Line plots displaying scaled log2fc values over the 14-day time course. Eight pathways of interest were explored here: A) Adipogenesis, B) Clear cell renal cell carcinoma pathways, C) Epidermal growth factor/Epidermal growth factor receptor signalling pathway, D) Endochondral ossification, E) Endochondral ossification with skeletal dysplasia’s, F) Gastrin signalling pathway, G) Metabolic reprogramming in colon cancer, and H) Vascular endothelial growth factor A/ Vascular endothelial growth factor receptor 2 signalling pathway. Here, individual genes found in the filtered miRNA-mRNA interactions for each pathway are plotted along a time course. Only genes (miRNAs or mRNAs) that have a scaled log2FC value of at least 1 at any point of the line plot are highlighted and labelled.

The most frequent significantly enriched signalling pathways for each time point based differential expression analyses conducted (DX/ D0) (D=day, D1, D3, D7, D10, D14) are presented. An X marks if the mechanistic pathway was significantly enriched (Benjamini-Hochberg adjusted P value < 0.05) at a that time point.

Chondrocytes were transfected for 24 hours with a non-target miRNA control, miR-199a-5p mimic or miR-199b-5p hairpin inhibitor. A) miR-199a-5p was overexpressed in chondrocytes and measured at day 7 of chondrogenesis. Measurements of chondrogenic biomarkers were taken from three human MSC donors, and technical repeats of each were carried out in quadruple. Averaged qPCR values were plotted along with the standard error of the mean values (SEM), where the control level was set at 100%. DMB measurements were taken at day 7 to determine GAG levels. B) miR-199b-5p was knocked-down prior to chondrogenesis initiation, after initiation measurements of ACAN, COL2A1 and SOX9 were taken at days: 0, 1, 3 and 7 of chondrogenesis, and DMB measurements were taken at day 7 to determine GAG levels. qPCR results for day 0 were undetectable for COL2A1. Percentages were calculated from the means of qPCR data with the control levels at each day termed 100% to allow visualisation, and averaged qPCR values were plotted with SEM. NS = not significant, * = <0.05, ** = < 0.01, *** = < 0.001. Samples numbers are as per A.

Results from RNAseq analysis of control miRNA, miR-199a-5p and miR-199b-5p knockdown experiments. A) Gene Set Enrichment Analysis (GSEA) was used to conduct Gene ontology (GO) analysis for the significantly differentially expressed genes found from miR-199a-5p or miR-199b-5p inhibition at day 0 and day 1 of chondrogenesis. Up to five activated and five suppressed pathways are displayed for each DE analysis. All GO terms shown have an adjusted P value of <0.05. Count size represents the genes found in a pathway and this determined the size of the circles. B) Volcano plots with the miRNAtap-selected 21 miR-199a/b-5p targets highlighted and some are labelled if space on the plot was available. Up-regulated targets are blue, down-regulated targets are red and invariant targets are black. Non-miR-199a/b-5p target genes were also displayed, and of these up-regulated genes are pink, down-regulated genes are light blue and non-significantly differentially expressed genes are light grey. The cut-off for significance was an adjusted (BH) P value <0.05. C) Scatter plot shows a more detailed account of the 21 miR-199a/b-5p targets found from RNAseq experiment.

A1 2

MSCs were transfected with hairpins to knockdown miR-199a-5p expression, miR-199b-5p expression or miR-199a/b-5p expression. FZD6 and ITGA3 expression was measured after miR-199a-5p and miR-199a/b-5p knockdown, and CAV1 was measured after miR-199b-5p knockdown. Measurements were taken at times 0, 1, 3 and 7 days after transfection. B1) Luciferase assays showing luciferase expression following co-transfection of miR-199a-5p target 3’ UTR reporter constructs for 24 hours. Expression was normalised to renilla luciferase and presented as percentage of control levels. B2) Mutagenesis experiments showing miR-199a-5p overexpression and control miRNA overexpression effects on control FZD6 and ITGA3 and in 3’-UTR mutated FZD6 and ITGA3. For A and B, values are the mean +/-SEM of data pooled from three independent experiments. NS = not significant, * = <0.05, ** = < 0.01, *** = < 0.001.

Initial kinetic modelling of miR-199a/b-5p regulation of chondrogenesis. A) Schematic of how miR-199a/b-5p modulation effects the predicted miR-199a/b-5p FZD6, ITGA3 and CAV1, and the chondrogenic biomarkers SOX9, COL2A1, ACAN and GAG. Made using BioRender. B) Simulations (blue lines) from the kinetic modelling were contrasted against the experimental data – if available (red line) and a MSE score is provided in these cases. Alternatively, if no experimental data was available, a dashed blue line displays the predicted behaviour of the gene. If multiple measurements were available, they have been displayed using red crosses. C) A more detailed model displaying how miR-199a/b-5p regulates chondrogenesis via FZD6, ITGA3 and CAV1 mRNAs, in GRN form. Here information from literature was added and miR-140-5p was added to the model. The GRN shown here is a minimalistic version of Supplementary Figure 1. This was used to inform the topology of a kinetic model which aimed to explain how miR-199a-5p and miR-199b-5p act as pro-chondrogenic regulators by down-regulating activity of FZD6, ITGA3 and CAV1 mRNAs. This GRN contained 18 species including: two proteins (TGFB3, SOX9), one phospho-protein (phospho-SOX9) three mRNAs (SOX9, ACAN, COL2A1), three miRNAs (miR-140-5p, miR-199a-5p, miR-199b-5p), two drugs (hpmiR-199a-5p, hpmiR-199b-5p), six Gene Activity (SRC, CAV1, FZD6, ITGA3, OtherTargets, OtherTargetsRegulator), and one phenotype (GAG). Each species has a sink and a source. Species are also shaped based on their properties: Proteins are rectangles, RNAs are rhombus, Phenotypes are hexagons, Drugs are oval, Gene Activity are rectangles with dotted lines. Species are also highlighted with a white box if they are found in the ECM or pink if they are found within a chondrocyte. Edges between species are solid if there is literature/ data supporting an interaction or dotted if there the interaction is hypothetical. Species are also colour coded: green if there is associated data, blue if there is some data and the rest has been inferred based on literature, or grey if there is no data associated with the species. D) Simulations from modelling the more detailed miR-199a/b-5p chondrogenesis model were shown here. Notations follow Figure 5B.