Development of the correlation between mitophagy-related genes and epilepsy

doi:10.21203/rs.3.rs-3595537/v1

Download PDF

Research Article

Development of the correlation between mitophagy-related genes and epilepsy

https://doi.org/10.21203/rs.3.rs-3595537/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Background

Approximately 30% of patients with epilepsy are refractory to medication. Dysregulation of mitochondrial autophagy in the central nervous system may be associated with the pathogenesis of epilepsy. Therefore, we conducted a global analysis to elucidate the biological roles of mitophagy-related genes in epilepsy.

Methods

We downloaded the GSE143272 dataset from the Gene Expression Omnibus (GEO) database and identified mitophagy-related genes (MRGs) from the Molecular Signatures Database. The differentially expressed MRGs between drug-naïve patients with epilepsy and healthy individuals were identified. A protein–protein interaction (PPI) network and diagnostic model were then constructed. Subsequently, we performed functional enrichment and clustering analysis to identify the different epilepsy subtypes. Weighted gene co-expression network analysis was used to identify significant modules closely related to epilepsy, and a PPI network was co-constructed with MRGs to obtain hub genes. Finally, we used CIBERSORT to further investigate the distribution pattern of peripheral blood immune cell subtypes in patients with epilepsy, as well as the correlation between hub genes, MRG expression, and immune cell infiltration.

Results

We identified 11 differentially expressed MRGs (5 upregulated and 6 downregulated). The MRG diagnostic model used in our study showed good diagnostic performance. We identified the top six significant modules that were closely related to epilepsy. By co-constructing PPI with MRGs, we obtained the top 10 hub genes and constructed a competing endogenous RNA (ceRNA) network. Furthermore, CIBERSORT analysis indicated that patients with epilepsy had a higher infiltration level of neutrophils, and the hub genes and differentially expressed MRGs were highly correlated with infiltrating immune cells.

Conclusions

Our findings highlight that mitophagy is associated with epilepsy and may provide a novel direction for the diagnosis and treatment of the disease.

bioinformatics

mitophagy-related genes

epilepsy

Epilepsy is one of the most prevalent neurological diseases, affecting over 70 million individuals worldwide ^[1]. Characterized by repeated seizures, it is often accompanied by emotional and cognitive dysfunction. Epilepsy can be categorized into different forms, and complicated clinical manifestations add to the difficulty of making an initial diagnosis. Once diagnosed, most patients receive anti-epileptic drugs (AEDs), such as carbamazepine, valproate, and levetiracetam. However, studies have demonstrated that AEDs only suppress seizures and do not inhibit the progression of neurological disorders ^[2], and one-third of patients might develop drug resistance ^[1]. Consequently, there is an urgent need to gain further insights into the pathogenesis of epilepsy, which may facilitate the identification of promising novel diagnostic biomarkers and therapeutic strategies for epilepsy.

Mitochondria are essential organelles that are classically seen as the powerhouse of the cell ^[3] and are the main source of reactive oxygen species (ROS) ^[4]. Mitophagy is a mitochondrial quality control mechanism that is crucial for cellular homeostasis, selectively removing damaged or dysfunctional mitochondria. Unsurprisingly, mitochondria play a vital role in modulating neuronal functions that are highly dependent on aerobic metabolism. In addition, an increasing number of studies have suggested that mitophagy is associated with various neurological diseases. The most significant genes involved in mitochondrial autophagy, such as phosphatase and tensin homolog-induced kinase 1 (PINK1), PRKN, BNIP3, BINP3L, and sequestosome 1 (SQSTM1), are correlated with several human neurological disorders, including Parkinson’s disease, ischemic stroke, Alzheimer’s disease, Huntington’s disease, and amyotrophic lateral sclerosis (ALS) ^{[5, 6]}. Furthermore, a few recent studies have also indicated that mitophagy may be related to epilepsy ^[7–12].

ROS generation contributes to the mechanisms underlying seizures, and there is an association between mitochondrial dysfunction and epilepsy pathogenesis ^[7]. Evidence for mitophagy dysregulation in epileptic seizures has been increasing in recent years. Most patients with primary mitochondrial disorders experience seizures. For example, tuberous sclerosis complex (TSC) in animal models can be controlled by restoring mitochondrial homeostasis ^[8]. However, patients with autism can have impaired mitophagy ^{[9, 10]}. A recent epilepsy study showed that PINK1-related mitophagy can be promoted and damaged hippocampal neurons can be repaired in epileptic rats ^[11]. Owing to the rapid development of sequencing technology, genes encoding proteins with mitochondrial localization or functions have been associated with the clinical manifestations of epilepsy ^{[6, 7, 12]}. However, the mechanism through which mitophagy contributes to epilepsy remains unclear.

According to previous studies, mitophagy is a novel and promising approach in epilepsy research. We hypothesized that mitophagy would be associated with epilepsy. Therefore, the present study comprehensively analyzed mitophagy-related genes (MRGs) in epilepsy using bioinformatics.

2.1 Data download and processing

We downloaded GSE143272^[13] from the Gene Expression Omnibus database, which is based on the GPL10558 Illumina HumanHT-12 V4.0 expression BeadChip. So no ethics approval and patients’ informed consent were needed for this present study. Blood samples were collected from drug-naïve patients with epilepsy (N = 34), those receiving anti-epileptic drug monotherapy (N = 57), and healthy controls (N = 50). On the basis of the molecular signature database (MSigDB) ^[14], 29 MRGs, including ATG5, FUNDC1, CSNK2A2, TOMM22, CSNK2A1, MAP1LC3A, and MFN2, were obtained. Of these, 28 matched the current dataset. The R software limma package ^[15] was used to perform differential expression analysis. Genes with a P-value < 0.05 were considered differentially expressed genes (DEGs). Discrepancy analysis results were visualized by constructing heatmaps and volcano plots using the pheatmap and ggplot2 packages.

2.2 Construction of the interaction network

In this study, the protein–protein interaction (PPI) network was constructed using the STRING database^[16]. Utilizing the CytoHubba plug-in in Cytoscape v3.9.1 software, the maximum neighborhood component (MNC) was calculated, and the top 10 MNCs were selected as hub genes. The starBase database ^[17] includes miRNA–ncRNA, miRNA–mRNA, RBP–RNA, and RNA–RNA data. miRNAs and lncRNAs that interact with these hub genes were predicted on the basis of parameters including Clade (mammal), Genome (human), Assembly (hg19), CLIP-Data (≥ 3), Degradome-Data (≥ 3), pan-Cancer (≥ 0), program Num (≥ 1), program (none) and Clade (mammal), Genome (human), Assembly (hg19), CLIP-Data (≥ 3), Degradome-Data (≥ 3), and pan-Cancer (≥ 0). Finally, we used the ggalluvial package to plot a Sankey diagram of the lncRNA–miRNA–mRNA ceRNA network.

2.3 Construction of the least absolute shrinkage and selection operator model

To evaluate the predictive power of MRGs in the normal and epilepsy groups, we used the glmnet package to perform ‎least absolute shrinkage and selection operator‎ (LASSO) regression analyses ^[18], screen MRGs associated with epilepsy, and construct a diagnostic risk model. Furthermore, we performed receiver operating characteristic (ROC) curve analysis to assess the predictive accuracy of the diagnostic model. To gauge forecast accuracy, we used the pROC package ^[19] to calculate the area under the curve (AUC).

2.4 Enrichment analyses

Gene Ontology (GO) analysis, a common approach for functional enrichment analysis, consists of biological processes (BP), molecular functions, and cellular components. The Kyoto Encyclopedia of Genes and Genomes (KEGG) database contains information regarding the interacting network of molecules or genes that can help identify the risk signaling pathways associated with diseases. GO and KEGG pathway enrichment analyses of DEGs were conducted using the clusterProfiler package ^[20]; a P-value < 0.05 was considered statistically significant. To study the BP differences between the different groups, we downloaded the “c2.cp.v7.2. symbols.gmt [Curated]” gene set from the MSigDB database ^[14] and conducted a gene set enrichment analysis (GSEA); a P-value < 0.05 was considered to be evidence of significant enrichment. In addition, according to the gene expression matrix of each sample, we used the Gene Set Variation Analysis (GSVA) package ^[21] to calculate related pathway scores and screened differentially expressed enrichment functions (or pathways) using the limma package.

2.5 Weighted gene co-expression network analysis

A co-expression network analysis was performed using the weighted gene co-expression network analysis (WGCNA) R package ^[22]. By identifying the Pearson correlation coefficient for genes, we constructed a weighted adjacent matrix and selected a suitable soft power threshold to generate a standardized scale network, which was then transformed into a topological overlap matrix (TOM). Furthermore, the TOM dissimilarity matrices were used as inputs for hierarchical cluster analysis. The dynamic tree-cutting method was used to construct a gene tree diagram and identify gene modules.

2.6 Clustering analysis

Consensus clustering analysis is a clustering algorithm based on resampling that is used to identify each member and its number of subgroups and verifies the reasonableness of clustering. We used the ConsensusClusterPlus package ^[23], adopted a consensus clustering method, chose “KB” as the clustering method and 1,000 as the number of iterations, and identified different disease subtypes based on gene expression profile data of the disease group.

2.7 Analysis of immune cell infiltration

We used the CIBERSORT package ^[24] to analyze immune cell infiltration. Using the gene expression characteristics of 22 types of immune cell subsets, we estimated the types and abundances of immune cells present in the mixed cells.

2.8 Statistical analysis

Statistical analysis was performed using the SPSS (version 24.0; IBM, New York, USA) and R software (version 3.6.2; the R Foundation, Vienna, Austria). Continuous variables are expressed as mean ± standard deviation. The chi-square test and Fisher’s exact test were used to compare categorical variables. An analysis of variance was used to detect differences in gene expression between clusters. We then calculated the ROC curve and Harrell’s concordance index (c index). P-values < 0.05 were considered statistically significant.

3.1 Study design

The flowchart of our study is shown in Fig. 1.

3.2 Differential analysis

Figure 2a shows a boxplot of the expression of 28 MRGs in the epilepsy and healthy control groups. Eleven MRGs with P < 0.05 were identified, of which five genes (MAP1LC3A, MAP1LC3B, PINK1, SQSTM1, and UBC) were upregulated and six genes (ATG5, RPS27A, TOMM22, TOMM40, TOMM5, and TOMM6) were downregulated in the epilepsy group. A heatmap of MRG expression in the epilepsy group versus in the healthy control group is presented in Fig. 2b. The correlations between MRGs in the epilepsy group are shown in Fig. 2c, among which MAP1LC3A, MAP1LC3B, PINK1, SQSTM1, TOMM22, TOMM40, TOMM5, TOMM6, and UBC were closely related to many other MRGs. Subsequently, the STRING database was used to construct a PPI network of 28 MRGs (Fig. 2e), and the interactions between genes were imported into the Cytoscape software. The MNC was calculated using the CytoHubba plug-in in Cytoscape, which identified the top 10 hub genes as PINK1, MFN1, MFN2, SQSTM1, VDAC1, FUNDC1, TOMM20, MAP1LC3B, MTERF3, and MAP1LC3A.

3.3 Construction of a disease model based on LASSO analysis

LASSO regression analysis was conducted on 28 MRGs, and an epilepsy diagnostic risk score model composed of 23 MRGs (ATG12, ATG5, CSNK2A1, CSNK2A2, CSNK2B, FUNDC1, MAP1LC3A, MAP1LC3B, MFN1, MFN2, MTERF3, PGAM5, PINK1, SQSTM1, SRC, TOMM20, TOMM22, TOMM40, TOMM5, TOMM7, UBB, UBC, and ULK1L) was obtained (Fig. 3). The ROC curve indicated that our model had a good diagnostic performance (AUC = 0.8912).

3.4 Enrichment analysis

On the basis of the differentially expressed MRGs, we performed GO annotation and KEGG pathway enrichment analyses (Tables 1 and 2; Fig. 4). The results of the GO annotation demonstrated that these differentially expressed MRGs were mainly related to BP such as macroautophagy, autophagy, processes utilizing autophagic mechanisms, autophagy of mitochondria, and mitochondrial disassembly (Fig. 4). The KEGG pathway enrichment analysis revealed that these DEGs were mainly enriched in pathways such as mitophagy-animal, ferroptosis, pathways of neurodegeneration-multiple diseases, shigellosis, Parkinson’s disease, ALS, autophagy-animal, ubiquitin-mediated proteolysis, Kaposi sarcoma-associated herpes virus infection, and autophagy-other. Moreover, the GSEA revealed that neutrophil degranulation, the pathway interaction database (PID) interleukin (IL)-6 and 7 pathways, erythropoietin receptor signaling, the PID p38 alpha beta pathway, angiogenesis, and the PID rac1 pathway were enriched in epilepsy samples (Table 3; Fig. 5).

Table 1

GO enrichment analysis
ONTOLOGY	ID	Description	GeneRatio	BgRatio	pvalue	p.adjust
BP	GO:0016236	macroautophagy	9/11	295/18670	2.90885E-15	1.66677E-12
BP	GO:0006914	autophagy	9/11	496/18670	3.22026E-13	6.1507E-11
BP	GO:0061919	process utilizing autophagic mechanism	9/11	496/18670	3.22026E-13	6.1507E-11
BP	GO:0000422	autophagy of mitochondrion	5/11	76/18670	4.43244E-10	5.07957E-08
BP	GO:0061726	mitochondrion disassembly	5/11	76/18670	4.43244E-10	5.07957E-08
BP	GO:1903008	organelle disassembly	5/11	101/18670	1.88763E-09	1.80269E-07
BP	GO:0006995	cellular response to nitrogen starvation	3/11	13/18670	4.33752E-08	3.10675E-06
BP	GO:0043562	cellular response to nitrogen levels	3/11	13/18670	4.33752E-08	3.10675E-06
BP	GO:0006605	protein targeting	6/11	432/18670	6.20886E-08	3.95297E-06
BP	GO:0000045	autophagosome assembly	4/11	93/18670	1.8536E-07	1.06211E-05
CC	GO:0005741	mitochondrial outer membrane	7/11	178/19717	1.3897E-12	7.86906E-11
CC	GO:0031968	organelle outer membrane	7/11	201/19717	3.28525E-12	7.86906E-11
CC	GO:0019867	outer membrane	7/11	203/19717	3.52346E-12	7.86906E-11
CC	GO:0098799	outer mitochondrial membrane protein complex	4/11	16/19717	9.5079E-11	1.59257E-09
CC	GO:0044455	mitochondrial membrane part	5/11	229/19717	8.83011E-08	1.18323E-06
CC	GO:0005776	autophagosome	4/11	93/19717	1.49225E-07	1.66634E-06
CC	GO:0098798	mitochondrial protein complex	4/11	262/19717	9.34412E-06	8.94366E-05
CC	GO:0005767	secondary lysosome	2/11	14/19717	2.5656E-05	0.000214869
CC	GO:0031307	integral component of mitochondrial outer membrane	2/11	20/19717	5.34696E-05	0.000395836
CC	GO:0031306	intrinsic component of mitochondrial outer membrane	2/11	21/19717	5.908E-05	0.000395836
MF	GO:0031625	ubiquitin protein ligase binding	6/9	290/17697	1.48246E-09	3.61847E-08
MF	GO:0044389	ubiquitin-like protein ligase binding	6/9	308/17697	2.12851E-09	3.61847E-08
MF	GO:0031386	protein tag	2/9	11/17697	1.26151E-05	0.000142971
MF	GO:0008320	protein transmembrane transporter activity	2/9	20/17697	4.3476E-05	0.000327793
MF	GO:0008565	protein transporter activity	2/9	22/17697	5.28297E-05	0.000327793
MF	GO:0022884	macromolecule transmembrane transporter activity	2/9	23/17697	5.78459E-05	0.000327793
MF	GO:1904680	peptide transmembrane transporter activity	2/9	32/17697	0.000113136	0.00054952
MF	GO:0042887	amide transmembrane transporter activity	2/9	47/17697	0.0002456	0.001043798
MF	GO:0002020	protease binding	2/9	128/17697	0.001807589	0.00682867
MF	GO:0010857	calcium-dependent protein kinase activity	1/9	10/17697	0.005075273	0.015814359
GO, gene ontology; BP, biological process; MF, molecular function; CC, cellular component.

Table 2

**KEGG enrichment analysis** KEGG, Kyoto encyclopedia of genes and genomes.
ID	Description		BgRatio	pvalue	p.adjust
hsa04137	Mitophagy - animal	5/8	68/8076	2.002E-09	4.00399E-08
hsa04216	Ferroptosis	3/8	41/8076	6.68331E-06	6.68331E-05
hsa05022	Pathways of neurodegeneration - multiple diseases	5/8	475/8076	3.32904E-05	0.000221936
hsa05131	Shigellosis	4/8	246/8076	5.34096E-05	0.000267048
hsa05012	Parkinson disease	3/8	249/8076	0.001445623	0.005782492
hsa05014	Amyotrophic lateral sclerosis	3/8	364/8076	0.004292776	0.014309253
hsa04140	Autophagy - animal	2/8	137/8076	0.007481116	0.019504801
hsa04120	Ubiquitin mediated proteolysis	2/8	140/8076	0.007801921	0.019504801
hsa05167	Kaposi sarcoma-associated herpesvirus infection	2/8	193/8076	0.014469572	0.032154603
hsa04136	Autophagy - other	1/8	32/8076	0.031276089	0.062552178

Table 3

GSEA enrichment analysis
Description	setSize	Enrichment Score	NES	pvalue
REACTOME_NEUTROPHIL_DEGRANULATION	380	0.4967	2.5430	0.0016
PID_IL6_7_PATHWAY	34	0.5524	1.9200	0.0018
WP_EPO_RECEPTOR_SIGNALING	21	0.6090	1.8893	0.0019
PID_P38_ALPHA_BETA_PATHWAY	25	0.5299	1.7291	0.0055
WP_ANGIOGENESIS	13	0.6101	1.6331	0.0235
PID_RAC1_PATHWAY	40	0.5277	1.9021	0.0018
WP_CANCER_IMMUNOTHERAPY_BY_PD1_BLOCKADE	17	-0.6079	-1.8565	0.0063
KEGG_RNA_DEGRADATION	45	-0.4857	-1.8671	0.0022
REACTOME_PD_1_SIGNALING	17	-0.6571	-2.0068	0.0042
PID_CD8_TCR_PATHWAY	48	-0.5261	-2.0195	0.0023
BIOCARTA_STATHMIN_PATHWAY	13	-0.7553	-2.0992	0.0020
BIOCARTA_CTL_PATHWAY	13	-0.8224	-2.2857	0.0020
GSEA, conducted gene set enrichment analysis; PID, pak inhibitory domain; NES, normalized enrichment score; IL, interleukin; EPO, erythropoietin; WP, wikipathways; RAC1, ras-related C3 botulinum toxin substrate 1; KEGG, kyoto encyclopedia of genes and genomes; PD1, programmed cell death protein 1; CD8, cluster of differentiation8; TCR, T cell receptor; CTL, cytotoxic T-lymphocyte.

Furthermore, the GSVA results (Table 4; Fig. 6) showed that GO microfibril, GO amino acid betaine metabolic process, TGFβ receptor signaling in epithelial to mesenchymal transition, and GSE35543 in vivo nTreg vs in vitro iTreg DN were enriched in the epilepsy group.

Table 4

GSVA enrichment analysis
	logFC	AveExpr	P Value	adj P Val
BILANGES_SERUM_SENSITIVE_VIA_TSC2	0.4161	0.0015	0.0000	0.0001
GO_AMINO_ACID_BETAINE_METABOLIC_PROCESS	-0.5406	-0.0379	0.0000	0.0001
GO_BOX_H_ACA_RNP_COMPLEX	0.5150	0.0193	0.0000	0.0002
GO_MICROFIBRIL	-0.7752	-0.0070	0.0000	0.0001
GO_NEGATIVE_REGULATION_OF_RETINOIC_ACID_RECEPTOR_SIGNALING_PATHWAY	0.5882	0.0493	0.0000	0.0001
GO_TETRAPYRROLE_METABOLIC_PROCESS	0.3773	0.0224	0.0000	0.0002
GO_TOXIC_SUBSTANCE_BINDING	0.5167	0.0047	0.0000	0.0002
GSE35543_IN_VIVO_NTREG_VS_IN_VITRO_ITREG_DN	-0.2730	-0.0116	0.0000	0.0002
REACTOME_TGF_BETA_RECEPTOR_SIGNALING_IN_EMT_EPITHELIAL_TO_MESENCHYMAL_TRANSITION	-0.4199	-0.0134	0.0000	0.0001
XU_CREBBP_TARGETS_UP	0.3624	-0.0073	0.0000	0.0002
GSVA, gene set variation analysis; logFC, log2 fold change; AveExpr, mean expression across groups; adj P Val, p value for differential expression after correcting for multiple testing; GO, gene ontology.

3.5 WGCNA

We then performed unsupervised clustering using systematic WGCNA, divided the samples into modules based on gene expression patterns, and obtained a genetic tree topology using homogeneous topological overlap-based dissimilarity clustering. The corresponding module colors were expressed as row colors, where each row color represented one color-coding module, which was identified and assigned different colors by merged dynamic tree cutting (Fig. 7a) and finally divided into 13 different modules (Fig. 7b). As shown in Fig. 7b, compared with the other modules, the blue module was positively correlated with the samples (with or without epilepsy), and the salmon, turquoise, cyan, green, light cyan, midnight blue, and gray modules were negatively correlated with the samples (with or without epilepsy). We obtained six hub modules related to epilepsy: turquoise, green, cyan, blue, light cyan, and midnight blue. The correlation between the characteristic gene values of these six hub modules and the samples is shown in Figs. 7c-h.

On the basis of the cluster dendrogram, we identified gene co-expression modules that were significantly associated with the disease group. An UpSet diagram of the top six gene co-expression modules and MRGs is shown in Fig. 8a. As shown in the Venn diagram of the MEturquoise modules with MRGs and differentially expressed MRGs (Fig. 8b), we identified four common genes (RPS27A, SQSTM1, TOMM5, and UBC). We then constructed the PPI network using the STRING database (Fig. 8c), imported the gene interactions into the Cytoscape software, used the CytoHubba plug-in in Cytoscape, and obtained the top 10 hub genes (RPS27A, UBC, Ribosomal protein S3 (RPS3), RPS9, RPL15, RPL8, RPS16, RPS23, RPS11, and RPL35), as shown in Fig. 8d. Finally, as depicted in Fig. 8e, we mapped the Sankey diagram that predicted miRNAs and lncRNAs interacting with hub genes using the starBase database.

3.6 Clustering analysis

To further investigate the underlying mechanism of epilepsy, we performed unsupervised consensus clustering on epilepsy samples on the basis of gene expression in the disease group, the results of which are shown in Figs. 9a-g. Figures 9h-i shows the gene expression profiles of the hub genes of the PPI network based on MEturquoise modules, differentially expressed MRGs, and MRGs in epilepsy subtypes A and B when clustered into two subtypes.

3.7 CIBERSORT immune infiltration analysis

Finally, we used the CIBERSORT algorithm to evaluate the abundance of immune cells in the epilepsy and normal samples (Figs. 10a-b). The results indicated that the patients with epilepsy had a higher infiltration level of neutrophils (P < 0.01) but lower resting memory CD4 + T cells and activated memory CD4 + T cells (P < 0.05). Furthermore, a correlation heatmap showed the correlation between the hub genes of the PPI network based on MEturquoise modules and differentially expressed MRGs, as well as between differentially expressed MRGs and immune cells (Figs. 10c-d).

Existing epilepsy research in the literature has far exceeded the “channelopathy era”, ^[25] and new treatment ideas are needed. Using bioinformatic analysis, we analyzed the correlation between mitophagy and epilepsy for the first time, and concluded that mitophagy was correlated with epilepsy. This conclusion agrees with recent experimental research showing that epilepsy in rats can be treated by the promotion of PINK1-related mitophagy ^[11].

Among the 28 MRGs, we found that five genes (PINK1, [microtubule-associated proteins 1A/1B light chain 3 beta 2] MAP1LC3B2, SQSTM1/p62, MAP1LC3A, and UBC) were upregulated in patients with epilepsy compared with healthy controls, whereas six other genes (RPS27A, ATG5, TOMM22, TOMM40, TOMM5, and TOMM6) were downregulated. A previous study suggested that at a normal level, mitophagy maintains the stabilization of cellular conditions, but at a high or low level, it contributes to disease ^[26]. The total expression of these genes was neither high nor low, demonstrating a double-edged sword effect of mitophagy on epilepsy.

PINK1 is a mitochondria-targeting neuroprotective kinase that protects cells from stress-induced mitochondrial dysfunction. It is one of the most widely studied MRGs. PINK1 is considered as the critical initiator of mitophagy, and its upregulated expression usually indicates beneficial mitophagy effects ^[27]. A previous experimental study confirmed that PINK1 overexpression alleviates hippocampal damage in rats with epilepsy ^[28]. Our study found that PINK1 expression was significantly increased in blood samples from patients with untreated epilepsy ^[11], which may indicate the beneficial effects of mitophagy on epilepsy.

Previous research suggested that the absence of MAP1LC3B2 decreases mitochondrial clearance ^[28]. Although there are no studies on the role of MAP1LC3B2 in epilepsy, its significantly higher expression in the epilepsy group in our study may indicate heightened mitophagy.

We also found that upregulation of SQSTM1/p62, an important MRG, is often used to monitor autophagic flux ^[29]. It has been discovered that the SQSTM1 protein is elevated in epileptogenic tubers in the TSC, which may contribute to the molecular signaling of the mammalian target of rapamycin complex (mTORC) and may lead to seizures ^[30]. Cukovic et al. found that in the developmental stages of rats with epilepsy, SQSTM1 is upregulated before apoptosis; thus, it may be involved in epilepsy-induced brain damage ^[31]. Another study identified SQSTM1 as a potential epilepsy-related gene ^[32]. MAP1LC3, which was the most upregulated gene in our study, has not yet been investigated in epilepsy. However, it has been identified as the hub gene of several forms of cancer ^[33–35] and was found to be downregulated in multiple tumors. Upregulation indicates mitophagy dysfunction and is often associated with adverse outcomes. However, further experimental studies are required to confirm its efficacy.

RPS27a was downregulated in the epilepsy group. Rps27a encodes a fusion protein consisting of ubiquitin and the ribosomal protein S27a and is associated with the regulation of cell proliferation and the cell cycle. RPS27a is linked to a range of brain diseases; for example, it shows increased expression in neurodegenerative diseases and is regarded as a key controller of microglial activation ^[36]. RPS27A was found to be upregulated in a bioinformatic study of multiple sclerosis (MS) and was identified as a hub gene from the MS blood samples ^[37]. The expression of RPS27A was enhanced in the early post-insult and latency phases in a mouse model of epilepsy in a separate study, and the researchers contended that this could enhance the activation of Toll-like receptor 4 (TLR4) and increases the risk of epilepsy ^[38]. The effect of mitophagy on cancer has been extensively studied by researchers, and the expression pattern of MRG in our epilepsy study appears to be in direct contrast to that reported in numerous cancer studies. For instance, in a variety of tumors, the expression of PINK1, MAP1LC3B2, SQSTM1/p62, and MAP1LC3A is suppressed, whereas that of RPS27A, ATG5, TOMM22, TOMM40, TOMM5, and TOMM6 is upregulated ^{[33, 35, 39]}. This suggests different mechanisms of mitophagy in these two types of diseases, which warrant further investigation.

Multiple molecular subtypes of epilepsy present with various clinical manifestations, making early recognition of some subtypes difficult. Early identification of the disease can provide multiple benefits to patients and their families. Mitophagy has been shown to be helpful for the diagnosis and outcome prediction of many diseases, including liver cancer ^[33], pancreatic cancer, ^[39] and subarachnoid hemorrhage ^[40]; however, whether it can be used as a biomarker of epilepsy is still unknown. In the present study, we constructed a diagnostic model of epilepsy using 23 MRGs obtained from the peripheral blood of patients with epilepsy. The model showed good diagnostic performance, as evaluated by ROC analysis (AUC = 0.8912), which was superior to that of many other epilepsy diagnosis models ^[41–44].

To study the role of mitophagy in epilepsy, we performed unsupervised clustering using WGCNA and obtained the top six significant modules closely related to epilepsy. The UpSet diagram shows that these MRGs participated in important modules. Next, we constructed a PPI network based on MEturquoise modules and differentially expressed MRGs and obtained hub genes, including RPS27A, UBC, RPS3, RPS9, RPL15, RPL8, RPS16, RPS23, RPS11, and PRL35. The role of RPS27A has been discussed above, and UBC is usually used as a reference gene in disease studies ^[45]. Previous studies have demonstrated an association between RPL8, RPL23, and epilepsy. Mitochondrial dysfunction has occurred in N(G)-nitro-L-arginine methyl ester–treated C57BL/6 mice, which manifested with epileptic seizures and changes in mood and behavior disorders; RPL8 and RPS23 were hub genes of these treated mice ^[46]. In another study, RPL8 was significantly increased in mouse models of pentylenetetrazole (PTZ)-induced epilepsy ^[47]. Several studies have suggested the association of RPS23 with neurodevelopmental disorders ^[48–50]. In addition, RPL15 is differentially expressed across all brain regions in Parkinson's Disease and is thought to be closely associated with other brain diseases ^[51]. RPS3 is an AKT substrate for neuronal survival, and overexpression of RPS3 in primary hippocampal neurons can induce neuronal apoptosis ^[52]. RPL35 induces neuroblastoma cell proliferation and high levels of RPL35 expression are associated with adverse outcomes in patients ^[53]. The roles of the remaining genes in epilepsy have not yet been studied and further research is required.

Increasing evidence suggests that immune disorders occur in patients with epilepsy ^[54–57]. Previous studies have suggested that the neutrophil/lymphocyte ratio significantly increases during the acute phase of epilepsy ^[54], peripheral blood mononuclear cells are involved in the pathogenesis of epilepsy ^55], and Treg cells are a potential treatment for epileptic cerebral tissue ^[56]. In patients with refractory epilepsy, blood-borne inflammatory myeloid cells and memory CD4 + and CD8 + T cells show obvious cellular infiltration in the brain tissue ^[57]. In our study, we found that patients with epilepsy had a higher infiltration level of neutrophils into the peripheral blood. This also supports previous observations that epilepsy can be treated by regulating neutrophils ^[−58–60]. In addition, we found that patients with epilepsy had lower levels of resting memory CD4 + T cells and activated memory CD4 + T cells in peripheral blood samples. It is important to note that Luo et al. also used a CIBERSORT algorithm to evaluate the abundance of immune cells in samples from GSE143272 and found that the epilepsy group had a higher infiltration level of activated mast cells, but lower activated memory CD4 + T cells, resting memory CD4 + T cells, and gamma delta T cells ^[43], without discovering infiltration difference of neutrophils. In contrast to our study, they chose all disease samples, including treated and untreated epilepsy samples. The untreated epilepsy samples that we chose could better represent the effect of immune cells when seizures occur. A major assumption is that the higher infiltration of activated mast cells and undiscovered infiltration differences in neutrophils may arise from the effect of AEDs. However, this hypothesis remains to be verified. Finally, we performed immune cell infiltration analysis with the hub genes and differentially expressed MRGs, and found that many of these genes were significantly correlated with several immune cells, including neutrophils, CD4 + T cells, and macrophages (Fig. 10c-d). This supports our view that mitophagy is related to epilepsy and that it may be possible to regulate the interaction between these genes and immune cells to treat epilepsy.

4.1 Limitations

This is the first study to conduct a bioinformatics analysis to elucidate the biological roles of MRGs in epilepsy. However, this study had certain limitations. We used representative MRGs, but a deeper understanding about mitophagy is gained, analyses using more accurate biomarkers will lead to more scientific results. Whether this conclusion can also be applied to experimental animal models of epilepsy and different disease stages of human epilepsy requires further investigation.

We performed for the first time a systematic analysis of the expression profiles of 28 MRGs in untreated patients with epilepsy using bioinformatics, and we found that mitophagy was correlated with epilepsy. These findings will allow us to further understand epilepsy and the effect of mitophagy, which will help in early diagnosis as well as provide a novel treatment direction for epilepsy.

AEDs, anti-epileptic drugs; ALS, amyotrophic lateral sclerosis; AUC, area under the curve; BP, biological processes; DEGs, differentially expressed genes; GO, gene ontology; GSEA, gene set enrichment analysis; GSVA, gene set variation analysis; IL, interleukin; KEGG, Kyoto Encyclopedia of Genes and Genomes; LASSO, least absolute shrinkage and selection operator; MNC, maximum neighborhood component; MRGs, mitophagy-related genes; MS, multiple sclerosis; mTORC, the mammalian target of rapamycin complex; PPI, protein–protein interaction; PTZ, pentylenetetrazole; ROC, receiver operating characteristic; ROS, reactive oxygen species; TLR4, Toll-like receptor 4; TOM, topological overlap matrix; TSC, tuberous sclerosis complex; WGCNA, weighted gene co-expression network analysis

Ethics approval and consent to participate: Not applicable

Consent for publication: Not applicable

Availability of data and materials: The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Competing interests: The authors declare that they have no competing interests

Funding: This research received no funding.

Authors' contributions: JXZ is first author. MW, HZ and VA contributed to the study design, implementation, manuscript discussion and critical revision. XPX made important revisions and confirmed final revision.All authors read and approved the final manuscript.

Acknowledgements: Not applicable.

Authors' information:

¹ Department of Pediatric, The First Afliated Hospital of Harbin Medical University, Harbin 150081, Heilongjiang, China

² Department of Thoracic Surgery, The Second Afliated Hospital of Harbin Medical University, Harbin 150081, Heilongjiang, China

Thijs RD, Surges R, O'Brien TJ, Sander JW. Epilepsy in adults. Lancet. 2019;393(10172):689–701.
Pitkänen A, Sutula TP. Is epilepsy a progressive disorder? Prospects for new therapeutic approaches in temporal-lobe epilepsy. Lancet Neurol. 2002;1(3):173–81.
Ma C, Xia F, Kelley SO. Mitochondrial targeting of probes and therapeutics to the powerhouse of the cell. Bioconjug Chem. 2020;31(12):2650–67.
Zorov DB, Juhaszova M, Sollott SJ. Mitochondrial Reactive Oxygen Species (ROS) and ROS-induced ROS release. Physiol Rev. 2014;94(3):909–50.
Ortiz-González XR. Mitochondrial dysfunction: A common denominator in neuro developmental disorders? Dev Neurosci. 2021;43(3–4):222–9.
Doblado L, Lueck C, Rey C, Samhan-Arias AK, Prieto I, Stacchiotti A, et al. Mitophagy in human diseases. Int J Mol Sci. 2021;22(8):3903.
Singh S, Singh TG, Rehni AK, Sharma V, Singh M, Kaur R. Reviving mitochondrial bioenergetics: A relevant approach in epilepsy. Mitochondrion. 2021;58:213–26.
Shulyakova N, Andreazza AC, Mills LR, Eubanks JH. Mitochondrial dysfunction in the pathogenesis of rett syndrome: implications for mitochondria-targeted therapies. Front Cell Neurosci. 2017;11:58.
Müller M. Disturbed redox homeostasis and oxidative stress:potential players in the developmental regression in Rett syndrome. Neurosci Biobehav Rev. 2019;98:154–63.
Condon KJ, Orozco JM, Adelmann CH, Spinelli JB, van der Helm PW, Roberts JM, et al. Genome-wide CRISPR screens reveal multitiered mechanisms through which mTORC1 senses mitochondrial dysfunction. Proc Natl Acad Sci U S A. 2021;118(4):e2022120118.
Tang G, Rios PG, Kuo SH, Akman HO, Rosoklija G, Tanji K, et al. Mitochondrial abnormalities in temporal lobe of autistic brain. Neurobiol Dis. 2013;54:349–61.
Napoli E, Song G, Panoutsopoulos A, Riyadh MA, Kaushik G, Halmai J, et al. Beyond autophagy: A novel role for autism-linked Wdfy3 in brain mitophagy. Sci Rep. 2018;8(1):11348.
Xiao D, Lv J, Zheng Z, Liu Y, Zhang Y, Luo C, et al. Mechanisms of microRNA-142 in mitochondrial autophagy and hippocampal damage in a rat model of epilepsy. Int J Mol Med. 2021;47(6):1–12.
Steinlein OK. Genetics and epilepsy. Dialogues Clin Neurosci. 2008;10(1):29–38.
Rawat C, Kushwaha S, Srivastava AK, Kukreti R. Peripheral blood gene expression signatures associated with epilepsy and its etiologic classification. Genomics. 2020;112(1):218–24.
Liberzon A, Birger C, Thorvaldsdóttir H, Ghandi M, Mesirov JP, Tamayo P. The Molecular Signatures Database (MSigDB) hallmark gene set collection. Cell Syst. 2015;1(6):417–25.
Ritchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, et al. Limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 2015;43(7):e47.
Kolde R. Kolde MRJRp: Package ‘pheatmap’. 2015, 1(7):790.
Wickham HJWIRCS. ggplot2. 2011, 3(2):180–185.
Li JH, Liu S, Zhou H, Qu LH, Yang JH. starBase v2.0: decoding miRNA-ceRNA, miRNA-ncRNA and protein-RNA interaction networks from large-scale CLIP-Seq data. Nucleic Acids Res. 2014;42:D92–7.
Brunson JC. ggalluvial: Layered Grammar for Alluvial Plots. J Open Source Softw. 2020;5(49):2017.
Robin X, Turck N, Hainard A, Tiberti N, Lisacek F, Sanchez JC, et al. pROC: an open-source package for R and S + to analyze and compare ROC curves. BMC Bioinformatics. 2011;12:77.
Yu G, Wang LG, Han Y, He QY. clusterProfiler: an R package for comparing biological themes among gene clusters. OMICS. 2012;16(5):284–7.
Hänzelmann S, Castelo R, Guinney J. GSVA: gene set variation analysis for microarray and RNA-seq data. BMC Bioinformatics. 2013;14:7.
Langfelder P, Horvath S. WGCNA: an R package for weighted correlation network analysis. BMC Bioinformatics. 2008;9:559.
Wilkerson MD, Hayes DN. ConsensusClusterPlus: a class discovery tool with confidence assessments and item tracking. Bioinformatics. 2010;26(12):1572–3.
Chen B, Khodadoust MS, Liu CL, Newman AM, Alizadeh AA. Profiling tumor infiltrating immune cells with CIBERSORT. Methods Mol Biol. 2018;1711:243–59.
Helbig I, Tayoun AA. Understanding geno types and phenotypes in epileptic encephalopathies. Mol Syndromol. 2016;7(4):172–81.
Gunst J, Derese I, Aertgeerts A, Ververs EJ, Wauters A, Van den Berghe G, et al. Insufficient autophagy contributes to mitochondrial dysfunction, organ failure, and adverse outcome in an animal model of critical illness. Crit Care Med. 2013;41(1):182–94.
Gan ZY, Callegari S, Cobbold SA, Cotton TR, Mlodzianoski MJ, Schubert AF, et al. Activation mechanism of PINK1. Nature. 2022;602(7896):328–35.
Wang Z, Gao L, Guo X, Feng C, Lian W, Deng K, et al. Development and validation of a nomogram with an autophagy-related gene signature for predicting survival in patients with glioblastoma. Aging. 2019;11(24):12246–69.
Zhu F, Li X, Tang X, Jiang J, Han Y, Li Y, et al. Neferine promotes the apoptosis of HNSCC through the accumulation of p62/SQSTM1 caused by autophagic flux inhibition. Int J Mol Med. 2021;48(1):124.
Dombkowski AA, Batista CE, Cukovic D, Carruthers NJ, Ranganathan R, Shukla U, et al. Cortical tubers: Windows into dysregulation of epilepsy risk and synaptic signaling genes by MicroRNAs. Cereb Cortex. 2016;26(3):1059–71.
Cukovic D, Bagla S, Ukasik D, Stemmer PM, Jena BP, Naik AR, et al. Exosomes in epilepsy of tuberous sclerosis complex: Carriers of pro-inflammatory MicroRNAs. Noncoding RNA. 2021;7(3):40.
Wang J, Lin ZJ, Liu L, Xu HQ, Shi YW, Yi YH, et al. Epilepsy-associated genes. Seizure. 2017;44:11–20.
Chen H, Wang J, Zeng R, Luo Y, Guo K, Wu H, et al. Development and validation of a novel mitophagy-related gene prognostic signature for hepatocellular carcinoma based on immunoscore classification of tumor. J Oncol. 2021;2021:5070099.
Guo Y, Shi J, Wang S, Zhao Z, Shang A, Li D et al. High Expression of mitochondrial autophagy-related gene FUNDC1 is associated with poor prognosis of colon cancer and defective immune cell infiltration. 2021. PREPRINT (Version 1) available at Research Square https://doi.org/10.21203/rs.3.rs-705751/v1.
Wang J, Qiu X, Huang J, Zhuo Z, Chen H, Zeng R, et al. Development and validation of a novel mitophagy-related gene prognostic signature for glioblastoma multiforme. BMC Cancer. 2022;22(1):644.
Khayer N, Mirzaie M, Marashi SA, Jalessi M. Rps27a might act as a controller of microglia activation in triggering neurodegenerative diseases. PLoS ONE. 2020;15(9):e0239219.
Kundu A. Identification of potential blood-based biomarkers for multiple sclerosis. 2022. PREPRINT available at https://doi.org/10.21203/rs.3.rs-705751/v1.
Walker A, Russmann V, Deeg CA, von Toerne C, Kleinwort KJH, Szober C, et al. Proteomic profiling of epileptogenesis in a rat model: Focus on inflammation. Brain Behav Immun. 2016;53:138–58.
Ammous Z, Rawlins LE, Jones H, Leslie JS, Wenger O, Scott E, et al. A biallelic SNIP1 Amish founder variant causes a recognizable neurodevelopmental disorder. PLOS Genet. 2021;17(9):e1009803.
Zhuo Z, Lin H, Liang J, Ma P, Li J, Huang L, et al. Mitophagy-Related gene signature for prediction prognosis, immune scenery, mutation, and chemotherapy response in pancreatic cancer. Front Cell Dev Biol. 2022;9:802528.
Youn DH, Kim BJ, Hong EP, Jeon JP. Bioinformatics analysis of autophagy and mitophagy markers associated with delayed cerebral ischemia following subarachnoid hemorrhage. J Korean Neurosurg Soc. 2022;65(2):236–44.
Chen QL, Xia L, Zhong SP, Wang Q, Ding J, Wang X. Bioinformatic analysis identifies key transcriptome signatures in temporal lobe epilepsy. CNS Neurosci Ther. 2020;26(12):1266–77.
Zhu Y, Huang D, Zhao Z, Lu C. Bioinformatic analysis identifies potential key genes of epilepsy. PLoS ONE. 2021;16(9):e0254326.
Luo X, Xiang T, Huang H, Ye L, Huang Y, Wu Y. Identification of significant immune-related genes for epilepsy via bioinformatics analysis. Ann Transl Med. 2021;9(14):1161.
Xia L, Liu L, Wang Q, Ding J, Wang X. Relationship between the pyroptosis pathway and epilepsy: a bioinformatic analysis. Front Neurol. 2022;12:782739.
Sadangi C, Rosenow F, Norwood BA. Validation of reference genes for quantitative gene expression analysis in experimental epileps. J Neurosci Res. 2017;95(12):2357–66.
Hendrickx JO, Adams C, Sieben A, Laukens K, Van Dam D, De Meyer GRY. Proteomic assessment of C57BL/6 hippocampi after non-selective pharmacological inhibition of nitric oxide synthase activity: implications of seizure-like neuronal hyperexcitability followed by tauopathy. Biomedicines. 2022;10(8):1772.
Vashishta A, Slomnicki LP, Pietrzak M, Smith SC, Kolikonda M, Naik SP, et al. RNA polymerase 1 is transiently regulated by seizures and plays a role in a pharmacological kindling model of epilepsy. Mol Neurobiol. 2018;55(11):8374–87.
Hetman M, Slomnicki LP. Ribosomal biogenesis as an emerging target of neurodevelopmental pathologies. J Neurochem. 2019;148(3):325–47.
Zhang YW, Liu S, Zhang X, Li WB, Chen Y, Huang X, et al. A functional mouse retroposed gene Rps23r1 reduces Alzheimer's beta-amyloid levels and tau phosphorylation. Neuron. 2009;64(3):328–40.
Ganapathi M, Padgett LR, Yamada K, Devinsky O, Willaert R, Person R, et al. Recessive rare variants in deoxyhypusine synthase, an enzyme involved in the synthesis of hypusine, are associated with a neurodevelopmental disorder. Am J Hum Genet. 2019;104(2):287–98.
Puthiyedth N, Riveros C, Berretta R, Moscato P. Identification of differentially expressed genes through integrated study of alzheimer's disease affected brain regions. PLoS ONE. 2016;11(4):e0152342.
Garcia-Gil M, Albi E. Nuclear lipids in the nervous system: what they do in health and disease. Neurochem Res. 2017;42(2):321–36.
Bell JL, Hagemann S, Holien JK, Liu T, Nagy Z, Schulte JH, et al. Identification of RNA-Binding proteins as targetable putative oncogenes in neuroblastoma. Int J Mol Sci. 2020;21(14):5098.
Hosseini S, Mofrad AME, Mokarian P, Nourigheimasi S, Azarhomayoun A, Khanzadeh S, et al. Neutrophil to lymphocyte ratio in epilepsy: A systematic review. Mediators Inflamm. 2022;2022:4973996.
Yamanaka G, Morichi S, Takamatsu T, Watanabe Y, Suzuki S, Ishida Y, et al. Links between immune cells from the periphery and the brain in the pathogenesis of epilepsy: A narrative review. Int J Mol Sci. 2021;22(9):4395.
Yue J, Xu R, Yin C, Yang H, Zhang C, Zhao D. Negative effects of brain regulatory T cells depletion on epilepsy. Prog Neurobiol. 2022;217:102335.
Xu D, Robinson AP, Ishii T, Duncan DS, Alden TD, Goings GE, et al. Peripherally derived T regulatory and γδ T cells have opposing roles in the pathogenesis of intractable pediatric epilepsy. J Exp Med. 2018;215(4):1169–86.
Barnes SE, Zera KA, Ivison GT, Buckwalter MS, Engleman EG. Brain profiling in murine colitis and human epilepsy reveals neutrophils and TNFα as mediators of neuronal hyperexcitability. J Neuroinflammation. 2021;18(1):199.
Gómez CT, Lairion F, Repetto M, Ettcheto M, Merelli A, Lazarowski A, et al. Cannabidiol (CBD) alters the functionality of neutrophils (PMN). Implications in the refractory epilepsy treatment. Pharmaceuticals (Basel). 2021;14(3):220.
Matsuo T, Komori R, Nakatani M, Ochi S, Yokota-Nakatsuma A, Matsumoto J, et al. Levetiracetam suppresses the infiltration of neutrophils and monocytes and downregulates many inflammatory cytokines during epileptogenesis in pilocarpine-induced status epilepticus mice. Int J Mol Sci. 2022;23(14):7671.

Download PDF

Version 1

posted

You are reading this latest preprint version

Development of the correlation between mitophagy-related genes and epilepsy

Status:

Version 1

Abstract

Background

Methods

Results

Conclusions

Figures

1. BACKGROUND

2. METHODS

2.1 Data download and processing

2.2 Construction of the interaction network

2.3 Construction of the least absolute shrinkage and selection operator model

2.4 Enrichment analyses

2.5 Weighted gene co-expression network analysis

2.6 Clustering analysis

2.7 Analysis of immune cell infiltration

2.8 Statistical analysis

3. RESULTS

3.1 Study design

3.2 Differential analysis

3.3 Construction of a disease model based on LASSO analysis

3.4 Enrichment analysis

3.5 WGCNA

3.6 Clustering analysis

3.7 CIBERSORT immune infiltration analysis

4. DISCUSSION

4.1 Limitations

5. CONCLUSION

Abbreviations

Declarations

References

Status:

Version 1