Adult human bone marrow-derived mesenchymal stem cells (hMSCs) are under study as therapeutic delivery agents that assist in the repair of damaged tissues. To achieve the desired clinical outcomes for this strategy requires a better understanding of the mechanisms that drive the recruitment, migration, and engraftment of hMSCs to the targeted tissues. It is known that hMSCs are recruited to sites of stress or inflammation to fulfill their repair function. It is recognized that toll-like receptors (TLRs) mediate stress responses of other bone marrow-derived cells. This study explored the role of TLRs in mediating stress responses of hMSCs. Accordingly, the presence of TLRs in hMSCs was initially established by reverse transcription-polymerase chain reaction assays. Flow cytometry and fluorescence immunocytochemical analyses confirmed these findings. The stimulation of hMSCs with TLR agonists led to the activation of downstream signaling pathways, including nuclear factor κB, AKT, and MAPK. Consequently, activation of these pathways triggered the induction and secretion of cytokines, chemokines, and related TLR gene products as established from cDNA array, immunoassay, and cytokine antibody array analyses. Interestingly, the unique patterns of affected genes, cytokines, and chemokines measured identify these receptors as critical players in the clinically established immunomodulation observed for hMSCs. Lastly, hMSC migration was promoted by TLR ligand exposure as demonstrated by transwell migration assays. Conversely, disruption of TLRs by neutralizing TLR antibodies compromised hMSC migration. This study defines a novel TLR-driven stress and immune modulating response for hMSCs that is critical to consider in the design of stem cell-based therapies.
Disclosure of potential conflicts of interest is found at the end of this article.
Toll-like receptors (TLRs) are a conserved family of receptors that recognize pathogen-associated molecular patterns and promote the activation of immune cells [1, , , –5]. To date, several TLRs (numbered 1–11) have been identified in humans. Agonists for TLRs include exogenous microbial components such as lipopolysaccharide (LPS) (TLR2 and TLR4), lipoproteins, and peptidoglycans (TLR1, TLR2, and TLR6); viral RNA (TLR3); bacterial and viral unmethylated CpG-DNA (TLR9); and endogenous molecules, including heat shock proteins (HSPs; TLR4) and extracellular matrix molecules (fibronectin and TLR4) [2, 3, 5, 6]. TLR agonist stimulation leads to the expression of inflammatory cytokines or costimulatory molecules by an MyD88 (a TLR adapter protein)-dependent or MyD88-independent signaling pathway and can promote chemotaxis of the stimulated cell. TLRs are differentially expressed on leukocyte subsets and nonimmune cells and appear to regulate important aspects of innate and adaptive immune responses [2, 7, , –10].
The ability of TLRs to recognize seemingly unrelated molecules shed from both pathogens (e.g., LPS) and injured tissues (e.g., HSP70) served as the premise for the proposed “danger model” of immune response . This model is based on the idea that the immune system responds to signals that represent potential harm to the host rather than to signals that are foreign to the host. In doing so, this model addresses the shortcomings of other immune recognition models that rely on the notion that host immune cells recognize only non-self molecules [3, 11, 12]. These latter models are limited since they fail to explain certain observed immune responses: mothers not rejecting fetuses that contain foreign proteins or tumor cells being tolerated despite producing non-self proteins. The “danger model” that relies upon TLRs and their ability to respond to a multitude of endogenously and exogenously derived and aberrantly shed molecules has spawned a great deal of interest from researchers in diverse fields, including but not limited to tumor immunology, inflammation, and vaccine development.
Initially, research on TLRs focused on their expression and signaling consequences in immune cells. However, recent reports indicate that other bone marrow-derived cells, including mesenchymal stem/progenitor cells (MSCs), are among the cells that express TLR proteins [7, –9]. MSCs are separated from other cells in the bone marrow by their tendency to adhere to plastic. MSCs are typically known to differentiate into osteoblasts, chondrocytes, and adipocytes in culture [13, 14]. These cells specifically home to damaged and inflamed tissues and contribute to their repair in part by secretion of immunomodulating cytokines, chemokines, and extracellular matrix proteins. Critically, these cells are immunosuppressive to the host and can be easily expanded to large numbers in culture . As a result of these and other qualities, human bone marrow-derived mesenchymal stem cells (hMSCs) are very attractive candidates in stem cell-based strategies for tissue repair and gene therapy. Numerous investigators have now demonstrated the successful recruitment and multiorgan engraftment capability of hMSCs in various animal models and human clinical trials [15, 16]. However, the success of this strategy has been limited since the net engraftment measured for the infused hMSCs in preclinical and clinical trials is relatively poor . Therefore, a better understanding of the precise molecular mechanisms governing the hMSC's stem cell fate, mobilization, and recruitment to the sites of engraftment is warranted to improve treatment efficacy.
To this end, our study sought to determine whether, as with other bone marrow-derived cells, hMSC migration and/or recruitment was also driven by TLRs. Recent reports concerning adult adipose tissue-derived, mesenchymal, and hematopoietic stem cells suggest that TLRs play a role in stem cell biology [7, –9]. However, these reports mainly focused on the role of TLRs in stem cell proliferation and their potential role in disrupting the differentiation capabilities of the stem cells. These reports did not focus on the role of TLRs in critical stress responses of stem cells as analyzed here. Note that stress or danger signal responses of hMSCs are defined here as one of the potential mesenchymal stem cell fates that is different from self-renewal, differentiation, or apoptosis and that drives their migration, invasion, and engraftment into damaged and inflamed tissues sites. This hMSC fate is initiated once the host encounters various conditions of tissue pathology, including mechanical injury (wounds), inflammation, infection, or cancer, that then drive the egress of the hMSCs from their niche and allow their migration into the circulation, their invasion across vessels, and their engraftment to the injured site to fulfill their repair function.
We report here that stimulation of TLRs within hMSCs leads to the activation of established TLR signaling pathways. Interestingly, these activated pathways mediate the secretion of discrete patterns of cytokines and chemokines depending on the TLR ligand used. This observation implicates these receptors in the immune modulating function established for hMSCs . We also show that TLR stimulation particularly promotes their migration capabilities. Thus, TLR stimulation may be one mechanism that specifically drives the recruitment, migration, and immune modulating function of the hMSCs at injured or stressed sites. Apart from establishing a new aspect of their biology, the identification of TLRs as critical mediators of stress responses within hMSCs also provides a novel target to exploit in the improvement of stem cell-based therapeutic strategies.
Materials and Methods
Human MSCs were obtained from our collaborators at the Tulane University Center for Gene Therapy led by Darwin J. Prockop, M.D., Ph.D. In addition, MSCs were obtained from two commercial suppliers, Cambrex (Walkersville, MD, http://www.cambrex.com) and Allcells (Emeryville, CA, http://www.allcells.com), to ensure variability of the starting cell population and to make certain that findings are universal and not unique to single donor pools derived from a unique source. These suppliers test the hMSCs for their homogeneity and provide test results for their differential potential to chondro-, osteo-, and adipogenic lineages. Once obtained, expanded, and established in our laboratory, the hMSCs are also verified for positive staining of CD90, CD105, CD106, CD164, CD56, CD166, CD29, and CD44 and negative antibody staining for CD45, CD14, CD31, CD34, HLA DR, and CD117. Consistent fibroblast-like morphology was monitored by microscopy. MSC cultivation consisted of growth in tissue culture-treated flasks or dishes in minimum essential medium α with GlutaMAX I (minimal essential medium-α; Invitrogen, Carlsbad, CA, http://www.invitrogen.com) supplemented with 16.5% fetal bovine serum (Atlanta Biologicals, Lawrenceville, GA, http://www.atlantabio.com). For serum-free cultivation, growth medium without serum supplementation was added to the MSC cultures for at least 18 hours prior to the beginning of the experiment. MSCs of a passage number no greater than 6 were routinely used in all the experiments to maintain consistency.
In this study, endotoxin or LPS, synthetic CpG-oligodeoxynucleotides (CpG-ODN), flagellin, and polyriboinosinic polyribocytidylic acid (poly[I:C]) represented exogenous TLR ligands. Fibronectin-derived fragments (III1c, 45 kDa) and the human secreted antimicrobial peptide LL-37 alone or in combination with LPS served as the endogenous danger signals. Typically, TLR ligands were used in the following concentrations: 1 μM poly(I:C) (Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com), 10 ng/ml LPS (Sigma-Aldrich), 5 μM flagellin (Sigma-Aldrich), 1 μM CpG-ODN (Integrated DNA Technologies, Coralville, IA, http://www.idtdna.com/Home/Home.aspx), 1 μg/ml fibronectin III1c or 45 kDa (Sigma-Aldrich), and 5 μg/ml LL-37 (Innovagen, Lund, Sweden, http://www.innovagen.se).
Reverse Transcription-Polymerase Chain Reaction
The hMSCs (70% confluence) were washed, and total RNA was isolated with 1 ml of TriReagent as standard (Invitrogen). Potential DNA contamination in the RNA sample was removed by Turbo DNA-free treatment (Ambion, Austin, TX, http://www.ambion.com). cDNA was elaborated from RNA using SuperScript II (Invitrogen). TLR transcripts were amplified using the primers published  (Integrated DNA Technologies). Human peripheral blood mononuclear cell (PBMC) RNA was used as a positive control. Polymerase chain reaction (PCR) assay cycles were as follows: 94°C for 2 minutes, 35 cycles of 94°C for 20 seconds, 56°C for 30 seconds, and 72°C for 30 seconds .
Human MSCs were harvested and analyzed by flow cytometry with a FACSCalibur flow cytometer (BD Biosciences, San Diego, http://www.bdbiosciences.com) as described previously . Intracellular antibody staining was achieved after fixation and permeabilization of the cells as indicated by the manufacturer (cytofix/cytoperm buffers; BD Biosciences, San Jose, CA).
Fluorescence immunocytochemistry was performed on cells grown to near confluence (70%) on chamber slides, fixed, and permeabilized with BD cytofix/cytoperm buffer (BD Pharmingen, San Diego, http://www.bdbiosciences.com/index_us.shtml). The primary antibodies diluted in stain buffer in the appropriate concentration (ratio of 0.5 μg of antibodies per 1 × 106 cells) were added for 1 hour at 4°C in the dark. Next, unbound primary antibody was washed twice with BD cytoperm wash buffer. The secondary antibody (Alexa 488-conjugated; Molecular Probes, Eugene, OR, http://probes.invitrogen.com) was diluted in stain buffer in the appropriate concentrations and added for 1 hour at 4°C in the dark. Slides were again washed twice prior to 4,6-diamidino-2-phenylindole staining and mounting with ProLong Gold antifade reagent (Molecular Probes). Micrographs were taken on a Zeiss Axioplan 2 fluorescence microscope (Carl Zeiss, Jena, Germany, http://www.zeiss.com) with Intelligent Imaging Innovations deconvolution hardware and software (SlideBook version 4, Intelligent Imaging Innovations, Denver, http://www.intelligent-imaging.com).
Western Blot Analysis
Cells were plated on 10-cm plates to 70% confluence at 37°C before treatments. The cells were washed twice with cold phosphate-buffered saline (PBS), and protein was isolated as standard (mammalian protein extraction reagent [Pierce, Rockford, IL, http://www.piercenet.com]) containing a protease inhibitor cocktail (Roche Applied Science, Indianapolis, IN, http://www.roche-applied-science.com), a phosphatase inhibitor cocktail (Sigma-Aldrich), and 1 mM phenylmethylsulfonyl fluoride (Sigma-Aldrich). Lysates were clarified by centrifugation at 14,000 rpm at 4°C. Protein concentration was measured by the Micro BCA Protein Assay Kit (Pierce). Lysates (50 μg) were resolved on 4%–12% SDS-polyacrylamide gels and transferred to nitrocellulose membranes (Invitrogen). Membranes were blocked with 5% nonfat dried milk in PBS containing 0.2% Tween-20 (PBST) for 1 hour at room temperature, and blots were incubated at 4°C overnight with the primary antibodies. The blots were then washed in PBST and incubated with species-specific IgG conjugated to horseradish peroxidase (1:5,000; Amersham Biosciences, Piscataway, NJ, http://www.amersham.com) for 1 hour at room temperature. Antigen-antibody complexes were visualized after exposure to x-ray film by enhanced chemiluminescence (Amersham Biosciences).
TLR Pathway cDNA Array Assay
hMSCs were treated with TLR ligands for 4 hours. Total RNA was isolated with RNAzol (Invitrogen) and analyzed as per the manufacturer's instructions (human TLR pathway-specific gene expression profiling system; SuperArray Bioscience Corporation, Frederick, MD, http://www.superarray.com) on an iCycler iQ5 Real-Time PCR Detection System (Bio-Rad, Hercules, CA, http://www.bio-rad.com). The raw data from both the control and the treated groups were obtained and uploaded into GEarray Analyzer software (SuperArray Bioscience) for analysis and verification of appropriate experimental procedures.
Cytokine Antibody Array Assay
Conditioned medium from TLR ligand-treated or untreated hMSC cultures were tested for cytokine, chemokine, and matrix metalloprotease inhibitor secretion by cytokine antibody arrays following the manufacturer's instructions (human cytokine array 3, MA6150; Panomics, Redwood City, CA, http://www.panomics.com).
Fluorescence Bead Assay
Conditioned medium from similarly treated hMSC cultures was tested for cytokine or chemokine secretion by fluorescence bead immunoassay per the manufacturer's instructions (Bender MedSystems, Inc., Burlingame, CA, http://www.bendermedsystems.com). Primary hMSC cultures were treated with various TLR ligands for 48 hours prior to harvesting and concentrating the spent culture medium with Centricon-10 (fivefold) and loading 100 μl per sample in triplicate. Following flow cytometry, the data were analyzed as indicated by manufacturer (FlowCytomixPro version 2.2; Bender MedSystems, Millipore Corp., Bedford, MA, http://www.millipore.com).
Transwell Migration Assay
Migration assays were performed in transwell inserts with 8-μm pore membrane filters (BD Biosciences, San Jose, CA). Human MSCs were grown to subconfluence (70%) prior to harvesting by trypsinization and labeling with CellTracker green (1 μM; Molecular Probes) for 1 hour at 37°C. Fluorescently labeled hMSCs (2.5–5 × 105 cells per well in 300 μl) were loaded onto the upper chamber, and 500 μl of hMSC growth medium with chemotactic factors or TLR ligands, as indicated, was loaded onto the bottom chamber. After overnight incubation, the upper sides of the filters were carefully washed with cold PBS, and nonmigrating cells remaining were removed with a cotton-tipped applicator. Fluorescence images of the migrating cells were collected using a Nikon TE300 inverted epifluorescence microscope (Nikon USA, Lewisville, TX, http://www.nikon.com) with software (DP Controller v188.8.131.52; Olympus, Tokyo, http://www.olympus-global.com). Each experiment was performed in triplicate with two separate hMSC donors. Data are expressed as numbers of counted, migrated cells per ×200 field micrograph for each sample well and normalized to those cell counts obtained for the untreated control.
Data are shown as average ± SEM. Multiple group comparison was performed by one-way analysis of variance (ANOVA) followed by the Bonferroni procedure for comparison of means. Comparison between any two groups was analyzed by the two-tailed Student's t-test or one-way ANOVA (Prism4; GraphPad Software, Inc., San Diego, http://www.graphpad.com). Values of p < .05 were considered statistically significant.
TLRs Are Expressed in hMSCs
Initially, a set of primers specific for human TLR1–10 cDNA was used in reverse transcription (RT)-PCR analyses to establish their expression in hMSCs . As a positive control, RNA from PBMCs was tested with the same primer sets in the assay. In agreement with recent reports, this strategy convincingly revealed the presence of TLRs 1, 2, 3, 4, 5, and 6 (Fig. 1A) . TLRs 1, 2, 4, 5, and 8 RNA expression were confirmed for PBMCs. Most striking was the expression of TLR3 RNA in hMSCs but not PBMCs and the lack of TLR8 expression in hMSCs, in contrast to PBMC expression.
Measurement of TLR protein expression in hMSCs was achieved by flow cytometry and immunofluorescence assays with specific antibodies to human TLR proteins. Flow cytometry analyses indicated that primary cultures of hMSCs contain ample amounts of MyD88 and TLRs 2, 3, 4, and 7 (Fig. 1B). TLR9 protein levels appeared to be the lowest measured (Fig. 1B). Verification of hMSC populations was also achieved by flow cytometry with established surface markers as described in Materials and Methods (Fig. 1B). Treatment of the hMSCs with the established TLR ligands generally resulted in diminished receptor levels in flow cytometry analysis, most likely because of receptor activation, internalization, and degradation, as evident from the leftward shift of the plot measured for the ligand-treated samples (Fig. 1C, blue line). Treatment of the hMSCs with the TLR9 agonist CpG-ODN was an exception and did not follow this pattern (data not shown). Consistent findings were noted in immunofluorescence assays (Fig. 1D). Note that TLR agonist treatment leads to endosomal-like partitioning for both TLR2 and TLR4. TLR3 expression was diffuse in the cytoplasm and along the cell's edge. TLR3 agonist treatment led to a more focused expression of TLR3 along the cell's edge, as well as to endosomal-like compartments.
TLR Stimulation of hMSCs Leads to Activation of Expected Downstream Signaling Molecules
TLRs within hMSCs were stimulated for 1 hour by various ligands and assessed by Western blot analysis to examine their downstream signaling capabilities (Fig. 2). Interestingly, treatment of the hMSCs with poly(I:C) the ligand for TLR3 resulted in the greatest activation of the nuclear factor κB (NF-κB) pathway. LPS treatments also led to increased phospho-IKKα/β expression, and this expression was dampened by combined LL-37 treatment, as previously reported . CpG-ODN (TLR9) treatment of hMSCs appeared not to affect these pathways. Analysis of phosphatidylinositol 3-kinase pathway stimulation upstream of both the MAPK and NF-κB pathways revealed that LPS, LL-37, poly(I:C), and fibronectin fragment (III1c) stimulation also activated this pathway.
TLR Stimulation in hMSCs Triggers the Induction of Cytokines, Chemokines, and Other Established TLR-Regulated Genes
To identify the genomic consequences of TLR stimulation within hMSCs, focused microarray analysis was performed. The hMSCs were treated with various TLR ligands for 4 hours prior to RNA isolation and array analysis of 84 TLR-related genes. The most dramatic fold changes observed for several genes in the treated over untreated controls are recorded in Table 1. The results are arranged by molecule type: TLR, then cytokines and chemokines, followed by other downstream signaling genes. Tumor necrosis factor α (TNFα) gene expression was enhanced for all the TLR ligands tested, confirming TLR stimulation within the hMSCs. Similar to the results presented above, unique expression patterns resulted for each TLR agonist used. Strikingly, regardless of the agonist used to treat the hMSCs, there was uniform vast induction of TLR3. Lower induction for TLR6 and TLR9 was also noted following LPS treatment of hMSCs. Poly(I:C) treatment of hMSCs led to almost exclusive TLR3 induction compared with other TLRs, whereas LL-37 treatment led to modest induction of TLR3, TLR4, and TLR6. Fibronectin III1C treatment followed the LPS profile but in a more dampened way. LPS treatment caused increased expression of many cDNAs, including CXCL10 interferon-gamma-inducible protein (IP10), interleukin (IL) 6, IL8, interferon (IFN) 1β, and NF-κB. Poly(I:C) exposure caused the highest induction of the chemokine ligand CCL2 or monocyte chemotactic protein 1 (MCP1). This treatment also caused induction of IRAK2, CXCL10 (IP10), IL6, IL8, IFN1β, and NF-κB.
Table Table 1.. %TLR stimulation in human MSCs (hMSCs) triggers the induction of cytokines, chemokines, and other established TLR-regulated genes
Enhanced chemokine and cytokine secretion by the hMSCs was demonstrated by both cytokine antibody array (Fig. 3A) and fluorescence bead array (Fig. 3B) assays. Conditioned medium from treated hMSC cultures tested by these methods also resulted in unique cytokine and chemokine secretion patterns depending on the ligand used. For comparison, those genes found in both the cDNA array and the cytokine antibody array are shown in boldface italic in Table 1. Cytokine antibody arrays confirmed that LPS treatment resulted in the increased secretion of CXCL10 (IP10), IL6, IL8, and TNFα. The increased secretion of IL12 was also noted following both poly(I:C) and CpG-ODN exposure of the hMSCs. IL4 secretion was uniquely noted following LL-37 treatment, whereas matrix metalloproteinase 3 was secreted following fibronectin-fragment treatment of the hMSCs (data not shown). Similar secretion patterns were observed in fluorescence bead array (Fig. 3B). Overall, the secretion pattern of LPS-treated cells appears to favor proinflammatory mediators such as IL1β, IL6, and TNFα, whereas poly(I:C)-mediated secretion patterns appear to favor anti-inflammatory mediators such as IL10 and IL12.
TLR Stimulation Within hMSCs Triggers Their Enhanced Migration
The effect of TLR stimulation on hMSC migration was examined by transwell migration assay. Several representative TLR ligands (as indicated) were added as chemoattractant on the bottom wells, and single-cell suspensions of hMSCs were loaded on top inserts. Consistently, poly(I:C) treatment enhanced the migration response from exposed hMSCs (Fig. 4A). LPS, CpG-ODN, and LL-37 resulted in moderate migration by the exposed hMSCs. Fibronectin (fragments III1C) and flagellin resulted in limited migration. Addition of the TLR ligands prior to migration assay resulted in improved migration (data not shown). As expected, minimal hMSC migration was measured in serum-free medium .
TLR3 Is Critical to the Migration Responses of the Stimulated hMSCs
The collective results indicated that TLR3 and its downstream signaling consequences are critical to hMSCs stress responses. Consequently, hMSCs were treated in the transwell migration assay with a neutralizing TLR3 antibody to test this notion. hMSCs were preincubated for 1 hour with a human TLR3-specific monoclonal antibody or an isotype IgG control prior to loading in transwell migration assay inserts. After overnight incubation, migration was quantified as indicated in Materials and Methods. Regardless of chemotactic factor used, preincubation of the hMSCs with anti-TLR3 antibody for 1 hour prior to migration assay inhibited their migration by at least 55% (Fig. 4B). By contrast, pretreatment of the hMSCs with an isotype IgG control led to minimal inhibition of hMSC migration (10%).
To understand the molecular details of how hMSCs sense stress, mobilize, and engraft at inflamed, injured, or pathological sites (tumors) to fulfill their repair function, we investigated TLRs and their downstream signaling consequences in these cells. Concurrent with our study, hematopoietic and other adult marrow-derived stem cells were reported to express TLRs [7, 9]. However, these reports mostly focused on the role of TLRs in stem cell proliferation and their potential role in disrupting the differentiation capability of these stem cells. These reports did not focus on the role of TLRs in critical stress responses of stem cells as described here. Specifically, since hMSCs are increasingly being used in cell-based therapies, the molecular details that drive their migration, recruitment, and engraftment is critical to improving controlled and desired clinical outcomes. We report here that stimulation of TLRs within hMSCs leads to activation of established TLR signaling pathways and increased secretion of cytokines and chemokines and promotes their migration. From these observations, we propose that TLR stimulation within hMSCs may be one critical mechanism that specifically drives their stress responses. Notably, it appears that of all the expressed TLRs in this cell population TLR3 stimulation specifically plays a significant role in hMSCs stress responses. This observation is unique to this bone marrow-derived cell and is different from immune cells, where TLR4 is the established LPS sensor responsible for affecting innate and adaptive immune responses.
In agreement with recent reports, hMSCs were found to express TLRs 1, 2, 3, 4, 5, 6, and 9 RNA by RT-PCR assays (Fig. 1A). Protein expression of the corresponding gene products was confirmed by flow cytometry and immunofluorescence assays [7, 9]. In addition, it was similarly found that hMSCs express reduced levels of TLR9 . Typically, overall TLR expression in hMSCs was high, with mean fluorescence intensities (MFIs) comparable to those of established hMSC markers (e.g., CD90). The trend of TLR expression was the same for all the donors tested (n > 7) and was as follows, from highest to lowest MFI: TLR5 ≫ TLR2 = TLR3 = TLR4 = TLR6 = TLR7 ≫ TLR9. Stimulation of the TLRs was found to lead to their transient internalization and degradation, indicating that TLRs are activated after ligand treatment of the hMSCs (Fig. 1B–1D). TLR9 appeared to be the only exception, with little degradation noted after ligand treatment of the hMSCs (data not shown).
We report for the first time that downstream signaling consequences of the stimulated TLRs within the hMSCs yielded specific activation patterns of the NF-κB, MAPK, and AKT pathways, consistent with the expected activation of established TLR-downstream pathways (Fig. 2). All of these signaling pathways are known to affect cell migration and invasion properties of the stimulated cell [19, –21]. Notably, treatment of the hMSCs with poly(I:C), the ligand for TLR3, resulted in the greatest activation of the NF-κB pathway. Similar effects for this ligand were reported for murine MSCs .
Exposure of the hMSCs to TLR ligands led to unique cytokine and chemokine secretion as established by cytokine antibody arrays and confirmed with bead arrays (Fig. 3A, 3B). LPS treatment of hMSCs distinctively led to the induced secretion of CCL5 (RANTES) and CXCL10 (IP10). Poly(I:C) treatment resulted in dramatically induced expression of CCL2 (MCP1). In our hands, the hMSCs secreted high levels of IL6 constitutively and indiscriminately after treatment of the cells with growth factors, hypoxia, erythropoietin, or various TLR ligands (unpublished observations) [17, 22]. By contrast, IL6 secretion was used previously as the criteria to primarily follow TLR2-related pathways in murine MSCs .
These observations support the notion that the different TLRs expressed within the hMSCs are competent in relaying distinct stress signals. Critically, many of the signaling consequences of TLR stimulation, such as secretion of CCL5 (RANTES), IL8, and TNFα, imply that TLRs may partially mediate the immune modulating responses established for hMSCs . Furthermore, these observations suggest that by stimulating specific TLR molecules, different immune responses might be elicited by the hMSCs at the site of engraftment. In fact, the prediction from our studies is that LPS pretreatment of hMSCs would create a proinflammatory milieu because of elevated IL6, TNF-α, and CCL5 (RANTES) secretion. Conversely, poly(I:C) pretreatment of hMSCs is expected to yield an anti-inflammatory environment because of increased IL10 and IL12 secretion. Intriguingly, a recent report dealing with the use of LPS moieties as vaccine adjuvants supports this notion . The improved adjuvant effect of the monophosphoryl lipid A (MPLA) moiety was attributed to the fact that this adjuvant, unlike lipid A (LPS), activates only one arm of the TLR signaling pathway. Instead of activating both the MyD88-dependent pathway that leads to inflammation and the TIR domain-containing adaptor inducing interferon beta (TRIF)-dependent pathway that leads to T-cell activation and promotes effective immune responses, this moiety only activates TRIF signaling and potentially avoids the harmful side effects caused by inflammation. Essentially, MPLA acts as a TLR3 ligand (poly[I:C]), since TLR3 signaling is unique and generally mediated exclusively through TRIF and not MyD88 signaling pathways.
It should be noted that although there was consistency in the ability of TLR ligands to induce discrete or unique cytokine and chemokine secretion patterns, depending on the ligand used, we did observe donor variability in these patterns. For example, the level of secretion of CCL5 (RANTES) following LPS treatment in one donor pool was by far the highest of the measured chemokines, whereas IL8 levels were higher than those of CCL5 (RANTES) secretion following LPS stimulation of another donor pool. These findings might be explained by subtle differences in TLR-signaling for each donor hMSC. In addition, our study confirms previous reports demonstrating that when LL-37 is combined with LPS treatment, the overall effect is an observed dampening of the LPS effect (Figs. 2, 3; Table 1) . Similarly, stimulation of hMSCs with fibronectin components yielded limited TLR-specific results, most likely reflecting the fact that endogenously derived agonists affect not only TLRs but also several other receptor classes on the cells.
Importantly, our study demonstrated that specific TLR stimulation drives the migration of the hMSCs (Fig. 4A). Although there was cursory mention of this potential effect by TLRs in a recent study of murine MSCs, this concept was not the focus of that report . In that study, the authors concluded from a rudimentary wound-healing assay that the TLR2 ligand impaired murine MSC migration . It is not surprising that there are differences in migration responses mediated by specific TLR ligands on MSC from different species, particularly in light of the fact that it was recently reported that TLR-mediated responses are both species-specific and cell-type-specific . Admittedly, the effect on hMSC migration by TLR stimulation could be improved from our original strategy. For instance, there is the possibility that the order of exposure to the hMSCs of the TLR ligands prior to or along with other stress signal molecules can be further explored to achieve more dramatic manipulation of hMSC migration and invasion capabilities. This point is particularly important since a recent report stated that CCL5 (RANTES)-driven hMSC migration was highly induced by pretreatment of the hMSCs with TNFα .
Several lines of evidence point uniquely to TLR3 as primarily mediating the stress migration responses within hMSCs. First, TLR3 appears to be highly expressed in this cell population, as demonstrated from the RT-PCR, flow cytometry, and immunofluorescence assay results presented (Fig. 1). Next, the dramatic effects of TLR3 ligand or poly(I:C) treatment of hMSCs that led to the specific induction of NF-κB, MAPK, and AKT pathways, as well as cytokine, chemokine, and other TLR pathway genes, support this notion (Figs. 2, 3; Table 1). Migration assays also demonstrated that poly(I:C) exposure of the hMSCs led to one of the most enhanced effects on hMSC migration when compared with other TLR ligands (Fig. 4A). This was confirmed by transwell migration (Fig. 4A) and Boyden chamber migration (data not shown) assays with several of the hMSC donors. This observation prompted further investigation. Thus, transwell migration assays performed with anti-TLR3 neutralizing antibodies resulted in consistent inhibition of hMSC migration (Fig. 4B). Surprisingly, this inhibition does not depend on specific TLR3 ligand exposure, suggesting that hMSC migration is dependent on TLR3 regardless of its activation. A detailed investigation is currently under way in the laboratory to strengthen support for these observations, as well as to study the possibility of manipulating this TLR pathway for the in vivo guided manipulation of infused hMSCs to improve their migration, invasion, and immune modulating responses at targeted sites. In summary, this study defines a novel TLR-driven stress and immune modulating response for hMSCs that is critical to consider in the design of stem cell-based therapies.
We demonstrated that stimulation of TLRs within hMSCs leads to the activation of established TLR signaling pathways. These activated pathways mediate the secretion of discrete patterns of cytokines and chemokines depending on the TLR ligand used. This observation critically implicates these receptors in the immune modulating function established for hMSCs. We also show that TLR stimulation particularly promotes their migration capabilities. Thus, TLR stimulation may be one mechanism that specifically drives the recruitment, migration, and immune modulating function of the hMSCs at injured or stressed sites. Along with establishing a new aspect of their biology, the identification of TLRs as critical mediators of stress responses within hMSCs also provides a novel target to exploit in the improvement of stem cell-based therapeutic strategies.
Disclosure of Potential Conflicts of Interest
The authors indicate no potential conflicts of interest.
We thank the members of the laboratory of Donald Phinney for their generous assistance with initially acquiring and cultivating the human mesenchymal stem cells. We also thank Joanna DeSalvo for her assistance in the preparation of the figures and Elizabeth Norton for insightful discussions of our work. This work was supported by NIH Grant AI056229, NIH Grant 1P20RR20152-01, and a research grant award from Cancer Association of Greater New Orleans.