In this study, 7 circRNAs, 22 miRNAs, and 34 target mRNAs in the circRNA-miRNA-mRNA ceRNA regulatory network were identified as crucial genes in ECA; in addition, the nuclear factor-k-gene binding (NF-κB) and phosphatidylinositol-3 kinase/protein kinase B (PI3K-Akt) signaling were identified as the most enriched pathways. Survival Cox regression analyses indicated that target LIMD2 in the ceRNA network may act as an independent predictor of poor OS in ECA patients. On further analyses of immune infiltration and autophagy genes related to target LIMD2, LIMD2 showed a close linkage with immunity and autophagy in ECA. Furthermore, we verified the high expression level of LIMD2 by qRT-PCR based on 20 pairs of ECA and normal samples. Hence, LIMD2 is a potential promising molecular marker of prognostic significance in ECA.
The ceRNA networks may play a vital role in the development of cancer [29]. Seven circRNAs, 22 miRNAs and 34 mRNAs were identified in the ceRNA network. The increase of has_circ_0000154 (circDCAF6) identified in our study was related to tumor invasion, positive lymph node metastasis, and higher TNM stages in GC patients. It could serve as an independent prognostic indicator [30]. A total of 13 identified mRNAs (HMGA2, CCL22, CDKN2D, CEACAM6, DKK2, KIT, KRT14, LEF1, LPCAT1, NCCRP1, NRXN1, PPP1R18) were found to be related to ECA. In a previous study, High-mobility group AT-hook 2 (HMGA2) was shown to regulate transcription by inducing structural alterations in the chromatin [31]. Several studies have shown that HMGA2 is re-expressed in most tumors and plays a vital role in tumorigenesis [32]. The stability of HMGA2 may be regulated by hepatitis B X-interacting protein (HBXIP) via the Akt-PCAF pathway, thereby promoting the growth of ECA cells [33]. All 22 DEmiRNAs in the ceRNA network have been reported to be related to cancer. MiR-141-3p plays an important role in various carcinomas [34], and it was found to be highly upregulated in ECA cells [35]. A recent study reported that miR-141-3p may inhibit the expression of pleckstrin homology domain leucine-rich repeat protein phosphatase-2 (PHLPP2), a negative regulator of the PI3K/AKT signaling, and could serve as an biomarker in ECA [36].
CircRNAs with RBP binding sites can act as sponges for RBPs and may indirectly modulate their functions [5]. In this study, the six identified RBPs (SRSF1, SRSF2, PCBP2, TIA1, FUS, FMR1) were closely related to ECA. PCBP2 performs multiple functions, such as stabilization of mRNAs and silencing or promotion of translation [37]. Several studies have shown that PCBP2 may promote tumor growth. Ye et al. found that PCBP2 regulates the proliferation and apoptosis of ESCC cells and may serve as a novel therapeutic target in ESCC [37]. Our findings are consistent with those of the previous study.
In our study, multiple diseases were enriched in ECA, including Adult T-cell lymphoma (ATL), Childhood Langerhans cell histiocytosis, and classical Hodgkin’s lymphoma that are strongly associated with immune cells. ATL is a T-cell lymphoproliferative tumor of mature CD4 + CD25 + T cells [38], and Langerhans cell histiocytosis is characterized by the accumulation of Langerhans cells, antigen-presenting cells [39]. Our results also revealed that many enriched GO and KEGG terms were closely related to immune response, including positive regulation of T cell activation and positive regulation of immune system process. GSEA showed that immunity was strongly associated with ECA, including systemic lupus erythematosus and T cell receptor signaling pathway. The NF-κB signaling pathway and PI3K/AKT signaling were the most enriched pathways. Zheng et al. depicted the entire immune landscape, including the innate and acquired immune cell map, in ESCC and adjacent tissues; their work revealed that ESCC is enriched in immune-suppressive cell mass [40]. Tong et al. revealed that 14-3-3ζ may enhance the invasion and growth of ESCC cells by inhibiting the S1PR2 protein expression via the NF-κB pathway [41]. GSVA indicated enrichment of ether lipid metabolism in ECA. Disorders of ether lipid metabolism are vital signs of tumors, which serves as the basis of tumor pathogenicity [42]. Cao et al. found that lncRNAs may interact with their adjacent coding RNAs to modulate ether lipid metabolism [43].
Furthermore, we found that overexpression of 12 genes was strongly related to poor OS. We included these 12 mRNAs in Cox regression analyses, and identified LIMD2 as a vital target. LIMD2 an oncogene in a variety of tumors [44], in previous studies, expression of LIMD2 in lung [45] and thyroid cancer tissues [46] showed a strong association with cell movement, metastatic potential, and tumor grade. Zhang et al. revealed that high expression of LIMD2 may enhance the progression of non-small cell lung carcinoma (NSCLC); in addition, overexpression of LIMD2 was closely related to lymph node metastasis, distant metastasis, and advanced stage. The survival time of NSCLC patients with overexpression of LIMD2 was shorter than that of patients with lower expression, indicating that LIMD2 may serve as a therapeutic target in NSCLC [45]. Araldi et al. found that LIMD2 may strengthen phosphorylation of kinases related to epithelial-mesenchymal transition (EMT) and invasion, and modulate the crucial steps in the metastatic cascade of thyroid cancer cells through the mitogen-activated protein kinase (MAPK) crosstalk [47]. Previous studies have demonstrated a close relationship between LIMD2 and T cells CD4 memory activated, and positive relationship of LIMD2 with nine autophagy-related genes. Yu et al. found that the immunity high group of cutaneous melanoma specimens had the highest level of CD4 memory activated T cells, and the OS rate was poor [48]. A recent research unraveled the connection between clinical information and immune signatures in GC. They noticed that the high-risk group showed greater proportion of CD4 memory activated T cells and M1 macrophages [49]. Bcl2-associated athanogene 3 (BAG3) was shown to be involved in multiple biological processes, including cell proliferation, cell vitality, and apoptosis [50]. Compared with Barrett’s metaplasia (BE) and normal samples, the amount of T lymphocytes with downregulation of bcl2 was notably increased in EAC [51].
Nevertheless, some limitations of our study should be considered while interpreting the results. Firstly, owing to the use of several datasets in our analyses, the effect of inter-batch differences on our results cannot be ruled out. In addition, further study with a larger sample size is required to verify the high expression of LIMD2 in ECA. Finally, further in vitro and in vivo biological experiments are required for in-depth characterization of the functions of the identified targets and potential mechanisms in ECA.