Immunomodulation by mesenchymal stem cells and clinical experience


Katarina Le Blanc, Center for Allogeneic Stem Cell Transplantation and Division of Clinical Immunology, Karolinska University Hospital Huddinge, F79, SE-14186 Stockholm, Sweden.
(fax: +46-87-466699; e-mail:


Mesenchymal stem cells (MSCs) from adult marrow can differentiate in vitro and in vivo into various cell types, such as bone, fat and cartilage. MSCs preferentially home to damaged tissue and may have therapeutic potential. In vitro data suggest that MSCs have low inherent immunogenicity as they induce little, if any, proliferation of allogeneic lymphocytes. Instead, MSCs appear to be immunosuppressive in vitro. They inhibit T-cell proliferation to alloantigens and mitogens and prevent the development of cytotoxic T-cells. In vivo, MSCs prolong skin allograft survival and have several immunomodulatory effects, which are presented and discussed in the present study. Possible clinical applications include therapy-resistant severe acute graft-versus-host disease, tissue repair, treatment of rejection of organ allografts and autoimmune disorders.


Mesenchymal stem cells (MSCs) were first characterized by Friedenstein and colleagues, who identified an adherent, fibroblast-like population in the adult bone marrow that could regenerate rudiments of bone in vivo [1–4]. Since then, MSCs have also been isolated from other tissues, including adipose, cord blood, and fetal liver, blood, bone marrow and lung [5–8].

Mesenchymal stem cells are rare noncycling cells in the human bone marrow [9–13]. Positive selection using antibodies that recognize marrow fibroblastic cells enrich for a discrete subpopulation of colony forming cells that retain the capacity to differentiate into adipose tissue, cartilage and bone in vitro [13–18].

Due to the rarity of the cells in bone marrow and the lack of a definitive marker that specifically isolates MSCs, the cells are commonly isolated by adherence to plastic and consecutive passage in tissue culture. MSCs may be expanded several fold in vitro, for instance after aspiration of bone marrow (Fig. 1). After expansion, they can be used for in vitro research, or clinical application.

Figure 1.

 Harvest of mesenchymal stem cells and ex vivo expansion for research and clinical use. (a) Aspiration of bone marrow, (b) mononuclear cells are separated by a density gradient, (c) the cells are cultured on flasks supplemented with 10% fetal calf serum. A selected batch for optimal expansion should be screened for, (d) when the cells are >70% confluent, the cells are detached by trypsin, (e, f) the cells are re-plated at a lower cell density, (g–i) the cells are harvested and may be used for research or injected i.v. to the patient.

Whether culture-expanded MSCs differ from their in vivo progeny remains uncertain, as proliferation on plastic surfaces could induce both phenotypic and functional changes [19, 20]. Certain subtypes of MSCs may also have a survival advantage in culture. Even single cell-derived clones of MSCs vary in their gene expression, differentiative capacity, expansion potential and phenotype [21–23]. The definition of MSCs generated ex vivo is a composite of morphological, phenotypical and functional characteristics. MSCs stain negative by flow cytometry for haematopoietic markers CD34, CD45 and CD14 and positive for CD29, CD73, CD90, CD105 and CD166. Addition of exogenous factors to the growth medium induces MSCs to differentiate [1, 21, 24].

It remains to be established if MSCs are true stem cells. Regeneration and maintenance of a whole tissue compartment has not been shown by transplantation of MSCs at a single cell level – a feature characteristic of a true stem cell. However, MSCs are clearly multipotent in vivo. After injection into newborn mice or infusion in utero, the cells engraft in multiple organs and demonstrate site-specific differentiation [25, 26]. MSCs can be detected in low levels in numerous tissues following intravenous infusion and preferentially home to sites of injury [27–30]. In experimental animal models, MSCs not only regenerate tissues of mesenchymal lineages, such as intervertebral disc cartilage [31], bone [32, 33], cardiomyocytes [34], and knee joint repair following menisectomy [35], but also differentiate into cells derived from other embryonic layers, including neurones [36], and epithelia in skin, lung, liver, intestine, kidney and spleen [29, 37, 38]. MSCs constitutively secrete cytokines important for haematopoiesis and promote engraftment of haematopoietic stem cells in experimental animal models [39–45]. The enhancing effect is most prominent when the dose of haematopoietic cells is limiting.

Mesenchymal stem cells have been proposed to have immunosuppressive properties and reduce inflammation. Human MSCs suppress lymphocyte alloreactivity in vitro in mixed lymphocyte cultures (MLC), through human leucocyte antigen (HLA)-independent mechanisms [46–49]. Intravenous administration of MSCs improves the outcome of renal, neural and lung injury in experimental animal models mainly through paracrine effects and a shift from the production of pro-inflammatory to anti-inflammatory cytokines at the site of injury [38, 50, 51].

Mesenchymal stem cells are one of the few normal cells that have so far been produced in large quantities needed for therapeutic development. Preliminary data suggest that the cells have properties that allow for transplantation across major histocompatibility complex (MHC) barriers. The ability to home to damaged tissue suggests that MSCs may be used for cartilage and bone repair. They may also be used for chemotherapy-induced toxicity such as radiation gastroenteritis and haemorrhagic cystitis. The immunomodulatory effects of MSCs may be used to repair tissue damage caused by the immune system in autoimmune-induced inflammatory bowel disease such as Crohn’s disease and ulcerous colitis, graft-versus-host disease (GVHD) of the gut, liver and skin after allogeneic haematopoietic stem cell transplantation (HSCT) and to prevent rejection of organ transplants.

This review focuses on recent findings that indicate a role for MSCs in immunomodulation and describes our current understanding of the mechanisms by which it occurs. We also present the first clinical trials using MSCs in this context.

MSC interactions with lymphocytes in vitro

The expression of MHC molecules on all cells on the body allows the immune system to distinguish self from nonself. MSCs share surface markers with the thymic epithelium and express adhesion molecules that are essential for T-cell interaction, including vascular cell adhesion molecule 1, intracellular adhesion molecule 2 and lymphocyte function-associated antigen 3 [21, 40, 52–54]. Adult human MSCs express intermediate levels of cell-surface MHC class I molecules. MHC class II is not detectable on the cell surface, but Western blotting on cell lysates shows that the cells contain intracellular deposits of class II alloantigens [55, 56]. γ-Interferon (IFN-γ) stimulation increased both class I and class II molecules [56]. MSCs differentiated into adipose, bone and cartilage cells express HLA class I, but the expression of class II can no longer be induced [55].

In the absence of immune suppression or tolerogenic mechanisms, allogeneic cells are rejected by the immune system. Cells expressing MHC molecules stimulate T cells directly only if they possess appropriate co-stimulatory molecules. Human MSCs do not express co-stimulatory molecules CD80 (B7-1), CD86 (B7-2) or CD40, even after IFN-γ stimulation. Allogeneic cells can also activate T cells through an indirect pathway where their MHC antigens are presented by professional antigen presenting cells (APC). Co-culturing with renal tubular epithelial cells and skin fibroblasts induces proliferation of allogeneic lymphocytes. However, in contrast, undifferentiated MSCs, MSCs exposed to IFN-γ fully expressing class II alloantigens and MSCs differentiated into adipocytes, osteocytes and chondrocytes, all fail to induce proliferation of allogeneic lymphocytes even in the presence of APCs or after provision of co-stimulatory signals using either CD28-specific antibodies and CD80 or CD86 gene transduction [45, 46, 48, 49, 55–58]. Alloreactivity can also be measured as IFN-γ production by activated lymphocytes. The results agree with the proliferation studies; human and rat MSCs do not elicit IFN-γ production by human peripheral blood mononuclear cells (PBMCs), whereas human and murine fibroblasts do [33, 45, 47, 56, 58]. The expression of lymphocyte activation-associated markers CD25 [interleukin-2-stimulated receptor (IL-2)], CD38 and CD69 decreased in the presence of MSCs [56, 59]. However, although MSCs may not induce lymphocyte activation, measured as proliferation, IFN-γ production or upregulation of activation-associated markers, priming of the lymphocytes still occurs. If T cells are co-cultured with MSCs and subsequently cultured with unseparated lymphocytes derived from the MSC donor, the stimulatory response is nearly identical to that of T cells precultured with lymphocytes derived from the same donor [56].

Mesenchymal stem cells not only fail to induce activation of CD4+ cells but also escape lysis by CD8+ cytotoxic lymphocytes [60, 61]. Even unseparated lymphocytes stimulated in vitro to target PBMCs derived from a specific donor, will lyse lymphocytes from that individual but not MSCs derived from the same donor [60]. MSCs were also comparatively more resistant to lysis by both alloreactive as well as peptide-specific MHC-class-I-restricted cytotoxic T cell clones (I. Rasmusson, M. Uhlin, K. Le Blanc and F.V. Levitsley, unpublished observation). Further analyses indicate that MSCs induce an abortive activation programme in fully differentiated CD8+ T cells so that major effector functions are not activated. MSCs also escape natural killer (NK) cell-specific lysis [60].

MSC–T cell interactions

Effects on T cells

Mesenchymal stem cells are immunosuppressive and inhibit the response of naïve and memory T cells in MLC and induced by mitogens [45–49, 55–57, 59–63]. Suppression is MHC independent and most marked if MSCs are added on the first day of the 6-day culture. The degree of suppression is dose dependent. Marked inhibition is observed when larger numbers of MSCs are present (MSC : lymphocyte ratio >1 : 10). By contrast, the addition of MSCs at a low ratio (1 : 100–1 : 10 000) often enhances proliferation [48, 57, 64]. The suppressive effect is retained by MSCs that have been induced to differentiate into osteocytes, adipocytes and chrondrocytes, and is further enhanced by pretreatment with IFN-γ [55]. MSC-mediated inhibition can cross species barriers. Porcine MSCs suppressed human lymphocytes stimulated by xenogeneic lymphocytes [64]. Human and mouse-derived MSCs also suppress xenoreactivity [64]. However, the suppressive mechanisms may differ between species. Human MSCs are suppressive when MSCs and lymphocytes are separated by a permeable membrane whereas rodent MSCs require cell–cell contact [47, 49, 56, 60, 62]. Human MSCs do not induce tolerance, anergy or apoptosis. After allostimulation in the presence of human or baboon MSCs, lymphocytes can be restimulated if the MSCs are removed [46, 56]. By contrast, if rodent MSCs have been present in the primary MLC for more than 24 h, murine T cells produce IFN-γ when restimulated but arrest in the G1 phase of cell division and fail to proliferate [65].

MSC effects on regulatory T cells and activation molecules

The subpopulation of naive CD4+ T cells that express CD25 are called regulatory T cells and have potent suppressor activity [66, 67]. MSCs increased the proportion of CD4+CD25hi, CD4+ CTLA4+ and CD4+CD25+CTLA4+ cells in IL-2-stimulated lymphocytes and MLC (Fig. 2) [68, 69]. In contrast, the number of CD25+ and CD38+ cells decreased in the presence of MSCs in mitogen-stimulated lymphocyte cultures [59, 70]. Depletion of CD25+ cells before stimulation with mitogen-activated monocytes did not affect the ability of MSCs to inhibit T-cell proliferation [58]. The induction of regulatory T cells may be mediated by different factors in alloreactive and mitogen-stimulated lymphocyte cultures as differences exist between the systems. MSCs also produce bone morphogenic protein-2, which mediates immunosuppression via the generation of CD8+ regulatory T cells [71].

Figure 2.

 The multiple effects of MSCs on immune cells. (a) MSCs increase the proportion of CD4+CD25+ cells and IL-10 production [63, 68, 69]. (b) MSCs decrease markers for activated T cells, CD25, CD69 and CD38 [59, 60]. MSCs delayed maturation of APC and decreased expression of HLA-DR. (c) Dendritic cell type 1 when stimulated had decreased TNF-α and IL-12, when co-cultured with MSCs [68]. (d) MSCs increased IL-10 secretion by LPS-stimulated dendritic cells type 2, CD4+ cell had decreased IL5-secretion. (e) T-helper cell type 1 IFN-γ production was significantly decreased by MSCs. (f) T-helper cell type 2 increased IL-4 secretion in the presence of MSCs. (g) MSCs inhibit mixed lymphocyte cultures and subsequent development of cytotoxic T cells by a soluble factor. (h) Several soluble factors are produced by MSCs, amongst them are IL-6, IL-8, stem-cell derived factor 1 (SDF1), vascular endothelial growth factor (VEGF). Soluble factors that have been suggested to inhibit T-cell activation are prostaglandin E2 [68], which induces regulatory T-cells, indoleamine 2,3-dioxygenase (IDO) [74], which is induced by IFN-γ which catalyses the conversion from tryptophan to kynurenine and inhibits T-cell responses [73]. Other soluble factors that have been suggested to inhibit T-cell responses are TGFβ1 [61, 62], hepatocyte growth factor [62] and IL-2. (i) Purified NK cells co-cultured with MSCs significantly decrease IFN-γ levels [68]. (j) MSCs were reported to decrease proliferation and immunoglobulin secretion of B cells when MSC were present in an equal proportion (1 : 1) [77]. Furthermore, chemokine receptor CXCR4, 5 and CCR 7 were significantly downregulated by MSCs. However, MSCs : lymphocytes in a lower concentration (1 : 10) stimulated B-cell antibody secretion [83].

MSC effects on T helper cells and cytotoxic T cells

In the presence of signals that favour T-helper cell 1 (TH1) development, such as CD3, CD28, IL-4, IL-2 and IL-12 stimulation, naïve T cells mature into IFN-γ-secreting cells. If MSCs are present in the culture, IFN-γ secretion is reduced (Fig. 2) [68]. Thus, MSCs induce a bias towards TH2 differentiation.

Mesenchymal stem cells suppress CD8+ T-cell-mediated lysis if added at the beginning of the MLC [60]. Cytotoxicity was not affected if MSCs were added in the cytotoxic phase [57, 60, 61, 69]. Lysis was partially abrogated by the addition of IL-2 [61]. MSCs may inhibit the afferent phase of alloreactivity and prevent the development of cytotoxic T cells. Once cytotoxic T cells are activated, MSCs are not effective. In vivo studies are necessary to elucidate this point.

Possible mechanisms of T-cell inhibition by MSCs

Human MSCs suppress the formation of CD4+ and CD8+ T cells by soluble factor(s) [49, 56, 60, 62]. The suppressive factor is not constitutively secreted by MSCs because cell culture supernatants do not suppress T-cell proliferation [45, 57, 59, 72]. Several soluble factors have been suggested but the available data are contradictory (Fig. 2). Antibodies against hepatocyte growth factor and transforming growth factor β1 (TGFβ1) partially restored proliferation of purified T cells but not PBMC [48, 61, 62]. Other suggested factors include cytokines IFN-γ, IL-10, tumour necrosis factor-α (TNF-α) and IL-2 [45, 46, 49, 56–58]. Liu et al. showed that the addition of antibodies specific for FasL and TGFβ1 attenuated suppression by MSCs in concanavalin A-stimulated MLC in a dose-dependant fashion, but anti-IL-10 had no effect [64].

Mesenchymal stem cells might inhibit T-cell proliferation through the production of indoleamine 2,3-dioxygenase (IDO). IDO is induced by IFN-γ, catalyses the conversion of tryptophan to kynurenine and inhibits T-cell responses by tryptophan depletion [73]. Meisel et al. demonstrated by Western blotting that human MSCs do not constitutively express IDO, but the expression is induced by IFN-γ [74]. IFN-γ also stimulates IDO enzyme activity in a dose-dependent manner. Significant IDO activity was detected in T cells stimulated with mitomycin C-treated PBMC in the presence of MSCs [74]. The addition of tryptophan restored T-cell proliferation, suggesting that IDO-activity can act as a T cell inhibitory effector mechanism. However, tryptophan depletion was not responsible for the immunosuppressive effect in MLC using unseparated mononuclear cells as responders [49]. Furthermore, tryptophan depletion leads to T-cell apoptosis and such cell death was not seen in two studies [49, 68].

Prostaglandin E2 (PGE2), which is synthesized by cyclooxygenase (COX) enzymes, induces regulatory T cells [75]. MSCs constitutively express both COX-1 and COX-2 [68, 76]. When purified T cells were co-cultured with MSCs, both COX-2 and PGE2 production increased [49, 68]. PGE2 synthesis inhibitors restored most of the proliferation of phytohaemagglutin-activated (PHA) lymphocytes co-cultured with MSCs [68].

Tse et al. studied alloreactive lymphocytes in contrast to mitogen-stimulated cultures. They found that neither MSC production of IL-10, TGFβ1, PGE2 or tryptophan depletion was responsible for the suppression in MLC [49].

There may be several reasons for the conflicting data. Whilst Tse et al. studied cultures of unseparated mononuclear cells as responders, purified T cells were used in the other studies. We previously showed that the results differed depending on how the lymphocytes were activated [63]. MSCs increased the transcription and translation of IL-2 and soluble IL-2 receptors when present in MLC, whereas the levels decreased after mitogen stimulation. When MSCs were added to the culture, IL-10 secretion increased in MLC but was unchanged in PHA-stimulated cultures. Addition of indomethacin, which inhibits PGE2 synthesis, partly restored MSC-mediated inhibition in lymphocytes stimulated with PHA, but not in MLC [63].

In many reports, only few experiments of each kind have been published, perhaps introducing a bias as the degree of immunomodulation by MSCs is highly variable [48]. Different culture conditions, kinetics and lymphocyte populations might also explain the contradictions. As mentioned above, species-specific differences also exist. The MSC doses have been variable. Some investigators used an MSC : lymphocyte ratio of ≥1 : 1 to demonstrate effects [45, 58, 62, 69, 77], whereas most studies used a ratio of 1 : 10 [48, 49, 55, 56, 59, 60, 62, 63, 69]. In dose–response experiments, high doses were immunosuppressive, whilst low numbers of MSCs sometimes enhanced proliferation [48, 64]. Nevertheless, a ratio of 1 : 10 of MSCs : lymphocytes in vitro is far from physiological, as the proportion of MSCs in the bone marrow was less than 1 : 10 000 mononuclear cells [78]. Thus, the more MSCs that are needed for an effect, the less likely the possibility that it is relevant in vivo. Such high doses cannot be given for therapeutic use.

MSC effect on APC

Mesenchymal stem cells reduce T-cell activation indirectly by reducing the formation of dendritic cells (DC) from monocytes [69]. Furthermore, MSCs inhibit differentiation and function of monocyte-derived DCs in a transwell system [79]. Inhibition was abrogated after removal of MSCs [58]. MSCs inhibit up-regulation of CD1A, CD40, CD80 (B7-1), CD86 (B7-2) and HLA-DR during DC maturation whilst CD83 increased [58, 68, 69, 80]. Importantly, DC isolated from cultures that were co-cultured with MSCs showed a reduced potential to activate CD4+ cells in MLC [79].

Mesenchymal stem cells reduced DC secretion of pro-inflammatory cytokines IFN-γ, IL-12 and TNF-α whilst production of the suppressive cytokine IL-10 increased (Fig. 2) [58, 63, 68, 79]. TNF-α inhibits DC maturation, migration and ability to stimulate alloreactive T cells, and may be an additional way of inducing immunosuppression by MSCs. In the presence of MSCs, IL-10-secreting plasmacytoid DCs, characterized by the expression of BDCA4 antigen, increased after stimulation with lipopolysacharide [68]. Similarly, mouse splenic stromal cells increased DC production of IL-10 and reduced IFN-γ, showing that MSCs differentiated to stromal cells have similar effects [81]. CD14+ monocytes activate MSCs to secrete soluble factors including IL-1β that inhibit alloreactive T-cells [70]. Most likely, the MSC effect on APC plays a role in the inhibition of T-cell activation.

MSC effect on natural killer cells

Mesenchymal stem cells reduced the secretion of IFN-γ by IL-2-stimulated NK cells (Fig. 2) [68]. MSCs did not inhibit NK-mediated lysis of K562 cells [60]. A ratio of 1 : 1 (MSC : blood lymphocytes) but not 1 : 10 inhibited both CTL and NK cell cytotoxic functions [69]. The study suggested that MSCs only in very high concentrations downregulated the expansion of CD8+ T cells and NK cells. Such high concentrations of MSCs cannot be used in vivo.

MSC effects on B cells

Mesenchymal stem cells have been shown to inhibit lymphocyte proliferation induced by B-cell mitogens, pokeweed mitogen and protein A from Staphylococcus aureus [48, 64, 65, 72, 77]. However, lymphocyte stimulation by pokeweed mitogen is T-cell dependent, and protein A stimulates proliferation of T and B cells [82]. Therefore, B-cell proliferation cannot be differentiated from that of T cells. MSCs inhibited B-cell stimulation by anti-CD40 and anti-IL-4 by a soluble factor [65, 77]. B-cell proliferation was arrested in the G0/G1 phase of the cell cycle, with no apoptosis. When purified B cells are activated, both the number of Ig-producing cells and the levels of IgM, IgG and IgA is reduced by MSCs if the cells are added at a ratio of 1 : 1 but not at lower ratios [77]. MSCs did not inhibit TNF-α, IFN-γ, IL-4 and IL-10 production. MSCs downregulated the expression of chemokine receptors CXCR4, CXCR5 and CCR7B as well as chemotaxis to CXCL12, the CXCR4 ligand, and CXCL13, the CXCR5 ligand, suggesting that high numbers of MSCs affect the chemotactic properties of B cells. However, MSCs at a 10-fold lower dose stimulated blood and splenic B cells to IgG production in Elispot [83]. When MSCs and spleen cells were separated by a semi-permeable membrane, IgG production was stimulated in unfractionated spleen cells, but not in enriched B cells. Thus, B-cell activation by MSCs may be caused by a soluble factor produced by accessory cells. Available data suggest that MSCs significantly effect B-cell inhibition as well as stimulation, depending on dose, source and test system (Fig. 2). These results may not support the potential therapeutic use of MSCs in autoimmune diseases, where B cells play a major role. High doses that may be dangerous to use for therapy were required for B-cell inhibition in vitro. At lower doses, MSCs were found to activate B cells. In vivo data are so far lacking.

MSC interactions with lymphocytes in vivo

Although MSCs do not induce an in vitro alloresponse, limited in vivo data are available. Transplanted allogeneic mismatched MSCs fail to induce an immune response and engraft in adult rodent, porcine and baboon experimental models [26, 30, 84]. Most clinical studies in humans have used MSCs from HLA-identical or near-identical donors. A low level of engraftment was observed after infusion of haploidentical-related MSCs to a patient with severe aplastic anaemia [85]. This was done so that MSCs by immunomodulation might reverse the T-cell defect thought to cause severe aplastic anaemia. Fully mismatched allogeneic fetal liver-derived MSCs were transplanted into an immunocompetent fetus in the third trimester of gestation [86]. The recipient, diagnosed with osteogenesis imperfecta, showed engraftment of donor cells in bone. Prior to transplant, proliferation of recipient lymphocytes against allogeneic PBL was detected in MLC, confirming immunocompetence of the fetus. By contrast, alloreactivity against donor MSCs was not detected either before or after transplant. In this single patient, the persistence of transplanted cells and the lack of immunoreactivity when patient lymphocytes were re-exposed to the MSC graft in vitro indicate that MSCs can be tolerated when transplanted across MHC barriers in humans.

It is clear that species-specific differences exist that may prevent engraftment of allogeneic or xenogeneic cells. Murine MSCs derived from Balb/c mice do not express alloantigens whereas C57BL/6 mice are positive for MHC class I but not for class II [47, 72, 87]. MSCs derived from both strains are weakly positive for CD80. Transfer of C57BL/6-derived MSCs to Balb/c mice elicited an immune response that was amplified by repeated challenge, resulting in rejection [87]. By contrast, the immortalized C3H10T1/2 (C3) cell line engrafted and formed bone after implantation in four different immunocompetent murine strains [71].

Human MSCs persist to adulthood after infusion in utero to fetal sheep and in the brain of albino rats [25, 88]. Immunological tolerance to mouse MSCs in rats was reported by Saito et al. [89] whereas human MSCs were rejected after infusion into immunocompetent rats [90]. Under these conditions, suppression of the activation of CD4+ and CD8+ T cells by MSCs may not be sufficient in preventing rejection in a discordant xenogeneic system. Immune responses following xenotransplantation include acquired and innate immunity, in which natural antibody, complement, NK cells and macrophages all play independent roles.

In vivo studies of MSCs

Homing of MSCs in animal models

Because MSCs may be used for tissue repair, it is of interest to know into which organs MSCs home. Allogeneic and autologous MSCs distributed to a wide range of tissues in baboons (Fig. 3) [30]. Radiolabelled MSCs injected intravenously in rats showed high radioactivity first in the lung and thereafter in the liver [28]. Low radioactivity was demonstrated in kidneys, spleen and bones. Similarly, human MSCs were shown to engraft in multiple tissues and demonstrate site-specific differentiation after intrauterine transplantation into sheep [25, 91]. In a mouse model of osteogenesis imperfecta, MSCs expressing normal type 1 collagen were infused, engrafted and normal collagen was detected [92]. In healthy experimental animals, MSCs home to most organs.

Figure 3.

 Detection of MSCs after intravenous injection in (a) experimental animals and (b) humans. MSCs transduced with green fluorescent protein distributed to a wide range of tissues, including intestine, pancreas, liver, lung, kidney, urethra, spleen, lymph node, thymus, skin, cerebellum, etc. following systemic infusion into nonhuman primates [30]. Gastrointestinal tissues harboured high concentrations of transgene per microgram of DNA. Kidney, lung, thymus and skin also contained high amounts of DNA. Engraftment ranged from 0.1% to 2.7%. Radiolabelled MSCs injected i.v. in rats showed high radioactivity first in the lung and thereafter in the liver [28]. After i.v. infusion, human MSCs were detected in the circulation in some, but not all, patients within the first hour of infusion, not thereafter [94]. Engraftment of MSCs was demonstrated in five of six patients with severe osteogenesis imperfecta at one or more sites, including bone, skin and marrow stroma [102]. MSC donor DNA was detected in the colon and lymph node at autopsy in a patient treated with MSCs for steroid-resistant GVHD [108].

Clinical studies using human MSCs

Mesenchymal stem cells have been brought into the clinic for several purposes: to differentiate and heal damaged tissues, to promote haematopoietic engraftment after transplant through the secretion of growth factors and for immunosuppression in GVHD. As the immunomodulatory mechanisms differ, for instance, between murine and human MSCs, animal models may not predict the clinical situation. For fear of adverse effects, including ectopic tissue formation and tumour development, severe cases were first considered for treatment. Two feasibility studies were performed by intravenous infusion of autologous human MSCs with no adverse reactions observed [93, 94] (Table 1).

Table 1. In vivo experience of human mesenchymal stem cells (MSCs)
DiseaseNumber of casesSource of MSCsOutcomeDetection of MSCsReference
  1. GVHD, graft-versus-host disease; SCID, severe combined immunodeficiency.

Haematological malignancies15AutologousNo adverse eventsNoLazarus et al. [93]
Breast cancer28AutologousSafe, no side effectsYesKoçet al. [94]
Inborn errors of metabolism11HLA-identicalNo immune response against donorNoKoçet al. [98]
Osteogenesis imperfecta3HLA-identicalNew dense bone formation; few fracturesYesHorwitz et al. [100]
Osteogenesis imperfecta6HLA-identicalEngraftment of gene-marked MSCsYesHorwitz et al. [102]
Osteogenesis imperfecta1HLA-mismatched fetal MSCRegularly arranged bone; few fractures Engraftment of mismatched MSCsYesLe Blanc et al. [86]
Acute myeloid leukaemia1HLA-haploidentical MSCsHaematopoietic engraftment with no GVHDNoLee et al. [103]
Leukaemia, aplastic anaemia, SCID*7HLA-identical or haploidenticalEnhanced engraftment of HSCTNoLe Blanc et al. [104]
Severe aplastic anaemia1Allogeneic MSCsEngraftment, improved stromaYesFouillard et al. [85]
Severe acute GVHD1Haploidentical MSCs twiceClearance of grade IV acute GVHD,PossibleLe Blanc et al. [108]
Hypophosphatasia, Hunter, vasculitis3Allogeneic bone fragmentsStroma cell engraftmentYesCahill et al. [107]
Leukaemia46HLA-identicalSafe, no side effectsYesLazarus et al. [106]
Malignancies, severe acute GVHD9HLA-identical, haploidentical and mismatchedComplete response of severe acute GVHD in 6/8YesRingdén et al. [109]
Malignancies, tissue toxicity10HLA-identical, haploidentical and mismatchedResolution of haemorrhagic cystitis, pneumomediastinum and colon perforationYesRingdén et al. [111]

Mesenchymal stem cells express high levels of arylsulphatase A and α-l-iduronidase [95]. The deficiency of these enzymes leads to failure to hydrolyse a distinct substrate, leading to its accumulation and dysfunction of multiple organs, the most severe being mental retardation. Arylsulphatase A deficiency is the cause of metachromatic leukodystrophy, and α-l-iduronidase deficiency is the cause of Hurler’s disease, disorders that may be prevented by allogeneic HSCT, which is the only potential cure [96, 97]. MSCs were expanded in vitro and given intravenously to patients with metachromatic leukodystrophy and Hurler’s disease, who had previously undergone HSCT to potentiate and enhance enzyme production in patients who still had some symptomatic disease after transplant [98]. In four of five patients with metachromatic leukodystrophy, there was clear evidence of improvement in nerve conduction velocity.

Mesenchymal stem cells may be used to treat bone disorders such as osteogenesis imperfecta [99]. Five patients with osteogenesis imperfecta treated with bone marrow transplantation had donor osteoblast engraftment, new dense bone formation, an increase in total bone mineral content, increases in growth velocity and reduced frequencies of bone fractures [100, 101]. This suggests that HSCT leads to engraftment of functional MSCs. Gene-marked MSCs, in order to identify the cells after infusion, were given to six children who had undergone HSCT for severe osteogenesis imperfecta [102]. Engraftment of MSCs in the bone and an acceleration of growth velocity were demonstrated. We performed in utero transplantation of male fetal HLA-mismatched MSCs to a female fetus with bilateral intrauterine femur fractures, diagnosed with severe osteogenesis imperfecta [86]. A bone marrow biopsy showed 0.3–7.4% Y-chromosome-positive cells by fluorescent in situ hybridization, indicating engraftment of the donor MSCs. The bone was regularly arranged, with configured bone trabecula lined by a columnar layer of normal osteoblasts. The patient has had fewer fractures than expected for a child with severe osteogenesis imperfecta.

Lee et al. reported a patient with acute leukaemia who received a peripheral blood stem cell graft together with MSCs from her HLA-haploidentical father treated with standard immunosuppression [103]. The patient had rapid engraftment with no acute or chronic GVHD and was well 31 months after transplantation. With conventional immunosuppression and a haploidentical nonmanipulated graft, the risk of rejection or severe GVHD is extremely high. MSCs were given in a pilot study together with HSCT in seven patients to enhance engraftment [104]. In three of these patients, MSCs were given because of previous graft failure and re-transplantation. MSCs were given at a dose of 1 × 106/kg and were HLA-identical in three cases and haploidentical in four cases. Neutrophils >0.5 × 109/L and platelets >30 × 109/L were achieved at a median of 12 days. All patients had 100% donor chimerism within 100 days. In one patient, Henoch-Schönlein purpura resolved. This study encourages controlled trials to use MSCs to enhance engraftment after HSCT.

An elderly woman with end-stage severe aplastic anaemia, suffering from pancytopenia with haemorrhages and infections, received MSCs derived from her HLA-haploidentical son on two occasions [85]. Engraftment of donor MSCs was detected by donor chimerism using polymerase chain reaction showing MSCs in the endosteum of a bone marrow biopsy specimen, but not in bone marrow aspirates. This is also in keeping with studies in baboons and rodents indicating that transplanted MSCs are located in the bone tissue rather than in the marrow cavity, and can be detected in bone biopsies but not in marrow aspirates [10, 105]. At the same time as HSCT, 46 patients received culture-expanded MSCs from their HLA-identical sibling donors [106]. MSC infusions were well tolerated [93, 106]. Moderate to severe acute GVHD was observed in 13 (28%) of 46 patients. Chronic GVHD was observed in 22 (61%) of 36 patients and 2-year progression-free survival was 53%. No MSC-associated toxicities were seen. Stromal cell chimerism was demonstrated in 2 of 19 examined patients at 6 and 18 months after transplantation. From this we can learn that MSCs are safe to give, but are difficult to detect after infusion, even in immunocompromised patients who have undergone HSCT.

At the same time as HSCT, three children with potentially fatal diseases, such as hypophosphatasia, Hunter’s disease and vasculitis, had bone fragments implanted intraperitoneally and into bone [107]. Donor osteoblast-like cells were also infused intravenously after transplant. Chimerism analysis of bone biopsies showed 25–60% donor stromal cell engraftment. This study shows that by bone implantation, donor stromal cells engraft.

We transplanted haploidentical MSCs to a patient with severe treatment-resistant grade IV acute GVHD of the gut and liver [108]. The aim was to use the tissue repair effect shown in vivo in animal models, and the immunomodulatory effects seen in vitro on human lymphocytes. The clinical response was striking with normalization of stool and bilirubin on two occasions, with voluminous haemorrhagic diarrhoea and highly elevated bilirubin being the most pronounced signs of GVHD in this patient. Subsequently, we summarized our experience of treating eight patients with steroid-refractory grades III–IV acute GVHD and one patient with extensive chronic GVHD [109]. Acute GVHD disappeared completely in six of eight patients and the survival curve was better than that of 16 patients with steroid-resistant biopsy-proven gastrointestinal GVHD, not treated with MSCs (= 0.03). These studies have been extended and in Europe, 55 patients have been treated for steroid-resistant acute GVHD with an overall response rate of 69% [110]. Nonresponders have died of progressive GVHD and several responders have died from infections with an overall survival of 23 of 55 (42%) from 2 months to 5 years. Although the experience is limited, MSCs seems a promising treatment for severe steroid-resistant acute GVHD.

Ten patients undergoing HSCT were treated with MSCs due to tissue toxicity [111]. In five patients, severe haemorrhagic cystitis cleared after MSC infusion. Gross haematuria disappeared after a median of 3 (1–14) days. Two patients with grade 5 haemorrhagic cystitis had reduced transfusion requirements after MSC infusions, but both died of multi-organ failure. MSC donor DNA was demonstrated in the urinary bladder in one of them. Two patients were treated for pneumomediastinum, which disappeared after MSC infusion. A patient with steroid-resistant GVHD of the gut experienced perforated diverticulitis and peritonitis that was dramatically reversed twice after infusion of MSCs. These preliminary data suggest that MSCs may also play a role in repairing severe tissue toxicity.

Future directions

Much of our knowledge of MSCs is derived from in vitro experiments. Larger clinical trials have just started. Preliminary data suggest that MSCs do not induce transplantation tolerance [108]. Therefore, studies are needed to optimally use MSCs and to combine their use with other immunosuppressive drugs. An MSC clinical expansion consortium has been created within the Developmental Committee of the European Group for Blood and Marrow Transplantation (EBMT). The purpose is to establish clinical MSC transplantation, to standardize the clinical expansion and release criteria of the expanded cells. A cell therapy registry has also been formed within the EBMT database, where both clinical outcome of treated patients as well as characteristics of the cells transplanted can be entered. The registry will include patients treated with cell therapy, other than haematopoietic stem and progenitor cells for any disorder, including cardiology, neurology and rheumatology. Double-blind, randomized studies using MSCs versus placebo have started in Europe and the USA in patients with steroid-resistant acute GVHD. In Europe, a double-blind, randomized study has also started to explore the use of MSCs to enhance engraftment in patients undergoing HSCT from unrelated donors. Studies are also under way in Europe to explore the use of MSCs to enhance engraftment of cord blood transplants and haploidentical haematopoietic transplants.

The healing and immunosuppression by MSCs may be due to their production of growth factors at the site of injury, rather than to the differentiation of the cells in replacing damaged tissue. After having fulfilled this function, it is unclear whether MSCs undergo apoptosis or rejection. In one patient treated with MSCs, MSC donor DNA could not be detected in any organ at autopsy a few weeks after infusion [109]. In another patient receiving MSCs from two donors, donor DNA from both donors was detected in lymph node and colon, target organs of GVHD, within weeks after infusion. More in vivo work is required to increase our understanding of how MSCs reduce inflammatory responses, distribute and differentiate after infusion in humans. Because of the disappointing outcome in patients with severe acute GVHD and the promising results in the pilot trial using MSCs, prospective trials using MSCs are needed in this setting. There are several clinically relevant questions that need to be addressed. What is the optimal timing, cell dose and additional immunosuppressive therapy? Because MSCs only inhibit the afferent phase of alloreactivity in vitro and not cytotoxicity by already developed cytotoxic T cells, additional treatment with anti-T-cell antibodies may be required in some cases [60]. This suggests that MSCs alone will not be sufficient in inducing immunosuppression and that systemic immunosuppressive therapy is also required.

Because of their nonspecific immunosuppressive activities, MSCs may also decrease immunity against viral, fungal and bacterial infections in vivo. For example, in one patient, gastrointestinal GVHD resolved after MSC infusion, but cytomegalovirus-associated gastroenteritis developed [109]. Another concern is that MSCs may harbour herpes viruses that may be transmitted from the donor to target tissues in a severely immunocompromised host. We found that CMV, herpes simplex virus (HSV) varicella zoster virus (VZV) and Epstein–Barr virus (EBV) are not detected in MSCs from herpes virus-seropositive donors [112]. MSCs can be infected by CMV and HSV, but not with EBV in vitro. Furthermore, MSCs may decrease immunity to tumours, such as EBV posttransplant lymphoproliferative disorders (EBV-PTLD), which was seen in one patient [109]. EBV-PTLD is seen in patients with severely impaired T-cell function, such as adoptive stem cell transplant patients given HLA-mismatched T-cell-depleted transplants. Because MSCs are immunosuppressive and may decrease GVHD, it is unknown whether MSCs will increase the risk of leukaemic relapse by abrogating the graft-versus-leukaemia effect [113]. GVHD has an antitumour effect, especially chronic GVHD [113–115]. In patients undergoing HSCT for leukaemia, leukaemic relapse is decreased in patients with GVHD. With increasing grade of acute GVHD, the risk of leukaemic relapse is reduced. However, severe acute GVHD is associated with a high mortality and therefore, the best leukaemia-free survival is seen in patients with mild acute GVHD [116].

Because MSCs have immunomodulatory effects against alloantigens, they may be used to treat rejections of organ allografts, such as kidneys, livers, hearts and intestinal transplants. Consistent with this, a rat cardiac allograft study showed that MSCs home to the site of allograft rejection [117]. It has also been suggested that MSCs may be used in autoimmune inflammatory bowel disease because of their immunomodulatory effects and their capacity for healing damaged gut epithelium [108]. Because of their capacity to differentiate into cartilage, it may be used for bone defects. Prospective randomized studies are now required to evaluate MSCs in patients with osteogenesis imperfecta [100–102]. We need more knowledge before MSCs can be used in autoimmune disorders. Autoimmune encephalitis in a murine model was abrogated when MSCs were administered before, but not after, disease onset [51].

Because MSCs evoke limited immunity, allogeneic MSCs may be used for acute disorders, such as heart infarctions, acute GVHD [108, 109] and acute exacerbations of inflammatory bowel disorders. However, for disorders where MSCs are not needed on an emergency basis, it may be preferred to culture-expand autologous MSCs.

In conclusion, from in vitro experiments, experimental animal models, safety and pilot studies in humans, MSCs have now taken the step into the first prospective clinical trials.

Conflict of interest statement

No conflict of interest was declared.


This study was supported by grants from the Swedish Cancer Society (0070-B05-19XAC, 4562-B02-02XBB, 4562-B05-XCC), the Children’s Cancer Foundation (2000/067, 03/039, 05/007), the Swedish Research Council (K2006-32X-05971-26-1, K2006-32XD-14716-04-1), the Cancer and Allergy Foundation, the Cancer Society in Stockholm, the Tobias Foundation, the Swedish Society of Medicine, the Sven and Ebba-Christina Hagbergs Foundation and the Karolinska Institutet.