Several reports have suggested that mesenchymal stem cells (MSCs) could exert a potent immunosuppressive effect in vitro, and thus may have a therapeutic potential for T cell-dependent pathologies. We aimed to establish whether MSCs could be used to control graft-vs-host disease (GVHD), a major cause of morbidity and mortality after allogeneic hemopoietic stem cell transplantation. From C57BL/6 and BALB/c mouse bone marrow cells, we purified and expanded MSCs characterized by the lack of expression of CD45 and CD11b molecules, their typical spindle-shaped morphology, together with their ability to differentiate into osteogenic, chondrogenic, and adipogenic cells. These MSCs suppressed alloantigen-induced T cell proliferation in vitro in a dose-dependent manner, independently of their MHC haplotype. However, when MSCs were added to a bone marrow transplant at a MSCs:T cells ratio that provided a strong inhibition of the allogeneic responses in vitro, they yielded no clinical benefit on the incidence or severity of GVHD. The absence of clinical effect was not due to MSC rejection because they still could be detected in grafted animals, but rather to an absence of suppressive effect on donor T cell division in vivo. Thus, in these murine models, experimental data do not support a significant immunosuppressive effect of MSCs in vivo for the treatment of GVHD.
Allogeneic hemopoietic stem cell transplantation (HSCT)3 is the most efficient treatment for many hematological malignancies and for primary immunodeficiencies. It relies on the elimination of the hemopoietic compartment by high dose chemotherapy and irradiation, and the reconstitution of a new hemopoietic system provided by the donor hemopoietic stem cells (1, 2). However, the transplants also contain mature T cells that can induce graft-vs-host disease (GVHD), a life-threatening complication of allogeneic HSCT (3). Indeed, these donor T cells are strongly activated after the recognition of alloantigens presented by the recipient APCs, and infiltrate several target organs, such as skin, liver, and gastrointestinal tract, where they exert cytotoxic effects. T cell depletion of the transplant markedly reduces the incidence and severity of GVHD. However, it leads to increased incidences of graft rejection (4), relapse of hematological malignancies (5), and severe infections, revealing the important role that donor T cells play in these phenomenon. Thus, a major goal in HSCT is to modulate alloreactivity by administering donor allogeneic T cells without causing GVHD while preserving graft-vs-leukemia and graft-vs infection effects. In this line, mesenchymal stem cells (MSCs) derived from bone marrow (BM) stroma have recently gained attention.
BM is a complex tissue containing hemopoietic stem cells in close contact with stromal cells constituting the BM microenvironment. Stromal cells have been reported to play a critical role in the regulation of hemopoiesis by promoting cell-to-cell interactions, expressing homing receptor, and constitutively secreting growth factors (6, 7, 8, 9). BM stroma of adult contains small numbers of MSCs that can be expanded in culture. First identified for their ability to differentiate into bone cells and adipocytes, further studies have demonstrated that MSCs can also differentiate in chondrocytes, tenocytes, skeletal myocytes, neurons, and cells of visceral mesoderm under appropriate conditions (10, 11). Some studies have shown that MSCs are capable of homing to different tissues and can survive in the long-term after in vivo administration (12, 13). This ability of MSCs to engraft, their particular function in the marrow microenvironment, and their differentiation potential have generated substantial interest as support for enhancing hemopoietic engraftment or as precursor cells for tissue reconstitution (14, 15).
Another great therapeutic potential for MSCs has arisen the observation that they could exert an immunosuppressive effect in vitro. It has been demonstrated that human, nonhuman primate, and mice MSCs can inhibit T cell proliferation induced either in a MLR or by nonspecific mitogens (16, 17, 18, 19). This suppression occurs regardless of the MHC of MSCs, stimulator and responder lymphocytes. The MSCs’ immunoregulatory effect seems at least in part mediated by the production of cytokines (16, 20), such as TGFβ-1and hepatocyte growth factors, and independent of the induction of apoptosis. These immunosuppressive properties of MSCs open attractive possibilities in the field of solid organ or HSCT. The purpose of this study was to set up a preclinical mouse model to evaluate the therapeutic potential of MSCs for preventing GVHD after allogeneic BM transplantation (BMT).
Materials and Methods
C57BL/6 (H-2b) and BALB/c (H-2d) 6- to 12-wk-old female mice were obtained from Iffa Credo. Ubiquitous expression of GFP in C57BL/6 mice was obtained by transgenesis, in which the GFP cDNA was under the control of a chicken β-actin promoter and CMV enhancer (21). Mice were housed in our animal facility and manipulated according to European Economic Community guidelines.
Murine MSC isolation and culture expansion
MSCs were generated from C57BL/6, C57BL/6 GFP+, and BALB/c mice. Mice were sacrificed by cervical dislocation, and femurs and tibiae were removed and cleaned of all connective tissue. BM cells were collected by flushing femurs and tibiae with 1× PBS buffer using a 26-gauge needle, filtrated, and washed twice by centrifugation at 1400 rpm for 7 min in 1× PBS/3% SVF. To initiate the MSC culture, cells were plated in 25- to 75-cm2 flasks (BD Biosciences) at concentration of 5 × 106/ml/cm2 nucleated cells in α-modified Eagle medium (Invitrogen Life Technologies) supplemented with 10% FCS (HyClone), 100 U/ml penicillin, and 100 mg/ml streptomycin (Invitrogen Life Technologies), and incubated at 33°C in a 5% CO2 atmosphere. In some experiments, cultures were supplemented with 1 ng/ml human platelet-derived growth factor (R&D Systems) and human fibroblast growth factor (R&D Systems). After 48 h, nonadherent cells were removed by washing with 1× PBS, and fresh medium was added. Then medium was changed weekly.
When the culture was near confluence, the monolayer cells were washed twice with 1× PBS, then lifted by incubation for 2–3 min at 37°C with a 0.25% trypsin solution containing 0.01% EDTA (Invitrogen Life Technologies). Trypsin was neutralized by the addition of fresh complete medium. The resulting suspension was then expanded by plating at 6000 cells/cm2 in 75-cm2 flasks. Same conditions were used for subsequent passages.
Immunostaining and FACS analysis
The following Abs were used for FACS analysis: CD45 FITC (YW62.3; Beckman Coulter), CD11b PE (M1-70.15; Caltag Laboratories), CD40 PE (1C10; BD Biosciences), CD80 biotin (16.10A1; BD Pharmingen), CD86 biotin (GL1; BD Pharmingen), ScaI FITC (D7; BD Pharmingen), H-2Kb biotin (AF6-88.5; BD Pharmingen), H-2Kd biotin (SF1-1.1; BD Pharmingen), I-Ab biotin (AF6-1201; BD Pharmingen), I-Ad biotin (2G9; BD Pharmingen), CD90.2 PE (30-H12; BD Pharmingen), and CD103 biotin (M90; BD Pharmingen). Twenty thousand events were acquired on a FACSCalibur and analyzed using Flow Jo or CellQuest software (BD Biosciences). Before use, MSC population was phenotyped by flow cytometry. MSCs were identified as negative for CD45 and CD11b. Donor T cells were stained with 2.5 μM CFSE (Sigma-Aldrich) to follow their cycle division in vivo in GVHD experiments.
Mixed leukocyte reaction
MSCs (0.25 × 106 to 2 × 106 cells) were plated in round-bottom 96-well plates (Corning Glass) in a total volume of 0.2 ml of complete α-MEM and maintained 1 day before the T cell proliferation assays. Splenocytes were isolated from C57BL/6 and BALB/c spleens after mechanical dissociation into 1× PBS/3% FCS. Erythrocytes were removed using 1-min NH4Cl 0.84% incubation, followed by two washes in RPMI 1640 supplemented with 10% FCS, 100 U/ml penicillin, 100 mg/ml streptomycin, l-glutamine (2 mM), HEPES (N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid, 25 mmol/L), and 5 × 10−5 M 2-ME (all obtained from Invitrogen Life Technologies). Responding splenocytes (1 × 105) and an equal number of irradiated (20 Gy) allogeneic stimulating splenocytes were added to the MSC culture after removing the complete α-MEM. T cell proliferation assays were performed in a total volume of 0.2 ml of complete RPMI 1640. MSCs were allogeneic or autologous to responder lymphocytes. Triplicate cultures were incubated at 37°C for 4 or 5 days. [3H]Thymidine with a sp. act. of 1 μCi (BD Biosciences) was added in 20 μl of complete RPMI 1640 to each well 18 h before harvest. The cells were harvested automatically on a filter using a Tomtec harvesting machine. The radioactivity was measured by means of a Microbeta liquid scintillation counter (PerkinElmer).
Twenty-four hours after lethal irradiation of BALB/c (8- to 12-wk-old, 8 Gy) recipient mice, they were transplanted with 3 × 106 BM cells from C57BL/6 donor mice. To induce GVHD, 5 × 105 CD3+ cells collected from pooled inguinal, brachial, axillary, and cervical lymph nodes (LN) were added to the BM transplant. The number of T cells in LN cell preparation was evaluated after staining with anti-CD3 PE-labeled mAb (clone 500A2; BD Pharmingen) and flow cytometric analysis, and then adjusted so that each mouse received 105 CD3+ T lymphocytes. Treated mice received either 5 × 105, 3 × 106, or 4 × 106 C57BL/6 MSCs (donor background) in 100–300 μl of 1× PBS. MSCs were administered i.v. 10–15 min before T cells plus BMT. Control groups comprised nongrafted irradiated mice or irradiated mice receiving only BM cells (that did not induce GVHD). The survival and appearance of mice were monitored daily, and body weight was measured twice per week. Kaplan-Meier survival curves were established for each group. When mice suffering from advanced stage disease were sacrificed for histological examination, this event was considered as a death in the survival curve.
After mice death or sacrifice, small and large bowel samples were fixed in 4% formaldehyde solution for several days and embedded in paraffin. For both organs, 5-μm sections were stained with H&E for histological examination. One pathologist analyzed slides in a blinded fashion to assess the intensity of GVHD. GVHD lesions in each bowel sample were graded according to a semiquantitative scoring system described by Hill et al. (22) with minor modifications. Six parameters were scored for the small bowel (villous blunting, crypt regeneration, crypt epithelial cell apoptosis, crypt loss, lamina propria inflammatory cell infiltrate, and mucosal ulceration), and for the large bowel (crypt regeneration, crypt epithelial cell apoptosis, crypt loss, surface colonocyte lesions, lamina propria inflammatory cell infiltrate, and mucosal ulceration). Each parameter was scored as follows: 0 as normal; 1 as focal and rare; 2 as focal and mild; 3 as diffuse and mild; 4 as diffuse and moderate; and 5 as diffuse and severe.
DNA extraction and PCR analysis
Genomic DNA for PCR analysis was prepared from tissues using phenol chloroform extraction. DNA analyses were performed by real-time quantitative PCR (Applied Biosystems). Amplification was performed using manufacturer-provided reagents following the standard recommended amplification conditions (Applied Biosystems). The primers and probe for eGFP were forward primer 5′-TCCGCCCTGAGCAAAGAC-3′ and reverse primer 5′-GAACTCCAGCAGGACCATGTG-3′; the probe labeled with fluorescent reporter was 5′-FAM-CCCAACGAGAAG-MGB-3′. As internal control, endogenous mouse apoliprotein B gene (APO) was also amplified. The primers and probe for APO gene were designed with Primer Express software (Applied Biosystems): forward primer 5′-CACGTGGGCTCCAGCATT-3′, reverse primer 5′-TCACCAGTCATTTCTGCCTTT-3′, and the probe 5′-FAM-CCAATGGTCGGGCACTGCTCAA-TAMRA-3′. The real-time PCR was performed using TaqMan Universal PCR master mix (Applied Biosystems) with 100 ng of DNA in triplicates, and the average threshold cycles (Ct) of the triplicates were used to calculate the fold change between a positive control DNA isolated from culture-expanded murine GFP+ MSCs diluted in GFP− DNA (C57BL/6 mice) and other samples. Ct for APO was used to normalize the samples. Relative quantification was calculated using the comparative Ct method (Applied Biosystems).
Capacity for the differentiation of MSCs
The capacity of MSCs to differentiate along adipogenic, chondrogenic, and osteogenic lineages was assessed, as described elsewhere (23). Briefly, adipogenic differentiation was induced by adding 1-methyl-3-isobutylxanthine (0.5 mmol/L), dexamethasone (1 μmol/L), insulin (10 μg/ml), and indomethacin (0.2 μmol/L) (all from Sigma-Aldrich) to MSC subconfluent cultures. Adipogenesis was measured by the accumulation of neutral lipids in fat vacuoles, stained with oil red O. To stimulate chondrogenic differentiation, MSCs were centrifuged in a 15-ml polypropylene tube and cultured in DMEM-high glucose (Invitrogen Life Technologies) with the following added: insulin (6.25 μg/ml), transferrin (6.25 μg/ml), linolenic acid (5.33 μg/ml), BSA (1.25 mg/ml), pyruvate (1 mmol/L), ascorbate-2-phosphate (0.17 mmol/L), dexamethasone (0.1 μmol/L), proline (0.35 mmol/L), and selenous acid (6.25 ng/ml) (all from Sigma-Aldrich). Differentiation was induced by adding tumor growth factor-3 (0.01 μg/ml; Sigma-Aldrich). Pellets were formalin fixed and embedded in paraffin. For osteogenic differentiation, cells were plated in 6-well culture plates at 3 × 103 cells/cm2 in MSC medium supplemented with dexamethasone (0.1 μmol/L), ascorbic acid (0.05 mmol/L), and glycerophosphate (10 mmol/L). Matrix mineralization was shown using the von Kossa method.
Statview software (Abacus Concepts) was used for statistical analysis. Kaplan-Meier survival curves were established for each group. Value of p is indicated when differences between two groups were statistically significant.
Isolation, in vitro expansion, and phenotypic characterization of MSCs
MSCs were isolated from BM cells of C57BL/6 or BALB/c mice by plastic adherence in long-term culture. BM cells were initially cultured at a concentration of 106 cells/cm2. Each time cells reached confluence, they were detached and replated at 6000 cells/cm2. We noted an initial phase of slow expansion until the fifth passage, because the number of cells increased by 1.8-fold at each passage. Then, between passage 6 and 7, the number of C57BL/6 and BALB/c MSCs increased by 5.8- and 3.5-fold, respectively (Fig. 1,A). At each passage, a cell aliquot was collected to evaluate the gradual enrichment of MSCs in the culture. For this, cells were stained with the hemopoietic lineage CD45 leukocyte-common Ag marker and the CD11b macrophage-specific marker, MSCs being characterized by a CD45−CD11b− phenotype. From the third passage, CD45+CD11b+ cells represented 54.4 and 38.6% of the cultured cells of C57BL/6 and BALB/c origin, respectively (Fig. 1,A). Less than 2% of cells were CD45+CD11b+ at passage 5 for C57BL/6 culture and passage 6 for BALB/c culture. We observed that cell expansion was markedly accelerated when cultures were constituted of >98% of CD45−CD11b− cells. Noteworthily, our procedure of selection of CD45−CD11b− cells from BM cells necessitated 5–7 mo of culture, depending on the strain used. During the whole culture period, MSCs neither expressed the costimulatory molecules CD40, CD80, and CD86, nor MHC class II (data not shown). Furthermore, MSCs are Sca-1+ and Flk-1− (data not shown) and, consequently, differ from BM-derived Flk-1+Sca-1+ MSCs previously described (24). When human human fibroblast growth factor and human platelet-derived growth factor cytokines were added to the cultures, the time necessary to reach >98% of CD45−CD11b− cells was reduced to 103 vs 126 days for C57BL/6 cells, and 127 vs 196 days for BALB/c cells, corresponding to passage 7 (data not shown). Culture-expanded MSCs exhibit a spindle-shaped fibroblastic morphology, as shown in Fig. 1,B. Additionally, passage 8 C57BL/6 MSCs were tested for their ability to differentiate along osteogenic, chondrogenic, and adipogenic lineages. Under appropriate inducing conditions, the culture could display: osteogenic differentiation as seen by matrix mineralization, chondrogenic differentiation as cell pellets exhibited positive safranin O staining with peripheral areas of oblate cells and central chondrocyte-like zone, and adipogenic differentiation evidenced by fat globules (Fig. 1, C–E). Then, after several months of culture, expanded MSCs kept their capacity to differentiate into multiple tissues.
In vitro immunosuppressive effect of MSCs on T cell proliferation
To test the suppressive effect of MSCs on T cell proliferation in vitro, we cultured 105 C57BL/6 or BALB/c splenocytes in the presence of BALB/c or C57BL/6 irradiated splenocytes, respectively. When C57BL/6 MSCs (syngeneic to responding cells) were added to the culture at ratios MSCs/CD3+ cells of 8/1, 4/1, or 2/1, T cell proliferation was inhibited in a dose-dependent manner (88, 77, and 49% inhibition, respectively) (Fig. 2,A). At the 1:1 ratio, the inhibition was less consistent and statistically nonsignificant. This suppressive effect was not dependent on the source of the MSCs, because it also occurred when C57BL/6 MSCs were used in BALB/c vs C57BL/6 MLR (allogeneic to responding cells; Fig. 2,B). Thus, C57BL/6 MSCs inhibit the proliferation of lymphocytes independently of their MHC for a ratio MSCs:CD3+ cells superior or equal to 2:1. We reproduced these experiments with MSCs of BALB/c background. In contrast to C57BL/6 MSCs, MSCs from BALB/c origin had a much weaker immunosuppressive effect, with <25% inhibition at a 1:1 ratio, but also independently of the MLR combination (data not shown). To determine whether MSCs could elicit a proliferative response by stimulating allogeneic lymphocytes, C57BL/6 or BALB/c splenocytes were cultured with allogeneic irradiated MSCs for 4 days at a 1:1 ratio. As shown in Fig. 2, C (BALB/c MSCs) and D (C57BL/6 MSCs), allogeneic irradiated MSCs did not induce T cell proliferative responses. In the same experiment, C57BL/6 and BALB/c splenocytes strongly proliferated when stimulated by irradiated allogeneic splenocytes (BALB/c and C57BL/6, respectively).
Absence of immunosuppressive effects of MSCs in vivo
Due to their strong immunosuppressive properties in vitro, we tested C57BL/6 MSCs for their capacity to prevent GVHD in mice. Lethally irradiated BALB/c females were grafted with 3 × 106 C57BL/6 BM cells. In these conditions, 100% of grafted mice survived in the absence of any sign of GVHD (Fig. 3,A), while all ungrafted animals died before day 15 (data not shown). When 5 × 105 CD3+ T cells from C57BL/6 donor mice were added to the BM transplant, all mice developed severe clinical signs of GVHD shortly after transplantation as attested by hunched posture, dull fur, weight loss (Fig. 3,B), and strong diarrhea, and died within 40 days (Fig. 3,A). We tested the immunosuppressive effect of MSCs in vivo using different MSCs/CD3+ donor T cell ratios (1/1, 6/1, 8/1). Whatever the number of MSCs added to the transplant, mice still developed clinical signs of GVHD, including weight loss (Fig. 3,B), and died with the same kinetics as the GVHD control group (Fig. 3,A). Thus, even at a ratio that provided a strong inhibition of the allogeneic response in vitro, infused MSCs did not prevent GVHD. Furthermore, the severity of GVHD histological signs in the large bowel was identical with the GVHD control mice, while it was decreased in the small intestine only for the higher MSCs:CD3+ cell ratio, as compared with the GVHD control group (Fig. 4; p < 0.01).
The failure of MSCs to suppress GVHD could be due to several factors, such as incomplete inhibition of T cell proliferation or differential effect of MSCs on CD4+ or CD8+ T cells in vivo. In vitro, previous reports suggested that CD8+ cells were the main target cells that did not proliferate in the presence of MSCs after allogeneic stimulation (16, 17, 18, 19). To assess the inhibitory activity of MSCs in vivo, we evaluated their ability to inhibit donor T cell division after their infusion together with donor T cells in GVHD experiments. For this, we first labeled donor T cells from C57BL/6 (Thy-1.1) mice with CFSE before infusion. Spleens and LNs of grafted animals were collected at day 4.5 and 6.5 postgrafting. In the absence of MSCs, CD4+ and CD8+ donor T cells massively and rapidly divided. More than 95% of CD4+ and CD8+ donor T cells present at these time points in the spleen of grafted animals had divided at least once. In the presence of 4 × 106 MSCs, the percentages of divided CD4+ and CD8+ donor T cells were not modified (Fig. 5). Similar observations were made in the LN of grafted animals (data not shown). Thus, MSCs did not inhibit alloreactive donor T cell division in allogeneic BMT setting.
Fate of infused MSCs in GVHD setting
The failure of MSCs to suppress GVHD could also be due to their rejection rapidly after BMT. Alternatively, differential homing of MSCs and T cells in target organs of GVHD or in secondary lymphoid organs could also be responsible for this absence of suppressive effect. We tested these hypotheses in another model of BMT in which MSCs originated from gene-modified (GFP) transgenic mice. We first verified that GFP-MSCs had the same phenotype and in vitro immunosuppressive activity as their wild-type counterpart (data not shown). We then evaluated the persistence and migration of GFP-MSCs by real-time quantitative PCRs, in different organs including target organs of GVHD. At days 2 and 6, BM, spleen, LN, liver, large bowel, and small intestine of grafted animals were collected, and DNA from these different tissues was extracted. At days 2 and 6, donor MSCs were always present in the BM of grafted animals (Fig. 6) and in the lungs, where they were mainly trapped (data not shown). This reveals that after their infusion, MSCs mainly migrated to their original localization. It also indicated that MSCs were not rejected nor destroyed till day 6 posttransplantation. However, although 4 × 106 GFP-MSCs were injected in irradiated mice, only traces of MSCs were detected in target organs of GVHD and in secondary lymphoid organs, some places in which they could exert their immunosuppressive effect (Fig. 6).
The in vitro immunosuppressive properties of MSCs on T cell proliferation have elicited an interest for these cells as candidates to modulate alloreactivity in solid organ transplantation or HSCT. Furthermore, current technology now permits the ex vivo purification and expansion of MSCs under clinical grade conditions (25, 26), while preserving their in vitro immunosuppressive effects (27). This has led to the launching of clinical trials in the field of HSCT.
We aimed at establishing an animal model to evaluate the use of MSCs for the prevention or treatment of GVHD following allogeneic HSCT. We first successfully developed culture conditions that permitted production of high numbers of highly purified MSCs. However, it should be noted that, contrary to human MSCs, the expansion of mouse MSCs was a very long and cumbersome process. The culture of murine MSCs is also frequently contaminated by hemopoietic progenitors that overgrow the cultures (28, 29) and murine MSCs can rapidly differentiate in culture. In our hands, MSCs were cultured under conditions close to those proposed by Peister et al. (30) in terms of medium and FCS concentrations. Despite these optimal conditions, several cultures spontaneously differentiated after some passages. Nevertheless, culture-expanded MSCs that we used in experimental BMT always exhibit a spindle-shaped fibroblastic morphology and could differentiate into osteogenic, chondrogenic, and adipogenic lineages. Additionally, they displayed the usual immunosuppressive properties of MSCs in MLR and failed to elicit proliferation of allogeneic T cells.
Surprisingly, when MSCs and T cells were transplanted, at a ratio that provides a strong inhibition of MLR in vitro, MSCs failed to prevent or even reduce the severity of GVHD as compared with mice receiving T cells alone. Several hypotheses can be raised to explain this result.
First, the immunosuppressive effect of MSCs observed in vitro could be artifactual, only related to in vitro culture conditions. However, there have been previous examples suggesting that mouse MSCs could exert an immunosuppressive activity in vivo. For instance, when B16 melanoma cells and MSCs were s.c. coinjected, MSCs were found at the periphery of the tumor and tumor growth was accelerated (19). Also, Zappia et al. (31) observed that murine MSCs ameliorated experimental autoimmune encephalomyelitis.
Second, the failure of MSCs to suppress GVHD could also be due to their rejection rapidly after BMT. Although described as poor stimulator cells (19, 20), recent publication showed that MSCs were rejected when grafted in nonirradiated allogeneic mice (32). This observation could represent a major drawback for the therapeutic use of MSCs. However, in the field of HSCT, recipients are lethally irradiated and, consequently, not capable to reject allogeneic cells. This is consistent with our observation that infused MSCs could be detected in recipient mice at least 6 days postgrafting. Also, the in vivo life span of in vitro expanded MSCs could be too short or, alternatively, the number of MSCs present in the target tissues of GVHD could be too low to exert an efficient immunosuppressive effect. In our model, we could only detect traces of MSCs in target organs of GVHD and in lymphoid organs. Consequently, MSCs are unlikely to interact with activated T cells. In this line, we observed that MSCs had no effect on donor T cell division neither on CD4+ nor on CD8+ T cells. Additionally, in a murine model of rheumatoid arthritis, Djouad et al. (33), who did not observe any clinical effect after injection of MSCs, could not detect MSCs in articular environment of the knee of grafted animals. This suggests that MSCs need a sufficient local concentration to exert their immunosuppressive effect in vivo. This was the case in a model of experimental autoimmune encephalomyelitis in which the presence of MSCs in lymphoid organs and in spinal cord was associated with the prevention of the disease (31). The problem of the low number of MSCs in target tissues could be circumvented by increasing the number of injected MSCs. However, this is difficult to achieve in mice, because producing high number of MSCs requires several months of culture.
Third, the in vivo inhibitory effect of MSCs could be only restricted to particular pathologies. For instance, Djouad et al. (33) suggested that the absence of clinical effect in their model of rheumatoid arthritis was probably related to high level of inflammation in this pathology, which may overcome the immunosuppressive properties of MSCs in vivo. This is compatible with the absence of clinical effect observed in this study in which a cytokine storm occurs in early stages of GVHD.
It should be mentioned that murine and human MSCs differ in several instances. First, ex vivo expansion of purified human MSCs is a short process, as opposed to murine MSCs. Second, at least in vitro, the immunosuppressive effect of human MSCs is much stronger (1000-fold) than that of murine MSCs (16, 17), rendering MSCs more readily usable in humans than in mice. This probably explains why clinical trials using MSCs to control GVHD have begun in the absence of published preclinical evidence of their in vivo efficacy. In this line, Lazarus et al. (27) reported in a preliminary report a phase I/II dose-escalation study evaluating the toxicity of MSCs infused together with HLA-identical HSCT originating from peripheral blood or BM of the same donor, in 15 patients with advanced hematological malignancies. No significant toxicity was observed, but no information on GVHD was reported 6 years later. In 2002, a second preliminary report realized on 31 patients indicated that the cotransplantation of MSCs together with HSCT from HLA-identical siblings could induce a reduction of acute and chronic GVHD (34). To our knowledge, this study remains unpublished in a peer-reviewed journal. Finally, a recent case report described the possibility to durably treat one patient with severe treatment-resistant grade IV acute GVHD by two injections of MSCs. Interestingly, MSCs came from the mother of the patient and was not related to the donor of blood stem cells (35). However, when immunosuppression was discontinued, the patient developed GVHD (36). Thus, there is still no publication demonstrating the direct effect of MSC on GVHD in humans. Rather, clinical information can only be found in abstracts or case reports that are difficult to interpret.
In conclusion, we could not evidence the therapeutic potential of MSCs for GVHD in mice. Because mouse models of GVHD are generally considered useful for investigating the potential of therapeutics, we thus believe that data from murine models can add up valuable information to quite a controversial field. However, only the results of the ongoing clinical trials will properly assess the therapeutic potential of human MSCs before new studies are started.
We acknowledge Olivier Boyer for critical reading of the manuscript, Jean-Jacques Mazeron et Gilbert Boisserie for irradiation of mice, Pierre Charbord and Lucie Lévêque for technical support, and Brice Chanudet for excellent care of mice. C57BL/6 GFP+ transgenic mice were provided by Prof. Masaru Okabe.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by a grant from Établissement Français du Sang.
Abbreviations used in this paper: HSCT, hemopoietic stem cell transplantation; BM, bone marrow; BMT, BM transplantation; Ct, threshold cycle; GVHD, graft-vs-host disease; LN, lymph node; MSC, mesenchymal stem cell.