CD106 Identifies a Subpopulation of Mesenchymal Stem Cells with Unique Immunomodulatory Properties

Mesenchymal stem cells (MSCs) reside in almost all of the body tissues, where they undergo self-renewal and multi-lineage differentiation. MSCs derived from different tissues share many similarities but also show some differences in term of biological properties. We aim to search for significant differences among various sources of MSCs and to explore their implications in physiopathology and clinical translation. We compared the phenotype and biological properties among different MSCs isolated from human term placental chorionic villi (CV), umbilical cord (UC), adult bone marrow (BM) and adipose (AD). We found that CD106 (VCAM-1) was expressed highest on the CV-MSCs, moderately on BM-MSCs, lightly on UC-MSCs and absent on AD-MSCs. CV-MSCs also showed unique immune-associated gene expression and immunomodulation. We thus separated CD106+cells and CD106−cells from CV-MSCs and compared their biological activities. Both two subpopulations were capable of osteogenic and adipogenic differentiation while CD106+CV-MSCs were more effective to modulate T helper subsets but possessed decreased colony formation capacity. In addition, CD106+CV-MSCs expressed more cytokines than CD106−CV-MSCs. These data demonstrate that CD106 identifies a subpopulation of CV-MSCs with unique immunoregulatory activity and reveal a previously unrecognized mechanism underlying immunomodulation of MSCs.


Introduction
In recent years, mesenchymal stem cells have attracted significant attention from basic and clinical investigators for their usefulness in the treatment of immune disorders, such as graftversus-host disease (GVHD) and autoimmune diseases [1,2]. MSCs constitute a specialized tissue microenvironment or niche, where they undergo self-renewal and multi-lineage differentiation [3]. MSCs were first isolated from bone marrow [4], and subsequently found in nearly every tissue attempted so far, including adipose [5], and several birth associated perinatal tissues including placenta [6], umbilical cord [7] and cord blood [8].
A number of studies have shown that MSCs derived from different tissues share many similarities but also exhibit some differences. Using identical culture conditions, major differences were observable in the frequencies, proliferation and differentiation potentials as well as biological functions [7,9]. Placental and umbilical cord derived MSCs are more primitive because they share more common genes with embryonic stem cells (ESCs) [10]. UC-MSCs express genes enriched in vascular endothelial growth factor and PI3K-NFkB canonical pathways, whereas BM-MSCs express genes involved in antigen presentation and chemokine/ cytokine pathways. UC-MSCs thus constitute a promising option for angiogenesis, whereas BM-MSCs for bone remodeling [11]. Furthermore, it has been found that the migration of placental and BM-MSCs was found to be 5.9 and 3.2 folds higher than that of UC-MSCs, respectively [12]. BM-MSCs show a stronger stimulating effect on megakaryocyte progenitor expansion than those from UC-MSCs [13] and display a better chondrogenic differentiation compared with other sources of MSCs [14]. In addition, the cytokine profiling is different based on the sources of MSCs [15]. MSCs from different tissues display different potential in proliferation, differentiation, immunomodulation and hematopoiesis supportive abilities, and these differences indicate a demand for effective preparation protocols tailored to each type of MSCs against different diseases [16].
One of the important biological functions of MSCs is the immunomodulation. MSCs can alter the function of T cells, B cells, dendritic cells and NK cells, and thereby exhibit potent immunosuppressive activity [17]. Generally, MSCs display their immunomodulatory activities through direct cell-cell contact and/ or secretion of soluble factors [17]. Cell -cell adhesion mediated by CD106 is known to be critical for T cell activation and leukocyte recruitment to the inflammation site and, therefore, plays an important role in evoking effective immune responses. CD106 is also reported as one of components of neural stem cells niche [18]. Moreover, CD106 is critical for MSCs-mediated immunosuppression [19] and for the binding of hematopoietic progenitor cells [20].
In the present study, we report a positive correlation between CD106 expression and immunosuppressive effect of CV-MSCs. We also show that TNF-a and IL-1b are required for CD106 + MSCs expansion. Our data suggest that CD106 could be used as a biomarker for a subpopulation of CV-MSCs with unique immunosuppressive activity.

Generation of human BM-MSCs, AD-MSCs, UC-MSCs and CV-MSCs
This study was approved by the Institutional Review Board of Chinese Academy of Medical Sciences and Peking Union Medical College. Bone marrow, adipose tissues, term placenta, umbilical cords, peripheral blood and cord blood were obtained from donors with written informed consent.
The isolation and expansion of MSCs were performed as described previously [7]. The basic culture medium for isolation of MSCs was the complete DF-12 medium (Gibco) containing 10% fetal calf serum (FCS) (HyClone), 2 mM glutamine, 100U/ml penicillin-streptomycin, and 10 ng/ml epidermal growth factor (EGF; Peprotech). Samples from 3 donors each were used to generate human BM-MSCs, AD-MSCs, UC-MSCs and CV-MSCs.

Separation of CD106 + CV-MSCs and CD106 2 CV-MSCs
Subpopulations of MSCs were separated using EasySepH PE Positive Selection Kit (StemCell Technologies). MSCs were detached by 0.025% trypsin and labeled with PE-anti CD106 antibody, and CD106 + and CD106 2 cells were separated according to manufacturer's instructions. In some cases, flow sorting was used. MSCs were labeled with PE-anti CD106 antibody, the positive and negative cells were separated using BD Influx (Becton Dickinson). For both of the methods, the purity of isolated cells was more than 90%.

Osteogenic and adipogenic differentiation
Osteogenic and adipogenic differentiation was carried out as described previously [7]. Briefly, cells were plated in 24-well plates at a density of 3000 cells/cm 2 . The medium was changed with  specific induction medium 24 hours later. For osteogenic induction, STEMPRO Osteogenesis Differentiation Kit (GIBCO) was used. For adipogenic induction, medium consisted of DMEM supplemented with 10% FBS, 1 mmol/L dexamethasone, 5 mg/ mL insulin, 0.5 mmol/L isobutylmethylxanthine (IBMX), and 60 mmol/L indomethacin was used. Reagents for adipogenic induction were purchased from Sigma. After 3 weeks of induction, the cells were stained using alizarin red S or oil red O solution.

Colony forming unit-fibroblast assay
For fibroblastic-like colony formation assay, 3 cells/cm 2 of CD106 + or CD106 2 CV-MSCs were seeded in T75 culture flask. The cell media were changed every 7 days. The culture was ended at day 14, and the adherent cells were washed twice with phosphate-buffered saline, and stained with 0.2% crystal violet in 80% methanol for 20 min at room temperature. Then the cells were washed again and the colonies were counted.

Isolation of human peripheral blood mononuclear cells (hPBMCs) and human cord blood mononuclear cells (hCBMCs) CD4 + T cells
Human PBMCs and CBMCs were isolated by Ficoll-Paque (Axis-Shield) density gradient centrifugation from blood of health volunteer donors. CD4 + T cells were purified using relevant magnetic MicroBead kits (Miltenyi Biotec) according to the manufacturer's instructions. The purity of CD4 + T cells was more than 95%.
Affymetric Genechip assay 2610 6 of CD106 + and CD106 2 CV-MSCs was collected and resuspended in Trizol (Invitrogen). The samples were stored in 280uC for genechip assay. The Microarray was done by CapitalBio Corporation. Data was analyzed using mas3.0 molecule annotation system.  Table 1.

RNA isolation, reverse transcription and real time PCR
In vitro stimulation of MSCs 10 5 cells/ml MSCs were seeded in to a 6 well plate in 2 ml for 24 hours, and then treated with IFN-c (30 ng/ml), IL-1b (10 ng/ ml) or TNF-a (30 ng/ml) for 24 hours. All of the pro-inflammatory cytokines were purchased from peprotech. ELISA Cell-free supernatants were collected and kept in refrigerator at 280uC. PGE 2 ELISA kit was purchased from Cayman Chemicals. IFN-c was tested using the kit from ebioscience. All of the ELISA assay kits were used following the supplier's instruction.

Statistical analysis
The data were analyzed for statistic significance using the GraphPad Prism software. Data are presented as mean 6 SEM. Student's unpaired t-test and ANOVA with bonferroni post hoc test were used to determine significance, p,0.05 was considered to be statistically significant.

Placental chorionic villi MSCs highly express CD106
To determine most significant differences among MSCs of different origins, we first compared the phenotypes of MSCs isolated from human term placental CV, UC, adult BM and AD using identical methods. Each type of MSCs was tested for 3    Figure 1A, B).

Gene expression pattern of placental chorionic villi MSCs is unique
To determine if different origins of MSCs affect Th1 specific cytokine IFN-c expression, we then compared levels of IFN-c using ELISA. PBMCs secreted IFN-c upon stimulation of PHA (10 mg/ml). All of these MSCs suppressed the expression of IFN-c in hPBMCs in the presence of PHA (Figure 2A). To exclude the possibility that the differences among MSCs were due to individual differences, we tested the UC-MSCs and CV-MSCs obtained from the same fetus origin. Interestingly, CV-MSCs (19.9960.8328% compared with PHA-actived hPBMCs) showed a significantly increased inhibitory effect on IFN-c production of the hPBMCs compared to UC-MSCs (48.7467.323% compared with PHAactived hPBMCs) ( Figure 2B), indicating that placenta CV contain a valuable source of MSCs with increased immunomodulatory activity.
To examine the mRNA expression of several immune factors in the four sources of MSCs, realtime PCR was performed. It is interesting to note that CV-MSCs expressed highest level of mRNA for COX-2 (Cyclooxygenase-2, inducible synthase for PGE 2 ), IL-1a, IL-1b, IL-6 and IL-8. However, the expression of TGF-b1 was similar in each type of MSCs ( Figure 2C, D). The expression of IDO1 mRNA was very low in all sources unless treated by pro-inflammatory cytokines like IFN-c. Importantly, CV-MSCs showed strongest ability to secrete PGE 2 among the four sources of MSCs (44.0068.814 ng/ml) ( Figure 2E). CD106 + CV-MSCs differ from CD106 2 CV-MSCs in their colony formation and immunosuppressive capacities CV-MSCs possess the strong immunomodulatory potential and express highly CD106. We thus hypothesized that CD106 + CV-MSCs maybe a novel subpopulation specialized in immunosuppression. To test this hypothesis, the CD106 + cells and CD106 2 cells were therefore separated from CV-MSCs, either by MACS or SORTING. Both of CD106 + cells and CD106 2 cells showed typical MSCs morphology ( Figure 3A) and osteogenic and adipogenic differentiation abilities ( Figure 3B), conforming their identical multipotent stem cell potential. Interestingly, CD106 + CV-MSCs possessed a five-fold decreased ability for colony formation than CD106 2 CV-MSCs ( Figure 3C), suggesting that the CD106 may appear on the MSCs at late developmental stage.
We then compared the ability of CD106 + CV-MSCs and CD106 2 CV-MSCs to inhibit the Th1 response by testing the   Figure 4A). In addition, CD106 + CV-MSCs suppress TNF-a production to a lower level than CD106 2 CV-MSCs ( Figure 4A). Furthermore, determination of intracellular cytokine contents on the CD4 + T cells of the co-cultures showed that both types of CV-MSCs decreased at different degree the expression of IFN-c and TNF-a, although CD106 + CV-MSCs were more effective ( Figure 4B and CD106 + CV-MSCs express higher levels of immunoregulatory factors than CD106 2 CV-MSCs To explore the mechanism underlying increased immunosuppressive function of CD106 + CV-MSCs, we performed microarray to analyze the gene expression pattern of CD106 + CV-MSCs and CD106 2 CV-MSCs, and real-time PCR was performed to confirm the results from microarray analysis. The expression of adherence molecules and cytokines were different in CD106 + and CD106 2 CV-MSCs, indicating different functions for two kinds of MSCs. Notably, the gene expression profiles associated with immunoregulatory factors were different, indicating different functions of these two subpopulations (Table 3 and 4). CD106 + CV-MSCs express higher levels of COX-2, IL-1a, IL-1b, IL-6 and IL-8 compared with CD106 2 CV-MSCs ( Figure 5A and B). CD106 + CV-MSCs showed an increased ability to secrete PGE 2 compared to CD106 2 CV-MSCs ( Figure 5C). To test the ability for another important immune modulator IDO1, IFN-c was used to stimulate the MSCs. IFN-c significantly up regulated the expression of IDO1 in 24 hours. CD106 + cells express higher level of IDO1 than CD106 2 cells in the presence of IFN-c ( Figure 5D). Together, CD106 + CV-MSCs expressed higher level of immunomodulating factors compared with CD106 2 CV-MSCs.

CD106 expression on CV-MSCs is controllable in response to propagation and cytokine induction
To determine if the CD106 expression on CV-MSCs is tissue specific and stable during long-term cultivation, we cultured CV-MSCs under different condition. It was interesting to observe that the expression of CD106 decreased after passaging in basic culture media ( Figure 6A). Accordingly, there was a reduction of mRNA expression of COX-2, IL-1a, IL-1b, IL-6 and IL-8 ( Figure 6B) and a reduction of PGE 2 production ( Figure 6C).
We then wanted to know if CD106 2 CV-MSCs could be induced to express CD106 and thereby increase their immunosuppressive ability. CD106 2 CV-MSCs were treated with IFN-c, TNF-a or IL-1b for 24 hours. The addition of pro-inflammatory cytokines resulted in an increased expression of COX-2, IL-1a, IL-1b, IL-6 and IL-8 in CD106 2 CV-MSCs (Figure 7). IFN-c apparently up regulated CD106 and IL-6 expression, but failed in other genes. Together, these results indicate that the expression of CD106 on MSCs is variable, and suggest that TNF-a, IL-1b, but not IFN-c, may be important for generation of CD106 + MSCs.

Discussion
In this study, using CD106 as a surface marker, we have successfully demonstrated a previously uncharacterized heterogeneity of MSCs isolated from chorionic villi. CD106 + CV-MSCs possess high immunomodulation activity, whereas CD106 2 CV-MSCs possess high colony formation capacity. In addition, we provide several lines of evidence that the immunosuppressive activity of MSCs is highly related to their ability for the secretion of immunoregulatory cytokines and PGE 2 . In CD106 + cells, a list of genes was up regulated, including COX-2, IL-1a, IL-1b, IL-6 and IL-8. The biological activity of CD106 + CV-MSCs was highly related to the stimulation of pro-inflammatory cytokines TNFa and IL-1b.
To our knowledge, CD106 is a cytokine-inducible cell surface protein capable of mediating adhesion to leukocytes expressing alpha 4 integrins. CD106-deficient mouse embryos were not viable and exhibited either of two distinct phenotypes. Half of the embryos died before embryonic day 11.5 and exhibited severe defects in placental development. The remaining CD106-deficient embryos survived to embryonic day 11.5-12.5 and displayed several abnormalities in their hearts development [21]. The role of CD106 in embryonic development is very important for the formation of the umbilical cord and placenta [22].
Human placenta is an important organ to provide nutrition, gas exchange, waste removal, endocrine and immune support during fetus development. The chorionic villi constitute a part of the border between maternal and fetal blood providing exchange between maternal and fetal circulation. Recently, the placenta has been recognized as an immune organ, creating a fetomaternal immune-privileged environment enabling the survival of the semiallogeneic fetus [23,24]. The underlying mechanism for the placental immune tolerance has been found to depend on regulatory T cells [25], negative signals from the PD1-PDL1 costimulatory pathway [26], IL-17-producing T cells (Th17) [27] and FAS-Ligand [28]. Th1 immune response should be suppressed, in order to maintain the fetomaternal tolerance. Wegmann and colleagues first developed the concept that there is a shift from a T helper 1 response to a T helper 2 bias during pregnancy that functionally induces maternal tolerance and suppression [29]. Administration of the Th1 interleukin IFN-c [30] leads to fetal loss and preterm labor in the mouse.
COX-2 expression in fetal membranes seems to be important for pregnancy, as multiple female reproductive failures were observed in COX-2 deficient mice [40]. COX-2 highly expressing CD106 + cells may have more functions but not only immune modulation. IL-1 is also critical for pregnancy. IL-1Ra prevents embryonic implantation by directly affecting the endometrial epithelium [41]. In addition, human placenta is a potent hematopoietic niche [42], the effect of CD106 + CV-MSCs on haemopoiesis and angiogenesis may be an interesting field for further study.
Pro-inflammatory cytokines like IFN-c, IL-1b and TNFa secreted by immune cells are critical to induce immunomodulatory ability of MSCs [43][44][45][46]. MSCs stimulated by proinflammatory cytokines were shown to be more effective to modulate the immune response in vitro [47] and in vivo [44]. CD106 + CV-MSCs showed similar gene expressions with CD106 2 CV-MSCs after stimulating with IL-1b or TNF-a, but not IFN-c. Interestingly, IL-1b and TNF-a have been shown to be important for reproductive [48,49], while IFN-c is low at the maternal-fetal interface [29]. To maintain or induce CD106 + CV-MSCs, stimulus in the microenvironments may be needed, and IL-1b or TNF-a might be candidate stimulus. Certainly other cytokine and cell-cell contact may also participate in the development of CV-MSCs.
In summary, this study reports for the first time that CD106 identifies a unique subpopulation of MSCs with powerful immunosuppressive activity, CD106 + MSCs mainly reside in human term placenta, which may contribute to fetomatenal tolerance. Our future work will be best directed toward designing therapeutic strategies for the treatment and prevention GVHD or other autoimmnune disorders as well as early pregnancy loss, preterm labour and transplant associated adverse outcomes.