The risk for the development of cancer and opportunistic infections is markedly increased for organ recipients receiving long-term immunosuppressive therapy.1 These severe side effects cannot be strictly avoided because they result from the desired state of general immunosuppression, which nonspecifically affects the entire immune response, not only the antigraft response. To potentially overcome this dilemma, cell-based approaches to immunotherapy could have the potential to induce tolerance toward a transplanted organ without the need of nonspecific immunosuppression or at least with a marked reduction in long-term immunosuppression.
Mesenchymal stem cells (MSCs) have recently emerged as promising candidates for cell-based immunotherapy promoting tolerance of solid allografts2–4 because they modulate the immune response in various ways. With respect to suppressing T cell proliferation in a clinically significant way, MSCs compete with other cell populations. It has been suggested that fibroblasts might possess immunoregulatory properties similar to those of MSCs and that immunoregulation might be a common feature of all stromal cells.5, 6 However, besides the advantages of MSCs with respect to cultivation, production, and storage, so far only MSCs have been applied successfully in experimental solid organ transplantation and clinical studies. Moreover, MSCs possess an additional regenerative potential and may therefore participate in the regeneration of marginal organs after transplantation, thus improving the overall clinical outcome. Therefore, MSCs, in our opinion, present the most suitable candidate cell population in comparison with fibroblasts, regulatory T cells, and tolerogenic dendritic cells (DCs). When we consider the introduction of a clinical protocol in transplantation medicine, it is beneficial that MSCs have been previously used in several large-scale clinical trials for different etiologies and are also available as off-the-shelf cell products manufactured by commercial providers.
MSCs were first characterized by Friedenstein and colleagues7–9 as adherent, fibroblast-like cells more than 30 years ago. Over the past 3 decades, a multitude of efforts have been made to characterize and study MSCs in a variety of applications. Although no unique marker has been established to isolate and identify MSCs, it has become clear that they do not express hematopoietic surface markers CD34 and CD45 but stain positive for CD44, CD29, CD105, CD73, and CD166.10, 11 Recent findings also indicate that MSCs express the pluripotency markers stage specific embryonic antigen-1, stage specific embryonic antigen-4, Oct-4, and Stro-1.12, 13 MSCs constitute a small proportion of the bone marrow14 but can also be isolated from several other tissues, such as adipose, cord blood, fetal liver, peripheral blood, and lung tissue.15–19 They can easily be purified by plastic adherence and can be expanded several-fold in vitro11, 20 without losing their differentiation capacity. MSCs not only can be differentiated into cells and tissues of the mesenchymal lineage, such as osteoblasts, chondrocytes, and adipocytes,11, 21–25 but also can turn into cells of distinct germinal sheets, such as neurons, lung, liver, and intestinal epithelium, splenocytes, and kidney cells.26–30 Additionally, MSCs express and secrete cytokines and growth factors, thereby regulating hematopoiesis31, 32 and the engraftment of transplanted hematopoietic stem cells in animal models.33–36 Besides these hematopoietic effects, MSCs deploy a bimodal immunofunction. They can exert immunosuppressive and immunostimulatory effects. The result of MSC influence appears to depend on the extent of the stimulus.37, 38 Mainly antiproliferative effects were detected when MSCs were cultured with lymphocytes in mixed lymphocyte reactions,14, 39–44 even when third-party MSCs were added or T cell proliferation was extensive in response to different mitogenic stimuli. Hence, the immunomodulatory effects of MSCs might be exploited for novel treatment regimens targeting autoimmune diseases, such as ulcerative colitis and graft-versus-host disease (GVHD),45 and to protect solid organ grafts from being rejected.
In this review, we focus on recent data describing the immunomodulatory potential of MSCs and try to shed light on possible immunological mechanisms underlying these MSC immune effects. We also summarize the information deducible from the first clinical trials using MSCs as immunomodulators to finally put forward a proposal for a clinical study protocol applying MSCs in clinical liver transplantation for further discussion.
Sources and Culture of MSCS
Because of the lack of a definite marker allowing their prospective isolation from a primary source, MSCs are usually isolated by plastic adherence and multipassage culture. In addition to bone marrow, MSCs have been isolated and expanded from several other tissues such as heart, spleen, perirenal fat, cord blood, fetal liver, and lung tissue.15–17, 19, 46 It has now been well established that MSCs from various sources have similar phenotypes and immunosuppressive properties in culture and comparable differentiation potentials. Thus far, MSCs have been isolated from many different species, including humans,11, 47–49 mice,47–49 rats,50 dogs,51 baboons, pigs, sheep, goats, rabbits, and cats.52 It can therefore be expected that they represent a conserved type of mammalian cell. However, some differences in MSC cultivation and expansion potential (eg, between murine and human or rat cells) have been reported. The isolation of murine mesenchymal stem cells (mMSCs) from bone marrow by standard methods usually results in a very heterogeneous cell population with a high degree of hematopoietic contamination.53 Therefore, several groups have undergone enormous efforts to develop standardized protocols to guarantee reproducible preparations of mMSCs.47, 48, 54 Nonetheless, also in our hands, the requirements to culture mMSCs differ from those for human and rat MSCs. In addition, mMSCs from various mouse strains have distinct growth rates, media requirements, surface epitopes,48 and differentiation capacities. The overall quality criteria for MSCs of any species are the absence of hematopoietic surface markers CD34 and CD45 and the presence of CD90, CD105, CD166, CD44, and Sca-1 expression11, 42 as well as a clear trilineage differentiation potential (osteoblasts, chondrocytes, and adipocytes).11, 22, 23, 25
A lot has recently been published on the immunomodulatory effects of MSCs in vitro. We and many others have thoroughly outlined that MSCs can inhibit the proliferation of lymphocytes stimulated by mitogens, antibodies, or alloantigens in a dose-dependent manner. This antiproliferative effect has been observed for stimulated peripheral blood mononuclear cells (PBMCs), splenocytes, and purified CD4 and CD8 T cell subsets of mice,55–57 rats,50 baboons,58 and humans.14, 39, 46 Furthermore, MSCs also inhibit the antigen-presenting function and maturation of DCs39, 59 and the proliferation of B cells57 and interact with other immune cells (reviewed by Uccelli et al.60). The origin of MSCs does not play a crucial role with respect to their immunosuppressive function because human MSCs harvested from the heart, spleen, fatty tissue, and bone marrow show similar antiproliferative properties.46 Even more interesting for a transplantation setting, the suppression of lymphocyte proliferation can be achieved with various MSC types, including recipient-derived, donor-derived, and third-party MSCs.
Despite the ongoing efforts to investigate MSC-mediated immunomodulation, the biological mechanism behind these observations remains largely unclear to date and in part controversial. Some authors claim cell-cell contact as a necessary prerequisite,2, 50, 56 whereas others have disproved cell-contact dependence in their settings.14 Thus, cellular and soluble factors most likely contribute to MSC-mediated inhibition of proliferation. This conclusion is also in agreement with a recent study by Shi et al.,55 who reported that the addition of an mMSC culture supernatant had a significantly higher inhibitory effect on murine splenocytes when MSCs were also present.55 Further contradictory data exist about the reversibility of MSC-mediated effects. Some authors have claimed that responder cells can be restimulated when separated from MSCs,14 whereas others have outlined the opposite.57
The role of cytokines has been intensively investigated, and several cytokines related to MSC-mediated immunosuppression have been identified. Several studies have shown that tumor necrosis factor α (TNF-α), interferon γ (IFN-γ), transforming growth factor β1, hepatocyte growth factor, interleukin 10 (IL-10), and the enzyme indoleamine 2,3-dioxygenase are involved in MSC-mediated inhibition.56, 61–63 The key players, however, are still disputed. Aggarwal and Pittenger39 and Ryan et al.62 gave evidence that prostaglandin E2 may be an important factor in their respective settings because the addition of prostaglandin E2 inhibitors partly abolished the MSC-mediated effects on human PBMC proliferation. For CD4 T cells, MSCs favor a shift from the T helper 1 subset to the T helper 2 subset,39 resulting in a higher frequency of anti-inflammatory cells. Recently, Sheng et al.56 introduced the finding that suppressive MSC effects depend on the presence of IFN-γ. In their system, proliferation of lymphocytes from IFN-γ−/− mice could not be inhibited by MSCs unless exogenous IFN-γ was added. Sheng et al. also demonstrated that IFN-γ induces the expression of programmed death ligand 1 on MSCs, and they showed that significant levels of IFN-γ and TNF-α, but no transforming growth factor β, IL-10, or IL-12, can be found in cocultures of activated murine splenocytes and mMSCs. Another convincing study by Ren et al.64 outlined that MSCs must be “licensed” by a combination of IFN-γ and TNF-α, IL-1a, or IL-1b to exert their immunosuppressive effect. Furthermore, using MSCs from inductible nitric oxide synthase−/− mice, they illustrated that nitric oxide production is a key factor for MSC-mediated inhibition of effector cell proliferation.
In contrast to the findings of MSC-mediated responder cell suppression, there is increasing evidence that MSCs can enhance responder cell proliferation in some settings. When the frequency of CD14+ cells exceeded 5%, human third-party MSCs increased the proliferative response in a mixed lymphocyte response.14 In addition, Le Blanc65 outlined that MSCs can stimulate human PBMCs alone or in allogeneic mixed lymphocyte cultures; we confirmed this finding in our laboratory. This stimulating effect was most prominent with very low numbers of MSCs. Casiraghi et al.2 demonstrated that MSCs containing hematopoietic stem cells cannot prolong allograft survival in a murine model of heterotopic heart transplantation, whereas MSCs alone can. Besides lymphocytes, MSCs also influence other cells of the immune system. Thus, MSCs reduce the ability of DCs to present antigen.59 Evidence is also accumulating that MSCs can induce a regulatory phenotype on DCs.66, 67 Therefore, MSCs suppress the proliferation of effector cells not only directly but also indirectly by preventing the initiation of the immune response through adequate antigen presentation.
In short, there is convincing evidence that MSCs can suppress the proliferation of alloreactive lymphocytes in vitro. The exact conditions that drive this strong suppressive effect, however, are undetermined so far and need further investigation.
MSCS in Immunological Disease Models
The in vivo immunomodulatory properties of MSCs were first described in a skin transplantation model.58 Later, MSCs were applied to ameliorate GVHD. Since then, contradictory results have been reported. In humans, MSCs were applied successfully to treat GVHD,68 whereas mice with GVHD did not respond to MSC infusions.69 Animal models featuring MSCs in solid organ transplantation emerged in 2006.50, 70 At that time, our group reported no effect on graft survival when MSCs were applied concurrently with cyclosporine after heart transplantation in rats. In contrast, other groups have demonstrated prolonged graft survival when MSCs were applied before transplantation with or without additional injections in corresponding models.70 We recently extended our initial findings and outlined that MSCs applied alone induce rejection, but their injection together with low doses of mycophenolate promotes long-term graft survival.3 Prolonged graft survival has been attributed to hemeoxygenase-1,4 to indoleamine 2,3-dioxygenase,3 and to the induction of regulatory T cells.2
The in vivo findings emphasize that the multitude of results from in vitro studies cannot be directly transferred into transplant models. MSCs have been applied at ratios of 1:1 to 1:5 with effector cells in vitro; corresponding high doses cannot be achieved in vivo because the injection of more than 2 million MSCs in small animals causes lung embolisms on a regular basis. This is due to the large size of MSCs, which presumably stick together and can obstruct the first capillary net that they encounter, at least in smaller animals and with non–good medical practice-produced MSCs because no thromboembolic complications have been reported in human studies so far. Moreover, many in vitro effects attributed to MSCs in culture depend on cell-cell contact or at least the close vicinity of MSCs and effector cells. The situation in vivo is therefore markedly different, as it remains unknown when MSCs have direct contact with immune cells and how they migrate through the host over time. In our own model, we detected allogeneic MSCs by Y-chromosomal polymerase chain reaction 4 days after application; intact living cells were never demonstrated. Additional evidence that we have found (unpublished observations, 2008) suggesting that MSCs are trapped in the lungs and liver (depending on the site of injection) makes it unlikely that MSCs exert long-term immunosuppressive functions themselves. It has been further reported that xenogeneic MSCs survive in the long term in nonobese diabetic/severe combined immunodeficient mice,71 and thus it appears possible that MSCs also survive in pharmacologically immunosuppressed organ transplant recipients. The best results in our animal model, however, were achieved when MSCs were applied before transplantation, when no immunosuppressive agents were present. In that setting, intact MSCs were not detected, and this suggests that in a similar clinical setting, MSCs will not be present in the organ recipient at the time of transplantation. Because MSCs seem to be cleared from the host over time, the question of their immunogenicity arises. Early reports indicated that MSCs are immunoprivileged and that even allogeneic cells are not targeted by the immune system despite their ubiquitous expression of major histocompatibility complex class I.65, 72 Later experiments suggested that allogeneic MSCs can indeed cause an immune response and can be rejected accordingly.73, 74 In that latter case, MSCs might act via soluble factors, or they more likely interact with immune cells during their short stay and license those to facilitate tolerance.
In short, the in vivo setting is far more complicated than culture experiments because the location of MSCs over time will fundamentally affect MSC functionality. Tracking MSCs in vivo is therefore one of the most critical tasks for identifying candidate cells that interact with MSCs. In this respect, DCs should be carefully considered as they possess a key role during the initiation of the immune response. Besides, DCs have been known to induce tolerance under conditions similar to those for MSCs,75 and it has been reported that DCs as well as MSCs up-regulate indoleamine 2,3-dioxygenase through autocrine IFN-γ secretion.76, 77 Thus, MSCs might act like DC-like cells or may induce the emergence of tolerogenic DCs before being rejected themselves.
Drugs that affect the immune system add extra complexity to the equation of MSC-induced tolerance as their interactions with MSCs are largely unknown. However, immunosuppressive drugs must be applied in the human transplant setting to ensure that patients receive the gold standard of immunosuppressive therapy after organ transplantation. Therefore, in a clinical study, MSCs will always be applied on top of an established immunosuppressive protocol. Moreover, different immunosuppressants seem to have opposing effects in MSC-treated animals: we have reported that graft survival is shortened in combination with cyclosporin50 but prolonged with mycophenolate in our model,3 and we have recently confirmed this observation through a series of in vitro experiments (unpublished data, 2008).
Important safety issues have to be taken into account when MSCs are applied in humans. Immunosuppressive drugs are associated with augmented tumor incidence.78 Hence, MSCs exhibiting immunosuppressive effects might also be related to cancer development. There is evidence that MSCs promote cancer metastasis, but induction of de novo tumors has not been reported so far.79 Tumor progression depends on the interaction of epithelial tumor cells with stromal cells. MSCs can differentiate into various kinds of stromal cells, and this suggests a theoretical involvement of MSCs in tumor formation. Despite these considerations, some reports suggest suppression of tumorigenesis by MSCs.80 Recently, it has been shown that the virus-specific T cell response is not affected after MSC infusions.81
A Clinical Scenario: MSC Therapy after Liver Transplantation
Studies evaluating cell-based therapies used to induce nonresponsiveness after solid organ transplantation will always be hampered by 2 major intrinsic problems. First, organ recipients will usually be heavily immunosuppressed during the induction phase to ensure optimal survival of the transplanted organ in the early days. Thus, any additional cell therapy will meet an environment of global recipient immunosuppression and will generally encounter high concentrations of calcineurin inhibitors (CNIs), steroids, antibodies, and antimetabolites, which might influence the function of transplanted cells. Second, solid organ transplantation has become a very safe and efficient treatment for end-stage organ failure, at least in the short term, so any alteration of clinical standards must meet significant ethical doubt.
It might seem contradictory on first impression that our study group suggests the implementation of an MSC-based immunosuppressive strategy for liver transplant recipients. Most of the recent attempts to introduce cell therapies have focused on renal transplantation and later stages after transplantation. We believe, however, that MSC therapy will be especially successful in the early induction phase after transplantation when the recipient immune system first encounters donor antigen. The rationale behind the implementation of such a protocol in liver transplant recipients is a combination of safety issues and aspects of liver immunobiology. Acute rejection after liver transplantation can be diagnosed promptly through biopsy and can almost always be controlled by a steroid bolus and an increase in immunosuppression without a clinical disadvantage for the patient in the long term. In addition, the intrinsic regenerative potential of the liver makes it likely that it would be susceptible to an increase of that potential by MSCs.
The problem of concurrent immunosuppression remains. In our hands, calcineurin inhibition after experimental solid organ transplantation abrogates the immunosuppressive effect of MSC therapy. Thus, a clinical protocol that involves early and/or ongoing calcineurin inhibition by, for example, cyclosporin plus MSC infusion is most likely prone to failure. In our opinion, the same will most likely be true when immunoregulatory cells other than MSCs are used together with CNIs. Besides, CNIs cause renal failure, hypertension, and hyperglycemia and increase the risk of malignancy like most other immunosuppressive drugs. Efforts have been made to reduce CNI treatment in organ transplantation protocols.82 The first promising results were achieved in patients undergoing liver transplantation with renal failure who received CNI-free immunosuppression with mycophenolate (A. A. Schnitzbauer, M. N. Scherer, and J. Rochon, unpublished results, 2008; A. A. Schnitzbauer, A. Doenecke, and J. L. Sothmann, unpublished results, 2008). This regimen is not yet accepted as the clinical standard, but we anticipate that it has a future in liver transplantation. Therefore, we suggest discussing a study protocol that applies CNI-free immunosuppression after liver transplantation that is further increased when needed; we call this approach bottom-up immunosuppression. This strategy will provide the best setting for analyzing the addition of MSC infusions in a clinical phase I/II study. It must be anticipated that some study participants will experience an episode of acute rejection. However, this event can be controlled by a steroid bolus and a simultaneous switch to high-dose CNI therapy in most cases. In addition to being a reasonable cell therapy protocol, this approach will allow us to define the frequency of early onset rejection as a primary study endpoint. As HCV patients are prone to early rejection, this population will most likely be excluded from a first clinical trial.
We and others have recently provided striking evidence that MSCs are powerful immunomodulators after solid organ transplantation in several preclinical models. The immunological mechanism behind this phenomenon remains unclear to date but likely involves the interaction of MSCs and MSC-secreted cytokines with specialized antigen-presenting cells or T cells of the recipient (Fig. 1). Our group and others are planning and executing clinical phase I and II studies to use the immunosuppressive potential of MSCs before and after solid organ transplantation. Network formation will be crucial for the success of MSC-based strategies (see http://www.misot.de).