Gene regulatory network analysis identifies MYL1, MDH2, GLS, and TRIM28 as the principal proteins in the response of mesenchymal stem cells to Mg2+ ions

Abstract

Magnesium (Mg)-based implants have emerged as a promising alternative for orthopedic applications, owing to their bioactive properties and biodegradability. As the implants degrade, Mg²⁺ ions are released, influencing all surrounding cell types, especially mesenchymal stem cells (MSCs). MSCs are vital for bone tissue regeneration, therefore, it is essential to understand their molecular response to Mg²⁺ ions in order to maximize the potential of Mg-based biomaterials. In this study, we conducted a gene regulatory network (GRN) analysis to examine the molecular responses of MSCs to Mg²⁺ ions. We used time-series proteomics data collected at 11 time points across a 21-day period for the GRN construction. We studied the impact of Mg²⁺ ions on the resulting networks and identified the key proteins and protein interactions affected by the application of Mg²⁺ ions. Our analysis highlights MYL1, MDH2, GLS, and TRIM28 as the primary targets of Mg²⁺ ions in the response of MSCs during 1-21 days phase. Our results also identify MDH2-MYL1, MDH2-RPS26, TRIM28-AK1, TRIM28-SOD2, and GLS-AK1 as the critical protein relationships affected by Mg²⁺ ions. By offering a comprehensive understanding of the regulatory role of Mg²⁺ ions on MSCs, our study contributes valuable insights into the molecular response of MSCs to Mg-based materials, thereby facilitating the development of innovative therapeutic strategies for orthopedic applications.

Keywords: Gene regulatory network analysis; Magnesium ions; Mesenchymal stem cells; Proteomics.

Graphical abstract

Fig. 1

The overview of the methodology implemented in this study. (A) Proteomics data were collected for the control and treatment groups, where the treatment group was exposed to 5 mM of Mg²⁺ ions. (B) The proteomics data were normalized to remove potential biases, and missing values were imputed using two the methods of KNN and MinProb. (C) Time-series analysis was performed to identify the most sensitive proteins to Mg²⁺ ions. (D) Network analysis was performed to reveal protein regulatory connections. (E) Model selection was employed to choose the best-performing GRN model according to the known links provided by the STRING database. (F) Protein role analysis was employed to obtain the proteins’ regulatory roles and to determine the proteins with a significant role change in response to Mg²⁺ ions. (G) Protein regulatory divergence analysis was performed to detect the regulatory links most affected by Mg²⁺ ions. Target proteins were identified using the cumulative information of F and G.

Fig. 2

The protocol of GRN inference using regression analysis. Time-series data (A) was discretized to obtain learning samples of $F_i$ (B). Next, the data was partitioned into sets of features and targets for each protein (C). Then, n regression models were trained using the expression data of n proteins in the network using RF and Ridge models (D).

Fig. 3

(A) The distribution of the missingness across different samples and measurement days. (B) The count of the DE proteins identified for different datasets and their interactions. Gene names are used instead of protein names. (C) Prediction scores of the protein expression for the regression-based models of RF and Ridge. Those models indicated as ✓ outperform the baseline random models (red dashed line). (D) The performance of different models in recovering the known links, measured by EPR score. The baseline models represent the scores obtained for the 1000 randomly generated networks. The numbers in percentage indicate the percentile rank of each model compared to the random models. The models chosen for short- and long-term responses are distinguished by their names encircled in a red oval. (E) The performance of the selected models of ShortTerm-MinProb-Portia and LongTerm-KNN-Portia compared to 100 models built using randomly chosen proteins as regulators. Red asterisks (*) represent the top 5% percentile ranks.

Fig. 4

The results of the protein role analysis for the selected model of (A) the short-term and (B) the long-term response. The dashed lines in the plots indicate the upper 75th percentile on each axes. Column (3) shows the significant protein role change across control and Mg²⁺ ions treatment. Gene names are used instead of protein names.

Fig. 5

The significance of divergence in the regulatory links due to Mg²⁺ ion treatment is given in the first column, A-a and B-a. The color of the edges shows the significance of the change from control to Mg²⁺ ions treatment. The active sum shows the regulatory influence of a protein on the rest of the network. The regulatory effect shows the strength of the regulation. These two complementary indications are only for the second column (A-b and B-b). Gene names are used instead of protein names. The graph was created using Cytoscape .