The two gene expression microarray datasets, GSE59701 and GSE88804, were obtained from GEO. Using the limma package in R software, a total of 2190 DEGs, including 1131 upregulated and 1059 downregulated genes, were obtained from two expression profile data (Table.1).
Table 1
Table 1
The number of upregulated and downregulated genes in ACC compared to normal tissue
references | GEO Accession number | Platform | Sample | DEG |
ACC | normal | UP | DOWN |
Gao R, et al. (2015) | GSE59701 | GPL6244 | 12 | 12 | 696 | 496 |
Andersson MK.et al. (2017) | GSE88804 | GPL6244 | 13 | 7 | 832 | 926 |
Total (Non-repetitive) | | | 25 | 19 | 1131 | 1059 |
The GO analysis and KEGG signaling pathway enrichment of the 2190 DEGs were performed using the Enrichr database. We considered adjusted P value < 0.05 as the threshold to get meaningful pathways. Seventeen statistically significant pathways have been upregulated (Fig. 1). PI3K-Akt signaling pathway, Cell cycle, central carbon metabolism in cancer, focal adhesion, ECM receptor interaction, Wnt, axon guidance, microRNAs in cancer, and Ras signaling pathways are important pathways in ACC obtained from enrichment.
Results exhibited 33 downregulated pathways with the adjusted P value < 0.05 (Fig. 2). They can be divided into three groups. One group is related to salivary secretion. The second group is related to lipid metabolism and adipocyte differentiation including, the PPAR signaling pathway, AMPK, and adipocytokine signaling pathway. These three pathways show a high expression correlation in the heat map (Fig. 3). The third group is related to immune response and inflammation that cluster together in the heat map (Fig. 3). They include Rheumatoid Arthritis pathway, TNF signaling pathway, NOD-like receptor signaling pathway, NF-kappa B signaling pathway, IL-17 signaling pathway, and Phagosome.
Analysis of significant enriched pathways including PI3K-Akt, Ras, Wnt and, cell cycle that identified by Enrichr shed more light on the procedure of tumorigenesis of ACC. Figure 4 shows the upregulated axis in ACC. Protein interactions are based on KEGG pathways.
The PPI network of 761 overlapped DEGs was constructed using the STRING database and Cytoscape software (Fig. 5). Hub genes were obtained using the four mentioned methods separately. Thirty genes with the highest degree are shown in orange and yellow and others are shown in blue. Twenty of these genes were confirmed by three other methods of network analyses. They include TP53, EZH2, NOTCH1, CTNNB1, GNG2, APP, MET, KIT, PLCG1, LEF1, that are upregulated ones and BMP4, PPARG, IGF1, C3, CCL5, COX2, PRKCA, ERBB4, ADIPOQ, and, EGF are downregulated hub genes.
The substantial participation of some of the twenty hub genes, such as TP53[18], NOTCH1[19], CTNNB1[20], MET[5], and KIT[21] in the ACC tumorigenesis, has already been studied. The others need future studies to clarify their role in the ACC. The calculation of AUC for 20 hub genes was performed to validate their potential as a diagnostic biomarker (Fig. 6a). The CTNNB1, NOTCH1, PLCG1, PRKCA, and TP53 genes have an AUC of more than 0.98, indicating their high specificity and sensitivity in distinguishing ACC samples from normal. Figure 6b shows the ROC curve for these five genes. To investigate the relationship between hub genes expression and MYB oncogene, the correlation between their expressions was calculated and shown by a heat map (Fig. 7). We found that in addition to TP53, CTNNB1, and NOTCH1, which have a decisive role in ACC[22],[23],[18], PLCG1 is a gene that its expression had a high correlation with MYB expression. This observation, together with the result obtained from the ROC curve propose this gene as a new diagnostic or therapeutic biomarker for future studies in ACC. A considerable role for PLCG1 in some cancers has been reported. In breast cancer, high expression of phosphorylated PLCG1 predicts metastasis in patients undergoing adjuvant chemotherapy[24]. In another study, PLCG1 inhibition induced programmed cell death in lung adenocarcinoma A549 cells[25].
EZH2 is another hub gene that appears to be particularly important in tumorigenesis. EZH2 participates in histone methylation of some tumor suppressor and inhibit them. The protein is only found in actively dividing cells [26], so it can be used as a diagnostic marker of dividing cells[27]. EZH2 can interact with Wnt signaling factors like MYC oncogene and cyclin D1 [28]. There is some FDA approve EZH2 inhibitors for treating different cancers[10]. Based on this evidence, investigation on EZH2 in ACC can be of great importance. In the following discussion, we will focus on the IGF-IR pathway and the PPARG pathway as two potential therapeutic pathways in ACC.
Genomic sequencing data, and cytogenetic maps revealed that most ACC cases have translocations that lead to the juxtaposition of NFIB, TGFBR3 and RAD51B super-enhancers either upstream or downstream of MYB locus. MYB oncogene binds these translocated super-enhancers, loops to the MYB promoter, and cause positive feedback that increases itself expression[29],[11]. Increased MYB transcriptional regulatory activity promotes tumor cell proliferation in ACC by regulating genes involved in the RNA processing, cell cycle, and DNA repair highlighting MYB as a potential therapeutic target[4]. Interestingly, Andersson et al. found that IGF-1R /IR inhibition with linsitinib decreased the MYB-NFIB fusion product [2]. As a result, MYB–NFIB expression can be regulated by inhibiting the IGF1R pathway [2],[5]. However, the use of IR/IGF-IR inhibitors in the treatment of ACC was not very successful. Despite the use of these drugs inhibited tumor growth in xenografted ACC, but it did not affect cancer cell apoptosis[5]. Moreover, IGF-IR inhibition had a short-term clinical response, and the patient became resistant to treatment after a few months [2]. The reason for this drug resistance is the interaction of the IGF-1R pathway with other signaling pathways [30]. The IGF system has two ligands; IGF-1 and IGF-2, and three receptor; IGF-1R (primarily), IGF-2R, and the insulin receptor (IR), that IR itself has two variants named IR-B and IR-A[31]. According to DEGs analysis, IGF-2, IR and IGF-1R have been upregulated and IGF-1 has been downregulated in ACC. Similar to previous studies on the IGFIR-AKT axis in ACC [2], we found upregulation of PI3K-AKT signaling pathways in gene enrichment. Also, the analysis of the increased pathways obtained from KEGG showed the pivotal role of PI3K-AKT and RAS signaling pathways (Fig. 4), which are downstream pathways of IGF-IR/IR [32]. Mitogen signaling by IR has been described in some tumor models and several examples have been provided in which the IGF1R or IR compensates the inhibition of the opposite receptor[31]. Recent evidence has shown that many cancer cell types, including prostate, colorectal, breast, and lung cancers express not only the IGF1R but also the IR-A, the isoform with high affinity for both insulin and IGF-2 and is also associated with a poor prognosis[33]. By activating IR-A, IGF-IR and IGF-1R/IR-A hybrid, IGF-2 can function as part of the drug resistance development system against IGF-1R inhibitors [31][34][35][36]. One solution is to study the effects of other IGFR/IR inhibitors[5]. However, IGF-1R/IR inhibitors cannot distinguish between IR-A and IR-B, and interfere with glucose metabolism leading to insulin resistance and hyperglycemia[30]. On the other hand, there is a link between hyperglycemia and cancer that may arise from preferential expression of IR-A[33]. IGF-II activates PI3K-Akt signaling in ACC through stimulation of IGF-IR and IR-A. One solution to these problems is to target IGF-II ligands directly. Because, in addition to have anti-proliferative activity, IGF-II inhibitors do not interfere with IR-B function. Dusigitumab (MEDI-573) is an IGF-1/IGF-2 co-neutralizing mAbs with a stronger binding affinity for IGF-2 than IGF-1. It has anti-proliferative activity in vitro and in vivo in preclinical models[37]. Therefore, IGF-2 could be a valuable new therapeutic target for ACC that has not been studied in ACC patients and requires future studies.
The second pathway that can be considered in ACC treatment is the PPARG pathway. Based on signaling pathway enrichment ACC's pathogenesis is mainly linked to lipid metabolism. Lipid metabolism signaling pathways including adipocyte signaling pathway, PPARG and AMPK have been downregulated in ACC. Figure 3 shows the high correlation of these pathways. Interestingly, there is a link between lipid metabolism and the IGF-1R pathway. IGF-1 promotes preadipocyte proliferation and differentiation, however IGF-IR abundance increase with adipocyte dedifferentiation[38]. IGF-2 has an inhibitory effect on the differentiation of visceral adipocytes that confirmed by reducing PPARG and ADIPOQ, two differentiation markers of adipocytes. Visceral adipocyte plays a substantial role in the pathogenicity of various diseases such as metabolic syndrome, type 2 diabetes, and cardiovascular risk[38]. IR-A is the predominant isoform in visceral preadipocytes and makes them more responsive to IGF-2. IR-B predominates in subcutaneous preadipocytes, so the binding of insulin to these cells has metabolic effects. Many types of tumors (breast, gastric, renal, colon, and ovarian) grow in the proximity of visceral adipocytes and induce dedifferentiation of visceral adipocytes into pre-adipocytes or reprogram them into cancer-associated adipocytes. Dedifferentiation of adipocytes causes the release of fatty acids into tumor microenvironment and supports the tumor growth [39]. If differentiation of these preadipocytes is induced again, the process of carcinogenesis may be prevented [40]. PPARG plays a critical role in adipocyte differentiation, insulin sensitization, lipid metabolism, and carcinogenesis[41]. PPARG pathway is strictly inhibited in ACC samples rather than normal samples (Fig. 2). We also found PPARG and ADIPOQ as hub genes in the PPI network. PPARG belongs to the nuclear hormone receptor superfamily named Peroxisome proliferator activated receptors (PPARs). Different PPAR pathways containing α, β, γ, and δ have been identified [39]. Only PPAR γ (PPARG) pathway has been inhibited in ACC. Several studies showed a significant reduction in PPARG expression in follicular thyroid cancer, esophageal cancer, cervical carcinoma, and colon cancer [41]. Activation of the PPARG pathway with its agonists may prevent tumor growth and proliferation by inhibition of PI3K and the Ras, downstream pathways of the Insulin/IGF axis [42]. Thiazolidinediones (TZD) are the most widely used synthetic agents bind to PPARG and activate it. After activation, PPARG moves to the nucleus and binds DNA to regulate the transcription of several genes, which ultimately increases the storage of fatty acids in adipocytes and differentiation of adipocytes. It also, reduces circulating fatty acids and improves insulin sensitivity[43] [44]. Ciglitazone, a synthetic PPARG ligand, prevents the proliferation of A549 cells (human alveolar adenocarcinoma cells)[41]. PPARG activation by rosiglitazone and pioglitazone substantially induced apoptosis and cell cycle G2 arrest in bladder cancer cells [45]. Activated PPARG performs its inhibitory role in cell growth and proliferation by improving cell differentiation [46]. Although the connection between the PPARG and IGF pathways is not clearly recognized but, the therapeutic function of PPARG is observed in tumors which IGF pathway is upregulated [47]. In light of these pieces of evidence, PPARG agonists may be potentially preventive and therapeutic agents in ACC. In support of this hypothesis, there is a report that found metformin usage significantly improved DFS( Disease-Free Survival) in ACC [48]. The use of these drugs can complement the effect of TKI drugs in ACC. Interestingly the use of metformin in A549 cells reduced PLCG1 levels and induced autophagy[49], so there is a need for further research to uncover the effect of PPARG activating drugs in the treatment of ACC.
The third group of pathways that have decreased with a high correlation in the ACC (Fig. 3) were pathways related to inflammation and the immune system, which included TNF signaling pathway, NF-kappa B signaling pathway, NOD-like receptor signaling pathway, Rheumatoid arthritis pathway, phagosome, adipocytokine signaling pathway, and IL-17 signaling pathway. Although, the progression and invasion of cancer cells are mediated by proinflammatory factors in the tumor microenvironment, tumor-derived factors sometimes disrupt the host immune system, leading to anti-inflammatory conditions in the tumor microenvironment. This immunosuppressive situation is associated with tumor progression and poor prognosis for patients with advanced cancer[50]. Identifying the immune system inhibition process in the ACC and the role of immunosuppressive factors derived from tumor cells in disease progression provides new insights about ACC treatment through the host immune system activation.