The GSE24432 microarray dataset from 40 obese women was analyzed for this study. After processing with GEO2R, 664 genes were obtained with criteria DEGs adj. p < 0.05 and |log2(FC)|> 0.3, with a FC range of 0.244 to 1.816. “Cellular Ageing” on GeneCards yielded 1892 protein-coding genes with a relevance score > 1.35. Venny reveals 50 intersecting genes (Fig. 1).

Identification of differentially expressed genes (DEGs) of GSE24432 and cellular aging datasets

To understand the biological functions of the 50 DEGs, we mapped the DEGs across different biological processes, cellular components, and molecular functions. The GO functional analysis was separated into three parts: Biological process (BP), cellular component (CC), and molecular function (MF). This provides a broader functional context of the DEGs from a mechanistic point of view based on the hierarchical structure of the GO terms. Based on the study, the DEGs enriched in BP are associated with processes such as translational initiation, SRP-dependent co-translational protein targeting to membrane, nuclear-transcribed mRNA catabolic process, nonsense-mediated decay, protein targeting to endoplasmic reticulum (ER), and establishment of protein localization to the ER (Fig. 2A). Regarding cellular components (CC), DEGs were primarily enriched in the cytosolic ribosome, ribosome, focal adhesion, cell-substrate junction, cytosolic large ribosomal subunit, and large ribosomal subunit (Fig. 2B). DEGs enriched in Molecular Function include structural constituent of the cytoskeleton, cyclin-dependent protein serine/threonine kinase regulator activity, structural constituent of ribosome, protein kinase regulator activity, GTPase activity, kinase regulator activity, 5S rRNA binding, cyclin-dependent protein serine/threonine kinase inhibitor activity, GTP binding, purine ribonucleoside binding, purine nucleoside binding, enzyme inhibitor activity, ubiquitin-protein transferase regulator activity, and ubiquitin-protein ligase binding (Fig. 2C). SRplot shows the combined GO from BP, CC, and MF by enrichment score (Fig. 3A) using the set criteria of p < 0.01. As we conducted the KEGG pathway analysis, we found that the DEGs are mainly enriched in Phagosome, Coronavirus disease, Salmonella infection, Cell cycle, ribosome, p53 signaling pathway, Chronic myeloid leukemia, Gap junction, Endocrine resistance, Bladder cancer, Motor proteins, Focal adhesion, Pathogenic Escherichia coli infection, AMPK signaling pathway, Melanoma, Glioma, and Pancreatic cancer. The 10 pathways with the corresponding genes are shown in Table 1.

Gene ontology analysis for biological process A, cellular components B, and molecular function C

GO result for three ontologies and functional analysis plot. A The horizontal plot of combined gene ontology and functional analysis from biological processes, cellular components, and molecular functions by enrichment score. B GO CHORD shows the relationship between GO term and 14 DEGs, which meet the cutoff criteria: p < 0.01, gene score ≥ 3, and GOterm ≥ 3. The DEGs are ordered based on their expression levels (log2FC) to highlight upregulated or downregulated pathways. Thick ribbons connect to specific GO terms, indicating a high concentration of genes associated with a biological function. The DEGs correspond to the combined gene ontology and functional analysis in B as follows: Ribosomal Protein, RPL; Nucleophosmin, NPM1; Tubulin Beta 6 Class V, TUBB6; Ribosomal Protein S15a, RPS15A; Tubulin Alpha 8, TUBA8; Tubulin Beta 2B Class IIb, TUBB2B; Cyclin-dependent Kinase Inhibitor 2A, CDKN2A; Actin Gamma 1, ACTG1; Tubulin Beta 3 Class III, TUBB3; RAS Like Proto-Oncogene A, RALA; Cyclin Dependent Kinase Inhibitor 1B, CDKN1B

The nutrient-sensing network, which includes extracellular ligands like insulins and IGFs, receptor tyrosine kinases with which they interact, and intracellular signaling cascades such as the PI3K-AKT and Ras-MEK-ERK pathways, and transcription factors FOXOs and E26, has been remarkably conserved throughout evolution [23]. Tyrosine kinase (TK) receptors bind extracellular growth factors like insulin or IGF1, activating mitogen signaling pathways via several downstream mediators, including PI3K, AKT, and the molecular machinery involved in protein translation and ribosome biogenesis (mTORC1, 4EBP1, S6K1, RPS6) [24]. The mechanistic target of rapamycin complex 1 (MTORC1) is responsive to various nutrients, stressors, and low energy conditions, thereby regulating the activity of numerous proteins, including transcription factors such as sterol regulatory element-binding protein (SREBP) and transcription factor EB (TFEB) [25]. This network is a principal regulator of cellular functions, encompassing autophagy, mRNA and ribosome biogenesis, protein synthesis, glucose metabolism, nucleotides, and lipids. Additionally, it plays a crucial role in mitochondrial biogenesis and proteasomal activity [26]. GO and functional analysis results revealed extensive crosstalk of cellular and molecular components and pathways associated with mRNA and ribosome biogenesis, AMPK signaling, and p53 signaling pathway after calorie restriction, as shown in Figs. 2 and 3A. AMPK and SIRT1 are essential nutrient sensors regulated by the ER, impacting cellular metabolism and energy homeostasis linked to cell proliferation, autophagy, and apoptosis [27]. While the PI3K/Akt pathway promotes mTOR-dependent cell growth and proliferation by stimulating nutrient uptake, the AMPK signaling pathway induces the p53-dependent stress response when energy levels are low [28]. Dietary restriction inhibits MTORC1, activates AMPK, SIRT1, and SIRT3, and increases adaptive cellular stress responses while suppressing the somatotrophic axis, extending longevity [29]. Given that cancer is linked to a metabolic state characterized by high energy demand, this implies that calorie restriction offers a possible cancer treatment.

Protein–protein interaction (PPI) networks can be used to understand better the molecular and cellular mechanisms associated with disease onset and progression and identify therapeutic targets. To understand how these DEGs drive biological reactions, we sorted the genes according to gene score > 3 and GO term > 3 to show which DEGs are significantly enriched in BP, CC, and MF. We obtained 14 DEGs most relevant to the GO terms (Fig. 3B). The GO CHORD shows how the gene interacts with the different GO terms from BP, CC, and MF and is ordered based on their expression levels (log2FC) to highlight upregulated or downregulated pathways. Thick ribbons connect to specific GO terms, indicating a high concentration of genes associated with a biological function. The GO CHORD showed that most genes are enriched in ribosome biogenesis and protein synthesis. Next, the 14 DEGs were submitted to STRING.db to generate functional and specific interactions between the proteins. The number of edges generated was 30, the average node degree was 4.29, and the average local clustering coefficient was 0.698. The PPI enrichment value was p = 2.61 × 10−7, showing that the nodes have more interactions among themselves than what would have been expected for a random set of proteins of the same size and degree distribution drawn from the genome. This indicates that the proteins are biologically connected (Fig. 4A). CytoHubba ranked the proteins based on their importance within the interactive PPI network. Using the MCC method to increase sensitivity and specificity, the proteins were ranked as follows: RPL5 and RPL11 = 1, RPL9 and RPL15 = 3, NPM1 = 5, RPS15A = 6, TUBB6 = 7, TUBA8 and TUBB2B = 8, CDKN2A = 10, ACTG = 11, TUBB3 = 12, RALA = 13 and CDKN1B = 14 (Fig. 4B). The result revealed that ribosomal proteins are highly ranked in the network, which suggests their involvement in the molecular mechanisms associated with calorie restriction. Furthermore, we observed upregulated expressions of RPS15A, RPL15, RPL11, RPL9, and RPL5, as well as CDKN1B and NPM1, while CDKN2A was downregulated by calorie restriction (Fig. 4C). These ribosomal genes encode proteins that are essential to ribosome synthesis. Ribosomes are crucial in protein synthesis, cell survival, growth, and proliferation. Nutrient-sensing pathways, such as insulin/IGF signaling, regulate ribosome biogenesis by stimulating rRNA transcription in the nucleolus. A modest reduction of these nutrient-sensing pathways extends lifespan across species, thereby linking ribosome biogenesis to aging [30]. Dysregulation of ribosome biogenesis at different stages is linked to cell cycle arrest, senescence, or apoptosis by impairing the expression of ribosomal proteins, leading to age-related degenerative diseases like cancer [31]. The NPM is another multifunctional protein whose depletion impairs ribosome biogenesis at multiple levels [32]. Our result showed that following calorie restriction intervention, expressions of NPM1 and other ribosomal protein genes, RPS15A, RPL15, RPL11, RPL9, and RPL5, showed consistent patterns of elevated expression (Fig. 4C). This agrees with the PPI network in Fig. 4A, showing the co-expression of NPM1 with the ribosomal proteins and suggesting potential regulatory function, association, and tissue-specific processes. As a histone chaperone that can stimulate rRNA transcription, NPM has been implicated in numerous cellular processes such as centrosome duplication, cell cycle regulation and genome stability maintenance [33]. The NPM exhibits a bidirectional role in apoptosis, which frequently vary according on the circumstances. Alterations in the NPM1 have been reported in numerous hematological malignancies [34]. The CDKN2A promotes NPM degradation, inhibiting ribosome biogenesis [35] (Fig. 4B). Moreover, in cases where NPM is upregulated, it is reported to be an attractive target for cancer therapy [31]. Our findings suggest that calorie restriction may regulate several molecular mechanisms involved in protein synthesis to increase life span and mitigate chronic disease by downregulating ribosome biogenesis and protein translation, regulating global proteostasis, and reducing disease severity.

PPI network from STRING.db A and Cytoscape B and differential expressions of DEGs represented by log2FC values C. The CytoHubba plugin identifies and ranks the nodes from the PPI network based on the Maximal Clique Centrality (MCC) method. Rank as follows: RPL5, 1; RPL11, 1; RPL9, 3; RPL15, 3; NPM1, 5; RPS15A, 6; TUBB6, 7; TUBA8, 8; TUBB2B, 8; CDKN2A, 10; ACTG1, 11; TUBB3, 12; RALA, 13; CDKN1B, 14. Ribosomal Protein, RPL; Nucleophosmin, NPM1; Tubulin Beta 6 Class V, TUBB6; Ribosomal Protein S15a, RPS15A; Tubulin Alpha 8, TUBA8; Tubulin Beta 2B Class IIb, TUBB2B; Cyclin-dependent Kinase Inhibitor 2A, CDKN2A; Actin Gamma 1, ACTG1; Tubulin Beta 3 Class III, TUBB3; RAS Like Proto-Oncogene A, RALA; Cyclin Dependent Kinase Inhibitor 1B, CDKN1B. PPI (A) color codes: from curated databases (aqua), experimentally determined (purple), gene neighborhood (green), gene fusions (red), gene co-occurrence (navy), text mining (yellow) co-expression (black), protein homology (maya)

To determine which of the DEGs genes could serve as a prognostic biomarker for colon cancer, we run the hub genes through the drug–gene interaction database (DGIdb) to understand the gene druggability information and available drugs and therapeutics for the hub genes. We identified several FDA-approved drugs from DGIdb for the genes. This step revealed that profiled drugs associated with these genes are treatments for different types of cancer, as shown in Table 2, with various side effects such as anemia, ascites, alopecia, constipation, edema, and fertility issues in girls and women. Next, we compared the distribution of the hub genes across different cancer hallmarks. Gene set enrichment analysis (GSEA) revealed the involvement of the hub genes in several hallmarks, such as tissue invasion and metastasis (p < 0.001), tumor-promoting inflammation (p < 0.001), resisting cell death (p < 0.01), and replicative immortality (p < 0.05) (Fig. 5 and Table 3). To validate the hub genes and discover potential prognostic biomarkers, we utilized the Kaplan Meier analysis and Cox proportional hazards regression to assess the correlation between the genes and survival of colon cancer. The study included 1167 patients and was not restricted by gender, adjuvant chemotherapy, or molecular subtypes of CRC. Although most patients receive conventional treatments like chemotherapy, radiation, or surgery, adenocarcinoma formation involves multiple genes and pathways [36]. This highlights the need for a comprehensive biomarker discovery to identify molecules involved in disease progression. Our results showed that higher expression of 4 hub genes, CDKN2A (FDR < 5%), RPL9 (FDR < 2%), TUBB6 (FDR < 1%), and RPS15A (FDR < 1%), and lower expression of CDKN1B (FDR < 1%), NPM1 (FDR < 1%), and RALA (FDR < 1%), correlated to shorter survival (Fig. 6). We cross-referenced these genes to the scAT expression of participants with obesity. We found that calorie restriction intervention decreased the expressions of CDKN2A and TUBB6 but increased CDKN1B and NPM1, which implies their mechanistic involvement in linking obesity to colon cancer. Genome-wide association studies (GWAS) revealed the association of CDKN2A locus with obesity-related diseases, including epicardial adipose tissue development, gestational diabetes, rapid decline in beta cell function, diabetic nephropathy progression, and coronary heart disease [37]. Since colon adenocarcinoma (COAD) is a common age-related digestive system tumor associated with obesity, this may support CDKN2A as a prominent biomarker for different CRC conditions and may be significant for the prognosis of cancer predisposition during early adulthood. Our result agrees with a previous study that showed considerably higher levels of CDKN2A expression in CRC tissues than in normal tissues, and elevated expression of CDKN2A was associated with poor CRC prognosis [38]. A similar experiment also revealed that CDKN2A was highly expressed (p < 0.001) in COAD compared to normal tissues, and the survival time was shorter for high CDKN2A expression when compared to low expression [36]. We further validated the genes on the UALCAN Data Analysis Portal, cBioPortal, and GEPIA2 databases. Among the four genes (CDKN2A, TUBB6, CDKN1B, and NPM1) whose expressions were regulated by calorie restriction intervention, we found that only CDKN2A was consistent throughout multiple databases with high expression significantly associated (p < 0.05) with shorter survival in COAD patients. This highlights CDKN2A as a more reliable prognostic molecular marker for COAD and a target for calorie restriction intervention (Fig. 7). However, based on the functional annotations and PPI analysis (Figs. 3 and 4), CDKN2A may also act as a master regulator of other hub genes by promoting or repressing their activity, thereby regulating several downstream molecular mechanisms and biological processes.

Cancer hallmark enrichment plot shows the enrichment of the cancer hallmarks when compared to the integrated cancer hallmark gene set and the distribution of genes across the different hallmarks to each other. Significant hallmarks are colored (p < 0.05)

Survival probability of colon cancer-associated genes TUBB6, CDKN2A, CDKN1B, NPM1, RALA, RPL9, and RPS15A. High expressions of 4 hub genes, CDKN2A (FDR < 5%), RPL9 (FDR < 2%), TUBB6 (FDR < 1%), and RPS15A (FDR < 1%), and low expressions of CDKN1B (FDR < 1%), NPM1 (FDR < 1%), and RALA (FDR < 1%), correlated to shorter survival. The four genes TUBB6, CDKN2A, CDKN1B, and NPM1 showed significant changes in gene expression (p < 0.05, |Log2FC|> 0.3) of participants with obesity after calorie restriction intervention compared to before. HR = hazard ratio, FDR = False Discovery Rate

Effect of CDKN2A expression level and prognostic significance analysis of COAD from UALCAN Data Analysis Portal A, cBioPortal B, and GEPIA2 databases C and D, gene expression profile E and boxplot F (p < 0.01) for normal (N) and tumor (T), and stage plot G (p = 0.00234). High expression of CDKN2A is associated with shorter survival in COAD patients. COAD = colon adenocarcinoma, HR = hazard ratio

The CDKN2A gene encodes multiple tumor suppressor 1 (MTS1), which belongs to the INK4 family [39]. p16INK4a inhibits CDK4/6 to prevent the phosphorylation of retinoblastoma protein, a tumor suppressor protein that promotes binding to transcription factor E2F and blocks G1 phase exit [37]. CDKN2A gene mutation results in cyclin D-CDK4 inhibition, resulting in abnormal cell proliferation [40], which explains the lower survival rates in CRC with increased CDKN2A levels. In addition, the INK4a-ARF locus encodes p16INK4a and p19ARF, which regulate Rb and p53, respectively. As a tumor suppressor protein, p53 mutations have been linked to ribosomal protein deletions (RPL5 and RPL11), increasing human cancer susceptibility [41]. Our study revealed that ribosomal protein genes are regulated by energy restriction intervention (Fig. 4C) and may be essential for p53 signaling by regulating cell stress response, inducing apoptosis, cell cycle arrest, or senescence. This agrees with a previous study which showed that energy restriction differentially regulated functional pathways like focal adhesion, apoptosis and p53 signaling (Mutch et al. 2011).

CDKN2A is a prototypical marker of cellular senescence. It is an independent prognostic factor that promotes the progression of CRC through epithelial–mesenchymal transition (EMT), a significant hallmark in tumor invasion and the process behind metastasis initiation that is involved in the apoptotic regulation of HT-29 cells [42]. The EMT has also been linked to beta-tubulin [43]. Tissue invasion and metastasis are among the cancer hallmarks associated with tubulin isotypes, according to Fig. 5 and Table 3. These isotypes regulate oncogenesis and possess prognostic influence in various solid tumors [44]. In bladder urothelial carcinoma cell lines, TUBB6 depletion reduced cell migration and invasion [45]. Several other studies showed the correlation of TUBB6 overexpression to tumor aggressiveness and shorter survival [46,47,48], as seen in Fig. 6. However, calorie restriction reduced the expression of TUBB6 and other tubulin isotypes (TUBB2B, TUBB3, and TUBA8) (Fig. 4C), which indicates that these genes could offer therapeutic targets for colon cancer by targeting specific cancer hallmarks like tumor-promoting inflammation, tissue invasion, and metastasis, resisting cell death and replicative immortality (Fig. 5). Because the overall survival rates of patients with high CDKN2A and TUBB6 are significantly lower, our result suggests that in addition to CDKN2A, TUBB6 influences patient survival and can be considered a prognostic molecular marker of colon cancer. Our result also showed that elevated expression of NPM1 was accompanied by the downregulation of CDKN2A expression following calorie restriction intervention, as shown in Fig. 4C. A recent study revealed that overexpression of NPM1 inhibits CDKN2A and increases cell proliferation in esophageal squamous cell carcinoma [49]. Other studies have also demonstrated the elevated expression of NPM1 in cancers associated with obesity, including CRC [50], kidney cancer [51], and liver cancer [52]. However, a major study revealed that nuclear retention of NPM1 increases CDKN1A and CDKN2A genes in acute myeloid leukemia [53]. This suggests that NPM1 exhibits context-dependent regulation of transcription, acting either as a transcriptional coactivator or corepressor, and a substantial tumor cell line depends on the expression [54]. The differential expressions of NPM1 and CDKN2A (Fig. 6) in colon cancer support the antagonistic effect of NPM1 on CDKN2A, which may indicate that the overexpression of NPM1 suppresses CDKN2A and limits its cell cycle functions.

Calorie restriction is well established as a longevity intervention in flies, rodents, and nonhuman primates, with a potential benefit in preventing malignancies and enhancing the efficacy of cancer treatments. However, to our knowledge, no single pharmacological agent has been developed to replicate all its benefits. This limitation stems from the incomplete understanding of the intricate molecular processes by which calorie restriction influences several biochemical pathways and regulates systemic metabolism. Several factors associated with calorie restriction intervention, including macronutrient sources, time of intake, optimal timing and duration, feeding schedule, inter-individual variability influenced by behavioral and environmental factors, and nutrigenetics, must be considered as significant variables that may have an impact on the outcomes. Preclinical models have provided valuable insights into how organisms adapt to fasting and how cancer cells respond to nutrient restriction. These discoveries have underscored calorie restriction as a feasible therapy for the prevention of cancer and chemotherapy toxicity across multiple malignancies [55]. However, the prolonged latency period for CRC presents challenges in determining the optimal window for dietary intervention [56]. Additionally, the risk of cancer-related weight loss may affect patient compliance and recruitment for prospective clinical studies. Since calorie restriction can inhibit tumor progression by altering cancer cell energy metabolism, delaying disease advancement, and improving survival outcomes, it has the potential as a therapeutic adjunct through metabolic reprogramming, ultimately enhancing treatment efficacy and patient prognosis.

A major limitation of our study is the small sample size, which may reduce its generalizability with the broader population. Further, the statistical power of the analysis is diminished by the small sample size, which increases the likelihood of Type II errors (failure to detect actual effects) and, when significant results are obtained, may lead to an overestimation of effect sizes. Additionally, the small sample size hinders the study’s ability to control confounding variables and reduces the generalizability of the findings. Future research should incorporate larger, more diverse cohorts to validate and expand these preliminary findings. Given the molecular heterogeneity of CRC and confounding factors such as age, tumor source, and size, these results should be interpreted with caution until validated through large-scale integrated analyses, in vivo studies, and clinical trials. However, a key strength of this study is its integration of multiple bioinformatics platforms for biomarker identification, enabling a rigorous stepwise comparison of existing and emerging databases. While colonoscopy remains the gold standard for CRC diagnosis, identifying non-invasive and more reliable early-stage diagnostic tools via biomarker discovery is essential for enhancing early detection and developing more personalized, effective treatment strategies.