Comprehensive Analysis of the Expression of Key Genes Related to Hippo Signaling and Their Prognosis Impact in Ovarian Cancer

The Hippo signaling pathway, one of the most conserved in humans, controlling dimensions of organs and tumor growth, is frequently deregulated in several human malignancies, including ovarian cancer (OC). The alteration of Hippo signaling has been reported to contribute to ovarian carcinogenesis and progression. However, the prognostic roles of individual Hippo genes in OC patients remain elusive. Herein we investigated the expression level and prognostic value of key Hippo genes in OC using online databases, followed by a qRT-PCR validation step in an additional patient cohort. Using the GEPIA database, we observed an increased level for TP53 and reduced expression level for LATS1, LATS2, MST1, TAZ, and TEF in tumor tissue versus normal adjacent tissue. Moreover, LATS1, LATS2, TP53, TAZ, and TEF expression levels have prognostic significance correlated with progression-free survival. The qRT-PCR validation step was conducted in an OC patient cohort comprising 29 tumor tissues and 20 normal adjacent tissues, endorsing the expression level for LATS1, LATS2, and TP53, as well as for two of the miRNAs targeting the TP53 gene, revealing miR-25-3p upregulation and miR-181c-5p downregulation. These results display that there are critical prognostic value dysregulations of the Hippo genes in OC. Our data demonstrate the major role the conserved Hippo pathway presents in tumor control, underlying potential therapeutic strategies and controlling several steps modulated by miRNAs and their target genes that could limit ovarian cancer progression.


Introduction
Ovarian cancer is usually diagnosed in advanced stages due to its nonspecific symptomatology and is the deadliest gynecological malignancy. Over 70% of cases are diagnosed with FIGO stage III or IV disease, where the five-year survival rate is 25% [1][2][3][4][5]. Numerically speaking, in 2018, the worldwide incidence of ovarian cancer (OC) increased to 295,414 cases, and the mortality followed with not less than 184,799 deaths [6,7]. Therefore, OC is the seventh most frequent type of cancer and the eighth most common cause of cancer death in women, with a case-to-fatality ratio nearly three times higher than breast cancer [2,3,8].

Gene Expression Profiling Interactive Analysis of the Key Hippo Genes
We performed the analysis of gene expression profiling on OC samples using the GEPIA database (http://gepia.cancer.pku.cn/, accessed on 4 January 2021) and the UAL-CAN database (http://ualcan.path.uab.edu/, accessed on 4 January 2021). The functions of the UALCAN database for OC are divided into expression analysis and custom data analysis according to race, age, tumor grade, and TP53 mutation status. This database includes gene expression data and survival information from a total of 1435 OC patients. The KM plotter analyzes individual genes against clinical outcomes such as relapse-free survival (PFS) and overall survival (OS). PFS is a beneficial outcome considering that it is not affected by later-line therapies and can be measured earlier than OS [31].

Survival Outcome Analysis of OC Patients
We used an online database (http://kmplot.com/analysis/, accessed on 4 January 2021) to determine the significance of gene expression in PFS. Six Hippo signaling genes: LATS1 and LATS2 (both tumor suppressor genes), MST1 (mammalian Ste20-like kinases 1), tumor suppressor gene TP53, transcriptional coactivator with PDZ-binding motif, TAZ, and TEF from TEAD/TEF family of transcription factors were entered into the database (https://kmplot.com/analysis/index.php?p=service, accessed on 4 January 2021) to attain Kaplan-Meier survival plots in which the number-at-risk is specified under the main plot by selecting the JetSet best probe set [32]. We calculated the hazard ratio (HR) and 95% confidence intervals and log rank; we considered a p-value of <0.05 to be statistically significant.

miRNA and Gene Validation in OC Samples
A total of 29 histologically confirmed OC patients admitted at The Oncology Institute Prof. Dr. Ion Chiricuta Cluj-Napoca, Romania, during 2018-2020 were enrolled in this study after the approval of the ethical committee. All patients signed the informed consent. The age of patients ranged between 22-75 years. All patients were staged according to the American Joint Committee on Cancer (AJCC) guidelines. Immediately following surgical excision, all tissue samples were snap-frozen in liquid nitrogen for RNA isolation and stored at −80 • C. Patients' clinical data are presented in Table 1.

Transcriptional Levels of Main Representative Members of Hippo Signaling in Patients with OC
Several genes related to Hippo signaling were identified in mammalian cells; among them, we selected 12 for analysis of the expression level. We compared the transcriptional levels of the main representative members of Hippo signaling within OC using the GEPIA database. We observed a statistically significant increased level for TP53 and a reduced expression level for LATS1, LATS2, MST1, TAZ, and TEF ( Figure 1A). The expression levels heatmap generated using cBioPortal is presented in Figure 1B. We observed a higher mutation rate for TP53 (66%), STK3 (11%), YAP1 (10%), and a less frequent mutation rate for LATS1, LATS2, MST1, TAZ, and TEF ( Figure 1C). LATS1 and TAZ expression levels decreased with stage, and we observed no significant differences in subgroup analysis for tumor grade and TP53 mutational status ( Figure S1). PIA database. We observed a statistically significant increased level for TP53 and a reduced expression level for LATS1, LATS2, MST1, TAZ, and TEF ( Figure 1A). The expression levels heatmap generated using cBioPortal is presented in Figure 1B. We observed a higher mutation rate for TP53 (66%), STK3 (11%), YAP1 (10%), and a less frequent mutation rate for LATS1, LATS2, MST1, TAZ, and TEF ( Figure 1C). LATS1 and TAZ expression levels decreased with stage, and we observed no significant differences in subgroup analysis for tumor grade and TP53 mutational status ( Figure S1).

Association between Gene Expression and Progression-Free Survival (PFS) in OC
To determine whether Hippo signaling gene expression is associated with prognosis in OC, we analyzed PFS between the high-and low-gene expression groups from the KM plotter online database. The results indicated that except for MST1, expression levels for LATS1, LATS2, TP53, TAZ, and TEF had a significant impact on PFS (p < 0.05, Figure 2). High expression of MST1, LATS1, and TP53 was associated with a better prognostic, whereas low expression of LATS2, TAZ, and TEF was associated with a better prognostic in OC. ics 2021, 11, 344 7 of 16  Subgroup analysis on tumor stage for LATS1 was not statistically significant. Low expression of LATS2 and TEF in advanced stages (III + IV) was associated with a significantly better prognostic; however, this was not significant for early stage OC (I + II). High expression of TAZ and MST1 was associated with a significantly better prognostic in early stage OC (I + II) but not in advanced stage OC (III + IV). High expression of TP53 was associated with a significantly better prognostic in both early (stage I + II) and advanced (stage III + IV) OC ( Figure 3 and Table 2). High expression of TP53 wild-type was associated with a favorable prognostic, whereas high expression of TP53 mutated was associated with an unfavorable prognostic. Subgroup analysis on TP53 mutational status revealed that MST1, LATS1, and TAZ were not associated with PFS. Low expression of LATS2 and TEF were associated with a significantly longer PFS rate in both groups (TP53 mutant and TP53 wild-type, Figure 4 and Table 3).

miRNA-mRNA Interaction Network Analysis via miRNET
The miRNA-mRNA interaction network revealed a correlation among genes involved in Hippo signaling, uncovering the high complexity of biological mechanisms, with each gene being targeted by a number of important miRNAs. Figure 5 presents the extracted miRNA-mRNA network, emphasizing those miRNAs that target at least three genes from the selected Hippo genes. The core of the network is TP53, interconnected with one of the most powerful tumor suppressor miRNAs, namely miR-34a-5p (known in literature to be related with OC prognostic), along with miR-25-3p, let-7c-5p, miR-15a-5p, and miR-125a-5p.

miRNA-mRNA Interaction Network Analysis via miRNET
The miRNA-mRNA interaction network revealed a correlation among genes involved in Hippo signaling, uncovering the high complexity of biological mechanisms, with each gene being targeted by a number of important miRNAs. Figure 5 presents the extracted miRNA-mRNA network, emphasizing those miRNAs that target at least three genes from the selected Hippo genes. The core of the network is TP53, interconnected with one of the most powerful tumor suppressor miRNAs, namely miR-34a-5p (known in literature to be related with OC prognostic), along with miR-25-3p, let-7c-5p, miR-15a-5p, and miR-125a-5p.

Validation of Key miRNA Related to Hippo Signaling by qRT-PCR
For the validation step, we selected the TP53 targets: miR-25-3p and miR-181c-5p. Validation of miRNAs was conducted using 29 tissue samples collected from OC, as well as 20 samples of distant normal tissues. For normalization of the miRNA data, U6 and RNU48 were used as internal controls, based on the ∆∆Ct method. miR-25-3p levels were overexpressed, whereas miR-181c-5p was downregulated ( Figure 6A). Additionally, a ROC (receiver-operating characteristic) curve was generated to assess the sensitivity and specificity of these genes, the highest AUC (area under the curve) value being for miR-25b-3p (0.8405) ( Figure 6B).

Validation of Key miRNA Related to Hippo Signaling by qRT-PCR
For the validation step, we selected the TP53 targets: miR-25-3p and miR-181c-5p. Validation of miRNAs was conducted using 29 tissue samples collected from OC, as well as 20 samples of distant normal tissues. For normalization of the miRNA data, U6 and RNU48 were used as internal controls, based on the ΔΔCt method. miR-25-3p levels were overexpressed, whereas miR-181c-5p was downregulated ( Figure 6A). Additionally, a ROC (receiver-operating characteristic) curve was generated to assess the sensitivity and specificity of these genes, the highest AUC (area under the curve) value being for miR-25b-3p (0.8405) ( Figure 6B).

Validation of Key Genes Related to Hippo Signaling by qRT-PCR
Validation of selected genes was conducted on the same patient cohort used for the evaluation of miRNAs. In order to further validate the gene expression alteration revealed using the GEPIA online tool, we performed qRT-PCR for LATS1, LATS2, and TP53, with the B2M gene as the endogenous control for normalization of the qRT-PCR data. Gene expression analysis showed that TP53 is overexpressed, whereas LATS1 and LATS2 levels were significantly underexpressed in tumor tissues versus normal adjacent tissues ( Figure 7A). For each evaluated gene, we generated a ROC curve to assess the sensitivity and specificity of these genes, the highest AUC value being for LATS1 (0.6984), LATS2 (0.8833), and TP53 (0.7056) ( Figure 7B). These qRT-PCR results further validate our earlier gene and miRNA expression profiles from the GEPIA database.

Validation of Key Genes Related to Hippo Signaling by qRT-PCR
Validation of selected genes was conducted on the same patient cohort used for the evaluation of miRNAs. In order to further validate the gene expression alteration revealed using the GEPIA online tool, we performed qRT-PCR for LATS1, LATS2, and TP53, with the B2M gene as the endogenous control for normalization of the qRT-PCR data. Gene expression analysis showed that TP53 is overexpressed, whereas LATS1 and LATS2 levels were significantly underexpressed in tumor tissues versus normal adjacent tissues ( Figure 7A). For each evaluated gene, we generated a ROC curve to assess the sensitivity and specificity of these genes, the highest AUC value being for LATS1 (0.6984), LATS2 (0.8833), and TP53 (0.7056) ( Figure 7B). These qRT-PCR results further validate our earlier gene and miRNA expression profiles from the GEPIA database.

Discussion
Ovarian cancer still has the highest mortality among gynecological malignancies. Although novel strategies for OC treatment continue to emerge, the effectiveness of novel treatments remains suboptimal. The Hippo family members play critical roles in tumorigenesis and inflammatory responses and have been reported to have critical prognostic significance in many cancer types. The prognostic roles and functions of Hippo-related genes and miRNA expression in OC have not yet been studied. In the current study, we comprehensively explored the expression patterns, prognostic values (PFS), genetic alterations, and potential functions of different Hippo members based on a variety of large online databases. Dysregulation of the Hippo pathway could be an important factor in the poor prognosis of ovarian cancer [34], a fact also confirmed by the present study.
The Hippo genes were shown to promote characteristics such as self-renewal, metastatic potential, and chemoresistance [35], displaying a poor prognosis in OC [26], a fact also sustained by the present study. In this study, we focused on the understanding of the

Discussion
Ovarian cancer still has the highest mortality among gynecological malignancies. Although novel strategies for OC treatment continue to emerge, the effectiveness of novel treatments remains suboptimal. The Hippo family members play critical roles in tumorigenesis and inflammatory responses and have been reported to have critical prognostic significance in many cancer types. The prognostic roles and functions of Hippo-related genes and miRNA expression in OC have not yet been studied. In the current study, we comprehensively explored the expression patterns, prognostic values (PFS), genetic alterations, and potential functions of different Hippo members based on a variety of large online databases. Dysregulation of the Hippo pathway could be an important factor in the poor prognosis of ovarian cancer [34], a fact also confirmed by the present study.
The Hippo genes were shown to promote characteristics such as self-renewal, metastatic potential, and chemoresistance [35], displaying a poor prognosis in OC [26], a fact also sustained by the present study. In this study, we focused on the understanding of the molecular mechanisms of this family of genes and targeting miRNAs (Figure 8), paving the way for further studies.  MST1/2 activation results in the phosphorylation and activation of their direct substrates LATS1/2 [36]. LATS1 and LATS1 kinase modules are strongly conserved over evolution [37], and our study reveals the downregulation of these kinases. Understanding LATS-mediated tumor suppression will probably facilitate tools for early detection, prognosis, and treatment of OC [37]. LATS1 expression levels might be a valuable survival indicator in ovarian serous carcinoma [38]. Both LATS1 and LATS2 expression levels significantly correlated with recurrence and stage [38], confirmed for LATS2 also by the present data.
miRNA has an important role in the interplay between the Hippo pathway and other signaling pathways such as the MAPK, Notch, Wnt, and TGFβ pathways [19,37,39]. MST1/2 activation results in the phosphorylation and activation of their direct substrates LATS1/2 [36]. LATS1 and LATS1 kinase modules are strongly conserved over evolution [37], and our study reveals the downregulation of these kinases. Understanding LATS-mediated tumor suppression will probably facilitate tools for early detection, prognosis, and treatment of OC [37]. LATS1 expression levels might be a valuable survival indicator in ovarian serous carcinoma [38]. Both LATS1 and LATS2 expression levels significantly correlated with recurrence and stage [38], confirmed for LATS2 also by the present data.
miRNA has an important role in the interplay between the Hippo pathway and other signaling pathways such as the MAPK, Notch, Wnt, and TGFβ pathways [19,37,39]. We should not underestimate the mutational effect on key Hippo genes that may affect the crosstalk with other signaling pathways [24]. In the modulation of the crosstalk, among them, the TP53 gene plays an important role [40]. The TP53 gene is frequently mutated in human cancers [41], including OC, where it was observed a direct connection between the TP53 mutational status and evolutionary conservation [5]. This may be related to the fact that the functional status of TP53 proteins dictates the subcellular localization, protein stability, and transcriptional activity of the core component of the Hippo pathway, YAP1. YAP1 and TP53 pathways are critical protectors of genome integrity in response to DNA damage. Interestingly, the YAP1 gene was not altered in OC [40].
The regulatory networks of TP53 and Hippo pathways are connected in a highly context-specific manner [39], affecting key cellular processes related to cell proliferation, apoptosis, or invasion [42][43][44]. OC patients presenting different mutated TP53 proteins show different chemotherapy responses and survival outcomes [42,45]. TP53 interacts directly with the LATS2 promoter to induce LATS2 expression [37].
The main TP53 target is represented by miR-34a, a transcript that has a powerful tumor suppressor role related to cell proliferation, particularly in advanced stages [43,46]. It was demonstrated that miR-34 family members are frequently downregulated in OC [43,44,46], in both TP53 wild-type and mutant subgroups [43,46], being inversely expressed with TP53 as shown in our study. miR-181 family members represent highly controversial transcripts, being overexpressed in OC and related to epithelial to mesenchymal transformation and resistance [47], including miR-181c [48]. Downregulation of miR-181c was also confirmed in our study. Additionally, overexpression of miR-25-3p was observed. An increased expression level for miR-25-3p is related to an unfavorable prognostic, indicative that this transcript holds prognostic value in OC [49]. Other studies revealed that miR-25-3p overexpression promoted OC cell proliferation and motility by targeting LATS2 [50] or apoptosis by targeting BIM [51]. This emphasizes the oncogenic potential of miR-25-3p in OC [50]. miR-25-3p overexpression promotes tumor metastasis by activation of EMT [52], a key mechanism involved in invasion and metastasis [53]. Hence, miR-25-3p can be considered as a cancer-specific transcript in OC, having prognostic value along with miR-7, miR-16, miR-93, miR-182, miR-376a, and miR-429 [52].
Hippo signaling provides multiple interactions with coding and non-coding genes that are regulated by a variety of mechanisms. Knowledge of the interconnections between the Hippo signaling pathway and other biological processes that result in tumorigenesis might promote the development of new therapeutic strategies in OC [54].

Conclusions
In summary, by using online databases, we accessed the expression levels and determined the prognostic roles of key Hippo signaling genes that were validated on an additional patient cohort.
These results indicate that key Hippo signaling genes and direct miRNAs (miR-25-3p and miR-181c-5p) have important prognostic value in OC. A better understanding of the heterogeneity and complexity of OC is needed to develop tools for new therapeutic strategies and accurately assess prognosis.