Gremlin Enhances the Determined Path to Cardiomyogenesis

Background The critical event in heart formation is commitment of mesodermal cells to a cardiomyogenic fate, and cardiac fate determination is regulated by a series of cytokines. Bone morphogenetic proteins (BMPs) and fibroblast growth factors have been shown to be involved in this process, however additional factors needs to be identified for the fate determination, especially at the early stage of cardiomyogenic development. Methodology/Principal Findings Global gene expression analysis using a series of human cells with a cardiomyogenic potential suggested Gremlin (Grem1) is a candidate gene responsible for in vitro cardiomyogenic differentiation. Grem1, a known BMP antagonist, enhanced DMSO-induced cardiomyogenesis of P19CL6 embryonal carcinoma cells (CL6 cells) 10–35 fold in an area of beating differentiated cardiomyocytes. The Grem1 action was most effective at the early differentiation stage when CL6 cells were destined to cardiomyogenesis, and was mediated through inhibition of BMP2. Furthermore, BMP2 inhibited Wnt/β-catenin signaling that promoted CL6 cardiomyogenesis. Conclusions/Significance Grem1 enhances the determined path to cardiomyogenesis in a stage-specific manner, and inhibition of the BMP signaling pathway is involved in initial determination of Grem1-promoted cardiomyogenesis. Our results shed new light on renewal of the cardiovascular system using Grem1 in human.

In this study, we performed GeneChip analysis to identify multiple extracellular determinants, such as cytokines, cell membrane-bound molecules and matrix responsible for cardiomyogenic differentiation, and evaluated the statistical significance of differential gene expression by NIA array analysis (http://lgsun. grc.nia.nih.gov/ANOVA/) [20], a web-based tool for microarray data analysis. We found that Grem1 enhances the determined path to cardiomyogenesis in a stage-specific manner, and that inhibition of the BMP signaling pathway is, at least in part, involved in initial determination of Grem1-promoted cardiomyogenesis.

GeneChip and statistical analysis
To identify cytokines and transcription factors responsible for cardiomyogenic differentiation, 69 human cells were analyzed, depending on gene expression levels, by GeneSpringGX software, and clustered into 30 groups (Fig. 1A, Table 1). Among the 30 groups, 21 groups included cells with a cardiomyogenic potential ( Fig. 1B: red numbers). To identify genes specific for these groups, hierarchical clustering was employed, using the average distance method. Genes with the lowest average expression E(G1) within the cluster that can differentiate into cardiomyocytes and genes with the highest average expression E(G2) outside the cluster were identified, as previously described [20][21][22]. Genes which have E(G1).E(G2) were estimated, using the False Discovery Rate (FDR,0.05). Grem1 was nominated as a cluster-specific cardiomyocyte-promoting gene in cells that could differentiate into cardiomyocytes following NIA array analysis (Fig. 1B). The gene expression profile reported in this paper has been deposited in the Gene Expression Omnibus (GEO) database (http://www.ncbi. nlm.nih.gov/geo: accession no. GSE8481, GSM41342-GSM41344, and GSM201137-GSM201145).

Immunocytochemistry of differentiated or undifferentiated CL6 cells
To examine CL6 cells for expression of cardiomyocytic protein, immunocytochemical analysis was performed. CL6 treated with Grem1 (125 ng/ml) and DMSO exhibited clear striation with immunostain using anti-cTnT or anti-a-actinin ( Fig. 2Da and b). The MF20-and cTnT-positive cells after exposure to Grem1 and DMSO formed clusters (Fig. 2Ea), compared with the cells after exposure to DMSO alone (Fig. 2Eb). CL6 cells treated with Grem1 alone were negative for MF20 and cTnT, but became positive for both markers following exposure to Grem1 (63 and 125 ng/ml) and DMSO (Fig. 2F). The beating area (Fig. 2B) showed a tendency similar to the MF20-and cTnT-positive area (Fig. 2F), thus there were positive correlations between them.  To determine if Grem1 (125 ng/ml) functions during the early or the late stage of differentiation, CL6 cells were treated with Grem1 for different time periods (Fig. 3A). Grem1 and DMSO were most effective on CL6 differentiation at 1-3 days (Fig. 3B, C) as assessed by percentages of MF20-positive area and beating area. Since Grem1 inhibits BMPs through direct binding [24], we hypothesized that BMP signaling is inhibitory to CL6 cardiomyogenesis during days 1-3. To confirm this hypothesis, RT-PCR analysis was performed to determine expression of the early mesodermal marker (BrachyuryT and Tbx6), cardiomyocyte-specific transcription factors (Csx/Nkx2.5), structural genes (b-MyHC), and Gapdh (Fig. 4A). DMSO induced the BrachyuryT and Tbx6 genes, and their expressions peaked at 3 days and then decreased; BMP2 down-regulated expression of these genes at 3-7 days. The Csx/ Nkx2.5 and b-MyHC genes started to be expressed at days 3 and 5, respectively, and their expression increased up to 14 days, at which time the timeframe analysis was terminated. BMP2 clearly inhibited expression of the Csx/Nkx2.5 and b-MyHC genes (Fig. 4A, lanes 1-7 versus lanes [8][9][10][11][12][13][14]. To examine cardiomyogenic differentiation, immunocytochemical analysis was performed on CL6 cells treated with the inducers. CL6 cells treated with DMSO and BMP2 for the first 3 days were negative for sarcomeric myosin (MF20) at 14 days, but became positive for sarcomeric myosin, following exposure to DMSO alone during days 1-3 (Fig. 4B). To determine if DMSO induces BMP production in CL6 cells, expression levels of Bmp2 and Bmp4 were determined by quantitative real-time RT-PCR analysis (Fig. 4C). DMSO clearly induced the Bmp2 and Bmp4 genes, and DMSO-induction was inhibited by BMP2 protein. The expression level of Bmp2 was highest during days 7-10 ( Fig. 4C: Bmp2) in DMSO-induced CL6 cells, and that of Bmp4 was highest during days 5-7 ( Fig. 4C: Bmp4).
To investigate BMP signaling on cardiomyogenic differentiation, we used the Id1 promoter-Lux plasmid that includes the luciferase gene driven by the Id1 promoter, known as a BMP target promoter (Fig. 4D). DMSO increased BMP signaling activity that peaked at 5 days (Fig. 4D, open square). BMP2 protein increased BMP signaling activity at 3 days (Fig. 4D, closed square), but lost BMP signaling activity at 5 days and later, implying that this loss of BMP signaling leads to lack of cardiomyogenic induction.
Since Wnt/b-catenin signaling is involved in CL6 cardiomyogenesis [23,25], we hypothesized that the BMP effect on CL6 cardiomyogenesis is mediated through Wnt/b-catenin signaling. Expression of Wnt3a, an activator of canonical Wnt signaling, was indeed detected in CL6 cells exposed to DMSO, and BMP2 significantly down-regulated Wnt3a expression at day 3 (Fig. 4E). By using the TOPflash plasmid [23] which includes the luciferase gene driven by two sets of three copies of the TCF recognition site, Wnt/b-catenin signaling was assessed to investigate the effect of BMP2. Wnt/b-catenin signaling activity increased at 48 h after treatment with DMSO. Activity was increased by DMSO treatment but decreased by BMP2 (Fig. 4F). Time course analysis revealed that Wnt/b-catenin activity peaked at 5 days after DMSO treatment, and decreased thereafter (Fig. 4G). BMP2 inhibited DMSO-induced Wnt/b-catenin activity throughout the experimental period (up to 14 days). These results imply that BMP signaling inhibits CL6 cardiomyogenesis at the early stage through inhibition of Wnt/b-catenin signaling.

Discussion
Our bioinformatics study using the results from the global gene expression analysis of human cells (GSM412342-41344 and GSM201137-201145 at http://www.ncbi.nlm.nih.gov/geo) nominated Grem1 as a candidate gene that may participate in cardiomyogenesis. By using CL6 embryonic cells as a model of cardiomyogenesis, we obtained two major findings: the first is that Grem1 enhanced cardiomyogenic differentiation of DMSOinduced CL6 cells at the early stage; the second is that Wnt/bcatenin and BMP signaling activity had developmental stagespecific effects on cardiomyogenesis (Fig. 5). Wnt/b-catenin activity at the early stage enhanced embryonic cell differentiation into cardiomyocytes, while suppressing this activity by BMP2 or BMP4 proteins as reported in the avian embryo [26]. In contrast, BMP signaling activity in the late stage enhanced cardiomyocytic differentiation. Grem1 regulated the stage-specific Wnt/b-catenin and BMP signaling activity on cardiomyogenesis.
Many studies have indicated that Grem1 is involved in cell differentiation and development, such as osteogenesis [27], lung morphogenesis [28], myogenesis [29], and limb formation [30], through inhibition of BMP2 and BMP4. Grem1-null mice show intact heart development, despite impairment of lung and kidney [31], and therefore Grem1 is considered not to be involved in cardiogenesis, or supplementary factors such as Noggin [32], with a similar function, may compensate Grem1 during development. Grem1 had an enhancing or promoting activity in in vitro cardiomyogenesis, as is the case with platelet-derived growth factor as a promoter of cell growth [33]. In this study, Grem1 was involved in cardiomyocyte differentiation. However Grem1 alone could not induce cardiomyocytic differentiation of CL6 cells in the absence of DMSO (Fig. 2C and F), suggesting that Grem1 is solely and DMSO. CL6 cells were passaged at 1.8610 5 cells in 6-well plate on Day 0. CL6 cells were exposed to Grem1 (125 ng/ml) and/or DMSO on the indicated day. Day when the cells were exposed to the inducers is shown by ''+'' (in gray cells for clarity). The medium including Grem1 and DMSO was changed every day. On day 14, the cells were immunocytochemically stained with MF20 antibody. (B) Myogenic differentiation of CL6 cells was estimated by sarcomeric myosin (MF20)-positive area. CL6 cells were treated with Grem1 (125 ng/ml) and DMSO for the indicated days. (C) Myogenic differentiation of CL6 cells was estimated by beating area. CL6 cells treated with DMSO and Grem1 (125 ng/ml) were incubated at indicated days. doi:10.1371/journal.pone.0002407.g003 a promoter of cardiomyogenic differentiation. One of the possible mechanisms for Grem1-enhanced cardiomyogenesis at the early stage is inhibition of the BMP signaling pathway [3]. Alternatively, Grem1-enhanced cardiomyogenesis may be mediated through proliferation of cardiac progenitor cells, as is the case of myogenic progenitor proliferation by Grem1 [34], and this possibility is supported by an increased number of sarcomeric myosin-positive CL6 cardiomyocytes (Fig. 2E and F).
The stage specificity of the Grem1 effect is possibly correlated with the biphasic and antagonistic effect of Wnt/b-catenin signaling on cardiomyogenesis, depending on the stage of development in vitro [25] and in vivo [35]. CL6 cells differentiated into cardiomyocytes via mesodermal induction by the Wnt/bcatenin signaling pathway at the early stage, and CL6 mesodermal cells differentiated into cardiomyocytes induced by BMP2 at the late stage. It is conceivable that embryonic cells, such as CL6 cells and ES cells, differentiate into cardiomyocytes by inhibiting BMP signaling via putative ''mesodermal cells'' or ''cardiomyogenic progenitors'', or differentiation stages corresponding to these cells (Fig. 5, Figure S2). The early stage process from embryonic cells to mesodermal cells was mediated via Wnt/b-catenin signaling (Fig. 4F, G), and was assessed by expression of BrachyuryT and Tbx6 genes (Fig. 4A), which are target genes for Wnt/b-catenin signaling [36]. BMP signaling antagonizes the cell fate-inducing activity of Wnt/b-catenin [37]. When embryonic cells or cardiomyogenic progenitors are induced to become mature cardiomyocytes by cytokines and growth factors, we must be careful with respect to the stage of cell differentiation because of the biphasic differential action of the factors which are dependent upon the differentiation stage.
In conclusion, we have demonstrated that Grem1 enhances the commitment or determined path to cardiogenic differentiation of CL6 teratocarcinoma cells. Apart from a role in development, Grem1 may serve a clinical use in cardiology, like granulocyte colony-stimulating factor that accelerates production of granulocytes in both peripheral blood and bone marrow. Nomination of Grem1 as a cardiomyogenic factor is based on hierarchical clustering analysis using global gene expression data of human cells. This bioinformatics approach may be useful for identifying morphogens/factors that can induce differentiation of other cell types/tissues/organs.

GeneChip analysis
GeneChip analysis was performed (Fig. 1A, Table 1) as previously described [38]. Human genome-wide gene expression was examined with the Human Genome U133A Probe array (GeneChip; Affymetrix), which contains the oligonucleotide probe set for approximately 23,000 full-length genes and expressed sequence tags, according to the manufacturer's protocol (Expression Analysis technical manual and GeneChip Small Sample Target Labeling Assay version 2 technical note [http://www. affymetrix.com/support/technical/index.affx]). Data analysis was performed by the GeneChip Operation System (Affymetrix) and GeneSpringGX software (Silicon Genetics). To normalize the staining intensity variations between chips, the average difference values for all genes on a given chip were divided by the median of all measurements on that chip. Hierarchical-clustering analysis was performed using a minimum distance value of 0.001, a separation ratio of 0.5, and the standard definition of the correlation distance.
Quantitative real-time RT-PCR was performed on an ABI Prism 7700 Sequence Detection System (Applied Biosystems), using 100 ng of cDNA in 25 ml reaction volume with 10 nmol/l of each primer, and 12.5 ml SYBR Green Realtime PCR Master Mix (TOYOBO). PCR primers for the genes of Bmp2, Bmp4, Wnt3a, and Gapdh (Table S1b) were obtained from PrimerBank (http:// pga.mgh.harvard.edu/primerbank/index.html). Calculations were automatically performed by ABI software (Applied BioSystems).

Transfection and luciferase assays
Cells (8.0610 5 ) seeded and cultured in 60 mm dishes (Becton Dickinson) were transfected 18 h after plating using Lipofectamine 2000 (Invitrogen) and PLUS reagent (Invitrogen) in Opti-MEM (Gibco). Transfection contained 1.0 mg of TOPflash plasmid (Upstate Biotechnology) for measurement of Wnt/b-catenin activity, or 5.0 mg of the Id1 promoter-Lux plasmid (provided by Dr Imamura and Dr Miyazono) for measurement of BMPinduced Id1 gene transcription, and 0.5 mg of pRL-CMV (Promega) as co-transfected control. Medium containing 10% FBS was changed 3 h after transfection and transfected cells (1.8610 5 ) were re-seeded in 6-well plates 24 h after transfection. After 18 h, CL6 cells were induced with BMP2 (100 ng/ml) and DMSO. CL6 cells were prepared for luciferase activity analysis using Dual luciferase reporter assay system (Promega).

Area calculation
The regions of interest (beating area, immunostaining area) were defined in Photoshop (Adobe systems) using the 'magic wand' tool. The total numbers of pixels identified were then counted using the histogram function. At least five different fields were measured for each dish.

Statistical analysis
Results, shown as the mean6SE, were compared by ANOVA followed by Scheffé's test, with P,0.05 considered significant. Figure S1 A semi-quantitative RT-PCR of cardiomyocytespecific genes. To investigate expression level of cardiomyocytespecific genes (Csx/Nkx2.5, Gata4, MyLC-2a, and MyLC-2v), a semi-quantitative RT-PCR was performed from CL6 cells treated with 1% DMSO and the indicated concentration of Grem1 for 14 days. Each RT-PCR product was electrophoresed in 2% agarose gel, and was measured using ImageJ software (http://rsb.info.nih. gov/ij/) to calculate the ratio of each gene to Gapdh. The expression level for each gene is determined relative to that of Gapdh, and expression level in CL6 cells treated with DMSO alone was regarded as 1.0. The relative expression levels were averaged from at least three independent experiments. Found at: doi:10.1371/journal.pone.0002407.s001 (1.04 MB DOC) Figure S2 Grem1 enhanced cardiomyogenic differentiation of mouse ES cells. Mouse ES cells (NCH1.5, C57BL/6J6129ter/Sv) were cultured on a mouse embryonic fibroblast feeder layer inactivated with 30 Gy c-irradiation in gelatin-coated 60 mm dishes (Becton, Dickinson). Cells were grown in KnockOut DMEM (Gibco) supplemented with 15% fetal bovine serum (Cell Culture Technologies), 2 mM GlutaMAX (Gibco), 0.1 mM nonessential amino acid (Gibco), 0.1 mM 2-mercaptoethanol (Gibco), penicillin, streptomycin, and 2,000 U/ml mouse leukemia inhibitory factor (LIF) (Chemicon). For cardiomyogenic differentiation, ES cells were exposed to 125 ng/ml Grem1 (R&D systems) for the three days. The cells were then trypsinized and cultured to form embryonic bodies (EBs) from a single cell using a threedimensional culture system (without LIF) on low cell binding dishes (96-well plate round bottom). This represented day 0 of EB formation. On the next day, the medium was replaced with the same medium without LIF. EBs were re-seeded on gelatin-coated 48-well plates with one EB per well, on day 8 after the start of EB formation. The cardiomyogenic induction was estimated by the beating EB number per total EB number, measured on day 12 under a phase-contrast microscope. Grem1 increased the percentage of beating EBs to 69.2%, as compared with 26