dlk1/FA1 Regulates the Function of Human Bone Marrow Mesenchymal Stem Cells by Modulating Gene Expression of Pro-inflammatory Cytokines and Immune Response-related Factors*

dlk1/FA1 (delta-like 1/fetal antigen-1) is a member of the epidermal growth factor-like homeotic protein family whose expression is known to modulate the differentiation signals of mesenchymal and hematopoietic stem cells in bone marrow. We have demonstrated previously that Dlk1 can maintain the human bone marrow mesenchymal stem cells (hMSC) in an undifferentiated state. To identify the molecular mechanisms underlying these effects, we compared the basal gene expression pattern in Dlk1-overexpressing hMSC cells (hMSC-dlk1) versus control hMSC (negative for Dlk1 expression) by using Affymetrix HG-U133A microarrays. In response to Dlk1 expression, 128 genes were significantly up-regulated (with >2-fold; p < 0.001), and 24% of these genes were annotated as immune response-related factors, including pro-inflammatory cytokines, in addition to factors involved in the complement system, apoptosis, and cell adhesion. Also, addition of purified FA1 to hMSC up-regulated the same factors in a dose-dependent manner. As biological consequences of up-regulating these immune response-related factors, we showed that the inhibitory effects of dlk1 on osteoblast and adipocyte differentiation of hMSC are associated with Dlk1-induced cytokine expression. Furthermore, Dlk1 promoted B cell proliferation, synergized the immune response effects of the bacterial endotoxin lipopolysaccharide on hMSC, and led to marked transactivation of the NF-κB. Our data suggest a new role for Dlk1 in regulating the multiple biological functions of hMSC by influencing the composition of their microenvironment “niche.” Our findings also demonstrate a role for Dlk1 in mediating the immune response.

Human bone marrow-derived mesenchymal stem cells (hMSC) 2 are a group of clonogenic cells present among the bone marrow stroma and capable of multilineage differentiation into mesoderm-type cells such as osteoblast, adipocyte, and chondrocyte (1) and possibly other non-mesoderm type cells (2). Moreover, hMSC provide supportive stroma for growth and differentiation of hematopoietic stem cells (HSC) and hematopoiesis (3). Understanding the mechanisms that control the differentiation decisions of hMSC is thus of the utmost importance from a basic bone biology point of view. It is also important for the clinical use of the hMSC in transplantation and regenerative medicine protocols (1). dlk1/Pref-1 (delta-like 1/pre-adipocyte factor-1) is a transmembrane protein of the EGF-like homeotic superfamily. It is expressed from an imprinted gene paternally expressed at 14q32. Its extracellular domain contains six cysteine-rich EGFlike repeats similar to those found in the Delta/Notch/Serrate family of signaling molecules (4,5). Dlk1 plays a critical role in modulating cell fate decisions throughout development (6). This is illustrated by the presence of high prenatal mortality, growth retardation, obesity, skeletal malformations, and abnormalities of hematopoiesis in mice deficient in Dlk1 (7,8). Also, in the human syndrome of maternal uniparental disomy 14 (where Dlk1 is silent), patients exhibit obesity, hypotonia, premature puberty, macrocephaly, short stature, and small hands (9). Dlk1 has been known for several years as a negative regulator of adipocyte differentiation (10). Recent data suggest that it is involved in many differentiation processes, including hematopoiesis (11), pancreatic islet cell differentiation (12), Schwann cell differentiation (13), and hepatic cell differentiation (14). Recently, we have identified Dlk1 as a novel regulator of hMSC differentiation (15). Cellular overexpression of Dlk1 or adding it as a soluble protein to hMSC led to inhibition of hMSC differentiation to osteoblasts or adipocytes (15). The extracellular domain of the dlk1 is cleaved by tumor necrosis factor-␣-converting enzyme (16) and was first identified in the amniotic fluid (17) and hence named fetal antigen 1 (FA1) (18). FA1 is the biologically active part of the molecule, and its serum levels change in some pathological conditions, e.g. growth hormone excess (acromegaly) (19). In contrast to other members of Delta/Notch/Serrate family, dlk1 protein does not have the DSL (Delta, Serrate, and LAG2) domain mediating Notch-Delta interaction, and thus it is not known whether dlk1 functions as a ligand or as a receptor (6). However, recent data based on a yeast two-hybrid system suggested that dlk1 interacts with Notch 1 and inhibits Notch signaling (20). Also, other signaling pathways have been demonstrated to be affected by the expression of Dlk1, e.g. mitogen-activated protein kinase (MAPK) (21) and insulin-like growth factor-I receptor-mediated p42/p44 MAPK activation (22). Despite this pleiotropic function of Dlk1 in many developmental and differentiation processes, little is known about the mechanisms mediating its regulatory function and its target genes.
To identify the mechanisms underlying the biological effects of Dlk1 in hMSC, we employed the DNA microarray approach to discover specific molecular pathways regulated by Dlk1. We found that expression of Dlk1 increased the production of a large number of inflammatory and immune response-related factors, including cytokines, chemokines, and complement factors by hMSC, and these changes were associated with its effects on hMSC differentiation into adipocytes, osteoblasts, and its hematopoiesis-support function.

EXPERIMENTAL PROCEDURES
Cell Culture-The establishment and the characterization of hMSC-TERT (immortalized hMSC cells, used as a control with no Dlk1 expression) and hMSC-dlk1 (Dlk1-overexpressing cells, derived from hMSC-TERT) cell lines have been described previously (15,23). Normal hMSC cultures were established from bone marrow aspirates obtained from young donors (n ϭ 5, males 25-30 years old) by aspiration from the iliac crest as described previously (24). A written consent was obtained from each participant, and the study was approved by the local Scientific-Ethical committee. All cells were cultured in a standard growth medium (SGM) containing minimal essential medium (MEM; Invitrogen) supplemented with 10% fetal calf serum (FCS) and 1% penicillin/streptomycin (Invitrogen) at 37°C in a humidified atmosphere containing 5% CO 2 .
Collecting the Conditioned Media and Purification of FA1-hMSC-TERT or hMSC-dlk1 cells were cultured at 3 ϫ 10 3 cells/cm 2 in Petri dishes in standard growth medium. At 80 -90% cell confluence, medium was replaced by serum-free MEM, and cells were cultured for 2 additional days. Conditioned media (CM) were then collected either from hMSC-TERT cells (control/CM) or hMSC-dlk1 (dlk1/CM), centrifuged at 1000 ϫ g for 5 min at 4°C to remove cell debris, and used neat or diluted 1:2 (v/v) or 1:4 (v/v) with serum-free MEM.
The full soluble ectodomain, active form of dlk1 protein named FA1, was purified from the collected serum-free CM of cultured hMSC-dlk1 (dlk1/CM) using immunospecific affinity chromatography as described previously (25). dlk1/CM depleted of FA1 (FA1 ؊ /CM) were also collected from the affinity chromatography eluents.
We were careful to avoid any endotoxin contaminations in our culture system or during FA1 preparation. To ensure that, the endotoxin levels were measured in all conditioned media and FA1 preparation as nondiluted research samples using the Endotoxin-Testing Service at Cambrex BioScience (test code 95-101; Cambrex BioScience Verviers S.p.r.l., Belgium). All conditioned media (control/CM, dlk1/CM, and FA1-/CM) and FA1 preparation used in this study were negative for contamination with endotoxins with less than 0.05 endotoxin unit/ml.
Study of the Effect of FA1-containing CM and Purified FA1 on the Cytokine Expression Profile of hMSC-Normal hMSC cells established from healthy donors were cultured in 6-well plates at 3 ϫ 10 4 cells/cm 2 in standard growth medium. At 90 -100% cell confluence, cells were cultured either in control/CM (obtained from hMSC-TERT), in dlk1/CM supplemented with 10% FCS, or in SGM supplemented with purified FA1 (1 or 5 g/ml) for 24 h.
Cell Differentiation Assays-To induce osteoblast differentiation, cells were cultured in 6-well plates at a density of 3 ϫ 10 4 cells/cm 2 in SGM for 24 h. At 70 -80% confluence, the medium was replaced by osteogenic medium, consisting of an SGM supplemented with 10 nM 1,25-dihydroxycholecalciferol (vitamin D 3 ) (kindly provided by Leo Pharma, Denmark). The medium was replaced every 3 days.
Proliferation Assay of Cultured Human B Cells-Peripheral blood mononuclear cells isolated from buffy coats of healthy blood donors were depleted of CD19ϩ cells using Dynabeads CD19 Pan B (Dynal, Norway) according to the manufacturer's instructions. Cell-attached magnetic beads were removed by magnet after overnight incubation with DETACHBEAD CD19 (Dynal, Norway). Isolated CD19ϩ cells were washed twice at 1,200 rpm for 10 min at room temperature in modified Eagle's medium (without phenol red) supplemented with 200 international units/ml penicillin and counted in trypan blue. Cells were then resuspended at 0.5 ϫ 10 6 /ml in RPMI (as a positive control), serum-free conditioned medium from hMSC-TERT (control/CM), or hMSC-dlk1 (dlk1/CM) supplemented with 10% FCS and stimulated with 20 ng/ml recombinant human IL-4 (R&D Systems, Abingdon, UK) and the indicated concentrations of recombinant human CD40L (R&D Systems) or left untreated (nonstimulated). Cells were seeded in flat-bottom 96-well microtiter plates (50,000 cells per well in 100 l of medium) and incubated at 37°C. At day 3 of culture, 100 l of fresh media supplemented with IL-4 and CD40L were added.
At day 6 of cell culture, metabolic activity of the cells was measured with an XTT assay (Biological Industries, Israel). Data were represented as fold induction of the proliferation of B cells over control CD19ϩ cells cultured in nonstimulated CM.
Microarrays-Both control (hMSC-TERT) and Dlk1-overexpressing cells (hMSC-dlk1) were cultured in triplicate at 3 ϫ 10 4 cells/cm 2 in Petri dishes in standard growth medium. At 90 -100% confluence, highly purified total cellular RNA was isolated from each of three independent cultures per cell line using an RNeasy kit (Qiagen Nordic, West Sussex, UK) according to the manufacturer's instructions. First-and second-strand cDNA syntheses were performed from 8 g of total RNA using the SuperScript Choice System (Invitrogen) according to the manufacturer's instructions. cRNA was synthesized from cDNA and biotinylated using the BioArray High Yield RNA transcript labeling kit (Enzo kit, Enzo Diagnostics, Farmingdale, NY), according to the protocol provided by the manufacturer, and hybridized for 16 -18 h at 45°C in a rotisserie to the Affymetrix Human Genome U133A 2.0 arrays. The arrays were scanned using the GeneArray scanner (Affymetrix).
Data analysis was performed using the dChip software (26). We first normalized our probe level data using the invariant set normalization procedure (26). The normalized probe level data were converted to model-based gene expression indexes (MBEI, log with base 2) for use as the values of gene expression for a total of 22,000 genes on the Affymetrix 133A 2.0 arrays. The paired t statistic was used to assess the genes that are differentially regulated in the two cell lines by assigning the hMSC-TERT cell line as reference and the hMSC-dlk1 cell line as the experiment group. The false discovery rate (27) was calculated for each of the genes to account for multiple testing. Significant genes were identified as those with a mean expression ratio between the experiment and the reference groups above 2 and below 2 and at the same time a false discovery rate of Ͻ0.05. Hierarchical clustering analysis was applied to cluster the genes as well as the samples using the average linkage method. Gene ontology classifications for all differentially expressed genes by Dlk1 in hMSC were performed using DAVID 2.0 software (28) and EASE software (29). Genes were classified according to their molecular function and relevant biological process. The microarray data were deposited in the Array Express public data base with assigned accession number (E-MEXP-560).
Total RNA Extraction and Real Time PCR Analysis-Total cellular RNA was isolated using a single step method with TRIzol (Invitrogen), according to the manufacturer's instructions. First strand complementary cDNA was synthesized from 4 g of total RNA using a revertAid H minus first strand cDNA synthesis kit (Fermentas, Copenhagen, Denmark) according to the manual instructions.
Real time PCR was performed using the iCycler IQ detection system (Bio-Rad) and SYBR Green I as a double strand DNAspecific binding dye. Thermocycling was performed in a final volume of 20 l containing 3 l of cDNA sample (diluted 1:20), 20 pmol of each primer, and 2ϫ iQ TM SYBR Green Supermix (Bio-Rad). The quantification of gene expression for each target gene and reference gene was performed in separate tubes using primers as shown in supplemental Table 1. We used a denatur-ing step at 95°C for 3 min and 40 cycles of 95°C for 30 s, 60°C for 30 s, and 72°C for 1 min. Each reaction was run in duplicate, and fluorescence data were collected at the end of the extension step in every cycle.
To ensure specific amplification, a melting curve was calculated for each PCR by increasing the temperature from 60 to 95°C with a temperature increment rate of 0.5°C/10 s. Fold induction and expression level for each target gene were calculated using the comparative C T method as follows: 1/(2 ⌬CT ), where ⌬C T is the difference between C T target and C T reference. After normalization to ␤-actin mRNA (User Bulletin No. 2, PerkinElmer Life Sciences), data were analyzed using optical system software version 3.1 (Bio-Rad) and Microsoft Excel 2000 to generate relative expression values.
In Situ Detection of NF-B Nuclear Translocation-Control cells (hMSC-TERT) were seeded at 3,000 cells/well in 96-well plates and cultured for 2 days in normal medium. Cells were exposed for 15, 30, 60, and 120 min at 37°C to nondiluted CM prepared either from control or hMSC-dlk1 cells. As a positive control for NF-B activation, some wells were treated with 1 g/ml LPS. In situ detection of NF-B nuclear translocation was carried out as described previously (30). Briefly, cells were fixed first in 4% paraformaldehyde for 10 min at room temperature and then with cold methanol for 15 min at Ϫ20°C. Cells were incubated with a mouse monoclonal anti-p65 antibody (Santa Cruz Biotechnology) in bovine serum albumin/PBS overnight at 4°C, washed several times in PBS, and subsequently incubated with a secondary AlexaFluor 568 goat antimouse IgG antibody (Invitrogen) in bovine serum albumin/PBS for 1 h at room temperature. After several washes, the nucleus of cells was counterstained with Hoechst 33258 (Sigma). Pictures of Hoechst and anti-p65 immunofluorescence were taken randomly in different areas of the wells with a coolsnap camera (Roper Scientific, Brock & Michelsen, Birkerød, Denmark) plugged to a stage-motorized inverted Axiovert 200 microscope (Zeiss, Brock & Michelsen). Cells showing a bright immunofluorescence for p65 in the nucleus were scored using MetaVue image analyzing software (Universal Imaging Corp., Brock & Michelsen). The results were presented as the number of cells with NF-B nuclear translocation in percentage of the total number of nuclei scrutinized.
Statistical Analysis-All values are expressed as means Ϯ S.D. Statistical analysis was performed by Student's t test using unpaired t test (two-tailed). p Ͻ 0.05 was considered significant.

RESULTS
To identify the genetic pathways regulated by the expression of Dlk1, we studied the basal gene expression pattern in Dlk1-overexpressing hMSC cells versus control hMSC-TERT cells using Affymetrix DNA microarrays. A total of 315 probe sets corresponding to 277 genes/expressed sequence tags were found to be differentially expressed in response to Dlk1 expression. Among these, 157 probes corresponding to 128 genes were up-regulated (Ͼ2-fold, p Ͻ 0.001), whereas158 probe sets corresponding to 149 genes were down-regulated (Ͻ2-fold, p Ͻ 0.001) in hMSC-dlk1 compared with control cells. Gene annotation based on biological processes and molecular functions revealed the pres-ence of 24% immune response-related genes, whereas the other 76% were assigned to different functional categories, such as signal transduction, cell adhesion, metabolism, apoptosis, and others ( Fig. 1, A and B). On the other hand, Dlk1 down-regulated genes were mainly annotated as DNA repli-cation-related genes and cell growth and/or maintenance genes (see the supplemental Table 2).
As shown in Table 1, the largest category of the up-regulated genes was annotated as immune regulatory factors, which include several potent pro-inflammatory cytokines, a A, hierarchical clustering analysis of up-regulated genes associated with Dlk1 expression in hMSC. A total of 158 probe sets corresponding to 128 genes was found to be significantly changed Ͼ2.0-fold by hMSC-dlk1 cells over hMSC under basal culture conditions. Microarray was performed in triplicate for each cell line. B, categorization of Dlk1-upregulated genes according to gene ontology. 128 genes were annotated using DAVID 2.0 software and by means of EASE software based on their biological process. C, validation of microarray data. Real time PCR was performed on RNA isolated from hMSC-TERT (control) and hMSC-dlk1 cells grown under basal culture conditions during their proliferation at 70% confluence (after 3 days of culture, opened bars) or 100% confluence (after 5 days of culture, closed bars). The expression level of each target gene was normalized for ␤-actin expression and represented as fold induction over control cells. Data are shown as means Ϯ S.D. of three independent experiments.

Dlk1 Regulates the Cytokines Expression of hMSC
large number of chemokines, the complement components C3 and C1 that are known as key factors for the activation of the complement pathway (31), prostaglandin E synthase, and prostaglandin-endoperoxide synthase 2 (COX-2) that are up-regulated in response to IL-1␤ (32) and KLRC1, a molecule expressed primarily by Natural Killer cells during infection and involved in the recognition of major histocompatibility complex class I HLA-E molecules (33). In addition, the apoptosis-annotated gene group included numerous genes such as CASP1, CASP8, TNFRSF9, TNFRSF21, and TNFAIP3 that function cooperatively with some cytokines to induce cell death and activate NF-B signaling (34). Given the gene expression pattern found, we decided to focus on studying Dlk1-up-regulated genes, especially the pro-inflammatory cytokines.
Validation of the Microarray Data-Real time PCR was performed to verify the data obtained by microarray analysis. We mainly compared the expression profile of 18 distinct cytokines-chemokines and some other inflammatory-related genes between control hMSC and hMSC-dlk1 cells under basal culture conditions during their proliferation at 70 and 100% confluence. As shown in Fig. 1C, the real time PCR analysis confirmed the increased gene expression of all 18 selected genes by Dlk1-overexpressing cells and the difference between the groups increased with time in culture. In addition, the magnitude of the difference in gene expression between groups was in many cases higher than that determined by microarray analysis. These discrepancies in fold differences are in agreement with data published by others, reporting that the dynamic range of the reverse transcription-PCR technique is generally greater than that of microarray hybridizations (35). Also, we confirmed the results obtained by the microarray analysis for another 15 randomly selected up-regulated genes (data not shown).
Dose-dependent Stimulation of Cytokine Gene Expression in hMSC by either Purified FA1 or CM from hMSC-dlk1 Cells-To confirm the results obtained from hMSC-TERT cells, we investigated the direct effects of dlk1 on the expression of inflammatory response-related genes in normal hMSC cells obtained from healthy donors. Serum-free CM collected from hMSC-dlk1 cells (FA1 concentration of 900 Ϯ 20 ng/ml; n ϭ 4) and control hMSC cells (containing no detectable FA1) were employed as such or diluted to 1 ⁄ 2 or 1 ⁄ 4 with standard medium and added to normal hMSC. We found that CM from Dlk1expressing cells markedly induced in a dose-dependent manner the same group of cytokines that were up-regulated in hMSC-dlk1 cells ( Fig. 2A).
To determine whether the CM-induced cytokine gene expression in hMSC was because of the direct effect of dlk1/ FA1 present in the CM, we examined the effect of two different concentrations of purified FA1 protein (1 or 5 g/ml) on normal hMSC. As shown in Fig. 2B, FA1 significantly induced the same cytokine gene expression profile as that triggered by whole dlk1-containing CM. These data confirm the direct effects of dlk1 in the up-regulation of proinflammatory cytokines production by hMSC.
The Inhibitory Effect of Dlk1 on the Differentiation of hMSC into Osteoblasts and Adipocytes Is Associated with a Mecha-nism Involving Dlk1-induced Cytokine Gene Expression-To examine whether the observed changes induced by Dlk1 on the immune response-related factors mediate its effects on the biological functions of hMSC, we studied the regulation of cytokine-chemokine expression profile during hMSC cell differentiation.
As shown in Fig. 3C, the expression patterns of cytokine genes induced by Dlk1 differed between hMSC induced to differentiate to osteoblasts or to adipocytes, despite IL-1␣, IL-1␤, CCL20, and KLRC2 being commonly up-regulated by Dlk1 in both differentiation processes.
To further investigate the possible biological role of dlk1induced cytokine production, we examined the effect of hMSC-dlk1 conditioned medium depleted of FA1 (FA1 ؊ /CM) on hMSC differentiation. As shown by Fig. 4, A and B, and similar to our previous studies (15), incubation of FA1 alone or CM obtained from dlk1-overexpressing cells (hMSC-dlk1) (containing FA1) inhibited hMSC differentiation into osteoblasts and adipocytes as assessed by real time PCR analysis (Fig. 4, A  and B). Interestingly, FA1-depleted CM (FA1 ؊ /CM) is shown to exhibit similar inhibitory effects in a dose-dependent manner (Fig. 4, B and C), suggesting that the presence of dlk1/FA1 in hMSC cells induces the production of several cytokines in their pericellular microenviroment (CM) that mediate these biological effects.
Regulation of Cytokine Gene Expression by Dlk1 in Response of hMSC to LPS-In addition, to study the effects of Dlk1 on cytokine production by hMSC, we investigated its effects on the response of hMSC to LPS. Because cellular treatment by LPS can induce the production of a large number of immune system-related factors and cytokines (38,39), we examined the modulatory effects of Dlk1 on this response. As shown in

Dlk1 Regulates the Cytokines Expression of hMSC
expression by hMSC for providing a supportive niche for HSC in bone marrow, we examined the effect of dlk1/CM versus control/CM on the B lymphocyte proliferation. Isolated B cells (CD19ϩ cells) from peripheral blood mononuclear cells were cultured either in CM from hMSC-dlk1 cells or hMSC-TERT (control), and the cell proliferation was measured calorimetrically by using an XTT-based assay. As shown in Fig. 5B, the FA1-containing CM markedly stimulated the proliferation of B lymphocytes in response to two different concentrations of CD40 ligand (2.5 and 10 g/ml) by 300 -350% over control CM. These results demonstrated the role of Dlk1-induced cytokine expression mechanisms in hematopoiesis by mediating the B cell proliferation.  MARCH 9, 2007 • VOLUME 282 • NUMBER 10

JOURNAL OF BIOLOGICAL CHEMISTRY 7345
Enhanced Cytokine Production by dlk1 Is Associated with Activation of NF-B Pathway-Many cell responses leading to inflammation are mediated through the induction and activa-tion of the transcription factor NF-B, which upon cell activation translocates into the nucleus and induces the transcription of several inflammatory cytokines (40). To investigate whether the expression of Dlk1 could affect the activation of NF-B, hMSC cells were treated with nondiluted CM from hMSC-TERT cells, CM from hMSC-dlk1 cells, or with LPS (1 g/ml), as a positive control. One hour later, the nuclear translocation of NF-B was assessed by immunofluorescence staining. Interestingly, cells treated with dlk1-containing CM displayed marked increase in the numbers of stained nuclei compared with cells treated either with control CM or LPS. These results demonstrate that dlk1 possesses the ability to significantly activate NF-B nuclear translocation in hMSC (Fig. 6).

DISCUSSION
In this study we demonstrate that one of the main effects of Dlk1 expression in hMSC is to enhance their expression of several pro-inflammatory cytokines and other immune response-related factors. In support of our findings, several studies showed that the expression of Dlk1 was associated with cellular processes characterized by the presence of an inflammatory response. For example, Dlk1 was found to be expressed at high levels by Murphy-Roths Lymphoma mouse blastema cells during ear wound healing (41), by transitamplified ductular (oval) cells during liver regeneration in liver injury model in mice (42), and by satellite cells during muscle regeneration (43). Furthermore, Dlk1 was highly expressed at an early stage of the progressive liver fibrosis in biliary atresia, a disease characterized by inflammation in the hepatobiliary system (44).

Dlk1 Regulates the Cytokines Expression of hMSC
These cytokines are known to be present in the bone marrow microenvironment and to exert a variety of biological roles, including regulation of immune cells functions, inflammation, and several metabolic functions (45). In addition, interleukins have been demonstrated to play a role in the control of differentiation and functions of bone cells (46) (see below). The expression of several chemokines (chemotactic cytokines) was also increased in hMSC-expressing dlk1. Chemokines are important secondary inflammatory mediators released by leukocytes and immune cells that control neutrophil recruitment and inflammation in vivo (47). They are also produced by bone marrow stromal cells and can support the proliferation and differentiation of hematopoietic cells (48,49). We identified several chemokine ligands that were up-regulated in response to Dlk1, including six CXC-and one CC-chemokine family genes as follows: CXCL1 (growth-related oncogene-␣/keratinocyte; KC), CXCL2 macrophage inflammatory protein-2/ (GRO␤), CXCL3 (GRO␥), CXCL6 (granulocyte chemotactic protein 2), CXCL8 (IL-8), CXCL11, and CCL20 (MIP-3␣).
Several apoptosis-related genes induced by inflammatory mediators were found to be increased in hMSC in presence of Dlk1. These apoptotic genes included caspase 1 (CASP1, IL1B-convertase), an activator of IL-1␤ (50) and which is involved in inflammation and wound healing (51), and caspase 8 (CASP8, FADD-like apoptosis regulator), which mediates death factor-induced apoptosis (52), and in addition, three members of the tumor necrosis factor family, including TNFAIP3 (53), TNFRSF9 (ILA), a regulator of T-lymphocyte proliferation and survival (54), and TNFRSF21 (death receptor 6, DR6), an activator of the cell death components via the activation of the CASP-8 pathway (55). Two inhibitors of apoptosis were also up-regulated, baculoviral inhibitor of apoptosis repeat-containing protein 3 (BIRC3) (56) and BCL2-related protein A1 (BCL2A1). These data suggest that dlk1 may modulate the apoptotic responses and corroborate the recent findings that dlk1 may function as a tumor suppressor gene with the ability to suppress tumor growth and increase anchorage-dependent cell death of cancer cells (57). Interestingly, overexpression of Dlk1 in hMSC did not affect their cell growth (15) suggesting that the dlk1 effects on cell growth may be cell type-dependent.
The increased inflammation-related cytokine expression profile in hMSC expressing dlk1 was also associated with the up-regulation of gene expression of several cell adhesion molecules. The production of several of these molecules is known to increase during the inflammatory response (58). Among these molecules are the

TABLE 2 Quantitative real time PCR analysis of Dlk1-up-regulated cytokines during osteoblast and adipocyte differentiation of hMSC
Cells were stimulated to differentiate into osteoblasts or adipocytes as described under "Experimental Procedures" (same experiment of Fig. 3). Data were represented as fold induction of up-regulated genes by Dlk1-overexpressing cells over control hMSC cells that differentiated under the same conditions. NC indicates not changed.

Gene
Adipocyte  (59). In addition to their role in inflammation, cell adhesion molecules are known regulators of cellcell and cell-matrix interactions that are important for multiple cellular functions of stem cells, including proliferation, differentiation, and migration (60). In bone, cadherins have been reported to mediate cell-cell interaction through adherens junctions that are important for both bone development and osteoblastic functions (61). Our experiments demonstrated that the inhibition of differentiation of hMSC into adipocytes and osteoblasts by dlk1 was associated with changes of a large number of cytokines and immune-related factors. The Dlk1 inhibitory effects on adipogenesis were associated with increased gene expression of IL-1␣, IL-1␤, IL-6, IL-8, CCL20, and COX-2. Some of these cytokines, e.g. IL-1 and IL-6, have been reported to inhibit adipogenesis of cultured MSC or pre-adipocytes (62). In addition, activation of TAK1-NIK/NF-B signaling by IL-1 in combination with tumor necrosis factor-␣ was shown to suppress transactivation of peroxisome proliferator-activated receptor ␥2 and inhibit adipogenesis of MSC (63). Similarly, we found that Dlk1 upregulated a wide range of cytokine and chemokine genes in hMSC during osteoblast differentiation, including IL-1␣, IL-1␤, IL-8, CXCL6, CCL20, and C3. Some of these cytokines, e.g. IL-1␣, IL-1␤, and IL-6, have been shown by others to inhibit osteoblast cell differentiation and to potentiate bone resorption (64) and are associated with osteoporotic bone loss in a variety of diseases, e.g. rheumatoid arthritis (65,66). The possible role of dlk1 in these diseases remains to be determined.
LPS in the presence of Dlk1 exerted synergistic effects on the gene expression of several cytokines and chemokines in hMSC. LPS is a bacterial endotoxin that can induce immune and inflammatory responses via up-regulating COX-2 and production of inflammatory cytokines in a variety of immune cells (67) and nonimmune cells (68). The response of hMSC to LPS suggests a possible important biological role of hMSC as part of the innate immune response against bone infection. LPS is an important component of several bacteria associated with chronic inflammatory processes in bon, e.g. osteomyelitis and periodontitis (69). Previous studies have demonstrated that LPS can inhibit osteoblast differentiation and enhance osteoclastic bone resorption (70). We found that several cytokines were up-regulated in the presence of LPS, including IL-1␣, IL-1␤, and IL-6 with known inflammatory and immune effects. Also, chemokine gene expression in hMSC, e.g. CXCL6 and CCL20, were up-regulated by LPS in the presence of dlk1. Recently, Honczarenko et al. (71) have shown that hMSC cells express certain CC and CXC chemokine receptors as well as their ligands, which have been reported to play a role in controlling the proliferation and differentiation of hMSC (71). The production of chemokines by osteoblastic cells has been demonstrated to play a role in the bone inflammatory response because of infection. Osteoblastic cells can secrete MCP-1, MCP-2, and MCP-3 following activation by bone bacterial pathogens in vitro, and these chemokines can in turn recruit and activate monocytes and selectively stimulate subpopulations of lymphocytes and Natural Killer cells (72). Moreover, osteoblastic cells produce other CXC chemokines (GCP-2, IL-8, GRO-␣, GRO-␥, and IP-10) that may play a role in the recruitment and activation of neutrophils (73). Because dlk1 can be produced in the bone marrow by hMSC (15) and CD34ϩ cells (74), it is possible that it can act to amplify these inflammatory responses during infection.
In addition to its role in adipocyte and osteoblastic differentiation, dlk1 enhanced the hematopoietic support function of hMSC as demonstrated here by enhancing the proliferation of human B cells. Several studies have implicated Dlk1 directly in hematopoiesis, where its expression by stromal cells of the bone marrow has been demonstrated to act as a positive regulator of HSC (11,75) and to mediate stromal-pre-B cell interactions (76). Also, the analysis of hematopoiesis in Dlk1 knock-out mice suggested the contribution of dlk1 in lineage-specific development of megakaryocytes and B cells (8). In addition, Dlk1 was elevated by CD34ϩ cells in patients with low risk myelodysplastic syndrome and acute myeloid leukemia as compared with normal individuals (8,77). One of the main functions of MSC is to provide a hematopoiesis-supporting microenvironment via the expression of many cytokines and regulatory factors (78). Thus, our data suggest that Dlk1 may be an important regulator of this function.
The molecular mechanisms of the biological effects of Dlk1 are not known. Our data suggest that dlk1 mediates its effects on hMSC indirectly and through changes in a large number of cytokines and immune molecules in the pericellular microenvironment. Based on our experiments, it is not possible to determine whether a specific "master" cytokine regulator is induced by dlk1 or that dlk1 affects the gene transcription of various cytokines simultaneously. However, we observed increased activation and translocation of the transcription factor NF-B in hMSC in the presence of Dlk1. Activation of NF-B plays a central role in the regulation of several cellular processes, such as inflammation, immune response, cell differentiation, and apoptosis via regulating the inducible expression of a wide variety of cytokines and chemokines (79). Interestingly, 60% of immune response-annotated Dlk1-induced genes are known targets for NF-B activation, including those encoding for cytokines (IL-1␣, IL-1␤, IL-6, and LIF), chemokines (IL-8, CXCL2, -3, -6, -11, and CCL20), complement 3, CTSS, SOD2, and COX-2 (80). Activation of NF-B by Dlk1 can thus be considered as an important mechanism by which Dlk1 mediates its regulatory effects on cytokine expression and consequently its regulatory function in hMSC. In supporting to this notion, several studies mentioned the inhibitory effect of NF-B-induced cytokines on osteoblast differentiation in MC3T3 cells (81), rabbit MSC cells (82), and human osteosarcoma Saos-2 cells (83) as well as on adipocyte differentiation (63). It is plausible that the effects of Dlk1 on the NF-B pathway may be mediated through its effects on Notch signaling. Dlk1 functions as an inhibitor of Notch signaling (20), and enhanced Notch signaling has been reported to inhibit NF-B activation via several mechanisms, including increased IB expression by Notch activation (84) and sequestering of NF-B in the cytoplasm by the intracellular Notch region liberated upon Notch activation (85).
Several factors are known to control the differentiation of MSC. For example, Core-binding factor 1 (CBFA1/Runx2), Ostrix (OSX), peroxisome proliferator-activated receptor ␥2 (86), and 14-3-3-binding protein (TAZ) (87) are transcription factors that initiate intrinsic genetic programs controlling the differentiation decision of MSC toward osteoblasts or adipocytes. However, these genetic programs must be subjected to environmental regulation, a concept known as the "stem cell niche" (88). Thus, both the signals present in the hMSC microenvironment niche and the intrinsic differentiation programs cooperate to control the biological functions of hMSC. Here we propose that dlk1 plays an important role in determining hMSC fate decisions of either self-renewal, lineage-specific differentiation, or immune-related responses by modulating the molecular composition of the hMSC microenvironment niche (Fig. 7). Furthermore, it is plausible that the presence of biologically active and free circulating dlk1 can mediate the interaction between bone cells and immune cells, suggesting a new role of Dlk1 in bone inflammation and immune regulation that needs further investigation.
Acknowledgments-We thank Dr. Charlotte H. Jensen for performing FA1 purification and enzyme-linked immunosorbent assay measurements and Tina K. Nielsen and Bianca Jorgensen for excellent technical assistance. We thank Dr. Torben Kruse for help with the microarray studies.  7. Schematic model of the cytokine-mediated mechanism used by dlk1/FA1 to regulate the multiple biological functions of hMSC in bone marrow. Dlk1 expression stimulates hMSC to produce several pro-inflammatory cytokines that affect the composition of their niche and shown to mediate fate decision of hMSC in bone marrow to self-renewal, differentiation, immune response, and hematopoietic supportive ability.