Toll-Like Receptors 3 and 4 Are Expressed by Human Bone Marrow-Derived Mesenchymal Stem Cells and Can Inhibit Their T-Cell Modulatory Activity by Impairing Notch Signaling



Bone marrow (BM)-derived mesenchymal stem cells (MSCs) are multipotent, nonhemopoietic progenitors that also possess regulatory activity on immune effector cells through different mechanisms. We demonstrate that human BM-derived MSCs expressed high levels of Toll-like receptors (TLRs) 3 and 4, which are both functional, as shown by the ability of their ligands to induce nuclear factor κB (NF-κB) activity, as well as the production of interleukin (IL)-6, IL-8, and CXCL10. Of note, ligation of TLR3 and TLR4 on MSCs also inhibited the ability of these cells to suppress the proliferation of T cells, without influencing their immunophenotype or differentiation potential. The TLR triggering effects appeared to be related to the impairment of MSC signaling to Notch receptors in T cells. Indeed, MSCs expressed the Notch ligand Jagged-1, and TLR3 or TLR4 ligation resulted in its strong downregulation. Moreover, anti-Jagged-1 neutralizing antibody and N[N-(3,5-difluorophenacetyl-l-alanyl)]-S-phenylglycine t-butyl ester (DAPT), an inhibitor of Notch signaling, hampered the suppressive activity of MSCs on T-cell proliferation. These data suggest that TLR3 and TLR4 expression on MSCs may provide an effective mechanism to block the immunosuppressive activity of MSCs and therefore to restore an efficient T-cell response in the course of dangerous infections, such as those sustained by double-stranded RNA viruses or Gram-negative bacteria, respectively.

Disclosure of potential conflicts of interest is found at the end of this article.


Bone marrow (BM) mesenchymal stem cells (MSCs) are multipotent nonhematopoietic progenitor cells that can differentiate into BM stromal cells, osteoblasts, adipocytes, chondrocytes, tenocytes, skeletal myocytes, neurons, and cells of visceral mesoderm [1, [2], [3], [4], [5]–6]. They typically do not express hematopoietic stem cell (HSC) markers, but they do express a quite specific pattern of molecules, such as SH2 (CD105), SH3, SH4 (CD73), CD106 (vascular cell adhesion molecule 1), CD54 (intercellular adhesion molecule 1), CD44, CD90, CD29, and STRO-1 [2, [3], [4], [5]–6].

In the last few years, it has become clear that MSCs also possess immunoregulatory properties, which have been extensively studied and characterized for their relevance in immune responses, as well as their potential usefulness in BM transplants. Mouse BM-derived MSCs can dramatically downregulate the response of naïve and memory antigen-specific T cells to their cognate peptide, and this effect is primarily cell contact-dependent [78]. By contrast, human MSCs may influence different effector cells, including CD4+ and CD8+ T cells [9, [10], [11], [12]–13], natural killer (NK) cells [9, 14, 15], B cells [9, 16], monocytes, and dendritic cells (DCs) [17, [18]–19], and their effect seems to be mainly dependent upon the release of soluble factors, such as transforming growth factor (TGF)-β1, hepatocyte growth factor [10, 14], prostaglandin E2 (PGE2) [14, 19], interleukin (IL)-10 [20], indoleamine 2,3-dioxygenase (IDO) [9, 12], and interferon-γ (IFN-γ) [21]. The primary mechanisms involved in the MSC-mediated suppressive activity on immune effector cells and the role of MSC-derived stromal cells in normal lymphoid development are still partially unknown. However, it has been shown that MSC infusion significantly prolongs the survival of MHC-mismatched skin grafts in baboons [22], reduces the incidence of graft-versus-host disease (GvHD) after allogeneic HSC transplantation in humans [23], and treats severe acute GvHD refractory to conventional immunosuppressive therapy [21].

Toll-like receptors (TLRs), which are broadly distributed on cells throughout the immune system [24, 25], are the best studied immune sensors of invading microbes, and their activation is essential for inducing the immune response and enhancing adaptive immunity against pathogens [26, 27]. Members of the TLR family are also involved in the pathogenesis of autoimmune, chronic inflammatory, and infectious diseases [28]. Thirteen TLR1 analogues have been identified (10 in humans and 13 in mice) that recognize a wide variety of pathogen-associated molecular patterns in bacteria, viruses, and fungi, as well as certain host-derived molecules [29]. For example, TLR2 recognizes bacterial lipoproteins, peptidoglycans, and lipoteichoic acids from gram-positive bacteria. TLR3 recognizes virus-derived double-stranded RNA, as well as its DNA analogue poly(I:C). TLR4 recognizes lipopolysaccharides (LPS) from gram-negative bacteria. TLR5 recognizes bacterial flagellin, and TLR9 recognizes the CpG motif of bacterial DNA. TLR7 and TLR8, which are phylogenetically related and form an evolutionary cluster with TLR9 [30], have recently been found to mediate recognition of viral single-stranded RNA [31, 32] and serve as receptors for small synthetic guanosine-based antiviral molecules, such as loxoribine [33]. TLRs can also form complexes with their ligands that result in a greater level of specificity. For example, heterodimers of TLR1/2 recognize triacetylated bacterial lipopeptides, whereas TLR2/6 recognizes diacetylated Mycoplasma lipopeptides [34]. A new member of the TLR family, TLR11, has recently been identified in mice, which responds specifically to uropathogenic bacteria [35]. The natural ligands for TLR10 have not yet been identified. Recent studies revealed the existence of endogenous TLR ligands, such as heat shock protein and high mobility group box 1, which are released by necrotic cells [36].

TLRs are type I transmembrane glycoproteins containing an extracellular domain composed of numerous leucine-rich repeats and an intracellular region containing a terminal inverted repeats (TIR) homology domain [37, 38]. The TIR domains interact with several TIR domain-containing adapter molecules (MyD88, TIRAP, TRIF, and TRAM) that activate a cascade of events resulting in the induction of transcription factors [39]. The common signaling feature among all TLRs is the activation of the transcription factor nuclear factor κB (NF-κB), which has been implicated in controlling the expression of inflammatory cytokines and cell maturation molecules. A subset of TLRs induces the production of type I IFNs [40, 41], which mediate antiviral, growth inhibitory, and immunomodulatory responses. Recently, a novel nonimmune role for TLRs has been reported, which is related to the maintenance of epithelial homeostasis through proliferation and tissue repair after direct injury to the epithelium [42] and stimulation of cell-cycle entry and progression in fibroblasts [43]. Moreover, it has recently been shown that the specific, ligand-mediated triggering of some TLRs, which are expressed by both murine and human MSCs, may control their proliferation and differentiation [44, 45].

The discovery of TLR expression by MSCs prompted us to investigate the potential link between TLR signaling and the MSC-mediated immunoregulatory functions. Thus, we demonstrated that human BM-derived MSCs express high levels of TLR3 and TLR4 and low levels of TLR1, TLR2, TLR5, and TLR6, whereas they do not express TLR7, TLR8, TLR9, and TLR10. Accordingly, LPS and poly(I:C), but not CpG ODN and R848, were able to induce NF-κB activation in MSCs, as well as cytokine and chemokine production by these cells. Flow-cytometric analysis of MSCs revealed no differences in the expression of CD34, CD80, CD86, CD105, and CD106 between untreated and LPS- or poly(I:C)-treated cells, as well no influence on their differentiation potential, inasmuch as they retained their ability to differentiate toward osteoblasts, chondrocytes, and adipocytes. However, we showed that the addition in culture of LPS or poly(I:C), but not of CpG ODN or R848, could significantly reduce the suppressive activity of MSCs on T-cell proliferation. This effect was mediated by the downregulation in MSCs of Jagged-1, a Notch receptor ligand that has been implicated in the inhibition of T-cell proliferation.

Materials and Methods

Reagents and Antibodies

The medium used for T-cell culture was RPMI 1640 (Seromed, Berlin,, supplemented with 2 mM l-glutamine, 1% nonessential amino acids, 1% pyruvate, 2 × 10–5 M 2-mercaptoethanol (Gibco, Grand Island, NY, Anti-CD3, CD4, CD8, CD16, CD19, CD105, CD106, CD90, CD31, CD45, CD14, CD34, CD80, CD86, CD119, human leukocyte antigen (HLA) (class I and II), and mouse IgG1 and IgG2a isotype control monoclonal antibodies (mAbs) were purchased from BD Biosciences (San Diego,, anti-TLR3 and -TLR-4 mAbs were purchased from Alexis Biochemical (San Diego, The neutralizing anti-Jagged-1 goat polyclonal IgG was obtained from R&D Systems Inc. (Minneapolis, TO-PRO-3 nuclear dye was purchased from Molecular Probes (Eugene, OR, Anti-CD29 mAb was purchased from Ancell (Bayport, MN,

Generation of MSCs

MSCs were obtained from BM aspirates of healthy donors, recruited after informed consent was obtained. BM mononuclear cells were separated by density gradient centrifugation (Lymphoprep; Nycomed, Oslo, Norway, and cultured in 25-cm2 flasks (BD Biosciences) at a concentration of 30 × 106 nucleated cells in 5 ml of Dulbecco's modified Eagle's medium, GlutaMAX I, 10% heat-inactivated adult bovine serum, 100 U/ml penicillin, and 100 μg/ml streptomycin (Gibco). Cultures were incubated at 37°C in a 5% CO2 atmosphere. After 72 hours, nonadherent cells were removed. When 70%–80% adherent cells were confluent, they were trypsinized (0.05% trypsin at 37°C for 5 minutes; Gibco), harvested, and expanded in larger flasks. Cells that were expanded for at least 6 weeks, >99% of them displaying a homogeneous immunophenotypic pattern, were used for the experiments.

Flow Cytometric Analysis

MSC immunophenotypic analysis was performed as detailed elsewhere [7]. Briefly, MSCs were detached using EDTA buffer, washed, and resuspended at 106 cells per milliliter. One hundred microliters of cell suspension was incubated at +4°C for 10 minutes with 15% fetal calf serum (FCS), followed by incubation with the specific mAb at +4°C for 30 minutes. Cells were then washed with phosphate-buffered saline (PBS) plus 0.5% bovine serum albumin (BSA) and analyzed by flow cytometry (BDLSR II; BD Biosciences). CD4+ T cells, obtained from peripheral blood mononuclear cells (PBMCs) by high-gradient magnetic-activated cell sorting (MACS; Miltenyi Biotec, Bergisch Gladbach, Germany,, were incubated with anti-CD3, CD4, CD8, CD14, CD16, and CD19 mAbs at +4°C for 15 minutes, washed with PBS plus 0.5% BSA, and analyzed by flow cytometry.

MSC Differentiation Assay

The differentiation potential of MSCs was assessed by testing their ability to differentiate into adipocytes, osteoblasts, and chondrocytes, as previously described [9]. Adipocytic differentiation was achieved after a 2-week culture of MSCs with adipogenic medium containing 10−6 M dexamethasone, 10 μg/ml insulin, and 100 μg/ml 3-isobutyl-1-methylxantine (Sigma-Aldrich, Milan, Italy, Osteoblastic differentiation was achieved after a 2-week culture with osteoblastic medium containing 10−7 M dexamethasone, 50 μg/ml ascorbic acid, and 10 mM β-glycerophosphate (Sigma-Aldrich). Chondrocytic differentiation was achieved after a 2-week culture with chondrocytic medium (CM) containing 10−7 M dexamethasone and 10 ng/ml TGF-β (Sigma-Aldrich), which were added to a pellet of 2.5 × 105 MSCs centrifuged at 1,500 rpm for 10 minutes. Oil red O, von Kossa, and toluidine blue dyes were used to identify adipocytes, osteoblasts, and chondrocytes, respectively.

Isolation of CD4+ T Cells from PB

Negative selection from PB of CD4+ T cells was performed by MACS (Miltenyi Biotec), as described elsewhere [9]. The purity of the negatively selected populations was consistently ≥97%.

Proliferation Assays

Purified CD4+ human T cells (105 cells) were stimulated with 105 irradiated (9,000 rad) allogeneic T-cell-depleted PBMCs and 1 μg/ml anti-CD3 mAb (BD Biosciences), with or without different numbers of MSCs (2 × 104, 104, and 103 cells per well), in the absence or presence of poly(I:C) (Sigma-Aldrich), LPS (Sigma-Aldrich), or R848 (−Resiquimod, S28463; 3M Pharmaceutical, St. Louis, On day 4, cultures were pulsed for 8 hours with 0.5 μCi (0.0185 MBq) of 3H-thymidine radionuclide (Amersham Biosciences, Little Chalfont, U.K., and harvested, and radionuclide uptake was measured by scintillation counting.

Enzyme-Linked Immunosorbent Assay for Cytokines and Chemokines

The amounts of IL-10, IL-8 (BD Biosciences), IL-6, IL-4 (R&D Systems), and IFN-γ (Endogen, Woburn, MA, were evaluated by homemade sandwich enzyme-linked immunosorbent assays (ELISAs) using commercial pairs of mAbs. The amounts of CXCL10, PGE2, IL-1β, TGF-β1 (R&D Systems), and CXCL4 (PF4 Asserachrom; Roche Diagnostics, Mannheim, Germany, were measured by commercial ELISAs.

IDO Activity Measurement

IDO activity was measured by determining kynurenine content by a homemade system, as described elsewhere [9]. In brief, 100 μl of the cell supernatant was added to 25 μl of trichloroacetic acid 30% (vol/vol), vortexed, and incubated for 30 minutes at 50°C to hydrolyze N-formylkynurenine to kynurenine. After centrifugation for 10 minutes at 10,000g, 100 μl of supernatant was transferred into a 96-well flat-bottomed plate and mixed with 100 μl of 2% freshly prepared Earlich's reagent (Sigma-Aldrich) (p-dimethylbenzaldheide in glacial acetic acid).

After 10 minutes of incubation, absorbance was read at a 492-nm wavelength with a microplate reader. Kynurenine concentration was calculated using a serially diluted standard curve of l-kynurenine (Sigma-Aldrich) and then by subtracting control levels (complete culture medium plus 10% FCS, which received the same treatment as the samples).

Quantitative Analysis of NF-κB Translocation

Nuclear translocation of transcription factors was quantified in nuclear extracts from highly purified MSCs by using TransAM ELISA-based kits specific for human-activated NF-κB (Active Motif, Carlsbad, CA, Briefly, 3 × 105 MSCs from three healthy donors were cultured in six-well plates with 1 ml of complete medium supplemented with 10% heat-inactivated FCS (HyClone, Logan, UT, in the absence or presence of LPS, poly(I:C), or R848 for 30 minutes. At the end of the culture, cells were extensively washed with PBS solution; after removal of the cytoplasmic fraction, nuclear extracts were obtained by using the Nuclear Extract kit (Active Motif) according to the manufacturer's protocols. One microgram of nuclear extract was plated on 96-well plates coated with the immobilized oligonucleotide containing the activated NF-κB consensus site (59-GGGACTTTCC-39), followed by the incubation with an antibody recognizing an epitope on a p50 subunit. A horseradish peroxidase-conjugated secondary antibody provided a colorimetric readout that was quantified by means of spectrophotometry. Raji nuclear extract was used as positive control, as suggested by the manufacturer. To monitor the specificity of the assay, a wild-type consensus oligonucleotide was used as competitor for NF-κB binding.

RNA Extraction and Quantitative Real-Time Reverse Transcription-Polymerase Chain Reaction

Total RNA was extracted by using the RNeasy kit and treated with DNase I to eliminate any genomic DNA contamination (Qiagen, Hilden, Germany, cDNA from each sample was synthesized from 1 mg of total RNA with TaqMan Reverse Transcription Reagents (Applied Biosystems, Foster City, CA,, according to the manufacturer's protocol. Real-time polymerase chain reaction (PCR) was performed with an ABI Prism 7900HT Sequence Detection System (Applied Biosystems), according to the manufacturer's instructions. All PCR amplifications were performed by MicroAmp optical 96-well reaction plate with TaqMan Universal Master Mix and with Assay-on-Demand (Applied Biosystems). Each assay was carried out in duplicate and included a no-template sample as negative control. Reverse transcription (RT)-negative samples were used to demonstrate that the signals obtained were RT-dependent. Relative expression of mRNA levels was determined by comparing experimental levels with a standard curve generated with serial dilution of cDNA obtained from human PBMCs. β-Actin was used as a housekeeping gene for normalization.

Confocal Microscopy

Human MSCs were cultured on Lab-Tek II chamber slides (Nalge Nunc International, Naperville, IL, for 3 days; after this period, cells were fixed in formaldehyde (2% in PBS pH 7.2), permeabilized with 0.05% Triton (Sigma-Aldrich) for 10 minutes, and incubated at room temperature with polyclonal rabbit IgG (1 mg/ml) to block the not specific sites of antibody (Ab) binding. After 15 minutes, cells were incubated with a fluorescein isothiocyanate (FITC)-conjugated mAb specific for TLR3 or a mouse IgG1 isotype control (both 20 μg/ml) for 30 minutes; in the same buffer, the TO-PRO-3 dye was added (0.2 μM) for the nuclear counterstaining. Cells were then washed in PBS for 5 minutes, and the slides were mounted with Vectashield mounting medium (Vector Laboratories, Burlingame, CA,

For the identification of Jagged-1, fresh or poly(I:C)-pretreated, adherent MSCs were incubated with a Jagged-1-specific goat IgG Ab or with control goat IgG, in the presence of an anti-CD105 mouse IgG1 mAb (as control of membrane staining) for 30 minutes (all the Abs were used at 10 μg/ml). After incubation, cells were washed and fixed as described previously. Cells were then incubated in the presence of Alexa Fluor 546-conjugated rabbit anti-goat Abs and Alexa Fluor 488-conjugated rabbit anti-mouse IgG1 Abs (both 5 μg/ml); nuclei were stained with TO-PRO-3 as previously described.

A confocal study was carried out by using an LSM 510 META confocal microscope (Carl Zeiss, Jena, Germany, Images were acquired with a ×40 objective (corresponding to a magnification of ×400).

Single confocal images of cells were obtained at nuclear equatorial level by using the 1 Airy unit formula for the adjustment of the pinhole diameter, corresponding to an optical slice of 0.7 μm (for FITC emission images). Images were acquired and analyzed by using the LSM 5 software (Carl Zeiss).


Statistical comparison of the proliferation assay arms was carried out according to the Student t test. Differences were considered statistically significant with p < .05.


Human BM-Derived MSCs Express High Levels of Functionally Active TLR3 and TLR4

MSCs displayed the constitutive expression of CD105, CD29, CD90, CD106, and HLA class I but not of CD80, CD86, CD45, CD34, or CD31 (Fig. 1A). The same cells showed multilineage differentiation potential as assessed by culturing in adipogenic, osteogenic, or chondrogenic medium (data not shown; [9]).

Figure Figure 1..

Immunophenotype and TLR expression of bone marrow (BM)-derived mesenchymal stem cells (MSCs). (A): Human proliferating adherent cells derived from BM of healthy donors were evaluated by flow cytometry for the expression of different surface markers associated with MSC immunophenotype; one representative analysis is reported. (B): The expression of mRNA specific for TLR1 to TLR10 was evaluated by quantitative real-time reverse transcription-polymerase chain reaction and normalized for the housekeeping gene β-actin; mean values of relative quantity of mRNA ± SD of eight different experiments are represented. (C): Protein expression of TLR3 and TLR4 was evaluated on MSCs by flow cytometric analysis, by using fluorescein isothiocyanate (FITC)-conjugated or Alexa Fluor 488-conjugated monoclonal antibodies (mAbs) (or appropriate isotype controls), respectively. (D): The cytoplasmic localization of TLR3 was also assessed by using confocal microscopy with an appropriate FITC-conjugated mAb (green) versus an isotype control; nuclear counterstaining was performed with TO-PRO-3 dye (red). Magnification, ×400. (C) and (D) show representative cases of three separate experiments. Abbreviations: HLA, human leukocyte antigen; TLR, Toll-like receptor.

To investigate the expression of TLRs in human BM-derived MSCs, in vitro-expanded MSCs obtained from healthy subjects were assessed by real-time RT-PCR for TLR mRNA content by using specific primers for TLR1–TLR10. As shown in Figure 1B, TLR3 and TLR4 mRNAs were consistently and highly expressed by MSCs; by contrast, low levels of TLR1, TLR2, TLR5, and TLR6 and no TLR7, TLR,8, TLR9, or TLR10 mRNAs were detected in BM-derived MSCs (Fig. 1B). The protein expression of TLR3 and TLR4 on in vitro-cultured human MSCs was demonstrated by flow cytometry (Fig. 1C). Moreover, intracytoplasmatic localization of TLR3 was also assessed by confocal microscopy analysis (Fig. 1D).

It is well known that TLR activation induces in immune cells the translocation of NF-κB from the cytosol to the nucleus, which results in NF-κB-dependent gene expression [24, 25]. We therefore investigated NF-κB activity in MSCs following their exposure to TLR3 or TLR4 ligands. As shown in Figure 2A, poly(I:C) (a TLR3 ligand) and LPS (a TLR4 ligand), but not R848 (a ligand for TLR7 and TLR8), could induce NF-κB translocation to the nucleus within 30 minutes, demonstrating that TLRs expressed by MSCs were functional. TLR activation in DCs also leads to the secretion of a variety of cytokines and chemokines [24, 25]. Therefore, to provide additional evidence that the TLRs expressed by MSCs were active, we assessed by ELISA the presence of IL-1β, IL-6, IL-8, TGF-β1, CXCL10, and CXCL4 in culture supernatants of MSCs stimulated by TLR3, TLR4, or TLR7/8 ligands. Among the TLR ligands tested, only LPS and poly(I:C) were able to induce IL-6, IL-8, and CXCL10 secretion in culture supernatants, whereas TGF-β was not affected, and there was no secretion of IL-1β or CXCL4 (Fig. 2B). We then asked whether TLR3 and TLR4 ligation on MSCs could have some influence on their immunophenotype and differentiation potential. The addition in culture of LPS or poly(I:C) did not change the expression of CD34, CD80, CD86, CD105, and CD106 in BM-derived MSCs, as assessed by flow cytometry (Fig. 3A). Moreover, pretreatment of MSCs for 5 days with LPS or poly(I:C) did not impair their ability to differentiate into adipocytes, chondrocytes, or osteoblasts (Fig. 3B). These data clearly demonstrate that human BM-derived MSCs express both TLR3 and TLR4 and that the triggering of these receptors by specific ligands induces cell activation and production of some cytokines and chemokines, without affecting either MSC immunophenotype or differentiation potential.

Figure Figure 2..

Triggering of Toll-like receptor (TLR) 3 and 4 by their specific ligands result in nuclear factor κB (NF-κB) nuclear translocation and in cytokines production by MSCs. (A): Nuclear fractions of MSCs cultured for 30 minutes in the absence or presence of poly(I:C) (TLR3 ligand), LPS (TLR4 ligand), and the control ligand R848 (TLR7/8, not expressed by MSCs) were assessed by enzyme-linked immunosorbent assay (ELISA) technique for the expression of NF-κB (p50 subunit). NF-κB induction is expressed as the ratio between the levels measured in the presence of each TLR ligand and the levels of NF-κB measured in the absence of such ligands. Columns represent mean values ± SD of four different experiments. *, p < 0.005. (B): Supernatants from 24-hour cell cultures obtained in the absence (medium) or presence of poly(I:C), LPS, R848, or the combination of the proinflammatory cytokines IFN-γ and TNF-α (positive control) were harvested for the determination of cytokine production by ELISA method. Mean values ± SD of five different experiments are shown. *, p < .01; **, p < .005. Abbreviations: IFN-g, interferon-γ; IL, interleukin; LPS, lipopolysaccharide; TGF, transforming growth factor; TNF, tumor necrosis factor.

Figure Figure 3..

The Toll-like receptor 3 ligand poly(I:C) interferes with neither the phenotype nor the differentiation potential of MSCs. (A): Immunophenotype was determined by flow cytometry on MSCs after 5 days of culture in the absence (medium) or presence of R848, poly(I:C), or LPS. One representative case of three is reported. (B): In analogous experiments, MSC were cultured for 5 days in the absence (medium alone) or in the presence of poly(I:C) or the control ligand R848. After this period, cells were cultured for an additional 2 weeks in AM, OM, or CM; oil red O, von Kossa, and toluidine blue dyes were used to identify adipocytes, osteoblasts, and chondrocytes, respectively. One representative experiment of three is depicted. Abbreviations: AM, adipogenic medium; CM, chondrogenic medium; LPS, lipopolysaccharide; OM, osteogenic medium.

TLR3 and TLR4 Stimulation of BM-Derived MSCs Inhibits Their Capacity to Suppress CD4 T-Cell Proliferation

In a previous report, we have shown that human bone marrow-derived MSCs can suppress the proliferation of CD4+ T cells stimulated with T-cell-depleted allogeneic PBMCs [9]. We therefore examined the effect of LPS, poly(I:C), and R848 on the immunosuppressive activity of MSCs. To this end, we assessed the proliferation of purified CD4+ T cells in response to allogeneic stimulation, after culturing in the absence or presence of increasing numbers of MSCs without or with LPS, poly(I:C), or R848. As expected, the addition of MSCs significantly inhibited the proliferation of activated CD4+ T lymphocytes (Fig. 4A). However, the addition of LPS or poly(I:C), but not of R848, at a 1:10 MSC-CD4+ T-cell ratio, almost completely restored the CD4+ T-cell proliferation as assessed by thymidine uptake, cell counts, and CFDA-SE staining (Fig. 4A; data not shown). To exclude the possibility that restoration of T-cell proliferation was mediated by the effect of the TLRs ligands on CD4+ T cells or T-cell-depleted allogeneic PBMCs, MSCs alone were preincubated for 5 days with poly(I:C) or R848 and then collected, washed, and finally cocultured with allogeneic PBMC-stimulated CD4+ T cells. Even under these experimental conditions, we observed a significant reduction of the ability of MSCs to suppress the proliferation of CD4+ T cells exerted by poly(I:C) (Fig. 4B).

Figure Figure 4..

Poly(I:C) and LPS, but not R848, reduce the inhibitory activity of MSCs on T cells proliferation. (A): CD4+ T cells were stimulated by irradiated allogeneic T-cell-depleted peripheral blood mononuclear cells (black column) in the absence or presence of different numbers of MSCs (MSC/T-cell ratios of 1/5, 1/10, and 1/100; white, light gray, and dark gray columns, respectively). These experimental conditions were repeated in the absence (medium) or in the presence of poly(I:C), LPS, or R848. After 4 days of culture, 3H-thymidine was added to the culture medium for additional 8 hours, and then cells were harvested and assessed for radionucleotide uptake. Mean values ± SD of 10 different experiments are reported. *, p < .05; **, p < .01. (B): Human bone marrow-derived MSC were cultured for 5 days in the absence or presence of poly(I:C) or R848; cells then were harvested, washed, and cocultured with polyclonally stimulated CD4+ T cells at different MSC/T-cell ratios, in the absence of any Toll-like receptor ligands. Columns represent mean values ± SD of three different experiments. *, p < .05. Abbreviations: LPS, lipopolysaccharide; MSC, mesenchymal stem cell.

TLR3 and TLR4 Stimulation Does Not Influence IDO Activity or PGE2 Levels

In a previous study, by using competitive inhibitors of IDO activity, which catalyzes the conversion from tryptophan to kynurenine, we demonstrated that a partial but substantial inhibition of the suppressive activity exerted by MSCs on the proliferation of CD4+ T cells stimulated with T-cell-depleted allogeneic PBMCs was related to the IDO activity [9]. On the basis of this finding, we hypothesized a possible interference exerted by LPS and poly(I:C) on IDO activity, which could result in the inhibition of the MSC immunosuppressive function. To support this possibility, CD4+ T cells were stimulated with T-cell-depleted allogeneic PBMCs, in the absence or presence of increasing numbers of MSCs, with or without LPS or poly(I:C); after 4 days, the amounts of kynurenine in supernatants of these in vitro cultures were assessed. In addition, the cells were harvested and the extent of T-cell proliferation was evaluated. As shown in Figure 5A, the addition of poly(I:C) or LPS inhibited the suppressive activity of MSCs, but the amounts of kynurenine in the supernatants of the same cell cultures did not vary, thus indicating that the IDO pathway was not influenced by TLR3 or TLR4 binding (Fig. 5B).

Figure Figure 5..

Indoleamine 2,3-dioxygenase (IDO) activity in MSCs is not affected by poly(I:C) and LPS. (A): CD4+ T cells were stimulated under the same experimental conditions reported in the legend of Figure 4A except for the absence of R848. Mean values ± SD of four different experiments are reported. *, p < .05. (B): The culture supernatants obtained from such experiments were harvested after 4 days of culture, before the addition of 3H-thymidine, and used for the determination of kynurenine, a tryptophan metabolite, indicating the enzymatic activity of IDO (described in Materials and Methods). Mean values ± SD of four different experiments are shown. Abbreviations: LPS, lipopolysaccharide; mcMol, micromolar; MSC, mesenchymal stem cell; n.s., not significant.

The role of PGE2 in the suppressive activity exerted by human MSCs has been recently demonstrated [14, 19]. To exclude a possible influence of LPS and poly(I:C) on PGE2 production by MSCs, we measured PGE2 concentrations in the supernatants of the same cell cultures. Neither LPS nor poly(I:C) exhibited any effect on the ability of MSCs to produce PGE2 (data not shown).

Notch Signaling Is Involved in the Suppressive Activity of MSCs, and TLR Triggering Downregulates the Expression of the Notch Ligand Jagged-1

The Notch pathway is highly conserved in evolution and is generally involved in cell fate decisions during differentiation [46]. Notch signaling is dependent on the γ-secretase-mediated cleavage of the Notch receptor intracellular domain (NICD) and on its translocation into the nucleus, where it associates with the repressor CBF-1/RBP-Jκ, which is converted into an activator of transcription [47, 48]. Recently, it has also been reported that one of the Notch ligands, Jagged-1, induces inhibition of T-cell proliferation, activation and cytokine production [49, 50]. Interestingly, it has also been reported that Jagged-1-overexpressing DCs can modulate T helper cell differentiation and that this phenomenon can be suppressed by incubating the DCs with TLR ligands, which induce a dramatic downregulation of Jagged-1 [51]. On the basis of these findings, we asked whether Jagged-1 could be involved in the suppressive activity of MSCs on T-cell proliferation and whether its expression could be changed following ligation of TLRs on the same cells. To this end, we first evaluated by quantitative real-time RT-PCR the expression of Jagged-1 and Delta-4 (another Notch ligand) in human MSCs. As shown in Figure 6A, of the two Notch ligands examined, only Jagged-1 was significantly expressed. Therefore, purified CD4+ T cells stimulated with allogeneic cells were cultured with or without increasing numbers of MSCs, in presence or absence of different concentrations of a neutralizing anti-Jagged-1 polyclonal Ab. As shown in Figure 6B, the addition in culture of the anti-Jagged-1 Ab at concentrations of 20 and 10 μg/ml significantly reverted the ability of MSCs to suppress the proliferation of CD4+ T cells. To further support the involvement of the Jagged-1/Notch interaction in the suppressive activity of human MSCs, we performed additional experiments in which the γ-secretase inhibitor N[N-(3,5-difluorophenacetyl-l-alanyl)]-S-phenylglycine t-butyl ester (DAPT), which is also known to inhibit the release of NICD [52], was added to the MSC/T-cell cocultures. As shown in Figure 6C, the addition of DAPT significantly reduced the ability of human MSCs to suppress T-cell proliferation. On the basis of these data, we hypothesized that ligation of TLR3 or TLR4 on MSCs could induce the downregulation of the Notch ligand Jagged-1 on these cells and the consequent reduction of their suppressive capacity. To demonstrate this hypothesis, the effects of TLR ligation on the expression of Jagged-1 by MSCs were examined. As shown in Figure 7A, Jagged-1 mRNA expression was significantly reduced upon stimulation of MSCs with either poly(I:C) or LPS but not with R848. To prove that the Jagged-1 mRNA reduction was also accompanied by the reduced expression of the Jagged-1 protein, Jagged-1 expression was also evaluated by confocal microscopy on MSCs cultured in the absence or presence of the TLR3 ligand poly(I:C). As shown in Figure 7B, under basal conditions, Jagged-1 was expressed, and its expression was associated on the membrane with CD105; on the contrary, and in accordance with the mRNA data, Jagged-1 expression appeared to be drastically downregulated by the presence in culture of poly(I:C) (Fig. 7C). Taken together, these data provide strong evidence that the Jagged-1/Notch interaction may be involved in the suppressive activity of MSCs on T-cell proliferation and that this effect can be reverted by Jagged-1 downregulation induced by TLR3 or TLR4 ligation.

Figure Figure 6..

MSCs express high levels of Jagged-1 mRNAs, and the blocking of Notch transduction pathway in T cells reduces their susceptibility to the suppression exerted by MSCs. (A): The expression of Jagged-1 and Delta-4 was evaluated at mRNA level in MSCs by quantitative real-time reverse transcription-polymerase chain reaction and normalized for the cell number. Columns represent mean values of relative quantity of mRNA ± SD from four different experiments. (B): The suppressive activity of MSCs was evaluated on activated T cells in the absence (medium) or in the presence of different concentrations of a goat anti-Jagged-1-neutralizing polyclonal antibody. Mean values ± SD of six different experiments are reported. *, p < .05. (C): Proliferation assays of T cells stimulated with T-cell-depleted allogeneic peripheral blood mononuclear cells plus anti-CD3 monoclonal antibody were performed in the absence or presence of different numbers of MSCs, without (medium) or with the γ-secretase inhibitor DAPT (5μM). Columns represent cpm mean value ± SD of nine different experiments. **, p < .01; *, p < .05. Abbreviations: DAPT, N[N-(3,5-difluorophenacetyl-l-alanyl)]-S-phenylglycine t-butyl ester; MSC, mesenchymal stem cell.

Figure Figure 7..

Levels of Jagged-1 mRNAs are downregulated by Toll-like receptor (TLR) 3 and TLR4 triggering. (A): MSCs were collected after 24 hours of culture in the absence (medium) or presence of poly(I:C), LPS, or R848. The expression of Jagged-1 mRNA was evaluated by quantitative real-time reverse transcription-polymerase chain reaction and normalized for the cell number; mean values of relative quantity of mRNA ± SD of five different experiments are represented. *, p < .01. Jagged-1 protein expression was evaluated on MSCs by confocal microscopy after 4 days of culture in the absence (B) or presence (C) of poly(I:C). Cell membrane was stained with an anti-CD105 monoclonal antibody (red), and Jagged-1 was stained with a specific polyclonal goat antibody (green); the staining obtained with control goat IgG (control) is also reported. Nuclei were counterstained with TO-PRO-3 (blue). Magnification, ×400. One representative experiment of three is reported. Abbreviation: LPS, lipopolysaccharide.


MSCs exhibit multiple functions and are considered to be important for prospective cell-based therapy. Understanding the factors and mechanisms regulating MSC differentiation, self-renewal, and suppression on ongoing immune responses is crucial and could allow us to manipulate them for therapeutic use. We show here that some TLRs, and particularly TLR3 and TLR4, are expressed by BM-derived MSC and are functional, as shown by the ability of their ligands, poly(I:C) and LPS, respectively, to induce NF-κB activation in MSCs, as well as the production of cytokines, such as IL-6, or chemokines, such as CXCL10 and IL-8. These findings are consistent with the results recently reported by Hwa Cho et al., showing high expression of functional TLR2, TLR3, TLR4, and TLR6 in human MSCs isolated from BM and adipose tissue [44]. More importantly, we demonstrate that TLR3 or TLR4 ligation on MSCs can regulate their immunosuppressive activity on T lymphocyte proliferation. Of note, these effects occurred without altering the MSC immunophenotype or differentiation potential. Our data are apparently in contrast with those reported by Pevsner-Fischer et al. [45], which showed that the TLR2 ligand Pam3Cys does not affect the immunosuppressive activity exerted by murine MSC on T-cell proliferation. One explanation may be that murine and human models are different. Another, more likely possibility is that in our assay we used as targets whole populations of CD4+ T cells, whereas these authors evaluated the immunosuppressive effects of TLR ligation on MSCs by using a single T-cell line activated by its cognate antigen, which may be insensitive to the effects of TLR ligation on MSCs.

In this study, we also provide evidence that the effects of TLR ligation on MSCs did not influence at least some of the mechanisms that have been described to be responsible for the immunosuppressive activity of MSC on T cells so far, such as IDO activity and PGE2 production. There is a general agreement on the fact that the suppressive activity exerted by MSCs on cells of the immune system depends on both soluble factors [9, 10, 12, 14, 19, 20] and cell-to-cell contact-dependent mechanisms, whose molecular basis has not been identified yet [7, 8]. Here, we demonstrate that the suppressive effect of MSCs was mediated by a previously unknown mechanism based on Notch receptor signaling on T cells. Members of the Notch family of transmembrane receptors are critically involved in the control of differentiation, proliferation, and apoptosis in several cells types [46, 49, 50]. Previous reports have shown that the overexpression of the Notch ligand Jagged-1 on DCs can modulate the differentiation of T helper lymphocytes and that ligation of TLR on DCs induces a dramatic downregulation of Jagged-1 [51]. On the basis of this finding, we first hypothesized that Notch signaling could be involved in the suppressive activity of MSCs on T-cell proliferation. Notch signaling is mediated by the γ-secretase-mediated cleavage of the NICD, translocation into the nucleus, and activation of CBF-1/RBP-Jκ [46, [47]–48]. In agreement with our hypothesis, the immunosuppressive effect on T-cell proliferation was inhibited by both an anti-Jagged-1 neutralizing Ab and the γ-secretase inhibitor DAPT. In addition, we found that MSCs constitutively expressed Jagged-1 but not the other Notch ligand, Delta-4, and that the expression of Jagged-1 by MSCs was inhibited at both the mRNA and protein levels by ligation of either TLR3 or TLR4. Taken together, these data strongly suggest that ligation of TLR3 or TLR4 on MSCs inhibits their suppressive effect on T-cell proliferation by hampering their Jagged-1 expression and, therefore, impairing its signaling to Notch receptor expressed on T cells. Thus, we have not only identified a new mechanism responsible for the cell contact-dependent, immunosuppressive effects of MSCs on T cells, but we have also provided the first evidence for TLR involvement in such immunomodulatory activity.

The results of this study raise the question of the pathophysiological meaning of the presence of TLR on MSCs and of their involvement in the regulation of the effector T-cell response. It is known that the effective induction of T-cell responses against pathogens requires T cell receptor (TCR) stimulation, the involvement of costimulatory molecules, and the expression of cytokine receptors. In addition, a temporal downmodulation of immune regulatory cells and circuits responsible for immunological tolerance contributes to the support of an immune response suitable for pathogen elimination. In this process, stimulation by pathogen-associated molecular patterns (PAMPs) through TLRs on DCs has been identified as a crucial component in effective host defense [53]. Indeed, TLR triggering alerts the innate immune system of the presence of microbial agents and induces a maturation program in DCs that enables these cells to act as APCs for specific T lymphocytes in draining lymph nodes, leading to the adaptive immune response [54]. However, analysis of the expression of TLRs and the in vivo application of TLR ligands demonstrated that the role of TLRs in immune regulation is more complex than the simple induction of DC maturation. TLRs are indeed widely expressed by immune cells, including T and B lymphocytes [24], and even by nonimmune cells, such as epithelial and endothelial cells [55, 56]. Interestingly, naturally arising regulatory T (Treg) cells all express TLRs [57, [58]–59]. It is known that Treg cells are essential for the protection against autoimmune diseases, but they can also hinder immune response against cancer and infectious agents. Thus, the presence of TLRs on Treg cells raises the intriguing possibility that TLR triggering may also contribute to the modulation of immune inhibitory pathways, thereby allowing more powerful immune responses. Recently, it has been demonstrated that TLR2 triggering on highly purified Treg cells, in combination with IL-2 addition and TCR ligation, can result in the proliferation of otherwise anergic Treg cells both in vitro and in vivo [60, 61]. Moreover, TLR2 ligation temporarily abrogates the suppressive phenotype of Treg cells, thereby enhancing immune responses both in vitro and in an acute infection model in vivo [60, 61]. Accordingly, human Treg cells have been found to express high levels of TLR8, and its triggering on these cells prevents their suppressive phenotype [62]. Thus, TLR ligands seem to have the capacity to modulate the immune response even by acting directly on TLR present on Treg cells.

Our demonstration that human BM-derived MSCs also express TLRs and that their triggering by specific ligands not only can induce the production of proinflammatory cytokines and chemokines by these cells but also suppress their inhibitory activity on T-cell proliferation is consistent with the concept that during microbial infections a downregulation of all immunosuppressive cells and circuits (including those provided by MSCs) may be useful to let the effector response become more efficient against the invading agent.

The evidence that MSCs, completely comparable to those obtained from BM, can also be achieved from lymphoid tissues, such as spleen and thymus [63, 64], suggests that MSC-derived stromal cells could play an important role in T-cell priming in primary and secondary lymphoid tissues. Some preliminary results that we have obtained with spleen- and thymus-derived MSCs are consistent with a broad involvement of the TLR and Notch systems in the immunomodulatory effect of MSCs of different origin, thus revealing a common pathway of immune regulation in the microenvironmental stromal compartment (data not shown).

Finally, the demonstration that TLR ligation by PAMPs on MSCs impairs their immunosuppressive activity may also have practical implications in view of cell-based intervention strategies. Indeed, allogeneic MSCs, in addition to their use in regenerative medicine, are being clinically evaluated for treating GvHD, inasmuch as preliminary findings suggest that these cells might enhance transplant engraftment [21, 65, 66]. The demonstration that the immunosuppressive activity of MSCs is lost following triggering of the TLRs that they express suggests that infectious agents expressing PAMPs for TLR3, TLR4, but probably also for TLR2 and TLR6 [44], should be considered at particularly high risk during this type of treatment.

Disclosure of Potential Conflicts of Interest

The authors indicate no potential conflicts of interest.


This work was supported by the Italian Ministry of University and Scientific Research, Italian Association for Cancer Research, Italian National Research Council, Fondazione Cariverona, and the Ministry of Health of Tuscany Region. F.L. and R.A. contributed equally to this work.