Integrated multi-omics analysis reveals the molecular interplay between circadian clocks and cancer pathogenesis

Abstract

Circadian rhythms (CRs) are fundamental biological processes that significantly impact human well-being. Disruption of these rhythms can trigger insufficient neurocognitive development, insomnia, mental disorders, cardiovascular diseases, metabolic dysfunctions, and cancer. The field of chronobiology has increased our understanding of how rhythm disturbances contribute to cancer pathogenesis, and how circadian timing influences the efficacy of cancer treatments. As the circadian clock steadily gains recognition as an emerging factor in tumorigenesis, a thorough and comprehensive multi-omics analysis of CR genes/proteins has never been performed. To shed light on this, we performed, for the first time, an integrated data analysis encompassing genomic/transcriptomic alterations across 32 cancer types (n = 10,918 tumors) taken from the PanCancer Atlas, unfavorable prognostic protein analysis, protein-protein interactomics, and shortest distance score pathways to cancer hallmark phenotypes. This data mining strategy allowed us to unravel 31 essential CR-related proteins involved in the signaling crossroad between circadian rhythms and cancer. In the context of drugging the clock, we identified pharmacogenomic clinical annotations and drugs currently in late phase clinical trials that could be considered as potential cancer therapeutic strategies. These findings highlight the diverse roles of CR-related genes/proteins in the realm of cancer research and therapy.

Figure 1

The circadian rhythms and circadian clock. (A) The circadian timing system synchronizes central and peripheral clocks across the human body to adapt our physiology to environmental changes. Light is received by ipRGCs in the eyes, which sends electrical signals to the SCN through the retinohypothalamic tract. The peripheral nervous system and humoral signals convey information from the SCN to orchestrate peripheral clocks. Feeding schedule and exercise can also activate central and peripheral clocks. Finally, circadian rhythms regulate hormones, thermogenesis, immunity, metabolism, reproduction, fat storage, and stem cell development. (B) Neurotransmitters released by ipRGCs. Glutamate and PACAP cause membrane depolarization in the postsynaptic SCN neurons. Changes in cAMP and Ca²⁺ levels induce phosphorylation of the CREB protein, and expression of canonical clock components (i.e., PER1 and PER2), thereby resetting SCN cellular oscillators. GABA, an inhibitory neurotransmitter, decreases the sensitivity of non-image-forming behaviors at low light levels. Lastly, SCN neurons control peripheral clocks throughout the body via neuronal and hormonal signals. (C) The human molecular clock is composed of canonical clock components, clock-controlled genes, and clock-controlled pathways. The clock is operated through a network of transcription-translation feedback loops (positive, negative, auxiliary, and metabolic) that oscillate with a near-24-h cycle (see Introduction). ipRGC intrinsically photosensitive retinal ganglion cells, SCN suprachiasmatic nucleus, PACAP pituitary adenylate cyclase-activating polypeptide, GABA γ-aminobutyric acid, RRE RORA or NR1D1 response elements, CCC canonical clock components, NAM nicotinamide, NMN nicotinamide mononucleotide, NAD⁺ nicotinamide adenine dinucleotide.

Figure 2

OncoPrint of genomic and transcriptomic alterations across 32 TCGA PanCancer types. (A) Ranking of the most altered CR-related genes (n = 206) considering the mean f of alteration events (cutoff = 0.063) and the CR-cancer score > 0.9. The Mann–Whitney U test showed significant difference of genomic alterations between genes upper the mean f and lower the mean f (P < 0.001). The OncoPrint was performed using the data from the cBioPortal platform (http://www.cbioportal.org/)^,. (B) Mean f per alteration type and significant Bonferroni correction (P < 0.001) of mRNA downregulation, mRNA upregulation and CNV amplification in comparison with other alterations. (C) Ranking of the 50% (n = 16) most altered TCGA PanCancer types according to the mean f of alterations. (D) Ranking of the top ten TCGA PanCancer types with highest mean f of genomic alterations in CCCs. (E) Ranking of the top ten TCGA PanCancer types with highest mean f of genomic alterations in CCGs. (F) Ranking of the top ten TCGA PanCancer types with highest mean f of genomic alterations in genes mediating NMCRE. CR circadian rhythm, CCCs canonical clock components, CCGs clock-controlled genes, NMCRE neural mechanisms of circadian rhythmicity and its entrainment, CNV copy number variants, TCGA The Cancer Genome Atlas, mRNA messenger RNA, BRCA breast invasive carcinoma, LGG brain lower grade glioma, PRAD prostate adenocarcinoma, LUSC lung squamous cell carcinoma, SKCM skin cutaneous melanoma, STAD stomach adenocarcinoma, BLCA bladder urothelial carcinoma, SARC sarcoma, PAAD pancreatic adenocarcinoma, ESCA esophageal carcinoma, ACC adrenocortical carcinoma, MESO mesothelioma, UVM uveal melanoma, KICH kidney chromophobe, UCS uterine carcinosarcoma, DLBC lymphoid neoplasm diffuse large B-cell lymphoma, CHOL cholangiocarcinoma.

Figure 3

Circadian rhythm protein–protein interactome network. Network made up of 13 CCCs (sky blue nodes; mean degree centrality = 17.2), 37 (29%) CCPs (red nodes; mean degree centrality = 16.6), and 76 (37%) proteins involved in the NMCRE (green nodes; mean degree centrality = 18.0) with at least one high-confidence interaction (cutoff > 0.9) with cancer driver proteins (pink node; mean degree centrality = 19.4). The Mann–Whitney U test showed a correlation of degree centrality between cancer driver nodes and CR-related nodes (P > 0.05). The CR-related proteins with both the highest degree centrality and the CR-cancer scores > 0.9 were EP300, TP53, HDAC1, MAPK8, and BTRC. Lastly, the CR-PPi network was designed and visualized through the Cytoscape software v.3.10 (https://cytoscape.org/). CR circadian rhythm, PPi protein–protein interaction, CCCs canonical clock components, CCPs clock-controlled proteins, NMCRE neural mechanisms of circadian rhythmicity and its entrainment.

Figure 4

Circadian rhythm-related genes/proteins with unfavorable prognosis in different cancer types and distance score of shortest pathways to cancer hallmark phenotypes. (A) The Human Pathology Atlas details the CR-related genes with unfavorable prognosis and log rank P-value < 0.001 across 19 cancer types. Additionally, 15 CR-related genes had the highest CR-cancer scores (> 0.9). All data was taken from The Human Protein Atlas platform (https://www.proteinatlas.org/). (B) Box plots showing the mean distance scores of shortest paths per cancer hallmark phenotype, and the Bonferroni correction as multiple comparison test (P < 0.001) to show significant differences across cancer phenotypes. The shortest paths to cancer hallmark phenotype analysis reveals that 16 (14%) CCCs, 37 (32%) CCPs, and 64 (55%) proteins involved in the NMCRE have the shortest distance scores to cancer phenotypes. (C) CR-related proteins with both the shortest distance scores to cancer hallmark phenotypes and the highest CR-cancer scores (> 0.9). Lastly, the shortest paths to cancer hallmark phenotypes were analyzed by using data from CancerGeneNet (https://signor.uniroma2.it/CancerGeneNet/). CR circadian rhythm, CCCs canonical clock components, CCPs clock-controlled proteins, NMCRE neural mechanisms of circadian rhythmicity and its entrainment, THCA thyroid carcinoma, GBM glioblastoma multiforme, HNSC head and neck squamous cell carcinoma, LUSC lung squamous cell carcinoma, LUAD lung adenocarcinoma, BRCA breast invasive carcinoma, UCEC uterine corpus endometrial carcinoma, OV ovarian serous cystadenocarcinoma, CESC cervical squamous cell carcinoma and endocervical adenocarcinoma, KIRP kidney renal papillary cell carcinoma, KICH kidney chromophobe, KIRC kidney renal clear cell carcinoma, LIHC liver hepatocellular carcinoma, PAAD pancreatic adenocarcinoma, SKCM skin cutaneous melanoma, PRAD prostate adenocarcinoma, BLCA bladder urothelial carcinoma, STAD stomach adenocarcinoma, CRC colorectal carcinoma.

Figure 5

Integrated multi-omics data analysis and functional enrichment analysis. (A) Heatmap of CR-cancer scores per multi-omics approach to prioritize the 31 essential CR-related proteins significantly involved in cancer (cutoff > 0.9). (B) The functional enrichment analysis displays the most significant annotations (Benjamini–Hochberg FDR < 0.01) related to the GO biological processes (http://geneontology.org/), the Reactome signaling pathways (https://reactome.org/), and the WikiPathways (https://www.wikipathways.org/) on cancer. The results of this enrichment were visualized using a Manhattan plot and were obtained through the g:Profiler software version e101_eg48_p14_baf17f0 (https://biit.cs.ut.ee/gprofiler/gost). (C) Overall survival analysis comparing metastatic patients with genomic alterations in 10 essential CR-related genes (n = 15,194) versus unaltered patients (n = 10,465). Unaltered patients had a median month average (55.72) significantly higher than altered patients (31.64) showing a log rank test P < 0.001. CR circadian rhythm, GO gene ontology, WP WikiPathways, BP biological processes, FDR false discovery rate, MSK-MET Memorial Sloan Kettering–Metastatic Events and Tropism, CI confidence intervals.

Figure 6

Pharmacogenomic clinical annotations and clinical trials. (A) Identification of known and predicted oncogenic variants (n = 765) through the boostDM machine learning-based method. (B) Sankey plot of in silico drug prescription based on pharmacogenomic clinical annotations. (C) Overview of phase III and IV clinical trials. Sankey plot showing therapeutic targets, drugs, mechanisms of action, and cancer types involved in late phase clinical trials. Data of clinical trials and mechanisms of action were taken from the Open Targets Platform (https://platform.opentargets.org/), and the Drug Repurposing Hub (https://clue.io/repurposing). Lastly, Sankey plots were designed using the SankeyMATIC software (https://sankeymatic.com/ and https://github.com/nowthis/sankeymatic) CR circadian rhythm.