Identify Cross Talk Between Circadian Rhythm and Coronary Heart Disease by Multiple Correlation Analysis

Abstract

Disorder in circadian rhythm has been revealed as a risk factor for coronary heart disease. Several studies in molecular biology established a gene interaction network using coronary heart susceptibility genes and the circadian rhythm pathway. However, cross talk between genes was mostly discovered in single gene pairs. There might be combination sets of genes intergraded as a unit to regulate the network. To resolve multiple variables in coronary heart susceptibility genes controlling circadian rhythm pathways, a multiple correlation analysis was applied to the transcriptome. Nine genes, including CUGBP, Elav-like family member ( CELF ); sodium leak channel, nonselective ( NALCN ); protein phosphatase 2 regulatory subunit B gamma ( PPP2R2C ); tubulin alpha 1c ( TUBA1C ); microtubule-associated protein 4 ( MAP4 ); cofilin 1 ( CFL1 ); myosin heavy chain 7 ( MYH7 ); QKI , KH domain containing RNA binding ( QKI ); and maternal embryonic leucine zipper kinase ( MELK ), from coronary heart susceptibility were identified to predict the outcome of a linear combination of circadian rhythm pathway genes with R factor more than 0.7. G protein subunit alpha o1 ( GNAO1 ), protein kinase C gamma ( PRKCG ), RBX , and G protein subunit beta 1 ( GNB1 ) in the circadian rhythm pathway are characterized as combination variables to coexpress with coronary heart susceptibility genes.

1. Introduction

Coronary heart disease (CHD) is one of the greatest threats to global human health. As a chronic disease, CHD is a complex result of both heritable and acquired risk factors. Genetic studies revealed that the heritability of CHD is between 40% and 60% and more than 100 susceptibility loci have been identified so far (McPherson and Tybjaerg-Hansen, 2016; Khera and Kathiresan, 2017). Epidemiological investigation suggested several risk factors for CHD assessment, including smoking, obesity, high blood lipids, hypertension, diabetes, and stress (Calderon et al., 1999; Kivimäki et al., 2012; Lozano et al., 2012; Finegold et al., 2013). The mechanisms underlying these biological processes have been characterized to explain the interactions among them. In all biological processes, there is a pivotal pathway named circadian rhythm that can cross talk with others by regulating the expression of most genes in the genome (Alloy et al., 2017; Hansen et al., 2017; Simpson et al., 2017).

In human and animal models, recent studies demonstrated that circadian rhythm is involved in the incidence of hypertension, coronary artery disease (CAD), dyslipidemia, and other cardiovascular pathologies by transcriptional regulation. Closely related to circadian rhythm, cardiovascular organs have diurnal rhythm, such as heart rate, blood pressure, and endothelial function. Disorders in circadian rhythm are usually the onset of cardiovascular disease such as acute coronary syndrome and atrial arrhythmia. Current evidences have supported that CHD pathology is tightly related to intrinsic circadian rhythms (Takeda and Maemura, 2011). The human heart has a rhythmic expression of clock genes, including PER1, PER2, and BMAL1, whereas the amplitudes are more different from any other human tissues (Leibetseder et al., 2009). In Drosophila heart, the CLOCK–CYCLE heterodimer exists as the circadian oscillator (Gummadova et al., 2009).

It is well known that circadian rhythm disturbances lead to mood disorders and delayed sleep onset (Kennaway, 2010). The clock gene Bmal1 has long been associated with hypertension and metabolic dysfunction (Miyazaki et al., 2011). Heart response to hypertrophic stimuli is regulated by cardiomyocyte circadian clock (Durgan et al., 2011). Clock mutant mice displayed dampened diurnal amplitudes of arterial pressure and heart rate (Sei et al., 2008). Clock mutant mice, Bmal1 (−/−), Clock (mut), and Npas2 (mut), which has clock gene deletion, showed abnormal changes in blood pressure (Curtis et al., 2007). There are clear effects of circadian rhythm in CHD (Steidle-Kloc et al., 2016). Long-term myocardial structure and function are adversely affected by circadian rhythm disorder after murine myocardial infarction (Alibhai et al., 2014). Age-associated dilated cardiomyopathy could be developed by loss of the circadian clock gene Bmal1 through altered myosin heavy chain gene expression and damage of sarcomere structure (Lefta et al., 2012).

In contrast, loss of the circadian rhythm gene mPer2 could attenuate myocardial injury in mice (Virag et al., 2010). Furthermore, circadian rhythms could be also involved in drug response in hypertension and CHD disease (Lemmer, 2006). In general, the circadian rhythm genes have been proved to interact with CHD. However, the potential preventative and therapeutic strategies for susceptible populations through the regulation of this interaction remain to be studied.

The genome extensive association analysis for CHD has revealed several susceptibility genes for CHD, which play a key role in diverse biological pathways such as inflammation, innate immunity, and cholesterol metabolism (Broeckel et al., 2002). Single-nucleotide polymorphisms (SNPs) on circadian rhythm genes also indicate that they are CHD susceptibility genes. However, other cross talk between susceptibility genes and circadian rhythm remains to be characterized.

The transcriptome network is an effective approach to dissect the cross talk between CHD susceptibility genes and circadian rhythm since circadian rhythm is involved in transcriptional regulation of the genome. Several CHD-related transcriptome studies were performed by chip and next-generation sequencing. For example, a microarray was applied to display the mouse heart transcriptome in various heart diseases (Hong et al., 2008). Similar studies had been done in humans to reveal the divergent cardiovascular disease pathways (Zhao et al., 2012). Transcriptome network analysis from these datasets identified a number of potential candidate genes for heart failure (Chen et al., 2013). In addition, many transcriptional regulators in human ischemic cardiomyopathy and coronary artery calcification were revealed from the gene expression network (Sen et al., 2014; Herrer et al., 2015). These works suggested the link between CHD genes and circadian rhythm by correlation analysis between genes.

So far, only single gene pairs were considered in these coexpression analyses. It is valuable to study the potential combination of genes associated with the downstream pathway. However, limited studies have developed methods to solve this problem. For instance, canonical correlation analysis was carried out to determine the genetic regulatory units in the Childhood Asthma Management Program (Naylor et al., 2010). The same method was performed on AIDS susceptibility genes and metabolic pathways (Ye et al., 2016). Hypergeometric distribution statistics were executed on the coexpression of mitochondrial respiratory complexes (Chen et al., 2014).

Multiple correlation analysis statistically calculates the coefficient that determines a given variable as a linear function of a set of variables. It is suitable for mining the potential combination of gene expression levels related to downstream genes. It could be used to calculate the correlation between genes and predict the linear combination from the gene expression. In this study, we conducted multiple correlation analysis to identify linear combinations of circadian rhythm genes that correlated with CHD susceptibility genes. We discussed the most significant cross talk between CHD and circadian rhythm under transcriptional regulation.

2. Methods

2.1. Transcriptome datasets

The normalized transcriptome microarray datasets were downloaded from the website (http://coxpresdb.jp), including about 4000 experiments and 20,000 human genes. Circadian-related genes were defined by KEGG annotation (www.keg.jp) as members of circadian rhythms and circadian entrainment pathways. From published genome-wide association studies (GWAS), it was identified that the CHD susceptibility gene had a significant p value (p < 10E-5). The expression patterns of circadian-related genes and CHD genes were extracted from transcriptome datasets.

2.2. Strategy

In this study, we performed all analyses following the strategy in Figure 1. First, we collected two lists of CHD susceptibility genes and circadian rhythm genes from published literature and databases. Second, the expression patterns of these genes were arranged from genome transcriptome datasets, which used microarrays to determine the expression levels of about 20,000 genes in 4000 experiments. Third, the Pearson correlation coefficient, r, was determined between all gene pairs in lists. Setting a threshold for r, the remaining coexpression levels were considered as links between genes in a network. Next, we defined the expression of each single CHD gene as variable y and expression of circadian groups as an independent variable X (x₁, x₂, x₃…x_n). Then, a plurality of correlation coefficients (R²) was calculated to estimate the combination of X, which could provide an output of variable y. The F-test was used to assess the correlation. The coefficient β factor represents the combination of X. Circadian genes with top β factor were considered as important regulators of a certain CHD gene. Using their β factor and expression of top circadian genes, we predict the expression of a certain CHD gene. Compared with real microarray data, CHD genes were selected for further analysis by transient expression of top circadian genes in cells.

FIG. 1.

Strategy of study. Collection of CHD genes and circadian pathway genes; construction of the coexpression network among individual CHD and circadian pathway genes; and performance of multiple correlation analysis. CHD, coronary heart disease.

2.3. Multiple correlation analysis

The coefficient of multiple correlation is used to statistically fit the correlation between a given variable and a set of other variables by using a linear function. The best fitting function was judged by the coefficient of multiple correlation R², which has values between 0 and 1. Higher value of R² indicates that the independent variable is more adaptable to the linear function than the dependent variable. R² revealed the strength of the association, but the direction could be provided by the simple correlation coefficient.

R² is determined by the vector c = (r_x_1y, r_x_2y, …, r_xNy)^T of each simple correlation coefficient between the independent variables x_n and dependent variable y. It is given by a formula: R² = c^TR_xx⁻¹c. R_xx⁻¹ is the inverse matrix: \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*} {R_{xx}} = \left( { \begin{matrix} {{r_{{x_1}{x_1}}}} & {{r_{{x_1}{x_2}}}} & \ldots & {{r_{{x_1}{x_N}}}} \\ {{r_{{x_2}{x_1}}}} & \ddots & {} & \vdots \\ \vdots & {} & \ddots & {} \\ {{r_{{x_N}{x_1}}}} & \cdots & {} & {{r_{{x_N}{x_N}}}} \\ \end{matrix} } \right) \end{align*} \end{document}

2.4. Cell culture and transfection

Human embryonic kidney (HEK) 293T cells were grown in Dulbecco's modified Eagle's medium (changed every 2 days), including 10% fetal bovine serum, with a 5% CO₂ atmosphere at 37°C. Cells were cultured in six-well plates for transfections using Lipofectamine 2000 following the manufacturer's protocol. Lipofectamine was cultured in serum-free medium. Next, appropriate amounts of vector and Lipofectamine were incubated together and added to cell plates.

2.5. RNA isolation, complementary DNA synthesis, and qRT-PCR (quantitative real-time polymerase chain reaction) analysis

Total RNAs were extracted from cells using TRIzol Reagent (Life Technologies) according to the manufacturer's protocol. From each sample, 1 μg of total RNA was used to generate complementary DNA (cDNA) using the First Strand cDNA Synthesis Kit (TOYOBO). The gene expression level was determined by quantitative polymerase chain reaction (qPCR) on cDNA using TransStart Top Green qPCR SuperMix (Transgen Biotech). Real-time PCRs were performed on the iQTM5 Detection System (Bio-Rad). qPCR cycling conditions were carried out as follows: 50°C for 2 minutes, initial denaturation at 95°C for 10 minutes, and 40 cycles at 95°C for 30 seconds and 55°C for 1 minute, and extension at 72°C for 30 seconds.

3. Results And Discussions

3.1. Genes involved in circadian rhythm and entrainment in CHD susceptibility

Genes related to circadian rhythm and circadian entrainment were collected following KEGG annotations. These circadian genes include the core clock genes such as CLOCK, Baml1, Period, and Cry, which established feedback loops to drive the rhythmic expression pattern of clock genes. Cryptochrome genes CRY and period genes PER code the components of a heterodimer, which could lead CLOCK/BMAL1 to repress their own expression. During these processes, casein kinase epsilon (CKI epsilon) phosphorylates PER and CRY and causes their degradation. This is the first negative feedback loop in regulation of the circadian rhythm core. The CLOCK/BMAL1 heterodimer could activate target genes with the regulatory enhancer as a positive feedback loop (Supplementary Table S1).

In circadian entrainment, signal transduction established a general pathway, including calcium influx. Diverse signal transduction cascades, such as calmodulin kinase II (CaMKII) and neuronal nitric oxide synthase activity, and cAMP- and cGMP-dependent protein kinases, as well as mitogen-activated protein kinase (MAPK), are triggered by glutamate NMDA-induced calcium influx and other second messengers. Phosphorylation of cAMP response element-binding protein (CREB) occurs in the interaction between these signal cascades, whereas the cross talk is not yet clear. Again, PER1 and PER2 expression levels were induced by pCREB binding on the CRE of their promoters (Golombek and Rosenstein, 2010).

3.2. Genes involved in CHD susceptibility

CHD susceptibility genes were collected from published GWAS, which identify SNPs associated with the CHD-related phenotype. In total, 373 SNPs had significant p value <10E5 in results of eight studies. As these SNPs are amplified in both coding regions and noncoding regions, we used SNPs located in coding regions to define the CHD susceptibility genes. In general, 120 genes involved in different biological processes were selected as CHD susceptibility genes (Supplementary Table S2). These genes were enriched in rhythmic, developmental, response to stimulus, and metabolic processes. It suggests that CHD susceptibility is regulated by circadian rhythm.

3.3. Coexpression network

Using Pearson correlation (r > 0.6) to define the link among circadian-related genes and CHD susceptibility genes, a coexpression network was constructed to show gene interactions in transcriptional regulation (Fig. 2). In general, 47 genes from circadian-related genes and CHD susceptibility genes were identified in this coexpression network, such as adenosine deaminase, RNA-specific B2 (ADARB2); adenylate cyclase 1 (ADCY1); adenylate cyclase 2 (ADCY2); ADCYAP receptor type I (ADCYAP1R1); chrom 14 open reading frame 119 (C14orf119); calcium voltage-gated channel subunit alpha1 I (CACNA1I); CUGBP, Elav-like family member 5 (CELF5); cofilin 1 (CFL1); DLG-associated protein 1 (DLGAP1); formin 2 (FMN2); G protein subunit alpha i3 (GNAI3); G protein subunit alpha o1 (GNAO1); G protein subunit beta 1 (GNB1); glutamate metabotropic receptor 7 (GRM7); guanylate cyclase 1 soluble subunit alpha (GUCY1A3); guanylate cyclase 1 soluble subunit beta (GUCY1B3); microtubule-associated protein 4 (MAP4); maternal embryonic leucine zipper kinase (MELK); sodium leak channel, nonselective (NALCN); nitric oxide synthase 1 (NOS1); proprotein convertase subtilisin/kexin type 2 (PCSK2); phosphatase and actin regulator 1 (PHACTR1); QKI, KH domain containing RNA binding (QKI); RASD family member 2 (RASD2); ryanodine receptor 2 (RYR2); seizure-related 6 homolog-like (SEZ6L); synaptotagmin 2 (SYT2); tubulin alpha 1c (TUBA1C); and WD repeat domain 12 (WDR12).

FIG. 2.

Coexpression network of CHD genes and circadian pathway genes. Pearson correlation factor r > 0.6.

There are many genes belonging to the same family, such as potassium voltage-gated channel subfamily J member 5/6/9 (KCNJ5/6/9), G protein subunit gamma 3/5/7 (GNG3/5/7), glutamate ionotropic receptor AMPA type subunit 1/2/3/4 (GRIA1/2/3/4), glutamate ionotropic receptor NMDA type subunit 1/2A/2C (GRIN1/2A/2C), calcium/calmodulin-dependent protein kinase II alpha (CAMK2A), calcium/calmodulin-dependent protein kinase II beta (CAMK2B), protein phosphatase 2 regulatory subunit B gamma (PPP2R2C), protein kinase C beta (PRKCB), and protein kinase C gamma (PRKCG). This coexpression network is constructed by 170 link edges (interactions) among these genes. Some genes have few edges that lead to isolation from the network, such as WDR12-MELK, GUCY1B3-GUCY1A3, TUBA1C-CFL1-GNB1, and GNAI3-GNG5-C140rf119. Some genes have multiple edges that lead to a complex network, such as GRIA1/2, ADCY1/2, KCNJ6, and CAMK2A. In all 47 genes in the network, 19 belong to CHD susceptibility genes and 28 are circadian related. Thus, this network provides new insight into interactions between CHD susceptibility genes and circadian-related genes, whereas the KEGG pathway could not reveal this important information.

3.4. Multiple correlations between CHD susceptibility genes and circadian genes CHD susceptibility gene

To identify the coexpression between CHD susceptibility genes and a combination of circadian-related genes, we performed multiple correlation analysis on each CHD susceptibility gene. In general, nine CHD susceptibility genes were determined with R² more than 0.7 (Table 1 and Fig. 3A). Results showed that more than seventy percent of expression variation of CHD susceptibility genes in different conditions could be explained by a combination of all circadian-related genes. These genes are CUGBP, CELF5; NALCN; PPP2R2C; TUBA1C; MAP4; CFL1; myosin heavy chain 7 (MYH7); QKI; and MELK. The F statistic value for these correlations is on top of all calculations with a value >80 (Fig. 3B). In addition, theɛvalue for these correlations is minimal in all, which indicates that there is less error in using this correlation to predict gene expression (Fig. 3C).

FIG. 3.

Statistics of multiple correlation analysis. (A) R², multiple correlation factor indicates the extent of correlation. (B) F-test. (C) ɛ, estimated error variance. (D) Correlation coefficient matrix for circadian pathway genes.

Table 1.

List of Coronary Heart Disease Susceptibility Genes Coexpressing Multiple Circadian Genes with High Multiple Correlation Factor (R² > 0.7)

Gene ID	Gene name	Gene name	R²	F	Espsilon
60680	CELF5	CUGBP, Elav-like family member 5 (CELF5)	0.8764	259.4751	0.0525
259232	NALCN	Sodium leak channel, nonselective (NALCN)	0.7935	140.6506	0.0449
5522	PPP2R2C	Protein phosphatase 2 regulatory subunit B gamma (PPP2R2C)	0.7587	115.0836	0.0701
84790	TUBA1C	Tubulin alpha 1c (TUBA1C)	0.7459	107.4623	0.1008
4134	MAP4	Microtubule-associated protein 4 (MAP4)	0.7441	106.4660	0.0571
1072	CFL1	Cofilin 1 (CFL1)	0.7431	105.8888	0.0950
4625	MYH7	Myosin heavy chain 7 (MYH7)	0.7372	102.6823	0.1990
9444	QKI	QKI, KH domain containing RNA binding (QKI)	0.7292	98.5615	0.0756
9833	MELK	Maternal embryonic leucine zipper kinase (MELK)	0.7036	86.8806	0.3615

We summarized gene function from published literature. NALCN codes proteins as a component of the voltage-independent, nonselective cation channel, which plays an important role in resting sodium permeability for normal respiratory rhythm (Wang and Ren, 2009). It works with UNC80 to activate the leak cation channel in a G-protein-independent manner (Lu et al., 2007). M3 muscarinic receptors control this ion channel in pancreatic beta cells. NALCN could also respond to tetrodotoxin, which was implicated to be involved in neuronal excitability (Aslam, 2009). PPP2R2C is located in human chromosome 4p16, which is a brain abundant protein (Hu et al., 2000). Mutation in PPP2R2C leads to intellectual disability (Backx et al., 2010). It acts as a suppressor of tumorigenesis and as a regulatory subunit of the PP2 complex, which is particularly induced by viral transformation (Banerjee et al., 2007). COFILIN works in cell migration as it affects F-actin fibers and cell membrane (Ashworth et al., 2010). Dephosphorylation and activation of cofilin could control actin reorganization and polymerization (Wang et al., 2010).

This gene activity is associated with Aur-A overexpression in breast cancer tissues (Eijgelsheim et al., 2010). MYH7 has been identified with SNPs associated with resting heart rate in recent studies (Eijgelsheim et al., 2010). In Laing distal myopathy, tail mutation in MYH7 leads to myopathic profiles (Muelas et al., 2010). In embryonic and neural stem cells, MELK is a marker for cell survival, proliferation, and apoptosis (Choi and Ku, 2011). In general, the relationship between the CHD susceptibility genes and circadian rhythm is unclear. Our results provide the first clue for this interaction.

3.5. Circadian-related genes

Then, β factors, which are coefficients of circadian-related genes, were analyzed to reveal all possible linear combinations of circadian-related gene expression (Fig. 3D). Top four combinations were further collected in terms of high R² and statistical significance such as CELF5, NALCN, PPP2R2C, and TUBA1C (Fig. 4 and Table 2). Top two circadian-related genes with positive β factor coefficients were selected to make the linear combination. These are GNAO1 (2775) and PRKCG (5582), PRKG1 (5592) and GNAO1, ADCYAP1R1 (117) and GNAO1, and RBX1 (9978) and GNB1 (2782) for CELF5, NALCN, PPP2R2C, and TUBA1C, respectively. Using two circadian-related genes, we predict the output of the linear combination. Compared with true values of CHD susceptibility genes, the linear combination could provide significant prediction (Fig. 5). Thus, in each multiple correlation, these four CHD susceptibility gene expression levels could be correlated with a combination of two circadian-related genes.

FIG. 4.

Coefficient for circadian genes with high correlation factors (R² > 0.7).

FIG. 5.

Multiple correlations predicted the expression. The scatter points are the true levels of gene expression, the net is predicted by linear regression of multiple correlations.

Table 2.

Multiple Correlation Coefficient of Coronary Heart Disease Susceptibility Genes as Linear Combinations to Coexpress with Circadian Genes

	Circadian genes
CHD gene ID	CELF5	NALCN	PPP2R2C	TUBA1C
Constant interval	0.0000	0.0000	0.0000	0.0000
107	0.0243	−0.0077	0.0383	0.0324
108	0.0309	−0.0040	0.0069	−0.0487
109	−0.0198	0.0278	0.0466	0.1031
111	0.0869	0.0153	0.0463	0.0119
112	−0.0264	−0.0145	0.0155	0.0388
113	0.0280	0.0709	0.0881	0.0174
115	0.0364	−0.0026	−0.0316	−0.0442
117	0.0055	0.0600	0.1550	0.0500
406	0.0196	0.0025	−0.0185	−0.0458
775	−0.0961	−0.0563	0.0686	0.0694
776	−0.0026	0.0462	0.0674	−0.0356
810	−0.0469	−0.0501	0.1420	0.0447
815	0.0617	−0.0728	0.0956	−0.1168
816	0.0933	0.0580	0.0839	−0.0007
817	0.0159	−0.0009	0.0148	−0.0232
818	−0.0274	0.0102	0.0442	0.0763
1385	−0.0206	−0.0465	−0.0325	−0.1170
1407	0.0145	0.0012	−0.0126	0.0685
1408	0.0203	0.0061	−0.0778	−0.0715
1453	−0.0203	−0.0248	0.0871	0.1175
1454	0.0774	0.0246	0.0406	0.0658
2353	−0.0166	0.0023	−0.0045	−0.0022
2770	−0.0142	0.0293	0.0077	0.0310
2771	−0.0664	0.0267	−0.0356	0.0499
2773	0.0125	0.0058	0.0268	0.1076
2775	0.1537	0.1921	0.2329	0.2072
2776	0.0537	0.0595	0.0473	0.0600
2778	−0.0320	−0.0005	−0.0511	0.0693
2782	0.0292	−0.0185	0.0271	0.2635
2783	−0.0041	−0.0347	0.0217	0.0224
2784	−0.0280	0.0432	−0.0528	−0.0965
2785	0.0177	0.0258	0.0274	0.1053
2786	0.1028	0.0031	0.0148	−0.0081
2787	−0.0073	−0.0766	−0.0620	−0.0154
2788	−0.0269	0.0728	0.0099	−0.1056
2790	0.0046	0.0120	−0.0172	−0.0517
2791	−0.0030	−0.0091	0.0035	−0.0227
2792	0.0078	0.0364	0.0210	0.0107
2793	−0.0447	0.0099	−0.0017	0.0074
2890	0.1067	−0.0329	−0.1057	−0.0349
2891	0.0202	0.0202	0.0217	−0.0149
2892	0.1155	0.0841	0.0471	0.0380
2893	−0.0170	0.0870	−0.0705	0.0572
2902	0.0721	−0.0366	−0.0785	−0.0680
2903	0.0172	0.0384	−0.0262	0.0256
2904	0.1379	0.0224	0.0029	0.1474
2905	0.0613	−0.0340	−0.0101	−0.1030
2906	0.1023	0.0792	−0.0026	0.0259
2977	−0.0147	−0.0061	0.0728	0.0525
2982	−0.0309	0.0040	0.0209	−0.0729
2983	0.0681	−0.0217	−0.0300	0.0530
3708	0.0242	−0.0292	−0.0346	−0.0669
3710	−0.0750	0.0105	−0.0192	0.0026
3760	−0.0540	−0.0234	−0.0185	−0.0084
3762	−0.1478	−0.0742	−0.1597	−0.0613
3763	−0.0548	0.0261	0.0261	−0.0053
3765	−0.0547	0.0800	0.0142	0.0118
4543	0.0759	0.0433	−0.0146	−0.0461
4544	−0.0991	0.0196	−0.0014	0.0587
4842	−0.0607	−0.0474	0.0086	−0.0550
4862	0.0294	0.0389	−0.0059	−0.0157
5187	0.0087	0.0024	−0.0946	−0.0690
5330	−0.1017	−0.0365	−0.1063	−0.0514
5331	−0.0412	−0.0410	0.0284	−0.0301
5332	0.0124	0.0074	0.0062	0.0154
5562	−0.0350	−0.0209	−0.0941	−0.0732
5563	−0.0189	−0.0330	0.0301	0.0077
5564	0.0016	0.0053	−0.0331	−0.0167
5565	−0.0408	0.0255	−0.0176	−0.0267
5566	0.0239	0.0205	−0.1345	0.0838
5567	0.0031	0.0210	0.0147	−0.0098
5568	0.0148	0.0166	0.1311	0.0891
5571	−0.0134	−0.0202	−0.0067	0.0865
5578	−0.0147	0.0527	−0.0447	0.2059
5579	0.0881	−0.0347	0.0165	0.0291
5582	0.2182	0.0564	0.0500	0.0533
5592	−0.0716	0.0882	−0.0216	0.0146
5593	−0.0404	−0.0279	−0.0817	−0.0893
5594	0.0846	−0.0287	0.0172	0.0219
5595	0.0106	0.0041	0.0594	0.1471
6095	−0.0384	0.0244	0.0606	−0.1418
6096	0.0105	−0.0240	0.0752	−0.0038
6097	−0.0041	−0.0198	−0.0464	−0.1186
6261	0.0293	0.0149	−0.0067	−0.0079
6262	0.0181	0.0456	−0.0332	0.0024
6263	−0.0023	−0.0167	0.0196	0.0224
6500	−0.0075	0.0330	0.0790	0.0774
8454	−0.0116	−0.0312	0.0032	0.0772
8553	−0.0192	−0.0014	0.0027	−0.0054
8863	−0.0019	−0.0039	−0.0006	−0.1015
8864	0.0215	0.0121	0.0074	−0.0056
8911	0.1489	0.0257	0.1355	−0.0589
8912	−0.0117	−0.0243	0.1249	−0.0409
8913	0.1213	−0.0329	0.0538	−0.0034
8945	0.0558	0.0763	0.1312	0.0541
9252	−0.0530	−0.0274	−0.0012	0.0593
9575	0.0125	−0.0022	0.0064	−0.0042
9722	−0.0028	0.0155	−0.0932	0.0429
9978	−0.0375	0.0038	0.0065	0.2098
10681	0.0177	0.0148	0.0584	−0.0469
23236	0.0194	−0.0276	−0.0189	−0.0471
23291	−0.0479	0.0116	0.0833	−0.0082
26224	0.0202	0.0344	−0.0199	−0.1295
51422	−0.0057	−0.0265	−0.0324	−0.0576
51655	−0.0061	−0.0021	0.0094	−0.0188
51764	−0.0180	−0.0548	0.0276	−0.0132
51806	0.0169	−0.0017	0.0161	−0.0061
53632	−0.0361	−0.0091	−0.0369	0.0239
54331	−0.0271	−0.0172	−0.0876	−0.0414
55811	0.0465	0.0254	0.0866	−0.0184
55970	0.0092	−0.0203	−0.0069	0.0101
59345	−0.0329	0.0393	0.0198	0.1699
79365	−0.0286	0.0370	0.0006	0.0252
91860	−0.0230	−0.0559	−0.0568	0.0470
94235	0.0061	0.0664	0.0376	0.0364
163688	−0.0181	−0.0123	0.0277	0.0035

CHD, coronary heart disease.

GNAO1 encoded a component of the Go heterotrimeric G-protein signal-transducing complex, which plays a role in early-onset epileptic encephalopathy. In addition, studies found that GNAO1 mutation variants cause severe developmental delay that is associated with epileptic encephalopathy (Saitsu et al., 2016). Infantile-onset epilepsy has been proved to be the result of a missense mutation in GNAO1 in a clinical investigation (Law et al., 2015). Deep brain stimulation is related to the GNAO1 mutation in progressive dyskinesia (Kulkarni et al., 2016). Importantly, recent studies revealed the relationship between GNAO1 and CHD. Cardiac ventricular excitability is regulated by GαO-mediated signaling. Furthermore, cardiovascular reflexes require GαO-mediated signaling in the rostral ventrolateral medulla oblongata. There is a circuit comprising GαO-mediated signaling and RVLM, which plays the role of autonomic regulator of the heart. Mutation in GαO has a significant effect on cardiovascular-related processes such as responses to stress and electrical properties (Richard et al., 2016).

PRKCG belongs to the serine- and threonine-specific protein kinase family, which plays a key role with calcium and second messenger diacylglycerol. Research on osteosarcoma risk revealed an interaction between PRKCG and osteosarcoma. Its susceptibility could be affected by PRKCG SNP (Zhang et al., 2014; Lu et al., 2015). In the mouse model, Purkinje cell pathology is induced by overactivation of PRKCG (Ji et al., 2014). The role of PRKCG in CHD has been revealed in recent studies on hypobaric hypoxia-induced cardiac hypertrophy in rats (Uenoyama et al., 2010). In addition, PRKCG is involved in regulation of signaling in mouse coronary artery smooth muscle cells (Ansari et al., 2009). It was shown that PRKCG could be a candidate controller of contractility and heart failure susceptibility (Liu and Chen, 2009). Furthermore, human coronary microcirculation requires PRKCG to mediate microvascular reactivity (Sodha et al., 2007). Myocyte contractility after cardioplegic arrest also requires PRKCG function (Mukherjee et al., 2006).

PRKG1 works in the nitric oxide/cGMP signaling pathway and many signal transductions. In pancreatic adenocarcinoma, an early-onset thoracic aortic disease, mutation on PRKG1 has been shown as a genetic marker (Karakhanova et al., 2014; Gago-Díaz et al., 2016). It was also believed that PRKG1 participates in CHD by certain mechanisms such as cardioprotection through prevention of PKG1α oxidation (Lawliss et al., 2015). Pulmonary vascular tone and hypoxic pulmonary hypertension are currently proved to be related to PRKG1 (Ramchandran et al., 2014). In cardiomyocytes, PRKG1 could affect cardiac hypertrophy through inhibiting apoptosis signaling (Fiedler et al., 2006; Lukowski et al., 2010).

ADCYAP1R1 is a homologous gene of the glucagon/secretin receptor family with an important role in the fear responses of the amygdala and hippocampus (Stevens et al., 2014). There is significant interaction between ADCYAAP1R1 and CHD. Deficiency of ADCYAAP1R1 in mice leads to pulmonary hypertension and right heart failure (Otto et al., 2004). Guinea pig cardiac neuron excitability could be induced by ADCYAAP1R1 through internalization and endosomal signaling (Merriam et al., 2013). It was well known that ADCYAAP1R1 is involved in activation of MEK/ERK signaling to induce guinea pig cardiac neuron excitability (Tompkins et al., 2016).

RBX1 is involved in ubiquitination processes by interaction with cullin proteins for cell cycle progression in which mutation could cause hypertension and hyperkalemia (Schumacher et al., 2015). Nonmuscle-invasive bladder transitional cell carcinoma showed overexpression of RBX1 (Wei et al., 2013). Nonsmall cell lung cancer also takes RBX1 as a prognostic factor (Xing et al., 2016). Related to CHD is hyperosmotic stress of salmon induced by RBX1 expression (Pan et al., 2002). It also controls pulmonary arterial hypertension through Ca²⁺ signaling.

GNB1 is a heterotrimeric guanine nucleotide-binding protein with function in signals between receptors and effectors. Recent studies suggested that it could be a target of urocortin-mediated cardioprotection (Barry et al., 2010).

3.6. Validation

To validate in silico results by multiple correlation analysis, we performed qRT-PCR to check four expression levels of CHD susceptibility genes after transient expression of a combination of multiple correlated circadian-related genes in 293 cells. First, each circadian-related gene was separately transiently overexpressed by transfection. Then, two circadian-related genes were treated at the same time. Results showed that the expression of CEFL5 was increased about 2 times when 293 cells overexpressed 2775 (GNAO1) or 5582 (PRKCG) only. It was increased 7 times when 293 cells overexpressed both 2775 (GNAO1) and 5582 (PRKCG) (Fig. 6A). In another case, the expression of either 5592 (PRKG1) or 2775 (GNAO1) was induced, and NALCN expression was not increased (Fig. 6B). While expression of TUBA1C was increased about 3 times when 293 cells overexpressed only 9978 (RBX1) or 2782 (GNB1), it was increased 6 times when 293 cells overexpressed both 9978 (RBX1) and 2782 (GNB1) (Fig. 6C). The expression levels of CEFL5 and TUBA1C matched with multiple correlation analysis results. It indicates that there is transcriptional regulation from 2775 (GNAO1) and 5582 (PRKCG), 9978 (RBX1) and 2782 (GNB1), to CEFL5 and TUBA1C, respectively.

FIG. 6.

Validation of multiple correlations by RT-PCR after transient expression of combination gene sets. (A) Expression level of CEFL after transient expression of individual GNAO1, PRKCG, and both. (B) Expression level of NALCN after transient expression of individual PRKG1, GNAO1, and both. (C) Expression level of TUBA1C after transient expression of individual RBX1, GNB1, and both. GNAO1, G protein subunit alpha o1; GNB1, G protein subunit beta 1; NALCN, sodium leak channel, nonselective; PRKCG, protein kinase C gamma; TUBA1C, tubulin alpha 1c.

Furthermore, this regulation could be added by combining two circadian-related genes. On the other side, expression of NALCN does not meet the prediction from multiple correlation analysis. That could be caused by the different down–upstream position or other actual errors in the experiments. However, multiple correlation analysis could provide an insight into the combination of transcriptional regulation. This will benefit detailed construction of gene interaction between CAD susceptibility genes and circadian-related genes.

Footnotes

Author Disclosure Statement

The authors declare they have no financial conflicts of interest.

References

Alibhai

F.J.

, Tsimakouridze

E.V.

, Chinnappareddy

, et al. 2014. Short-term disruption of diurnal rhythms after murine myocardial infarction adversely affects long-term myocardial structure and function. Circ. Res. 114, 1713–1722.

Alloy

L.B.

, Ng

T.H.

, Titone

M.K.

, et al. 2017. Circadian rhythm dysregulation in bipolar spectrum disorders. Curr. Psychiatry Rep. 19, 21.

Ansari

H.R.

, Teng

, Nadeem

, et al. 2009. A(1) adenosine receptor-mediated PKC and p42/p44 MAPK signaling in mouse coronary artery smooth muscle cells. Am. J. Physiol. Heart Circ. Physiol. 297, H1032–H1039.

Ashworth

, Teng

, Kaufeld

, et al. 2010. Cofilin-1 inactivation leads to proteinuria—studies in zebrafish, mice and humans. PloS One, 5, e12626.

Aslam

2009. The NALCN ion channel is activated by M3 muscarinic receptors in a pancreatic beta-cell line. EMBO Rep. 10, 873–880.

Backx

, Vermeesch

, Pijkels

, et al. 2010. PPP2R2C, a gene disrupted in autosomal dominant intellectual disability. Eur. J. Med. Genet. 53, 239–243.

Banerjee

A.K.

, Read

C.A.

, Griffiths

M.H.

, et al. 2007. Clonal divergence in lung cancer development is associated with allelic loss on chromosome 4. Genes Chromosomes Cancer, 46, 852–860.

Barry

S.P.

, Lawrence

K.J.

, Soond

S.M.

, et al. 2010. New targets of urocortin-mediated cardioprotection. J. Mol. Endocrinol. 45, 69–85.

Broeckel

, Hengstenberg

, Mayer

, et al. 2002. A comprehensive linkage analysis for myocardial infarction and its related risk factors. Nat. Genet. 30, 210–214.

10.

Calderon

Jr. , Schneider

R.H.

, Alexander

C.N.

, et al. 1999. Stress, stress reduction and hypercholesterolemia in African Americans: a review. Ethnic Dis. 9, 451–462.

11.

Chen

, Hyun

T.K.

, Han

, et al. 2014. Coexpression within integrated mitochondrial pathways reveals different networks in normal and chemically treated transcriptomes. Int. J. Genomics, 2014, 452891.

12.

Chen

, Wang

H.Y.

, and Zeng

C.Y.

2013. Transcriptome network analysis of potential candidate genes for heart failure. Genet. Mol. Res. 12, 4687–4697.

13.

Choi

, and Ku

J.L.

2011. Resistance of colorectal cancer cells to radiation and 5-FU is associated with MELK expression. Biochem. Biophys. Res. Commun. 412, 207–213.

14.

Curtis

A.M.

, Cheng

, Kapoor

, et al. 2007. Circadian variation of blood pressure and the vascular response to asynchronous stress. Proc. Natl. Acad. Sci. U. S. A. 104, 3450–3455.

15.

Durgan

D.J.

, Tsai

J.Y.

, Grenett

M.H.

, et al. 2011. Evidence suggesting that the cardiomyocyte circadian clock modulates responsiveness of the heart to hypertrophic stimuli in mice. Chronobiol. Int. 28, 187–203.

16.

Eijgelsheim

, Newton-Cheh

, Sotoodehnia

, et al. 2010. Genome-wide association analysis identifies multiple loci related to resting heart rate. Hum. Mol. Genet. 19, 3885–3894.

17.

Fiedler

, Feil

R.F.

, Willenbockel

, et al. 2006. cGMP-dependent protein kinase type I inhibits TAB1-p38 mitogen-activated protein kinase apoptosis signaling in cardiac myocytes. J. Biol. Chem. 281, 32831–32840.

18.

Finegold

J.A.

, Asaria

, and Francis

D.P.

2013. Mortality from ischaemic heart disease by country, region, and age: statistics from World Health Organisation and United Nations. Int. J. Cardiol. 168, 934–945.

19.

Gago-Díaz

, Blanco-Verea

, Teixidó

, et al. 2016. PRKG1 and genetic diagnosis of early-onset thoracic aortic disease. Eur. J. Clin. Invest. 46, 787–794.

20.

Golombek

D.A.

, and Rosenstein

R.E.

2010. Physiology of circadian entrainment. Physiol. Rev. 90, 1063–1102.

21.

Gummadova

J.O.

, Coutts

G.A.

, and Glossop

N.R.

2009. Analysis of the Drosophila Clock promoter reveals heterogeneity in expression between subgroups of central oscillator cells and identifies a novel enhancer region. J. Biol. Rhythm. 24, 353–367.

22.

Hansen

S.N.

, Peics

, and Gerhart-Hines

2017. Keeping fat on time: circadian control of adipose tissue. Exp. Cell. Res. 360, 31–34.

23.

Herrer

, Roselló-Lletí

, Ortega

, et al. 2015. Gene expression network analysis reveals new transcriptional regulators as novel factors in human ischemic cardiomyopathy. BMC Med. Genomics. 8, 14.

24.

Hong

S.E.

, Park

, Cha

, et al. 2008. Identification of mouse heart transcriptomic network sensitive to various heart diseases. Biotech. J. 3, 648–658.

25.

, Yu

, Zhang

, et al. 2000. Molecular cloning and mapping of the brain-abundant B1gamma subunit of protein phosphatase 2A, PPP2R2C, to human chromosome 4p16. Genomics, 67, 83–86.

26.

, Hassler

M.L.

, Shimobayashi

, et al. 2014. Increased protein kinase C gamma activity induces Purkinje cell pathology in a mouse model of spinocerebellar ataxia 14. Neurobiol. Dis. 70, 1–11.

27.

Karakhanova

, Golovastova

, Philippov

P.P.

, et al. 2014. Interlude of cGMP and cGMP/protein kinase G type 1 in pancreatic adenocarcinoma cells. Pancreas, 43, 784–794.

28.

Kennaway

D.J.

2010. Clock genes at the heart of depression. J. Psychopharmacol. (Oxf). 24, 5–14.

29.

Khera

A.V.

, and Kathiresan

2017. Genetics of coronary artery disease: discovery, biology and clinical translation. Nat. Rev. Genet. 18, 331–344.

30.

Kivimäki

, Nyberg

S.T.

, Batty

G.D.

, et al. 2012. Job strain as a risk factor for coronary heart disease: a collaborative meta-analysis of individual participant data. Lancet, 380, 1491–1497.

31.

Kulkarni

, Tang

, Bhardwaj

, et al. 2016. Progressive movement disorder in brothers carrying a GNAO1 mutation responsive to deep brain stimulation. J. Child Neurol. 31, 211–214.

32.

Law

C.Y.

, Chang

T.L.

, Sun

Y.C.

, et al. 2015. Clinical whole-exome sequencing reveals a novel missense pathogenic variant of GNAO1 in a patient with infantile-onset epilepsy. Clin. Chim. Acta, 451, 292–296.

33.

Lawliss

M.S.

, Rocque

J.M.

, Angelopoulos

, et al. 2015. Prevention of PKG1alpha oxidation augments cardioprotection in the stressed heart. J. Clin. Invest. 125, 2468–2472.

34.

Lefta

, Campbell

K.S.

, Feng

H.Z.

, et al. 2012. Development of dilated cardiomyopathy in Bmal1-deficient mice. Am. J. Physiol. Heart Circ. Physiol. 303, H475–H485.

35.

Leibetseder

, Humpeler

, Svoboda

, et al. 2009. Clock genes display rhythmic expression in human hearts. Chronobiol. Int. 26, 621–636.

36.

Lemmer

2006. The importance of circadian rhythms on drug response in hypertension and coronary heart disease—from mice and man. Pharmacol. Ther. 111, 629–651.

37.

Liu

, and Chen

X.S.

2009. Protein kinase C{alpha}, but not PKC{beta} or PKC{gamma}, regulates contractility and heart failure susceptibility: implications for ruboxistaurin as a novel therapeutic approach. Circ. Res. 105, 194–200.

38.

Lozano

, Naghavi

, Foreman

, et al. 2012. Global and regional mortality from 235 causes of death for 20 age groups in 1990 and 2010: a systematic analysis for the Global Burden of Disease Study 2010. Lancet, 380, 2095–2128.

39.

, Su

, Das

, et al. 2007. The neuronal channel NALCN contributes resting sodium permeability and is required for normal respiratory rhythm. Cell, 129, 371–383.

40.

, Zhu

, Lian

, et al. 2015. Genetic variations in the PRKCG gene and osteosarcoma risk in a Chinese population: a case-control study. Tumour Biol. 36, 5241–5247.

41.

Lukowski

, Rybalkin

S.D.

, Loga

, et al. 2010. Cardiac hypertrophy is not amplified by deletion of cGMP-dependent protein kinase I in cardiomyocytes. Proc. Natl. Acad. Sci. U. S. A. 107, 5646–5651.

42.

McPherson

, and Tybjaerg-Hansen

2016. Genetics of coronary artery disease. Circ. Res. 118, 564–578.

43.

Merriam

L.A.

, Baran

C.N.

, Girard

B.M.

, et al. 2013. Pituitary adenylate cyclase 1 receptor internalization and endosomal signaling mediate the pituitary adenylate cyclase activating polypeptide-induced increase in guinea pig cardiac neuron excitability. J. Neurosci. 33, 4614–4622.

44.

Miyazaki

, Schroder

, Edelmann

S.E.

, et al. 2011. Age-associated disruption of molecular clock expression in skeletal muscle of the spontaneously hypertensive rat. PloS One, 6, e27168.

45.

Muelas

, Hackman

, Luque

, et al. 2010. MYH7 gene tail mutation causing myopathic profiles beyond Laing distal myopathy. Neurology, 75, 732–741.

46.

Mukherjee

, Apple

K.A.

, Squires

C.E.

, et al. 2006. Protein kinase C isoform activation and endothelin-1 mediated defects in myocyte contractility after cardioplegic arrest and reperfusion. Circulation, 114, 308–313.

47.

Naylor

M.G.

, Lin

, Weiss

S.T.

, et al. 2010. Using canonical correlation analysis to discover genetic regulatory variants. PloS One, 5, e10395.

48.

Otto

, Hein

, Brede

, et al. 2004. Pulmonary hypertension and right heart failure in pituitary adenylate cyclase-activating polypeptide type I receptor-deficient mice. Circulation, 110, 3245–3251.

49.

Pan

, Zarate

, and Bradley

T.M.

2002. A homolog of the E3 ubiquitin ligase Rbx1 is induced during hyperosmotic stress of salmon. Am. J. Physiol. Regul. Integr. Comp. Physiol. 282, R1643–R1653.

50.

Ramchandran

, Raghavan

, Geenen

, et al. 2014. PKG-1alpha leucine zipper domain defect increases pulmonary vascular tone: implications in hypoxic pulmonary hypertension. Am. J. Physiol. Lung Cell. Mol. Physiol. 307, L537–L544.

51.

Richard

, Joel

, Lutz

, et al. 2016. The role of GalphaO-mediated signaling in the rostral ventrolateral medulla oblongata in cardiovascular reflexes and control of cardiac ventricular excitability. Physiol. Rep. 4, e12860.

52.

Saitsu

, Fukai

, Benzeev

, et al. 2016. Phenotypic spectrum of GNAO1 variants: epileptic encephalopathy to involuntary movements with severe developmental delay. Eur. J. Hum. Genet. 24, 129–134.

53.

Schumacher

F.R.

, Siew

, Zhang

, et al. 2015. Characterisation of the Cullin-3 mutation that causes a severe form of familial hypertension and hyperkalaemia. EMBO Mol. Med. 7, 1285–1306.

54.

Sei

, Oishi

, Chikahisa

, et al. 2008. Diurnal amplitudes of arterial pressure and heart rate are dampened in Clock mutant mice and adrenalectomized mice. Endocrinology, 49, 3576–3580.

55.

Sen

S.K.

, Barb

J.J.

, Cherukuri

P.F.

, et al. 2014. Identification of candidate genes involved in coronary artery calcification by transcriptome sequencing of cell lines. BMC Genomics, 15, 198.

56.

Simpson

S.D.

, Ramsdell

J.S.

, III, Watson

W.H.

, et al. 2017. The draft genome and transcriptome of the Atlantic horseshoe crab, Limulus polyphemus. Int. J. Genomics, 2017, 1–14.

57.

Sodha

N.R.

, Feng

, Clements

R.T.

, et al. 2007. Protein kinase C alpha modulates microvascular reactivity in the human coronary and skeletal microcirculation. Surgery, 142, 243–252.

58.

Steidle-Kloc

, Schönfelder

, Müller

, et al. 2016. Does exercise training impact clock genes in patients with coronary artery disease and type 2 diabetes mellitus?. Eur. J. Prev. Cardiol. 23, 1375–1382.

59.

Stevens

J.S.

, Almli

L.M.

, Fani

, et al. 2014. PACAP receptor gene polymorphism impacts fear responses in the amygdala and hippocampus. Proc. Natl. Acad. Sci. U. S. A. 111, 3158–3163.

60.

Takeda

, and Maemura

2011. Circadian clock and cardiovascular disease. J. Cardiol. 57, 249–256.

61.

Tompkins

J.D.

, Clason

T.A.

, Hardwick

J.C.

, et al. 2016. Activation of MEK/ERK signaling contributes to the PACAP-induced increase in guinea pig cardiac neuron excitability. Am. J. Physiol. Cell Physiol. 311, C643–C651.

62.

Uenoyama

, Ogata

, Nakanishi

, et al. 2010. Protein kinase C mRNA and protein expressions in hypobaric hypoxia-induced cardiac hypertrophy in rats. Acta Physiol. (Oxf). 198, 431–440.

63.

Virag

J.A.

, Dries

J.L.

, Easton

P.R.

, et al. 2010. Attenuation of myocardial injury in mice with functional deletion of the circadian rhythm gene mPer2. Am. J. Physiol. Heart Circ. Physiol. 298, H1088–H1095.

64.

Wang

, and Ren

2009. UNC80 functions as a scaffold for Src kinases in NALCN channel function. Channels, 3, 161–163.

65.

Wang

L.H.

, Xiang

, Yan

, et al. 2010. The mitotic kinase Aurora-A induces mammary cell migration and breast cancer metastasis by activating the Cofilin-F-actin pathway. Cancer Res. 70, 9118–9128.

66.

Wei

W.M.

, Qiu

J.X.

, Liu

Z.H.

, et al. 2013. Overexpression of RING box protein-1 (RBX1) associated with poor prognosis of non-muscle-invasive bladder transitional cell carcinoma. J. Surg. Oncol. 107, 758–761.

67.

Xing

, Chen

K.B.

, Xuan

, et al. 2016. RBX1 expression is an unfavorable prognostic factor in patients with non-small cell lung cancer. Surg. Oncol. 25, 147–151.

68.

, Yuan

, Wang

, et al. 2016. A canonical correlation analysis of AIDS restriction genes and metabolic pathways identifies purine metabolism as a key cooperator. Comput. Math. Methods Med. 2016, 1–10.

69.

Zhang

, Hu

, Wang

H.K.

, et al. 2014. Single-nucleotide polymorphisms of the PRKCG gene and osteosarcoma susceptibility. Tumour Biol. 35, 12671–12677.

70.

Zhao

, Sheng

, and Huang

2012. A systematic analysis of heart transcriptome highlights divergent cardiovascular disease pathways between animal models and humans. Mol. Biosyst. 8, 504–510.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.09 MB

0.05 MB