PseudoCell: A Multivalued Logical Regulatory Network to Investigate Premature Senescence Dynamics and Heterogeneity

Abstract

Purpose: Premature cellular senescence is a pivotal process in aging and age-related diseases, triggered by various stressors. However, this is not a homogeneous phenotype, but a heterogeneous cellular state composed of multiple senescence programs with different compositions. Therefore, understanding the complex dynamics of senescence programs requires a systemic approach. We introduce PseudoCell, a multivalued logical regulatory network designed to explore the molecular intricacies of premature senescence.

Methods: PseudoCell integrates key senescence signaling pathways and molecular mechanisms, offering a versatile platform for investigating diverse premature senescence programs initiated by different stimuli.

Results: Validation through simulation of classical senescence programs, including oxidative stress-induced senescence and oncogene-induced senescence, demonstrates its ability to replicate molecular signatures consistent with empirical data. Additionally, we explore the role of CCL11, a novel senescence-associated molecule, through simulations that reveal potential pathways and mechanisms underlying CCL11-mediated senescence induction.

Conclusions: In conclusion, PseudoCell provides a systematic approach to dissecting premature senescence programs and uncovering novel regulatory mechanisms.

Keywords: aging; bioinformatics; cellular senescence; in silico modeling; premature senescence.

FIGURE 1

PseudoCell interface architecture. The interface is organized into three modules: User Interface, Core Simulator, and Importing/Exporting Module. The User Interface module is responsible for handling the user's experimental conditions. The Core Simulator module is responsible for carrying out the network update iterations based on the logic rules described. The Import/Export Module handles the functions for sharing the networks created by the community. The interface returns a matrix of activation frequency values as output, where the lines represent the samples (S1, S2….Sn) and the columns (n1, n2…nn) the network nodes.

FIGURE 2

In silico stimulation of OSIS, OIS Ras and OIS AKT1‐dependent leads to activation of molecular elements similar to those expected in biological models. (A) Heatmap demonstrating the systemic response of the network to 50% frequency stimulus of the CYBB node. In the graph, the lines represent the network nodes, while the columns represent the samples from both experimental groups. (B) Volcano plot representing differential activation nodes in OSIS group compared to the control. Red circles indicate nodes with a p value < 0.010 and a fold change greater than 0.050. Yellow circles represent nodes with only a p value < 0.010 and were not considered differentially activated. Gray circles represent no differentially activated nodes. (C) OSIS leads to the expression of DNA damage and cell cycle arrest markers. (D) OSIS leads to increase in p16 activation and decrease in CDK4. (E) OSIS leads to SASP factors expression. (F) Heatmap demonstrating the systemic response of the network to constitutive Ras activation. (G) Volcano plot representing differential activation nodes in Ras stimulated group compared with the control. (H) Heatmap demonstrating the systemic response of the network to constitutive AKT1 activation. (I) Volcano plot representing differential activation nodes in AKT1 stimulated group compared with the control. Data are presented as median and IQR. Significant differences considered p value < 0.001 (***) and p value < 0.0001 (****). Normality data distribution was assessed using the Shapiro–Wilk normality test. Mann–Whitney U‐test was used for nonparametric samples. NAF, node activation frequency.

FIGURE 3

In silico cell cycle arrest simulation in different premature senescent models. (A) Increased activation of the G1 Phase node in most senescence‐inducing stimuli and reduced mitosis. (B) Positive correlation between G1 cell cycle arrest and TP53 or p21 activity. Data distribution normality assessed by the Shapiro–Wilk test. Intergroup differences assessed by the Kruskal–Wallis test followed by comparison of each group's mean rank with the rank of the control group. The data presented are expressed as the median and Interquartile Range. Significant differences considered when p value < 0.001 (***) or p value < 0.0001 (****). NAF, node activation frequency.

FIGURE 4

CCL11 induces activation of nodes associated with PI3K/AKT, MAPK14, and cell cycle arrest pathways in PseudoCell's network. (A) The PI3K, AKT1, and MAPK14 nodes had their activity modulated in a frequency‐dependent manner when disturbed by CCL11. (B) Activation of TP53, p16, and p21 node after CCL11 stimulation. Data distribution normality assessed by Shapiro–Wilk test. Intergroup differences assessed by Kruskal–Wallis test followed by comparation of each group's mean rank with the rank of the control group. Significant differences considered when p value < 0.050 (*), p value < 0.010 (**), and p value < 0.0001 (****). NAF, node activation frequency.

FIGURE 5

In silico evaluation of increasing frequency of CCL11 node activation. (A) Heatmap representing alteration in activation frequency patterns for Control, CCL11 Low, CCL11 Medium, CCL11 High, and CCL11 Constitutive groups. (B) Volcano plot showing the differentially activated nodes in Controls in contrast to CCL11 Low, Medium, and Constitutive (p value < 0.010 and |FC| > 0.050). Negative FC values were considered downregulated, while positive values were considered upregulated. (C) Top enriched pathways in CCL11 High. (D) CCL11 activation frequencies induce an increase in DSB, ATM, H2AX, Superoxide, and CYBB nodes activation (p value < 0.0001). Data distribution normality assessed by Shapiro–Wilk test. Intergroup differences assessed by Kruskal–Wallis test followed by comparison of each group's mean rank with the rank of the control group. Significant differences considered when p value < 0.050 (*), p value < 0.010 (**), and p value < 0.0001 (****). NAF, node activation frequency.

FIGURE 6

Effects of CYBB node suppression on DNA damage response and senescence markers. (A) Heatmap representing alteration in activation frequency patterns for Control, CCL11 High, and CYBB Knockout (KO CYBB) groups. (B) CYBB knockout suppresses activation of nodes associated with DNA damage, DSB, ATM, and H2AX. (C) CYBB knockout reduces TP53 activation, but not p21. Data are presented as median and IQR. Data distribution normality was assessed by Shapiro–Wilk test. Intergroup differences were assessed by Kruskal–Wallis test. Significant differences were considered p value < 0.010 (***), and p value < 0.0001 (****). NAF, node activation frequency.