4.5. AgingGenes and AgingNet Reviews

In this work, two analyzes were carried out essentially. The first, analysis of AgingGenes, aims to identify highly age-related transcripts and assess how their biological functions may be related to immunosenescence, taking into account their behaviour throughout life.

‌ The second, co-expression analysis, consists of identifying which transcripts interact with each other throughout life, which biological processes they possibly represent and how their behaviour over time may be related to immunosenescence. Such interactions between transcripts were represented through a co-expression network, called AgingNet.

4.5.1. Age-Related Transcripts: AgingGenes

A total of 56 AgingGenes were found, among which 10 have positive linear correlations (pos), 6 with negative linear correlations (neg) and 40 with non-linear correlations (mic). Among the mic, 3 subgroups of transcripts with distinct expression profiles were identified, consisting of 7, 25 and 8 transcripts each and called micA, micB and micC, respectively. The following figures (Figures 4.6 A and B) illustrate the genes corresponding to each AgingGenes found in the StringDB database (SZKLARCZYK et al., 2015).

Figura 4.6 A. AgingGenes com Correlação Linear. (A) genes com correlações lineares positivas (pos) e (B) negativas (neg). As linhas representam interações entre genes detectadas pelo String.

Figura 4.6 B. AgingGenes com Correlação Não-Linear (mic). A, B e C representam os genes com correlações não-lineares micA, micB e micC, respectivamente. As linhas representam interações entre genes detectadas pelo String.

Of all AgingGenes, only 5 genes were found in GenAge and correspond to genes TERF2, MED1, PIK3CA, MYC and PDGFRB, with non-linear correlations (micA, micB, micC) and negative and positive linear correlations, respectively. Interestingly, each of these 5 genes fits into a distinct type of age correlation.

Next, to better understand which biological processes the AgingGenes are involved in, a pathway enrichment analysis was performed for each class of transcripts as described in Section 3.5.4. Three large enrichment structures were found, illustrated in Figures 4.7 A, B and C.

Figura 4.7 A. Vias do Reactome Enriquecidas para os AgingGenes (Estrutura 1). Cada círculo representa uma via enriquecida, sendo que suas dimensões refletem o valor de enriquecimento (Combined Score) e a cor representa em quais classes dos AgingGenes tal via foi enriquecida.

Figura 4.7 B. Vias do Reactome Enriquecidas para os AgingGenes (Estrutura 2). Cada círculo representa uma via enriquecida, sendo que suas dimensões refletem o valor de enriquecimento (Combined Score) e a cor representa em quais classes dos AgingGenes tal via foi enriquecida

Figura 4.7 C. Vias do Reactome Enriquecidas para os AgingGenes (Estrutura 3). Cada círculo representa uma via enriquecida, sendo que suas dimensões refletem o valor de enriquecimento (Combined Score) e a cor representa em quais classes dos AgingGenes tal via foi enriquecida.

After exploring the possible biological processes in which the AgingGenes are related, their behaviour throughout life was analyzed. For this, a representative expression profile was created for each class of AgingGenes, illustrated in Figures 4.8 A and B. Such profiles represent the median expression of all transcripts for each age, separated by their class. Simultaneously, the analysis of detection of state changes was carried out and the ages where changes occurred are indicated by triangles, coloured according to the sub-module. In addition, the dimension of each triangle represents the intensity of change, calculated as the ratio (log2FC, log2 Fold-Change) between the expression of the age where the change of state occurred and the mean expression of the ages between 19 and 25 years old.

Figura 4.8 A. Perfil de Expressão dos AgingGenes com Correlações Lineares. A linha pontilhada representa o perfil mediano de expressão reescalado por z-score para, a linha completa representa a regressão loess utilizando todos os transcritos de cada classe e sua sombra representa a variabilidade. As cores representam cada tipo de correlação linear. Os triângulos representam momentos onde ocorreram alterações de estado e sua dimensão representa a intensidade da alteração (|log2FC|).

Figura 4.8 B. Perfil Mediano de Expressão para os AgingGenes com correlações não-lineares. A linha pontilhada representa o perfil mediano de expressão reescalado por z-score, a linha completa representa a regressão loess utilizando todos os transcritos de cada classe e sua sombra representa a variabilidade. As cores representam cada subgrupo dos AgingGenes não-lineares. Os triângulos representam momentos onde ocorreram alterações de estado e sua dimensão representa a intensidade da alteração (|log2FC|).

When evaluating the median expression profile of AgingGenes with linear correlation, there is a more accentuated and progressive increase in the pos class from 35 years old, where it undergoes its first small change of state at 36 years old and continues with increasingly intense changes until the age of 70, where it undergoes its greatest change. The profile of the neg class, on the other hand, presents a more troubling behaviour, starting to show decays at the age of 30, its first change of state, and again, more intensely, at 36, together with the pos class. The micA class, related to telomere maintenance, appears to be the class with the greatest disturbances suffered throughout life. After 10 years since the last changes suffered by the pos and neg class AgingGenes (36 years old), the changes seem to intensify leading to a state of clear disturbance until 56 years old, in which all classes again undergo state changes. The micC class has a positive trend with age, undergoing major changes around the age of 60 years. The micB class, on the other hand, has a profile similar to the micA class, but with less intense changes (except for the period between the years 85 and 90). The micC class, which has a positive trend concerning age, is related to cell proliferation, cell cycle pathways and mitosis. Furthermore, it is enriched for both CD28-dependent PI3K/Akt signalling pathways and signalling by growth factors such as FGFR and EGFR. The behaviour of these pathways agrees with the neg class, which indicates a decrease in the expression of genes associated with the entry of the S stationary phase. This class is also related to NOTCH1 signalling in cancer. At the same time, the micA class, the profile that suffers the steepest declines during life, is enriched for telomere maintenance pathways, such as telomere terminal packing, and especially enriched for the senescence-inducing pathway for telomere-related stresses and DNA damage. The micB class, with more subtle downward trends in expression during life, is enriched for Tumor Necrosis Factor (TNF) signalling pathways, both by TNFR1 (pathways listed in table Annex A - Enrichment of AgingGenes) and by TNFR2, involved in cytokine signalling through the activation of NF-kB through the non-canonical pathway. Cytokine signalling was also enriched for the pos class. The pos class is also related to PI3K/AKT signalling in cancer and, along with the micC class, is also enriched for the constitutive signalling pathway by aberrant PI3K in cancer.