The Inhibitor of DNA Binding Family Regulates the Prognosis of Ovarian Cancer

Background: The inhibitor of DNA binding or differentiation (ID) protein family contributes to the carcinogenesis and progression of various cancers. However, its mechanistic role in tumor initiation and progression of ovarian cancer (OC) has remained unclear. Methods: We used the Oncomine, GEPIA, Kaplan-Meier plotter, cBioPortal, SurvExpress, PROGgene V2 server, TIMERdatabase, and FunRich to evaluate the expression and predictive prognostic value of individual IDs members’ mRNA in patients with OC. Results: Our results revealed that the mRNA transcripts of all ID family members were markedly downregulated in OC compared to normal tissue. Aberrant expression of ID 1/3/4correlated with cancer aggressiveness and clinical in OC patients. The prognostic value of ID members was also explored within the subtypes, pathological stages, clinical stages, and TP53 mutational status. The group with a low risk IDs showed a relatively good overall survival (OS) in comparison to the high-risk group. In contrast, the expression level of IDs was signicantly associated with the levels of in ﬁ ltrating B cells and macrophages. Finally, enrichment analysis showed that ID co-expressed genes were involved in ID-, c-MYC, TNF-, and Wnt signaling pathways. Conclusion: These results indicate that ID1/3/4 may be exploited as promising prognostic biomarkers and therapeutic targets in OC patients. 21 cancer types based on the Gene Expression Omnibus (GEO), the European Bioinformatics Institute (EBI) and TCGA. In the present study, PROGgeneV2 was used to validate the relationship between the expression of IDs and prognostic outcomes in OC. The HRs and the corresponding 95% CIs were used to assess the prognostic eciency of IDs on OC. HRs and 95% CIs for OS were directly obtained from PROGgeneV2. Different data sets were meta-synthesized using STATE 14.0 software (State Corporation, College Station, TX, USA). The heterogeneity among studies was estimated with the v2-based Q-test and Higgins’ I 2 statistic. A p-value < 0.05 for the Q-test or I2 > 30% indicated signicant heterogeneity, and the random-effects model was used; otherwise, the xed-effects model was used. OC patients. The group with low risk ID signature presented a markedly good OS relative to the high-risk group. In contrast, the expression levels of IDs were signicantly correlated with the levels of inltrating B cells and macrophages. Finally, enrichment analysis showed that ID co-expressed genes were involved in ID-, c-MYC-, TNF-, and Wnt signaling pathways. These results indicate that ID1/3/4 may be exploited as promising prognostic biomarkers and therapeutic targets in OC patients.

Recently, it was reported that several ID proteins are aberrantly expressed in OC samples in comparison with normal tissues, and the level of ID is closely related to poor differentiation, advanced stage, enhanced malignant potential, and worse clinical pathological features of OC (12)(13)(14) (15). Elevated expression levels of ID1 and ID3 were found to be a strong predicator of shorter survival in OC (14,15). These reports show that ID could be a promoter of OC progression and tumorigenesis. More importantly, animal experiments showed that partial loss function or knockdown of ID1 and ID3 decreased proliferation, anchorage-independent growth, increased apoptosis, and reduced survival in various human cancer cells (16,17). There is reason to believe that IDs may be novel therapeutic genes and potentially versatile therapeutic targets for OC. Regrettably, the distinct roles of the individual ID proteins in OC are not fully known. In our study, we comprehensively analyzed the relationships between the four ID subtypes and OC based on several large databases such as cBioPortal, Kaplan-Meier plotter (KM plotter), Gene Expression Pro ling Interactive Analysis (GEPIA), SurvExpress, TIMER, and FunRich, to determine the expression patterns, genetic alterations, immune in ltrations, molecular function, and prognostic signature of ID proteins in OC.

Ethics statement
All protocols and experiments in this study conformed to the Declaration of Helsinki and were approved by the Academic Committee of the First People's Hospital of Yichang. The data used in this study were obtained from published reports.

Oncomine analysis
The Oncomine (www.oncomine.org) contains massive cancer-related microarray datasets of DNA or RNA sequences.
It is frequently used in genome-wide expression studies (18). Herein, it was employed to reveal the transcriptional pro le of ID family members in patient specimens from different cancer types and healthy controls. Moreover, the Student's t-test was used to compare the expression levels between the two groups. Signi cant expressions were those with fold-change = 1.5; P-value = 0.001. GEPIA dataset analysis GEPIA (http://gepia.cancer-pku.cn) provides a platform for analyzing RNA sequencing dataset covering on 9,736 tumors and 8,587 normal specimens in the Genotype-tissue Expression dataset (GTEx) and the Cancer Genome Atlas (TCGA) projects. GEPIA is highly interactive and enables users to adjust various functions, such as dimensionality reduction analysis, correlation analysis, survival analysis, tumor/normal differential expression analysis, similar gene detection, and pro ling plotting based on the pathological stage or type of cancer (19).
TCGA and cBioPortal analysis TCGA (http://cancergenome.nih.gov/) comprises pathological and sequencing datasets for 30 types of cancers (20). On the other hand, cBioPortal is a freely-accessible cancer genomic web platform (http://www.cbioportal.org/), which may be used for integrative analysis and multi-functional visualization for clinical pro les and data of cancer genomics (21). In this study, the dataset "Ovarian Serous Cystadenocarcinoma (TCGA, Provisional)" was used. The frequency of ID family gene alterations, copy number variance, and mRNA expression z-scores (RNA Seq V2 RSEM) were assessed using cBioPortal in line with the guidelines provided on the cBioPortal webpage.

Functional enrichment analysis
FunRich is an open access gene interaction network analysis tool and enables comprehensive functional annotation of various biological processes (22). In the present study, processes and pathways enrichment analyses of ID family proteins were performed using FunRich to identify genes associated with ID expression. In addition, the Gene Ontology (GO) terms for cellular component (CC), molecular function (MF), and biological process (BP) categories, as well as the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways enrichment analysis of the 50 closely related genes signi cantly associated with ID expression were performed through FunRich.
Kaplan-Meier plotter analysis KM plotter (www.kmplot.com) platform to estimate the prognostic performance of ID mRNA expression. This database constitutes the survival information and gene expression datasets of 1,816 patients with OC (23). We then subcategorized patients into low and high expression groups on the basis of the median expression value, and assessed the progression-free survival (PFS), and overall survival (OS) of OC patients employed the Kaplan-Meier survival plot, with log-rank p-value, a hazard ratio (HR) with 95% con dence intervals (CI). Further sub-classi cation was performed; TP53 mutation status, histological subtypes and pathological subtypes, foe subgroup analysis.

SurvExpress analysis
The SurvExpress (http://bioinformatica.mty.itesm.mx/SurvExpress), a web-based resource, is commonly used for risk assessment and survival multivariate analysis using gene expression data (24). This database was employed herein for risk assessment and survival analysis to identify key ID gene signatures in OC. A prognostic index established was utilized to group patient samples into high or low-risk groups in reference to the median value of the index by employing the maximized risk algorithm. The log-rank p-value, log-rank test with HR with 95% CI was utilized for statistical analysis of the equality of survival curves.

PROGgeneV2 analysis
The PROGgeneV2 is a web based tool available at www. compbio.iupui.edu/proggene. It contains data from 134 cohorts from 21 cancer types based on the Gene Expression Omnibus (GEO), the European Bioinformatics Institute (EBI) and TCGA. In the present study, PROGgeneV2 was used to validate the relationship between the expression of IDs and prognostic outcomes in OC. The HRs and the corresponding 95% CIs were used to assess the prognostic e ciency of IDs on OC. HRs and 95% CIs for OS were directly obtained from PROGgeneV2. Different data sets were meta-synthesized using STATE 14.0 software (State Corporation, College Station, TX, USA). The heterogeneity among studies was estimated with the v2-based Q-test and Higgins' I 2 statistic. A p-value < 0.05 for the Q-test or I2 > 30% indicated signi cant heterogeneity, and the random-effects model was used; otherwise, the xed-effects model was used.

TIMER analysis
The extent of immune in ltration among various types is often estimated using the TIMER (https://cistrome.shinyapps.io/timer platform) (25). This tool was therefore utilized to assess the correlation of IDs expression with six immune in ltrates (DCs, macrophages, neutrophils, CD4 + T cells, B cells, and CD8 + T cells) in OC using Spearman's correlation analysis. On the basis of this correlation module, we established scatter plots between a pair of user-de ned genes for each type of cancer, and the expression of each gene was presented with log2 RSEM.

Results
Mapping the mRNA expression pro le to IDs in OC samples The human genome contains genes encoding four ID family members. The Oncomine database was employed to compare the expression pattern of ID genes in cancer samples and normal tissue samples, and the results are presented in Fig. 1 and Table 1. Notably, cancer samples displayed the lowest expression of ID1 mRNA among the three datasets (26,27). It was reported that ID1 is decreased in ovarian serous adenocarcinoma when compared to normal samples in the Yoshihara datasets (27) and Hendrix (26). In the TCGA dataset also, ID1 also downregulated in ovarian serous cystadenocarcinoma in comparison to normal samples, with a fold change of -2.901. Similarly, the transcriptional level of ID2 was signi cantly downregulated in patients with OC in the three datasets (26-28). In the Hendrix (26) and Yoshihara dataset (27), ID2 was signi cantly downregulated with fold changes of -1.523 and − 6.008, respectively, in ovarian serous adenocarcinoma while it was downregulated in ovarian serous surface papillary carcinoma with a fold change of -11.999 in the Welsh dataset (28). A similar trend was also found for ID3. The ID3 mRNA expression was markedly lower in multiple types of ovarian cancer compared to that in normal tissues in the Welsh (28), Yoshihara (27), and Hendrix (26) datasets. The mRNA level of ID4 was signi cantly lower in ovarian carcinoma and ovarian serous cystadenocarcinoma than that in the normal samples in the Bonome (29) and TCGA datasets. We also investigated the mRNA expression levels of IDs in OC compared to that in normal tissue using the GEPIA dataset. As shown in Fig. 2A -D, ID1-ID3 mRNA transcripts were relatively low in OC tissues compared to normal ovarian tissues, however, only the levels of ID2 and ID3 showed marked differences between OC and normal tissues.
In addition, analysis of the GEPIA dataset indicated that the mRNA level of ID was not related to the different stages of OC ( Fig. 2E -H).
Genetic alteration rate of IDs and co-expressed genes in OC patient samples Gene variations of IDs in OC were examined on the cBioPortal. As shown in Figure. 3, a total of 594 patients and 606 samples from the TCGA provisional dataset of ovarian serous carcinoma were analyzed. The genetic alteration rates of ID1, ID2, ID3, and ID4 were 10, 6, 3, and 15%, respectively (Fig. 3A). We further explored the impact of IDs genetic alterations on the prognosis of OC. Notably, no signi cant association between the prognosis of OC with ID gene alteration or without alteration based on the TCGA provisional dataset (p values, 0.404 and 0.759, Fig. 3B, C).

Functional enrichment analysis of IDs and co-expressed genes in patients with OC
We subsequently compiled a list of the expressed IDs and the 50 closest co-expressed genes predicted by analyzing GO and KEGG in Funrich. As shown in Fig. 4, the BP of IDs and their co-expressed genes were dramatically concentrated in processes related to regulation of nucleic acid metabolism, nucleobase, nucleoside nucleotide and regulation of gene expression, peptidolysis, proteolysis, organogenesis, and regulation of immune response (Fig. 4A). The MF of these genes were mainly transcription regulator activity, protease inhibitor activity, protein binding, antigen binding, and protein serine/threonine phosphatase activity (Fig. 4B). For the CC, the genes were correlated with nuclear membrane, protein kinase, CK2 complex, nucleus, junctional sarcoplasmic reticulum membrane, and connexon complex (Fig. 4C). Additionally, the KEGG analysis revealed signi cant enrichment of genes in ID-, c-MYC-, TNF-, and Wnt signaling pathways (Fig. 4D).

Prognostic value of IDs in OC samples
We subsequently assessed the correlation of individual IDs with different clinical pathology parameters such as pathological grade, clinical stage, and TP53 mutation status of OC. The results presented in Fig. 5 and Table 2 indicates that high mRNA expression of ID1 and ID3 predicted worse PFS and OS in serous OC patients. In contrast, the mRNA level of ID4 predicted favorable OS. In endometrioid OC, the expression of ID1 and ID3 showed a strong correlation with good PFS. As shown in Table 3, in OC patients with pathological grade III, elevated ID1 and ID3 correlated with poor PFS and OS. In patients with pathological grade II, ID3 correlated with poor PFS and OS. In addition, upregulated ID1 was linked to poor OS and upregulated ID2 correlated with poor OS in pathological grade I patients. As shown in Table 4, in clinical stage III patients, increased expression of ID1 and ID3 was associated with worse OS, and elevated ID2 was associated with poor PFS. In clinical stage IV OC patients, elevated ID1 was associated with worse OS and high ID3 expression was related to poor OS and PFS in this subgroup. As shown in Table 5, Moreover, high expression of ID1 and ID3 was related was associated with worse PFS and OS in OC patients carrying mutated TP53, and high ID1, ID2, and ID3 expression was associated with worse OS in OC patients with wildtype TP53.

Prognostic value of ID signatures in patients with OC
The SurvExpress platform was used to establish a prognostic index based on ID expression. A total of 1,609 patients from three ovarian cancer datasets with large sample sizes were analyzed using the SurvExpress platform. High/low risk groups were categorized by prognostic risk algorithm in each dataset. The survival analysis and Kaplan-Meier plotter between high risk (red) and low risk (green) groups and the heat map of the expression of IDs in each dataset are shown in Fig. 6. The results showed that the expression of each ID member was distributed between high and low risk groups. More importantly, the low risk group displayed a signi cantly good OS in comparison with the high risk group in the ovarian Meta-base: 6 cohorts with 22

Validate the prognostic value of IDs in patients with OC in different data sets
We applied PROGgeneV2 to validate the he prognostic value of IDs in patients with OC in different data sets. The results showed that seventeen data sets with 2,585 subjects reported the data of relationship between ID1 and OS in patients with OC. The pooled result showed that increased ID1 expression was signi cantly correlated with worse OS (HR: 1.08, 95% CI: 1.01-1.14, p = 0.017), with signi cant heterogeneity (I 2 : 36.5%, Ph = 0.065) (Fig. 7A). In addition, the same 17 data sets reported data on the association between ID2 and OS in patients with OC. Meta-analysis of these 17 sets showed that there was no signi cant correlation between the expression of ID2 and the OS of OC  . 7B). Simultaneously, there were 18 data sets containing 2,663 OC patients the prognostic value of ID3 and ID4 in OS. As shown in Fig. 7C, the elevated ID3 expression was signi cantly associated with unfavorable OS (HR: 1.10, 95% CI: 1.04-1.16, p < 0.001) and no signi cant heterogeneity was observed (I 2 : 11.5%, Ph = 0.317). At last, the results presented in Fig. 7D indicates that increased ID4 expression was positively correlated with better OS (HR: 0.90, 95% CI: 0.84-0.97, p < 0.001), with extreme heterogeneity (I 2 : 55.0%, Ph = 0.003).

Immune in ltration analysis of IDs in patients with OC
We explored the correlation between ID expression and immune in ltration levels in OC using correlation modules in TIMER. As shown in Fig. 8

Discussion
Key among the genes that encode the helix-loop-helix (HLH) family of transcription factors is the ID, abundant in stem and progenitor cells (1). To date, it is known that ID proteins are encoded by four ID genes in the ID family in vertebrates: ID1-4, all of which encode the corresponding four ID family members (1,2). These genes are located in different chromosomes and show inconsistent expression pro les and functions (30). Emerging evidence suggests that ID proteins play vital roles in tumorigenesis in several types of malignancies mediated by their ability to regulate cell-cycle, cell differentiation, epithelial-mesenchymal transition, chemoresistance, and immunomodulation (1)(2)(3)5). Yet, the speci c roles of the four ID members in OC are obscure. This study evaluated the prognostic value and expression of ID family genes by investigating various large databases. Our study presents the rst silico and bioinformatics analysis of the ID family.
ID1 is the most widely characterized component of the HLH transcription factor family (31). Studies show that the molecular functions of ID-1 included induction of cell proliferation, increasing DNA synthesis, and interaction with various oncogenes (32). Aberrant expression of the ID1 protein has not only been detected in multiple types of human cancers, but is also correlated with tumor stages and clinical outcome (33,34). Furthermore, ectopic expression of ID1 in human cancer cells increases serum-independent cell growth, enhances primary tumor G1/S phase formation and metastatic potential, and protects tumor cells against apoptosis. Conversely, inhibition or inhibition of ID1 in human cancer cells has been shown to suppress cell proliferation, induce cellular senescence, induce G2/M cell-cycle arrest, reduce tumor colony formation or multiplicity, and increase lifespan (35,36). In OC, Schindl et. al. found that ID1 expression correlates with the malignant potential of OC and is correlated with aggressive behavior, differentiation of tumor cells, and clinical prognosis (12). Several studies have found that increased ID1 may promote cancer cell proliferation and enhance endothelial progenitor cell angiogenesis through regulation or facilitation of EGFR and TGFβ1 expression, and activation NF-κB/MMP-2 and PI3K/Akt signaling pathways in OC cells (6, [37][38][39]. In addition, the study by Li ZD et al demonstrated that apigenin can suppress the expression of ID1, resulting in inhibition of tumorigenesis in human OC A2780 cells (40). Thus, ID1 represents a promising therapeutic target for OC. In our study, the Oncomine and GEPIA datasets indicated that the expression of ID1 was suppressed in human OC.
The Kaplan-Meier plotter and PROGgeneV2 analysis revealed a high mRNA expression of ID1, and this was correlated with poor OS in all OC patients. These data re ect the heterogeneity of ID1 expression in mRNA and protein levels, and point to the oncogenic function of ID1.
ID2 belongs to the HLH transcription factor family, which promotes proliferation and invasive growth in multiple solid cancers, e.g., hepatocellular cancer, breast cancer, thyroid cancer, pancreatic cancer, and OC (1,41). Like ID1, several studies have shown that ID2 promotes the proliferation of human cancer cells by inhibiting cell apoptosis, enhancing cancer stemness of pre-malignant cells, or mediating m6A modi cations (42)(43)(44). Conversely, reduced ID2 expression increases apoptosis, reduces cell proliferation, and decreases tumor initiation in human cancer cells (16, 45). However, currently there are very few reports on ID2 and OC development in the literature. An earlier study showed the ID2 gene as a candidate for inherited predisposition to breast and ovarian cancer in Jewish women (46). Moreover, the study by Meng et. al. reported that elevated ID2 expression in ERα-positive epithelial tumor cells promoted the invasiveness of cells via a non-canonical pathway independent forming dimers with basic helix-loophelix factors (47). In this study, ID2 mRNA expression was found to be lower in OC samples than in normal ones, and elevated ID2 expression was strongly related to poor PFS in all patients with OC. Prognostic analysis in patients with OC in different data sets, however, overall effect did not show any signi cant correlation between ID2 expression and OS. The oncogenic effects, predictive value, and potential molecular targets of ID2 in OC remain to be investigated further.
ID3, associated with HLH transcription factors, has been recognized as a key regulator of cell development, senescence, differentiation, proliferation, stemness, and migration (1, 48). It has been con rmed that ID3 and ID1 can compensate for each other and have similar biological functions (5). Previous studies have demonstrated that aberrant expression of ID3 is associated with advanced tumor stage and poor prognosis in many types of human cancers. In animal experiments, although Id1+/-Id3-/-or Id1-/-Id3+/-mice grow to adulthood, they are unsuitable to implanted tumor xenografts because these mice lack the capacity to recruit hematopoietic precursors and endothelial (48,49). Furthermore, double knockdown of ID1 and ID3 has been shown to decrease proliferation and anchorageindependent growth, increase apoptosis, and reduce survival in various human cancer cells (16, 17). More importantly, ID3 knockdown improved the survival duration of animals in a seeding model of medulloblastoma. It also compromised the progression of leptomeningeal seeding and the growth of primary tumors (50). Elsewhere, it was recognized that BMP4 signaling is active in ovarian cancer cells where it maintains ID3 expression. This was con rmed by the use of BMP4 blocker Noggin, which decreased endogenous ID3 expression (51). In this study, we also demonstrated that the expression of ID3 in OC tissues was lower than that in normal tissues, and ID3 overexpression was associated with reduced OS and PFS in OC patients. Because ID3 undergoes epigenetic inhibition in multiple cancers, it is therefore thought to be a tumor suppressor.
In comparison with the other ID proteins, ID4 possesses a polyproline domain at its C terminus and a polyalanine domain at its N terminus. Although it harbors the HLH domain, ID4 does not display similar expression and function with D protein (5). Numerous studies have shown that the phenotypic changes and molecular pathways regulated by ID4 are, in general, not like those regulated by ID1, ID2, and ID3. Interestingly, ID4 seems to function as a tumor suppressor in multiple cancers and as a tumor promoter in a small subset of cancers (30,52,53). The proposed tumor-suppressing effects of ID4 draw on observations that ID4 undergoes epigenetic silencing in several solid cancers such as esophageal, gastric, pancreatic, colorectal, cholangiocarcinoma and lung cancer. However, ID4 has been reported to be elevated in some small cancers, such as OC, prompting researchers to re-classify it as a tumor promoter (4,52,54,55). Mice de cient in ID4 develop some types of cancers in their lifetime, and the lack of ID4 results in follicular dysplasia and increased atretic follicles due to decreased estrogen biosynthesis (30). It is worth noting that a recent study showed that administration to mice harboring an ovarian tumor with an ID4-speci c tumorpenetrating nanocomplex was capable of suppressing the growth of established tumors and signi cantly improved survival (7). In the current study, unlike ID1, ID2, and ID3, the expression of ID4 was higher in OC tissues than that in normal tissues, and high ID4 expression was signi cantly correlated with better OS in OC patients, thereby indicating its tumor promoter role in OC.
We also attempted to examine the mechanisms and roles of members of the ID family, we also used the cBioPortal database to explore the mutations in the ID family. The results showed that the genetic alteration rate of the ID family members varied from 3% to 15% for individual genes based on the TCGA provisional dataset, however, there was no signi cant association between the prognosis of OC with ID gene alteration or without alteration. We then constructed a network of ID family members and 50 of the closest co-expressed genes. The results of the functional analysis indicated that these genes were mainly enriched in tumor-related pathways, including the ID-, c-MYC-, TNF-, and Wnt signaling pathways. In addition, two major highlights of this study were the ID signature and immune in ltration analysis. In the ID signature analysis, the prognostic values of ID signature in patients with OC were evaluated in three datasets based on the SurvExpress platform. The method overcomes the problem single gene with the expression optimal cutoff for prognostic analysis cannot fully re ect the optimal differentiation of survival bene ts and performance of potential biomarkers. In the immune in ltration analysis, we explored the correlation between IDs expression with six immune in ltration levels in OC via correlation modules in TIMER. Our results showed that ID expression showed a strong correlation with in ltrating levels of B cells and macrophages, which further con rmed that the biological role of ID may be associated with immune regulation. However, the underlying molecular mechanisms and regulation steps remain largely unexplored.

Conclusions
This study reveals that IDs exhibited diverse expression pro les between OC and normal samples. Aberrant expression of ID1/3/4 was correlated with cancer aggressiveness and prognosis in OC patients. The group with low risk ID signature presented a markedly good OS relative to the high-risk group. In contrast, the expression levels of IDs were signi cantly correlated with the levels of in ltrating B cells and macrophages. Finally, enrichment analysis showed that ID co-expressed genes were involved in ID-, c-MYC-, TNF-, and Wnt signaling pathways. These results indicate that ID1/3/4 may be exploited as promising prognostic biomarkers and therapeutic targets in OC patients. Abbreviations: