Abstract
Bone marrow-derived mesenchymal stromal cells (MSC) possess an immune plasticity manifested by either an immunosuppressive or, when activated with IFN-γ, an APC phenotype. Herein, TLR expression by MSC and their immune regulatory role were investigated. We observed that human MSC and macrophages expressed TLR3 and TLR4 at comparable levels and TLR-mediated activation of MSC resulted in the production of inflammatory mediators such as IL-1β, IL-6, IL-8/CXCL8, and CCL5. IFN-α or IFN-γ priming up-regulated production of these inflammatory mediators and expression of IFNB, inducible NO synthase (iNOS), and TRAIL upon TLR activation in MSC and macrophages, but failed to induce IL-12 and TNF-α production in MSC. Nonetheless, TLR activation in MSC resulted in the formation of an inflammatory site attracting innate immune cells, as evaluated by human neutrophil chemotaxis assays and by the analysis of immune effectors retrieved from Matrigel-embedded MSC injected into mice after in vitro preactivation with cytokines and/or TLR ligands. Hence, TLR-activated MSC are capable of recruiting immune inflammatory cells. In addition, IFN priming combined with TLR activation may increase immune responses induced by Ag-presenting MSC through presentation of Ag in an inflammatory context, a mechanism that could be applied in a cell-based vaccine.
Bone marrow-derived mesenchymal stromal cells (MSC)4 are mesenchymal progenitors thought to give rise to cells that constitute the hematopoietic microenvironment. MSC can be facilely isolated and expanded from the adherent cell fraction of bone marrow aspirates and serve as precursors for the generation of a variety of mesodermal tissues, including bone, cartilage, and muscle. In addition to their mesenchymal plasticity, MSC are able to regulate the immune system in a manner that depends on their state of activation (1). Resting MSC are described as strongly immunosuppressive, especially for the repression of allogeneic immune responses and autoimmune diseases (2, 3). However, in response to IFN-γ stimulation, MSC acquire MHC class II expression (4) and are capable of Ag presentation (5, 6, 7) and to induce a T lymphocyte-mediated immune response in vivo (5). This suggests that MSC possess an immune plasticity with a default-suppressive phenotype and, when appropriately activated, an option to stimulate an immune response. This immune plasticity appears akin to that observed with dendritic cells (DC), which in the absence of maturation can induce T cell tolerance (8).
DC maturation can be achieved by activation of TLR. Twelve TLR have been identified in mammals and are expressed preferentially by immune cells. TLR are pattern recognition receptors that recognize lipids, carbohydrates, peptides, or nucleic acids specifically expressed by various pathogens. On the one hand, TLR activation critically initiates the inflammatory and subsequent adaptive immune responses. In macrophages and DC, activation with TLR ligands can result in an increase in MHC class I- and II-mediated Ag processing, expression of costimulatory molecules, and the production of inflammatory mediators. Among TLR-induced cytokines, biologically active IL-12p75, which is a polypeptide composed of the IL-12p35 and IL-12/IL-23p40 chains, is a key factor in the induction of Th1 and CTL cells, production of opsonic Abs, and activation of macrophages and NK cells that secrete high levels of IFN-γ (9). On the other hand, the expression and response to TLR is modified in the context of an immune response. For instance, stimulation with individual TLR induces the production of an excess of IL-12p40 in monocyte-derived DC and macrophages (10). Optimal production of IL-12p35 and active IL-12p75 is observed in response to TLR ligands after priming with IFN-γ (11) or IFN-α (12), costimulation with CD40L (13), or to a specific combination of TLR ligands (12, 14). These combined stimulations result in the activation of different signaling pathways and transcription factors that are particularly effective for the additive or synergistic up-regulation of IL-12 (15). In addition, inflammatory cytokines can regulate the expression of TLR, such as IFN-α known to increase expression of TLR3 in human primary DC, macrophages (16), or endothelial cells (17).
Observations of TLR expression and function on MSC revealed that mouse MSC (mMSC) expressed all TLR mRNA, with the exception of TLR9. A TLR2 ligand, inhibited mMSC differentiation while sparing their ability to suppress allogeneic T cell activation (18). Recently, the protective effect of i.v. injected syngeneic mMSC was reported against rapid septicemia induced by cecal ligation and puncture (19). In this model, mMSC located in the lungs where they recruited macrophages and reprogrammed them for increased IL-10 release, an effect mediated by the combined production of NO by LPS-activated mMSC and macrophages. As for human MSC (hMSC), one study showed that hMSC expressed only TLR3 and TLR4 and exposure to TLR ligands decreased their ability to suppress allogeneic T cell proliferation, an effect that correlated with the down-regulation of Notch ligand Jagged-1 expression on hMSC (20). The present work focused on assessing TLR expression and activation in MSC compared with “professional APC” such as primary macrophages for the induction of an inflammatory state, as well as regulation of these properties by IFN-α or IFN-γ priming.
Materials and Methods
Reagents
Recombinant human (rh) and mouse (rm) IFN-α, IFN-γ, and TNF-α were from Invitrogen, rhIFN-α2b was from Schering (Intron A), and rhGM-CSF was a generous gift from Cangen. Specific TLR activation was achieved using Pam3CSK4 (InvivoGen), polyinosinic-polycytidylic acid (poly(I:C); Sigma-Aldrich) for TLR3, LPS from Escherichia coli
Cell isolation and culture
hMSC from donors 240L, 5066R, and 5068L (24- year-old (yo) male, 22 yo female, and 24 yo male, respectively) were given to us by D. J. Prockop (Tulane University, New Orleans, LA). hMSC from donors 52F, 291, 292, 293, 294, and 302 (52 yo female, 73 yo male, 85 yo female, 73 yo female, 69 yo female, and 48 yo male, respectively) were established from bone marrow aspirates from patients undergoing hip surgery replacement at the Montreal Jewish General Hospital, as previously described (5). hMSC culture medium was α-MEM with 2 mM l-glutamine, 16.5% FBS, and 100 U/ml penicillin and streptomycin (PS). All tissue culture reagents were from Wisent Technologies. hMSC were expanded for up to four passages by plating cells every 7 days at 100 cells/cm2. Karyotype analyses of hMSC from donors 240L, 5066R, 5068L, 292, and 293 were established on 20 metaphasic cells and were shown to be normal in the absence of consistent numerical or structural chromosome anomalies (D. J. Prockop, personal communication, and data not shown). mMSC were obtained from C57BL/6 mice as described elsewhere (21). MSC were tested for the absence of CD31+ endothelial cells and CD45+ or CD34+ hematopoietic cells, expression of CD44, CD73, CD90, and CD105, and capacity to differentiate into adipocytes and osteocytes as described previously (21). Unless indicated otherwise, hMSC and mMSC were cultured at 100 cells/cm2 or 500 cells/cm2, respectively, for 4 days before activation. PBMC (four donors: MA-A, MA-B, MAL, and 692) were isolated from fresh apheresis with informed consent from healthy donors. Human monocyte-derived macrophage-enriched cell preparations were obtained by plating PBMC at 2 × 106 cells/ml in IDMEM supplemented with 2% human serum A/B, 100 U/ml PS, and 500 U/ml rhGM-CSF for 5–6 days before harvesting adherent cells. Flow cytometry analysis performed as described in Ref. 22 demonstrated that >95% adherent cells were CD14+ and/or CD64+. U937 (ATCC CRL-1593.2) and Mono-mac 6 cells are human monocyte and macrophage-like immortalized cells, respectively, and were cultured in RPMI 1640 with 2 mM l-glutamine, 10% FBS, and 100 U/ml PS. Primary HUVECs were obtained from Cambrex. Mouse peritoneal macrophages were exuded from C57BL/6 female retired breeder mice (Charles Rivers Laboratories) by sterile lavage and plated in RPMI 1640, 10% FBS, and 100 U/ml PS for 16 h before extensive washing to remove nonadherent cells.
Flow cytometry analysis
The following Abs specific for human molecules were used for flow cytometry analysis: biotin-coupled anti-CD90 (clone 5E10), CD45 (clone HI30), and TLR4 (clone HTA125); FITC-conjugated anti-CD64 (clone 10.1), CD105 (clone 8E11; Chemicon International) and CD64 (clone 10.1); PE-conjugated anti-β2-microglobulin (clone TÜ99), CD31 (clone WM-59), CD73 (clone AD2), CD80 (clone L307.4), TLR3 (clone TLR3.7; eBioscience), and TLR4 (clone HTA125; eBioscience); allophycocyanin-conjugated anti-CD4 (clone RPA-T4), CD34 (clone 581), and CD44 (clone G44-26); and isotypic controls. Mouse specific Abs were: FITC-conjugated anti-CD11c (clone HL3), PE-conjugated anti-CD117 (clone 2B8), CD14 (clone MΦP9), and NK1.1 (clone PK136), PerCP-Cy5.5-conjugated anti-Ly-6G/6C (clone RB6-8C5), and allophycocyanin-conjugated anti-CD3ε (clone 145-2C11). Cell permeabilization was performed using the Perm/Wash buffer. Flow cytometry analysis was performed on 20,000 events using a FACSCalibur cytometer and data were analyzed using CellQuest software. Except where indicated, Abs, reagents, and apparatus were from BD Biosciences.
Real-time RT-PCR
Primer sequences (5′–3′ forward and reverse, respectively) specific for human mRNA were: TLR1, GCACCCCTACAAAAGGAATCTG and GGCAAAATGGAAGATGCTAGTCA (107 bp); TLR2, CTGGTAGTTGTGGGTTGAAGCA and GATTGGAGGATTCTTCCTTGGA (102 bp); TLR3, TTAAAGAGTTTTCTCCAGGGTGTTTT and AATGCTTGTGTTTGCTAATTCCAA (124 bp); TLR4, CCCCTTCTCAACCAAGAACCT and ATTGTCTGGATTTCACACCTGGAT (120 bp); TLR5, TGCTAGGACAACGAGGATCATG and GAGGTTGCAGAAACGATAAAAGG (114 bp); TLR6, AGGCCCTGCCCATCTGTAA and GCAATTGGCAGCAAATCTAATTT (100 bp); TLR7, GCTATTGGGCCCATCTCAAG and TCCACATTGGAAACACCATTTTT (112 bp); TLR8, TCAGTGTTAGGGAACATCAGCAA and AACATGTTTTCCTTTTTAGTCTCCTTTC (134 bp); TLR9, GGGAGCTACTAGGCTGGTATAAAAATC and GCTACAGGGAAGGATGCTTCAC (103 bp); TLR10, TTTACTCTGGGACGACCTTTTCC and ATAAGCCTTACCACCAAAAGTCACA (100 bp); IL-8, CTGTTAAATCTGGCAACCCTAGTCT and CAAGGCACAGTGGAACAAGGA (64 bp); IL-12A, ATGCTCCAGAAGGCCAGACA and CCTCCACTGTGCTGGTTTTATCT (100 bp); IL-12B, TCATCAGGGACATCATCAAACC and CAGGGAGAAGTAGGAATGTGGAGTA (131 bp); iNOS/NOS2A, GGCTCCTTCAAAGAGGCAAA and CATCTCCCGTCAGTTGGTAGGT (100 bp); and 18S rRNA, TTACCAAAAGTGGCCCACTA and GAAAGATGGTGAACTATGCC (200 bp). Human primers specific for CCL5, TNF-α, TRAIL, IFN-β, and IFN-α2 mRNA were as previously described (22). Primer sequences (5′–3′ forward and reverse, respectively) specific for mouse mRNA were: Cxcl1/KC, CACCTCAAGAACATCCAGAGCTT and GTGGCTATGACTTCGGTTTGG (53 bp); Cxcl2/Mip-2, CCCTGGTTCAGAAAATCATCCA and GCTCCTCCTTTCCAGGTCAGT (48 bp); Ccl2, GTCTGTGCTGACCCCAAGAAG and TGGTTCCGATCCAGGTTTTTA (61 bp); Ccl3, CTCCCAGCCAGGTGTCATTTT and TTGGAGTCAGCGCAGATCTG (59 bp); Ccl4, AGCCCTGATGCTTCTCACTGA and GGGCCAGGAAATCTGAACGT (59 bp); Ccl5, TCCAATCTTGCAGTCGTGTTTG and TCTGGGTTGGCACACACTTG (58 bp); Ccl7, GCTGCTTTCAGCATCCAAGTG and GCAGCATGTGGATGCATTG (59 bp); Ccl8, CCAGACCAAGCAGGGTATGTC and CATGTACTCACTGACCCACTTCTGT (43 bp); and Ccl9, CGGAGAGTTCAGAGATGCATTG and TTGTTTGTAGGTCCGTGGTTGT (62 bp). Quantitative RT-PCR assays were performed in duplicate on an Applied Biosystems 7500 Fast Real-Time PCR system thermal cycler and SYBR Green Mastermix (Applied Biosystems) as described elsewhere (5, 6, 7). Data were analyzed by the comparative threshold cycle method, based on the relative expression of target gene mRNA vs 18S rRNA levels as a reference, using the SDS version 1.3.1 Relative Quantification Software. Expression levels of TLR mRNA were also determined using GAPDH or β-actin mRNA levels as alternate references and similar results were obtained (data not shown). Specificity of PCR amplification was tested by melting curves and agarose gel analysis. The absence of genomic DNA contamination was demonstrated routinely by analysis of PCR performed with total RNA and with each of the primer sets.
Immunoblot analysis
Immunoblot analyses were performed as described previously (5, 6, 7). Primary Abs were specific for: c-Rel (N), RelA/p65 (C-20), α-tubulin (TU-02), IFN regulatory factor (IRF) 1 (C-60) (all from Santa Cruz Biotechnology).
Growth response to TNF-α, IFN-α, or IFN-γ
hMSC were seeded in 96-well plates at 300 cells/cm2 in complete medium. Three days later, cells were treated or not with TNF-α (1–10 ng/ml), IFN-α (100–10 000 U/ml), or IFN-γ (10–300 U/ml) for 4 days before performing a MTT assay using the Cell Titer96 reagent (Promega) to measure cell viability. Conversion of tetrazolium compound to a 490-nm absorbing formazan product was measured on an ELX800 microplate reader (Bio-Tek Instruments).
Neutrophil chemotaxis assay
hMSC were untreated or primed with 30 U/ml IFN-γ for 18 h and activated with 100 ng/ml LPS for 2 h, extensively washed to remove IFN-γ and LPS, and supernatants (conditioned medium) were harvested 48 h later. Neutrophils were enriched from heparinized peripheral blood from healthy volunteers by Ficoll density gradient centrifugation (room temperature, 400 × g, 45 min) and erythrocytes were removed by hypotonic shock in 168 mM NH4Cl, 10 mM KHCO3, and 100 μM EDTA. Cells were washed at 120 × g for 10 min with the break off to remove platelets. Human neutrophils were then isolated by negative selection using a human neutrophil enrichment kit (StemCell Technology). The 24-well cell invasion assay (Millipore) was used to examine the migration of neutrophils according to the manufacturer’s protocol. Briefly, this modified Boyden chamber is composed of a well (lower compartment) containing an insert (upper compartment) that is separated by an 8-μm pore size polycarbonate membrane. Before cell seeding, Transwell permeable supports and cells were preincubated for 30 min and 1 h, respectively, in serum- free medium. Neutrophils (4 × 104
3, 5 × 103, 104, 2 × 104, and 4 × 104 neutrophils.Immune effector infiltration analysis
mMSC (3 × 106 cells) were untreated or primed with 30 U/ml IFN-γ for 18 h and activated with 1 μg/ml LPS for 2 h, extensively washed to remove IFN-γ and LPS, mixed to 300 μl of Matrigel (BD Biosciences) at 4°C, and injected s.c. in mice. After 2 days, implants were surgically removed and incubated at 37°C with a solution of 1.6 mg/ml collagenase type I (Sigma-Aldrich) and 200 μg/ml DNase I (Sigma-Aldrich) until complete dissolution of the Matrigel. The number of cells with a diameter larger than 7 μm (to exclude erythrocytes) was determined by automatic counting using a Z2 Coulter Particle Counter and Size analysis (Beckman Coulter) and cells were analyzed for the presence of Ly-6G/6C+, NK1.1+, CD11c+, CD117+, CD14+, or CD3+ cells by flow cytometry.
Statistical analysis
Otherwise indicated, experiments with hMSC were repeated at least twice using two different donors in each, giving a total of four independent data. Two groups were compared with the two-tailed unpaired Student t test using Prism software (GraphPad). Three groups and more were compared with the one-way ANOVA test and Dunnett’s multiple comparison or Bonferroni post tests.
Results
Constitutive TLR expression pattern in hMSC
To analyze the constitutive expression pattern of TLR in hMSC, adherent cell fractions from bone marrow aspirates performed in normal donors were selected for expression of CD44, CD73, CD90, and CD105, osteogenic and adipogenic differentiation capacity, and the absence of CD31+ endothelial cells and CD45+ or CD34+ hematopoietic cells, (Fig. 1⇓A). RNA were prepared from hMSC (four donors), unfractioned PBMC (two donors) that contain a mixture of immune cells expressing various TLR, and the monocytic immortalized cell lines U937 and Mono-mac 6. Real-time RT-PCR analyses of TLR1 to TLR10 mRNA levels demonstrated that hMSC expressed only TLR3 and TLR4 at levels comparable to hematopoietic cells (Fig. 1⇓B). TLR5-, TLR6-, and TLR9-encoding mRNA were detected in all hMSC donors. Since expression levels were low compared with fresh PBMC, the functional responses to corresponding ligands were not investigated. Cell culture density (tested from 100 to 2000 cells/cm2) had no effect on the basal TLR expression in hMSC (data not shown).
TLR3 and TLR4 expression in hMSC. A, hMSC characterization (donor 293). Top panel, Cells were analyzed by flow cytometry for surface expression of CD31, CD34, CD44, CD45, CD73, CD90, and CD105. The gray line represents the isotype control and the black line represents the specific Ab. Bottom panel, Cells were tested for their capacity to differentiate into bone and fat in the presence of osteocyte and adipocyte differentiation medium (original magnification, ×50 for nontreated cells and ×100 for differentiated cells). B, TLR1 to TLR10 mRNA levels. Total DNase-treated RNA from hMSC (donors 240L, 5068L, 5066R, and 52F), PBMC (donors MA-A and MA-B), and human monocytic immortalized cell lines U937 and Mono-mac 6 were prepared and processed for quantitative RT-PCR analysis of TLR1 to TLR10 expression. Relative quantification (RQ) was calculated by normalizing the expression of target genes to reference 18S rRNA levels. Normalized gene expression values were compared between samples displaying detectable target gene expression. The minimal normalized detectable target gene expression level was given the value of 1, and other values show fold increase expression compared with this minimal level. Shown are the means of triplicates ± SDs. ∗, Not detectable expression. Normalization to GAPDH and β-actin mRNA levels gave similar expression profiles (data not shown). C, TLR3 protein expression. hMSC (donors 240L and 293) and primary human macrophages (hMφ, donor 692) were analyzed for surface expression of FcγR1/CD64 and intracellular expression of TLR3 by flow cytometry as in A. D, Effect of cell permeabilization on TLR3 staining in hMSC (donor 240L) and U937 cells as analyzed by flow cytometry as in A.
Flow cytometry analysis of TLR3 and TLR4 expression was next performed. Two commercially available Ab clones directed against TLR4 were tested but did not allow the detection of significant expression of TLR4 in hMSC as well as FcγR1/CD64+ primary macrophages or U937 cells (data not shown). Expression of TLR4 was hence tested for responsiveness to TLR ligands (see below). Analysis of TLR3 expression demonstrated that hMSC homogenously expressed TLR3 at levels comparable to macrophages (Fig. 1⇑C). Previous studies demonstrated that TLR3 is found at the endosomal membrane and an internalization of exogenously added dsRNA is necessary for TLR3 signaling in DC (23). In a human lung fibroblast cell line, however, TLR3 was observed at the cell surface (23). In this study, in both hMSC and U937 cells, detection of TLR3 expression was enhanced by cell permeabilization and intracellular staining (Fig. 1⇑D), suggesting similar and intracellular localization of TLR3 in hMSC and macrophages.
Human MSC produce IL-1β, IL-6, IL-8, and CCL5, but not IL-12p75 in response to TLR3 or TLR4 ligands
The presence of functional TLR in hMSC was assessed by incubating cells with specific TLR3 or TLR4 ligands, respectively, poly(I:C) (a synthetic dsRNA), or LPS, and measuring production of early inflammatory mediators such as IL-1β, IL-6, IL-8/CXCL8, and β chemokine CCL5, as well as the Th1-inducing cytokine IL-12p75. As previously described (3), we observed that hMSC constitutively produced high levels of IL-6 (>10 ng/48 h/106 cells; Fig. 2⇓, A and B). In response to stimulation with LPS (1–1000 ng/ml) or poly(I:C) (0.02–20 μg/ml), hMSC from donors 292 (Fig. 2⇓A) and 240L (data not shown) produced elevated levels of IL-6 compared with macrophages and comparable amounts of CCL5. The fold induction in the increase of IL-6 production in response to poly(I:C) or LPS was nevertheless similar in both cell types. Although doses as low as 0.02 μg/ml poly(I:C) or 1 ng/ml LPS induced secretion of CCL5 in hMSC, optimal responses for IL-6 production were observed at 20 μg/ml poly(I:C) or 1 μg/ml LPS (Fig. 2⇓A). We hence used the latter doses for subsequent analyses. Response study in two other hMSC donors demonstrated that poly(I:C) or LPS activation induced production of inflammatory IL-1β and IL-6 cytokines and IL-8 chemokine (Fig. 2⇓B). In sharp contrast, levels of IL-12p75 were below detection in supernatants of resting or TLR ligand-activated hMSC, as tested by flow cytometry-microbead array or ELISA (data not shown). In addition, activation of hMSC with Pam3CSK4, a TLR1 and TLR2 ligand, did not up-regulate IL-6 or CCL5 production by opposition to macrophages (data not shown), confirming the specificity of the TLR expression and responsiveness in hMSC.
hMSC activated with TLR3 or TLR4 ligands produce inflammatory cytokines or chemokines. A, Dose response to poly(I:C) or LPS. hMSC (donor 294) or primary human macrophages (hMφ, donor 692) were stimulated with poly(I:C) (0.02–20 μg/ml) or LPS (1–1000 ng/ml) for 48 h before analysis of IL-6 and CCL5 levels in supernatants by ELISA. Shows are means of triplicates ± SDs. ∗, Not done. B, Poly(I:C) or LPS-induced IL-6, IL-1β, and IL-8 production. hMSC (donors 292 and 240L, plated at 500 cells/cm2 for 2 days) were treated with poly(I:C) (20 μg/ml) or LPS (1 μg/ml) for 24 h. Cytokine levels in supernatants were then measured by flow cytometry using a cytometric bead array. C, Poly(I:C)-induced IL-8 mRNA expression. hMSC (donor 294, plated at 500 cells/cm2 for 2 days), PBMC (donor MAL), and the 293T human embryonic cell line were treated with poly(I:C) (20 μg/ml) or LPS (1 μg/ml) for 20 h and processed for analysis of IL-8 expression as detailed in Fig. 1B legend. RQ, Relative quantification.
TLR3 is not the sole receptor specific for dsRNA, as most mammalian cells express RIG-I-like receptors (e.g., RIG-I, MDA5, and LGP2). Although TLR3 binds to endosomal dsRNA, RIG-I-like receptors are cytosolic RNA helicases able to sense dsRNA resulting from viral replication or electroporation of poly(I:C), leading to the production of inflammatory mediators in most cell types (24, 25, 26). Hence, we compared the response to poly(I:C) between 293T cells, which express RIG-I but not TLR3 or TLR4, and hMSC. Poly(I:C) or LPS added to 293T cell cultures did not increase expression of IL-8, in contrast to hMSC and PBMC (Fig. 2⇑C), suggesting that the response to poly(I:C) overlaid to cell cultures is mediated by TLR3 in hMSC.
These data support the hypothesis that the constitutive expression of TLR4 and TLR3 in hMSC resulted in a response to their specific ligand that was predominantly characterized by secretion of inflammatory cytokines and chemokines, with the exception of IL-12p75.
Exposure to TNF-α, IFN-α, or IFN-γ in hMSC regulates the expression of TLR in hMSC
To study the TLR response in the context of an immune or inflammatory response, we investigated the effects of exposure to TNF-α, IFN-α, or IFN-γ on hMSC growth and TLR expression. IFN-γ induced a marginal but statistically significant growth inhibition (<20% in OD values at all tested doses, p values <0.05) in two hMSC donors, as detected by the measure of mitochondrial activity in an MTT assay (supplemental Fig. 15). At all tested doses, TNF-α significantly inhibited growth in only one donor (35–40% decrease in OD values, p values <0.001) and IFN-α in both donors (25–49% decrease in OD values, p values <0.001). Automatic cell counting performed on hMSC from four donors cultured with 3 ng/ml TNF-α, 1000 U/ml IFN-α, or 100 U/ml IFN-γ for 48 h showed 5–20% decrease in cell numbers (data not shown). Thus, hMSC displayed moderate to medium growth retardation when cultured with TNF-α, IFN-α, or IFN-γ.
Exposure to TNF-α, IFN-α, or IFN-γ had no significant effects on expression levels of TLR1, TLR4, TLR5, TLR8, TLR9, and TLR10 mRNA in hMSC, as tested by real-time RT-PCR after 6 h (data not shown) or 24 h (Fig. 3⇓A and data not shown). In contrast, TNF-α up-regulated expression of TLR2 and TLR7, IFN-α strongly increased the expression of TLR3, and IFN-γ increased the expression of TLR2 and TLR3 (Fig. 3⇓A). Up-regulation of TLR3 expression by IFN-α and IFN-γ in hMSC was also observed by intracellular staining (Fig. 3⇓B). Importantly, in three tested hMSC donors (240L, 5068L, and 293) quantification of IFN-α or IFN-γ effects on TLR3 expression assayed by RT-PCR could not be paralleled to protein levels determined by flow cytometry. In the case of hMSC donor 240L, a 16- or 350-fold increase in TLR3 mRNA levels was observed upon IFN-γ or IFN-α activation, but this corresponded to only a 1.3- and 1.4-fold increase in protein levels, respectively. Overall, these results suggest that single treatments with TNF-α, IFN-α, or IFN-γ resulted in up-regulation of the expression of TLR2, TLR3, or TLR7 in hMSC.
Effect of TNF-α, IFN-α, and IFN-γ on TLR expression in hMSC. A, TLR mRNA levels. hMSC were activated with 3 ng/ml TNF-α, 1 000 U/ml IFN-α, or 100 U/ml IFN-γ for 6 h and then processed for quantification of TLR expression as detailed in Fig. 1B legend. RQ, Relative quantification. B, TLR3 protein expression. hMSC (donor 240L) were treated with 1 000 U/ml IFN-α or 100 U/ml IFN-γ or not (−) for 48 h and analyzed for TLR3 expression by flow cytometry, as detailed in Fig. 1C legend. B7H1 expression was tested as an additional control for IFN stimulation.
Combination of TLR3 or TLR4 activation with IFN-α or IFN-γ priming increases the inflammatory response in hMSC
To further address the regulation of the TLR response by cytokines in hMSC, we analyzed the effects of preexposure to IFN-α or IFN-γ on TLR3 or TLR4 ligand-induced expression of a set of genes encoding cytokines, chemokines, and proinflammatory mediators, namely, IL-12A (encoding IL-12p35), IL-12B (encoding IL-12p40), TNFA, CCL5/RANTES, IFNB, IFNA2, TRAIL, and NOS2A/iNOS. For this task, hMSC from donors 292, 240L, and 302 (Fig. 4⇓) and 291 and 5068L (data not shown) as well as primary macrophages were stimulated with IFN-α or IFN-γ for 18 h before the addition of poly(I:C) or LPS for 6 h for analysis of gene expression by real-time RT-PCR or 48 h for protein secretion by ELISA.
Combination of TLR3 or TLR4 activation with IFN-α or IFN-γ priming increases hMSC inflammatory response. hMSC (donors 292, 240L, and 302) were cultured at 500 cells/cm2 for 4 days. PBMC and primary macrophages (hMφ) were obtained from blood aphereses (donors MAL and 692). A, Expression of inflammatory genes. Cells were primed or not with IFN-α (2 000 U/ml) for 18 h and treated or not with poly(I:C) (20 μg/ml) or LPS (1 μg/ml) for 6 h. Total DNase-treated RNA was prepared and processed for quantitative RT-PCR analysis of IL-12A, IL-12B, TNFA, CCL5, IFNB, TRAIL, and iNOS expression, as described in Fig. 1B legend. RQ, Relative quantification. B, Cells were primed or not with IFN-γ (30 U/ml) for 18 h, treated or not with poly(I:C) (20 μg/ml) or LPS (1 μg/ml), and processed as in A. C, Inflammatory cytokine and chemokine production. Cells were primed or not with IFN-γ (30 U/ml) for 18 h, treated or not with poly(I:C) (20 μg/ml) or LPS (1 μg/ml) for 48 h, and processed for quantification of IL-12p75, TNF-α, IL-6, IL-8, and CCL5 levels by ELISA. D, Neutrophil chemotaxis. hMSC donors (292, 294, and 302) were activated or not with IFN-γ (30 U/ml) for 18 h and LPS (100 ng/ml) for 2 h and extensively washed to remove IFN-γ and LPS. Conditioned medium (CM) from the three hMSC donors (nontreated or IFN-γ + LPS-activated) were harvested 48 h later and pooled in equal volumes. Human neutrophils were isolated from peripheral blood and were placed in the upper compartment of a Transwell filter system. Conditioned medium was added in triplicate to the bottom wells in the absence or presence of anti-human IL-8 or isotype control (goat IgG). Neutrophils migration was quantified 4 h later. Results are shown as the total number of migrated cells (means ± SEM; ∗, p < 0.05 and ∗∗, p < 0.01).
Treatment with IFN-α alone had no effect on expression levels of IL-12A, IL-12B, TNFA, CCL5, IFNB, TRAIL, and iNOS (Fig. 4⇑A). Activation of hMSC with poly(I:C) or LPS increased expression of TNFA, CCL5, IFNB, and TRAIL (p values <0.05). Expression of IL-12A and IL-12B was not detected in poly(I:C) or LPS-activated MSC, in contrast to macrophages. Pretreatment with IFN-α and poly(I:C) acted in synergy for the up-regulation of IL-12A, TNFA, CCL5, IFNB, and iNOS but not of IL-12B mRNA levels in hMSC.
When poly(I:C) or LPS was added to hMSC preexposed to IFN-γ, a synergistic up-regulation of CCL5, TRAIL, and iNOS was observed in hMSC (p values <0.05) at levels comparable to or higher than macrophages (Fig. 4⇑B). TNFA was also synergistically up-regulated; however, its expression in hMSC was minimal compared with macrophages. Of note, sole activation with IFN-γ strongly up-regulated expression of IL-12A in hMSC. Activation with IFN-γ and poly(I:C) or LPS, however, did not further increase expression of IL-12A and did not induce expression of IL-12B, in clear contrast to macrophages.
ELISA analyses using cell supernatants harvested 48 h after stimulation confirmed that in macrophages high levels of TNF-α production was induced by LPS alone; while IL-12p75 secretion was synergistically increased by IFN-γ and LPS (Fig. 4⇑C). In hMSC, production of TNF-α and IL-12p75 was detected upon dual stimulation with IFN-γ and poly(I:C) or to a lesser extent with IFN-γ and LPS. Consistent with gene expression analyses, TNF-α and IL-12p75 levels were at least 50- and 3- fold lower, respectively, in supernatants from hMSC compared with macrophages. In contrast, a robust production of CCL5 was induced by LPS alone in hMSC (p values <0.001), which was increased at least two times by preexposure to IFN-γ (p values of IFN-γ- or LPS-treated vs IFN-γ plus LPS-treated hMSC <0.001; Fig. 4⇑C). Similarly, production of IL-6 and IL-8 were increased by the dual stimulation of IFN-γ and LPS in hMSC and macrophages (Fig. 4⇑C). Because IL-8 is a specific neutrophil-attracting chemokine, we tested the capacity of activated hMSC to mediate neutrophil recruitment. For this task, hMSC were untreated or primed with IFN-γ for 18 h and activated with LPS for 2 h, extensively washed to remove IFN-γ and LPS, and supernatants (conditioned medium) were harvested 48 h later. Neutrophil chemotaxis was tested in a Transwell system with purified human blood neutrophils in the upper compartment and supernatants from hMSC in the bottom compartment. Neutrophils migrated in higher numbers toward compartments with conditioned medium from IFN-γ-primed LPS-activated hMSC compared with nontreated hMSC (p value <0.01), an effect that was significantly inhibited by the neutralization of IL-8 in supernatants from IFN-γ-primed LPS-activated hMSC but not from nontreated cells (Fig. 4⇑D).
Overall, mRNA and protein expression analyses suggested that the association of TLR3 or TLR4 ligands with priming with IFN-α or IFN-γ increased the production of inflammatory cytokines and chemokines, such as IL-6, CCL5, and IL-8 in hMSC. The most striking difference between hMSC and macrophages, however, was the low levels of production of IL-12p75 and TNF-α by the former.
Basal and IFN-α- or IFN-γ-regulated expression of c-Rel, RelA, and IRF-1 in hMSC
Given the lack of production of IL-12p75 by TLR-activated hMSC, we tested whether hMSC and macrophages differently expressed latent transcription factors that, when activated by TLR signaling kinases, are involved in the expression of IL-12A and IL-12B. Notably, it was shown in monocyte-derived DC or macrophages that TLR-induced activation of IL-12A and IL-12B promoters is critically dependent on the binding of c-Rel-p50 NF-κB complexes (27, 28, 29). Activation of ubiquitous RelA-p50 NF-κB complexes, in contrast, was reported sufficient for expression of inflammatory cytokines such as IL-1 or IL-6 (29). In the present study, we observed that hMSC from three donors displayed basal expression of RelA but almost undetectable expression of c-Rel compared with primary macrophages (Fig. 5⇓A).
c-Rel and IRF-1 expression in hMSC. A, Basal c-Rel expression. Samples of WCE from hMSC (donors 291, 292, and 240L, plated at 2000 cells/cm2 for 3 days) and primary human macrophages (hMφ, donor MAL) were run on 4–20% SDS-PAGE gels and subsequently subjected to immunoblot analysis of RelA (p65), c-Rel, or α-tubulin (control) expression. B, IFN-α- or IFN-γ-induced IRF-1 expression. hMSC (donor 240L, cultured as in A) were treated or not with IFN-α (2000 U/ml) or IFN-γ (30 U/ml) for either 24 or 48 h. Samples of WCE were run on 4–20% SDS-PAGE gels and subsequently subjected to immunoblot analysis of IRF-1 or α-tubulin (control) expression.
IFN-γ-induced IRF-1 DNA binding was demonstrated to act in synergy with c-Rel/p50 for optimal IL-12A promoter activation in primary mouse macrophages (30). Likely, in hMSC, IRF-1 was strongly up-regulated by 24 h after exposure to IFN-γ and to a much lesser extent to IFN-α (Fig. 5⇑B). In addition, we observed that c-Rel levels were unaffected by exposure for 24–48 h to IFN-γ or IFN-α (data not shown). Overall, these results suggest that the absence of basal expression of c-Rel in hMSC compared with macrophages hindered the up-regulation of IL-12A and IL-12B mRNA by TLR ligands. In IFN-γ-stimulated hMSC, IRF-1 was likely to contribute to the activation of IL-12A promoter, as seen above (Fig. 4⇑C); however, the absence of induction of IL-12B may explain the low level production of active IL-12p75.
In vivo immune cell infiltration induced by TLR-activated MSC
Supernatants from hMSC activated with IFN-γ and LPS contained chemokines involved in the attraction of neutrophils (Fig. 4⇑C) as well as unfractioned PBMC (data not shown). To address the in vivo immune recruitment induced by MSC upon TLR activation, we first characterized the response of mMSC to TLR ligands. In distinction to hMSC, mMSC were previously shown to express all TLR, with the exception of TLR9 (18). In this study, we observed that C57BL/6 mMSC exposed to poly(I:C) or LPS increased IL-6 production, as detected by ELISA on supernatants of mMSC activated for 48 h (Fig. 6⇓A). IFN-α or IFN-γ priming combined with TLR activation resulted in a 19- to 180-fold additional increase in IL-6 production compared with single treatments (p < 0.001). Real-time PCR analyses performed on mMSC primed or not for 18 h with IFN-γ and/or activated with LPS for 4 h demonstrated that the dual treatment increased transcription of mRNA encoding the neutrophil-attracting mouse chemokines CXCL1 and CXCL2, as well as monocyte-macrophage or leukocyte chemokines such as CCL2, CCL5, CCL7, CCL8, and CCL9, with the exception of CCR1 and/or CCR5-binding chemokines CCL3 and CCL4 (Fig. 6⇓B). By contrast, secretion of TNF-α or IL-12 in activated mMSC remained low compared with macrophages (Fig. 6⇓A).
In vivo inflammatory response induced by TLR-activated MSC. A, IL-12, TNF-α, and IL-6 production by TLR3- or TLR4-activated mMSC. C67BL/6 mMSC or primary peritoneal macrophages (mMφ) were primed or not with IFN-α (2000 U/ml) or IFN-γ (30 U/ml) for 18 h and treated or not with poly(I:C) (20 μg/ml) or LPS (1 μg/ml) for 48 h. Supernatants were then processed for quantification of IL-12p75, TNF-α, and IL-6 levels by ELISA. ∗, Not done. B, Chemokine gene expression. mMSC were primed or not with IFN-γ (30 U/ml) for 18 h and treated or not with LPS (100 ng/ml) for 4 h. Total DNase-treated RNA was prepared and processed for quantitative RT-PCR analysis of Cxcl1, Cxcl2, Ccl2, Ccl3, Ccl4, Ccl5, Ccl7, Ccl8, and Ccl9 as described in Fig. 1B legend. ∗, Not detectable expression. Analysis on mouse spleen cells demonstrated expression of Ccl3 and Ccl4 genes (data not shown). RQ, Relative quantification. C, Immune infiltration induced by TLR-activated mMSC. mMSC were primed with IFN-γ (30 U/ml) for 18 h and activated with LPS (1 μg/ml) for 3 h. After extensive washing, cells (3.106) were mixed to Matrigel and injected s.c. in C57BL/6 mice. After 2 days, implants were excised and dissolved. Cells larger than 7 μm were automatically counted and analyzed for the presence of Ly6C/6C+ granulocytes, NK1.1+ NK cells, CD11c+ DC cells, and CD3+ T cells by flow cytometry. Results were reproduced in two other sets of experiments comparing nontreated to IFN-γ + LPS-treated mMSC (data not shown). Data represent values obtained with individual mice (∗, p < 0.05; ∗∗, p < 0.01; and ∗∗∗, p < 0.001, n = 5).
In vivo inflammatory chemoattracting properties of MSC were assessed by injecting C57BL/6 mice with extensively washed and Matrigel-embedded mMSC. Two days later, plugs were excised, dissolved by collagenase, and cells larger than 7 μm, to exclude erythrocytes, were automatically counted. Increases of 114 ± 30% and 146 ± 35% in cell numbers (not statistically significant) were observed in implants after injection of IFN-γ- or LPS-treated MSC compared with nontreated MSC. Combined treatment with IFN-γ priming and LPS resulted in an increase of 189 ± 52% in cell numbers (p < 0.01; data not shown). In all plugs, infiltrating immune cells consisted of Ly6C/6C+ granulocytes, i.e., neutrophils and eosinophils, NK1.1+ NK cells, CD11c+ DC cells, and CD3+ T cells as analyzed by flow cytometry (Fig. 6⇑C). No CD14+ monocytes/macrophages and CD117+ mast cells were detected (data not shown). Implantation of LPS-activated mMSC, especially cells primed with IFN-γ, resulted in a significant increase in granulocytes and NK cell infiltration (Fig. 6⇑C). Only one of the mice injected with IFN-γ plus LPS-treated mMSC displayed an increase in infiltrating CD3+ T cells (Fig. 6⇑C). Immune cell infiltration was reduced by 5 days (data not shown), suggesting it correlated with the transient production of inflammatory mediators by activated mMSC. These reports support the notion that mMSC closely parallel hMSC for their response to TLR3 or TLR4 activation, which was boosted by IFN-α or IFN-γ priming, resulting in an in vivo inflammation and recruitment of innate immune effectors.
Discussion
Since TLR are main players in the induction of the innate and subsequent adaptive immune responses, we studied the effects of TLR activation in MSC. We observed that hMSC and macrophages expressed TLR3 and TLR4 at comparable levels. Activation of hMSC with poly(I:C) or LPS resulted in the production of inflammatory mediators such as IL-1, IL-6, IL-8, and CCL5 that was increased by IFN-α or IFN-γ priming, resulting in the attraction of innate immune cells in vivo. The most notable difference between hMSC and macrophages was the lack of production of IL-12 or TNF-α by the former.
TLR engagement in macrophages or DC results in the increased production of inflammatory mediators, chemokines, and T- and B-activating cytokines such as IL-1α, IL-1β, IL-6, IL-8, IL-12, CCL5/RANTES, type I IFN, TNF-α, antitumoral and antimicrobial molecules such as NOS2A/iNOS-induced NO or TRAIL, as well as Ag processing and presentation (31, 32). In this study, we observed that hMSC expressed high basal levels of TLR3 and TLR4 and responded to their respective ligand by the production of inflammatory cytokines IL-1β and IL-6 as well as the chemokines IL-8 and CCL5. IFN-γ priming increased both TLR3- and TLR4-mediated inflammatory responses, as detected by the measure of TRAIL and iNOS mRNA transcripts as well as of the production of IL-6, IL-8, and CCL5 (Fig. 4⇑). IFN-α priming in hMSC and macrophages increased expression of TLR3 and responsiveness to poly(I:C). By contrast, TNF-α priming up-regulated TLR7 expression in hMSC; however, hMSC remained unresponsive to guardiquidimod, a TLR7 ligand (data not shown).
The most notable difference between MSC and macrophages was the low levels of IL-12B expression as well as the marginal production of IL-12p75 and TNF-α by the former upon TLR3 or TLR4 activation, even after IFN-α or IFN-γ priming. TLR3 and TLR4 use MyD88-Toll/IL-1 receptor domain-containing adaptor protein and Toll/IL-1 receptor domain-containing adaptor protein inducing IFN-β adaptor molecules, respectively, to activate NF-κB transcription factors (33). It was suggested that the constitutive expression of the c-Rel NF-κB subunit by DC and macrophages plays a determinant role for their capacity to produce of IL-12 and to induce T cell cross-priming (27). By contrast, the ubiquitous p65/RelA NF-κB subunit is able to activate IL-1, IL-6, and IL-8 promoters. In line with these observations, we report that hMSC displayed low basal levels of c-Rel compared with macrophages, while expression of RelA was detected in both cell types. Further investigations are required to identify mechanisms explaining the lack of production of TNF-α by hMSC. It was observed that the human TNFA promoter contains p50/RelA binding sites but its activation seems to rather depend on an enhancer complex containing NFAT, Ets/Elk, Sp1, ATF-2-Jun, and the coactivator proteins CBP and p300 (34).
Several mechanisms may account for the effect of IFN-α or IFN-γ on the tonic signaling through TLR. As reported in DC (9), we observed in hMSC that IFN-γ alone induced the transcriptional activation of IL-12A. IFN-γ-induced IRF-1 DNA binding was shown to act in synergy with c-Rel/p50 for optimal IL-12A promoter activation (30). This synergy was not clear when studies were performed with IL-12B instead of IL-12A (30). In the present study, we observed that IFN-γ, and to a lesser extent with IFN-α, up-regulated levels of IRF-1 in hMSC. Among other factors possibly implicated in the IFN-γ and TLR pathway cross-talk in MSC is IRF-8. IRF-8 expression is up-regulated by IFN-γ and IRF-8 was described to bind to the mouse il-12b promoter (35), as well as to increase TLR signaling and activation of ERK and JNK by interacting with TNFR-associated factor 6 (36).
What are the physiological outcomes and possible therapeutic applications of in vitro pretreatments with TLR ligands combined or not with IFN priming in MSC? On the one hand, we observed here that in vitro stimulation of mMSC with LPS before injection in Matrigel led to increased recruitment of NK and granulocytes. Infiltration was more noticeable when mMSC were primed with IFN-γ before LPS activation. Supernatants from IFN-γ-primed and LPS-activated hMSC induced human neutrophil Transwell migration. Correspondingly, ELISA or gene expression analyses reported that production of chemokines or expression of encoding mRNAs was increased by the dual stimulation in hMSC and mMSC. Among factors produced by TLR-activated MSC and deemed for a role in chemoattraction of these innate immune effectors are IL-8/CXCL8 in humans or CXCL1 and CXCL2 in mice, and CCL2, CCL5, CCL7, CCL8, and CCL9 that are chemokines produced during inflammation and acting as essential chemotactic and activators for neutrophils (37) and monocytes, DC, NK, or some T cell subsets (38), respectively. In addition, IFN-α and IFN-γ stimulation was previously reported to up-regulate Ag- presenting functions in MSC (5, 6, 7); however, this up-regulation was unaffected by the dual stimulation with TLR ligands (data not shown). Hence, these data suggest that in vitro pharmacological modulation of TLR3- or TLR4-mediated activity on MSC could be exploited as a means to further increase their IFN-γ-up-regulated APC functions through the formation of an appropriate inflammatory milieu after immunization with MSC-based vaccines. On the other hand, we observed that IFN-primed MSC were not able to produce IL-12 in response to TLR stimulation, which may hamper their capacity to directly boost Ag-specific Th1 responses upon Ag presentation compared with classical APC such as DC.
In the context of bacterial infection, two studies have reported the beneficial effect of i.v. MSC injections to reduce the septic shock caused by a generalized inflammatory state in humans who developed hemorrhagic cystitis and peritonitis after allogeneic hematopoietic stem cell transplantation (39) or in a mouse model of peritonitis (19). The latter study reported that after injection in septic mice, most mMSC localized in the lung where they were surrounded by macrophages. In vitro studies suggested that the concomitant activation of macrophages and MSC by LPS led to the additive release of NO and PGE2, which reprogrammed LPS-induced activation of macrophages for increased production of the anti-inflammatory IL-10 (19). Altogether, these observations and the present study suggest that LPS-mediated activation of MSC results in the recruitment of innate immune cells. The immunological outcome of this recruitment and interaction between MSC and innate immune cells is, however, difficult to predict since it may be regulated by the kinetics of activation of each cell type by bacterial toxins, as well as the concurrent cytokine milieu in which the cells are found. In conclusion, we propose that IFN-γ priming of MSC skews toward an Ag-presenting phenotype, an effect that may be amplified by concurrent agonist action of TLR activation leading to the recruitment of innate immune effectors, suggestive of a physiological role of activated MSC in immune defense in the presence of a proinflammatory cytokine milieu.
Acknowledgments
We thank Drs. Margaret Wolfe and Darwin J. Prockop (Tulane University) for the distribution of human MSC and useful information on their phenotype.
Disclosures
The authors have no financial conflict of interest.
Footnotes
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 a Terry Fox Foundation New Frontiers Program Project Grant and by the Canadian Institute for Health Research Operating Grant MOP-15017. M.F. is a PhD candidate at McGill University and is the recipient of a Canadian Institute for Health Research Studentship and J.G. is a Fonds de Recherché en Santé du Québec Chercheur Senior.
↵2 R.R.-M. and M.F. contributed equally to this study.
↵3 Address correspondence and reprint requests to Dr. Jacques Galipeau, Department of Medicine and Oncology, Sir Mortimer B. Davis Jewish General Hospital & Lady Davis Institute for Medical Research, McGill University, 3755 Cote Sainte-Catherine Road, Montreal, Quebec, Canada H3T 1E2. E-mail address: jacques.galipeau{at}mcgill.ca
↵4 Abbreviations used in this paper: MSC, mesenchymal stromal cell; DC, dendritic cell; poly(I:C), polyinosinic:polycytidylic acid; mMSC, mouse MSC; hMSC, human MSC; rh, recombinant human; rm, recombinant mouse; yo, year old; PS, penicillin and streptomycin; IRF, IFN regulatory factor; iNOS, inducible NO synthase; NOS, NO synthase.
↵5 The online version of this article contains supplemental material.
- Received November 18, 2008.
- Accepted April 16, 2009.
- Copyright © 2009 by The American Association of Immunologists, Inc.