Bronchopulmonary dysplasia is one of the most common complications arising in preterm infants, especially in those born underweight and those of small gestation weeks. It has been reported that up to 70% of babies born before 26 wk of gestation will develop BPD [25]. The progression of BPD is known to be driven by multiple mechanisms, with the participation of a few important protein and signaling pathways, such as thec vascular endothelial growth factor (VEGF), interleukin (IL), and phosphatidyl inositol-3-enzyme-serine/threonine kinase (PI3K-AKT) signaling pathway[26]. Therefore, it is important to clarify the pathophysiology of BPD and discover means of early diagnosis and treatment-related biomarkers. Bioinformatics analysis and efficient microarray might be conducive to our understanding of the molecular mechanisms of disease occurrence and development, thus helping the exploration of genetic alternations and identification of underlying diagnostic biomarkers.
In this study, we screened out 19 significant DEMs, of which 10 were shown to be upregulated, whereas 9 downregulated. Rcesults of functional enrichment analysis indicated that these significant DEGs were associated with the virus infection, antigen processing and presentation, B-cell receptor, phagosome, hematopoietic cell lineage, and CAMs signaling pathways in BPD. Among these signaling pathways, the CAMs pathway was also obtained in the analysis of GSE32472[22] and GSE125873[23]. In these 2 series they both identified the signaling pathway of T-cell receptor which was not obtained in our study, maybe due to the use of different grouping methods and sampling time. Key DEGs, such as CD19, CD22, CD72, CD74, MS4A1, and FCGR2B were identified as hub genes in PPI networks. Moreover, through the construction of the PPI network, we could recognize key genes, with which miRNAs might interplay with. Despite filtering the genes with the potential targets of the 19 significant DEMs, we could still identify 140 upregulated and 33 downregulated genes. Hence, considering the total number of DEMs, the enormous and complex miRNA-mRNA regulatory network could be unimaginable. The hub genes of a network are known to always be important, resembling "seeds", that could combine different signal pathways.
Furthermore, some of these DEGs were validated and found to be correlated with BPD. One of the DEGs, called adrenomedullin (ADM), which was found to be downregulated in our study, was shown to be regulated by hsa-miR-423-5p, hsa-miR-3940-5p, hsa-miR-767-5p, and hsa-miR-4301. Moreover, ADM has been shown to have potent angiogenic, anti-inflammatory, and antioxidant properties. Zhang et al. reported that ADM deficiency in human pulmonary microvascular endothelial cells (HPMEC) resulted in significantly increased the generation of hyperoxia-induced reactive oxygen species and cytotoxicity compared with ADM sufficient HPMEC, finally causing BPD [27]; however this finding remains to be validated. Likewise, WNT 3/16 were demonstrated to be upregulated in our study through many miRNAs, such as the underexpressed hsa-miR-767-5p, hsa-miR-5681b, hsa-miR-423-5p, hsa-miR-3940-5p, and hsa-miR-3960.The WNT family have also been found to be associated with the development of BPD. Hyperoxia is known to increase the expression of WNT2b, WNT 5a, WNT 9a, and WNT 16, and decrease the expression of WNT 4, WNT 10a, and WNT 11 [28]. The WNT family of proteins includes a large number of members that control a variety of developmental processes, including cell fate, proliferation, polarity, and migration [29]. Li et al. found that patients with BPD were characterized by an increased activity of Wnt/β-catenin [30]. Similar to that, we also found an increased expression of WNT 16 in BPD, but the mechanisms by which WNT3 might cause BPD remain to be explored. Among the identified DEGs, we also found the upregulated TLR10, which was shown to be regulated by miRNA, such as the underexpressed hsa-miR-767-5p, hsa-miR-5681b, hsa-miR-423-5p,hsa-miR-3940-5p, and overexpressed hsa-miR-33a-5p, and hsa-miR-337-5p, to be enriched in the immune response process term of the GO-BP category. Toll Like Receptors (TLRs) are known to play an important role in regulating inflammation, maintaining mucosal homeostasis and preventing bacterial invasion [31]. Rising evidence has implied that the TLR signaling pathway is the pivotal component of the pulmonary homeostatic program that abrogates lung inflammation and injury[32]. Many studies were aimed at Toll-interleukin 1 receptor domain-containing adaptor protein (TIRAP). Researchers have also found that TLR5 and TLR4 were associated with the occurrence of BPD via the MyD88-dependent pathway[33, 34], and TLR10 was reported to active the TRL4 signaling pathway. So, TLR10 might also be related to the occurrence of BPD; another finding that requires confirmation.
In this study, we screened out 19 DEMs, suggested to modulate the expression of DEGs and contribute in the regulation of many pathways. Besides, we also found that single miRNA could interplay with many mRNA species, as well as that a single mRNA could also interplay with many miRNA species. Although most of them have not been reported in the mechanisms so far studied in patients with BPD, we could still obtain some information from existing studies. One such case was the hsa-miR-15b-5p, one of the identified DEMs in our study. Zhang et al. found that miR-15b-5p was upregulated in BPD mice[35]. Fu et al. have also reported that it has a protective action against oxidative stress in HG-stimulated podocytes [36], while Ezzie et al. found that it was increased in patients with chronic obstructive pulmonary disease (COPD) and could potentiate the progression of fibrosis in lung tissues [37].As such, overexpression of the hsa-miR-15b-5p in the BPD blood samples in our results, might indicate a similar underlying association. Another DEM, hsa-miR-301a-3p, which was overexpressed in our study, was demonstrated to modulate DEGs, such as TLR10, CD72, and BMP3. This effect has been previously observed in animal experiments. The study by Dong et al. showed the overexpression of miR-301a in a murine model of hyperoxia-induced bronchopulmonary dysplasia[38]. Therefore, hsa-miR-301a-3p might also play a role in the mechanism of BPD development in infants, which remains to be validated.
Although we investigated the miRNA-mRNA regulatory pathway in BPD using bioinformatics methods, our study had some limitations that should be clarified. First, the samples were limited and might have led to high false-positive rates and one-sided results. Therefore, it is required to improve the detection power by integrating more datasets in future studies. Second, the source of microarray data was only from blood samples. Body fluids that could be noninvasively obtained in the clinic, such as sputum and urine might also contain miRNAs. Third, to confirm the mechanisms of hub genes related to BPD, it will be helpful to add some in vitro or in vivo experiments to validate our results. Forth, due to the absence of clinical data, we were unable to assess the relationship between DEMs and the severity of BPD. More clinical and demographic characteristics of infants with BPD are required for further analysis. Finally, experimental evidence, obtained from wet research, such as western blot, real-time PCR and immunohistochemistry assays are required to better delineate the role of hub genes and the potential mechanisms of BPD.
In this study, we found multiple miRNA-mRNA regulatory pathways and potential biomarkers of BPD, in line with our current knowledge of the pathophysiology of this disease. We believe that this hypothesis-generating study offers a new insight into the molecular mechanisms of BPD through the and identification of several latent biomarkers that could be used toward its diagnosis and treatment.