Identification of Pathways Mediating Growth Differentiation Factor5-Induced Tenogenic Differentiation in Human Bone Marrow Stromal Cells

Abstract

To date, the molecular signalling mechanisms which regulate growth factors-induced MSCs tenogenic differentiation remain largely unknown. Therefore, a study to determine the global gene expression profile of tenogenic differentiation in human bone marrow stromal cells (hMSCs) using growth differentiation factor 5 (GDF5) was conducted. Microarray analyses were conducted on hMSCs cultures supplemented with 100 ng/ml of GDF5 and compared to undifferentiated hMSCs and adult tenocytes. Results of QuantiGene® Plex assay support the use and interpretation of the inferred gene expression profiles and pathways information. From the 27,216 genes assessed, 873 genes (3.21% of the overall human transcriptome) were significantly altered during the tenogenic differentiation process (corrected p<0.05). The genes identified as potentially associated with tenogenic differentiation were ARHGAP29, CCL2, integrin alpha 8 and neurofilament medium polypeptides. These genes, were mainly associated with cytoskeleton reorganization (stress fibers formation) signaling. Pathway analysis demonstrated the potential molecular pathways involved in tenogenic differentiation were: cytoskeleton reorganization related i.e. keratin filament signaling and activin A signaling; cell adhesion related i.e. chemokine and adhesion signaling; and extracellular matrix related i.e. arachidonic acid production signaling. Further investigation using atomic force microscopy and confocal laser scanning microscopy demonstrated apparent cytoskeleton reorganization in GDF5-induced hMSCs suggesting that cytoskeleton reorganization signaling is an important event involved in tenogenic differentiation. Besides, a reduced nucleostemin expression observed suggested a lower cell proliferation rate in hMSCs undergoing tenogenic differentiation. Understanding and elucidating the tenogenic differentiation signalling pathways are important for future optimization of tenogenic hMSCs for functional tendon cell-based therapy and tissue engineering.

Fig 1. The candidate tenogenic markers (COL-I, TNMD, TNC and SCX) expression of GDF5 (100 ng/ml)-induced hMSC on day 4 (A, B, C) and day 10 (D, E, F) by immunofluorescence imaging.

The extent of candidate tenogenic markers expressions were increased in GDF5 treated hMSC compared to the untreated control. An increase in the intensity of the expression of these markers was also observed in day 10 GDF5-induced hMSCs compared to that of day 4. Images were captured at 63X objective and a scale bar (50 μm) was depicted on the right bottom corner of the overlay images.

Fig 2. Overview of microarray analysis: principle component analysis (PCA) and Limma analysis.

PCA analysis was performed on all samples and all probes to characterize the variability present in the data. The results showed a distinct separation between all the groups. The PCA was visualized in 2D view (A) and 3D view (B), with the different colour coded for different groups; and the 3D view (C) with the colour coded for different individual donor (In the legend, individual 1 to 6 were the bone marrow donors and individual 7 to 12 were the tendon donors). Image B and C showed that the arrays were grouped according to their experimental groups (treatment) but not according to the donor variation. (Group 1: Control hMSC, Group 2: Day-4 GDF5-induced hMSC, Group 3: Day-10 GDF5-induced hMSC, Group 4: tenocytes). The microarray experiments were designed to detect differential expression of transcripts with GDF5 treatment and were compared with Venn diagrams. The list of the significantly (corrected p-value) up- and down- regulated genes, were used to detect the altered candidate tenogenesis genes within the GDF5-treated groups (Group 2 and 3) as depicted in the intersections or uniqueness; between all comparisons with control hMSC (as depicted in D) and tenocytes compared to all the other groups (as depicted in E). The numbers in each section or intersections of the circles represented the total number of significantly differentially up- or down- regulated genes for the pairwise comparisons (as denoted above or below each circle). The numbers in green and red fonts indicated the significantly up- and down-regulated genes, respectively. (G1: Control hMSC; G2: Day-4 GDF5-induced hMSC; G3: Day-10 GDF5-induced hMSC; G4: tenocytes).

Fig 3. Expression levels of selected candidate tenogenic and non-tenogenic marker genes (total of nine genes) based on microarray and QuantiGene® Plex 2.0 Assay.

(A) The graphical representation of genes expression patterns in hMSCs in response to GDF5 induction; with their respective log₂ ratio based on microarray analysis. (B) Gene expression profiles independently validated using QuantiGene® Plex assay presented in log₂ ratio. Expression variation for each gene was visualized with standard deviation. Considering that Scx and Mmp3 were derived from three biological samples compared to that of microarray which was derived from six bioloigical samples, the variation in the expression profiles was not unexpected. Whereas the Pparγ, Sox9, Comp and Bglap were all weakly expressed genes (the non-tenogenic marker genes), thus the differences detected may result from the detected limit and sensitivity of the different platforms, which subsequently affected interplatform reproducibility of differentially expressed genes. (C) Expression levels of the ECM related, candidate tenogenic and non-tenogenic marker genes based on microarray analysis. The graphical representation of genes (n = 16) displaying changes in expression patterns in hMSC in response to GDF5 induction with their respective log₂ ratio based on microarray analysis. The genes which showed at least fold change of 2 (log₂ ratio = 1, red dotted line) and fold change of less than 0.5 (log₂ ratio = -1, green dotted line) were regarded as significantly up- and down- regulated genes respectively.

Fig 4. Cytoskeleton reorganization in hMSCs visualized by AFM.

Representative AFM topography scanning of control hMSCs (left upper), hMSCs at day 4 of GDF5 induction (right upper), hMSCs at day 10 of GDF5 induction (left lower) and tenocyte. In the topography images, brighter colour indicated higher distance off of substrate (cover slip). The panel on the right side of each image indicated the height scale (z-scale) of the topography. There was a marked difference in the topography (cytoskeleton organization) of the control hMSCs compared to the hMSCs exposed to GDF5. The topography of control hMSCs had a larger z-scale; apparently possessed higher morphology. Both control hMSCs and day-4 GDF5-induced hMSCs showed detailed treelike web structure of presumably the actin network under the cell membrane especially at the leading edges which strongly attached on the cover slip (black arrowheads). The GDF5-induced hMSCs possessed more flatten morphology because they adhered more strongly via the stress fibers that could be visualized just under the surface of the cell membrane (white arrowheads). The detailed structure of presumably the actin cytoskeleton (actin bundles or stress fibers) could be observed in the day-10 GDF5-induced hMSCs and tenocytes.

Fig 5. Actin cytoskeleton reorganization and nucleostemin (NST) expression in hMSCs upon GDF5 induction captured with confocal laser scanning microscope.

Representative images of sequential scanning: nucleus stained with Hoescht 33342 (first panel on the left), nucleostemin (NST) (with indirect FITC stain; second panel) and actin fibres (direct staining which specifically stained cellular F-actin; third panel) and the merged image of all channels (last panel on the right). Scale bar = 50μm (at 100x objective).