PPARgamma2 nuclear receptor controls multiple regulatory pathways of osteoblast differentiation from marrow mesenchymal stem cells

Abstract

Rosiglitazone (Rosi), a member of the thiazolidinedione class of drugs used to treat type 2 diabetes, activates the adipocyte-specific transcription factor peroxisome proliferator-activated receptor gamma (PPARgamma). This activation causes bone loss in animals and humans, at least in part due to suppression of osteoblast differentiation from marrow mesenchymal stem cells (MSC). In order to identify mechanisms by which PPARgamma2 suppresses osteoblastogenesis and promotes adipogenesis in MSC, we have analyzed the PPARgamma2 transcriptome in response to Rosi. A total of 4,252 transcriptional changes resulted when Rosi (1 microM) was applied to the U-33 marrow stromal cell line stably transfected with PPARgamma2 (U-33/gamma2) as compared to non-induced U-33/gamma2 cells. Differences between U-33/gamma2 and U-33 cells stably transfected with empty vector (U-33/c) comprised 7,928 transcriptional changes, independent of Rosi. Cell type-, time- and treatment-specific gene clustering uncovered distinct patterns of PPARgamma2 transcriptional control of MSC lineage commitment. The earliest changes accompanying Rosi activation of PPARgamma2 included effects on Wnt, TGFbeta/BMP and G-protein signaling activities, as well as sustained induction of adipocyte-specific gene expression and lipid metabolism. While suppression of osteoblast phenotype is initiated by a diminished expression of osteoblast-specific signaling pathways, induction of the adipocyte phenotype is initiated by adipocyte-specific transcriptional regulators. This indicates that distinct mechanisms govern the repression of osteogenesis and the stimulation of adipogenesis. The co-expression patterns found here indicate that PPARgamma2 has a dominant role in controlling osteoblast differentiation and suggests numerous gene-gene interactions that could lead to the identification of a "master" regulatory scheme directing this process.

Fig. 1. Factorial design for the study of PPARγ-induced cell differentiation in mesenchymal stem cell cultures

Marrow stromal cells transfected with PPARγ2 (U-33/γ2) or a control vector (U-33/c) were subjected to rosiglitazone (Rosi) to produce four different cell states. Replicate experiments were performed 2, 24 and 72h after induction with Rosi to generate a total of 24 samples. Initial cell plating is showed as open circles and harvest points are shown as black circles.

Fig. 2. Generation of contrast sets

A gene universe was selected based on mappings to Entrez gene identifiers (see Experimental Procedures). When multiple differentially expressed transcripts (DETs) mapped to a single Entrez gene, the probe set with the largest F_s statistic across all conditions was chosen to represent the gene. Time-specific contrasts were used to find significant differences due to activation of the U-33/γ2 cell type [U-33/γ2 vs. (Rosi + U-33/γ2)] (Set 1), the presence of Rosi in U-33/c cells [U-33/c vs. (Rosi + U-33/c)] (Set 2), or between cell types [U-33/c vs. U-33/γ2] (Set 3). Overlapping and non-overlapping groups of time-specific differences between cell states are represented with Venn diagrams and referred to as contrast sets. Numbers in Venn diagrams refer to the number of differentially expressed transcripts.

Fig. 3. Clustering of expression estimates due to activation of U-33/γ2 cells with Rosi

(A) A heat map and k-means clustering of log₂ transformed expression estimates standardized across conditions for activated U-33/γ2 versus inactivated U-33/γ2 cells at each harvest point. Time (2, 24 or 72h) is indicated with a black triangular symbol above each heat plot. (B) A modified GO slim analysis was used to describe prevailing biological processes in each cluster (see Experimental Procedures). The percentage of genes from each cluster in (A) that can be interpreted with the modified GO slim procedure is shown next to each bar. In (B) the total number of genes shown for each cluster may exceed the number of genes in each cluster from (A) due to the many-to-one and one-to-many mapping relationships between genes and GO terms.

Fig. 4. Clustering of expression estimates due to differences between cell types

(A) Heat map and k-means clustering of log₂ transformed expression estimates standardized across conditions for un-induced U-33/γ2 versus U-33/c cells at each harvest point. Time (2, 24 or 72h) is indicated with a black triangular symbol above each heat plot. (B) A modified GO slim analysis was used to describe prevailing biological processes in each cluster (see Experimental Procedures). The percentage of genes from each cluster in (A) that can be interpreted with the modified GO slim procedure is shown next to each bar.

Fig. 5. “Early” responding gene transcripts to Rosi treatment in U-33/γ2 cells

Directed acyclic subgraph of biological process gene ontology terms over-represented at 2h due to activation of U-33/γ2 cells with Rosi. The numbers are node identifiers and the colors indicate significant biological processes (p < 0.01). blue – anatomical structure morphogenesis, yellow – signal transduction, light blue – lipid metabolism, red – apoptosis, orange – immune system, green – inflammatory response, beige – non-specific or general processes, and white – not significant. Arrow and dash designators underneath “leaf” nodes indicate prevailing changes for significant genes in each process at 2, 24 and 72h, respectively. Arrows indicate up- and down-regulation while a “-“ indicates that an equal number of genes in the node were up- and down-regulated at a given time point.

Fig. 6. Schematic representation showing the number of gene expression changes involved in osteoblast-specific signaling pathways and phenotype regulators as function of treatment duration with Rosi

Numbers in boxes represent number of pathway-specific genes with expression changes at each analyzed time point. Arrows indicate whether the expression of listed gene was up- or downregulated at the given time point.