Metabolism as an early predictor of DPSCs aging

Abstract

Tissue resident adult stem cells are known to participate in tissue regeneration and repair that follows cell turnover, or injury. It has been well established that aging impedes the regeneration capabilities at the cellular level, but it is not clear if the different onset of stem cell aging between individuals can be predicted or prevented at an earlier stage. Here we studied the dental pulp stem cells (DPSCs), a population of adult stem cells that is known to participate in the repair of an injured tooth, and its properties can be affected by aging. The dental pulp from third molars of a diverse patient group were surgically extracted, generating cells that had a high percentage of mesenchymal stem cell markers CD29, CD44, CD146 and Stro1 and had the ability to differentiate into osteo/odontogenic and adipogenic lineages. Through RNA seq and qPCR analysis we identified homeobox protein, Barx1, as a marker for DPSCs. Furthermore, using high throughput transcriptomic and proteomic analysis we identified markers for DPSC populations with accelerated replicative senescence. In particular, we show that the transforming growth factor-beta (TGF-β) pathway and the cytoskeletal proteins are upregulated in rapid aging DPSCs, indicating a loss of stem cell characteristics and spontaneous initiation of terminal differentiation. Importantly, using metabolic flux analysis, we identified a metabolic signature for the rapid aging DPSCs, prior to manifestation of senescence phenotypes. This metabolic signature therefore can be used to predict the onset of replicative senescence. Hence, the present study identifies Barx1 as a DPSCs marker and dissects the first predictive metabolic signature for DPSCs aging.

Figure 1

Extraction summary and preliminary screening. (a) A simplified summary of the isolated DPSC lines and the initial project design to model DPSCs aging in vitro. (b) Preliminary screening of cell lines with cell surface markers, CD29 and CD146 using western blot shows differential expression of these markers. (c,d) Fluorescent cytometry of candidate cell lines (DPSC 29,44,43,45) demonstrates high percentage of mesenchymal markers; CD29+CD146+ cells (c) and CD44+Stro1+ cells (d). (e) A representative graph shows that the percentage of “double positive” were comparable with commercial DPSC and MSC cell lines from Lonza. (n = 3 per cell line). Graph error bars are the means ± standard error of the mean (SEM).

Figure 2

Difference in growth kinetics within the candidate cell lines. (a) Number of population doublings in 24 hours calculated in each consecutive passage indicates that two of the candidate cell lines demonstrated replicative senescence which is a classical hallmark of in vitro ageing (Cells were passaged to a density of 1:3 in each passage). (b) Quantification of average cell surface area in early and late passages of DPSC 44 and 45. Significance was determined by unpaired Student’s t-test; n = 70 cells per cell line were measured; ****p < 0.0001; Graph error bars are the means ± SEM (c) flow cytometric analysis of senescence-associated-β-gal activity in early and late passages of DPSC 44 and 45.

Figure 3

Osteo/odontogenic and Adipogenic potential of DPSCs. (a) Osteo/odontogenic cells were stained for extracellular calcifications with Alizarin red stain. Adipogenic cells were stained for neutral lipids with BODIPY and Oil Red O stain. (a) Spectrometric quantification of Alizarin stain normalized with cell numbers (DAPI staining) showing all the candidate cell lines were able to differentiate into osteoblasts/odontoblast at different levels except DPSC 43 which did not survive the osteogenic differentiation. (c) Spectrometric quantification of Oil Red O stain normalized with cell numbers (DAPI staining) showing all the cell lines were able to differentiate into adipocytes at different levels and are highly resembling the differentiation levels of commercial cell lines. (d–g) Treatment with TeSRmeta for 4 days did not have any impact on the differentiation potential of the candidate cell lines, except for DPSC 43 in which it enhanced the survivability and differentiation to osteoblasts/odontoblasts. (n = 3 per cell line). Graph error bars are the means ± SEM.

Figure 4

The gene expression of DPSCs compared to MSCs, Fibroblasts and hESCs. (a) Principal component analysis (PCA) of DPSCs, MSCs and hESCs showed that DPSCs and MSCs grouped together. (b)Principal component analysis (PCA) of DPSCs with skin fibroblasts and MSCs from literature showed that DPSCs have distinct expression profiles from foreskin fibroblasts. (c) Volcano plot of genes differentially expressed in MSCs/DPSCs vs. hESCs. (d) Volcano plot of genes differentially expressed in MSCs/DPSCs compared to fibroblasts. (e) Volcano plot of genes differentially expressed in DPSCs vs. MSCs.

Figure 5

BARX1 gene as a new specific marker for DPSCs. (a–c) Immunofluorescent staining of Barx1 in skin fibroblasts, adult DPSCs (DPSC 29), and deciduous DPSCs (DPSC 292) at low power magnification. (d–f) Showing the nuclear localization of Barx1 transcription factor at high power magnification. Exposure settings and laser intensity of the confocal microscope were adjusted and normalized for fibroblasts, and same settings were used for the DPSCs. (g) Quantitative PCR reveals absence of BARX1 gene in BM-MSCs and in skin fibroblasts. Fold change normalized to DPSC Lonza. Significance was determined by unpaired Student’s t-test; n = 3–6 per cell line; **p < 0.01; ****p < 0.0001; Graph error bars are the means ± SEM.

Figure 6

Metabolic assays showing the metabolic differences between the candidate DPSCs lines. (a,b) Mitostress Assay shows that DPSCs have higher mitochondrial activity than human embryonic stem cells (*p = 0.002). (c,d) From the selected candidate cell lines, DPSC 29 and 44 use palmitate inferring their ability to use fatty acid as an energy source. DPSC 43 and 45 have lower ability to use fatty acid as an energy source. (e,f) Glycolysis stress assay conducted on the selected candidates shows that DPSC 43 and 45 have significantly lower glycolytic capacity compared to DPSC 29, 44 and commercial DPSCs. Significance was determined by unpaired Student’s t-test; n = 3–12 per cell line; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001; Graph error bars are the means ± SEM.

Figure 7

Proteomic analysis within the candidate cell lines. (a) Proteomic workflow used to identify differentially regulated protein expression in young (early passage) versus old (late passage) DPSC in rapid aging (lines 43, 45) or slow aging (lines 29, 44) cells. (b) Fold changes of replicative senescence markers, CDKN2A (known as P16) and CDKN1A (known as P21) in late passages of rapid aging cell lines (43 & 45) and slow aging cell lines (29 & 44). (c,d) Proteomics GO terms enrichment for slow aging (29 & 44) and rapid dividing (43 & 45) cell lines. (e) String analysis of enriched muscle contraction proteins expressed in late passages of rapid aging cell lines. (f) Fold change of TGF-β pathway related proteins higher in rapid aging compared to slow aging at early passages (P value < 0.05) (left), and String analysis of TGF-β pathway proteins expressed in early passages (P3) rapid aging cells at higher level than slow aging early passages (right).

Figure 8

Activation of TGF-β pathway leads to formation of more actin stress fibers. (a) A simplified flowchart that shows the process of utilising VerSeDa secretome prediction utilities to analyse the predicted secretome in DPSC. (b) Signaling pathway impact analysis (SPIA) showing relatively activated or inhibited signaling pathways in rapid aging cell lines’ predicted secretome compared to slow aging cell lines. (False discovery rate < 0.1). (c,d) Activation of TGF-β pathway by adding Activin to the DPSCs growth media for 2-days increases actin stress fiber formation (d), compared to control (c). (e–h) Representative confocal images of DPSC 44 and DPSC 45 comparing the size and intensity of the actin fibers at passage 4, and passage16. (i) Quantification of stress fibers in early passage (P3) DPSC Lonza in control, and after treating with Activin for 2-days. Significance was determined by unpaired Student’s t-test; n = 100 cells were counted per condition; *p = 0.03; Graph error bars are the means ± SEM. (j) Quantification of cells with senescence phenotype at passage 16 in DPSC 44 and DPSC 45. Significance was determined by unpaired Student’s t-test; n = 100 cells were counted per condition; *p = 0.0005; Graph error bars are the means ± SEM. (k) Hypothetical model: DPSCs transcriptome and metabolic analysis suggest that low glycolysis and fatty acid oxidation (FAO) and upregulation of TGF-β activity at early passages are predictive for rapid aging phenotype in later passages.