Infusion of Mesenchymal Stem Cells and Rapamycin Synergize to Attenuate Alloimmune Responses and Promote Cardiac Allograft Tolerance

Authors


* Corresponding author: Hao Wang, hwang1@uwo.ca

Abstract

The inherent immunosuppressive properties and low immunogenicity of mesenchymal stems cells (MSCs) suggested their therapeutic potential in transplantation. We investigated whether MSCs could prolong allograft survival. Treatment involving infusion of MSCs into BALB/c recipients 24 hours after receiving a heart allograft from a C57BL/6 donor significantly abated rejection and doubled graft mean survival time compared to untreated recipients. Furthermore, combination therapy of MSCs and low-dose Rapamycin (Rapa) achieved long-term heart graft survival (>100 days) with normal histology. The treated recipients readily accepted donor skin grafts but rejected third-party skin grafts, indicating the establishment of tolerance. Tolerant recipients exhibited neither intragraft nor circulating antidonor antibodies, but demonstrated significantly high frequencies of both tolerogenic dendritic cells (Tol-DCs) and CD4+CD25+Foxp3+T cells in the spleens. Infusion of GFP+C57BL/6-MSCs in combination with Rapa revealed that the GFP-MSCs accumulated in the lymphoid organs and grafts of tolerant recipients. Thus, engraftment of infused MSCs within the recipient's lymphoid organs and allograft appeared to be instrumental in the induction of allograft-specific tolerance when administered in combination with a subtherapeutic dose of Rapamycin. This study supports the clinical applicability of MSCs in transplantation.

Introduction

Therapeutic application of mesenchymal stem cells (MSCs) has generated particular interest in the field of transplantation, as a potential means to improve graft outcomes and eliminate requirements for lifelong immunosuppression. MSCs are rare multipotent residents of the bone marrow (BM) that can rapidly expand ex vivo without loss of their multilineage differentiation potential (1–6). Of stromal cell origin, MSCs support expansion and differentiation of hematopoietic cells (7–9), as well as migrate to injured tissues and differentiate (10–12). Furthermore, MSCs enhance hematopoietic stem cell (HSC) engraftment after HSC transplantation, as well as serve as precursors of tissue reconstruction (13–17).

Numerous in vitro studies have reported that MSCs are immunoregulatory and alter differentiation, maturation and cytokine secretion profiles of dendritic cells (DCs) (18–20), B cells (21,22) NK cells (23) as well as T cells (24–28). These MSC-mediated effects appear to be independent of classical MHC class I and II restriction and have been attributed to direct cell–cell contact (26,27,29,30) and soluble factors produced by the MSCs (24,31,32). MSCs are considered immunoprivileged due to their low and/or lack of surface expression of MHC class I/II, as well as co-stimulatory molecules (33,34). Detection of viable allogeneic MSCs in immunocompetent recipients emphasizes this low immunogenicity capacity (12,35). In vivo studies have shown that allogeneic MSCs prolonged skin allograft survival in baboons (28), prevented allogeneic tumor cell rejection in mice (36,37), and attenuated graft-versus-host disease (38,39). However, mechanisms underlying effects of MSCs have not been clearly identified and some conflicting results have been reported (33,40,41).

With the aim of using MSCs in a therapeutic protocol involving solid organ transplantation, we currently examined the effects of MSCs in combination with Rapa, in a C57BL/6-to-BALB/c heart allograft model. We found that MSC monotherapy inhibited acute graft rejection and in combination with low-dose Rapa induced donor-specific allograft tolerance. In tolerant recipients, MSCs migrated to the transplanted heart and various lymphoid organs, suggesting that MSCs interact at these sites with immune cells and attenuate their function. Indeed, a high frequency of Tol-DCs and CD4+CD25+Foxp3+T cells was accompanied by the absence of antidonor antibodies. These results suggested that the combination of MSCs+Rapa provides a promising therapeutic strategy in clinical transplantation.

Materials and Methods

Animals

Male adult C57BL/6 (H-2b), BALB/c (H-2d) and C3H (H-2k) mice (Jackson Laboratories, Bar Harbor, MA) were used as donors, recipients and third-party donor controls, respectively. Enhanced green fluorescence protein (eGFP)-transgenic C57BL/6 mice (GFP+C57BL/6; Jackson Laboratories) were used as a source of MSCs for in vivo tracking. Animals were housed in the Animal Care Facility, University of Western Ontario, and handled according to the Canadian Council on Animal Care guidelines.

Preparation of MSCs and Rapamycin for in vivo treatment

MSCs were prepared from C57BL/6, GFP+C57BL/6, BALB/c, and C3H mice for in vivo injection. BM cells were flushed from femurs and tibiae, washed and cultured in low-glucose DMEM supplemented with 2% fetal bovine serum, penicillin/streptomycin (Invitrogen, Burlington, ON), epidermal growth factor and platelet-derived growth factor (both at 10 ng/ml; R&D Systems, Minneapolis, MN). After 48 hours, nonadherent cells were discarded. Upon 90% culture confluence, adherent cells were trypsinized (0.25%, Sigma-Aldrich, Burlington, ON), harvested and expanded by plating at 5000 cells/cm2. After 2–3 similar passages, harvested MSCs were depleted of CD45+ cells using CD45-positive selection (Miltenyi Biotec Inc. Auburn, CA). CD45 cell effluents (MSCs) were phenotypically characterized before i.v. injection. Rapamycin (Wyeth Pharmaceuticals, Markham, ON) was prepared as emulsion in 100% olive oil.

Platelet preparation, characterization and analysis

Platelets (PLTs) from naïve C57BL/6 mice were prepared as previously described (42,43). Briefly, whole blood was collected in the presence of citrate–phosphate–dextrose–adenine (CPDA; Sigma-Aldrich), pooled and separated by two rounds of differential centrifugation. PLTs were resuspended in CPDA-saline (1/100; PBS; Invitrogen) and adjusted to 107 cells/ml. Prior to i.v. injection and within 4 hours of isolation from the donor to minimize loss of MHC class I expression, the leukoreduced PLTs were characterized for purity (>94% PLTs) and MHC class I by flow cytometry on a FACS Calibur (Becton Dickinson, Mountain View, CA) (43,44), after co-staining with fluorescein isothiocyanate (FITC)-conjugated anti-CD61 and phycoerythrin (PE)-conjugated anti-H2Db antibodies (BD Biosciences, Franklin Lakes, NJ and Cedarlane Laboratories Ltd, Hornby, ON, respectively).

Heterotopic cardiac transplantation

Intra-abdominal heterotopic cardiac transplantation was performed as previously described (45,46). In this study, transplant surgery involved the transfer of fully MHC-mismatched hearts from C57BL/6 (H-2b) donors to BALB/c (H-2d) recipients. Heartbeat of the grafts was monitored and evaluated daily by direct abdominal palpation in double-blind fashion to detect the state of cardiac health/rejection (45,46).

Experimental groups

(A) Survival studies:  Heart transplant recipients were randomly assigned to four groups (n = 16): Group 1, untreated; Group 2, Rapa alone (2 mg/kg/day; p.o., POD0-13); Group 3, donor C57BL/6 MSCs (1 × 106, i.v.) 24 hours after heart transplantation; Group 4, combination treatment of MSCs and Rapa. Two more treatment groups received Rapa and infusion of MSCs from (i) syngeneic BALB/c and (ii) third-party C3H MSCs. To look for potential ‘blood transfusion’ effects, two control groups were included involving BALB/c mice (n = 4) receiving equivalent numbers of fresh leukoreduced donor platelets (C57BL/6 PLTs; 1 × 106 i.v.), 24 hours after heart transplantation, either alone, or with Rapa (2 mg/kg/day; p.o., POD0–13).

(B) Antigen-specific tolerance:  Heart recipients treated with C57BL/6-MSCs+Rapa received full thickness skin grafts (1 × 1 cm2) from C57BL/6 or C3H mouse strains on POD100 and monitored daily (n = 4) (45,46). Rejection was defined as scar formation and/or epidermis sloughing.

(C) MSC migration:  Twelve heart recipients were infused with GFP+C57BL/6-MSCs (GFP-MSCs)+Rapa and sacrificed on days 7, 30, or 100 (n = 4). Equivalent numbers of GFP-MSCs were injected into naïve BALB/c (n = 4) and assessed 100 days later.

Graft histology

Heart graft samples fixed in 10% formaldehyde were embedded in paraffin and sectioned at 8 μm for H&E staining (n = 8) (47). The sections were examined and scored in a blinded fashion for severity of rejection by a pathologist (B.G.) (48,49).

Immunohistochemistry

Cryosections embedded in Tissue-Tek O.C.T (Skura Finetek, Torrance, CA), mounted on gelatin-coated slides were stained using an avidin–biotin immunoperoxidase method (Vector Laboratories, Burlingame, CA) (46) (n = 8). Intragraft T-cell infiltration was detected using anti-mouse CD4 and CD8 mAbs (clone YTS 191.1.2, Cedarlane Laboratories Ltd and clone 53–6.7, BD Biosciences, respectively), while antibody deposition was detected with anti-mouse-IgG and IgM antibodies (Cedarlane). Negative controls were performed by omitting the primary antibodies. Antibody reactivity was evaluated on five randomly selected high-powered bright-phase microscope fields of each tissue section obtained from eight animals/treatment groups. Intragraft IgG/IgM deposition were quantified based on density of positively staining of an antibody titration series within a given section (mm2), while intragraft CD4+/CD8+ cells were quantified by counting positively stained cells and dividing by the specific section area (cells/mm2; Empix Imaging, Mississauga, ON) (46).

Determination of donor-reactive antibody and cellular phenotypic expression

For all FACS analysis, cells were analyzed on a FACS Calibur flow cytometer (Becton Dickinson). All FITC-, PE- and CyChrome-conjugated goat or rat anti-mouse antibodies were purchased from BD Biosciences (Franklin Lakes, NJ), Cedarlane Laboratories or eBioscience (San Diego, CA). All flow cytometric analyses were compared to their appropriate control goat or rat Ig isotypes.

Circulating anti-C57BL/6 allogeneic and anti-BALB/c syngeneic IgG/IgM antibodies were evaluated in recipient sera by flow cytometry following incubation with allogeneic donor (n = 8) or syngeneic splenocytes (n = 6) (47). Recipient mouse sera (1:25 dilution in PBS) were incubated with mouse splenocytes, 30 min at 37°C (46). Cells were then incubated with either anti-mouse IgG-FITC or IgM-FITC and co-stained with anti-CD3-PE, 1 hour at 4°C. CD3-gated splenic cell populations were analyzed for FITC-positive fluorescence with MFI indicative of individual alloreactive isotype levels. CD3+T cells were used for alloreactive antibody detection to minimize backgrounds of antibody binding to Fc-receptors of other cell types.

Immediately prior to mouse intravenous injection, MSCs were stained and analyzed by FACS for surface expression of CD45/CD3/CD19/CD40/CD80/CD86/CD14/CD31/CD44/MHC-I/MHC-II/Ly-6G-Sca-1/Flk-1/CD13.

Phenotypic analysis of CD11c+-enriched DC sub-populations (positive CD11c+ microbead selection technology; Miltenyi Biotec Inc., Auburn, CA) from various treatment groups consistently involved double-staining with anti-mouse CD11c-FITC (clone HL3), in combination with PE-labeled anti-mouse surface I-Ab (MHC II), CD40, CD80 or intracellular IL-12 (p70). CD11c+ DC subsets were analyzed for cell percentages positive for the second DC marker, as well as mean fluorescence intensity (MFI).

CD4+CD25+Foxp3+T cells were identified using an anti-mouse Foxp3-PE staining set, along with CD4-FITC and CD25-Cy5 antibodies. Acquisition involved CD4+ gating.

MLR

To assess T-cell function and donor-specificity, ex vivo CD3+ T cells (untouched by negative microbead-based selection removing CD45R/CD49b/CD11b/Ter-119 cells; Miltenyi Biotec; Auburn, CA) serving as responders (5 × 105/well) were incubated for 96 hours in the presence/absence of Mitomycin C-treated (50 μg/ml; Sigma-Aldrich) naïve splenic CD11c DCs (positive selective isolation using CD11c+ microbeads; 5 × 104/well) used as stimulators from naïve C57BL/6, C3H or BALB/c mice (DC:T ratio of 1:10). To test functional capacity of various ex vivo DC populations, various recipient (BALB/c) splenic CD11c+-enriched DCs (treated with Mitomycin C) serving as stimulators (5 × 104/well) were incubated with naïve C57BL/6 splenic CD3+ T cells (5 × 105/well; DC:T ratio of 1:10). MLR replicates were set up such that 72-hour supernatants could be collected for cytokine analysis by ELISA, and 96-hour T-cell proliferations could be monitored. T-cell proliferations were assessed using [3H]-Thymidine (GE Healthcare, Baie d’Urfé, Quebec) incorporation, counted (Wallac Betaplate counter; Turku, Finland) and reported in counts per minute (cpm).

ELISA

Quantitative analyses of IFN-γ, IL-2 and IL-10 were performed by enzyme-linked assays using commercially available kits (R&D Systems; Minneapolis, MN).

Statistical analysis

Cardiac graft survival was reported in terms of median survival time and comparative analysis was accomplished via the Kaplan–Meier cumulative survival method, with survival differences between groups determined using Log-rank (Mantel-Cox) testing. Histological and immunohistological findings were analyzed nonparametrically using ANOVA-on-rank. Statistical comparisons of cytokine levels (ELISA), antidonor antibody levels (FACS), T-cell proliferations (MLR), as well as cell phenotype differences (FACS) were performed using one-way ANOVA. Differences with p-values ≤ 0.05 were considered significant.

Results

Phenotypic characterization of MSCs

To exclude residual hematopoietic cells from influencing in vivo experiments, we used BM-derived MSCs after 2–3 expansion cycles, followed by stringent depletion of CD45+ cells for injection. As demonstrated in Figure 1, adherence to this culture protocol resulted in MSCs negative for CD45, CD3, CD19, but weakly positive for CD31. As expected from their source of origin and the expansion procedure followed, the cultured MSCs expressed significant levels of Flk-1, Sca-1, CD13 and CD44. In addition, the MSCs were found to express moderately low levels of MHC class I, low levels of MHC class II and CD40, but no detectable amounts of surface CD80, in agreement with MSC phenotypes previously described (50,51). The resultant MSCs had a spindle-shaped fibroblastic morphology and maintained their multilineage differentiation potential by maintaining their ability to differentiate into osterogenic, chondrogenic and adipogenic lineages when cultured in various specific differentiation media (data not shown). Thus, these expanded bone-marrow-derived mouse cells possess both the phenotype and differentiation ability associated with MSCs and thus were considered as such.

Figure 1.

Phenotypic analysis of C57BL/6 mouse MSCs by flow cytometry. MSCs were cultured and isolated from the bone marrow of wild-type C57BL/6 or enhanced green fluorescence protein (eGFP)-transgenic C57BL/6 mice. For all in vitro and in vivo experiments, MSCs were collected and used after 2–3 cycles of expansion (the third generation), and depleted of CD45+ cells, to avoid contamination with residual hematopoietic stems cells. The phenotype of MSCs were determined by staining with mAbs against CD45, CD3, CD19, CD31, Flk-1, Scal-1, CD13, CD44, MHC-I, MHC-II, CD40, CD80, analyzed by flow cytometry and indicated with thick black histogram lines. Thin gray histogram lines represent staining with the corresponding isotype controls.

MSCs inhibit acute cellular rejection and in combination with Rapa achieved long-term heart allograft survival

To elucidate whether MSCs alone, or in combination with short-term Rapa treatment, could modulate acute rejection, MSCs from donor C57BL/6 mice were injected into BALB/c recipients 24 hours after heterotopic heart transplantation. Untreated recipients rapidly rejected grafts in 7.5 days (7–8) (Figure 2A) with typical pathological features of acute antibody-mediated rejection (AMR) characterized by intravascular thrombosis and interstitial hemorrhage mixed with acute cellular rejection (CMR; Figure 2B-a) involving intragraft infiltration of CD4+/CD8+ cells (Figure 2C). Rapa monotherapy considerably reduced AMR and increased median graft survival to 16.5 days (15–19) (Figure 2A), but reduced CMR only marginally (Figure 2B-b and C). MSC monotherapy like Rapa doubled heart allograft survival to 14.0 days (12–18) (Figure 2A) and significantly inhibited CMR, but rejection was due to AMR seen as thrombosis and interstitial hemorrhage (Figure 2B-c), and only a few CD4+/CD8+ cell infiltrates (Figure 2C). In contrast, combination therapy achieved long-term allograft survival (100 days, compared to the untreated and monotherapy groups, n = 8; *p < 0.01) (Figure 2A) with normal histology on POD 7 and 30 (data not shown), as well as POD100 (Figure 2B-d and C). In addition, we found that MSCs derived from BALB/c (syngeneic) or C3H (third party) mice attenuated heart graft rejection, and in combination with Rapa also achieved long-term C57BL/6 heart graft survival exhibiting normal histology (data not shown). Interestingly, transfusion of an equivalent number of fresh leukoreduced donor (C57BL/6) platelets—bearing MHC class I on their surface—administered either alone or in the presence of Rapa exhibited median survival times of 9.5 days (9–10) and 18.5 days (18–19), respectively, and thus did not significantly prolong heart graft survival past the corresponding controls of untreated cardiac recipients (vs PLTs), or those receiving Rapa alone (vs PLTs+Rapa). Together these results demonstrated that although MSC monotherapy was capable of significantly slowing acute rejection, neither MSCs nor Rapa alone could prevent outright acute rejection and that only the combination of MSCs+Rapa could achieve allograft tolerance, independent of the MHC-genotype of the MSCs.

Figure 2.

Graft survival and histology after MSCs or MSCs + Rapa treatment. (A) Heart allograft survival. The differences in survival times between recipients treated with MSCs (1 × 106 i.v., POD1) in combination with Rapa (2 mg/kg/day; p.o., POD0–13) were statistically significant compared to the untreated and monotherapy groups (n = 8; Log rank test) (*p < 0.01). Two additional control groups involving BALB/c mice (n = 4) received fresh leukoreduced donor platelets (1 × 106 i.v., POD1) either alone or with Rapa (2 mg/kg/day; p.o., POD0–13). (B) Histology of cardiac allografts in BALB/c recipients. Grafts were harvested at the time of rejection or at POD 100 and evaluated by H&E staining of paraffin sections (n = 8). Heart grafts from untreated (a), Rapa treated (b), MSCs treated (c) recipients and combination of MSCs and Rapa (d) are compared. Arrows indicate intravascular and/or interstitial changes in heart grafts (400×). (C) Immunoperoxidase staining for intragraft CD4+ and CD8+ cells among untreated and treated mice (n = 8) at the time of rejection or at study endpoint (POD100) was analyzed by counting all positively stained cells in the section, divided by the section area examined (cells/mm2; Empix Imaging, Mississauga, ON). Statistical significance as determined by ANOVA on rank is indicated (*p < 0.01, vs untreated group at endpoint; **p < 0.01, combination therapy group on POD100 vs other groups at endpoint).

MSCs inhibit intragraft and circulating alloreactive antibody levels

Based on previous reports of human MSCs modulating syngeneic B cells in vitro (21), we determined whether mouse MSCs modulated the humoral reponse in our in vivo transplant model, by monitoring intragraft antibody deposition and circulating antidonor antibody levels. Our investigations revealed that untreated recipients had extensive deposition of antidonor IgG within transplanted allografts (Figure 3A, B). In contrast, both monotherapies reduced intragraft IgG antibody deposition as measured at rejection endpoint (Figure 3A, B), with MSCs or Rapa alone attenuating intragraft antibody levels with similar efficacy. Most significant were the observations of long-term surviving grafts of combination therapy recipients which had no detectable IgG deposition at the early timepoints of POD 7 and 30 (data not shown), as well as the study endpoint of POD100 (Figure 3A). Similar humoral immune response modulation was observed upon measurement of serum antidonor antibody levels, where both MSCs and Rapa alone were found to significantly downregulate allospecific IgG production when compared to the untreated group at endpoint rejection (Figure 3C). Furthermore, the circulating antidonor IgG levels in long-term surviving recipients remained at low levels over the course of the study, at the early time points of POD8, 15 and 30 as well as on POD100 (Figure 3B), consistent with the absence of IgG deposition in the grafts. In this model, although circulating antidonor IgM antibody levels were lower than IgG levels in all groups, we still found that compared to untreated mice, the serum antidonor IgM levels were significantly decreased in both the MSC monotherapy and MSC+Rapa combination therapy groups at the time of sacrifice (Figure 3B). As expected, both the IgM and IgG anti-recipient (syngeneic) antibody reactivities were extremely low or undetectable (data not shown). Thus, in particular, the significant attenuation of the donor-specific IgM levels indicated a direct effect of the MSCs on B cells, since IgM production is not dependent on T-cell help (52).

Figure 3.

Intragraft antibody deposition and circulating antidonor antibody levels in recipients following heart transplantation. (A) Representative immunoperoxidase staining for IgG and IgM deposition within the heart grafts of untreated (a, b); Rapa (c, d); MSCs (e, f); and MSCs+Rapa (g, h), following staining and microscopic analysis of all animals within each group (n = 8). (B) Intragraft IgG and IgM deposition among untreated and treated mice (n = 8) at the timepoint of rejection or at study endpoint (POD100) graphically presented as the percentage of positively stained areas within a given section. The percentage of positively stained areas within various graft sections was compared for statistical significance using ANOVA on rank as indicated (*p < 0.01, vs untreated group at endpoint; **p < 0.01, combination therapy group on POD100 vs other groups at endpoint). (C) Serum levels of antidonor antibodies in BALB/c recipients. Sera were collected from untreated or treated recipients at the indicated time points (n = 8) reacted with donor strain splenotytes and evaluated for antibody staining by flow cytometry following CD3+-gating. The results are expressed as mean fluorescent intensity (MFI) ± SEM. Statistical significance of the differences in antidonor antibody levels of recipients were all compared using one-way ANOVA (*p < 0.01, vs untreated group; **p < 0.01, combination therapy group at different timepoints vs other groups at endpoint).

MSCs in combination with Rapa induce donor-specific graft tolerance

To study in vitro and in vivo whether the immunologic tolerance induced in the long-term surviving recipients was donor specific, both MLR and skin grafting were performed. The MLR assay revealed that CD3+ T cells from long-term surviving BALB/c mice possessed a low proliferative response against donor C57BL/6 DCs in MLR when compared to T cells of naive BALB/c mice—similar to the low proliferation levels witnessed upon co-incubation of the MSC+Rapa T cells with syngeneic BALB/c DCs—and yet maintained a normal proliferative response against third-party C3H DCs (Figure 4).

Figure 4.

MSCs in combination with Rapa induce donor-specific tolerance. Splenic CD3+ T cells (untouched using negative microbead selection) from either long-term surviving recipients or naïve BALB/c mice were used and studied as responder cells in MLR (n = 4) in the presence or absence of Mitomycin C-treated CD11c+ DCs obtained from naive C57BL/6 (B6) donor, third-party C3H, or syngeneic BALB/c mice used as stimulator cells (DC:T ratio of 1:10). The proliferative responses were assessed by [3H]-Thymidine incorporation, expressed in counts per minute (cpm) ± SEM and compared for statistical significance using one-way ANOVA (*p < 0.01, long-term MSCs+Rapa T cells co-incubated with donor B6-DCs vs long-term surviving T cells co-incubated with third-party C3H-DCs, naïve BALB/c T cells mixed with B6-DCs and naïve BALB/c T cells co-incubated with third-party C3H-DCs. Data shown are representative of four separate experiments.

Full thickness skin grafts from either C57BL/6 donor or C3H third-party donors were transplanted onto long-term surviving recipients on POD100 (n = 4 per group) to further confirm donor-specific tolerance. Tolerant recipients accepted C57BL/6 donor skin grafts for over 100 days, but rejected the C3H third-party skin grafts within 11 ± 1 days (POD111 ± 1). Interestingly, similar long-term skin graft survival (>100 days) was observed in BALB/c long-term surviving heart recipients receiving MSCs from either third-party C3H mice or recipient (syngeneic) BALB/c mice in combination with Rapa (data not shown). Thus, acceptance upon secondary challenge with a skin graft (i.e. antigen specificity) was dependent on the origin of the initial heart graft and was independent of the MHC-genotype of the MSCs.

MSCs block DC maturation and in combination with Rapa increase the frequency of Tol-DCs and CD4+CD25+Foxp3+T cells

MSCs have been shown to inhibit the upregulation of maturation markers on DCs in vitro (18,19). Thus, we investigated whether similar MSC modulation of DCs occurred in vivo. A comparison of splenic CD11c+DCs of the treatment groups versus the untreated group revealed that MSC monotherapy significantly blocked CD11c+DC maturation indicated by decreased CD40, CD80 expression and IL-12 secretion (Figure 5A and B) and concomitantly lowered their allogeneic stimulatory capacity (Figure 5C). However, CD11c+DCs isolated from tolerant mice were the most significantly attenuated in terms of phenotype and functionality (Figure 5A–C), compared to all other groups. In fact, despite previously reported DC-modulating effects of Rapa (53), low dose of Rapa used as monotherapy had little effect on DC phenotype and stimulatory capacity (Figure 5A–C). In addition, upon studying the functional effects on T cells of the various ex vivo DC populations, by assessing cytokine secretion in an MLR assays, CD11c+DCs isolated from MSC monotherapy-treated mice revealed their ability to significantly inhibit IFN-γ and IL-2 secretion (Figure 5D). Further significantly decreased levels of IL-2 and IFN-γ production and, conversely, increased IL-10 cytokine secretion were found in MLR supernatants of CD3+T cells stimulated with DCs isolated from tolerant recipients (Figure 5D).

Figure 5.

Figure 5.

Identification of Tol-DCs in tolerant recipients. DC Phenotype. Splenic CD11c+ DCs isolated from tolerant BALB/c recipients on day 100 or from other groups at the endpoint (n = 16) were double-stained with FITC-labeled anti-mouse CD11c mAb and various PE-labeled anti-mouse CD40, CD80, MHC class II and IL-12 (p70) antibodies. CD11c+-gated DCs were analyzed by flow cytometry for expression of the various DC markers. (A) Representative dot plots of CD80 staining of CD11c+-gated DCs following at least four determinations in four separate experiments. Intensity of CD80 staining is indicated in mean fluorescent units in the top right corner of each dot plot. (B) Expression levels of surface CD40, CD80, MHC class II, as well as intracellular IL-12 on CD11c+-gated DCs (n = 4) were analyzed by flow cytometry and expressed in terms of MFI ± SEM (bars) and in percentages of positively stained cells (◆), with values compared for statistical significance using one-way ANOVA (*p < 0.05, vs untreated group; **p < 0.01, combination therapy group on POD100 vs other groups at endpoint except naïve control). The results are representative of four independently performed experiments. (C) DC Function. MLR involving CD11c+ DCs isolated from various BALB/c cardiac transplant recipients as effectors. Splenic CD11c+ DCs were isolated from untreated, monotherapy-treated or tolerant MSC + Rapa-treated BALB/c recipients (n = 4) and were used as stimulators after Mitomycin-C treatment. Splenic CD3+ T cells from naïve allogeneic donor C57BL/6 (solid black bars), third-party C3H (light gray bars), or syngeneic BALB/c (dark gray bars) were used as responders, and studied in combination with DCs or alone as a control (DC:T ratio of 1:10). Results were expressed in counts per minute ± SEM (cpm) and compared for statistical significance using one-way ANOVA (*p < 0.05, MSCs alone vs untreated group; **p < 0.01, combination therapy group on POD100 vs all other transplant recipient DCs at endpoint). Data shown are representative of four separate experiments. (D) Tolerant DCs promote Th2 differentiation. C57BL/6 splenic CD3+ T cells were isolated and incubated with CD11c+ DCs derived from spleens of long-term surviving BALB/c recipients, and untreated or monotherapy-treated allograft-rejecting BALB/c mice (n = 4). After 72 h incubation, supernatants were harvested and cytokine levels were performed by ELISA. Mean cytokine concentrations (ng/ml ± SEM) were compared for statistical significance using one-way ANOVA (*p < 0.05, vs untreated group, **p < 0.01, combination therapy group on POD100 vs other groups at endpoint) and are representative of four separate experiments.

Figure 5.

Figure 5.

Identification of Tol-DCs in tolerant recipients. DC Phenotype. Splenic CD11c+ DCs isolated from tolerant BALB/c recipients on day 100 or from other groups at the endpoint (n = 16) were double-stained with FITC-labeled anti-mouse CD11c mAb and various PE-labeled anti-mouse CD40, CD80, MHC class II and IL-12 (p70) antibodies. CD11c+-gated DCs were analyzed by flow cytometry for expression of the various DC markers. (A) Representative dot plots of CD80 staining of CD11c+-gated DCs following at least four determinations in four separate experiments. Intensity of CD80 staining is indicated in mean fluorescent units in the top right corner of each dot plot. (B) Expression levels of surface CD40, CD80, MHC class II, as well as intracellular IL-12 on CD11c+-gated DCs (n = 4) were analyzed by flow cytometry and expressed in terms of MFI ± SEM (bars) and in percentages of positively stained cells (◆), with values compared for statistical significance using one-way ANOVA (*p < 0.05, vs untreated group; **p < 0.01, combination therapy group on POD100 vs other groups at endpoint except naïve control). The results are representative of four independently performed experiments. (C) DC Function. MLR involving CD11c+ DCs isolated from various BALB/c cardiac transplant recipients as effectors. Splenic CD11c+ DCs were isolated from untreated, monotherapy-treated or tolerant MSC + Rapa-treated BALB/c recipients (n = 4) and were used as stimulators after Mitomycin-C treatment. Splenic CD3+ T cells from naïve allogeneic donor C57BL/6 (solid black bars), third-party C3H (light gray bars), or syngeneic BALB/c (dark gray bars) were used as responders, and studied in combination with DCs or alone as a control (DC:T ratio of 1:10). Results were expressed in counts per minute ± SEM (cpm) and compared for statistical significance using one-way ANOVA (*p < 0.05, MSCs alone vs untreated group; **p < 0.01, combination therapy group on POD100 vs all other transplant recipient DCs at endpoint). Data shown are representative of four separate experiments. (D) Tolerant DCs promote Th2 differentiation. C57BL/6 splenic CD3+ T cells were isolated and incubated with CD11c+ DCs derived from spleens of long-term surviving BALB/c recipients, and untreated or monotherapy-treated allograft-rejecting BALB/c mice (n = 4). After 72 h incubation, supernatants were harvested and cytokine levels were performed by ELISA. Mean cytokine concentrations (ng/ml ± SEM) were compared for statistical significance using one-way ANOVA (*p < 0.05, vs untreated group, **p < 0.01, combination therapy group on POD100 vs other groups at endpoint) and are representative of four separate experiments.

Since CD4+CD25+Foxp3+ proportions had previously been reported to increase upon co-culturing with MSCs in vitro (24), we examined whether CD4+CD25+Foxp3+ cells may have contributed to the heart allograft acceptance we had witnessed in vivo. Slight increases in the splenic CD4+CD25+Foxp3+T cell populations of both the Rapa and MSC monotherapy groups were observed (Figure 6). In contrast, tolerant recipients demonstrated a significant increase in CD4+CD25+Foxp3+T cell frequency compared to all other groups (Figure 6). Similar patterns of enhanced Tol-DC and CD4+CD25+Foxp3+ frequencies were witnessed in the lymph nodes of tolerant animals (data not shown).

Figure 6.

Frequency of CD4+CD25+Foxp3+ T cells generated in tolerant recipients. Splenocytes were harvested from long-term surviving BALB/c recipients, and untreated or monotherapy-treated allograft-rejecting BALB/c mice. (A) Representative dot plots of CD25 and Foxp3 staining of CD4+-gated cells following at least four determinations in four separate experiments. Percentage of CD4+CD25+Foxp3+T cells is indicated in the top right corner of each dot plot. (B) Frequency of splenic CD4+CD25+Foxp3+T cells (n = 4) were analyzed by flow cytometry and expressed graphically in terms of mean percentages (± SEM) that were statistically compared using one-way ANOVA (*p < 0.01, vs all other groups). Results are representative of four independent experiments.

Migration of infused MSCs in tolerant recipients

To facilitate the visualization of MSCs and follow their migration upon infusion into cardiac recipients, the source of MSCs was changed from wild-type C57BL/6 to GFP+C57BL/6 mice. GFP expression did not affect morphology, phenotype and in vitro immunosuppressive activity of the GFP-MSCs (data not shown), in agreement with documented eGFP-transfection studies of MSCs (33,54). As demonstrated in Figure 7, following administration of GFP-MSCs in combination with Rapa into cardiac transplant recipients, we detected substantial numbers of GFP-MSCs in the thymus by day 7, but few in the bone, spleen, lymph nodes and heart graft of the recipients. Increased numbers of GFP-MSCs were detected in the bone, spleen and lymph nodes by POD30 and POD100. As compared to spleen and lymph nodes, the thymus had dramatically increased numbers of GFP-MSCs on POD7, 30 and 100. In bone tissue, GFP-MSCs were present in the osseous and BM. Importantly, cardiac muscle and blood vessels of the grafts contained MSCs by POD30 that increased by POD100. However, GFP-MSCs were not found in native hearts, kidneys, livers and lungs of the recipients (data not shown). In naïve recipients, minor MSC engraftment occurred but without increases in number over a 100-day period (data not shown).

Figure 7.

GFP-MSCs migrate into central and secondary lymphoid organs as well as heart allografts. MSCs expanded from GFP+C57BL/6 mice were injected intravenously (1 × 106 cells) into cardiac allograft BALB/c recipients in combination with short-term Rapa treatment and monitored over time for migration of the GFP-MSCs (n = 4). The animals were sacrificed on POD 7 (A, D, G, J, M), 30 (B, E, H, K, N) and 100 (C, F, I, L, O, P). Organs were removed, washed in PBS, and then embedded in Tissue-Tek OCT compound. Sections were cut at 8 μm on a cryostat and cells with GFP-fluorescence were detected using a fluorescence microscope. Bone (A, B, C), thymus (D, E, F), spleen (G, H, I), lymph node (J, K, L), heart graft (M, N, O, P). 200× magnification was used for microscopy.

Discussion

In this current study, adoptively transferred MSCs attenuated acute allograft rejection and synergized with Rapa to promote cardiac allograft tolerance in vivo. To our knowledge, this is the first report of a combination therapy involving MSCs altering the immune response through interaction and regulation of immune cells in a solid organ transplant model. The importance of our observations is strengthened by the fact that long-term graft survival and tolerance observed upon MSC+Rapa therapy was (a) antigen-specific and (b) was achieved regardless of MHC origin of the MSCs.

Firm establishment that cardiac transplant tolerance observed following MSCs+Rapa treatment was MHC class I independent, was further demonstrated in this study when leukoreduced PLTs, expressing MHC class I, were administered in an intravenous transfusion regimen mimicking that of the MSCs in both cell number and timing. Neither PLTs alone, or in combination with Rapa, significantly prolonged heart graft survival as achieved by MSCs alone or MSCs+Rapa in our study—confirming that immunoregulation exerted by MSCs cannot be simply attributed to a ‘blood transfusion’ effect—previously correlated with MHC class I expression alone (42,43,55). Since our PLT-transfusion regimen mirrored our current MSC protocol in terms of a single injection of a limited cell number, as well as in terms of timing of administration at 24 hours postcardiac transplantation, our results likely differed from the previously described 4-week transfusion protocol of 100-fold more PLTs per injection prior to skin transplantation (43). Indeed, MHC class I independence of the MSC immunomodulatory effects witnessed in our study, agrees with previous reports of their MHC independence (26,27,29,30,33,34).

Despite the current noted benefit of MSCs as a monotherapy, administration of MSCs alone was not sufficient to completely inhibit the allogeneic response. Similarly, despite the observed benefit of Rapa monotherapy in attenuating allograft rejection, use of low-dose Rapa was unable to permanently maintain the heart graft. Only combination therapy of MSCs+Rapa synergized to induce graft tolerance. The currently witnessed synergistic success of MSCs+Rapa in achieving antigen-specific tolerance was in contrast to findings that MSCs failed to prolong rat cardiac allograft survival when used in combination with low-dose CsA (56). An important distinction between protocols is that initiation of Rapa immunosuppression in our study on POD0, versus use of CsA on POD3 (56), could provide necessary time for MSCs to engraft and mediate their suppressive function. In addition, it has been well documented that CsA inhibits Treg generation, while Rapa enhances Treg production (47,53,57). Thus, Rapa has greater potential to synergize with MSCs to support Treg generation (58,59).

In our study, the synergistic benefits of MSCs+Rapa therapy were multi-faceted with modulation of the immune system at the levels of DCs, T and B cells. Indeed, immunogenicity of the CD11c+DC population was significantly reduced in terms of phenotype and function, with DCs exhibiting multiple tolerogenic properties ex vivo—similar to MSC effects on DCs observed previously in vitro (18,19,24). In addition, DCs from the combination therapy group appeared to modulate a T-effector cell bias in favor of a Th2 response, reducing IL-12 production and enhancing IL-10-–a cytokine shift previously observed upon MSC administration (24) and associated with tolerance in transplantation (60,61). In addition to the contribution of Tol-DCs and CD4+CD25+Foxp3+T cells to development of a tolerogenic state (62–66), it is likely that reduced antidonor humoral responses also played a role (67,68). Thus far, MSC modulation of B cells has only been reported in vitro, where human MSCs significantly affected B-cell differentiation and chemotactic behavior (21). In our current model, significant attenuation of IgM levels suggested but did not prove that MSCs may directly affect B cells in vivo, since IgM production is T-cell independent and thus antidonor IgM modulation was not simply a reflection of MSCs influencing T–B cell interactions (52,69). However, attenuation of B cells and antibody levels by MSCs could also stem from various indirect MSC effects, such as altering the amount of free alloantigen due to their overall suppression of graft damage (65,70), or MSC modulation of DC antigen processing/presentation efficacy (18,19,24), etc.

The currently demonstrated success of MSCs+Rapa inducing recipient allo-antigen-specific Tol-DCs suggests the treatment's relevance particularly in terms of influencing indirect antigen presentation. With indirect pathway involvement in graft rejection being recognized in the clinic as substantial—particularly at later times after transplantation (65,71)—the current achievement of tolerance with MSCs+Rapa is significant, despite the need to still discern the treatment's effects on the direct presentation pathway. We propose that under the influence of MSCs+Rapa therapy, recipient DCs processing apoptotic donor-(C57BL/6)-derived cells, expressing sub-optimal co-stimulatory and/or adhesion molecules become tolerogenic, and present donor-specific peptides to recipient T cells in a manner that leads to C57BL/6-specific tolerance, but not universal tolerance (i.e. C3H-specific) (65,71–73). Based on their currently and previously demonstrated independence from MHC restriction (26,27,29,30,33,34), we predict that MSCs—irregardless of their source—may also influence direct allorecognition by modulating passenger donor DCs within the allograft to become tolerogenic, reducing their co-stimulatory capacity similar to that witnessed in this study with recipient DCs.

Significant MSC engraftment within the cardiac graft of tolerant recipients, but the absence from the native heart, suggests that MSC migration and/or proliferation is influenced by tissue injury and/or inflammatory signals. We propose that migration of infused MSCs to allogeneic heart grafts may serve two purposes. Firstly, MSCs may shield heart graft alloantigens from the recipient's immune system through their capacity to differentiate into cardiac muscle and vascular endothelium (4,74–77). Secondly, localized MSCs may protect the heart through immunoregulation involving direct MSC–lymphocyte contact (26,27,29,30) and/or MSC-released soluble factors (24,31,32).

In this study, MSCs revealed several unique properties. First, MSCs need time to mediate their immunosuppressive function in vivo. The initial phase of Rapa immunosuppression appears to be necessary to provide early blunting of the allogeneic immune response to enable successful MSC engraftment within the recipient's lymphoid organs and graft. Second, MSCs play their immunoregulatory function at least in part, within the recipient's lymphoid organs. After transfer, MSCs migrate to central and peripheral lymphoid organs upon heart transplantation, where they can reside over 100 days and potentially influence immune cell responses through cell-to-cell interactions or through cytokine secretions, toward a tolerogenic phenotype. Third, MSCs exert their function in a MHC genotype-independent manner and thus could be used as a ‘universal’ cell source. In conclusion, our present study provides substantial proof that MSCs can permanently engraft immune organs of allogeneic hosts and induce immune-specific graft tolerance by modulating function of DCs, T and B cells. Therefore, our work not only provides a novel approach to inducing organ transplantation tolerance but also provides new insights into mechanisms of MSC action in vivo.

Acknowledgments

The authors are grateful to Dr. Gill Strejan for critical review of the manuscript. The authors also acknowledge the longstanding support of Dr. Robert Zhong (deceased), remembered for his unwavering focus on translational transplant research. This work was supported by Heart and Stroke Foundation of Ontario (NA5938 and T6318), Roche Organ Transplantation Research Foundation, The Kidney Foundation of Canada, Canada Foundation for Innovation, and the Multi-Organ Transplant Program Research Grant, London Health Sciences Centre.

Ancillary