Identification of the potential biomarkers associated with circadian rhythms in heart failure

Introduction

Heart failure (HF) emerges as a syndrome with complex clinical symptoms elicited by the damage to the heart’s structure and/or function due to various pathogenic factors (McDonagh et al., 2021), which indicates the terminal stage of various heart diseases accompanied with a permanent deterioration in prevalence, incidence, mortality, and re-hospitalization rates. The prevalence of HF was reported as 1–3% in the general population among developed countries, and increased to about 5–9% in the elderly (van Riet et al., 2016; Virani et al., 2020).

Chronic heart failure (CHF) leading to disabling symptoms and high hospitalization rates, which exerts a heavy burden on patients, followed by their families, and society’s healthcare resources. Previous research has suggested a possible association of the disrupted circadian rhythm with the accelerated HF development, while the underlying mechanisms remain to be elucidated. All living organisms on the earth exhibit the circadian rhythms with a 24-h cycle, which are manipulated by the biological clock that provides basic timing information to manipulate biochemical, physiological, and behavioral processes with the adaptation to the environment and its variations. It has been demonstrated that approximately 13% of genes and 8% of proteins show rhythm in the heart, and disruptions of the molecular clock could lead to significant cardiovascular disease (Storch et al., 2002), which may also perturb the delicate balance with severe repercussions on HF outcomes (Alibhai et al., 2014; Song et al., 2022). As a result, a comprehensive elucidation of the link between circadian clock-related genes and the heart will contribute to developing a novel diagnostic and therapeutic approach to HF.

The current study was carried out to identify the circadian rhythms-related hub genes and evaluate their diagnostic value in HF, providing reference for detecting and treating the disease depending on the genomic expression profiles. Differentially expressed genes (DEGs) in HF in comparison to healthy samples were screened out according to the analysis datasets GSE57338 from the Gene Expression Omnibus (GEO). Subsequently, circadian rhythm-related genes were overlapped in the Circadian Gene DataBase (CGDB) to obtain the circadian rhythm-related DEGs. To explore the potential molecular mechanisms underlying at the pathogenesis of HF, we enriched the functions and pathways. The protein-protein interaction (PPI) network was used to analyze the rhythm-related DEGs, and the MCODE in Cytoscape software to identify hub genes. Hub genes also exhibited a high diagnostic accuracy according to the receiver operating characteristic analysis (ROC) validated in the external validation dataset. Eventually, we determined the mRNA expression levels of hub genes by quantitative RT-PCR in clinical samples of human peripheral blood. The rhythms-related hub genes for circadian may provide reference for potential biomarkers and therapeutic targets for HF.

Materials and Methods

Results

GO and KEGG enrichment analysis of circadian rhythms-related DEGs

By intersecting the circadian rhythms-related genes with DEGs, a total of 723 circadian rhythms-related DEGs were determined (Fig. 2), with the top 10 results of GO functions depicted in Fig. 3A. The GO category indicated the prominent enrichment of circadian rhythms-related DEGs in functions of manipulating lipid metabolic process, rhythmic process and circadian rhythm. KEGG analysis demonstrated the role of circadian rhythms-related DEGs in the Circadian rhythm, AMPK pathway and HIF-1 pathway (Fig. 3B).

PPI network construction and module analysis

The PPI network was employed to examine the circadian rhythm-related DEGs, through which 278 nodes and 404 protein interactions were identified (Fig. 4A). Afterwards, MCODE was employed to extrude the prominent module in the PPI network, which isolated three core modules. In detail, module A (score = 6.857) covered 24 interactions and eight protein nodes (Fig. 4B), module B (score = 5) for ten interactions and five nodes (Fig. 4C), and module C (score = 5) for ten interactions and five nodes (Fig. 4D). The eight MCODE Score Top1 module genes, CRY1, ARNTL, PER2, BHLHE41, CRY2, PER3, NPAS2 and PER1 were selected as hub genes for further analysis.

Identification of hub genes

Four down-regulated hub genes (ARNTL, CRY1, NPAS2 and PER1) and four up-regulated hub genes (PER2, PER3, CRY2 and BHLHE41) were identified in HF samples (Fig. 5A). According to the ROC curves, the genes of ARNTL, PER3, CRY2, BHLHE41 and NPAS2 exhibited high accuracy with AUC > 0.7 (Fig. 5B).

Subsequently, we validated the expression and diagnostic value of ARNTL, PER3, CRY2, BHLHE41 and NPAS2 in GSE5406. The genes in GSE5406 exhibited a similar expression trend to those in the training set GSE57338 (Fig. 6A), with a more prominent AUC values over 0.7 (Fig. 6B). As a result, ARNTL, PER3, CRY2, BHLHE41 and NPAS2 were identified as potential diagnostic markers for HF.

Validation of hub genes

The variation of mRNA level in hub genes ARNTL, PER3, CRY2, BHLHE41 and NPAS2 was detected using the peripheral blood of the participants by quantitative RT-PCR, which for CRY2 and BHLHE41 showed significant increase, whereas decrease for ARNTL and NPAS2 in the HF group in comparison to the control (P < 0.001); PER3 mRNA showed a slight higher expression in the HF group (n = 21) in comparison to the control group (n = 10), while no significant difference was found (P = 0.118) (Fig. 7), which is similar to the outcomes from our bioinformatics analysis.

Discussion

Increasing evidence points to the association of circadian disruption with various diseases, not except for HF. Circadian rhythms-related genes (CCGs) generally show the differential expression in heart tissues, which may exert an essential role in HF initiation and progression, as demonstrated by mouse models (Lefta et al., 2012; Duong et al., 2019). Epidemiological evidence has also demonstrated the significant correlation of the disruption of circadian rhythms in humans with the increased risk of HF (Yamada et al., 2015; Li et al., 2021), that circadian rhythm disturbances may promote the development and progression of HF. As a result, there is an urgent requirement for a detailed understanding of the biological mechanism, serving as the reference for developing novel prevention, diagnosis, and treatment for HF.

In the present study, 10,163 DEGs were identified by a series of bioinformatics analyses based on gene expression profiles obtained from GSE57338, covering 4,615 up-regulated and 5,548 down-regulated genes in HF patients compared to healthy controls. After overlapping in the Circadian Gene DataBase (CGDB), we obtained 723 circadian rhythms-related DEGs. GO analysis indicated a function enrichment of circadian rhythms-related genes, such as regulating lipid metabolic processes, rhythmic processes, and circadian rhythm. Studies have validated the circadian rhythm-related lipid metabolism disorder as a critical step in HF development and progression (Wende et al., 2016; Crnko et al., 2019; Song et al., 2022), which reached an accordance to our bioinformatics analysis. Besides, the significantly enrichment of most circadian rhythms-related DEGs in AMPK signaling and the HIF-1 signaling pathway was revealed by the KEGG pathways analysis, in which AMPK serves as a primary regulator that integrates energy homeostasis with the circadian regulation of vital metabolic pathways (Jordan & Lamia, 2013; Lee & Kim, 2013).

Moreover, some studies have revealed the AMPK involvement in maintaining apoptosis, autophagy, and energy metabolism in degenerating hearts (Kops et al., 2002; Puigserver et al., 2003; Lagouge et al., 2006; Ronnebaum & Patterson, 2010; Schilling & Kelly, 2011; Kim et al., 2012). Hypoxia-inducible factor 1 (HIF1), an oxygen homeostasis regulator, has been validated to play a role in glucose metabolism and mitochondrial respiration (Peek et al., 2017), which could also be activated by the tissue hypoxia, showing the potential as an intriguing factor to mediate the relationship between metabolic energy disorder and HF (Fillmore, Mori & Lopaschuk, 2014; Warbrick & Rabkin, 2019). As outlined above, a metabolic shift is finely tuned by the metabolic energy sensors, including AMPK and HIF-1, which manipulate the cardiomyocyte molecular clock through the availability of energy and nutrient. The underlying mechanism of dysregulation of circadian rhythm could be associated with HF development and progression.

We selected eight overlapping genes as the most significant hub genes in GSE57338 according to the PPI network, which were, CRY1, ARNTL (also known as BMAL1), PER2, BHLHE41 (also known as DEC2), CRY2, PER3, NPAS2, and PER1. Further, ROC curves were plotted for evaluating the diagnostic values of hub genes, which identified five hub genes with high accuracy that showed AUC >0.7, that were, down-regulated (ARNTL, NPAS2) and up-regulated (PER3, CRY2, BHLHE41) genes in HF samples. Available data suggest that CLOCK (or NPAS2)-BMAL1/ARNTL heterodimer could activate the transcription of the negative elements, Period (Per1-3) and Cryptochrome (Cry1-2), as well as other clock-controlled genes that enable maximum advantage provided to cardiac function (Zhang & Jain, 2021). Our study demonstrates the association of HF with energy metabolism that is linked with circadian rhythm disorder. According to the study of Young et al., Cardiomyocyte-specific Bmal1 knockouts (CBK) mice exhibited metabolic dysfunction (depressed glucose utilization and increased fatty acid oxidation) (Young et al., 2014). Kohsaka et al. (2014) found that BMAL1 (-/-) mice tended to exhibit the mitochondrial dysfunctions in the heart. Moreover, the development of age-related dilated cardiomyopathy was also found in BMAL1 (-/-) mice, characterized by the thinned myocardial wall, dilated left ventricle, and the decreased cardiac output (Lefta et al., 2012). Neuronal PAS Domain Protein 2 (NPAS2) serves as an essential member of the clock gene that manipulates the biological clock’s rhythm. Studies have demonstrated a close relation of NPAS2 to cardiovascular diseases. For instance, Englund et al. (2009) found a close association with hypertension, which is also a risk factor for HF. A previous study has displayed a uniform converge of the primary dysregulated pathway(s) in the NPAS2 (-/-) animals on lipid metabolism (O’Neil et al., 2013), which is essential for the homeostasis maintenance of cardiac energy metabolism. In addition, NPAS2 (-/-) mice showed a lacked or blunted bimodal activity and sleep patterns in comparison to wild-type mice (Dudley et al., 2003). In contrast, Clock mutants of wild-type mice display the increased physical activity and food intake at rest and fasting (Turek et al., 2005), suggesting the nonredundant functions of Clock and Npas2 paralogs concerning energy balance (Huang et al., 2011).

Cryptochrome2 (CRY2) shows a 10-fold greater expression compared to CRY1 in the heart, which means that CRY2 is more involved in the heart’s clock negative feedback loop mechanism than CRY1 (Young, Razeghi & Taegtmeyer, 2001). Machicao et al. (2016) identified that CRY2 as a novel switch in hepatic fuel metabolism, which promotes the increased triglyceride storage and reduced glucose production, as well as to regulate muscle metabolism according to several studies (Ehrenborg & Krook, 2009; Jordan et al., 2017). Moreover, the genetic ablation of CRY1/2 exhibited an association to the exercise capacity increase in mice, implying it function to modulate metabolic flexibility in muscles, to achieve a balanced daily energy between supply and demand (Jordan et al., 2017). The clock repressor PER2 manipulates the lipid metabolism by modulating PPAR-gamma and mitochondrial rate-limiting enzymes. (Grimaldi et al., 2010; Neufeld-Cohen et al., 2016). It was demonstrated that the peak point of PER3 expression were positively correlated with physical activity and peak oxygen uptake in older adults (Takahashi et al., 2017). The stabilization of PER2 triggers a HIF-controlled metabolic switch that promotes the myocardial adaptation to ischemia (Eckle et al., 2012), confirming to the hypothesis that PER possibly optimizes the mitochondrial reaction to transformations in heart energy supply and request. BHLHE41, also known as DEC2, serves as a negative regulator of the circadian rhythm that suppresses Clock/Bmal1 expression (Honma et al., 2002). As reported, the dominant energy sensor phosphorylation of AMP-activated protein kinase (pAMPK) is markedly promoted in DEC knockout mice (Sato et al., 2016). The hypoxia-inducible factor-1α (HIF-1α) activates the transcription of Dec1 and Dec2 that play a role in regulating the lipid metabolism by inhibiting PPAR transcription (Yun et al., 2002; Sato et al., 2008; Park & Park, 2012; Chen & Yang, 2014). Consequently, the disruption of normal CCRGs expression may disturb the cardiac energy homeostasis, which might be related to the occurrence, development, and prognosis of HF (Yan et al., 2021). Based on the integrated bioinformatical studies, we conducted a comprehensive and novel analysis of clock gene expression profiles in patients with HF. In particular, the ARNTL, CRY2, BHLHE41, and NPAS2 are identified as the potential biomarkers associated with the circadian rhythm in HF. Meanwhile, the down-regulated genes, ARNTL and NPAS2, and the up-regulated genes, CRY2 and BHLHE41, could serve as the promising therapeutic targets for HF. Additionally, the significantly altered differential genes involved in AMPK and HIF-1α pathways were found in HF patients. The detection of clock genes involved in HF as a highly relevant topic that will exert implications for research and translational value with its diagnostic relevance.

There exist several limitations in the present study. We employed the data from publicly available databases, which resulted in a proneness to selection bias of the analysis. The DEGs were identified in the GSE57338 dataset, which provided the study samples mainly from heart tissue, while we employed it to validate the hub genes obtained from peripheral blood. It is required to verify the expression of hub genes in heart tissue samples, which may lead to the more reliable results. In addition, the number of enrolled patients was relatively limited here, which may affect study results to a certain extent. Therefore, more extensive prospective studies are required, as well as further experimental investigation to more comprehensively understand how ARNTL, CRY2, BHLHE41, and NPAS2 contribute to HF pathogenesis.

Conclusions

Our study demonstrated ARNTL, CRY2, BHLHE41 and NPAS2 as potential biomarkers of circadian rhythm in HF, with the potential to provide novel markers or targets for the diagnosis and treatment of HF.

Supplemental Information