the activation of the enzyme indoleamine 2,3dioxygenase (IDO), which results in both a local deprivation of the essential amino acid tryptophan and the production of inhibitory molecules known as kynurenines, leading to attenuated T cell proliferation 49 . T Reg cells can also produce IL10 (REF. 50), which is an immunosuppressive cytokine that can inhibit APC activity and promote the conver sion of T cells into T regulatory type 1 (T R 1) cells (see below), and this is likely to be important for regulatory activity in the draining lymphoid tissue and the allograft. Indeed, blocking the activity of CTLA4 or IL10 in vivo prevents T Reg cellmediated regulation in transplantation models 6,45,51 (FIG. 1a). IL10 can also be produced by other leukocyte populations, notably DCs and regulatory B cells (see below). Transforming growth factorβ (TGFβ) 52,53 and IL35 (REF. 54) are also important for the functional activity of T Reg cells (FIGS 1,2).
For T Reg cells to function efficiently they must be in the right place at the right time during an immune response 55 . T Reg cells initially function in the drain ing lymphoid tissue to inhibit the activation of naive T cells and the development of an effectortype adaptive immune response 56 . Then, by migrating to the allograft, they reduce the impact of ischaemia-reperfusion injury 57 . Later in the response, the allograft is an important site for providing T Reg cells with continuous exposure to graft derived antigens 58 , thereby creating an environment that is permissive to graft acceptance 59,60 .
In addition to T Reg and induced T Reg cells, several CD4 + FOXP3regulatory T cell populations have been described. These include T R 1 cells, which are a distinct population of peripherally induced T Reg cells that develop in the presence of IL10 (see above). T R 1 cells regulate responses through the FOXP3independent secretion of IL10 and TGFβ, leading to bystander regulation of effector T cells 61 (FIG. 1b), and they can function to inhibit GVHD and allograft rejection 62 . It is likely thatdepending on the microenvironment created in vivo, the degree of histoincompatibility between the donor and recipient, and the specific type of tissue transplanted -different CD4 + T Reg cell populations are involved in inhibiting destructive immune responses to the allograft.

CD8 + regulatory T cells.
CD8 + T cells can suppress the activity of selfreactive T cells 29,63 as well as effector T cell responses to alloantigens 64 . In the 1980s, hypothetical suppressor cascades involving the participation of mul tiple leukocyte subsets and various secreted mediators were proposed, but none of these was validated 65,66 . Consequently, interest in CD8 + T cellmediated regu lation waned. However, the recent characterization of CD8 + CD28and IL10producing CD8 + T Reg cells has revived interest in this field. CD8 + CD28 -T cells inhibit APCmediated T cell activation by direct cell contactdependent mechanisms 67 and are most likely to be an endstage population. Cells with this pheno type have been identified in renal transplant recipients who have received alemtuzumab as an induction therapy. Alemtuzumab -a monoclonal antibody that targets CD52 + cells -results in profound leukocyte deple tion when used as an induction therapy in transplant recipients. Leukocyte repopulation occurs gradually over time in response to the lymphopenic environment created following alemtuzumab therapy, with differ ent leukocyte subsets repopulating at different rates 68 . CD8 + CD28 -T cells that are present in renal transplant recipients treated with alemtuzumab induction therapy could have a role in suppressing the immune response to donor alloantigens in the longer term 69 .
By contrast, a population of IL10producing CD8 + T Reg cells, which have been referred to as 'CD8 + T R cells' , can be generated from naive CD8 + T cells and inhibit primary T cell responses through an IL10dependent mechanism. These IL10producing CD8 + T Reg cells share many similarities with the T R 1 cells mentioned above, in that they are anergic (that is, refractory to restimulation by their cognate antigen), their generation depends on IL10 and they suppress primary T cell responses through IL10dependent mechanisms (FIG. 1c). IL10 producing CD8 + T Reg cells have been described in vivo in a patient with longterm acceptance of an allogeneic kidney transplant 70 . These findings suggest that there may be two distinct subsets of CD8 + T Reg cells, which use dis tinct mechanisms and are generated in vivo in response to different microenvironments. A kinetic analysis to deter mine the relative presence of each subset in transplant recipients treated with different immunosuppressive drug regiments is required to understand their relative roles in controlling immune responses to donor alloantigens after organ transplantation.

CD4 -CD8regulatory T cells.
A subset of double negative T cells -which do not express CD4, CD8 or NK1.1 but do express CD3 and the αβ T cell receptorhas been shown to inhibit immune responses mediated by effector CD4 + and CD8 + T cells and to prevent allo graft rejection 71,72 , GVHD 73 and diabetes development 74 . Doublenegative T cells use a variety of mechanisms to mediate suppression. For example, they kill T cells in an antigenspecific manner via the CD95-CD95L pathway, they downregulate the expression of the costimulatory molecules CD80 and CD86 by DCs, they induce DC apop tosis, and they acquire alloantigens from DCs by trogocytosis (FIG. 1c). Recently, interferonγ (IFNγ) expres sion was found to be induced in doublenegative T cells by autologous tolerogenic DCs (see below), leading to the accumulation of doublenegative T cells in the spleens of rats that were operationally tolerant to a heart allograft. Notably, blockade of IFNγ resulted in allograft rejection 75 .  Figure 1 | Mechanisms used by adaptive regulatory immune cells in transplantation. a | Naturally arising thymus-derived regulatory T (T Reg ) cells that can respond to donor alloantigens through cross-reactivity will be present in the recipient at the time of transplantation. These cells are recruited to the allograft, where they can suppress ischaemia-reperfusion injury. Moreover, in the draining lymphoid tissue, T Reg cells inhibit T cell proliferation. Regulatory B (B Reg ) cells and tolerogenic dendritic cells (DCs) can promote the development of induced T Reg cells from naive T cells. These induced T Reg cells promote tolerance to the allograft through various mechanisms, including the production of interleukin-10 (IL-10) and transforming growth factor-β (TGFβ), the inhibition of antigen-presenting cell (APC) function, and effects on amino acid availability and energy metabolism. b | T regulatory type 1 (T R 1) cells are a subset of regulatory T cells that do not express forkhead box P3 (FOXP3). They are induced in the presence of IL-10, which is produced by T Reg cell, tolerogenic DC and B Reg cell populations, either in the draining lymphoid tissue or in the allograft. T R 1 cells can suppress pro-inflammatory activities of both APCs and effector T cells. c | In the presence of IL-10, naive CD8 + T cells can be converted into CD8 + T Reg cells that function in a similar manner to T R 1 cells. CD8 + CD28cells can inhibit APC function to promote immune regulation. d | Double-negative (DN; that is, CD4 -CD8 -) T cells function by downregulating the expression of co-stimulatory molecules by DCs, thereby inhibiting the ability of DCs to stimulate pro-inflammatory immune responses and, instead, inducing the apoptosis of DCs. In addition, DN T cells can acquire alloantigens through trogocytosis. This enables them to present antigens to effector T cells in a manner that promotes T cell apoptosis. CTLA4, cytotoxic T lymphocyte antigen 4; IDO, indoleamine 2,3-dioxygenase; LAG3, lymphocyte activation gene 3; PDL1, PD1 ligand 1; TNF, tumour necrosis factor.

Graft-versus-tumour response
An immune response mounted against host tumour cells by donor T cells (mainly cytotoxic CD8 + T cells) that are derived from an allogeneic bone marrow transplant.

Plasmacytoid DC
(pDC). An immature dendritic cell (DC) with a morphology that resembles that of a plasma cell. pDCs produce large amounts of type I interferons in response to viral infection.

Transitional B cells
Transitional B cells are short-lived immature B cells, typically found in the spleen, that either die or are selected into the peripheral mature B cell repertoire. Transitional B cells can be subdivided into three subsets (T1, T2 and T3 cells) based on different phenotypical and functional characteristics.

Alternatively activated macrophages
Macrophages that are induced by T H 2-type cytokines, such as IL-4 and IL-3, and that are associated with immune responses to parasites and tissue-repair programmes.
Human doublenegative T cells share phenotypical and functional features with their mouse counterparts, including the ability to acquire peptide-MHC complexes from APCs and the ability to induce the apoptosis of antigenspecific CD8 + T cells 76 . In haematopoietic stem cell transplantation (HSCT), a deficiency of double negative T cells is associated with the occurrence of GVHD. This provides some evidence, albeit limited, that doublenegative T cells with regulatory activity can participate in the development of peripheral tolerance in humans 77 .
NKT cells. NKT cells have been implicated in both transplant rejection and tolerance 78 owing to their unique capacity for rapid and early production of pro inflammatory or antiinflammatory cytokines in response to their cognate glycolipid antigens presented on CD1 molecules. CD4 − CD8 − T cells from mouse bone marrow, the majority of which are NKT cells, were found to inhibit acute lethal GVHD by augmenting the proliferation of donorderived T Reg cells in an IL4dependent manner 79,80 . Furthermore, in a mouse model of HSCT, the adoptive transfer of highly puri fied (>95% pure) NKT cells suppressed GVHD and decreased the production of IFNγ and tumour necro sis factor (TNF) by donorderived T cells, but left the graft-versus-tumour response intact 81 .

γδ T cells. γδ T cells are nonconventional T cells that have important roles in antitumour and antiviral immune responses. An altered distribution of Vδ1 and
Vδ2expressing γδ T cell subsets has been observed in operationally tolerant liver transplant recipients com pared with agematched nontransplant controls 16,82 , but experimental data on whether γδ T cells can protect the allograft are limited. A tissuehoming population of γδ T cells has been shown to exert local regulatory activ ity in nontransplant settings 83 , but a specific functional role of γδ T cells in mediating transplant tolerance has yet to be described.

Regulatory B cells
Interest in the role of regulatory B cells in transplan tation was stimulated by clinical rather than experi mental data. Despite the absence of donorspecific allo antibodies, increased numbers of B cells and higher expression levels of B cellassociated genes, such as CD20 and CD79B, were found in immunosuppression free renal transplant recipients 18,19,21 . In mice, regula tory B cells express high levels of CD1d, CD21, CD24 and IgM and moderate levels of CD19, although some heterogeneity may exist 84 . Human regulatory B cells, classified as CD19 + CD24 hi CD38 hi , constitute a small subset of the total B cell pool and share some prop erties with their mouse counterparts, including an immature phenotype (TABLE 1). One of the character istics of B cells with regulatory activity is their ability to secrete IL10, a cytokine known to favour immune regulation, as discussed above (FIG. 1a). The ability of IL10secreting B cells to regulate an immune response was first demonstrated in autoimmunity [85][86][87] . CD40 stimulation appears to be required to stimulate IL10 production and is necessary for the activation of regu latory B cell functions 88 . CD40activated human CD19 + B cells can induce the generation of T Reg cells from allogeneic CD4 + T cells 89 , but whether this is depend ent on B cellderived IL10 remains to be clarified. In comparative studies using cells from the same donor, B cellinduced T Reg cells showed more potent sup pressive activity than plasmacytoid DC (pDC)induced T Reg cells 90 .
It remains to be clarified how regulatory B cells are generated in response to alloantigens, and also when and where they function in transplant recipi ents. In mice, the ligation of T cell immunoglobulin and mucin domaincontaining protein 1 (TIMD1; also known as HAVCR1) on B cells promotes population expansion and regulatory activity 91 , suggesting a poten tial therapeutic strategy for increasing the number of regulatory B cells in vivo. Other studies have shown that IgM + B cells, but not IgG + B cells, form clusters within kidney allografts in tolerant rats, and this find ing has been interpreted as indicating the presence of B cells with regulatory activity 92 . Our own work has investigated the evolution of regulatory B cell responses during leukocyte reconstitution in kidney transplant recipients treated with alemtuzumab. We found that regulatory B cells appear transiently in the peripheral blood following transplantation 93 , and their presence is potentially related to the later presence of T Reg cells in transplant recipients receiving induction therapy 94,95 . The link between the functions of regulatory B cells and T Reg cells in protection against allograft rejection requires further study.
Some other B cell populations may have the capacity to induce immunological unresponsiveness to specific alloantigens in vivo. Transitional B cells are poor APCs in vitro, as they exhibit little costimulatory function 96 .
In addition, small resting B cells have been shown to promote tolerance to allografts by presenting donor derived alloantigens to the host immune system in a nonstimulatory form 97 .

Regulatory macrophages
Macrophages can have both protective and pathological functions and can be divided into subgroups on the basis of their tissue location and their functional properties 98 . In transplantation, macrophage activation occurs initially as a result of the tissue injury associated with ischaemiareperfusion and can contribute to early graft damage 99 . By contrast, alternatively activated macrophages can inhibit the production of proinflammatory cytokines by classically activated macrophages and can contrib ute to wound healing and tissue repair. In organ trans plantation, this repair process is important in the early posttransplant period, as wound healing allows tissue homeostasis to be reestablished. However, later in the course of transplantation, tissue repair responses may be less desirable, as they can contribute to delayed allo graft failure by causing the occlusion of blood vessels within the allograft, a process referred to as transplant arteriosclerosis.

B-1 cells
IgM hi IgD low MAC1 + B220 low CD23 − cells that are predominantly found in the peritoneal and pleural cavities.
The size of the B-1 cell population is kept constant owing to the self-renewing capacity of these cells. B-1 cells recognize self components, as well as common bacterial antigens, and they secrete IgM antibodies that tend to be of low affinity and broad specificity.
'Regulatory macrophages' may represent an addi tional, distinct population, whose main physiological role is to dampen proinflammatory immune responses 100 . However, so far no stable convenient surface markers for regulatory macrophages have been identified. Regulatory macrophages produce large amounts of IL10 but, unlike alternatively activated macrophages, do not express arginase 1 and are not dependent on signal transducer and activator of transcription 6 (STAT6) signalling. Interactions with T Reg cells can induce macrophages to acquire the properties of alternatively activated or regula tory macrophages 101 (FIG. 2). Furthermore, the interaction of macrophages with B-1 cells results in the formation of a regulatory macrophage population 102 .
Human regulatory macrophages isolated from the peripheral blood are characterized by their morphology, their cellsurface phenotype (TABLE 1) and their ability to suppress T cell proliferation in vitro. In a pilot clini cal study, human regulatory macrophages were shown to reduce the need for immunosuppressive drugs when administered to kidney transplant recipients 103 . Host macrophages can also have a protective effect following transplantation. Indeed, reducing the pool of the host macrophages in recipient mice increased the expansion of donor T cell populations and aggravated GVHD after allogeneic HSCT 104 .
Tolerogenic dendritic cells DCs are crucial for priming antigenspecific T cell responses, including those to alloantigens 105 . However, they can also promote tolerogenic responses 1,106-108 (TABLE 1).
Initially, immature conventional myeloid DCs that express low levels of MHC class II and costimulatory molecules at their cell surface were identified as the dominant type of DC with the capacity to induce T cell tolerance 109 . Indeed, immature DCs can promote toler ance to solidorgan allografts and bone marrow grafts 106 . For example, a single injection of immature donor derived DCs 7 days before the transplantation of an MHCmismatched heart allograft extends the survival of the allograft 110 or prolongs it indefinitely 111 . Moreover, the injection of donorderived DCs prevents the rejec tion of MHCmismatched skin grafts 112 and protects recipient mice from developing lethal acute GVHD 113 . The tolerogenic effects of immature DCs can be enhanced by administering the cells together with other immuno modulatory agents, such as drugs that block the CD40-CD40L costimulatory axis 114 .
pDCs can also promote tolerance in transplanta tion 115 . In experimental models, pDCs acquired allo antigens in the allograft and then migrated to the draining lymphoid tissue, where they induced the gener ation of T Reg cells 116 . In mice, prepDCs appear to be the principal cell type that facilitates haematopoietic stem cell engraftment and the induction of donorspecific skin graft tolerance in allogeneic recipients 117 .
Higher ratios of pDCs to myeloid DCs were found in the peripheral blood of paediatric liver transplant recip ients on no or very low immunosuppression compared with those on normal maintenance doses. In the same study, a similar trend was observed between patients receiving lowdose immunosuppressive therapy dur ing prospective immunosuppressive drug weaning and patients on maintenance immunosuppression 118 . In addition, higher levels of expression of PD1 ligand 1 (PDL1) and CD86 by pDCs were found to correlate with elevated numbers of CD4 + CD25 hi FOXP3 + T Reg cells in liver transplant recipients who were free from immunosuppressive drug regimens 119 . These data suggest that pDCs may contribute to immune regulation in liver transplant recipients (FIG. 1a).
In summary, both myeloid DCs and pDCs can pro mote tolerance to alloantigens, and DC maturation in itself does not appear to be the distinguishing feature that separates immunogenic DC functions from tolerogenic ones 120 . However, despite the tolerogenic functions of DCs discussed above, the use of DCs to facilitate the induction of operational tolerance is not without risk. DCs are argu ably better known for their ability to prime the immune system. Indeed, DCs pulsed with antigens are being used clinically as vaccines to stimulate immune responses to tumour antigens. Using DCs as a cellular therapy in trans plantation may therefore carry the risk of sensitizing the recipient. One possible approach to reducing this risk is to combine DC administration with costimulatory mol ecule blockade, with the objective of presenting donor alloantigens to T cells in a manner that will induce T cell tolerance or nonresponsiveness to the allograft.

Myeloid-derived suppressor cells
Myeloidderived suppressor cells (MDSCs) are a hetero geneous population of progenitor cells that have been associated with many suppressive immune functions. MDSCs can accumulate in tissues during inflammation, and may subsequently differentiate into macrophages, DCs and granulocytes. The expansion and activation of MDSC populations are regulated by factors produced by other cells present in the same microenvironment, including stromal cells, activated T cells and, in tumours, the tumour cells themselves.
Several MDSC subsets have been described in both mice and humans 121 . Despite their heterogeneity, most MDSCs express common phenotypical markers, which include GR1 and CD11b in mice, and CD11b, CD33, CD34 and low levels of MHC class II molecules in humans (TABLE 1). Activated MDSCs suppress both the proliferation of and cytokine production by effector T cells, B cells and NK cells in vitro through various mech anisms, for example by expressing inducible nitric oxide synthase (iNOS; also known as NOS2) and arginase 1. MDSCs can also inhibit T cell proliferation and modify T cell differentiation pathways, for example by promoting T Reg cell differentiation in a process requiring IFNγ and IL10 (REF. 122) (FIG. 2). Interestingly, interactions between MDSCs and macrophages result in a shift towards an alternatively activated macrophage phenotype 123 .
MDSCs have been shown to promote tolerance to alloantigens, and there is direct evidence of a tolerogenic role for MDSCs in heart and islet allografts in mice 124,125 and for iNOSexpressing MDSCs in a rat kidney allo graft model 126 . The mechanisms used by MDSCs to promote tolerance to alloantigens require further clarification. Some evidence suggests that they may act partly through the induction or sparing of T Reg cells 126 . In addition, MDSCs have been found to upregulate their expression of haem oxygenase 1, an enzyme that inhib its DC maturation, promotes IL10 production and has cytoprotective properties 127 .

Mesenchymal stromal cells
Mesenchymal stromal cells (MSCs) are a subpopulation of multipotent cells in the bone marrow that support haemato poiesis and possess immunomodulatory and reparative properties 128 (TABLE 1). Bone marrowderived MSCs have the ability to migrate to sites of inflammation, including to an allograft 129 . When MSCs are exposed to an inflammatory microenvironment, they are capable of reg ulating many immune effector functions. Notably, MSCs have been shown to promote the generation of T Reg cells both in vitro and in vivo through mechanisms involving prostaglandin E2, TGFβ and cell-cell contact 130,131 .

Anti-thymocyte globulin
Polyclonal antibodies specific for human T cells that are produced by immunizing rabbits or horses.

Calcineurin inhibitor
An immunosuppressive drug that blocks calcineurin (a phosphatase that is necessary for the nuclear translocation of the transcription factor NFAT) and thus restricts T cell activation.
The impact of MSCs on the generation of T Reg cells may be indirect, as DCs that are repeatedly exposed to MSCs are maintained in an immaturelike state 132 . In transplantation, the retrieval and transplantation of the allograft inevitably results in ischaemia-reperfusion injury, which creates an inflammatory microenviron ment in the graft. The recruitment of MSCs to the graft in the early posttransplant period could potentially induce the generation of T Reg cells from T cells that are also recruited to the allograft (FIG. 2). In addition, MSCs may promote the acceptance of allogeneic islets of Langerhans by secreting matrix metalloproteinases 129 . There is also evidence that MSCs can suppress graft rejection by inhibiting alloantibody production 133 .
Interaction of distinct regulatory cell populations Organ and tissue retrieval inevitably results in tissue damage, and ischaemia-reperfusion injury is associ ated with the surgical procedures required to transplant a solidorgan graft. This tissue damage triggers the activation of the innate immune system, resulting in inflammation in the allograft within hours after trans plantation 1 . As already highlighted, MDSCs and MSCs migrate to the site of inflammation and are activated by the proinflammatory cytokines that are present 128 . Thus, early in the response to a transplant, these popu lations may infiltrate the allograft and exhibit immuno modulatory activity (FIG. 2). Donorderived DCs and macrophages that are resident in the allograft, as well as leukocytes that are attracted to the site of inflammation, might be influenced by the activity of either MDSCs or MSCs, which may promote some cells to develop regulatory properties. The inflammatory environment in the allograft also triggers the maturation of resident donorderived DCs and their migration to the draining lymphoid tissue, where they initiate T cell activation 134 . Importantly, the absence of secondary lymphoid tissue is sufficient to prevent the rejection of vascularized organ allografts in naive recipients 135 . Activated T cells return ing to the allograft may encounter a tolerogenic micro environment created by the presence of MDSCs and MSCs, and thus they may give rise to induced T Reg cells.
Although the numbers of thymusderived T Reg cells and regulatory B cells that can respond to donor allo antigens are relatively low before transplantation, both cell populations have the potential to migrate to the allograft, where they can contribute to modulating the immune response (FIG. 1). In addition, depending on the microenvironment, donor alloantigens presented by donorderived or recipient APCs could drive the expansion and generation of T Reg cell or regulatory B cell populations. However, it is important to remember that despite the occurrence of active regulation in the allo graft and periphery, rejection or GVHD will dominate in the absence of any other forms of immunosuppressive therapy. Thus, the regulatory immune cells that pre existed in the recipient or were generated during the course of the response are not sufficiently powerful or present in sufficient numbers early in the evolution of the response to control the high frequency of leukocytes capable of destroying the graft.
Immunosuppression and regulatory cells Most transplant recipients are treated with a combi nation of immunosuppressive drugs and biological agents to control the rejection and GVHD responses. The combination of drugs used varies depending on the type of transplant as well as on the protocols used by individual transplant programmes. In some centres, the recipients of kidney transplants may be given an induction therapy using a monoclonal antibody or poly clonal antibody preparation, such as alemtuzumab or anti-thymocyte globulin, at the time of transplantation. This treatment markedly depletes most of the leuko cyte populations in the peripheral blood and can affect the lymphoid organs, creating lymphopenia. However, not all leukocyte populations are affected to the same extent; for example, memory T cells are not depleted following alemtuzumab administration 68,136 . Leukocyte repopulation is triggered by lymphopenia, and different subsets are repopulated at different rates. Interestingly, leukocyte depletion has the potential to tip the balance in favour of immune regulation by creating a situation in which regulatory immune cells outnumber effector cells. In alemtuzumabtreated transplant recipients, there is evidence that T Reg cells and regulatory B cells are present among the repopulating leukocytes, and these cells may contribute to the prevention of rejec tion [93][94][95] . Antithymocyte globulin has also been shown to promote the generation of T Reg cells 137 .
Irrespective of whether leukocytedepletion ther apy is used, transplant recipients are treated with other immunosuppressive drugs: most commonly a calcineurin inhibitor, such as tacrolimus or cyclosporine A, and an antiproliferative agent, such as mycophenolate mofetil. Both types of drug inhibit the activation of effector T cells, which explains the rationale for their use in clinical transplantation, but they can also inhibit the generation and function of regulatory immune cells. Interestingly, T Reg cells are found in the peripheral blood of kidney transplant recipients treated with calcineurin inhibitors 138,139 , as well as in liver transplant recipients following weaning and withdrawal from treatment with tacrolimus and steroids 16 . Many transplant recipients are now also treated with a monoclonal antibody specific for CD25, the αchain of the IL2 receptor. T Reg cells express high levels of CD25 (TABLE 1), and IL2 is required for the generation and expansion of these cell populations. Thus, the impact of CD25specific antibody therapy on the generation and function of regulatory cells, in par ticular T Reg cells, is debated. Some transplant recipients are treated with rapamycin, an immunosuppressive drug that targets the mammalian target of rapamycin (mTOR) pathway and was shown to support the generation of T Reg cells ex vivo 140 and to promote their function in vivo 44 .
Thus, the precise cocktail of immunosuppressive agents used in a transplant recipient may have differential effects on both effector and regulatory immune cells.
The use of combinations of immunosuppressive agents in clinical transplantation highlights the chal lenges associated with designing protocols that include the infusion of regulatory immune cell populations. As shown by adoptive transfer studies, the infusion of Nature Reviews | Immunology

Graft-versus-leukaemia effect
The antitumour activity of donor T cells against residual leukaemic cells of the graft recipient following allogeneic bone marrow transplantation. regulatory immune cells shortly before or at the time of transplantation, or even during a graft rejection episode, clearly has the potential to inhibit the activity of effec tor cells and promote graft acceptance 40,46 . The infusion of distinct populations of regulatory cells also enhances the generation and function of host regulatory mecha nisms (FIGS 1,2). Coadministration of immunosuppres sive drugs has the potential to enhance the functional properties and the generation of regulatory immune cells, but some combinations of drugs may also inhibit immune regulation. Thus, the competing actions of each component of an immunosuppressive protocol need to be considered carefully as new protocols are defined for clinical trials, and more work is needed to define the complex effects that different immunosuppressive agents may have. This information is essential for the effective clinical translation of cell therapies in transplantation, as discussed below.
Clinical translation of regulatory cell therapies Cellular therapies using T Reg cells, regulatory macro phages and MSCs to suppress rejection or GVHD are being developed for application to clinical trans plantation (FIG. 3).
T Reg cell therapy. In animal models, the transfer of T Reg cells, with or without manipulation ex vivo, has proved to be a very effective strategy for controlling acute 6,141-144 and chronic 142,145 allograft rejection and for prevent ing GVHD 146 . These studies provide proofofconcept data to support the clinical translation of this approach. However, defining the human T Reg cell source, the specific population and the expansion strategy that are most effective presents many challenges.
In the setting of bone marrow transplantation, a few clinical studies have infused T Reg cells in an attempt to limit GVHD (TABLE 2). Expanded popula tions of thymusderived T Reg cells were used to treat two patients who developed GVHD following bone marrow transplantation, and clinical improvement was seen in both patients 147 . In a doseescalation study, CD4 + CD25 + FOXP3 + cells isolated from cord blood and expanded ex vivo for 24 hours in the pres ence of beads coated with CD3 and CD28specific antibodies were infused into adult patients receiving HSCT. No safety concerns were reported and, when the data were compared with the results from historical controls, a slight reduction in the incidence of GVHD grades II-IV was observed 148 . Expanded populations of donorderived CD4 + CD25 + T Reg cells have also been infused into HLAhaploidentical HSCT recipi ents, again in a doseescalation study. Encouragingly, GVHD was prevented in the absence of any immu nosuppressive treatment regimen, and there was no loss of the graft-versus-leukaemia effect 149 . In addition, lymphoid reconstitution occurred and the patients had improved immunity to opportunistic pathogens. In organ transplantation, the ONE Study -a multi centre Phase I/II study funded by the European Union FP7 programme -will investigate the safety of infusing ex vivoexpanded T Reg cells and T R 1 cells into kidney transplant recipients.
An alternative approach, involving the use of IL2 and rapamycin to enhance T Reg cell function in vivo in mice, has been shown to be effective for the treatment of acute GVHD 150 . Lowdose IL2 therapy was also used success fully to treat patients with chronic GVHD 151 , resulting in an increase in the median number of T Reg cells to more than five times the baseline levels. Importantly, the low dose IL2 therapy expanded T Reg cell populations with out affecting the function of conventional CD4 + T cells. Whether the same would hold true in recipients of solidorgan transplants remains to be determined.
One of the concerns about the use of T Reg cells clini cally is whether global immunosuppression will occur. If so, it could be argued that they offer no advantage over current immunosuppressive regimens. In mice, following bone marrow transplantation, animals injected with T Reg cells during solidorgan transplan tation were able to control cytomegalovirus 152 and influenza virus 153 infections. Importantly, in humans there is no evidence for an increase in susceptibil ity to infections in patients infused with T Reg cells 149 . Moreover, T Reg cell infusion can prevent GVHD while maintaining the graftversusleukaemia effect that occurs during bone marrow transplantation 24 . Clearly, this parameter will require careful monitoring in future clinical studies.
A second concern is the heterogeneity of the FOXP3 + T Reg cell populations that have been charac terized to date. Currently, there are no markers that can accurately discriminate human T Reg cells from other T cells. For example, although T Reg cells have been associated with a CD25 hi CD127 low phenotype, activated T cells also upregulate CD25 and downregulate CD127.
Thus, populations isolated on the basis of expression of one or both of these markers may contain contami nating effector T cells. The addition of immunosup pressive agents, such as rapamycin, to the expansion cultures has been shown to increase the purity of T Reg cells by eliminating nonT Reg cells present in the starting population 140 . As such, the immunosuppressive drugs used in transplantation may reduce the impact of con taminating cells. As mentioned above, careful consid eration will need to be given to the immunosuppressive drug regimen.
Regulatory macrophage cell therapy. Regulatory macro phages isolated from the organ donor have been administered intravenously to two recipients of kidney transplants from living donors. This had no deleteri ous impact on graft function, and enabled tacrolimus immunosuppressive therapy to be reduced within the first 24 weeks after transplantation 103 . Following administration, the regulatory macrophages remained viable and trafficked from the lungs via the blood to the liver, spleen and bone marrow. Studies to define the mechanisms responsible for graft survival revealed that at 1 year after transplantation the patterns of gene expression found in the peripheral blood of the patients treated were similar to those described for immuno suppressionfree kidney transplant recipients 18 (TABLE 3). A followup study using regulatory macrophages in kidney transplant recipients will be performed as part of the ONE Study.
Mesenchymal stromal cell therapy Bone marrowderived MSCs used in cell or organ transplantation can be autologous or derived from the donor or a third party. Autologous MSCs are clearly the safest option for clinical cell therapy in terms of the rel ative risk of rejection or GVHD. However, there are cir cumstances in which healthy autologous HLAmatched cells or haploidentical cells will not be available, and in such situations thirdparty allogeneic MSCs could provide an immediate source of cells ready for clini cal cell therapy. Intravenous administration has been shown to be a suitable route for MSC infusion, but an additional possibility would be to infuse MSCs into the donor organ before transplantation or to cotransplant MSCs at the site of the allograft 129 .
The number of bone marrowderived MSCs infused in the clinical studies reported to date has ranged from 4 × 10 5 to 1 × 10 7 per kg body weight 154,155 , with no signifi cant correlation between the dose of MSCs received and the clinical outcome. Single, double and repeated infu sions have been administered with no obvious pattern to the outcome observed in each variation of the proto col. For example, some patients responded to a second infusion following a nonresponse to the first, whereas others failed to respond even after multiple infusions 154 . Not all clinical studies using MSCs to modulate immune reactivity have reported positive data. A significant pla cebo effect was observed in Phase III clinical trials that used thirdparty MSCs to treat GVHD 156 . By contrast, treatment with MSCs was found to correlate with a sig nificant improvement in patients with steroidresistant liver or gastrointestinal GVHD 156 . The activation status of the MSCs at the time of infusion may explain these inconsistencies. In kidney transplantation, the infusion of autologous MSCs into patients resulted in a lower incidence of acute rejection, a decreased risk of oppor tunistic infection and an improvement in the estimated renal function at 1 year posttransplantation 157 .
Although over 100 clinical trials investigating the immunomodulatory and proreparative effects of MSCs are in progress (see ClinicalTrials.gov), this form of cell therapy is still at an early stage of development. The results of these trials will undoubtedly provide further insight into the application of therapeutically administered MSCs in transplantation.
Other forms of cell therapy Other forms of cell therapy also have potential in clini cal transplantation, as highlighted by the experimental data reviewed above. Clinical trials using therapeutic DC vaccines to stimulate immune responses in patients with cancer and to restore selftolerance in patients with type 1 diabetes are in progress 158 . The implementation of human DC therapy in clinical transplantation still requires several key parameters to be optimized and finalized. These parameters include the methodology for DC iso lation and purification 159 , the source of DCs, the route and timing of administration, and the most appropriate form of adjunctive immunosuppressive therapy.

Conclusion
Immune regulation is a complex process that involves multiple mechanisms and, more often that not, the cooperation of distinct cell populations. Transplantation is no exception to this rule, and multiple regulatory cell types can have roles at different stages of the allo graft response. Interestingly, one of the common fea tures of many regulatory leukocytes is their ability to produce IL10, a cytokine that can create a microen vironment that facilitates regulation and that may function to enhance the generation and function of regulatory immune cells throughout the posttrans plant period. Transplant recipients are treated rou tinely with immunosuppressive drug therapies, some of which may inhibit the generation or function of regulatory immune cells. Currently, lifelong treatment with these drugs is normally used to prevent the risk of graft rejection. Understanding the impact of immu nosuppressive drugs on the function and generation of regulatory immune cells is an important step for the successful clinical translation of novel cell therapies. The goal of introducing regulatory immune cell ther apy in clinical transplantation is to improve longterm outcomes after transplantation.