LVNC is a primary genetic cardiomyopathy with uncertain etiology [27], presenting various genetic mutations in the majority of cases. In the current study, we identified that 2452 mRNAs and 1415 lncRNAs were significantly differentially expressed between 6 LVNC patients and 6 healthy controls. Enrichment analyses were respectively performed on gene sets from GSEA and gene modules from WGCNA. Finally, a lncRNA-miRNA-mRNA interaction network was constructed. LINC01515, AC092718.2, AC007216.4 were recognized in LVNC as the upstream lncRNA effectors with complete regulatory axes.
GSEA revealed that NLRP3 inflammasome (GO), bile secretion and thyroid hormone synthesis (KEGG) were enriched in LVNC, consistent with the former study [15]. In addition, the “interstitial matrix”, “endocytosis” terms were positively enriched. Endocytosis plays a vital role in homeostasis of plasma membrane, regulating communication between cells and their environment [28]. The ubiquitination and endocytosis of NOTCH ligands were reported to activate downstream signaling pathways, facilitating myocardium compaction and trabeculation maturation [29]. Considering interstitial matrix, interstitial fibrosis is a magnificent pathological characteristic of cardiac remodeling throughout the process of cardiomyopathy, manifesting as excessive deposition of extracellular matrix proteins [30]. Thus, the mechanism underlying LVNC and its adverse effect on cardiac remodeling and heart failure may be partially elucidated.
In contrast, pathways involving acyl-CoA and ketone body metabolism were negatively correlated with LVNC. Diverse deficiencies of key enzymes through the metabolism of acyl-CoA were reported in association with LVNC, including malonyl coenzyme A decarboxylase [31], 3-hydroxy-3-methylglutaryl-coenzyme A lyase (HMG-CoA) [32] and short-chain acyl-CoA reductase [33]. Meanwhile, HMG-CoA also participates in the last step of ketogenesis. LVNC patients caused by HMG-CoA deficiency still exhibits with nonketotic hypoglycemia, limiting cardiac energy supply and facilitating the development of heart failure [34].
WGCNA, a bioinformatic method describing correlation patterns among genes, is naturally adaptive with ceRNA network. We deemed WGCNA as the essential system of selection for ceRNA components. The enriched KEGG pathway “endocytosis” in GSEA was reassured by the lightyellow module of weighted co-expression analysis on mRNA. Genes in the lightyellow module were significantly correlated with “endocytic vesicle” and “early endosome membrane”. In addition, genes in the greenyellow module were enriched in three smooth muscle contraction pathways. In the majority of LVNC patients, sarcomeric gene variants are commonly identified [35], which codes for sarcomere and functions as the principal contractile unit of striated muscle. Moreover, Kern et al [36] introduced that Adamts9, a highly conserved versican-degrading protease, is differentially expressed in vascular smooth muscle cells and the lack of Adamts9 leads to a comparatively milder noncompaction of left ventricle. The compaction process in normal cardiac embryogenesis involves rearrangement of the myocardial blood supply, during which vascular smooth muscle cells undergo intense proliferation and migration [37]. However, the compaction process develops improperly in LVNC, leading to the disorder of intramural perfusion [12].
From WGCNA on mRNAs, we found that the blue module gained highest correlation with LVNC. Relevant pathways of the blue module concentrated on cell polarity. Defects of cell polarity regulators, such as Scrib [38], DAAM1 and DAAM2 [39], were verified to participate in cardiomyocyte differentiation and maturation as well as sarcomere assembly. A recent study reported that the polarity of cardiomyocyte guided oriented cell division and directional migration, thus corresponding disorder would disturb trabeculation [40]. These findings provided new insights into early pathogenesis and the mechanism underlying distinctive hypertrabeculation in LVNC.
We screened three lncRNAs presenting interaction with LVNC related miRNA. The biological function of AC092718.2 and AC007216.4 remains unknown. LINC01515 was reported to regulate α-synuclein levels in cerebrospinal fluid [41] and function through lncRNA-miRNA-mRNA axis in leukemia [42] and nasopharyngeal carcinoma [43]. Through MCC method, 9 hub mRNAs (CDK6, RAB40C, CCNT2, NKIRAS2, HUWE1, SPARC, MYCN, COL4A2, MAZ) were identified as core components of constructed ceRNA network. Considering their significant expression change, employment of these genes alone or in combination as biomarkers may make the screening and diagnosis of LVNC more convenient.
There are several limitations of the current study: Firstly, we only included 12 blood samples (6 LVNC patients and 6 healthy controls) from GEO database due to the lack of other public data. Small sample size decided an inevitable between-group heterogeneity, weakening the results of WGCNA. Concurrently, tissue difference between blood and heart also interfered with LVNC related expression profiling. Besides, we didn’t reserve other modules with high correlations in WGCNA for further analysis. Our ceRNA network might discriminately and partially describe the RNA interaction mode in LVNC. Depressingly, the lack of clinical information of involved patients also limited further analysis for prognostic value of identified hub genes. Moreover, the integral analysis was conducted in silico. Thus, provision of public data from LVNC samples is urgently needed and further study can verify the findings of our work in an experimental way.