In the present study, we used three gene expression profiling datasets in the GEO database for the identification of overlapped DEGs in ACC compared with normal adrenal tissues. Together, 10 up-regulated and 28 down-regulated DEGs were screened out. Clustering analysis revealed a certain inconsistency of the expression of DEGs within the same dataset, especially in GSE90173 and GSE19776 datasets (Fig. 2A&B), highlighting the heterogeneity of ACC. The phenomenon that the GSE14922 dataset showed obvious and clear stratification might be partially attributed to the small number of specimens involved (Fig. 2C). Next we aimed to identify novel hub genes among these 38 DEGs within a series of procedures. 7 DEGs shown known associations with ACC on Pubmed were ruled out first (Table 1), among which CDK1, CCNB1, PRC1, and CDKN1C were involved in the regulation of cell cycle while IGFBP6 interacted with IGFs. The expression values and prognostic significance of the remaining 31 novel DEGs were cross-checked among the three datasets, the GEPIA2 database, and the starBase database to increase data reliability and reduce false positives, as a result of which 7 up-regulated and 2 down-regulated DEGs were identified. To further narrow down the range of promising hub genes, 7 up-regulated DEGs were inputted into the GeneCards database and two hub genes were retrieved showing gene-tumor associations. Together, the identification of two up-regulated and two down-regulated novel hub genes in ACC was completed. The survival analysis indicated that RACGAP1 had the greatest impact on the OS of ACC, followed by TPX2, FMO2, and HOXA5 (Fig. 3). Since the staging system in ACC is the most important factor in the prediction of prognosis, we analyzed the expression profile of four novel hub genes in different stages of ACC and found that TPX2 and RACGAP1 were positively correlated with the staging, indicating a promising role of such genes in the stratification and the prediction of prognosis in ACC [11]. Interestingly, co-expression analysis revealed that the association between TPX2 and RACGAP1 was the strongest among all contrast groups and that the expression of HOXA5 was almost completely independent of that of RACGAP1 and TPX2 (Fig. 5). Therefore, a PPI network was constructed to visualize the possible connections of four novel genes and seed genes in ACC (Fig. 6). Surprisingly, HOXA5, TPX2, and RACGAP1 were somehow all associated with TP53, indicating that at least two pathways were involved. And the phenomenon that RACGAP1 was associated with TP53 through TPX2 partially explained the aforementioned strong association between TPX2 and RACGAP1, implying that future studies concerning TPX2 and RACGAP1 in ACC should be warranted.
TP53 encodes the transcription factor p53, which plays an important role in the inhibition of cancer cells. Along with the accumulation of p53, TP53 inactivation mutation, the incidence of which is around 15–70%, is one of the most common molecular events in the formation of ACC and is closely associated with a larger volume, more advanced stage, a higher rate of metastasis, and shorter disease-free survival in ACC [16–21]. Furthermore, TP53 germline mutations are more likely to occur in pediatric ACC and are associated with worse outcomes [22], emphasizing the magnitude of genetic counseling in patients with ACC, especially in those of suspected hereditary cancer syndrome [23]. It has also been reported that ACC is an integral part of Li Fraumeni Syndrome (LFS), which is a familial cancer predisposition caused by germline mutations of TP53 [19]. Since TPX2, RACGAP1, and HOXA5 were found to be associated with TP53 in this study, it could be possible that these three novel hub genes might play important roles in the formation of ACC.
TPX2 gene encodes a microtubule-associated protein, which serves a vital role in mitotic spindle formation and chromosome segregation process [24, 25]. The overexpression of TPX2 has been reported in many studies and related to poor clinical outcomes of multiple malignancies, such as lung cancer, thyroid carcinoma, clear cell renal carcinoma, gastrointestinal cancer, and hepatocellular cancer, implying the possibility that TPX2 can be used as a prognostic marker and potential therapeutic target in these tumors [26–34]. Furthermore, TPX2 was reported to be involved in the invasion of tumor cells by the elevation of MMP2 and MMP9 through AKT signaling in hepatocellular carcinoma [35, 36]. With the help of MYC, TPX2 could also cooperate with AURKA to drive the tumorigenesis of colon cancer, providing a promising therapeutic option for MYC-driven tumors by inhibition of the AURKA-TPX2 axis [37, 38]. In terms of the association between TPX2 and p53, a novel regulatory circuitry of TPX2-p53-GLIPR1 was found in bladder cancer to regulate proliferation, migration, invasion, and tumorigenicity, while silencing of TPX2 could activate p53 signaling and inhibit the proliferation of breast cancer cells [39, 40]. Together, these data suggest that TPX2 might be a promising target for the prediction of prognosis and therapeutic approaches in ACC and future studies concerning TPX2 in ACC should be warranted.
RACGAP1, which is located in the mitotic spindle and participates in Rho mediated inactivation of various signals, is a member of GTPase activating proteins (GAPs). RACGAP1 is a known regulator of cytokinesis during the normal cell cycle and plays an important role in cell growth regulation, differentiation, cell transformation, invasion, migration, oncogenesis, progression, recurrence, and metastasis [41, 42]. RACGAP1 has been correlated with poor outcomes in various tumor types, including breast cancer, hepatocellular carcinoma, gastric cancer, colorectal cancer, epithelial ovarian cancer, esophageal carcinoma, and invasive cervical cancer [43–48]. Interestingly, RACGAP1 was reported to be a target gene of mutant p53 [49, 50]. Furthermore, AURKA is reported to be directly correlated with the expression of RACGAP1, which is a modulator of the classical Wnt signaling pathway [44]. Since AURKA has been recognized to be closely associated with TPX2 and the Wnt signaling pathway plays an important role in ACC, whether the association between TPX2 and RACGAP1 through AURKA exists and whether RACGAP1 could function through Wnt signaling pathways in ACC are worth exploring [3, 37].
HOXA5 is a member of the HOX family that participates in various cellular processes. The methylation status of HOXA5 has also been suggested to be associated with tumor prognosis in multiple studies [51, 52]. In breast cancer, the loss of HOXA5 expression, resulting from methylation of its promoter regions, could lead to loss of p53 expression and the functional activation of Twist, and therefore promotes breast tumorigenesis [53–55]. Furthermore, the cooperation between HOXA5 and p53 is found to be able to inhibit the invasion of tumor cells, at least partially owing to the decrease of MMP2 activity in NSCLC. [56]. Low expression of HOXA5 is associated with high levels of nuclear β-catenin, which is the hallmark of the Wnt signaling pathway in colorectal carcinoma [57]. Moreover, HOXA5 could also negatively regulate the Wnt canonical pathway in the developing lungs [58]. Since RACGAP1 is a modulator of the classical Wnt signaling pathway [44], whether there is a connection between RACGAP1 and HOXA5 remains to be explored. Together, it is highly promising that the regulation of HOXA5 could provide a potential therapeutic approach in ACC [59].
FMO2, a major isoform mainly and greatly expressed in the human lung, is a member of a superfamily of monooxygenase genes. FMO2 has been shown to regulate oxidative stress levels by producing metabolites that promote the release of ROS in the form of hydrogen peroxide [60]. However, there is little evidence to support the association between FMO2 and tumorigenesis. In experimental animals, the expression of the FMO2 gene is regulated by sex hormones, and putative glucocorticoid responsive elements are found to be located in the 59-flanking regions of the rabbit FMO2 gene, indicating that FMO2 may be regulated by the excessive release of hormones in ACC [61]. Whether FMO2 contributes to the development of ACC and how it functions deserves future investigation.
There were still some limitations in this study. The sample size of the datasets we used was limited, and it might be more statistically significant to adopt a larger sample size for statistical analysis. In addition, this study was completely based on in silico evidence, which needed to be verified by experiments and real-world clinical samples. And due to the default limitations of the online analysis tools we used, unexpected variances might occur during the analyses. In spite of these limitations, we still believe that this study can still provide researchers with some potential research targets in ACC.
In conclusion, this study identified four novel hub genes using integrative analysis of gene expression profiling in ACC. TPX2, RACGAP1, and HOXA5 may function through the regulation of p53 while FMO2 may be associated with the excessive release of hormones. Therefore, this study provided potential prognostic and therapeutic targets for future experimental and clinical investigation of ACC.