Lung adenocarcinoma is one of the most common tumor types in the respiratory system, with a poor prognosis and a high metastasis rate. However, the mechanism of LUAD metastasis is still unclear. In this study, we uncovered the prognostic ceRNA and immune cells between primary LUAD with or without distant metastasis, and based on them, we constructed two prediction nomograms. Both the two nomograms achieved a good reliability. We found that has-miR376a-3p and PIK3R1 are related based on the ceRNA network. In addition, we found the potential regulatory networks among NR3C2, PIK3R1, PTPN13, SPRY2 and mast cells resting, which may take part in the distant metastasis of LUAD.
The ceRNA networks are composed of protein-coding mRNAs and ncRNAs. The latter includes miRNAs and lncRNAs. Generally, they play crucial roles in the tumorigenesis and metastasis (20). In this study, we identified the ceRNA networks that regulate LUAD metastasis with 31 DEGs. The prediction nomogram was then established and the AUC value of 3-year survival and 5-year survival was 0.727 and 0.723, respectively. Based on the hypergeometric testing and correlation analysis, has-miR376a-3p (miRNA) and PIK3R1 (protein-coding RNA) were shown to be associated with each other. Besieds, the correlation analysis between ceRNAs and immune cells uncovered that NR3C2, PIK3R1, PTPN13, SPRY2 were positively associated with mast cells resting. Therefore, in our study, we focused on the roles of has-miR376a-3p, NR3C2, PIK3R1, PTPN13, SPRY2 and mast cells resting.
miRNAs are small, single-stranded, noncoding RNAs that can post-transcriptionally regulate the expression of various oncogenes and tumor suppressor genes. miR376 is one important microRNA family for cancer formation and progression and has been widely reported in gliomas, leukemia, breast and ovarian cancers(22). miR376a is a member of miR376 family and participates in the regulation of macroautophagy(23). In this study, we found has-miR376a-3p may regulate PIK3R1 in the distant metastasis of LUAD.
PIK3R1 is the predominant regulatory isoform of PI3K and one of the most common alterations involved in the tumorigenesis(24). It is a tumor-suppressor gene and its gain-of-function often leads to developmental defects in helper and regulatory T-cell subsets(25). In ovarian cancer, the copy number loss of PIK3R1 (p85α) most commonly occurs and PIK3R1 loss activates AKT and p110-independent JAK2/STAT3 signaling(26). In the NSCLC, PIK3R1 is involved in the suppression of miR-486-5p on tumor cell growth(27). Our data of has-miR376a-3p regulating PIK3R1 is also a potential mechanism of LUAD. In addition, we also uncovered that PIK3R1, NR3C2, PTPN13, SPRY2 and mast cells resting may interact with each other and regulate LUAD metastasis.
NR3C2, a tumor-suppressor gene, is associated with the cancer progression of renal, colonal and pancreatic cancer(28). It also has the prognostic effects. For example, its decreased expression is related to the poor prognosis in renal carcinoma(29). Many miRNAs take part in the regulation of NR3C2. In colon adenocarcinoma, miR‐4709 overexpression could facilitate cancer proliferation and invasion via downregulating NR3C2(28). In pancreatic cancer, NR3C2 is also reported to be the target of miR-135b-5p, and mediate the tumor cells migration, invasion and epithelial-mesenchymal transition(30). In NSCLC cells, NR3C2 is also regarded to be the potential tumor suppressing gene, similar to our study(31).
PTPN13, a tumor-suppressing gene, is the member of tyrosine phosphatase gene superfamily(32). In lung cancer, PTPN13 is negatively correlated with grade and stage. It can inhibit the anchorage-independent growth and invasion by inactivating the EGFR and Her2/ErbB2 (33, 34). The suppressing role of PTPN13 is also related to IRS‐1/PI3K/Akt signaling pathway(35), which may be regulated by the above-mentioned PIK3R1. Besides, PTPN13 is also regulated by miR‐30e and suppresses tumor cell proliferation and colony formation via EGFR/AKT signaling in lung cancer(36). The metastasis role of PTPN13 was reported to be associated with P-FAK level, which had a significant positive correlation with the TNM classification(35).
The Sprouty (SPRY) family are important modulators of receptor tyrosine kinase (RTK) signaling. As a member of the family, SPRY2 works as a bimodal regulator (37). Its roles in tumorigenesis and metastasis are contrasting different in various cancer contexts. On the one hand, it could negatively regulate HGF-mediated ERK and AKT signaling in human leiomyosarcoma; on the other hand, SPRY2 induces mesenchymal-epithelial transition (MET) activation to promote cell migration and invasion in colonic adenocarcinomas(38, 39). In NSCLC, Spry2 antagonizes RTK-induced signaling at different levels and inhibits the tumor growth, indicating feasibility for the usage of Spry in targeted gene therapy of NSCLC(40).
Mast cells is an important part of the tumor microenvironment (TME). In the proliferation and metastasis of tumor cells, mast cells also play important roles. The proliferation rate of A549/LLC cells was markedly increased by mast cells(41). The cell-to-cell contact interactions between mast cells and tumor cells can result in the activation of mast cells, which in turn increase phosphorylation of ERK1/2 MAPK/Akt in a PI3K dependent fashion and promote tumor growth(42). In addition, the exosomes from A549 cells which contain stem cell factor (SCF) can be taken up by mast cells and activate mast cells through SCF-KIT signal transduction, thereby leading to mast cells degranulation and the release of tryptase. Tryptase can subsequently accelerate the tumor cell proliferation and migration through the JAK-STAT signaling pathway(43). This phenomenon was also reported in the breast cancer metastasis(44).
Although this study provides potential roles of ceRNAs and immune cells in LUAD metastasis, it still has some limitation. First, as a bioinformatics study, the results have not been examined by cell and molecular experiments. Thus, our further plan is to verify it by biological experiments. Second, the samples in this study are from America, and the applicability of prediction model needs further validation by European and Asian data. Third, the sample size of RNA-sequencing data is relatively limited and further large samples should be used to validate the prediction models.