Tumor Suppressive Role of the PRELP Gene in Ovarian Clear Cell Carcinoma

Ovarian clear cell carcinoma (OCCC) has a poor prognosis, and its therapeutic strategy has not been established. PRELP is a leucine-rich repeat protein in the extracellular matrix of connective tissues. Although PRELP anchors the basement membrane to the connective tissue and is absent in most epithelial cancers, much remains unknown regarding its function as a regulator of ligand-mediated signaling pathways. Here, we obtained sets of differentially expressed genes by PRELP expression using OCCC cell lines. We found that more than 1000 genes were significantly altered by PRELP expression, particularly affecting the expression of a group of genes involved in the PI3K-AKT signaling pathway. Furthermore, we revealed the loss of active histone marks on the loci of the PRELP gene in patients with OCCC and how its forced expression inhibited cell proliferation. These findings suggest that PRELP is not only a molecule anchored in connective tissues but is also a signaling molecule acting in a tumor-suppressive manner. It can serve as the basis for early detection and novel therapeutic approaches for OCCC toward precision medicine.


Introduction
Ovarian cancer is a leading cause of gynecological cancer mortality worldwide [1]. In the early stages, only a few subjective symptoms are present, and no screening method for ovarian cancer has been established. Even if patients with ovarian cancer initially respond favorably to first-line platinum-based chemotherapy, it is estimated that ≥80% of these patients will eventually relapse [2]. Although several target therapies have been applied for ovarian cancer, including poly (ADP-ribose) polymerase (PARP) inhibitors and vascular endothelial growth factor (VEGF) inhibitors, a selectable therapeutic approach for relapse or treatment resistance remains limited [3]. For the early detection of ovarian cancer and to develop novel therapeutic approaches toward precision medicine, further investigation of the molecular mechanisms underlying ovarian tumorigenesis and progression is required.
Epithelial ovarian cancers comprise five main histological subtypes: high-grade serous, endometrioid, clear cell, mucinous, and low-grade serous carcinomas. High-grade serous ovarian cancer (HGSOC) is the most common type of ovarian cancer. Accumulating pathological, epidemiological, and molecular evidence has revealed that fallopian tube secretory epithelial cells are the likely progenitors of HGSOC [4,5]. Except for TP53, BRCA1, and BRCA2, point mutations in oncogenes or tumor suppressor genes are relatively uncommon in HGSOC. Instead, several chromosomal structural variations have been reported in HGSOC [6]. Approximately 50% of HGSOC cases show defects in the homologous recombination (HR) DNA repair pathway, which is a key determinant of platinum sensitivity and provides a rationale for the use of PARP inhibitors [7,8]. In contrast, ovarian clear cell carcinoma (OCCC) accounts for approximately 10% of all ovarian cancers. Endometriosis is reported to be a risk factor for OCCC and coexists in more than 50% of cases [9,10]. Although OCCC is estimated to originate from ovarian endometriotic epithelial cells, the exact mechanism of its tumorigenesis has not been fully elucidated and awaits further study. OCCC shows a distinctive molecular pathogenetic pathway and intrinsic chemoresistance, which makes this entity unique compared to the other subtypes [11,12]. However, a therapeutic strategy specific to OCCC has not yet been established due to its relative rarity, and OCCC is usually treated in the same way as HGSOC. Patients with advanced or recurrent OCCC experience poorer clinical outcomes compared to those with HGSOC.
Small leucine-rich proteoglycans (SLRPs) are a family of 17 known proteoglycans, which are secreted proteins in the extracellular matrix (ECM). Regarding their function, SLRPs not only modify ECM organization but also serve as regulators of ligand-mediated signaling pathways [13][14][15][16]. Using a public dataset, we have previously shown that the PRELP gene, one of the secreted ECMs, is markedly downregulated in most epithelial cancers. Moreover, its ectopic expression in bladder cancer cell lines inhibited the transforming growth factor-beta and the epidermal growth factor pathways, induced cell-cell adhesion, and reversed epithelial-mesenchymal transition (EMT) [17].
In this study, we aimed to investigate the molecular mechanisms underlying OCCC tumorigenesis toward the early detection and novel therapeutic strategy for OCCC. We performed genetic and epigenetic analyses of the PRELP gene using OCCC cell lines and clinical samples. To our knowledge, this is the first study to focus on the PRELP gene in OCCC.

Clinical Materials
Both normal ovarian tissues and tumor tissues of the same patient were collected during primary debulking surgery. Fresh-frozen samples of OCCC (n = 7) were obtained from collections at the University of Tokyo Hospital and the National Cancer Center Hospital (NCCH). Fresh-frozen samples of HGSOC (n = 1), low-grade serous ovarian carcinoma (n = 1), and carcinosarcoma (n = 1) were obtained from collections at the NCCH (Supplementary Table S1). The clinical specimens were stored at -80 • C and embedded into the OCT compound, followed by frozen sectioning and RNA extraction. The histology was determined based on the pathologist's assessment according to the 2020 World Health Organization classification of female genital tumors.

Database Analysis
To compare the expression of tumor and normal tissues in the ovary, we reanalyzed using the same pipeline of RNA sequencing (RNA-seq) from The Cancer Genome Atlas (TCGA) and the Genotype-Tissue Expression project datasets on the UCSC Xena platform (http://xena.ucsc.edu/, accessed on 2 May 2022) [18]. RNA-seq, DNA copy number, gene mutation, and clinical data of 316 patients with HGSOC in the TCGA cohort were sourced from the cBioPortal for Cancer Genomics (http://www.cbioportal.org/, accessed on 2 May 2022) [19][20][21]. The individual data used to generate the graphs are listed in Supplementary  Tables S2-S4.

Plasmids
The procedure for plasmid construction was performed as reported previously [22,23]. The PRELP-myc cDNA was PCR amplified with the following primers-PRELP-myc_F_NheI: 5 -ACC CAA GCT GGC TAG CCA CCA TGA GGT CAC CCC TCT GCT G -3 , PRELP-myc_R_NotI: 5 -CAG CAC AGT GGC GGC CGC TCG AGT CTA GAC TAT AGT TCT AGA GGC TCG A-3 -and cloned into the modified Edit-R Inducible Lentiviral Plasmid at the NheI and NotI sites. All plasmids were verified by Sanger sequencing.

Cell Viability Assay
The general procedure for the cell viability assay was performed as reported previously [22,23]. Conditionally PRELP-expressing cells were plated in 96-well plates at the following concentrations: 1500 cells/well for RMG-I and 1000 cells/well for OVTOKO and SKOV3. At the indicated time, 10 µL of the Cell Counting Kit-8 (343-07623, Dojindo, Kumamoto, Japan) reagent was added to each well. Cell viability was measured by detecting the absorbance at 450 nm using Multiskan FC (Thermo-Fisher Scientific, Waltham, MA, USA).

Soft Agar Colony Formation Assay
We performed a soft agar colony formation assay, which is an anchorage-independent growth assay, as described previously [25]. Briefly, twice the usual amount of serum and antibiotics was added to this solution and was mixed with the 1.2% SeaPlaque agarose solution (50100, Lonza, Basel, Switzerland) to yield a final 1× growth medium with 0.6% agarose solution. A sterile solution of 0.7% SeaPlaque agarose was mixed with the aforementioned 2× growth medium to yield 1× growth medium with 0.35% agarose. A suspension of 5000 cells was added to a 1.5 mL aliquot of the 1× growth medium with 0.35% agarose solution, and the resultant mixture was plated out in a 6-well dish on top of the 1× growth medium with a 0.6% agarose layer, which was prepared in advance. The plates were incubated for approximately 2-3 weeks until visible colonies appeared. The colonies were stained with 1-mg/mL p-nitroblue tetrazolium chloride (144-01993, FUJIFILM Wako Pure Chemical Co., Osaka, Japan) for 18 h. Phase-contrast images were acquired using Celldiscoverer 7 (ZEISS, Oberkochen, Germany) and analyzed using ImageJ (National Institutes of Health, MD, USA) [26].

RNA-Seq
Conditionally PRELP-expressing cells derived from the OVTOKO and SKOV3 cell lines were treated with 1-µg/mL DOX for 72 h. The cells were washed with ice-cold PBS (-), and the total RNA was extracted using QIAzol Lysis Reagent and RNeasy Plus Mini Kit (73404, Qiagen, Venlo, Netherlands), according to the manufacturer's instructions. Oligo (dT)-conjugated beads were used to isolate mRNA. After mixing with the fragmentation buffer, mRNA was fragmented, and cDNA was synthesized using the mRNA fragments as templates. Short DNA fragments were purified and resolved with elution buffer for end reparation. Subsequently, the short DNA fragments were ligated with adapters. The suitable fragments were selected for the PCR amplification as templates. During the QC steps, Agilent 2100 Bioanalyzer was used to quantify the sample library. Finally, the DNA libraries were sequenced using BGISEQ-500.

PRELP Gene Expression and Genomic Aberrations
To investigate the association between the PRELP mRNA expression and genomic aberrations, we first analyzed the correlation between PRELP gene expression and somatic mutations and copy number aberrations (CNAs) using a comprehensive genomic dataset of patients with ovarian cancer (see Materials and Methods). We found that PRELP mRNA expression was significantly repressed in both primary and recurrent ovarian cancers compared with that in normal tissues (p < 0.0001) (Figure 1a and Supplementary Table S2). Deletions in the PRELP gene were found in 6.9% of the cases, and the PRELP gene was amplified in 53.3% of the cases. These alterations were moderately correlated with PRELP mRNA expression (Figure 1b and Supplementary Table S3). PRELP retained its wild-type form in 99.7% of the cases (Figure 1c and Supplementary Table S4). These results suggest that PRELP mRNA expression is suppressed in ovarian cancer, which is not dependent on genetic deletion and mutation but rather epigenetic mechanisms, such as DNA methylation or histone modification.
Next, we evaluated PRELP gene expression using clinical tissues that matched with those from the patients with ovarian cancer, including OCCC, HGSOC, low-grade serous carcinoma, and carcinosarcoma. PRELP mRNA expression was significantly repressed in ovarian cancer tissues by approximately 1/100th compared with that in normal tissues (p < 0.01) (Figure 2a). Additionally, we compared PRELP gene expression between an immortalized human ovarian endometriotic epithelial cell line (HMOsisEC10) and OCCC cell lines, including ES2, OVISE, OVTOKO, RMG-I, and SKOV3. All OCCC cell lines showed a significantly reduced mRNA expression in PRELP compared with HMOsisEC10 ( Figure 2b). These results indicate that the expression of PRELP was drastically altered in all the clinical specimens and cell lines we examined.

Cell Viability upon PRELP Expression
To investigate the effects of PRELP expression on cell viability in OCCC, we stably introduced the PRELP gene into the RMG-I, OVTOKO, and SKOV3 cell lines. The transduced PRELP gene contained an inducible promoter (Tet-On system), which allowed us to control the timing of the PRELP gene expression by adding DOX to the cultured medium. Western blotting confirmed that the PRELP protein was expressed following the addition of DOX in all the cell lines ( Figure 3, upper side). Under this condition, cell proliferation was significantly inhibited by PRELP expression (Figure 3, bottom side).
The soft agar colony formation assay with or without DOX treatment in PRELPinducible SKOV3 showed that the induction of PRELP reduced the colonies (Supplementary Figure S1), suggesting the requirement of PRELP suppression for anchorage-independent growth. Note that the other PRELP-inducible RMG-I and OVTOKO cell lines could not be evaluated because they did not form colonies with or without DOX treatment. (c) PRELP mRNA expression (z-scores relative to diploid samples; y-axis) is plotted against a wildtype (n = 315) or missense mutation (n = 1; x-axis).
Next, we evaluated PRELP gene expression using clinical tissues that matched with those from the patients with ovarian cancer, including OCCC, HGSOC, low-grade serous carcinoma, and carcinosarcoma. PRELP mRNA expression was significantly repressed in ovarian cancer tissues by approximately 1/100th compared with that in normal tissues (p < 0.01) (Figure 2a). Additionally, we compared PRELP gene expression between an immortalized human ovarian endometriotic epithelial cell line (HMOsisEC10) and OCCC cell lines, including ES2, OVISE, OVTOKO, RMG-I, and SKOV3. All OCCC cell lines showed a significantly reduced mRNA expression in PRELP compared with HMOsisEC10 ( Figure 2b). These results indicate that the expression of PRELP was drastically altered in all the clinical specimens and cell lines we examined.

Cell Viability upon PRELP Expression
To investigate the effects of PRELP expression on cell viability in OCCC, we stably introduced the PRELP gene into the RMG-I, OVTOKO, and SKOV3 cell lines. The transduced PRELP gene contained an inducible promoter (Tet-On system), which allowed us  Statistical analysis was performed using Student's t-test. ** p < 0.01; *** p < 0.001; **** p < 0.0001.

ChIP-Seq in Clinical Tissues of OCCC
The aforementioned results show that PRELP suppression in OCCC is likely mediated by epigenetic mechanisms, such as DNA methylation or histone modification. Given that there were no obvious CpG islands in the PRELP promoter region in the UCSC genome browser [29] (data not shown), we considered that an alteration of histone modifications might be involved in PRELP suppression. To this end, we performed a series of ChIP-seq on frozen normal ovarian tissues (n = 2) and OCCC tissues (n = 5) to investigate whether or not PRELP suppression was caused by the loss of active marks or the gain of suppressive marks. To investigate the molecular mechanisms underlying OCCC tumorigenesis, normal ovarian tissues with the same genetic background as OCCC tissues need to be analyzed. Therefore, normal ovarian tissues were collected from the opposite healthy ovaries of the same patients to avoid genetic and epigenetic alteration in those regions by inflammation and compression. The pathologist confirmed the histology of the normal ovarian tissues obtained. We found a loss in the active marks H3K4me3 and H3K27ac peaks around the upstream region of the PRELP gene in OCCC (Figure 4, blue and red). Interestingly, we also noticed the loss of CTCF in OCCC, which might reflect the promoter-proximal CTCF binding mediated by distal enhancer-dependent gene activation (Figure 4, green) [30]. In contrast, we did not observe a noticeable gain of the suppressive mark H3K27me3 in OCCC (Supplementary Figure S2). According to these results of our clinical ChIP-seq analyses using normal ovary and OCCC tissues, we can suggest that the loss of active histone markers in the promoter region of PRELP results in the loss of PRELP expression during tumorigenesis in OCCC. Statistical analysis was performed using Student's t-test. ** p < 0.01; *** p < 0.001; **** p < 0.0001.

ChIP-Seq in Clinical Tissues of OCCC
The aforementioned results show that PRELP suppression in OCCC is likely mediated by epigenetic mechanisms, such as DNA methylation or histone modification. Given that there were no obvious CpG islands in the PRELP promoter region in the UCSC genome browser [29], we considered that an alteration of histone modifications might be involved in PRELP suppression. To this end, we performed a series of ChIP-seq on frozen normal ovarian tissues (n = 2) and OCCC tissues (n = 5) to investigate whether or not PRELP suppression was caused by the loss of active marks or the gain of suppressive marks. To investigate the molecular mechanisms underlying OCCC tumorigenesis, normal ovarian tissues with the same genetic background as OCCC tissues need to be analyzed. Therefore, normal ovarian tissues were collected from the opposite healthy ovaries of the same patients to avoid genetic and epigenetic alteration in those regions by inflammation and compression. The pathologist confirmed the histology of the normal ovarian tissues obtained. We found a loss in the active marks H3K4me3 and H3K27ac peaks around the upstream region of the PRELP gene in OCCC (Figure 4, blue and red). Interestingly, we also noticed the loss of CTCF in OCCC, which might reflect the promoter-proximal CTCF binding mediated by distal enhancer-dependent gene activation (Figure 4, green) [30]. In contrast, we did not observe a noticeable gain of the suppressive mark H3K27me3 in OCCC (Supplementary Figure S2). According to these results of our clinical ChIP-seq analyses using normal ovary and OCCC tissues, we can suggest that the loss of active histone markers in the promoter region of PRELP results in the loss of PRELP expression during tumorigenesis in OCCC.

Gene Sets Altered by the Overexpression of PRELP in OCCC Cell Lines
If PRELP affects the ligand-mediated pathways, fluctuations in several relevant genes should be observed. To this end, we performed RNA-seq on conditionally PRELPexpressing cells derived from the OVTOKO and SKOV3 cell lines. Indeed, we found that more than 1000 genes were significantly altered by PRELP expression (Figure 5a and Supplementary Tables S5 and S6), indicating that PRELP is a regulator of various genes. Furthermore, many genes associated with the PI3K-AKT signaling pathway were altered by PRELP expression and ranked at the top in the KEGG pathway enrichment analysis (Figure 5b,c and Supplementary Table S9). A similar trend was observed in SKOV3 cells that overexpressed PRELP (Supplementary Figure S3 and Supplementary Tables S7, S8 and S10). These results suggest that PRELP regulates the PI3K-AKT signaling pathway, which might have inhibitory effects on cell proliferation.

Gene Sets Altered by the Overexpression of PRELP in OCCC Cell Lines
If PRELP affects the ligand-mediated pathways, fluctuations in several relevant genes should be observed. To this end, we performed RNA-seq on conditionally PRELPexpressing cells derived from the OVTOKO and SKOV3 cell lines. Indeed, we found that more than 1000 genes were significantly altered by PRELP expression (Figure 5a and Supplementary Tables S5 and S6), indicating that PRELP is a regulator of various genes. Furthermore, many genes associated with the PI3K-AKT signaling pathway were altered by PRELP expression and ranked at the top in the KEGG pathway enrichment analysis (Figure 5b,c  and Supplementary Table S9). A similar trend was observed in SKOV3 cells that overexpressed PRELP (Supplementary Figure S3 and Supplementary Tables S7, S8 and S10). These results suggest that PRELP regulates the PI3K-AKT signaling pathway, which might have inhibitory effects on cell proliferation.

Discussion
In this study, we provided a novel perspective on the molecular features of the pathogenesis of OCCC. We first showed that the PRELP gene expression is repressed using public datasets of clinical tissues from patients with ovarian cancer compared with normal tissues. Furthermore, all OCCC cell lines we tested exhibited a significant downregulation of the PRELP expression compared with ovarian endometriotic epithelial cells and the putative progenitor of OCCC, which reflects that PRELP repression may be involved in early OCCC tumorigenesis. Thus, studying PRELP as a biomarker for the early detection of OCCC is intriguing.
How is PRELP expression suppressed during OCCC tumorigenesis? Deletions and somatic mutations in the PRELP gene are relatively rare (Figure 1), making it unlikely that genetic alterations are the primary cause of the suppression of PRELP expression. Indeed, ChIP-seq analyses revealed that using normal ovary and OCCC tissues and transcriptionally active marks, such as H3K4me3 and H3K27ac, were diminished at the PRELP gene promoter region in OCCC compared with those in normal ovary tissues, indicating that the PRELP gene is repressed, at least in part, by an epigenetic mechanism.
Epigenetic mechanisms have been shown to contribute profoundly to OCCC tumorigenesis. For example, mutations in the AT-rich interaction domain 1A (ARID1A) and phosphatidylinositol-4,5-bisphosphate 3-kinase (PIK3) catalytic subunit alpha (PIK3CA) have been frequently detected in OCCC cases [31][32][33]. ARID1A encodes the BAF250 protein as a subunit of the switch/sucrose non-fermentable chromatin-remodeling complex that facilitates the epigenetic regulation of the chromatin structure and transcription factors [34,35]. BAF250 protein loss can be detected in atypical endometriosis but not in distant endometriotic lesions [34]. ARID1A and PIK3CA mutations may contribute to OCCC tumorigenesis through sustained interleukin (IL)-6 production [35]. Moreover, Yano et al. showed that histone deacetylases, HDAC6 and HDAC7, are more strongly expressed in OCCC than in other subtypes of ovarian cancer [36]. How these epigenetic alterations and PRELP suppression contribute to OCCC tumorigenesis are currently unclear; however, one scenario is that the interplay between ARID1A and HDACs could significantly contribute to OCCC tumorigenesis [37]. Therefore, it is possible that the dysregulation of HDACs directly suppresses PRELP gene expression.
Using conditionally, PRELP-expressing cells derived from OCCC cell lines in which the ARID1A protein expression was repressed [38], we revealed that PRELP significantly altered the gene expression of more than 1000 genes. Particularly, it regulates a set of genes related to the PI3K-AKT signaling pathway ( Figure 5). How PRELP can alter these pathways is unclear, even though it is a secreted ECM protein. However, recent proteomic studies have suggested that PRELP interacts with two growth factor receptors: insulinlike growth factor I receptor and low-affinity nerve growth factor receptor (p75NTR) [39], suggesting that PRELP functions as a ligand. Although many questions remain unanswered, elucidating the subcellular localization of PRELP may provide clues to the answers to these questions.
Although the aforementioned results provide the unexpected finding that PRELP is not only a connective tissue-anchored molecule but also a tumor-suppressive signaling molecule in OCCC, many unresolved issues remain. For example, experiments, such as the cell-cycle analysis, are needed to elucidate more detailed molecular mechanisms. In future studies, we plan to comprehensively analyze the protein expression level of a group of genes involved in the PI3K-AKT signaling pathway with or without PRELP using immunoblotting. Moreover, the anticancer mechanisms of the PRELP protein in OCCC have not been elucidated. The xenograft mouse model in the OCCC cell line is one of the next studies we wish to conduct for further validation of the function of PRELP. Lastly, due to the limited number of clinical samples used in this study, conducting clinical studies is impossible, such as validating a biomarker. Most of the cohort databases of ovarian cancer correspond to HGSOC patients, and there is no big dataset of OCCC patients. Several PI3K/AKT/mammalian target rapamycin (mTOR) pathway inhibitors have been used in clinical trials for OCCC [40]. To validate the clinical function of PRELP, we have to perform a prospective study to establish a dataset of OCCC patients for genetic and epigenetic analyses.
Although we analyzed pathways at the induction of PRELP expression, specifically in OCCC, there have been some studies reporting PRELP in other cancer types. HDAC inhibitors show anti-cancer effects by partially regulating the function of PRELP in bladder cancer. Furthermore, the acetylation of lysine residue 5 of histone H2B in the PRELP gene promoter region is a marker for the restoration of PRELP expression [23]. The overexpression of PRELP correlates with better patient survival and inhibits both cell proliferation and migration in hepatocellular carcinoma [41]. A peptide corresponding to the N-terminal heparin-binding domain of PRELP inhibits osteoclastogenesis in breast cancer metastases [42]. Thus, PRELP may regulate a set of genes related to the PI3K-AKT signaling pathway regardless of carcinoma type to some extent. However, no single data type, such as somatic mutations or gene expression, can capture the complexity of all the factors relevant to understanding phenomena such as cancer. Recently, integrated genetic and epigenetic dataset analyses using machine learning algorithms have emerged [43]. Our future directions may focus on epigenetics combined with RNA-seq and ChIP-seq, which may become part of the data to predict clinical outcomes. Elucidating the roles of PRELP in cancer may help understand the clinical significance of the classified genes identified by machine learning algorithms.

Conclusions
This study demonstrated that the PRELP gene was broadly repressed in OCCC, which was mediated by epigenetic mechanisms involving the loss of active histone marks. Furthermore, induced PRELP expression in OCCC cell lines inhibited cell proliferation, presumably via the PI3K-AKT signaling pathway. These findings suggest that PRELP is not only a molecule anchored in the connective tissues but is also a signaling molecule that acts in a tumor-suppressive manner in OCCC. Further investigation of the molecular mechanisms underlying OCCC tumorigenesis can serve as the basis for early detection and novel therapeutic strategies for OCCC toward precision medicines.
Supplementary Materials: The following supporting information can be downloaded at https: //www.mdpi.com/article/10.3390/jpm12121999/s1. Figure S1: Soft agar colony formation assay. Figure S2: H3K27me3 marks in the PRELP gene locus. Figure S3: MA plots, KEGG pathway enrichment analysis, and Volcano plots in SKOV3. Table S1: Clinical sample characteristics. Table S2: PRELP mRNA expression and ovarian tissue types. Table S3: PRELP mRNA expression and CNAs. Table S4: PRELP mRNA expression and somatic mutations. Table S5: Upregulated genes upon PRELP overexpression in OVTOKO. Table S6: Downregulated genes upon PRELP overexpression in OVTOKO. Table S7: Upregulated genes upon PRELP overexpression in SKOV3. Table S8: Downregulated genes upon PRELP overexpression in SKOV3. Table S9: The KEGG pathway in PRELP-overexpressed OVTOKO. Table S10: The KEGG pathway in PRELP-overexpressed SKOV3. Table S11: Information on certified cell lines. Raw Data S1: Uncropped immunoblot data are shown in Figure 3.  Institutional Review Board Statement: All methods were performed according to the Ethical Guidelines for Medical and Health Research Involving Human Subjects. For the use of specimens in this research, informed consent was obtained from all patients, and the study was approved by the Institutional Review Boards of the NCCH (approval ID: 2016-496) and the Human Genome, Gene Analysis Research Ethics Committee of University of Tokyo Hospital, Tokyo, Japan (approval ID: G0683-18).
Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.
Data Availability Statement: Raw data for RNA-seq have been deposited in the DNA Data Bank of Japan with DRA014790. Raw data from ChIP-seq using human-derived tissue samples will not be disclosed at this time for ethical reasons. All codes in this study are available upon request.