SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Objective

Adult mesenchymal stem cells (MSCs) represent promising tools for therapeutic applications such as tissue engineering and cellular therapy. Recent data suggest that, due to their immunosuppressive nature, MSCs may be of interest to enhance allogeneic hematopoietic engraftment and prevent graft-versus-host disease. Using a murine model of rheumatoid arthritis (RA), this study investigated whether the immunosuppressive properties of MSCs could be of therapeutic value to inhibit reactive T cells in autoimmune diseases such as RA.

Methods

In mice with collagen-induced arthritis (CIA), we injected various doses of C3 MSCs at the time of immunization or booster injection, and subsequently evaluated the clinical and immunologic parameters. The immunosuppressive properties of MSCs were determined in vitro in mixed lymphocyte reactions with or without the addition of tumor necrosis factor α (TNFα).

Results

In the CIA model of arthritis, MSCs did not confer any benefit. Both the clinical and the immunologic findings suggested that MSCs were associated with accentuation of the Th1 response. Using luciferase-expressing MSCs, we were unable to detect labeled cells in the articular environment of the knee, suggesting that worsening of the symptoms was unlikely due to the homing of MSCs in the joints. Experiments in vitro showed that the addition of TNFα was sufficient to reverse the immunosuppressive effect of MSCs on T cell proliferation, and this observation was associated with an increase in interleukin-6 secretion.

Conclusion

Our data suggest that environmental parameters, in particular those related to inflammation, may influence the immunosuppressive properties of MSCs.

Mesenchymal stem cells (MSCs) are adult progenitor cells present in the bone marrow that are able to differentiate into several lineages, such as chondrocytes, osteoblasts, tendinocytes, myocytes, and adipocytes (1). Due to their differentiation potential and the ease of expansion, they are largely studied for their use in tissue engineering for bone and cartilage repair. Numerous reports have focused on their phenotypic and functional characterization (for review, see refs. 2 and3) and, in particular, it was shown that MSCs display immunosuppressive capacities. This was first shown in vitro, in experiments in which MSCs were able not only to escape recognition by alloreactive T cells, but also, when added in mixed lymphocyte reactions (MLRs), to suppress the proliferation of T cells (4–6). Our group recently demonstrated that this immunosuppressive effect acts through the generation of CD8+ regulatory T cells (6). In vivo, intravenous injection of MSCs was also shown to prolong graft survival in major histocompatibility complex–mismatched recipient baboons (7).

The immunosuppressive features of MSCs are of clinical relevance in allogeneic transplantation because it is expected that the incidence and/or severity of graft-versus-host disease (GVHD) will be reduced. Consistent with this notion, a recent clinical study showed that infusion of haploidentical MSCs in a patient with severe treatment-resistant, grade IV, acute GVHD resulted in a striking immunosuppressive effect as long as 1 year after treatment (8). Other applications of MSCs, such as in autoimmune diseases, may be of interest because they may be used in total bone marrow transplantation (BMT), which consists of both hematopoietic stem cells and stromal cells (9). In the MRL/lpr mouse model of lupus, it was reported that total BMT resulted in complete absence of the disease at 40 weeks after treatment and increased survival at 1 year (10). When the adherent cells were removed before transplantation, 75% of the mice died within 90 days. These data suggest that use of MSCs may be of potential interest for the treatment of diseases in which suppression of T cell activation is of major importance.

Rheumatoid arthritis (RA) is an autoimmune disease that is characterized by chronic inflammation of the joints. Cumulative evidence suggests that CD4+ T cell–mediated autoimmune responses play a critical role in the pathogenesis of RA, in conjunction with the activity of B cells and macrophages that infiltrate the synovium (11). Tumor necrosis factor α (TNFα) is secreted by monocyte/macrophages and fibroblasts and plays a central role in RA by inducing a cascade of cytokines, including interleukin-1 (IL-1), IL-6, IL-15, IL-18, and granulocyte–macrophage colony-stimulating factor. Recent therapeutic approaches have involved the use of inhibitors capable of blocking either the binding of TNFα or IL-1 to its cell-surface receptors or the response of CD4+ T cells. Three anti-TNF drugs (a chimeric mouse/human and a fully human antibody to TNF, and a TNF receptor–immunoglobulin fusion molecule) have been proven to be effective and safe in appropriate and well-conducted clinical trials and have shown effectiveness in slowing and even arresting the progression of radiographic damage (12).

Therapeutic agents with activity against T cells, including leflunomide, CTLA-4Ig, anti-CD4 antibodies, cyclosporine, tacrolimus, and T cell receptor Vβ-chain vaccination strategies, have also been studied in RA. Combination therapies that include any of these T cell–activation inhibitors in conjunction with non–T cell–specific agents, such as methotrexate, antimalarial agents, or anti-TNF biologic agents, may prove to be the most effective strategies in controlling this complex disease (13). The collagen-induced arthritis (CIA) model, induced with bovine type II collagen (CII) in DBA/1 mice, is the prototype model of RA (14). This model thus provides a suitable tool to investigate strategies aimed at inhibiting the proliferation of the T cell compartment and may provide a preclinical model for treatment evaluation.

The aim of our study was to determine whether the immunosuppressive properties of MSCs could be of therapeutic value in autoimmune diseases, by inhibiting autoreactive CD4+ T cell proliferation. We used the CIA model to explore the effect of MSCs on the disease course. The results show that the injection of MSCs did not produce any beneficial effect on arthritic symptoms. These observations were shown, at least in vitro, to be related to the presence of TNFα, which resulted in reversion of the immunosuppression mediated by MSCs.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Cell culture and transfection experiments.

The murine MSCs C3H10T1/2 (C3) were grown in complete medium, consisting of Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, 100 units/ml penicillin, and 100 μg/ml streptomycin. Transfection of C3 cells was performed according to the standard calcium phosphate coprecipitation procedure. The hygromycin-resistant pREP10 vector expressing the viral IL-10 (vIL-10) gene under control of the Rous sarcoma virus promoter, named pREP10-vIL-10, was kindly provided by Dr. C. Verwaerde (Institut Pasteur, Lille, France). The neomycin-resistant pCMV-Luc+-Neo vector encoding the luciferase gene under control of the cytomegalovirus promoter was a kind gift from Dr. P. Balaguer (15). Stable clones were obtained after selection using either 400 μg/ml of hygromycin or 1 mg/ml of G418 (Invitrogen, Cergy, France).

Quantification of luciferase activity.

The activity of firefly luciferase was quantified in cell extracts obtained from cells or tissues. These cell extracts were prepared using the Luciferase Reporter Gene Assay, High Sensitivity (Roche Diagnostics, Neuilly-sur-Seine, France), carried out according to the supplier's protocol. Luciferase activity was normalized according to the cell number or to the weight of recovered tissue.

Quantification of cytokines by enzyme-linked immunosorbent assay (ELISA).

Secretion of murine IL-1β, IL-2, IL-4, IL-6, IL-10, interferon-γ (IFNγ), and TNFα in culture supernatants was determined by ELISAs (BD Biosciences, Le Pont de Claix, France). Viral IL-10 production was quantified using the ready-SET-Go kit (Clinisciences, Montrouge, France).

Mixed lymphocyte reactions.

MLRs were performed as previously described (6). When necessary, MSCs were added to the MLR to obtain a 300-μl final volume. Whenever tested, recombinant murine TNFα (R&D Systems, Lille, France) was added at concentrations of 50–500 ng/ml. Each experiment was performed in triplicate and repeated at least 3 times.

Animals.

Adult (8–10-week-old) male DBA/1 mice were grown in our animal facilities, and experiments were conducted in accordance with the recommendations of the European Convention for the Protection of Vertebrate Animals Used for Experimentation. Between 5 and 11 mice per group were included depending on the experiment, and experiments were done at least twice.

Induction of arthritis.

Bovine CII was emulsified with an equal volume of Freund's complete adjuvant. Mice were injected at the base of the tail with 100 μl of emulsion containing 100 μg of CII. On day 21, animals received a booster of CII emulsion in Freund's incomplete adjuvant. C3 cells (106 or 4 × 106 cells/100 μl phosphate buffered saline) were injected in the tail vein either on the day of primary immunization or at the time of the booster injection. For the homing experiments, cells were inoculated after arthritis onset (day 32).

Development of CIA was assessed every 2–3 days. Paw swelling was assessed by measuring the thickness of the hind paws using a caliper. The maximal thickness observed in either of the 2 hind paws of each mouse was also determined during the time course of the disease. The clinical score was assessed using the following system: grade 0 = no swelling, grade 1 = ≥0.1 mm increase in paw swelling, grade 2 = ≥0.2 mm increase in paw swelling, grade 3 = extensive swelling (≥0.3 mm) with severe joint deformity, and grade 4 = pronounced swelling (≥0.45 mm) with pronounced joint deformity. After the animals were killed (between day 42 and day 48 after immunization), the hind limbs were collected for radiography and fixed for histology. Radiologic and histologic scoring were performed as described previously (16).

Assessment of in vitro T cell function.

Splenocytes were collected and cultured at a density of 5 × 105 cells/well in 200 μl of RPMI medium, supplemented with 1% autologous serum, 2 mM glutamine, 0.1 mM nonessential amino acids, 20 mM HEPES, 1 mM sodium pyruvate, and 5 × 10−5M 2-mercaptoethanol in 96-well round-bottom plates, in the presence or absence of 5 μg/ml concanavalin A. For proliferation assays, cells were cultured for 3 days, and 1 μCi/well of 3H-thymidine was added for an additional 18 hours of culture. For cytokine production, supernatants from 2 × 106 splenocytes were collected after 24 hours of culture.

Measurement of serum anti-CII antibody levels.

Serum samples were collected on day 45 for the detection of anti-CII IgG1 and IgG2a antibodies by ELISA, as described previously (16). Results were expressed as the ratio of IgG2a antibody levels to IgG1 antibody levels at a 1:200 dilution of serum.

Statistical analysis.

Statistical comparisons were done with the Student's t-test or an impaired Mann-Whitney test to compare nonparametric data for statistical significance. Percentage comparisons were done using the chi-square test. All data were analyzed by the program Instat (Graphpad, San Diego, CA).

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Dose-dependent effects of systemic injection of MSCs on paw swelling in the CIA model.

Since we had previously shown that injection of MSCs could display immunosuppressive effects (6), we wondered whether MSCs could prevent or diminish arthritic symptoms in the CIA model. We first tested the dose of MSCs to determine a potential dose-response effect. We observed that 4 × 106 MSCs, versus 106 MSCs, led to an increase in paw swelling (results not shown). We then investigated whether the time of injection, at immunization or at booster (day 21), may influence the disease course.

We observed a worsening of the symptoms when cells were injected closer to the time of onset of CIA (data not shown). We thus combined the 2 doses of cells and the 2 times of injection in the same experiment. According to the assessment of paw swelling, a statistically significant increase could be observed in the mice receiving the highest dose of MSCs on day 21 (Figure 1). The maximal paw thickness also tended to be worsened in this group, although the differences were not statistically significant (Table 1). A significantly higher incidence of arthritis, together with higher radiologic and histologic scores, was observed in the groups of mice injected with high doses of MSCs on day 21 (Table 1). These data suggest that MSCs do not display any beneficial effect in mice with CIA and tend to worsen clinical symptoms when injected at a high dose and close to the onset of arthritis.

thumbnail image

Figure 1. Effects of dose and time of injection of C3 mesenchymal stem cells on paw swelling in collagen-induced arthritis. The results are representative of 1 of 2 experiments with the number of mice varying from 5 to 10 in each group. ∗ = P < 0.1 versus controls. d0 = day 0; d21 = day 21.

Download figure to PowerPoint

Table 1. Clinical parameters and cytokine levels in treated mice versus control mice with collagen-induced arthritis
 ControlDay 0Day 21
106 C34 × 106 C3106 C34 × 106 C3
  • *

    P < 0.001 versus controls.

  • P < 0.01 versus controls.

  • P < 0.05 versus controls.

  • §

    Cytokine levels were determined in the supernatants of 2 × 106 splenocytes stimulated with concanavalin A for 24 hours. Results are the mean level of cytokines expressed in the supernatant of splenocytes from pooled spleens in each group. IFNγ = interferon-γ; IL-2 = interleukin-2; TNFα = tumor necrosis factor α.

Clinical parameter     
 % arthritis incidence677017*90*83
 Maximal paw thickness, mean ± SD mm2.29 ± 0.412.37 ± 0.362.22 ± 0.042.33 ± 0.352.57 ± 0.48
 Days to arthritis onset, mean ± SD32.5 ± 1.2235 ± 5.5132.0035.22 ± 5.0232.5 ± 1.22
 % with severe radiologic score7366100*8667
 % with severe histologic score505383*7173
 IgG2a:IgG1, mean ± SD0.26 ± 0.340.37 ± 0.790.25 ± 0.550.77 ± 1.130.57 ± 0.83
Cytokine, pg/ml§     
 IFNγ2611,1306284841,652
 IL-2273746363353645
 TNFα611286229163
 IL-1β3142323457
 IL-101168163
 IL-4461046454151

In parallel, we investigated the mechanisms underlying the severity of CIA in this experiment. Because the severity of CIA is reflected by the switch from a Th1 to a Th2 response (17), we measured the levels of IgG2a and IgG1 anti-CII antibodies in the serum. In 3 of 4 groups injected with C3 cells, the IgG2a:IgG1 ratios were higher than in the control group of arthritic mice (Table 1).

We also analyzed the cytokine production by spleen cells to discriminate between a Th1 and Th2 response of effector cells (17). Proinflammatory cytokines (IFNγ, TNFα, IL-1β, and IL-2) were either stable or enhanced in all arthritic groups as compared with those in the control group, but the highest increases were observed in the group receiving 4 × 106 C3 cells on day 21 (Table 1). In all treated arthritic groups compared with the control group, the levels of antiinflammatory cytokines (IL-4 and IL-10) were also enhanced. Nevertheless, the ratio of IFNγ:IL-4 was enhanced by 2 fold (mean values between 8.9 and 10.9 among the treated groups) as compared with the control mice (mean 5.7), which is indicative of a higher Th1 response. Taken together, these data demonstrate the accentuation of the Th1 helper response when MSCs are injected in arthritic mice.

Effects of IL-10–expressing MSCs on CII-induced arthritic symptoms.

Because MSCs do not display any immunosuppression in the CIA model, we investigated whether the use of genetically modified MSCs, expressing an antiinflammatory cytokine, could improve the course of the disease. Since our group and other investigators have already shown the efficacy of the viral form of IL-10 (vIL-10) as a possible candidate for the treatment of arthritis (16, 18–21), we derived clones of C3 MSCs that expressed vIL-10. In MLRs, we assessed the immunosuppressive effect of the highest producer clone (580 pg IL-10/106 cells/24 hours), hereafter referred to as C3 IL-10, by comparing the effect of decreasing concentrations of naive C3 cells with that of IL-10–expressing C3 cells. Using this experimental construct, we found that secretion of IL-10 significantly enhanced the antiproliferative activity of engineered MSCs (Figure 2A).

thumbnail image

Figure 2. Effects of engineered C3 mesenchymal stem cells (MSCs) expressing interleukin-10 (IL-10) on collagen-induced arthritis (CIA). A, Immunosuppressive effect of naive compared with IL-10–expressing C3 MSCs (C3-IL10). The proliferative activity of responder splenocytes in a mixed lymphocyte reaction (MLR) using allogeneic splenocytes (allo) is attributed a 100% value. C3 or C3-IL10 MSCs were added to the MLR at different cell concentrations. Bars show the mean and SD. B, Mean increase in paw swelling (in mm) in the groups of CIA mice receiving naive or IL-10–engineered C3 MSCs on the day of type II collagen immunization, compared with the group of CIA control mice. No statistically significant differences between groups were recorded. The results are representative of 1 of 2 experiments with the number of mice between 6 and 7 in each group.

Download figure to PowerPoint

We then compared the effect of C3 IL-10 cells with that of C3 cells in the CIA model. To this aim, we injected 106 cells in the tail vein of mice at the time of CII immunization. This dose and time of cell injection were chosen to prevent an aggravating effect due to the presence of MSCs, while still permitting significant IL-10 secretion. As shown in Figure 2B, all groups of mice developed arthritis as recorded by the increase in paw swelling. Similarly, no statistically significant difference was observed between the C3- and C3 IL-10–treated mice and the control mice, in neither the clinical score (mean ± SD 5.3 ± 4.23, 8.0 ± 6.29, and 3.6 ± 3.91, respectively), maximal paw thickness (2.20 ± 0.2 mm, 2.41 ± 0.39 mm, and 2.28 ± 0.13 mm, respectively), nor the time to onset of CIA (36.4 ± 5.9 days, 35.6 ± 6.7 days, and 35.7 ± 2.7 days, respectively). The percentages of mice with severe histologic scores (24%, 56%, and 12.5% among the C3-treated, C3 IL-10–treated, and control mice, respectively) and severe radiologic scores (17%, 33%, and 0%, respectively) were even enhanced in the treated groups. Thus, no beneficial effect on the disease could be observed when MSCs were engineered to express IL-10.

Worsening of RA symptoms by intraarticular MSC injection, without homing of MSCs in the joints.

Previous reports have suggested that infiltration of MSCs from bone marrow into the synovium precedes the inflammatory cell accumulation and clinical onset of arthritis (22, 23). We thus wondered whether MSCs injected either systemically or at ectopic sites could migrate to the arthritic joints. To this aim, we developed stable clones of C3 cells expressing the firefly luciferase gene. One clone, C3 Luc-4, was selected according to its high luminescence intensity. The sensitivity of detection of C3 Luc-4 cells in vitro was ∼103 cells, corresponding to 9.7 relative luciferase units. We then checked that the transfection procedure did not interfere with their immunosuppressive properties, and confirmed that C3 Luc-4 cells were able to inhibit T cell proliferation in the same range as naive C3 cells (results not shown).

We then investigated whether MSCs will migrate to the arthritic joints when administered at the onset of CIA (day 32), and whether the route of cell administration can influence the homing of MSCs. We injected the C3 Luc-4 cells either intravenously, intraperitoneally, intramuscularly, or intraarticularly in both naive mice and CIA mice. Arthritis, as measured by the extent of paw swelling (results not shown) and the percentage of mice with severe radiologic and histologic scores (Figure 3B), developed in all of these animals regardless of the route of cell administration. However, the clinical score was higher in mice receiving the C3 Luc-4 cells by intraarticular injection (Figure 3A).

thumbnail image

Figure 3. Assessment of MSC migration to the joints when administered at the onset of CIA (day 32), and effects of route of cell administration. A, Clinical score recorded during the onset of CIA in mice receiving 106 C3 Luc-4 MSCs through various routes of injection. The naive group comprised the pool of nonimmunized mice that received 106 C3 Luc-4 MSCs through various routes of injection on day 32. Arrow indicates the day of MSC injection at the beginning of CIA (day 32). B, Radiologic score expressed as the percentage of mice injected with C3 Luc-4 MSCs and displaying severe scores (score >1), and histologic score expressed as the percentage of mice displaying severe scores (score ≥2). Bars show the mean and SD. C and D, Detection of the luciferase activity in various tissues of naive (C) or CIA (D) mice injected with C3 Luc-4 MSCs through various routes. Data are represented as the relative luciferase units/gm of tissue. ∗ = P < 0.05; ∗∗ = P < 0.01; ∗∗∗ = P < 0.001. IA = intraarticular; IM = intramuscular; IV = intravenous; IP = intraperitoneal (see Figure 2 for other definitions).

Download figure to PowerPoint

Detection of the luminescent cells was performed in cellular extracts from patellae pouches obtained when the mice were killed, but no signal could be observed, suggesting that cells, if not absent, were at least undetectable by this method (<0.1% of injected cells). In contrast, cells extracted from the tissues were detected among all samples tested but at various levels; they were barely detectable in the heart, liver, marrow, and kidney (results not shown) but were frequently recovered from the muscle, lung, spleen, and brain (Figures 3C and D). Tissue distribution seemed to be unaffected by the route of administration but was related to the status of the mice. Thus, the muscle was the main target in arthritic mice, but high luciferase activity was detected in the lung after intravenous injection, which was expected after systemic infusion (Figure 3C). The lung and the muscle were the 2 main targets in naive mice (Figure 3D), with a high number of cells recovered in the muscle after intramuscular cell injection.

Involvement of TNFα in the loss of the immunosuppressive properties of MSCs.

To understand why MSCs do not display any immunosuppressive effects in CIA, we tested the potential role of the inflammatory environment on MSC behavior. We therefore investigated the role of various cytokines (TNFα, IFNγ, IL-1β) and lipopolysaccharide (LPS) on the antiproliferative effect induced by MSCs. No effect was observed when IFNγ and IL-1β were added to the MLR, and only a slight effect was observed with LPS (results not shown). In contrast, although C3 MSCs totally inhibited the allogeneic response of T lymphocytes in MLRs, they were unable to suppress the proliferative response when cultured in the presence of TNFα, since no statistically significant difference was observed compared with the allogeneic response (Figure 4A). TNFα alone had no effect on the allogeneic reaction (results not shown), whereas the reversion of immunosuppression was observed when TNFα was added at 50 or 500 ng/ml. Moreover, T cell proliferation was statistically higher than that observed when MSCs were present alone.

thumbnail image

Figure 4. Reversion of the immunosuppressive effect of C3 MSCs in the presence of tumor necrosis factor α (TNFα). A, The proliferative activity of responder splenocytes in an MLR using allogeneic splenocytes (allo) is attributed a 100% value. C3 MSCs were added to the MLR with or without TNFα at the concentrations of 50 ng/ml (TNF50) or 500 ng/ml (TNF500). ∗∗∗ = P < 0.001. B, Levels of various cytokines in the supernatants of C3 MSCs incubated for 24 hours in the presence or absence of 50 or 500 ng/ml of TNFα (TNF50 or TNF500, respectively). Bars show the mean and SD. See Figure 2 for other definitions.

Download figure to PowerPoint

We then determined the expression profile of cytokines secreted by MSCs in the absence or presence of TNFα. No variations were observed for most of the cytokines tested, since low levels of IL-1β were measured (Figure 4B) and IL-2, IL-4, IL-12, and IFNγ were absent (results not shown). However, the concentrations of IL-6 were greatly enhanced in the presence of TNFα (up to 16-fold increase with the highest dose of TNFα) (Figure 4B). Thus, the reversion of MSC-induced immunosuppression and the high levels of IL-6 in the presence of TNFα might account for the absence of a beneficial effect of MSCs in CIA.

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

The immunosuppressive properties of MSCs have been established (4–6) and, recently, treatment of GVHD with MSCs has been reported (8), suggesting that MSCs have therapeutic potential to inhibit the unwarranted host immune response. In the present study, we investigated whether MSCs might be of therapeutic value in autoimmune inflammatory disorders. We showed that MSCs provide no benefit in the CIA model of arthritis. Indeed, we observed a switch in the behavior of MSCs depending on the inflammatory environment, alloreactivity or autoimmunity, and we showed in vitro that TNFα is responsible for the reversion of the immunosuppressive effect of MSCs, possibly accounting for the lack of amelioration of RA.

CIA is a well-established model for RA that has been used by many investigators to test the effects of various treatments aimed at inhibiting the T cell response (14). We used C3 cells because we have previously shown that they share the immunosuppressive characteristics of primary MSCs (6). Unexpectedly, in the CIA model, no beneficial effect on arthritis was observed at the doses and times of MSC injection tested. Together with the clinical findings, the immunologic parameters tended to reverse, further suggesting that instead of being inhibited, the T cell response was activated. This activation was unlikely the result of CII expression by MSCs that could thus act as a boost, because we have previously demonstrated that naive C3 cells were unable to differentiate in vivo into CII-expressing chondrocytes (24). These findings suggest that environmental parameters might influence the properties of MSCs, since inflammation reverses immunosuppression.

Among the cytokines produced by macrophages and fibroblasts, TNFα has been found to play a pivotal role in RA (12). We thus investigated whether TNFα and/or the subsequent cytokine cascade could influence the immunosuppressive properties of MSCs. We found that in the presence of TNFα, MSCs were unable, at least in vitro, to inhibit the proliferation of allogeneic T cells, and this was associated with an increase in the level of IL-6. In culture supernatants, IL-6 was increased at least 12–16-fold in the presence of TNFα and, although detected in patellae pouches, was barely detectable (between 30 and 150 pg/ml) in the sera of CIA mice. Low levels of systemic IL-6 have already been reported in the CIA model (25). In our culture conditions, C3 cells secreted basal levels of IL-1β, whereas no IL-2, IL-4, or IFNγ was detectable, which confirms previous data (26). These cytokines were not enhanced upon addition of TNFα (results not shown). On the contrary, IL-12 was undetectable with or without TNFα induction, although it has been shown to be expressed by primary human MSCs (26). Whether the increase in IL-6 was sufficient to account for the aggravation of arthritis or whether this cytokine may influence the properties of MSCs still needs to be investigated. Detection of high levels of IL-6 both in RA patients and in a mouse model of arthritis has been reported (27, 28), and IL-6 is one of the cytokines measured as a reflection of arthritis aggravation. These observations suggest that IL-6 plays a role in the worsening of CIA in the presence of MSCs.

Human MSCs have been isolated from synovial membrane (29) and shown to maintain their multilineage differentiation potential in vitro (30) and in vivo in a model of skeletal muscle repair (31). MSCs have also been identified in the synovial fluid of patients with arthritis (32) and in articular cartilage (33). These data suggest that MSCs are present both in normal joints and arthritic joints. Other reports even have suggested that in the presence of TNF, marrow-derived MSCs could accumulate in the synovial membrane and bone marrow and initiate the clinical onset of arthritis (23). We investigated whether exogenously added MSCs could influence the clinical course of arthritis by migrating to the joints and participating in pannus formation. Using luciferase-expressing MSCs, enzymatic activity has been detected in all organs tested, with variations depending on tissue and individual variability, except in the knee joints. It is likely that the sensitivity limit of ∼1,000 cells in the tissue extract is too high to permit the detection of the injected cells. Nevertheless, clinical, radiologic, and histologic parameters were significantly worsened when cells were injected intraarticularly. This suggests that although MSCs were not detected in the patellae pouches, part of the cells may still have contributed to the aggravation of CIA symptoms.

Although we could not detect luciferase-positive MSCs in the knee joints, we wanted to determine whether the homing capacities of MSCs varied depending on the route of injection and the status of mice (arthritic versus naive). In CIA mice, principally targeted tissues were the muscle, lung, spleen, and brain, with the muscle being the first targeted tissue except in the case of intravenous injection, and, to a lesser extent, the marrow, heart, kidney, and liver were targeted. No striking difference was observed in comparison with naive mice, although the lung and muscle were the 2 tissues mainly targeted in this latter group. Similar findings were previously obtained using murine bone marrow adherent cells injected in irradiated mice (34, 35). A recent study in a baboon model demonstrated a less abundant engraftment of ex vivo–expanded MSCs when the recipients were not previously conditioned by lethal total body irradiation (36). Thus, depletion of the hematopoietic compartment by irradiation may help cell engraftment but does not seem to modify the biodistribution. Similarly, when comparing our experiments with those of Pereira et al (34, 35), distribution of infused cells appears to be unmodified in naive and irradiated mice. However, a more precise quantification of infused cells in the various tissues needs to be performed. Our data further suggest that MSCs could distribute in a similar manner and in many tissues, following local or systemic infusion, and this could occur independent of the inflammatory context.

IL-10 is an antiinflammatory cytokine whose beneficial role has been shown in CIA (16, 19, 20). Although IL-10–expressing MSCs have a greater in vitro immunosuppressive potential, when injected into mice, they behave similarly to naive MSCs, since both forms were ineffective in reducing the clinical parameters of CIA. Secretion of IL-10 by the genetically modified MSCs was undetectable in the sera of the injected mice (results not shown). It was previously described that using recombinant adenoviral gene transfer, a significant benefit was associated with a level of 30 ng/ml IL-10 in the animal sera, whereas levels of ∼600 pg/ml had only a minimal effect (16). Although we cannot compare the delivery systems used in both experiments, one may still speculate that the dose of IL-10 necessary to display a beneficial effect has to be in the range of nanograms per milliliter in the bloodstream. Nevertheless, we cannot exclude the possibility that the local delivery of IL-10–engineered MSCs directly into the joints might have a therapeutic effect. Use of human MSCs expressing the soluble TNF receptor II has recently been reported in the SCID model of arthritis induced by anti-CII antibodies, and investigators reported a reduction of arthritic symptoms (Barry F: personal communication). Thus, by blocking TNFα, this therapeutic molecule not only prevents the cytokine cascade responsible for cell proliferation and degradation in the joints, but also possibly preserves the immunosuppressive effect of MSCs.

Our results thus suggest that although nonengineered MSCs seem to be unsuitable for the treatment of inflammatory diseases, use of engineered MSCs could be of interest to effectively deliver a therapeutic molecule. These data need to be confirmed in other autoimmune disease models.

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

We thank A. Pillon and P. Ballaguer (INSERM U540, Montpellier, France) for the kind gift of the pCMV-Luc+-Neo vector, and C. Verwaerde (Institut Pasteur, Lille, France) for the generous gift of the pREP10-vIL-10 vector. We are grateful to Denis Greuet for providing excellent animal care, Michèle Radal (Centre de Recherches et de Lutte contre le Cancer Val d'Aurelle, Montpellier, France) for performing the histologic work, and Dr. Taourel and colleagues at Lapeyronie Hospital in Montpellier for carrying out the radiographic assessments.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES