About SeneQuest

Why?

Information on gene-to-senescence associations is crucial for increasing our understanding of the molecular mechanisms driving senescence. Currently this information seems to be scattered in small databases as well as in the literature. We believe that the creation of a central data hub gathering all the available information, is of paramount importance and will greatly facilitate information retrieval as well as the formation of more robust evidence driven research hypotheses. This was the reason for the creation of Senequest, a literature-based evidence database of genes related to senescence. This data resource is made freely available to the scientific community through http://senequest.net/.

How?

The Quertle Artifical Intelligence AI-powered literature search engine [https://quertle.com/] was used to identify publications that contained genes that are up- or down-regulated in senescence. Quertle uses AI to recognize various entities such as genes, proteins, diseases, etc and the relationships among them for unstructured data sources such as the biomedical literature. Given that the AI has yet to match the human capability for information retrieval from such unstructured and diverse sources as the biomedical literature, it is expected that the identification of all relevant literature through such tools will be to some extent incomplete. Apart from Quertle a second list was created out of Entrez GeneRIF entries [https://www.ncbi.nlm.nih.gov/gene/about-generif] that included the keyword "senescence". Approximately 3300 publications containing at least one link between the expression status of the factor under investigation (gene-protein or miRNA) and senescence were extracted and manually curated. In this list all high-throughput studies available during the preparation of this manuscript were analyzed. Additional information regarding cell-lines applied, tissue types and disease status mentioned in the publication were included. If more than one gene were found related to senescenceper examined publication, this information was also extracted and utilized. 13,015such associations (up- or down-regulation) with senescence were identified. Filtering for unique relations resulted in 7,969 genes encoding proteins and miRNAs. Each gene in the database is connected with multiple literature evidence, which is displayedin the form of PubMed IDs, describing the relationship of the gene’s expression status(up- or down-regulation)with senescence.Interactions of genes are also stored in the database and the user can search for interactants of a specific gene that are also connected with senescence. Additionaly, Gene Ontology (GO) codes are associated with each gene. SeneQuest provides the ability for the user to search for senescence-associated genes that are linked to a specific GO-term or any of its descendants. All evidence is linked to one or multiple PubMed IDs that the user can immediately view by selecting the corresponding links. Finally cell-line and tissue type information are also stored in the database for each gene-to-senescence association and is searchable through the the user interface. Given its dynamic nature, SeneQuest will be continuously updatedby members of the International Cell Senescence Associationand new information from additional databases (such as http://www.mirbase.org/) will be constantly incorporated. Database updates will be made public twice a year.

Prediction of key regulated genes associated with senescence and heart disease

In the current update of SeneQuest, we have included a new section allowing us to correlate senescent cells with various diseases. The first clinical context in which the effect of senescence is addressed are cardiovascular diseases such as coronary, atherosclerotic or dilated cardiomyopathy. In the following updates, similar sections pertaining to additional disease types will be included. The available body of evidence on the impact of senescence on heart tissue has suggested diverse outcomes with regards to disease, which is mainly attributed to the different cell types affected in the various CVD types. As the effects of senescence on cardiac homeostasis seem to be cell-type specific, it is currently challenging to determine whether depletion of senescent cells in CVDs may yield beneficial or detrimental phenotypes. Following a machine learning computational approach (Figure 1), we predicted key regulatory genes associated with both senescence and heart disease in different cardiac disease settings. To that end, we analyzed a large repertoire of publically available or published datasets ranging from bulk RNA-seq, ribosome profiling, single cell RNA-seq (scRNA-seq) to singe cell ATAC-seq (scATAC-seq) in human and mouse cells (1;2;3;4;5). We initially sought to estimate the relative RNA abundance of all human and mouse genes, based on bulk and single cell RNA data (1;2;3;4;5) (Figure 1). In addition, from the single cell data we determined which genes can be used as markers of the different cardiac cell subtypes, which was accomplished via the use of Monocle and Seurat packages (6;7). To additionally encompass data regarding the regulation processes upon heart failure, we also included open chromatin regions at the single cell level from single cell ATAC-seq data (2;4) as well as translation efficiency changes from ribosome profiling data from the study of van Heesch et al. (5). Hence, for each gene we identified the cell types where it is predominantly expressed (8) and generated a score (hereafter named regulation score) representing its expression change in the failing heart versus normal cardiac tissue; we, then, proceeded in raking genes based on that score (Table 1). To derive an accurate correlation between senescence and heart disease we obtained the top 75% of ranked genes (~500 genes) based on the regulation score and used the machine learning Endeavour (9), Cytoscape (10) and ConsensusPathdb (11) platforms to associate them with senescence and human heart diseases (cardiomyopathy, myocarditis and ischemic infarction) (12) (Figure 2). The prioritized set of genes associated with senescence and heart failure was projected back to the different cell types from the scRNA-seq or bulk RNA-seq analyses and their abundance was estimated in each cellular subtype (Figure 3). As the majority of available data pertain to the left atrium and ventricle, the current analysis was focused on those areas. Furthermore, we analyzed scRNA-seq data from neonatal mouse hearts subjected to myocardial infarction (MI), in order to identify potential differences in senescence-associated gene expression due to regeneration processes. For that purpose we used data from neonatal mice whose heart tissue maintains regenerative potential one day post MI (P1) versus neonatal mice whose heart has lost regenerative capacity 8 days post MI (P8) (2). The differences observed between P1 versus P8 regarding senescent-associated genes are correlated with several metabolic pathway alterations (Figure 4), which is in line with previous interpretations of the particular dataset suggesting that the differential activation of metabolic pathways in senescent cells may have a diverse impact on the recovery processes (2). Figure 5 clearly shows that a wide range of predicted senescence-associated markers in cardiomyopathy display different expression profiles when comparing the P1 and P8 cohorts. Overall, the aim of this analysis was to implement a machine learning approach to identify a potentially differing expression pattern of predicted senescence-related regulatory genes among the various cardiac cell populations, upon conditions leading to heart failure versus normal heart tissue. The results of the analysis demonstrated that key senescence-related genes may alter their expression in cardiac subtypes depending on the cell type and the disease state (Figures 1-3 and Table 1). Furthermore, besides cardiac disease per se senescence-related genes may alter their expression in response to heart regeneration cues (Figures 4, 5 and Table 1). Our analysis comprises a senescence-related gene mapping attempt in the failing heart; the rapid accumulation of “-omics” data in the field of CVDs and senescence is expected to further increase the accuracy of predictions carried out via machine learning approaches, thus culminating in the development of better targeted therapeutic strategies. The above results (Figures and Table) can be found on the following link: https://github.com/VGlabUOA/prediction_senescence_CVD

References:
[1] Wang, L. I., Yu, P., Zhou, B., Song, J., Li, Z., Zhang, M., ... & Hu, S. (2020). Single-cell reconstruction of the adult human heart during heart failure and recovery reveals the cellular landscape underlying cardiac function. Nature cell biology, 22(1), 108-119
[2] Wang, Z., Cui, M., Shah, A. M., Tan, W., Liu, N., Bassel-Duby, R., & Olson, E. N. (2020). Cell-type-specific gene regulatory networks underlying murine neonatal heart regeneration at single-cell resolution. Cell reports, 33(10), 108472.
[3] Litviňuková, M., Talavera-López, C., Maatz, H., Reichart, D., Worth, C. L., Lindberg, E. L., ... & Teichmann, S. A. (2020). Cells of the adult human heart. Nature, 588(7838), 466-472.
[4] Ruiz-Villalba, A., Romero, J. P., Hernandez, S. C., Vilas-Zornoza, A., Fortelny, N., Castro-Labrador, L., ... & Prósper, F. (2020). Single-cell RNA sequencing analysis reveals a crucial role for CTHRC1 (Collagen Triple Helix Repeat Containing 1) cardiac fibroblasts after myocardial infarction. Circulation, 142(19), 1831-1847.
[5] van Heesch, S., Witte, F., Schneider-Lunitz, V., Schulz, J. F., Adami, E., Faber, A. B., ... & Hubner, N. (2019). The translational landscape of the human heart. Cell, 178(1), 242-260.
[6] Trapnell, C., Cacchiarelli, D., Grimsby, J., Pokharel, P., Li, S., Morse, M., ... & Rinn, J. L. (2014). The dynamics and regulators of cell fate decisions are revealed by pseudotemporal ordering of single cells. Nature biotechnology, 32(4), 381-386.
[7] Hao, Y., Hao, S., Andersen-Nissen, E., Mauck III, W. M., Zheng, S., Butler, A., ... & Satija, R. (2021). Integrated analysis of multimodal single-cell data. Cell.
[8] Cao Y, Wang X and Peng G (2020) SCSA: A Cell Type Annotation Tool for Single-Cell RNA-seq Data. Front. Genet. 11:490. doi: https://doi.org/10.3389/fgene.2020.00490.
[9] Tranchevent, L. C., Ardeshirdavani, A., ElShal, S., Alcaide, D., Aerts, J., Auboeuf, D., & Moreau, Y. (2016). Candidate gene prioritization with Endeavour. Nucleic acids research, 44(W1), W117-W121.
[10] https://cytoscape.org/
[11] http://cpdb.molgen.mpg.de/
[12] http://www.informatics.jax.org/phenotypes.shtml

Senescent immune cells within the tumor microenvironment

In order to examine whether immune cell senescence occurs within the tumor microenvironment an extensive in silico tracking of senescent-like immune cells in human NSCLCs has been performed. Particularly, using as a source a number of investigations (for further information please refer to https://github.com/VGlabUOA/DISC), we have analyzed four gene sets (associated with Replicative Senescence (RS), Ras-Induced- Senescence (RIS), Oncogene-Induced-Senescence (OIS) and Oxidative Stress (OS) induced Senescence) to identify a common senescence gene signature. This signature was subsequently applied onto a scRNA-seq (Kim et al, 2020) and a TCR-seq (Guo et al, 2018) analysis to track senescent-like immune cells.
All analysis can be seen using the following link: https://github.com/VGlabUOA/DISC

The site will be updated twice a year by members of the International Cell Senescence Association
-ICSA consortium