The administration of ex vivo culture-expanded mesenchymal stromal cells (MSCs) has been shown to reverse symptomatic neuroinflammation observed in experimental autoimmune encephalomyelitis (EAE). The mechanism by which this therapeutic effect occurs remains unknown. In an effort to decipher MSC mode of action, we found that MSC conditioned medium inhibits EAE-derived CD4 T cell activation by suppressing STAT3 phosphorylation via MSC-derived CCL2. Further analysis demonstrates that the effect is dependent on MSC-driven matrix metalloproteinase proteolytic processing of CCL2 to an antagonistic derivative. We also show that antagonistic CCL2 suppresses phosphorylation of AKT and leads to a reciprocal increased phosphorylation of ERK associated with an up-regulation of B7.H1 in CD4 T cells derived from EAE mice. CD4 T cell infiltration of the spinal cord of MSC-treated group was robustly decreased along with reduced plasma levels of IL-17 and TNF-α levels and in vitro from restimulated splenocytes. The key role of MSC-derived CCL2 was confirmed by the observed loss of function of CCL2−/− MSCs in EAE mice. In summary, this is the first report of MSCs modulating EAE biology via the paracrine conversion of CCL2 from agonist to antagonist of CD4 Th17 cell function.
Multiple sclerosis (MS),3 a chronic inflammatory condition characterized by the infiltration of peripheral immune cells to the CNS, can lead to disability due to major damage to neuronal axons (1). It was long believed that cells of the Th1 lineage were solely responsible through IFN-γ production for the initiation and progression of MS and its murine equivalent: experimental autoimmune encephalomyelitis (EAE) (2). Today, there is a general consensus that CD4 T cells of the Th17 lineage are also meaningfully implicated in EAE onset and, via their production of IL-17 and possibly other cytokines (3), play a major role in the induction of the demyelination process (4). Nevertheless, EAE lesions do not rely only on proinflammatory cytokines but rather on the dynamic migration of inflammatory cells from the periphery to the CNS. More specifically, the critical importance of chemokines during EAE was defined through various neutralization experiments (5). As a result, CCL2 was identified as one of the major chemokines orchestrating the dynamic migration of inflammatory cells to the CNS (5, 6). In line with this observation, the absence of CCR2 on immune competent cells rendered mice resistant to EAE development (7). Taken together, these data strongly suggest the necessity of CCL2/CCR2 interaction for EAE development and progression, in addition to the importance of Th1 and Th17 cytokines in promoting inflammation. Therapeutic interventions that modulate the CCL2/CCR2/Th17 maladaptive pathways would be of interest for patients suffering of MS. Along this line of thought, much effort has been invested in the use of cellular products as a pharmaceutical for immunomodulation and suppression of a pathological autoimmune response. This is best exemplified by the development of protocols for use of ex vivo-expanded autologous T regulatory cells for treatment of autoimmune ailments, including EAE (8). Mining a similar vein, MSCs are the focus of much preclinical and clinical research from regenerative medicine to immunomodulation (9, 10). Their ability to inhibit allogeneic T cell activation has been exemplified by various murine studies in which MSC-based cell therapy proved its clinical use for immunosuppression (11, 12, 13, 14, 15, 16). In particular, the induction of T cell anergy following MSC administration to EAE mice improved their disease score, suggesting profound suppressive properties believed to be achieved via unidentified secreted soluble factors (11, 17). MSCs are known to produce an array of cytokines and chemokines, including CCL2 (18). Furthermore, MSCs secrete a variety of matrix metalloproteinases (MMPs) (19, 20, 21) known to cleave a wide spectrum of target molecules, including CC chemokines (21, 22). Indeed, MMP-mediated cleavage of the four N-terminal amino acids of CCL2 converts CCL2 from an agonist to an antagonist of T cell chemotaxis and activation (21). We have recently reported that MSC-derived CCL2 undergoes paracrine cleavage by MSC-derived MMPs, leading to the generation of the antagonistic form of CCL2 (hereafter known as mpCCL2) with potent suppression of Ig production by Ag-activated CCR2-expressing B lymphocytes (20). The fact that MSCs modulate their chemokine secretome to form anti-inflammatory polypeptides, led us to hypothesize that MSCs may mediate part of their suppression via inhibition of most, if not all, CCR2-expressing immune competent cells. Thus, we propose that MSCs may mediate their in vivo EAE suppressive effects by modulating the biology of CCL2-driven proinflammatory lymphoid cells. Indeed, Th17 CD4 T cells causative of MS/EAE, characterized by a unique CCR4/CCR6 profile (23), are also known to express CCR2 both in humans and mice afflicted with autoimmune diseases (24, 25, 26). We here validate the previous reports that autologous MSCs can serve as a cellular pharmaceutical to attenuate the clinical and pathological manifestation of EAE and further demonstrate that this effect is mediated via MSC-derived CCL2 secretion, its subsequent processing by MMPs to an antagonistic form and its profoundly suppressive effect on proinflammatory Th17 T cells causative of EAE.
Materials and Methods
Female wild-type (WT) BALB/c, C57BL/6, CCL2−/−, and retired breeder mice were purchased from The Jackson Laboratory. Recombinant human MMP1 and actinonin were purchased from Sigma-Aldrich. STAT3 and AKT Abs were purchased from Cell Signaling Technology. ERK Abs were purchased from Santa Cruz Biotechnology. Anti-CCL2-neutralizing Ab and its isotype, rCCL2, CCR2/5 primers as well as IFN-γ, TNF-α, IL-6, and IL-17 ELISAs were purchased from R&D Systems. CD4 T cell enrichment kit was purchased from StemCell Technologies. All FACS Abs and IL-6-neutralizing Ab were purchased from BD Pharmingen. The ProteinChip Ab capture kit was purchased from Bio-Rad. The 5-76 antagonist CCL2 was synthesized by Genecust.
Whole bone marrow from femurs and tibias of WT female C57BL/6 or CCL2−/− C57BL/6 mice was harvested and placed in culture in DMEM containing 10% FBS and antibiotics (complete medium). After a period of 5–6 wk a homogeneous polyclonal population of MSCs appeared then was phenotyped by FACSCalibur Cytometer (BD Biosciences) using R-PE-conjugated anti-CD31, CD44, CD45, CD73, CD90, CD105, MHC class I (MHCI), and MHC class II (MHCII). MSCs plasticity was tested by the induction of differentiation into mesenchymal lineages using adipocytic or osteogenic differentiation media as show previously (9).
Spleens of BALB/c and C57BL/6 mice were harvested and used to generate a hemoglobin-free single cell suspension (splenocytes). BALB/c (105) and C57BL/6 (105) splenocytes were cocultured in round-bottom 96-well plates in triplicates with or without MSC supernatant concentrated 20× using Amicon’s (Millipore) treated with CCL2-neutralizing Ab or isotype control Ab. Three days later, the coculture was centrifuged, and 100 μl of cell supernatant were used for IFN-γ measurement by ELISA.
Surface-enhanced laser desorption ionization-time of flight (SELDI-TOF) analysis
PG20 arrays (ProteinChip Ab capture kit; Bio-Rad) were placed in a bioprocessor (Ciphergen Biosystems), and CCL2 Ab was applied to each spot. Abs were cross-linked to the array surface by applying 30 μl of cross-linking reagent (ProteinChip Ab capture kit; Bio-Rad). Ag capture from conditioned medium (CM) was performed by applying 20 μl of sample per spot, or rCCL2 at a final concentration of 1 ng/μl was applied per spot. Following incubation overnight at 4°C on an agitating plate, arrays were washed once with wash buffer, rinsed twice with PBS, and rinsed twice with 1 mM HEPES. Chips were analyzed on the SELDI Protein Chip System Series 4000 Enterprise Edition (Ciphergen Biosystems) and a protocol created using CiphergenExpress version 3.0.6. Data were acquired with laser intensity set at 3500 nano-joules and focus mass to 8 kDa.
EAE induction and MSC treatments
Purified synthetic peptides of MOG35–55 1 mg/ml (Sheldon Biotech Center, McGill University) was emulsified (1:1 volume ratio) in complete Freund’s adjuvant (Cedarlane Laboratories) containing 4 mg/ml Mycobacterium tuberculosis H35RA (Difco Laboratories) and injected s.c. at the base of the tail. Animals also received pertussis toxin (Sigma-Aldrich) immediately after the s.c. injection by i.p. injections, repeated 2 days later. Mice were clinically scored every 2 days as follow: 0, no disease; 1, floppy tail; 2, hind limb weakness; 3, partial hind limb paralysis; 4, complete hind limb paralysis; and 5, moribund stage. Following the appearance of EAE symptoms, mice were scored for disease severity, stratified, and assigned to separate test groups to obtain equally weighted average disease scores before experimental interventions consisting of i.p. injection of WT or CCL2−/− MSC (2 × 106cells/injection). All mice were bled a month later (day 34) for cytokine analysis and MOG-specific Ab titer.
Generation of mpCCL2 and ex vivo restimulation of splenocytes with MOG35–55
To generate mpCCL2 in vitro, 10 ng of pure recombinant human MMP1 were added directly to 50 μg of pure rCCL2 in PBS for a period of 4 h at 37°C. mpCCL2 or 5-76 was detected by Western blot (WB) using tricine gels as reported previously (20, 27). Both forms were directly used (at the chosen concentration) in assays without further modifications. Spleens from MOG35–55-immunized C57BL/6 mice were collected and mechanically disrupted in complete RPMI 1640 medium. Viable splenocytes were cultured for 3 days with or without MOG35–55 (1 μg/ml) in the presence of rCCL2, mpCCL2, concentrated WT MSC CM (alone or actinonin), or CCL2−/− MSC CM. Supernatants were analyzed for IL-17 by ELISA according to the manufacturer’s instructions. Splenocytes derived from control or MSC-treated groups (sacrificed at day 48) were stimulated with MOG35–55, then their supernatant was analyzed by ELISA for IL-17 and TNF-α. In addition, B7H-1 and CD86 cell surface expression of MOG35–55- stimulated splenocytes was analyzed by flow cytometry. For IL-17 intracellular staining, cells were first labeled with CD4 before fixation for 20 min. After the wash, stained cells were incubated for 30 min with permeabilization buffer containing anti-IL-17A Ab then analyzed by flow cytometry after the last wash.
EAE-derived CD4 T cells purification and Western blotting
Splenocytes from MOG35–55-immunized C57BL/6 mice were used to purify CD4 T cells according to the manufacturer’s instructions. Purity of CD4 T cells was confirmed by flow cytometry. For CCR2/CCR5 expression, RT-PCR was performed on RNA extracted using the AllPrep DNA/RNA Mini kit (Qiagen). For STAT3 phosphorylation (pSTAT3) analysis by WB, whole-cell lysate from enriched CD4 T cells treated with CM derived from WT or CCL2−/− MSCs was separated on 4–20% gradient SDS-PAGE (Invitrogen) and blotted with pSTAT3 or total STAT3 Abs according to the manufacturer’s instructions.
APC assays were performed using peritoneal macrophages collected from retired breeder C57BL/6 mice. Upon binding and washing from nonadherent cells, macrophages were treated with MOG35–55 for 24 h, then washed and fixed using paraformaldehyde before the addition of MOG35–55-specific enriched CD4 T cells under different treatment conditions. Supernatants were collected and centrifuged 72 h later for IL-17 analysis by ELISA. The assay was repeated with the use of anti-IL-6-neutralizing Ab.
Spinal cord histology and immune infiltrate analysis
EAE mice were perfused with 20 ml of PBS before the removal of their spinal cord of sacrificed mice (day 48). For histological analysis, spinal cords were fixed in 10% formalin before luxol fast blue/crystal violet costaining or CD4 immunohistology. Flow cytometry analysis of CD4 T cells was performed as follows. Spinal cords were mechanically disrupted to generate single-cell suspension. Cells were then stained with CD4-PE and analyzed by FACSCalibur Cytometer (BD Biosciences). Results are shown as absolute cell numbers.
B7H.1 induction and in vitro response analysis
EAE-derived CD4 T cells were cultured for 24 h in the presence of media only, rCCL2, mpCCL2, WT MSC CM, or CCL2−/− CM. B7H.1 expression was then assessed by flow cytometry using B7H.1-PE Ab. CD4 T cells stimulated with media only, rCCL2 or mpCCL2 for 15 min where then lysed, and run on a 4–20% gradient gel for phospho-AKT (pAKT) or phospho-ERK (pERK) analysis. Total AKT or total ERK was used as loading control for these WBs. In vitro stimulation of EAE-derived splenocytes with MOG35–55 was performed as explained earlier but in the presence of B7H.1-neutralizing Ab or its appropriate isotype. Three days later, the culture media were collected and analyzed for IL-17 by ELISA.
Values of p were calculated by paired Student’s t test.
A homogeneous MSC population was obtained from C57BL/6 mice after five to six passages in vitro. Their phenotypic analysis by flow cytometry confirmed their expression of CD44, CD73, and CD105, whereas CD31, CD45, CD90, MHCI, and MHCII could not be detected (Supplemental Fig. 1A).4 The same phenotype was obtained for MSCs derived from CCL2−/− mice (data not shown). When cultured in appropriate differentiation media, MSCs gave rise to adipocytes and osteoblasts (Supplemental Fig. 1B).4 To confirm the presence of CCL2 and derivatives in MSC secretome, a SELDI-TOF analysis was performed on WT or CCL2−/− MSC CM (Fig. 1⇓A). As predicted, we were unable to detect CCL2 in the secretome of CCL2−/− MSC CM (Fig. 1⇓A, lower panel). As a positive control (Fig. 1⇓A, top panel), we analyzed pure bacteria-derived bioactive recombinant mouse CCL2. As per manufacturer specification, this rCCL2 with a m.w. of 8525 has an intact QPDA N terminus and also bears a spontaneous C-terminal cleavage. This polypeptide’s MW is consistent with CCL2 1–73, as predicted by best fit with amino acid sequence. Our analysis of this protein gave a MW of 8531 in good concordance with the manufacturer’s specifications. Analyzing MSC CM (Fig. 1⇓A, middle panel), we found that MSCs produce at least four CCL2 variants (MWs of 8521, 8380, 8254, and 8137, respectively). To demonstrate the immunosuppressive property of the secretome of WT MSCs, 20× concentrated MSC CM was added to a two-way MLR using live C57BL/6 and BALB/c splenocytes. A significant decrease in IFN-γ (Fig. 1⇓B) was obtained with CM derived from WT MSCs, whereas the addition of CCL2-neutralizing Ab partially abolished the suppressive effect. These data identify MSC-derived CCL2 (or one of its derivatives) as a direct inhibitor of splenocyte activation as assayed by MLR.
Effect of MMP-cleaved CCL2 on MOG35–55-stimulated splenocytes
Since lymphocyte-derived IL-17 is believed to be a major component of EAE physiopathology, we tested whether full-length rCCL2 and its MMP-processed derivative—mpCCL2—modulated IL-17 production by lymphocytes. First, we confirmed the presence of the cleaved form of rCCL2 following rMMP1 processing (Fig. 1⇑C). Splenocytes harvested from mice symptomatic with MOG35–55-induced EAE were restimulated ex vivo with MOG35–55 in the presence of rCCL2, mpCCL2, as well as CM from WT and CCL2−/− MSCs. A potent reduction in IL-17 was noticed using WT MSC CM along with a slight but significant inhibition with the CCL2−/− MSC CM (Fig. 1⇑D). The addition of actinonin (a broad-action MMP inhibitor) to WT MSC CM completely restored IL-17 secretion to comparable levels as the MOG35–55 stimulation only, suggesting that MMP-processing of CCL2 is essential for its suppressive effects (Fig. 1⇑D). Working with pure rCCL2 protein, we demonstrate that the full-length rCCL2 has no effect on IL-17 production by MOG35–55 stimulated EAE splenocytes, whereas CCL2 in vitro processed by human MMPs (akin to the extracellular processing of CCL2 by MSC-derived MMPs) leads to significant suppression of IL-17 secretion (Fig. 1⇑E). Interestingly, the use of the synthesized CCL2 5-76 antagonist at equimolar concentrations with rCCL2 plus rMMP1 leads to a more robust inhibition of IL-17 from EAE-derived CD4 T cells (Fig. 1⇑E).
Biochemical effect of mpCCL2 on MOG35–55-specific Th17 CD4 T cells
As detailed above, we show that MMP-processed CCL2 either in its recombinant form or secreted by MSCs can suppress IL-17 production by Ag-specific restimulated lymphoid cells. However, the CD4 T cell subset of lymphocytes is critical for development of EAE. Therefore, we here examine the biochemical effect of mpCCL2 directly on purified MOG35–55-specific IL-17-secreting CD4 T cells. It is important to note that Th17 CD4 T cells were characterized in humans to be CCR2+CCR5−, whereas CCR2+CCR5+ are known to be part of the Th1 lineage and thus secrete IFN-γ (24). Our purified CD4 T cells derived from EAE mice expressed CCR2 but were negative for CCR5 as confirmed by RT-PCR (Fig. 2⇓A). Since pSTAT3 is required for IL-17 secretion and CCR2 engagement by CCL2 is known to lead to STAT3 activation, purified EAE CD4 T cells were probed for pSTAT3 following the different stimuli. WT MSC CM completely suppressed pSTAT3 as opposed to CCL2−/− MSC CM (Fig. 2⇓B). To understand the significance of these findings on IL-17 secretion, purified CD4 T cells were cultured in the presence of fixed syngeneic peritoneal macrophages presenting MOG35–55 under the different test conditions. As such, macrophages act as stimulators without the capacity of secreting soluble factors that might interfere with the assay. As expected, both mpCCL2 and WT MSC CM significantly inhibited IL-17 secretion from responder CD4 T cells, whereas CCL2−/− MSC CM promoted it (Fig. 2⇓C). We propose that the IL-17 induction by CCL2−/− MSC may unmask the effect of MSC-derived IL6 (Fig. 2⇓C) since IL-6 neutralization blocked IL-17 production to a comparable level to MOG stimulation only (Fig. 2⇓D). Taken together, these data demonstrate that MOG35–55-specific CD4 T cells exposed to mpCCL2 respond by suppressing pSTAT3 and IL-17 production.
MSCs ameliorate EAE in a CCL2-dependent manner in vivo
To address whether MSC-derived CCL2 is operative in suppressing EAE symptoms in mice, we compared the therapeutic use of WT MSCs and CCL2−/− MSCs in EAE mice. We found that i.p. administration of WT MSCs in symptomatic mice significantly reduced disease score over time, whereas CCL2−/− MSCs had no significant impact akin to the PBS control group (Fig. 3⇓A). Analysis of mouse serum at day 34 post-EAE onset revealed elevated levels of circulating IL-17 and TNF-α in mice given PBS or CCL2−/− MSCs, whereas a significant reduction for both cytokines was observed following the administration of WT MSCs (Fig. 3⇓B). In addition, a 4-fold decrease in anti-MOG35–55 Ab titer was achieved with WT MSCs administration as compared with PBS or CCL2 −/− MSC groups (Fig. 3⇓C).
MSCs block CD4 T cell infiltration to the spinal cord
On the basis of the evidence that MSCs alleviate EAE symptoms, we examined whether there was a neuropathological correlate. The spinal cords of all treated mice were analyzed for histological evidence of inflammation. We found immune infiltrates in the spinal cords of PBS or CCL2−/− MSC-treated EAE mice with clear demyelination as shown by the luxol fast blue/crystal violet costaining (Fig. 4⇓A). Interestingly, immune cells were found surrounding the spinal cord derived from the WT MSC treated group, more specifically in the dura sheath, with no evident spinal cord infiltration (Fig. 4⇓A). CD4 T cell immunohistology revealed a similar pattern (Fig. 4⇓A), and their increased number infiltrating the neuraxis was confirmed by flow cytometry of dissociated neural tissue (Fig. 4⇓B).
In vitro recall response to MOG35–55 in MSC-treated mice
We further tested the response of splenocytes derived from MSC-treated mice to MOG35–55 and documented a strong reduction in the proliferative ability of cells (Fig. 5⇓A) while expressing higher levels of the negative regulatory molecule B7H-1 (Fig. 5⇓B). A significant decrease of CD86 levels were also observed in splenocytes-derived from MSC-treated group (Fig. 5⇓B). In addition, a reduction of both IL-17 and TNF-α from MSC-treated EAE mice was observed as opposed to the remaining groups (Fig. 5⇓C). To confirm the decrease in IL-17 secretion, splenocytes derived from EAE-treated mice were stimulated with MOG35–55 then stained for intracellular IL-17A. Approximately 9% of CD4 T cells were IL-17 positive in the PBS-treated group as opposed to 15% following CCL2−/− MSC administration. In the WT MSC group, ∼2% of CD4 T cells produced IL-17 (Fig. 5⇓D).
EAE-derived CD4 T cells up-regulate B7H.1 following exposure to mpCCL2
Because of the increase B7H.1 expression following restimulation of MSC-treated mice with MOG35–55, we assessed the level of B7H.1 expression following culture of purified CD4 T cells under the different test conditions. We observed a robust up-regulation of B7H.1 expression (Fig. 6⇓A) upon the addition of WT MSC CM or mpCCL2 (61 and 56%) when compared with CCL2−/− MSC CM or rCCL2 (22 and 28%). This mpCCL2-dependent increase of B7H.1 prompted us to analyze the activation status of AKT and ERK that are known to play a role in B7H.1 regulation and expression. Using purified EAE-derived CD4 T cells, pAKT or pERK (Fig. 6⇓B) were assessed following media or rCCL2 or mpCCL2 treatments. Interestingly, we noticed a strong inhibition of pAKT and increased pERK with mpCCL2. To test the potential inhibitory effects of B7H.1 on IL-17 secretion, splenocytes derived from EAE mice were stimulated with MOG35–55 in the presence or absence of B7H.1-neutralizing Abs. We found an up-regulation of IL-17 secretion with mpCCL2, CCL2−/− MSC CM supplemented with mpCCL2, or WT MSC CM upon the addition of B7H.1-neutralizing Abs as compared with isotype control (Fig. 6⇓C). Taken together, these data demonstrate that mpCCL2 suppresses pAKT and activates pERK, which correlates with B7H.1 up-regulation in CD4 T cells. Furthermore, we find that B7H.1 is materially important in suppressing IL-17 production in bystander cells.
Naive CD4 T cells can acquire a Th17 phenotype, characterized by the production of IL-17, upon their TCR engagement in the presence of TGF-β and IL-6 (28). Their expansion could be further enhanced by IL-1β, TNF-α, and IL-23, all capable of triggering inflammation and, under certain circumstances, buttressing a maladapted autoimmune response (29). Interestingly, IL-17 neutralization by Abs (30), soluble receptor constructs (31), the use of IL-17-deficient mice (32), and autovaccination (33) was beneficial in inhibiting autoimmune pathology such as joint destruction in rheumatoid arthritis, diminishing symptoms in inflammatory bowel as well as improve EAE. In addition to IL-17, specific chemokines modulation strategies have also demonstrated encouraging results in blocking the migration of inflammatory cells to inflamed sites, therefore limiting disease progression. Such observations were supported by various studies in which MIP1α (34), CCL2 (35), or RANTES (36) neutralization attenuated the manifestations of autoimmunity. Since Th17 CD4 T cells secrete IL-17 as well as other proinflammatory cytokines and are responsive to CCL2 (37, 38), we hypothesized that the suppression of these lymphocytes via CCL2 antagonism could allow a robust blockade of IL-17-driven inflammation and cellular infiltration of the CNS. In this regard, MSCs were of great interest for two reasons. The study by Zappia et al. (12) demonstrated a significant improvement of EAE disease score following i.v. injection of syngeneic, ex vivo-expanded MSCs via the induction of T cell anergy by an unknown mechanism. Second, the remarkable capacity of MSCs in converting some factors of its own secretome into antagonist molecules via MMP paracrine cleavage—a processed coined “degradomics” (39)—as demonstrated with CCL2, suggests a novel mechanism by which MSCs can modulate inflammation (20). In an attempt to determine whether MSC processing of its own chemokine secretome could be linked to their therapeutic benefit in EAE, we find that a causal link indeed exists. In addition to their mesenchymal plasticity, we have demonstrated that CM collected from MSCs has the capacity to partially block allogeneic in vitro T cell activation as initiated in a two-way MLR, a property that was lost upon the addition of CCL2-neutralizing Abs. We further demonstrate that the MMP-processed form of CCL2 is required for suppression of MLR, in keeping with our prior observation made under similar experimental condition on B cells (20). Interestingly, we find that the secretome of mouse MSCs contains at least four CCL2 variants consistent with unglycosylated forms with MW spanning 8137 to 8521. Taking in to consideration that the predicted MW of fully unglycosylated mouse CCL2 1–73 is 8525, these derivatives are all consistent with N-terminal cleaved variants of unglycosylated CCL2 and possible CCL2 1–69. These data are reminiscent of what was observed in the secretome of human mononuclear cells, where a substantial fraction of CCL2 1-76 was unglycosylated and accompanied by unglycosylated N-terminal truncated variants 5-76, 6-76, and 1-69. Only the 1-76 and 1-69 variants maintain agonist properties, the N-terminal truncated variants, 5-76 and 6-76, are antagonists (40).
This observation of CCL2 effect does not rule out the absence of the other reported soluble suppressive factors (13, 14, 41); however, it reinforces the notion that a direct MSC/T cell contact is not absolutely required for the subsequent induction of suppressive factors as shown with the NO case (42). Indeed, we observed that CCL2−/− MSC CM led to a measurable decrease in IL-17 secretion by CD4 T cells in vitro (yet less than WT MSCs).
From a cell biochemistry perspective, we focused on the interplay of CCL2 and its MMP-processed derivative on CD4 Th17 T cells associated with the immune pathology of EAE. In terms of intracellular T cell signaling, a link between IL-17 production and response to CCL2 is the activation of STAT3, which is known to selectively mediate Th17 differentiation by directly binding to IL-17 promoter (43). We found that WT MSC CM completely abrogated pSTAT3 once added on Th17 CD4 T cells as opposed to CCL2−/− MSC CM, suggesting the active recruitment of a cellular phosphatase as part of the response to mpCCL2. APC assays using fixed peritoneal macrophages presenting MOG35–55 cultured in the presence of EAE-derived CD4 T cells led to a reduction in IL-17 production under WT MSC CM or mpCCL2 conditions, whereas enhancement was obtained if CCL2−/− MSC CM is used. The latter observation was unexpected and could be explained by MSC secretion of IL-6 that could directly affect responding CD4 T cells. This piece of data suggests that the presence of the truncated form of CCL2 binds to Th17 CD4 T cells leading to a dominant-negative effect despite the presence of the proinflammatory signal delivered by IL-6 and possibly others. It has recently been proposed by Haak et al. (3) that IL-17A and IL-17F, as produced by Th17 cells, may not be necessary or sufficient to induce EAE. However, it is possible—even likely—that Th17 cells produce unidentified factors that may play a meaningful role in these cells’ ability to induce, maintain, and transfer encephalitogenic properties. Our interpretation of the sum of data published on the link between Th17 cells and EAE appears to preponderantly favor a causal role for the cells themselves to be involved, but the role of individual factors produced by these cells (IL-17A, IL-17F, and others) remains to be defined. Although CCR2 expression is not specific or restricted to Th17 or Th1 cells, its overlap is sufficient to permit for depletion of pathogenic lymphomyeloid cells. We propose that whichever may be the precise mechanism by which helper Th1 and Th17 helper cells mediate their encephalitogenic properties, these appear to be hobbled by antagonistic CCL2 as produced by MSCs as described here.
The control of Ag-specific T cell activation is regulated through various mechanisms, including costimulation via CD80/CD86, soluble factors such as cytokines, activation-induced cell death, anergy, or through negative signals delivered by regulatory costimulatory pathways (44, 45). In particular, the importance of the negative regulatory molecule B7H.1 and its interaction with the programmed death-1 (PD-1) receptor has been recently highlighted. This B7H.1/PD-1 engagement blocks proliferation and IL-2 secretion, in addition to a robust induction of CD4/CD8 cell cycle arrest (46). Critical roles have been attributed to B7H.1, especially in the EAE model where its neutralization or inhibition resulted in more severe disease score and increased secretion of proinflammatory cytokines (47). Our MOG35–55-stimulated splenocytes derived from WT MSC-treated EAE mice proliferated weakly in vitro compared with CCL2−/− MSC or PBS-treated EAE mice and showed increased expression of B7H.1. This prompted us to study the expression levels of other immune regulatory molecules such as CD80 and CD86 where the latter was down-regulated in the WT MSC group. This unbalance in B7H.1/CD86 expression could explain the poor responsive levels of restimulated splenocytes and their reduced ability to secrete both IL-17 and TNF-α. Such observations correlate well with the study of Zappia et al. (12) because B7H.1 is known to block T cell proliferation and IL-2 secretion. Indeed, up-regulation of B7H.1 during Th1-driven inflammation has been shown to serve as a negative feedback mechanism for controlling pathogenic T cell responses (47). Furthermore, an increase of MOG35–55-specific CD4 T cell responses derived from B7H.1−/− mice was observed reflecting, therefore, the multiple negative regulatory functions of B7H.1 in limiting the expansion and differentiation of naive CD4 T cells, as well as regulating the reactivation of effector CD4 T cells during autoimmunity (46, 47). In support of this claim, we demonstrate that mpCCL2 has the capacity to induce the expression of B7H.1 on CD4 T cells via the inhibition of pAKT and activation of ERK as reported in multiple myeloma cells (48). As such, we believe that the poor in vitro responsiveness of T cells that we observe in our model is due to the inhibitory effects exerted by B7H.1. An additional element of proof was the increased IL-17 secretion of MOG35–55 restimulated splenocytes cultured with WT MSC CM or mpCCL2 following the addition of B7H.1-neutralizing Ab. It is also important to note that both B7H.1 and PD-1 are coexpressed on T and B cells and might therefore allow negative regulation of T/T or T/B cell interactions, hence creating a blockade at the level of T cell help during the stimulation/differentiation of Ag-activated B cells (49). Our observation of a 4-fold decrease of MOG titer in the serum of WT MSC-treated EAE confirms this concept in addition to the possibility that a direct interaction of WT MSC-derived CCL2 with CCR2-expressing CD138+ plasmablasts or plasma cells could lead to their subsequent inhibition (20). Nevertheless, the substantial reduction in CD86 levels could also amplify the poor responses obtained at the cellular and humoral levels as reported previously (45).
We investigated the therapeutic outcome following i.p. injection of WT vs CCL2−/− MSCs in EAE mice with pre-established EAE. We observed a significant and stable improvement of EAE score following WT MSC administration with decreased CD4 T cell infiltration to the CNS. In contrast, the use of CCL2−/− MSCs did not alleviate EAE symptoms and was associated with inflammatory demyelination. This observation complements a recent report finding that MSC suppressive effect in a mouse model of GVHD is dependent upon their response to IFN-γ, production of NO, and secondary enhancement of chemokine secretome (49). Taken together, we can speculate that MSC’s response to IFN-γ, as reported, leads to enhanced CCL2 production in vivo, and presumably its antagonistic derivative, and accounts for the near totality of anti-EAE suppressive effect. If there were supplementary suppressor molecules generated by MSCs such as CCL7, NO, or any of the myriad factors suggested to date (50, 51), these failed to provide any meaningful clinical reversal of EAE disease in mice treated with CCL2−/− MSCs. Reduced plasma levels of both TNF-α (a surrogate marker of Th1 activity) and IL-17 correlated well with the scores since minimal amounts were detected in the serum of WT MSC-treated mice and upon MOG35–55 restimulation in vitro of their splenocytes. The study by Zappia et al. (12) claimed that i.v. administration of MSCs led to T cell anergy because EAE-derived splenocytes failed in responding to MOG35–55 unless IL-2 was added to the system. Our experiments did not reveal any sign of T cell anergy because MOG35–55 stimulation of splenocytes from WT MSC-treated mice proliferated weakly but in a dose-dependent manner. This discrepancy could be due to the different MSC delivery routes and doses administered. Whereas i.v. MSC injection lead to their early accumulation in the spleen and cervical draining lymph nodes and at a later time point in the subarachnoid space of the spinal cord (12), it is difficult to administer >500,000 cells in an i.v. bolus to a mouse because these mice tend to suffer a high rate of sudden death at even modestly higher doses. In contrast, i.p. administration of MSCs allows delivery of doses up to 10,000,000 cells without any noticeable side effect to experimental mice. The effectiveness observed following i.p. administration of MSCs reinforces the model by which they exert their effect via paracrine delivery of suppressive CCL2 acting upon distant disease sites such as CNS, yet we cannot rule out the possibility that a subset of i.p.-delivered MSCs redistribute to spleen, lymph nodes, and subarachnoid space. Furthermore, we cannot exclude the possibility of MSCs recruiting and/or interacting in vivo with unidentified “veto-like” immune inhibitory cells that can directly or indirectly interfere with the immune pathology seen in these EAE mice. It has been shown that chemokine receptor expression profile is linked to the stage of maturation of CD4 cells. Naive CD4 T cells are null for CCR2, yet fully differentiated CD4 cells will express CCR2 (52). We may therefore speculate that MMP-processed CCL2 will inhibit mainly memory/effector CD4 T cells rather than their naive CCR2-null progenitors. Our work and that of Zappia et al. (12) validate the use of syngeneic MSCs to treat EAE mice. We and others (53) have shown that MHC-mismatched allogeneic MSCs are immune rejected by immune competent murine recipients. Therefore, it remains to be seen whether the pharmaceutical use of MHC-mismatched MSCs—as opposed to autologous cells—will be of optimal benefit in patients suffering from MS. In conclusion, we demonstrated the specific capacity of MSCs in regulating Th17 CD4 T cell activation and migration in EAE mice via the production of MMP-processed CCL2. We can speculate that MSCs may serve well for treatment of autoimmune ailments driven by a pathological CCR2+ CD4 helper mechanism.
We thank Denis Rodrigue, Julie Hinsinger, and Micheline Fortin from the histopathology facility at the Institute for Research in Cancer Immunotherapy (Montreal University) for tissue processing and staining.
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.
↵1 This work was supported by Canadian Institute of Health Research Grant MOP-15017 (to J.G.). M.R. is a recipient of a Fonds de Recherches en Santé du Québec (FRSQ) Scholarship, and J.G. is a FRSQ chercheur-boursier sénior.
↵2 Address correspondence and reprint requests to Dr. Jacques Galipeau, Division of Hematology/Oncology, Jewish General Hospital, McGill University, Montreal, Canada 3755 Cote Ste-Catherine Road, Montreal, Quebec, Canada, H3T 1E2. E-mail address:
↵3 Abbreviations used in this paper: MS, multiple sclerosis; CM, conditioned medium; EAE, experimental autoimmune encephalomyelitis; MHCI, MHC class I; MHCII, MHC class II; MMP, matrix metalloproteinase; pAKT, phospho-AKT; PD-1, programmed death-1; pERK, phospho-ERK; pSTAT3, STAT3 phosphorylation; SELDI-TOF, surface-enhanced laser desorption ionization-time of flight; WB, Western blot; WT, wild type.
↵4 The online version of this article contains supplemental material.
- Received November 26, 2008.
- Accepted March 5, 2009.
- Copyright © 2009 by The American Association of Immunologists, Inc.