Accumulating evidence has shown that circRNAs and their mediated ceRNA regulatory networks play a vital role in pathogenesis and progression of various human cancers, including PTC. hsacirc0058124 was previously identified to be upregulated in PTC and associated with poor prognosis. By sponging hsa-miR2185p, hsacirc0058124 promoted proliferation, invasion, and metastasis of PTC via the NOTCH3/GATAD2A signaling axis [21]. Through regulation of ABCA9 and MTA1 via sponge hsa-miR-1179 and hsa-miR-1205, upregulated hsa_circ_0039411 promoted proliferation, migration, and invasion of PTC cells and inhibited cell apoptosis [22]. CircRNAs are resistant to exonase and RNase R, enabling them to be more stable than other types of ncRNAs as well as play an important regulatory role in cancer cell. The current understanding of the circRNA-related ceRNA network in PTC is limited, requiring further study.
In this study, DE-circRNAs, DE-miRNAs, and DE-mRNAs with potential ceRNA regulatory relationships in PTC were identified through comprehensive analysis of targeting relationship prediction and differential expression data. Based on the further correlation analysis of expression between DE-miRNAs and DE-mRNAs with potential ceRNA regulatory relationships, a circRNA-miRNA-mRNA regulatory network was constructed in PTC. DE-mRNAs within the ceRNA network associated with PFI of PTC were further identified to construct a ceRNA regulatory subnetwork associated with PFI. Six hub DE-mRNAs, namely CLCNKB, FXBO27, FXYD6, RIMS2, SPC24, and CDKN2A, were identified to be most significantly related to the PFI of PTC, which were regulated by three DE-miRNAs, including hsa-miR-146b-3p, hsa-miR-139-5p, and hsa-miR-139-3p. They were subsequently regulated by eleven DE-circRNAs via the ceRNA mechanism, including hsa_circ_0003645, hsa_circ_0089153, hsa_circ_0005699, hsa_circ_0007146, hsa_circ_0038718, hsa_circ_0001658, hsa_circ_0008784, hsa_circ_0000965, hsa_circ_0001917, hsa_circ_0008354, and hsa_circ_0049271. These newly identified key circRNA-miRNA-mRNA regulatory relationships may help elucidate the molecular mechanism of progression in post-surgical PTC. Moreover, they may provide novel therapeutic targets for treatment of recurrence in PTC. Hub DE-mRNAs identified in the ceRNA network were potential predictors of PFI in PTC.
Although most patients with PTC have a relatively good prognosis, a portion of patients will eventually develop post-surgical recurrence. These high-risk PTC patients should be distinguished early enough to adopt more aggressive treatment options, additional adjuvant radioactive iodine therapy, thyroid hormone suppression therapy with a higher dosage, and timely intervention to prevent post-surgical progression if necessary. In contrast, for the remaining low-risk PTC patients, aggressive treatment options should be avoided to improve quality of life. Therefore, it is critical to accurately predict the post-surgical risk of progression in patients with PTC. Currently, the ATA guidelines recommend the use of AJCC staging system and MACIS score to predict the risk of postoperative mortality. ATA risk stratification system was recommended to assess post-surgical risk of recurrence. The existing risk assessment tools are not sufficiently accurate and a novel prediction system should to be established to more accurately predict the prognosis of PTC patients. In the current study, we established a circRNA associated ceRNA network in the PTC. Based on this, six hub PFI related DE-mRNAs of the PTC were identified. A six-DE-mRNA signature was subsequently established, which was able to distinguish high-risk PTC patients from the low-risk ones and could accurately predict the PFI. Nomogram has been widely applied in oncology to assess the prognosis of cancer patients [23]. Nomogram can integrate various prognostic factors, including molecular and clinical parameters, and provide a visual graphical interface for personalized prediction of clinical events. In this study, to establish a more accurate assessment tool for prognosis evaluation in post-surgical PTC, a prognostic nomogram was established incorporating the DE-mRNA signature and clinicopathological parameters. This nomogram was able to accurately predict PFI of PTC and was better than the ATA risk stratification system, MACIS score and AJCC staging system recommended by ATA guidelines.
Currently, 11 DE-circRNAs were identified to regulate the six hub DE-mRNAs via the ceRNA mechanism in the study. The role of hsa_circ_0089153 in PTC has previously been reported. hsa_circ_0089153 was upregulated in clinical specimens of PTC [24]. In vitro experiments indicated that downregulation of hsa_circ_0089153 expression could inhibit proliferation, migration, and invasion of PTC cells. The luciferase reporter assay confirmed that hsa_circ_0089153 sponged hsa-miR-145 to mediate upregulation of ZEB2 expression, thereby playing a carcinogenic role in PTC. hsa_circ_0089153 was also reported to be upregulated in gastric cancer and bladder cancer [25, 26]. Our study indicated that upregulated hsa_circ_0089153 may sponge hsa-miR-139-3p, and upregulated SPC24 and CDKN2A expression to promote PTC progression. hsa_circ_0003645 was previously identified to be upregulated in non-small cell lung cancer tissue [27]. hsa_circ_0038718 was reported to be upregulated in hepatocellular carcinoma [28]. Our result suggested that hsa_circ_0003645 and hsa_circ_0038718 may also play an oncogenic role as a sponge of hsa-miR-139-3p. The function of hsa_circ_0001917 in PTC has not been reported. However, hsa_circ_0001917 was also identified to be upregulated in hepatocellular carcinoma [28]. Our result suggested that the upregulated hsa_circ_0001917 may sponge hsa-miR-139-5p, promoting PTC via upregulation of RIMS2. The function of hsa_circ_0005699 in cancer was controversial. hsa_circ_0005699 was previously reported to downregulate MCM8 and NCAPD2 expression by acting as a sponge of hsa-miR-504, functioning as tumor suppressor in gastric cancer [29]. Our study suggested that hsa_circ_0005699 may promote PTC as a sponge of hsa-miR-139-3p and upregulate SPC24 and CDKN2A expression. The function of the remaining six circRNAs in cancer is still known. Our study suggests that the upregulation of hsa_circ_0007146 may promote PTC through the sponge of hsa-miR-139-3p. hsa_circ_0001658, hsa_circ_0008784, and hsa_circ_0000965 may upregulate RIMS2 expression as a sponge of hsa-miR-139-5p, playing an oncogenic role in PTC. Moreover, downregulated hsa_circ_0008354 and hsa_circ_0049271 may sponge hsa-miR-146b-3p and suppress the expression of CLCNKB, FBXO27, and FXYD6. The function of these circRNAs in PTC require further experimental validation.
Six hub DE-mRNAs markedly associated with the PFI of PTC were identified in the current study. Upregulated SPC24 and CDKN2A were identified as targets of hsa-miR-139-3p. The tumor suppressing role of hsa-miR-139-3p has been identified in multiple tumors [30]. SPC24 is an important component of kinetochore-associated NDC80 complex. This complex mediates chromosome segregation and spindle checkpoint activity. SPC24 plays an important role in maintaining the integrity of kinetochore [31]. SPC24 was previously identified to be highly expressed in anaplastic thyroid cancer. Knockdown of SPC24 expression inhibited cell growth and invasion and promoted tumor cell apoptosis [32]. SPC24 was also reported to be highly expressed in hepatocellular carcinoma and was an independent predictor of survival [33]. Downregulation of SPC24 inhibited growth, invasion of tumor cells, and promoted apoptosis. The oncogenic role of SPC24 was also identified in breast cancer and lung cancer [34, 35]. Here, SPC24 was identified to be negatively regulated by hsa-miR-139-3p. This regulatory role between SPC24 and hsa-miR-139-3p was previously observed in bladder cancer [36]. CDKN2A is traditionally known as a tumor suppressor gene coding for two proteins, including the p16INK4a and p14arf [37]. CDKN2A is involved in cell cycle regulation. However, mounting data suggest that CDKN2A may play a dual role in multiple tumors. The p14arf that it codes plays an important role in invasion and metastasis and is associated with a poor prognosis [38]. Upregulation of p14arf has been identified in multiple hematological malignancies, aggressive types of B-cell lymphomas and bladder cancers. [39–41]. The oncogenic function of p14arf is associated with the autophagy regulation [38]. Downregulation of p14arf was shown to inhibit progression of lymphomas with the MYC mutation [42]. Particularly in PTC, p16INK4a and p14arf coded by CDKN2A were both identified to be upregulated in thyroid tumorigenesis [43]. Wild type p14arf has been observed to delocalize into the cytoplasm in aggressive PTC. Here, we further identified that CDKN2A was upregulated in PTC and was associated with shorter PFI, and hsa-miR-139-3p was the potential negative regulator of CDKN2A. The function of CDKN2A in PTC progression requires further study. RIMS2 was also identified to be upregulated in PTC and was potentially regulated by hsa-miR-139-5p. hsa-miR-139-5p was identified to be downregulated in the primary tumor and further in PTC metastasis [44]. hsa-miR-139-5p was also able to be sponged by circBACH2 and relieved the suppression of the target gene LMO4 in PTC [45]. RIMS2 was identified as a novel target of hsa-miR-139-5p in this study. RIMS2 is a Rab effector and scaffold protein associated with exocytosis [46]. It was recently reported to have a high mutation rate in melanoma and mantle cell lymphoma [47, 48]. In this study, RIMS2 was identified as a hub DE-mRNA in the ceRNA network and was associated with PFI of PTC. The function of RIMS2 in PTC requires further experimental validation.
In this study, downregulated CLCNKB, FBXO27, and FXYD6 were identified to be novel targets of hsa-miR-146b-3p in PTC. hsa-miR-146b-3p was previously reported to be upregulated in PTC and positively associated with central lymph node metastases [49]. hsa-miR-146b-3p may promote invasion and metastasis of PTC by targeting NF2 [50]. hsa-miR-146b-3p also targeted PAX8 in modulating the differentiated phenotype of PTC [51]. FXYD6 is known as a specific modulator of Na, K-ATPase, and is expressed in multiple epithelial cells of the inner ear [52]. In accordance with the current study, FXYD6 was previously identified to be downregulated in PTC and was associated with poor prognosis [53]. CLCNKB is a voltage-gated chloride channel participating in the regulation of cell volume, membrane potential stabilization, signal transduction, and transepithelial transport [54]. CLCNKB was previously reported to be downregulated in renal carcinoma [55]. Hypermethylation in deletions of CLCNKB in renal carcinoma further indicated its tumor suppressing role in cancer [56]. FBXO27 is a component for substrate-recognition in the SCF-type E3 ubiquitin ligase complex [57]. Its role in cancer is currently unknown. In this study, CLCNKB, FBXO27, and FXYD6 were identified to be negatively regulated by hsa-miR-146b-3p. Together with hsa_circ_0008354 and hsa_circ_0049271 formed an important part of the PFI related ceRNA subnetwork we established. The potential tumor suppressing role of CLCNKB, FBXO27, and FXYD6 in PTC requires further validation.
To the best of our knowledge, the ceRNA network we established and the prognostic DE-mRNA signature we proposed has not been reported previously. The nomogram incorporating the DE-mRNA signature and clinical parameters was robust in predicting PFI of PTC. DE-circRNAs, DE-miRNAs, and DE-mRNAs were potential therapeutic targets for prevention and treatment of recurrent PTC. We acknowledge that our study inevitably has some limitations. Firstly, sequencing data and follow-up data of PTCs in our study were based on TCGA-THCA dataset. Most patients were from North America. Therefore, caution should be exercised in extrapolating this conclusion to other populations. Secondly, the targeting relationship of ceRNA is based on bioinformatic speculation and requires further experimental validation. Finally, the biological function of certain DE-circRNAs, DE-miRNAs, and DE-mRNAs require investigation with experiments in PTC.