Antiinflammation Effect of Human Placental Multipotent Mesenchymal Stromal Cells Is Mediated by Prostaglandin E²via  a Myeloid Differentiation Primary Response Gene 88-dependent Pathway

• Human term placenta is a rich source of multipotent mesenchymal stromal cells, which might represent a novel approach to antiinflammatory therapy, but the mechanisms involved are not well understood

• The antiinflammation effect of human placental multipotent mesenchymal stromal cells is mediated, at least in part, by prostaglandin E2 via  a myeloid differentiation primary response gene 88-dependent pathway

UP-REGULATION of inflammatory molecules, including tumor necrosis factor-α (TNF-α), interleukin (IL)-6, macrophage inflammatory protein-2 (MIP-2), intercellular adhesion molecule 1 (ICAM-1), and prostaglandin E²(PGE²), is essential in mediating the development of systemic inflammatory response during sepsis.1,–,3Endotoxin-induced up-regulation of inflammatory molecules is regulated by nuclear factor-κB (NF-κB) and mitogen-activated protein kinases (MAPKs).4,5It is well-established that cellular recognition of endotoxin and subsequent endotoxin binding mediated by the complex of toll-like receptor 4 (TLR4) and myeloid differential-2 (MD-2), a protein that is associated with extra-cellular domain of TLR4, is important for the activation of NF-κB and MAPKs.6,–,8In addition, the interaction between TLR4 and myeloid differentiation primary response gene 88 (MyD88) following TLR4/MD-2 complex activation is also crucial in mediating the activation of NF-κB and MAPKs induced by endotoxin.9,10 

Multipotent mesenchymal stromal cells (MSCs), a group of stem cells isolated from tissues such as postnatal bone marrow, adipose tissue, placenta, and scalp tissue, have been shown to possess the capacity for prolonged self-renewal and retain the potential to differentiate into a variety of more specialized cell types.11Moreover, data from both human and animal models confirmed that MSCs are able to escape immune recognition and subsequent allogenic rejection because of the apparent hypoimmunogenic nature of MSCs.12Bone marrow MSCs have been shown to possess significant immune modulatory capacity.12,–,14Previous studies clearly demonstrated that bone marrow MSCs not only inhibit the production of proinflammatory cytokines but also promote the production of antiinflammatory cytokines (e.g. , IL-10).13,14However, MSCs are a rare population in adult human bone marrow, and the number of bone marrow MSCs significantly decreases with aging.15 

In contrast to adult bone marrow, human term placenta is a rich source of MSCs.15,–,17The differentiation potentials of human placental MSCs (referred to as hPMSCs in this study) are similar to those of bone marrow MSCs.15Human placental MSCs are also hypoimmunogenic.15Previous data indicated that hPMSCs also possess significant immune modulatory capacity.17However, the question of whether hPMSCs could modulate the up-regulation of inflammatory molecules during sepsis remains unstudied. To elucidate further, we thus conducted this cellular study with the hypothesis that hPMSCs could inhibit the up-regulation of inflammatory molecules and activation of NF-κB and MAPKs in endotoxin-activated murine macrophages. The effects of hPMSCs on modulating endotoxin binding, TLR4/MD-2 complex activation, and the interaction between TLR4 and MyD88 in endotoxin-activated murine macrophages were also studied.

Human placental MSCs were isolated from human term placenta and grown, as we previously reported.18Placental tissue was obtained after women gave informed consent and all experiments were approved by the Institutional Review Board of Mackay Memorial Hospital (MMH-I-S-226), Taipei, Taiwan. Immortalized murine macrophages (RAW264.7 cells) were also grown, as we previously reported.19RAW264.7 cells and hPMSCs were incubated in a humidified chamber at 37°C in a mixture of 95% air and 5% CO². To induce the production of the molecules to be investigated, confluent RAW264.7 cells with or without hPMSCs' coincubation were stimulated with lipopolysaccharide (100 ng/ml, E. coli  Serotype 0127:B8 endotoxin; Sigma-Aldrich, St. Louis, MO), as we previously reported.20 

Freshly harvested culture media were analyzed for the concentrations of TNF-α, IL-6, IL-10, MIP-2, ICAM-1, and prostaglandin E²using enzyme-linked immunosorbent assays (ELISA) (Colorimetric ELISA Kits of TNF-α, IL-6, IL-10, ICAM-1, and PGE²; Pierce Biotechnology, Inc., Rockfold, IL; MIP-2 ELISA kit; R&D Systems, Inc., Minneapolis, MN), following the manufacturer' instructions.

Preparations of nuclear and cytosolic extracts were performed as we previously reported.20After separation by gel electrophoresis, the proteins were transferred to nitrocellulose membranes (Bio-Rad Laboratories, Hercules, CA). For nuclear extracts, the nitrocellulose membranes were incubated overnight at 4°C in primary antibody solution of phosphorylated NF-κB p65 (Phospho-NF-κB p65 [Ser536], 1:500 dilution; Cell Signaling Technology, Inc., Danvers, MA) or Histone H3 (internal standard, 1:500 dilution; Cell Signaling Technology) to facilitate assaying NF-κB nuclear translocation. For cytosolic extracts, the membranes were incubated overnight at 4°C in primary antibody solution of cytosolic inhibitor of NF-κB (I-κBα) (1:500 dilution; Cell Signaling Technology), phosphorylated I-κBα (Phospho-I-κBα[Ser32], 1:1,000 dilution; Cell Signaling Technology), or Actin (internal standard; 1:5,000 dilution; Millipore Corporation; Burlington, MA) to facilitate assaying I-κBα phosphorylation. For cytosolic extracts, the membranes were also incubated overnight at 4°C in primary antibody solution of phosphorylated extracellular regulated kinase (ERK) (1:500 dilution, polyclonal p-ERK1/2 antibody; Santa Cruz Biotechnology, Inc., Santa Cruz, CA), phosphorylated c-jun N-terminal kinase (JNK) (1:500 dilution, polyclonal p-JNK1/2 antibody; Santa Cruz Biotechnology), phosphorylated p38 MAPK (1:200 dilution, polyclonal p-p38 MAPK antibody; Santa Cruz Biotechnology), or Actin (1:5,000 dilution; Millipore) to facilitate assaying MAPK activation. Horseradish peroxidase conjugated antimouse IgG antibody (Amersham Pharmacia Biotech, Inc., Piscataway, NJ) was used as the secondary antibody. Bound antibody was detected by chemiluminescence (ECL plus kit; Amersham Pharmacia Biotech) and chemiluminescence film (Hyperfilm, Amersham Pharmacia Biotech). The protein band densities were quantified using densitometric technology (Scion Image for Windows; Scion Corp, Frederick, MD).

To further elucidate the effects of hPMSCs on NF-κB nuclear translocation, immunofluorescent staining was performed. The protocol for immunofluorescent staining was adapted from a previous report.21In brief, cell cultures were grown on glass coverslips. After reaction, cell cultures were fixed, permeabilized, and blocked. Coverslips were incubated 30 min in primary antibody solution of phosphorylated NF-κB p65 (Phospho-NF-κB p65 [Ser536], 1:100 dilution; Cell Signaling Technology). After washing, cells were incubated with fluorescent rhodamine isothiocyanate-conjugated secondary antibody (Jackson ImmunoResearch Inc., West Grove, PA) for 30 min. Nuclear counterstaining with diamidino-2-phenylindole (Pierce Biotechnology) was then performed and the cells were imaged using a confocal microscope (TCS SP5 AOBS; Leica Microsystems CMS GmbH, Mannheim, Germany).

NF-κB activation is also evidenced by increases in NF-κB-DNA binding activity.22To faciliate investigation, cell cultures were grown in culture dishes. After reaction and harvesting, nuclear extracts were prepared as aforementioned. EMSA was then performed using a chemiluminescence EMSA kit (Panomics Inc., Fremont, CA), as previously described.23 

Protocols for assaying endotoxin binding to macrophages and membrane TLR4/MD-2 complex expression using flow cytometry were adapted from previous reports.24,25In brief, cell cultures were treated with phosphate buffered saline (PBS) or fluorescein-labeled lipopolysaccharide (LPS-FITC, 100 ng/ml; Sigma-Aldrich) to facilitate endotoxin binding assay. After reaction, the cells were washed with PBS and resuspended in medium. Then, the binding of LPS-FITC to RAW264.7 cells was analyzed using a flow cytometer (Cytomic FC50 Flow Cytometer; Beckman Coulter, Inc., Fullerton, CA). In addition, cell cultures were treated with PBS or lipopolyaccharide to facilitate TLR4/MD-2 complex expression assaying. After reaction, the cells were also washed with PBS and resuspended in medium. Then, the cells were incubated for 45 min with primary antibody solution of TLR4/MD-2 antibody (1:150 dilution, monoclonal TLR4/MD-2 antibody; BD Pharmigen, San Diego, CA). A FITC-labeled antimouse IgG antibody (1:200 dilution; BD Pharmigen) was used as the secondary antibody. Then, TLR4/MD-2 complex cell surface expression was measured by a flow cytometer (Beckman Coulter).

Protocols for immunoprecipitation assay were adapted from a previous report.26In brief, cell lysates (1 mg of protein) were incubated with TLR4 antibody (2 μg; Invivogen Corp., San Diego, CA) at 4°C for 24 h. Then, 50% protein A-agarose beads (10 μl; Sigma-Aldrich) were added to cell lysates and mixed at 4°C for 24 h. After collection, the immunoprecipitates were washed and then subjected to electrophoresis followed by protein transfer to nitrocellulose membranes (Bio-Rad). The membrane was then blotted with the primary antibody solution of MyD88 (1:200 dilution; Santa Cruz Biotechnology). Then, the procedures of blocking, secondary antibody conjugation, and bound-antibody detection were performed, as previously described in immunoblotting assay.

For outcomes measured at a single time-point, one-way ANOVA was used to determine the differences between the groups. Post hoc  pair-wise comparisons were performed using the Tukey test. For outcomes measured serially over time, repeated measures ANOVA was performed to determine the between-subject effects (i.e. , group effects), within-subject effects (i.e. , time effects), and between-subject by within-subject effects (i.e. , group by time interaction effects). The significance level (α) was set at 0.05. Multiple comparisons between groups at each time-point were performed with the Bonferroni correction for the significance level (i.e. , 0.05 divided by n comparisons). All data were derived from three independent experiments performed in duplicates (n = 6). Predictive Analytics SoftWare Statistics 18.0 (IBM SPSS; IBM Cooperation, Armonk, NY) was used to analyze the data.

Series of preliminary studies, i.e. , low, middle, or high cell number of hPMSCs (i.e. , cell number: hPMSCs vs . RAW264.7 cells = 1:10⁴, 1:10³, or 1:10², respectively) added to RAW264.7 cells at 4 h before, immediately after, or 4 h after lipopolysaccharide administration (i.e. , pre-, co-, and posttreatment, respectively) were preformed to elucidate the effects of hPMSC administration timing and hPMSC cell number on inflammatory molecule IL-6 production in RAW264.7 cells. Our preliminary data revealed that the effect of lipopolysaccharide on inducing IL-6 up-regulation in RAW264.7 cells could only be mitigated by pretreatment with high cell number of hPMSCs. We thus decided to employ this hPMSC administration protocol for the rest of the study.

RAW264.7 cells alone or RAW264.7 cells with hPMSCs were treated with PBS or lipopolysaccharide (designated as the PBS+R, PBS+R+S, LPS+R, and LPS+R+S group, respectively). After reaction for 0, 6, 12, 18, and 24 h, culture media were harvested and assayed. Changes in inflammatory molecules' up-regulation were summarized in figure 1. Repeated measures ANOVA revealed significant group (P < 0.001), time (P < 0.001), and interaction effects (P < 0.001) in TNF-α, suggesting that the magnitude of changes in TNF-α over time was significantly different among groups. Repeated measures ANOVA also revealed that the magnitudes of changes in IL-6, MIP-2, ICAM-1, PGE², and IL-10 over time were significantly different among groups (all P < 0.001).

Multiple comparisons between groups revealed that the basal TNF-α concentrations of these four groups were comparable (all P > 0.008). Moreover, the TNF-α concentrations of the PBS+R+S group were comparable with those of the PBS+R group throughout the experiment (all P > 0.008). In contrast, the TNF-α concentrations of the LPS+R and the LPS+R+S groups harvested at 6, 12, 18, and 24 h after reaction were significantly higher than those of the PBS+R group (all P < 0.001). Moreover, the TNF-α concentrations of the LPS+R+S group harvested at 6 and 12 h after reaction were significantly lower than those of the LPS+R group (P = 0.002 and 0.003). Multiple comparisons between groups revealed similar trends in IL-6, MIP-2, and ICAM-1, and IL-10. In addition, the IL-6 concentrations of the LPS+R+S group harvested at 18 h after reaction were significantly lower than those of the LPS+R group (P = 0.008). Moreover, the MIP-2 and ICAM-1 concentrations of the LPS+R+S group harvested at 18 and 24 h after reaction were significantly lower than those of the LPS+R group (MIP-2: P < 0.001 and = 0.004; ICAM-1: P = 0.008 and 0.006). In contrast, the IL-10 concentrations of the LPS+R+S group harvested at 6, 12, 18, and 24 h after reaction were not significantly different from those of the LPS+R group (all P > 0.008).

Multiple comparisons between groups further revealed that the basal PGE²concentrations of the PBS+R+S and LPS+R+S groups were comparable and both were significantly higher than that of the PBS+R group (P = 0.007 and 0.006) and the LPS+R group (both P = 0.005). Moreover, the PGE²concentrations of the PBS+R+S group were significantly higher than those of the PBS+R group throughout the experiment (all P < 0.008). In addition, the PGE²concentrations of the LPS+R group harvested at 6, 12, 18, and 24 h after reaction were significantly higher than those of the PBS+R group (all P < 0.001). Moreover, the PGE²concentrations of the LPS+R+S group harvested at 12, 18, and 24 h after reaction were significantly lower than those of the LPS+R group (P = 0.008, 0.002, and 0.004).

In addition, hPMSCs treated with PBS or lipopolysaccharide were also included to serve as the control for hPMSCs. Repeated measures ANOVA revealed that the magnitudes of changes in TNF-α, IL-6, MIP-2, ICAM-1, PGE2, and IL-10 over time in hPMSCs treated with PBS or lipopolysaccharide were not significantly different from those of the PBS+R groups (data not shown).

As our data revealed low-grade PGE²production in the PBS+R+S and the LPS+R+S groups, the harvested culture media were further analyzed using human PGE²ELISA kit (MyBioSource LLC, San Diego, CA) to determine the source of PGE². Our data revealed that the human PGE²concentrations of the PBS+R and LPS+R groups were almost undetectable (fig. 2A). Moreover, the human PGE²concentrations of the PBS+R+S and LPS+R+S groups were low (approximately 8–10 pg/ml) (fig. 2A).

To elucidate the role of cell–cell contact in this regard, noncell contact culture using a transwell microplate (24 mm transwell with 0.4 μm pore polyester membrane insert; Corning Inc., Corning, NY)27was employed and the culture media were assayed. Our data revealed that the human PGE²concentrations of the noncell contact groups were also low (fig. 2B). In contrast, the basal total PGE²concentrations of the noncell contact groups were comparable (P > 0.05, fig. 2B). Moreover, in the noncell contact groups, the total PGE²concentrations of the LPS+R group harvested at 18 h after lipopolysaccharide were comparable with those of the LPS+R+S group (P > 0.05, fig. 2B). In addition, in the noncell contact groups, the MIP-2 concentration of the LPS+R+S group was significantly lower than that of the LPS+R group (P = 0.039, fig. 2C). However, in the noncell contact groups, the concentrations of TNF-α, IL-6, and ICAM-1 of the LPS+R+S group were comparable with those of the LPS+R group (P > 0.05, fig. 2C). Moreover, the concentrations of TNF-α, IL-6, MIP-2, and ICAM-1 of the noncell contact LPS+R+S group were significantly higher than those of the cell contact LPS+R+S group (all P < 0.05, fig. 2C).

Our data also revealed that the IL-10 concentrations of the LPS+R+S and the LPS+R groups in both the cell contact and the noncell contact groups were comparable (P > 0.05, fig. 2C). To elucidate whether this phenomenon was also true in primary macrophages culture, we then employed primary macrophages culture (designated as PM) isolated from C57Bl mouse (i.e. , 10 ml sterile ice-cold 1× PBS intraperitoneal injection followed by washing with PBS and resuspension in RPMI 1640 [Sigma-Aldrich] supplemented with 10% fetal bovine serum)28to facilitate investigation. The culture media were harvested and the IL-10 concentrations were assayed after reaction for 6 h, according to previous data.29Our data revealed that the IL-10 concentrations of the LPS+PM+S and the LPS+PM groups in both the cell contact and the noncell contact groups were also comparable (P > 0.05, fig. 2C).

RAW264.7 cells alone, RAW264.7 cells with hPMSCs, or RAW264.7 cells with hPMSCs plus the selective cyclooxygenase-2 (i.e. , the enzyme that mediates PGE²production) inhibitor NS-398 (1 μM; Sigma-Aldrich) were treated with lipopolysaccharide (designated as the LPS+R, LPS+R+S, and LPS+R+S+N groups, respectively). NS-398 was administered at 30 min before hPMSCs, according to previous data.29 

As expected, the concentrations of TNF-α, IL-6, MIP-2, ICAM-1, and PGE²of the LPS+R+S group were significantly lower than those of the LPS+R group (all P < 0.05, fig. 3). In contrast, the concentrations of TNF-α, IL-6, MIP-2, and ICAM-1 of the LPS+R+S+N group were significantly higher than those of the LPS+R+S group (P = 0.024, 0.018, 0.015, and 0.031, respectively) whereas the PGE²concentration of the LPS+R+S+N group was significantly lower than that of the LPS+R+S group (P = 0.003; fig. 3). Moreover, the IL-10 concentrations of the LPS+R, LPS+R+S, and LPS+R+S+N groups were not significantly different (all P > 0.05, fig. 3).

Repeated measures ANOVA revealed significant group, time, and interaction effects in phosphorylated NF-κB p65 and I-κBα (all P < 0.001). Multiple comparisons between groups revealed that the concentration of phosphorylated NF-κB p65 of the LPS+R+S group harvested at 15 and 45 min after reaction were significantly lower than those of the LPS+R group (P < 0.001 and 0.007, fig. 4A). The concentration of phosphorylated I-κBα of the LPS+R+S group harvested at 30, 45, and 60 min after reaction were also significantly lower than those of the LPS+R group (P = 0.006, 0.001 and 0.003, fig. 4B). Moreover, the phosphorylated NF-κB p65 concentrations of the LPS+R+S+N group harvested at 30 and 45 min after reaction were significantly higher than those of the LPS+R+S group (P = 0.005 and 0.006, fig. 4A). The phosphorylated I-κBα concentrations of the LPS+R+S+N group at 30, 45, and 60 min after reaction were significantly higher than those of the LPS+R+S group (P = 0.005, 0.004, and 0.002, fig. 4B).

According to the above-mentioned data, we chose to harvest the cells at 30 min after reaction to facilitate immunofluorescent staining and EMSA assays. Our data revealed that the fluorescence intensity of the LPS+R+S group was weaker than that of the LPS+R group, whereas the fluorescence intensity of the LPS+R+S+N group was stronger than that of the LPS+R+S group (fig. 4C). In addition, the EMSA data (fig. 4D) paralleled the immunofluorescent staining data.

Repeated measures ANOVA revealed significant group, time, and interaction effects in phosphorylated ERK, JNK, and p38 MAPK (all P < 0.001). Multiple comparisons between groups revealed that the concentration of phosphorylated ERK of the LPS+R+S group harvested at 45 and 60 min after reaction were significantly lower than those of the LPS+R group (P = 0.006 and 0.004, fig. 5). In addition, the concentration of phosphorylated JNK and p38 MAPK of the LPS+R+S group harvested at 30 min after reaction were significantly lower than those of the LPS+R group (P = 0.007 and 0.003, fig. 5). Moreover, the phosphorylated ERK concentration of the LPS+R+S+N group harvested at 60 min after reaction was significantly higher than that of the LPS+R+S group (P = 0.005, fig. 5). The phosphorylated p38 MAPK concentration of the LPS+R+S+N group harvested at 30 min after reaction was also significantly higher than that of the LPS+R+S group (P = 0.003, fig. 5). In contrast, the concentration of phosphorylated JNK of the LPS+R+S+N group was not significantly different from those of the LPS+R+S group (P > 0.017, fig. 5).

Endotoxin binding to RAW264.7 cells would result in a significant increase in mean fluorescence intensities (MFI) when assayed by flow cytometry.30Our data revealed that the 15-min MFI of the LPS+R group was significantly higher than those of the PBS+R group (P < 0.001, fig. 6A). However, the 15-min MFI of the LPS+R+S and LPS+R+S+N groups were comparable to that of the LPS+R group (fig. 6A). Moreover, endotoxin-induced TLR4/MD-2 complex activation would result in down-regulating the membrane TLR/MD-2 complex expression, i.e. , a significant decrease in MFI when assayed by flow cytometry.30Our data revealed that the 15-min MFI of the LPS+R group was significantly lower than those of the PBS+R group (P = 0.023, fig. 6B). However, the 15-min MFI of the LPS+R+S and LPS+R+S+N groups were also comparable with those of the LPS+R group (P > 0.05, fig. 6B).

Repeated measures ANOVA revealed significant group (P < 0.001), time (P < 0.001), and interaction effects (P = 0.004) in MyD88 protein concentrations in a TLR4-immunoprecipitation complex in cell lysates. The MyD88 protein concentrations of the LPS+R+S group harvested at 30 and 60 min after reaction were significantly lower than those of the LPS+R group (P = 0.006 and 0.005, fig. 7). Moreover, the MyD88 protein concentration of the LPS+R+S+N group harvested at 30 min after reaction was significantly higher than that of the LPS+R+S group (P = 0.008, fig. 7).

Inflammatory molecules' up-regulation is essential in mediating the development of systemic inflammatory responses and the subsequent multiple organ dysfunction during sepsis.1,–,3,31This concept was further supported by previous data that therapies aiming at decreasing inflammatory molecules' up-regulation could reduce the pathologic sequelae of sepsis.32Data from this study provide clear evidence to demonstrate the inhibition effects of hPMSCs on the up-regulation of inflammatory molecules in activated murine macrophages. Our findings thus support the possibility of clinical application of hPMSCs in the treatment of sepsis.

Data from this study also confirmed that hPMSCs significantly inhibited the activation of NF-κB and MAPKs, i.e. , the crucial upstream pathways that regulate the expression of inflammatory molecules during sepsis.4,5These data support the concept that hPMSCs may act through inhibiting the activation of NF-κB and MAPKs to exhibit their inhibition effects on the endotoxin-induced inflammatory molecules' up-regulation. Endotoxin-induced up-regulation of NF-κB and MAPKs is mediated by the TLR4/MD-2 complex-mediated cellular recognition of endotoxin.6,7However, our data revealed that the TLR4/MD-2 complex-mediated cellular recognition of endotoxin was not affected by hPMSCs. These data indicated that the observed inhibition effect of hPMSCs on the activation of NF-κB and MAPKs may involve other mechanisms.

Previous data demonstrated that the activation of the TLR4/MD-2 complex could subsequently lead to the activation of several crucial cellular signaling pathways, especially the MyD88-dependent signaling pathway.9,10Because the MyD88-dependent signaling pathway links the TLR4/MD-2 signaling and the activation of NF-κB and MAPKs,9,10it is likely that the effects of hPMSCs on inhibiting NF-κB and MAPKs activation may involve the MyD88 signaling pathway. Our data confirmed that the endotoxin-induced interaction between TLR4 and MyD88 was inhibited by hPMSCs. These data thus support the concept that hPMSCs may act through inhibiting the MyD88 signaling pathway rather than the TLR4/MD-2 complex-mediated cellular recognition to exhibit their effects on inhibiting inflammatory molecules' up-regulation and activation of NF-κB and MAPKs.

Data from this study confirmed the antiinflammation effect of hPMSCs. However, our data revealed that pretreatment and high cell number are required for hPMSCs to exhibit their antiinflammation effects. In addition, our data revealed that the cell–cell contact between hPMSCs and murine macrophages before lipopolysaccharide or PBS (i.e. , hPMSCs pretreatment) could induce low-grade but significant PGE²expression in murine macrophages. PGE²has long been considered as one of the proinflammatory molecules.3However, abundant data have revealed that PGE²also possesses potent antiinflammation effects.33,34Previous data further indicated that the dual effects of PGE²were concentration-dependent and PGE²at low concentrations possessed potent antiinflammation capacity.35As our data revealed low-grade PGE²expression in murine macrophages pretreated with hPMSCs, it is likely that the low-grade PGE²production induced by the cell–cell contact between hPMSCs and murine macrophages may be actively involved in mediating the antiinflammation effect of hPMSCs. This concept was confirmed by this study, as our data revealed that the effects of hPMSCs on inhibiting the up-regulation of inflammatory molecules, the activation of NF-κB and MAPKs, and the interaction between TLR4 and MyD88 could be reversed by the selective cyclooxygenase-2 inhibitor NS-398. Judging from these data, we thus conclude that the antiinflammation effect of hPMSCs is mediated by PGE²via  a MyD88-dependent pathway.

Certain study limitations do exist. First, though the MyD88-dependent signaling pathway is the predominant pathway that links the TLR4/MD-2 signaling and the activation of NF-κB and MAPKs,9previous data showed that the MyD88-independent signaling pathway (i.e. , the toll-IL-1 receptor-domain-containing adapter-inducing interferon-β and its related adaptor molecule) also participates in the linkage between TLR4/MD-2 signaling and the NF-κB activation.36As this study did not elucidate the effects of hPMSCs on the expression of the MyD88-independent signaling pathway, the question of whether the antiinflammatory effect of hPMSCs also involves the MyD88-indepent signaling pathway remains to be elucidated. Second, previous data indicated that bone marrow MSCs could induce the expression of the antiinflammatory cytokine IL-10 in macrophages, one of the crucial mechanisms that have been proposed to mediate the antiinflammation capacity of bone marrow MSCs.13,14,29However, our data revealed that hPMSCs did not pose significant effects on regulating the expression of IL-10 in both the immortalized cell line RAW264.7 cells and the primary macrophages culture. These data seemed to support the concept that hPMSCs posed no significant effects on regulating IL-10 expression in macrophages. The mechanism(s) underlying the observed discrepancy between bone marrow MSC and hPMSCs remains to be elucidated. Third, this study focused on the reactions within the first 24 h after endotoxin, (i.e. , the early phase of sepsis). It is well established that sepsis and septic shock are associated with overwhelming inflammatory response in the early phase and subsequent immune suppression in the later phase.37Judging from our data, it is likely that incorporation of hPMSCs as part of the therapies in the early phase of sepsis and septic shock should be beneficial. However, the question of whether hPMSCs could exert beneficial effects when administered in the later phase of sepsis and septic shock remains to be elucidated. Moreover, our data revealed that high cell numbers were essential for hPMSCs to exhibit the antiinflammation capacity. However, the safety of high cell number of hPMSCs remains to be elucidated. More data are needed before further conclusions can be drawn. Fourth, placenta is a rich source of hPMSCs. Nevertheless, limited accessibility of hPMSCs to clinicians remains one main obstacle for clinical application of hPMSCs. To overcome this obstacle, we aimed to develop therapies beyond hPMSCs (i.e. , therapies that can exert similar protective effects but without actually using hPMSCs). As our data highlighted the crucial role of PGE²and the MyD88 pathway in mediating the antiinflammation capacity of hPMSCs, we speculate that therapies aiming at modulating PGE²and/or MyD88 expression should be a beneficial therapeutic strategy against sepsis/septic shock, and thus warrants further investigation.

In summary, our data revealed that the antiinflammation effect of hPMSCs is mediated, at least in part, by PGE²via  a MyD88-dependent pathway.