Integrated transcriptome-proteome analyses of human stem cells reveal source-dependent differences in their regenerative signature

Abstract

Mesenchymal stem cells (MSCs) are gaining increasing prominence as an effective regenerative cellular therapy. However, ensuring consistent and reliable effects across clinical populations has proved to be challenging. In part, this can be attributed to heterogeneity in the intrinsic molecular and regenerative signature of MSCs, which is dependent on their source of origin. The present work uses integrated omics-based profiling, at different functional levels, to compare the anti-inflammatory, immunomodulatory, and angiogenic properties between MSCs from neonatal (umbilical cord MSC [UC-MSC]) and adult (adipose tissue MSC [AD-MSC], and bone marrow MSC [BM-MSC]) sources. Using multi-parametric analyses, we identified that UC-MSCs promote a more robust host innate immune response; in contrast, adult-MSCs appear to facilitate remodeling of the extracellular matrix (ECM) with stronger activation of angiogenic cascades. These data should help facilitate the standardization of source-specific MSCs, such that their regenerative signatures can be confidently used to target specific disease processes.

Keywords: immunomodulatory; integrative; mesenchymal stem cell; multi-omics; proteomics; secretome; transcriptomics.

Graphical abstract

Figure 1

Transcriptomic analysis between UC-MSCs and adult-MSCs Principal-component analysis (PCA) of transcriptomic datasets for each MSC source (n = 3 donors per source) and corresponding heatmap of normalized fragments per kilobase of exon per million mapped fragments (FPKM) values from 2,127 statistically significant DEGs (Log |FC| > 2; FDR <0.05) between all sources of MSCs.

Figure 2

Transcriptomic enrichment analysis (A) Venn diagram showing the distribution of 2,127 DEGs in adult-MSCs compared with UC-MSCs. (B) GO enrichment analysis indicates the cellular localization of the 730 common adult-MSC DEGs. (C) All significant (FDR < 0.05) pathways upregulated in UC-MSCs compared with adult-MSCs. (D) Circos plot depicts the relationship between enriched immune-regulatory pathways and their corresponding genes. Clockwise from top: pathways and genes are ordered by their number of interactions. Ribbon size encodes cell value associated with row/column segment pair. Column segment value and ribbon color are decided by the number of interactions. (E) All significant (FDR < 0.05) pathways upregulated in adult-MSCs compared with UC-MSCs. (F) Protein network of genes involved in extracellular matrix (ECM) organization pathway identified in adult-MSCs. Yellow circles indicate ECM proteolysis enzymes (MMPs, ELANE, and PLG) identified in our genomic screen. All other significant genes identified in our screen are highlighted with a black border. No fill circles indicate genes that are not validated experimentally.

Figure 3

Proteomic analysis for AD-MSCs, BM-MSCs, and UC-MSCs (A) PCA of all identified proteins for each MSC source (n = 3 donors per source) and the corresponding heatmap of normalized Z scores from 1,341 statistically significant proteins (Z > 2; FDR < 0.05) enriched among all three MSC sources. (B) GO enrichment analysis for 1,341 significant proteins demonstrating their cellular localization. (C) Top 10 pathways identified from gene enrichment analysis of proteins identified in the extracellular fraction.

Figure 4

Functional correlation analysis between transcriptomics and proteomics (A) Functional correlation analysis for UC-MSCs. Venn diagram shows the distribution of differentially abundant proteins (DAPs) upregulated in UC-MSCs compared with adult-MSCs. Bar plot shows the overlap of pathways between transcriptomics and proteomics datasets upregulated in UC-MSCs. (B) Correlation heatmap between transcriptomics and proteomics of genes for functionally overlapped pathways for UC-MSCs. (C) Functional correlation analysis for adult-MSCs. Venn diagram shows the distribution of DAPs upregulated in adult-MSCs compared with UC-MSCs. Bar plot shows the overlap of pathways between transcriptomics and proteomics datasets upregulated in adult-MSCs. (D) Correlation heatmap between transcriptomics and proteomics of genes for functionally overlapped pathways for adult-MSCs.

Figure 5

Comparison of growth factors and immune mediators in the secretome of UC-MSCs and adult-MSCs (A) PCA and heatmap of mean fluorescent intensity (MFI) values of (A) growth factors and (B) immune mediators secreted in CM from all three MSC sources (n = 3 donors per source). Comparison of gene ratio frequency (number of genes in our screen versus all background genes associated with that pathway) between significant (FDR < 0.05) angiogenic pathways (i.e., canonical and immune modulatory) enriched in (C) UC-MSCs and (D) adult-MSCs.

Figure 6

Comparison of immunomodulatory factors in the secretome of UC-MSCs and adult-MSCs (A and B) PCA and heatmap of MFI values of (A) anti-inflammatory factors; and (B) immunosuppressive factors, secreted in culture media (CM) from all three MSC sources (n = 3 donors per source). (C) Protein-protein interaction maps (PPI score >1.0 × 10⁻¹²) of 24 immunomodulatory factors were identified using the String database and ordered using the number of edges associated with each gene (degree sorted) in Cytoscape for AD-MSCs, BM-MSCs, and UC-MSCs. Genes with significant MFI values are color coded based on their anti-inflammatory (red), immunosuppressive (yellow), and anti-inflammatory/immunosuppressive dual roles (orange). Gray represents no change in MFI.

Figure 7

Differences in immune and angiogenic response between UC-MSCs and adult-MSCs (A) Representative flow cytometry figures showing classical (CD11b+CD16⁻) and non-classical (CD11b+CD16+) monocytes populations in RAW 264.7 macrophages under resting (M(0)) and activated (M(Φ)) states (INFγ (150 ng/mL) + TNFα (50 ng/mL)) and following treatment with different MSC sources. (B) Changes in population of monocytes (CD11b+CD16^+/−) (top) and Macrophages (M1, CD86⁺; M2, CD206+) (bottom) following activation with INFγ (150 ng/mL) + TNFα (50 ng/mL) and treatment with different MSC sources. (C) Changes in expression of CD202b (top) and CD31(PECAM1) (bottom) in MS-1 endothelial cells following activation with INFγ (150 ng/mL) + TNFα (50 ng/mL) and treatment with different MSC sources. (D) Changes in Caspase-3/7 activity (top) and apoptosis (bottom) for HK-2 cellular injury models using LPS or hypoxia (ischemia)/normoxia (reperfusion). For all experiments, three different donors (n = 3) for each MSC source were used as replicates. Data represents mean ± SEM. Statistical analysis was computed using one-way anova or Fisher's t-test; p∗ < 0.05, p∗∗ < 0.01 and p∗∗∗ < 0.001.”