Genomic landscape of endometrial stromal sarcoma of uterus.

Although recurrent gene fusions such as JAZF1-JJAZ1 are considered driver events for endometrial stromal sarcoma (ESS) development, other genomic alterations remain largely unknown. In this study, we performed whole-exome sequencing, transcriptome sequencing and copy number profiling for five ESSs (three low-grade ESS (LG-ESS) and two undifferentiated uterine sarcomas (UUSs)). All three LG-ESSs exhibited either one of JAZF1-SUZ12, JAZF1-PHF1 and MEAF6-PHF1 fusions, whereas the two UUSs did not. All ESSs except one LG-ESS exhibited copy number alterations (CNAs), many of which encompassed cancer-related genes. In UUSs, five CNAs encompassing cancer-related genes (EZR, CDH1, RB1, TP53 and PRKAR1A) accompanied their expressional changes, suggesting that they might stimulate UUS development. We found 81 non-silent mutations (35 from LG-ESSs and 46 from UUSs) that included 15 putative cancer genes catalogued in cancer-related databases, including PPARG and IRF4 mutations. However, they were non-recurrent and did not include any well-known mutations, indicating that point mutations may not be a major driver for ESS development. Our data show that gene fusions and CNAs are the principal drivers for LG-ESS and USS, respectively, but both may require additional genomic alterations including point mutations. These differences may explain the different biologic behaviors between LG-ESS and UUS. Our findings suggest that ESS development requires point mutations and CNAs as well as the gene fusions.

To further characterize ESS genomes and extend the knowledge on genetic mechanisms for ESS development, the following questions were investigated in this study: (i) whether LG-ESS, HG-ESS and UUS genomes have driver mutations or CNAs; (ii) whether there are any previously uncharacterized gene fusions in either LG-ESS or HG-ESS or UUS; and (iii) whether there are any differences in expression profiling among LG-ESS, HG-ESS and UUS. For these, we analyzed genomes of LG-ESSs and UUSs (Table 1) by whole-exome sequencing (WES), wholetranscriptome sequencing and microarray-comparative genomic hybridization (a-CGH) in this study.
We analyzed gene expression profiles of the five ESSs using the whole-transcription data, which were subsequently compared to their CNA profiles. A normal (non-malignant) endometrial tissue was used to compare the gene expression level with ESS tissues. Five cancer-related genes in the UUSs exhibited a positive correlation between CNAs and gene expressions (down-regulation of genes with CNA loss: CDH1, RB1, TP53 and PRKAR1A; up-regulation of a gene with CNA gain: EZR) ( Figure 2). Compared to the normal endometrium, the ESSs showed increase of EZR expression (5.63 fold-change) and decreases of RB1 (0.15 fold-change), CDH1 (0.15 and 0.13 fold-changes in case 4 and case 5, respectively), TP53 (0.04 fold-change) and PRKAR1A (0.11 fold-change) expressions. All of the copy number losses except RB1 were heterozygous deletion while RB1 loss was homozygous deletion (Figure 2).

Whole-exome sequencing profiles
The five ESSs (three LG-ESSs and two UUSs) were also analyzed by WES. Mean coverage of the sequencing depth was 116X (range: 88-143X) for normal samples and 142X (range: 128-155X) for tumor samples. A total of 81 non-silent point mutations and indels (median: 10, range: 6-36) were identified in the five ESSs (Supplementary  Table S3). Both distribution of sequence composition and relative fraction of mutation spectra were not significantly different among the five ESSs (P > 0.05). Missense mutation with C:G > T:A transition was the most common type ( Figures 3A and 3B).
To address whether the mutations found in our study could be causally implicated in ESS development, we queried the cancer Gene Census and found four mutations (SRGAP3, EBF1, IRF4 and PPARG). Also, by analyzing the CHASM to distinguish driver and passenger mutations, we identified ten putative cancerrelated mutations (THSD7A, TBC1D14, TTK, SLC38A1, CTNNA2, SUPT6H, DCC, MDGA2, CNNM1 and MYO19) [18]. Finally, in the pan-cancer database, we detected SRGAP3 and ZFP36L2 as putative cancer-related genes [19]. Together, we detected 15 putative cancer-related genes with somatic mutations that could be causally implicated in ESS development ( Figure 3C). Of them, SRGAP3 was co-detected in the cancer Gene Census and the pan-cancer database ( Figure 3C). Sanger validation of the mutations including four putative cancer-related genes (SRGAP3, PPARG, DCC and ZFP36L2) is illustrated in Supplementary Figure S2. Of the mutated genes identified, three genes, which included a cancer-related gene MYO19, were found to have expressional changes in the same cases by the transcriptome analysis (Supplementary Figure S3).

DISCUSSION
The aim of this study was twofold. First, we attempted to disclose any somatic genetic alterations other than the gene fusions in ESSs. Second, we for the first time attempted to disclose genetic features of UUSs that by definition do not harbor any of the ESS-specific fusions. We found that the ESSs analyzed in this study harbored 6-36 non-silent somatic mutations per genome, but they did not include well-known mutations (e.g., TP53, KRAS and PIK3CA). As for the CNAs, all except one ESS harbored CNAs, many of which encompassed well-known driver genes. Importantly, some cancerrelated genes in UUSs showed a close correlation between CNAs and gene expression changes, strongly suggesting their implications on ESS tumorigenesis. The UUSs showed bigger median values of CNAs, non-silent mutations and putative cancer-related genes than the LG-ESSs (Table 4), but the differences were not significant, probably due to the small number of the ESSs analyzed. Together, this study identified that ESSs harbored not only ESS-specific fusions but also somatic mutations and CNAs encompassing driver genes. Our findings suggest a possibility that gene fusions alone may not fully develop ESSs as identified in other tumors [20].
We identified five genes in UUSs that shared CNA and gene expression changes (down-regulation and CNA loss: PRKAR1A, CDH1, RB1 and TP53, up-regulation and CNA gain: EZR). CDH1 is a tumor suppressor gene that encodes E-cadherin and is known to be associated with various malignancies [21,22]. Loss of CDH1 results in dysfunction of cell-cell adhesion, allowing for abnormal cell-cell interaction such as epithelial-mesenchymal transition (EMT) [23]. In addition, CDH1 behaves as a negative regulator in Wnt signaling [24]. ESSs are related to up-regulation of Wnt signaling, for example, exhibiting a down-regulation of SFRP4, a negative regulator in Wnt signaling [22,25]. In our study, we also found that both SFRP4 and CDH1 expressions were decreased in the ESSs, further suggesting the importance of Wnt signaling in ESSs. Recent research advances yielded a number of US Food and Drug Administration (FDA)-approved drugs that may change Wnt signaling [26]. Our study may provide further rationale for performing future studies that may explore the use of Wnt modulators in ESSs. TP53 is the most common tumor suppressor gene that is frequently inactivated in most cancers [27]. Tumor suppressor genes generally follow "two-hit hypothesis", being bi-allelically inactivated by point mutation, deletion and promoter hypermethylation [28]. In spite of the earlier report on   LG-ESS: low-grade ESS Recurrent fusion transcripts are frequently detected in ESSs [4,5,[9][10][11][12][13]. JAZF1-SUZ12, JAZF1-PHF1 and MEAF6-PHF are gene-fusions that have been detected most frequently in LG-ESSs [5,12,13] and YWHAE-FAM22 gene fusion is a key factor in defining HG-ESS [31]. By definition, UUSs are high-grade ESSs without known gene fusions [2]. As expected, we were able to find JAZF1-SUZ12, JAZF1-PHF1 and MEAF6-PHF fusions in each of the three LG-ESSs. For the two ESSs without any known ESS-specific fusions, we were not able to detect relative fractions of sequence spectra of five endometrial stromal sarcomas are shown. C. Fifteen putative cancer-related genes with somatic mutations. Blue box denotes the genes that overlap the cancer Gene Census genes [17], yellow box denotes the genes that were detected by the CHASM analysis [18] and red box denotes the gene catalogued in the pan-cancer driver database [19]. Of them, SRGAP3 was detected in both of the cancer Gene Census and the pan-cancer driver database (hatched). any novel fusions, identifying that they were UUSs per se. Our data suggest that the UUS with a driver fusion might, if any, be very rare. We found PPARG and IRF4 mutations in the ESSs. PPARG encodes peroxisome proliferator-activated receptor gamma (PPAR-γ/PPARG) [32] that has tumor suppressor functions in many endocrine organs including breast, prostate and pituitary gland [33][34][35]. In uterus, PPARG activation inhibits growth and survival of human endometriosis cells by suppressing estrogen biosynthesis [36]. IRF4 encodes a transcription factor in interferon regulatory factor family. A chromosomal translocation involving IRF4 and the IgH locus, t(6;14)(p25;q32) is considered a cause of multiple myeloma [37]. IRF4 is required for endometrial decidualization [38]. Together, these data suggest a possible rationale that PPARG and IRF4 mutations might be involved in ESS development.

Table 4: Summary of comparison data between LG-ESS and UUS genomes LG-ESS (n = 3) UUS (n = 2) P value
In this study, ESSs harbored at least one genetic alteration (somatic mutations or CNAs or gene fusions) that may stimulate ESS tumorigenesis ( Figure 4). Also, we observed that all of the ESSs carried either somatic mutations or CNAs-harboring driver genes, albeit not recurrent. It suggests that non-recurrent alterations may cooperate together for the ESS turmorigenesis.
Our data are based on the analysis of five ESS genome pairs (three LG-ESSs and two UUSs). The small sample size is due to the rarity of ESS [1,2]. Further investigation with larger sample size across diverse ethnic groups would reveal additional information, e.g., discovery of potential additional driver mutations in ESSs and additional novel fusions in LG-ESSs and HG-ESSs. In addition, a larger cohort would provide genomic features of metastatic ESSs compared to the primary ESSs.
In summary, our data for the first time attempted the integrative analyses of whole-exome, whole-transcriptome and a-CGH in ESS genomes. Previous studies on ESS genomes mainly focused on gene fusions. However, our data indicate that fusions are not the only genetic alteration occurred during ESS development. Somatic mutations, CNAs as well as gene fusions alone or together might contribute to ESS development. Our data also suggest that CNAs may be a major genomic alteration for UUS development while gene fusions are the major genomic alteration for LG-ESS. Our findings may provide a useful resource for understanding this heterogeneous disease and identifying genomic clues for differential diagnosis and therapy options for ESS.

Endometrial stromal sarcoma tissues
Normal and tumor tissues from five ESS patients were obtained from the tissue banks of Korean Gynecologic Cancer Bank (Seoul, Korea), Guro Hospital of Korea University (Seoul, Korea) and Busan National University Hospital (Busan, Korea). We also used normal endometrial tissue from a healthy woman. All of the six samples were from Koreans. Approval for this study was obtained from the institutional review board at the Catholic University of Korea, College of Medicine. Clinicopathologic features of the five ESS patients are summarized in Table 1. Initially, frozen tissues from the tissue banks were cut, stained with the hematoxylin/eosin and examined under microscope by a pathologist, who identified areas rich in ESS tumor cells in the frozen tissues. In order to procure matched normal tissues from each ESS patient, we used peripheral blood lymphocytes or normal tissues that were confirmed to be free of tumor cells by microscopic examination. Purities of the tumor cells were approximately 70%. For genomic DNA and RNA extraction from the frozen tissues, we used the DNeasy Blood & Tissue Kit (Qiagen, Hilden, Germany) and mirVanamiRNA Isolation Kit (Invitrogen, Carlsbad, CA, USA), respectively.

Whole-exome sequencing
Using genomic DNA from five ESSs and matched normal cells, we performed exome-capture using the Agilent SureSelect Human All Exome 50Mb kit (Agilent Technologies) according to the manufacturer's instructions. DNA libraries were constructed according to the protocol provided by the manufacturer and whole exome-sequencing (WES) was performed using an Illumina HiSeq2000 platform to generate 101bp paired-end reads. Burrows-Wheeler aligner was used to align the sequencing reads onto the human reference genome (hg19). The aligned sequencing reads were evaluated by using Qaulimap [39]. Supplementary Table S4 shows the information of sequencing alignments (e.g., the number of reads and sequencing coverage). Processing and the management of the sequencing data were performed as described elsewhere [20]. In brief, somatic variants were identified by using MuTect [40] and SomaticIndelDetector [41] for point mutations and indels. ANNOVAR package was used to select somatic variants located in coding sequences and to predict their functional consequences, such as silent or non-silent variants [42]. Then, we used the CHASM analysis program with 'uterus' option for cancer tissue type (FDR < 0.3) in order to identify the putative cancer-related mutations [18]. In order to validate noticeable somatic mutations, genomic DNA from tumor areas and matched normal tissues from each case were amplified by PCR and sequence analyses were performed.

Transcriptome sequencing for gene fusion and expression profiling analyses
The mRNA of five ESSs and normal proliferative endometrium of a woman was converted into a 272bp to 289bp-sized cDNA library using TruSeq RNA Library Preparation Kit (Illumina). Whole-transcriptome sequencing was performed using an Illumina HiSeq2000 platform. Sequencing reads were mapped onto the human reference genome (GRCh37, hg19). Gene fusions were identified by searching for the spanning reads and split reads by using the deFuse program [43]. Transcriptome sequencing data were analyzed using TopHat (http:// ccb.jhu.edu/software/tophat/index.shtml) for alignment, Cufflinks for assembly [44] and a known set of reference transcripts from Ensembl v. 65 (http://www.ensembl.org) for expression estimation. The expression levels were quantified as fragments per kilobase of exon per million fragments mapped (FPKM).

Reverse transcription-polymerase chain reaction and sequencing
RT was performed using oligo-(dT) primer and SuperScript III reverse transcriptase (Invitrogen). PCR was performed with Pfu DNA polymerase (Promega) according to the manufacturer's instruction. The thermal cycling included one cycle at 95°C for 2 min followed by 35 cycles of 95°C at 0.5 min, 55-61°C for 0.5 min, 72°C for 1 min and a final extension of 72°C for 5 min. Details of the primer pairs and corresponding genes are available in Supplementary Table S5. PCR products were visualized on 2% agarose gel and subsequently analyzed by direct DNA sequencing.

DNA copy number profiling
DNA copy number profiling was performed using the Agilent Sure Print G3 Human CGH Microarray 180K. The genomic DNA of five ESS tissues and matched normal tissues was hybridized onto the array according to the manufacturer's instructions. Background correction and normalization for array images was performed using Agilent Feature Extraction Software v10.7.3.1. The RankSegmentation statistical algorithm in NEXUS software v7.5 (Biodiscovery Inc., El Segundo, CA) was used to define the CNAs of each sample; a log2 ratio larger than 0.3 was identified as gain and lower than −0.3 as loss.
specimen and performed clinical review. All authors have read and approved the manuscript for publication.