Comparative Transcriptomics Identifies Potential Stemness-Related Markers for Mesenchymal Stromal/Stem Cells

doi:10.21203/rs.3.rs-961875/v1

Mesenchymal stromal/stem cells (MSCs) are multipotent cells residing in multiple tissues with the capacity for self-renewal and differentiation into various cell types. These properties make them promising candidates for regenerative therapies. MSC identification is critical in yielding pure populations for successful therapeutic applications; however, the criteria for MSC identification proposed by the International Society for Cellular Therapy (ISCT) is inconsistent across different tissue sources. This study aimed to identify potential markers to be used together with the ISCT criteria to provide a more accurate means of MSC identification. Thus, we carried out a comparative analysis of the expression of human and mouse MSCs derived from multiple tissues to identify the common differentially expressed genes. We show that six members of the proteasome degradation system are similarly expressed across MSCs derived from bone marrow, adipose tissue, amnion, and umbilical cord. Also, with the help of predictive models, we found that these genes successfully identified MSCs across all the tissue sources tested. Moreover, using genetic interaction networks, we showed a possible link between these genes and antioxidant enzymes in the MSC antioxidant defense system, thereby pointing to their potential role in prolonging the life span of MSCs. Our results suggest these genes can be used as stemness-related markers.

Scientific Communication

Biotechnology and Bioengineering

Comparative Transcriptomics

Potential Stemness-Related Markers

Mesenchymal Stromal/Stem Cells

Mesenchymal stem cells (MSCs) are multipotent adult stem cells that can be isolated from a variety of tissues such as bone marrow (BM)¹, adipose tissue (AT)², amnion (AM)³, umbilical cord (UC)⁴, and many more. Due to the myriad sources of MSCs, the International Society for Cellular Therapy (ISCT) proposed minimum criteria by which these MSCs can be identified. These criteria include 1) plastic-adherence of cells in vitro, 2) expression of specific cell surface markers (CD105, CD90, and CD73), and lack of expression of others (CD45, CD14, CD19, CD34, CD11b, CD79alpha, and HLA-DR), and 3) ability to differentiate into osteoblasts, chondroblasts, and adipocytes in vitro⁵. Unfortunately, growing evidence shows that these criteria are not consistent across different tissues and species since they define only general functional and morphological characteristics⁶. As a result, scientists have resorted to using additional “stemness” or “stemness-related” genes as markers to aid in the correct identification of MSCs⁷. Proper identification of MSCs is crucial to producing pure populations, thereby increasing their use in regenerative therapies. Thus, MSCs is an attractive tool for regenerative therapies, as their ease of isolation and ability to differentiate into multiple lineages make them ideal candidates for this purpose.

MSCs play a critical role in tissue maintenance, regeneration, and homeostasis in vivo⁸. Generally, MSCs remain quiescent, relying on glycolysis to produce energy for their metabolic needs⁹; however, upon tissue injury or loss, MSCs are activated to regenerate the damaged tissue and exit quiescence in favor of a more proliferative state. This highly proliferative state must maintain the balance between replenishing downstream lineages and replenishing the stem cell pool. As they begin to proliferate, energy demands increase, and glycolysis shifts to oxidative phosphorylation; this shift is accompanied by an increase in the production of reactive oxygen species (ROS)¹⁰. Oxidative phosphorylation is, indeed, a much more efficient means of generating ATP than glycolysis and can produce up to fifteen times more ATP. However, this is a double-edged sword since excess ROS can impair self-renewal and proliferation of MSCs^11,12.

For this reason, MSCs have an active antioxidant defense system. It has been demonstrated that MSCs constitutively express high levels of antioxidant enzymes, such as superoxide dismutases, catalases, and glutathione peroxidases¹³. These enzymes repair oxidatively damaged proteins, but some become oxidatively modified or damaged irreversibly. The cell has systems that recognize and remove these irreversibly damaged proteins and consequently prevent their buildup¹⁴. One of these systems is the proteasome degradation system, which plays a vital role in the degradation of oxidized and damaged proteins, preventing their accumulation and subsequent cellular dysfunction¹⁵.

The proteasome 26S is a multicatalytic degradation complex composed of a core particle (the 20S) and one or two regulatory particles (19S). The 20S core comprises four rings, two of which are composed of seven alpha subunits, while the other two rings are composed of seven beta subunits. The 19S regulator is comprised of a base (containing six ATPase and two non-ATPase subunits) and a lid (containing up to 10 non-ATPase subunits)¹⁶. The proteasome’s primary function in the cell is to degrade unneeded or damaged proteins by proteolysis; this can be carried out in either a ubiquitin-dependent manner through the 26S pathway or a ubiquitin-independent way through the 20S pathway. Recently, the proteasome has gained a lot of attention, and it has been shown to play an essential role in preserving the self-renewal and stemness of human MSCs. Kapetanou and colleagues showed that senescence and loss of stemness in human MSCs are accompanied by a sharp decline in proteasome content and activity¹⁷. Furthermore, they showed that the expression of some proteasome subunits is possibly affected by pluripotency factors. Taken together, these observations support their hypothesis of a relationship between proteostasis and stem cell function, where proteostasis is critical in maintaining proper protein levels, leading to efficient pluripotency and self-renewal maintenance.

In this study, we carried out a comparative analysis of RNA-Seq data of MSCs derived from different tissues of origin (BM, AT, UC, AM) and various species (human, mouse) to yield a list of common differentially expressed genes (DEGs). We provide evidence that members of the proteasome show similar patterns of expression across all MSC samples. Furthermore, we offer a possible relationship between the proteasome and antioxidant enzymes in protecting MSCs from oxidative stress, highlighting their importance in MSC survival. Finally, we demonstrate that six proteasomal degradation systems can be used as supplementary stemness-related markers for MSC identification through predictive models.

RNA-seq analysis

We constructed clustering maps using the t-distributed stochastic neighbor embedding (t-SNE) statistical method to visualize the transcriptomic similarities and differences between the samples. The clustering maps showed all human MSC samples clustering together in a distinct cluster apart from tissue-specific cell (TSCs) samples (Fig. 1a). Likewise, the mouse MSC samples clustered together, while TSC samples clustered independently (Fig. 1b). Next, we compared the gene expression of human-derived MSCs against their tissue-specific counterparts and identified 20,973, 24,365, 8,296, and 29,197 DEGs for h_AM_MSCs, h_BM_MSCs, h_AT_MSCs, and h_UC_MSCs, respectively. We constructed a Venn diagram to visualize the common DEGs shared by the MSCs derived from the four different tissue sources. The common DEGs made up 4.6% of the examined DEGs, equivalent to 2,181 common DEGs (Fig. 2a). Similarly, for the mouse datasets, we compared the gene expression of mouse-derived MSCs with their tissue-specific counterparts and identified 14,843 DEGs for m_AT_MSCs and 12,783 DEGs for the m_BM_MSCs. We found that the common DEGs made up 78.8% of the examined DEGs, equivalent to 12,178 common DEGs (Fig. 2b). Finally, we compared both lists of common DEGs to determine if the human and mouse MSCs shared any common DEGs. The Venn diagram showed that the two species had 1,583 (13.3%) DEGs in common (Fig. 2c). Interestingly, the heatmap of the 1,583 common DEGs showed three main clusters: a cluster that included all the human MSC samples, another cluster that included all the mouse-derived MSC samples plus m_BM_TSCs, and, finally, the last cluster included the rest of the TSC samples. Each of these clusters included subclusters that grouped the triplicate of each tissue type (Fig. 3).

DEGs ontology and enrichment analysis

In the gene ontology analysis, We produced a list of 157 enriched gene ontology (GO) terms with a threshold P-value of 10⁻³. ReViGO generated a scatter plot of the enriched GO terms organized according to their significance (p-value) and uniqueness. Following visualization, we identified a unique GO term (GO:2000736) to regulate stem cell differentiation with a significant p-value of 8.69E-4, a q value of 4.64E-2, and an enrichment score of 1.48. Moreover, this GO term had a frequency of 0.010%, a log10 p-value of –3.0610, and a uniqueness score of 0.70 (Fig. 4). Other GO terms were more general, less unique, and not explicitly specific to stem cell function. The GO term GO:2000736 included 23 genes that belonged to the proteasomal degradation pathway (Fig. 5) (Table1).

Table 1

List of DEGs belonging to the proteasomal degradation pathway identified in the GO enrichment analysis.
HUGO ID	Systemic ID	Gene
PSMA1	α6	proteasome (prosome, macropain) subunit, alpha type, 1
PSMA5	α1	proteasome (prosome, macropain) subunit, alpha type, 5
PSMA7	α4	proteasome (prosome, macropain) subunit, alpha type, 7
PSMB1	ß6	proteasome (prosome, macropain) subunit, beta type, 1
PSMB2	ß4	proteasome (prosome, macropain) subunit, beta type, 2
PSMB3	ß3	proteasome (prosome, macropain) subunit, beta type, 3
PSMB4	ß7	proteasome (prosome, macropain) subunit, beta type, 4
PSMB5	ß5	proteasome (prosome, macropain) subunit, beta type, 5
PSMB6	ß1	proteasome (prosome, macropain) subunit, beta type, 6
PSMB7	ß2	proteasome (prosome, macropain) subunit, beta type, 7
PSMC1	Rpt2	proteasome (prosome, macropain) 26s subunit, atpase, 1
PSMC2	Rpt1	proteasome (prosome, macropain) 26s subunit, atpase, 2
PSMC4	Rpt3	proteasome (prosome, macropain) 26s subunit, atpase, 4
PSMC5	Rpt6	proteasome (prosome, macropain) 26s subunit, atpase, 5
PSMD1	Rpn2	proteasome (prosome, macropain) 26s subunit, non-atpase, 1
PSMD2	Rpn1	proteasome (prosome, macropain) 26s subunit, non-atpase, 2
PSMD3	Rpn3	proteasome (prosome, macropain) 26s subunit, non-atpase, 3
PSMD4	Rpn10	proteasome (prosome, macropain) 26s subunit, non-atpase, 4
PSMD5	Rpn4	proteasome (prosome, macropain) 26s subunit, non-atpase, 5
PSMD7	Rpn8	proteasome (prosome, macropain) 26s subunit, non-atpase, 7
PSMD8	Rpn12	proteasome (prosome, macropain) 26s subunit, non-atpase, 8
PSMD13	Rpn9	proteasome (prosome, macropain) 26s subunit, non-atpase, 13
PSMD14	Rpn11	proteasome (prosome, macropain) 26s subunit, non-atpase, 14

Gene expression analysis

Now that our attention was drawn to these 23 genes, we wanted to take a closer look at their behavior. To begin with, we inspected their expression patterns in both species. We found that the 23 genes appeared to be upregulated in both the human and mouse MSC samples, except for PSMD4, which was downregulated in the mouse MSC samples compared to the TSC samples (Fig. 6a and b). PSMD4 had a P-value of 6.27E-03 and a fold change of 0.136934557 in m_AT_MSCS, while in m_BM_MSCs, it had a P-value of 9.79E-03 and a fold change of –0.224675507. Since most genes were upregulated, we attempted to explore further the interplay between these genes and other systems that assist the proteasome in antioxidant defense, namely, antioxidant enzymes.

Consequently, members of the superoxide dismutase, glutathione peroxidase, peroxiredoxin, thioredoxin, and peroxidasin families were cross-referenced against the 1,583 common DEGs identified earlier. We found SOD3, GPX7, GPX8, PRDX2, PRDX4, TXN2, and PXDN werepresent in our list of common DEGs. We found that the majority of the transcripts of these genes was upregulated but not all. Out of the seven genes, four (GPX7, GPX8, PXDN, TXN2) were upregulated in all analyzed MSC samples. Each of the other three (SOD3, PRDX2, and PRDX4) was upregulated in all the MSC samples except human AT-MSCs and BM-MSCs and mouse BM-MSCs, respectively (Fig. 6c and d).

Gene interaction network

Next, we wanted to shed more light on the interaction between these antioxidant enzymes and the 23 members of the proteasomal degradation system identified earlier. We employed the help of Cytoscape and the Genemania database to understand the interplay between these antioxidant enzymes and the proteasomal genes. A gene interaction network was generated, and it showed all 23 genes of the proteasomal degradation pathway to be co-expressed together and co-expressed with the antioxidant genes. Specifically, it showed PSMA7 to be co-expressed with GPX7, which in turn is co-expressed with GPX8, SOD3, and PXDN. Also, it showed that PRDX2 is co-expressed with PSMA7, PSMB3, PSMB6, PSMB7, PSMC4, PSMD3, and PSMD8. Furthermore, TXN2 was co-expressed with PSMA1, PSMA7, PSMB3, and PSMB6. Finally, PRDX4 was co-expressed with PSMA5, PSMB1, PSMB2, PSMB5, PSMB6, PSMC1, PSMC2, PSMC5, PSMD1, PSMD8, and PSMD14 (Fig. 7).

Predictive model

Finally, we wanted to test the genes’ efficiency in predicting the identity of MSCs across the different tissue sources in both human and mouse species. To test this, AutoWEKAClassifier performed 486 evaluations of available classifiers and found Random Forest to be the best classifier with the best error rate. The Random Forest tree classifier was used to train the model with 10-fold cross-validation, and the trained model was finalized. The final model was loaded to test its performance in predicting stem cell type on the testing data. We used the upregulated proteasomal genes as attributes, and we removed PSMD4 from the list since it had inconsistent expression across both species. We proceeded with the other 22 genes and ran the Random Forest model. The model tested the data 40 times and showed that MSCs were correctly classified in all 40 instances. To test if all 22 genes contributed equally to the classification process, we ran the gain ratio attribute selection evaluator in WEKA. We found six genes to be the top contributors in the classification: PSMB5, PSMB1, PSMD14, PSMC4, PSMA1, and PSMD8 (Supplementary Table 3). We repeated the Random Forest model using these six genes and these six genes were enough to correctly classify the MSCs all 40 instances (Supplementary Table 4).

Ever since the discovery of MSCs by Friedenstein et al.¹, researchers have debated their identity; however, the criteria proposed by the ISCT still fail to describe MSCs adequately and shows discrepancies across species and tissues of origin. As the focus shifted to stemness and stemness-related gene expression to aid in identifying MSCs, the search for adequate markers has intensified. Here, we show that members of the proteasome degradation system can be used as potential stemness-related markers to identify MSCs.

In this study, we integrated the RNA-seq data of MSCs derived from four different human tissues (AM, BM, AT, and UC) and two different mouse tissues (AT, BM). Differential expression analysis presented us with a list of 1,583 DEGs common to MSCs and TSCs across all tissue types in human and mouse. Further gene ontology enrichment analysis categorized these genes into GO terms, one of which was the GO term for regulating stem cell differentiation. This GO term was chosen for its uniqueness, high significance, and its specificity to stem cell processes. It included 23 members of the proteasome degradation system. Compelling evidence suggests a pivotal role for the proteasome in maintaining the pluripotency of mouse and human embryonic stem cells by supporting the clean-up of proteins oxidatively damaged during differentiation¹⁸.

The proteasome is an essential component of protein quality-control systems and plays a critical role in cellular homeostasis. It is involved in the degradation of abnormal, oxidized, or otherwise damaged proteins¹⁹. The accumulation of oxidized proteins in cells leads to their decreased life span²⁰. Moreover, it has been demonstrated that dysfunction of the proteasome is heavily implicated in cell ageing²¹. A recent study revealed that impairment of proteasome function resulted in an accumulation of oxidatively modified proteins in senescent Wharton’s jelly (WJ) and adipose-derived human adult mesenchymal stromal/stem cells. More importantly, this study showed that these cells’ senescence is accompanied by a decline in proteasome content and activities, coupled with the concurrent loss of their stemness¹⁷. Although the degradation of oxidized proteins can occur by ubiquitin-dependent (26S-proteasome) and ubiquitin-independent (20S-proteasome) mechanisms^22–24, various studies have shown that the 20S proteasome might be the major machinery involved in this process^25,26. Here, we show that members of the 20S proteasome (PSMA1, PSMA5, and PSMA7) of the alpha subunits and all members of the beta subunit (PSMB1, PSMB2, PSMB3, PSMB4, PSMB5, PSMB6, and PSMB7) are not only differentially expressed in MSCs but are also upregulated.

Furthermore, we show that members of the 19S proteasome base (PSMC1, PSMC2, PSMC4, PSMC5, PSMD1, and PSMD2) and lid (PSMD3, PSMD5, PSMD7, PSMD8, PSMD13, and PSMD14) are also differentially expressed and upregulated in MSCs in comparison with TSCs. However, our results also showed that PSMD4 expression in mouse MSC samples was downregulated. PSMD4’s central role in the 19S lid is to recognize polyubiquitinated protein substrates and detach the ubiquitin molecules from them for their subsequent degradation through the 26S proteasomal pathway²⁷. PSMD4 is not the only ubiquitin receptor in the 19S lid. PSMD2 is another ubiquitin receptor that recognizes and binds both ubiquitin and ubiquitin-like proteins²⁸. We found PSMD2 to be upregulated in our mouse MSCs. It could be that mouse MSCs rely mainly on PSMD2 to recognize polyubiquitinated protein substrates, thereby rendering PSMD4 dispensable.

Studies have shown that during MSC proliferation, ROS are produced as byproducts of oxidative metabolism. However, increased ROS levels may lead to a decrease in cell survival and have also been implicated in cell senescence¹¹. To counteract these detrimental effects, the cell has antioxidant defense systems activated by high ROS levels. Increased ROS concentration causes Nrf2 (a stress-responsive transcription factor) to dissociate from its inhibitory complex with Keap1. This enables Nrf2 to accumulate and translocate to the nucleus, where it binds to antioxidant-response elements (ARE), thus promoting the expression of several antioxidant²⁹ and proteasomal genes³⁰. These antioxidant enzymes and the proteasome degradation machinery work together as a defense against damaging high ROS levels. We demonstrated that members of the proteasome degradation machinery were upregulated across the MSCs samples tested. We also demonstrated that GPX7, GPX8, TXN2, and PXDN antioxidant genes were differentially expressed and upregulated in MSCs compared to TSCs. Additionally, we provided evidence that these genes are co-expressed with the proteasome degradation machinery members by data mining gene interaction databases. Taken together, this points to the efficiency of MSCs in counteracting oxidative stress, in which the proteasome is integral.

Finally, to show the competence of these proteasome genes in identifying MSCs, we employed the aid of predictive models. Predictive models have been used robustly to identify a general MSC phenotype that could distinguish MSCs from other cell types. A recent study showed that gene expression levels in prediction models increase the classification accuracy of the combined set of traditional MSC cell surface markers³¹. Using the Random Forest model, we showed that the expression of six proteasome genes could accurately distinguish MSCs from their tissue-specific counterparts. Of these six genes, PSMB5, PSMB1, and PSMD14 have been linked to stem cell function. As previously mentioned, PSMB1 and PSMB5 are catalytic subunits of the 20S proteasome, and reducing their expression leads to a decrease in cell proliferation and an increase in replicative senescence in hBMSCs³². Likewise, Kapetanou and colleagues reported a similar decline in the expression of these two genes in senescent WJ-MSCs. They also show that PSMB5 overexpression rescues these senescent cells from age-related reduction in proteasome expression and function, improving their stemness and extending their lifespan¹⁷. Finally, PSMD14 is essential for proper 26S assembly³³; it also plays a role in cleaving polyubiquitin chains at a proximal site and recycling ubiquitin chains³⁴. PSMD14 is a crucial regulator of stem cell maintenance. A reduction in its levels leads to a marked decrease in Oct4 protein expression, accompanied by abnormal morphology in embryonic stem cells³⁵. However, no data currently exists on the three remaining genes (PSMC4, PSMA1, and PSMD8) that link them to any stem cell function.

Our study carried out a comprehensive comparative analysis of MSCs RNA-seq data across two species and six different tissue types to ascertain potential identity markers. Our results showed that six members of the proteasomal machinery are promising candidates in identifying MSCs. Moreover, we shine a light on their association with antioxidant enzymes in defending MSCs against high ROS levels, thereby maintaining their proliferation and self-renewal. These six genes can be used as additional stemness-related markers to refine and enhance the accuracy of MSC identification, which is a critical step in ensuring the yield of a pure population for consequent applications in regenerative therapies.

RNA-seq datasets and processing

RNA-seq datasets were obtained from the Gene Expression Omnibus (GEO) ³⁶ and Array Express databases³⁷. We collected transcriptomic data for human MSCs derived from umbilical cord (h_UC_MSCs), amnion (h_AM_MSCs), bone marrow (h_BM_MSCs), and adipose tissue (h_AT_MSCs) and their tissue-specific counterparts (h_UC_TSCs, h_AM_TSCs, h_BM_TSCs, and h_AT_TSCs). Mouse transcriptomic data included bone marrow-derived MSCs (m_BM_MSCs) and adipose tissue-derived MSCs (m_AT_MSCs), along with their tissue-specific counterparts (m_BM_TSCs and m_AT_TSCs) (Supplementary Table 1). We processed triplicates of each cell type except for h_AM_TSCs, h_AM_MSCs, and h_UC_MSCs, for which we managed to obtain quadruplicates, bringing our total number of samples to 27 human samples and 12 mouse samples (Supplementary Table 2). Data for all cell types were converted from the Sequence Read Archive (SRA) format into the FASTQ format using the SRA Toolkit version 2.10.8 for downstream analysis³⁸. Moreover, data were filtered; any read with a length less than 50 bp was excluded. Adapter sequences were detected and trimmed using fastp version 0.19.5³⁹.

RNA-seq data analysis to find DEGs

We used Kallisto version 0.46.1 for pseudo-alignment and the quantification of abundances of transcripts from the RNA-Seq data⁴⁰. Human data were pseudo-aligned to the human reference transcriptome GCA_000001405.15_GRCh38, while mouse data were pseudo-aligned to the mouse reference transcriptome GCA_000001635.9_GRCm39 provided by the Genome Reference Consortium⁴¹. Abundances produced by Kallisto were processed by Sleuth version 0.29.0 for the differential expression analysis⁴². Venn diagrams of the common DEGs were constructed using Venny version 2.0⁴³. Finally, the t-distributed stochastic neighbor-embedding clustering and heatmaps were generated using R 3.6.1 (www.r-project.org).

DEGs ontology and enrichment analysis

Biological processes encompassing the DEGs were identified based on GO enrichment analysis using the GOrilla database⁴⁴. The p-value threshold was set at 10e-3. Afterward, we visualized the enriched GO terms using ReViGO, and a scatter plot was produced showing the log10 p-value and log size of each GO term⁴⁵.

Generation of gene interaction network

To further investigate the interactions between the DEGs, we constructed a gene interaction network by mining interaction networks from the GEO, BioGRID, IRefIndex, and I2D using the GeneMANIA Cytoscape plugin⁴⁶. This step produced an annotated Cytoscape network of functional interactions between the DEGs.

Predictive model

Lastly, to assess the selected DEGs’ ability to identify MSCs, we built a predictive model using the Waikato Environment for Knowledge Analysis (WEKA) software version 3.8.4⁴⁷. The gene expression values were converted into the ARFF file format, where our genes of interest were used as attributes in the training dataset. We used the AutoWEKAClassifier package (https://github.com/automl/autoweka) to find the best classification model for our provided dataset automatically. WEKA was also used for attribute selection.

Author Contribution

Conceived the study: A.A. and M.G. Designed the study: A.A., A.M. and M.G. Collected the data: M.G. and A.E-H. Analyzed the data: M.G., A.E-H., A.M., and A.A. Wrote the paper: M.G., and A.E-H. Review and Editing the paper: A.M., and A.A.

Competing interest

The authors declare that they have no competing interests.

Friedenstein, A. J., Petrakova, K. V., Kurolesova, A. I. & Frolova, G. P. Heterotopic of bone marrow. Analysis of precursor cells for osteogenic and hematopoietic tissues., 6, 230–247 (1968).
Zuk, P. A. et al. Human adipose tissue is a source of multipotent stem cells. Molecular biology of the cell, 13, 4279–4295 (2002).
Alviano, F. et al. Term amniotic membrane is a high throughput source for multipotent mesenchymal stem cells with the ability to differentiate into endothelial cells in vitro. BMC Developmental Biology, 7, 11 (2007).
Erices, A., Conget, P. & Minguell, J. J. Mesenchymal progenitor cells in human umbilical cord blood. British journal of haematology, 109, 235–242 (2000).
Dominici, M. et al. Minimal criteria for defining multipotent mesenchymal stromal cells. The International Society for Cellular Therapy position statement., 8, 315–317 (2006).
Peister, A. et al. Adult stem cells from bone marrow (MSCs) isolated from different strains of inbred mice vary in surface epitopes, rates of proliferation, and differentiation potential., 103, 1662–1668 (2004).
Zhao, W., Li, Y. & Zhang, X. Stemness-Related Markers in Cancer. Cancer Transl Med, 3, 87–95 (2017).
Minguell, J. J., Erices, A. & Conget, P. Mesenchymal Stem Cells. Experimental Biology and Medicine, 226, 507–520 (2001).
Sart, S., Song, L. & Li, Y. Controlling Redox Status for Stem Cell Survival, Expansion, and Differentiation. Oxidative Medicine and Cellular Longevity 2015, 105135 (2015).
Liang, R. & Ghaffari, S. Stem cells, redox signaling, and stem cell aging. Antioxidants & redox signaling, 20, 1902–1916 (2014).
Ko, E., Lee, K. Y. & Hwang, D. S. Human umbilical cord blood-derived mesenchymal stem cells undergo cellular senescence in response to oxidative stress. Stem Cells Dev, 21, 1877–1886 (2012).
Choo, K. B. et al. Oxidative stress-induced premature senescence in Wharton's jelly-derived mesenchymal stem cells. Int J Med Sci, 11, 1201–1207 (2014).
Valle-Prieto, A. & Conget, P. A. Human mesenchymal stem cells efficiently manage oxidative stress. Stem Cells Dev, 19, 1885–1893 (2010).
Jung, T. & Grune, T. The proteasome and its role in the degradation of oxidized proteins. IUBMB Life, 60, 743–752 (2008).
Chondrogianni, N. et al. Protein damage, repair and proteolysis. Mol Aspects Med, 35, 1–71 (2014).
Tanaka, K. The proteasome: overview of structure and functions. Proc Jpn Acad Ser B Phys Biol Sci, 85, 12–36 (2009).
Kapetanou, M., Chondrogianni, N., Petrakis, S., Koliakos, G. & Gonos, E. S. Proteasome activation enhances stemness and lifespan of human mesenchymal stem cells. Free Radic Biol Med, 103, 226–235 (2017).
Schröter, F. & Adjaye, J. The proteasome complex and the maintenance of pluripotency: sustain the fate by mopping up? Stem Cell Res Ther, 5, 24–24 (2014).
Raynes, R., Pomatto, L. C. D. & Davies, K. J. A. Degradation of oxidized proteins by the proteasome: Distinguishing between the 20S, 26S, and immunoproteasome proteolytic pathways. Molecular Aspects of Medicine, 50, 41–55 (2016).
Reeg, S. & Grune, T. Protein Oxidation in Aging: Does It Play a Role in Aging Progression? Antioxid. Redox. Signal, 23, 239–255 (2014).
Chondrogianni, N. et al. Central Role of the Proteasome in Senescence and Survival of Human Fibroblasts: INDUCTION OF A SENESCENCE-LIKE PHENOTYPE UPON ITS INHIBITION AND RESISTANCE TO STRESS UPON ITS ACTIVATION. Journal of Biological Chemistry, 278, 28026–28037 (2003).
Goldberg, A. L. Protein degradation and protection against misfolded or damaged proteins. Nature, 426, 895–899 (2003).
Shang, F., Nowell, T. R. & Taylor, A. Removal of Oxidatively Damaged Proteins from Lens Cells by the Ubiquitin-Proteasome Pathway. Exp. Eye Res, 73, 229–238 (2001).
Bader, N., Jung, T. & Grune, T. The proteasome and its role in nuclear protein maintenance. Exp. Gerontol, 42, 864–870 (2007).
Silva, G. M. et al. Redox control of 20S proteasome gating. Antioxid Redox Signal, 16, 1183–1194 (2012).
Jung, T., Höhn, A. & Grune, T. The proteasome and the degradation of oxidized proteins: Part II – protein oxidation and proteasomal degradation. Redox Biol, 2, 99–104 (2014).
da Fonseca, P. C., He, J. & Morris, E. P. Molecular model of the human 26S proteasome. Molecular cell, 46, 54–66 (2012).
Chojnacki, M. et al. Polyubiquitin-Photoactivatable Crosslinking Reagents for Mapping Ubiquitin Interactome Identify Rpn1 as a Proteasome Ubiquitin-Associating Subunit. Cell chemical biology, 24, 443–457446 (2017).
Dai, X. et al. Nrf2: Redox and Metabolic Regulator of Stem Cell State and Function. Trends in Molecular Medicine, 26, 185–200 (2020).
Kwak, M. K., Wakabayashi, N., Greenlaw, J. L., Yamamoto, M. & Kensler, T. W. Antioxidants enhance mammalian proteasome expression through the Keap1-Nrf2 signaling pathway. Molecular and cellular biology, 23, 8786–8794 (2003).
Rohart, F. et al. A molecular classification of human mesenchymal stromal cells. PeerJ, 4, e1845 (2016).
Yu, H., Lai, H. J., Lin, T. W. & Lo, S. J. Autonomous and non-autonomous roles of DNase II during cell death in C. elegans embryos.Bioscience Reports35 (2015).
Quinet, G., Gonzalez-Santamarta, M., Louche, C. & Rodriguez, M. S. Mechanisms Regulating the UPS-ALS Crosstalk: The Role of Proteaphagy., 25, 2352 (2020).
Lu, L. et al. Potential role of 20S proteasome in maintaining stem cell integrity of human bone marrow stromal cells in prolonged culture expansion. Biochemical and Biophysical Research Communications, 422, 121–127 (2012).
Buckley, S. M. et al. Regulation of Pluripotency and Cellular Reprogramming by the Ubiquitin-Proteasome System. Cell stem cell, 11, 783–798 (2012).
Barrett, T. et al. NCBI GEO: archive for functional genomics data sets—update. Nucleic Acids Res, 41, D991–D995 (2012).
Athar, A. et al. ArrayExpress update – from bulk to single-cell expression data. Nucleic Acids Res, 47, D711–D715 (2018).
Leinonen, R., Sugawara, H. & Shumway, M. & Collaboration, o. b. o. t. I. N. S. D. The Sequence Read Archive. Nucleic Acids Res, 39, D19–D21 (2010).
Chen, S., Zhou, Y., Chen, Y. & Gu, J. fastp: an ultra-fast all-in-one FASTQ preprocessor., 34, i884–i890 (2018).
Bray, N. L., Pimentel, H., Melsted, P. & Pachter, L. Near-optimal probabilistic RNA-seq quantification. Nature Biotechnology, 34, 525–527 (2016).
O'Leary, N. A. et al. Reference sequence (RefSeq) database at NCBI: current status, taxonomic expansion, and functional annotation. Nucleic Acids Res, 44, D733–745 (2016).
Pimentel, H., Bray, N. L., Puente, S., Melsted, P. & Pachter, L. Differential analysis of RNA-seq incorporating quantification uncertainty. Nature methods, 14, 687–690 (2017).
Oliveros, J. C. VENNY. An interactive tool for comparing lists with Venn Diagrams(2007).
Eden, E., Navon, R., Steinfeld, I., Lipson, D. & Yakhini, Z. GOrilla: a tool for discovery and visualization of enriched GO terms in ranked gene lists. BMC Bioinformatics, 10, 48 (2009).
Supek, F., Bošnjak, M., Škunca, N. & Šmuc, T. R. E. V. I. G. O. Summarizes and Visualizes Long Lists of Gene Ontology Terms. PLOS ONE, 6, e21800 (2011).
Montojo, J. et al. GeneMANIA Cytoscape plugin: fast gene function predictions on the desktop., 26, 2927–2928 (2010).
Frank, E., Hall, M. A. & Witten, I. H. in Data Mining: Practical Machine Learning Tools and Techniques (ed Morgan Kaufmann, 2016).

No competing interests reported.

supplementarydata.docx

Comparative Transcriptomics Identifies Potential Stemness-Related Markers for Mesenchymal Stromal/Stem Cells

Status:

Version 1

Abstract

Figures

Introduction

Results

RNA-seq analysis

DEGs ontology and enrichment analysis

Gene expression analysis

Gene interaction network

Predictive model

Discussion

Materials And Methods

RNA-seq datasets and processing

RNA-seq data analysis to find DEGs

DEGs ontology and enrichment analysis

Generation of gene interaction network

Predictive model

Declarations

Author Contribution

Competing interest

References

Additional Declarations

Supplementary Files

Status:

Version 1