Genetic alterations in endometrial cancer by targeted next-generation sequencing

Experimental and Molecular Pathology 100 (2016) 8–12

Contents lists available at ScienceDirect

Experimental and Molecular Pathology journal homepage: www.elsevier.com/locate/yexmp

Genetic alterations in endometrial cancer by targeted next-generation sequencing Ya-Sian Chang a,b, Hsien-Da Huang c,d, Kun-Tu Yeh e, Jan-Gowth Chang a,b,f,⁎ a

Epigenome Research Center, China Medical University Hospital, Taichung, Taiwan Department of Laboratory Medicine, China Medical University Hospital, Taichung, Taiwan Department of Biological Science and Technology, National Chiao Tung University, Hsinchu, Taiwan d Institute of Bioinformatics and Systems Biology, National Chiao Tung University, Hsinchu, Taiwan e Department of Pathology, Changhua Christian Hospital, Changhua, Taiwan f School of Medicine, China Medical University, Taichung, Taiwan b c

a r t i c l e

i n f o

Article history: Received 9 November 2015 Accepted 24 November 2015 Available online 25 November 2015 Keywords: Endometrial cancer Next-generation sequencing PTEN mutation IL-7 signaling pathway

a b s t r a c t Many genetic factors play important roles in the development of endometrial cancer. The aim of this study was to investigate genetic alterations in the Taiwanese population with endometrial cancer. DNA was extracted from 10 cases of fresh-frozen endometrial cancer tissue. The exomes of cancer-related genes were captured using the NimbleGen Comprehensive Cancer Panel (578 cancer-related genes) and sequenced using the Illumina Genomic Sequencing Platform. Our results revealed 120 variants in 99 genes, 21 of which were included in the Oncomine Cancer Research Panel used in the National Cancer Institute Match Trial. The 21 genes comprised 8 tumor suppressor candidates (ATM, MSH2, PIK3R1, PTCH1, PTEN, TET2, TP53, and TSC1) and 13 oncogene candidates (ALK, BCL9, CTNNB1, ERBB2, FGFR2, FLT3, HNF1A, KIT, MTOR, PDGFRA, PPP2R1A, PTPN11, and SF3B1). We identified a high frequency of mutations in PTEN (50%) and genes involved in the endometrial cancer-related molecular pathway, which involves the IL-7 signaling pathway (PIK3R1, n = 1; AKT2, n = 1; FOXO1, n = 1). We report the mutational landscape of endometrial cancer in the Taiwanese population. We believe that this study will shed new light on fundamental aspects for understanding the molecular pathogenesis of endometrial cancer and may aid in the development of new targeted therapies. © 2015 Elsevier Inc. All rights reserved.

1. Introduction Endometrial cancer is the fourth most common gynecological malignancy in Europe and the United States. More than 280,000 women are diagnosed each year worldwide and 74,000 women die from endometrial cancer annually. It is also the second most common gynecological cancer in Taiwan. Endometrial cancer is broadly divided into two groups: endometrioid carcinoma and uterine serous carcinoma (Bokhman, 1983). Close to 80– 90% of endometrial cancers are endometrioid carcinomas and 2–10% are uterine serous carcinomas (Dedes et al., 2011). Although uterine serous carcinoma is in the minority, it is much more aggressive than endometrioid carcinoma and has a poor outcome (del Carmen et al., 2012; Hamilton et al., 2006). Therefore, understanding the genomic alterations associated with this disease may provide opportunities for genome-guided clinical trials and drug development. Next-generation sequencing (NGS) technology has emerged as a powerful tool to investigate the genetic etiology of diseases. NGS has ⁎ Corresponding author at: Epigenome Research Center, China Medical University Hospital, 2 Yuh-Der Road, Taichung 404, Taiwan. E-mail address: [email protected] (J.-G. Chang).

http://dx.doi.org/10.1016/j.yexmp.2015.11.026 0014-4800/© 2015 Elsevier Inc. All rights reserved.

been applied in a variety of ways, such as whole genome sequencing, targeted capture, high-throughput RNA sequencing, and chromatin immunoprecipitation followed by sequencing. The ability to generate a huge quantity of sequencing data also presents the challenge of deciding which variants to validate. Several organizations, such as the College of American Pathologists, Centers for Disease Control and Prevention, and U.S. Food and Drug Administration, have published guidelines for clinical NGS analysis. Over the past two years (2012–2014), a number of studies have reported the genomic landscapes of endometrial cancer. The first study to characterize the genomic landscape of uterine serous carcinoma was described by Kuhn et al. (2012). Among the most frequently mutated genes in 10 uterine serous carcinomas were TP53, PIK3CA, FBXW7, and PPP2R1A. Subsequently, Gallo et al. found frequent somatic mutations not only in FBXW7, but also in ubiquitin ligase complex and chromatin remodeling genes (Le Gallo et al., 2012). Zhao et al. then decoded the exomes of a uterine serous carcinoma cohort five times larger than those reported earlier. They identified a significantly increased burden of mutation in 14 genes, including the previously reported and wellknown cancer genes TP53, PIK3CA, PPP2R1A, KRAS, PTEN, FBXW7, and CDKN1A (Zhao et al., 2013). The Cancer Genome Atlas (TCGA) project revealed 14 pathogenic driver genes of uterine serous carcinoma

Y.-S. Chang et al. / Experimental and Molecular Pathology 100 (2016) 8–12

(TP53, PIK3CA, FBXW7, PPP2R1A, CHD4, CSMD3, SLC9A11, PTEN, COL11A1, PRPF18, SPOP, CDH19, HIST1H2AM, and CELP) (Kandoth et al., 2013). The first whole exome sequencing study of endometrioid carcinoma investigated 13 cases. Ten tumor suppressor genes (ARID1A, INHBA, KMO, TTLL5, GRM8, IGFBP3, AKTIP, PHKA2, TRPS1, and WNT11) and two oncogenes (ERBB3 and RPS6KC1) were identified as potential candidate driver genes (Liang et al., 2012). In addition, the frequent occurrence of mutations in PTEN (64%), PIK3CA (59%), ARID1A (55%), CTNNB1 (32%), MLL2 (32%), FBXW7 (27%), RNF43 (27%), APC (23%), FGFR2 (18%), and EGFR (14%) among endometrioid carcinomas was confirmed by Kinde et al. based on the exome sequencing results of 22 cases (Kinde et al., 2013). The aim of the current study was to identify genetic alterations in endometrial cancer in the Taiwanese population. We performed deep sequencing (N500 ×) to detect the mutational status in 578 cancerrelated genes (NimbleGen Comprehensive Cancer Panel) using freshfrozen tissues from 10 Taiwanese patients with endometrial cancer. 2. Materials and methods 2.1. Sample preparation and DNA extraction The specimens consisted of 10 fresh-frozen endometrial cancer tissues that were submitted for targeted sequencing. DNA was isolated using proteinase K and a QIAamp® DNA Micro Extraction Kit (QIAGEN) according to the manufacturer's protocol. This study was approved by the Institutional Review Board (KMUH-IRB-970488). 2.2. Library preparation and amplification, targeted capture, and Illumina-based sequencing Genomic DNA (1 μg) was fragmented using a Covaris S2 Focusedultrasonicator (Covaris, Woburn, MA), and quality control (QC) was performed using an Agilent Bioanalyzer 2100 (Agilent Technologies) to ensure a fragment size range from 200 to 400 bp. Fragmentation was followed by end repair, A-tailing and sequencing adapter ligation using an Illumina TruSeq DNA Sample Preparation Kit. The adapterligated DNA was amplified via selective, limited-cycle polymerase chain reaction (PCR) for a total of eight cycles. The prepared library (1 μg) was hybridized for 64 to 72 h to the NimbleGen SeqCap EZ Designs-Comprehensive Cancer library (Roche Diagnostics). The hybridized product was amplified for 18 PCR cycles using Roche postcapture primers. QC was performed on the amplified product using an Agilent Bioanalyzer DNA 1000 Kit to ensure that the final library fragment size ranged from 150 to 400 bp; the product was quantified using KAPA SYBR® Fast ABI Prism® 2 × qPCR Master Mix (KAPA Biosystem) to ensure a sufficient yield for sequencing. For paired-end 100 bp × 2 sequencing on the Illumina HiSeq 2000 instrument, captured libraries were denatured and loaded onto an Illumina cBot instrument at 9.5 pM for cluster generation according to the manufacturer's instructions. Ten libraries were sequenced per HiSeq lane. A 1% PhiX control DNA (Illumina) was added into libraries as the internal control. On each flow cell, one lane contained PhiX DNA as a control. 2.3. Bioinformatic analysis To filter poor-quality reads, FASTX-Toolkit (http://hannonlab.cshl. edu/fastx_toolkit) was employed to process the raw read data files. There are two steps for sequence quality processing. The command was “fastq_quality_filter –Q33 –q 30 –p 70”; “–q 30” indicates that the minimum quality score to keep is 30 and “–p 70” indicates that the minimum percent of bases must have “–q” quality greater than or equal to 70%. Sequences were retained if both forward and reverse sequencing reads passed the first step. Bowtie2, an efficient sequence alignment tool (Langmead et al., 2009), was used to align the retained reads with the human genome

9

(Grch38.p2). According to the sequence alignment, reads having only one chromosomal location were retained for further analysis. The Genome Analysis Toolkit (GATK) (McKenna et al., 2010), a widely used genetic variants discovery tool, was employed to identify genetic variants according to the sequence alignment results. Several databases and tools that provide information on genetic variants were used to annotate identified genetic variants. dbSNP (b144) (Sherry et al., 2001) is a database that collects and provides information on genetic variants within different species. ClinVar (Landrum et al., 2014) is a database that collects significant clinical information on genetic variants from patients. Cosmic (v73) (Forbes et al., 2015) is a database that collects somatic mutation information in human cancers. Polyphen2 (Adzhubei et al., 2010) is a software that predicts the level of influence of non-synonymous substitution in the human proteins. Five criteria were used to identify the genetic variants of cancer driver genes: 1) a read coverage greater than 500, 2) non-synonymous genetic variants, 3) the occurrence of genetic variants was not over 50%, 4) the genetic variants were considered or predicted as pathogenic or possibly/probably functionally impaired, and 5) the global minor allele frequency of the genetic variants was less than 0.02. 2.4. Sanger sequencing validation Primers for Sanger sequencing validation were designed using Primer3 software. The PCR primers used are described in Supplemental 1. PCR amplifications were performed using Pro Taq Plus DNA Polymerase (Protech Technology Enterprise, Taiwan) following the manufacturer's instructions. Sanger sequencing was performed on an ABI Prism 3130 Genetic Analyzer. 3. Results 3.1. Pathway analysis of mutated genes in endometrial cancer We sequenced each sample on the Illumina HiSeq Platform to an average sequencing depth of 535.79. We generated a mean of 16 M raw reads per sample, of which 98.45% to 99.71% were aligned to the human reference genome (Grch38, Table 1). Table 2 shows an overview of our approach to identify variants. After data filtering, we identified 120 variants in 99 genes, including 107 missense variants, 7 nonsense variants and 6 frame shift variants (Supplemental 2). Functional annotation of the 99 mutated genes was performed by use of the Annotation, Visualization, and Integrated Discovery (DAVID) database. Eighteen of the 99 mutated genes were found in the Kyoto Encyclopedia of Genes and Genomics (KEGG) cancer pathways (hsa05200), including CREBBP, CTNNB1, FGFR2, FLT3, FOXO1, MTOR, MSH2, MSH6, PTCH1, PTEN, PIK3R1, PDGFRA, PML, KIT, TPR, TP53, AKT2, and ERBB2 (p = 2.6 × 10−9, false discovery rate = 2.7 × 10−6,

Table 1 Alignment and coverage statistics for 10 endometrial cancer patients. Patient ID

Total reads

Reads mapped to genome

Covered ≧ 500× (%)

Average target coverage

F114T F123 F132 F134 F146 F147T F150T F152T F92T 03-3812T Average

17,197,961 20,545,225 14,534,452 15,572,857 14,826,104 11,099,160 16,221,035 16,256,250 17,218,187 17,730,300 16,120,153

17,079,751 20,459,077 14,426,666 15,507,377 14,693,849 10,925,387 16,154,454 16,180,893 17,150,449 17,676,534 16,027,138

59.2% 68.8% 46.6% 44.7% 48.9% 23.4% 46.8% 46.6% 52.6% 51.1% 48.9%

569.37 691.13 479.77 523.69 491.33 354.19 531.05 534.26 585.51 597.55 535.79

10

Y.-S. Chang et al. / Experimental and Molecular Pathology 100 (2016) 8–12

Table 2 Overview of our approach to identifying cancer driver genes in endometrial cancer. Type of prioritization filter All variants Non-synonymous variants The occurrence of variants b 50% ClinVar or Polyphen2 predicted the variants to be pathogenic or possibly/probably functionally impaired Global minor allele frequency b 0.02 in dbSNPv144

Remaining variants (n) 4057 418 337 142 120

Fisher's exact test = 4.3 × 10−10). The results were further validated by exome sequencing and Sanger sequencing (Fig. 1). Moreover, we identified several cellular pathways that were altered in endometrial cancer tissues (Table 3).

Table 3 Mutated pathways in endometrial cancer. Pathways involved in carcinogenesis

Mutated genes

hsa04150: mTOR signaling pathway hsa04722: Neurotrophin signaling pathway hsa04530: Tight junction

RICTOR, MTOR, PIK3R1, RPTOR, TSC1, AKT2 MAP3K1, NTRK3, PIK3R1, PYPN11, TP53, AKT2 CTNNB1, MYH9, PTEN, PPP2R1A, MLLT4, AKT2 FOXO1, MTOR, PIK3R1, RPTOR, TSC1, AKT2 MTOR, PIK3R1, ABL2, AKT2, ERBB2 CTNNB1, FLT1, PTEN, PIK3R1, PDGFRA, AKT2, ERBB2 PIK3R1, ATM, TP53, AKT2 CREBBP, PIK3R1, PTPN11, TYK2, AKT2

hsa04910: Insulin signaling pathway hsa04012: ErbB signaling pathway hsa04510: Focal adhesion hsa04210: Apoptosis hsa04630: Jak-STAT signaling pathway hsa04144: Endocytosis

FGFR2, FLT1, PDGFRA, RABEP1, KIT

4. Discussion 3.2. Mutation status of the NimbleGen Comprehensive Cancer Panel in endometrial cancer Notably, 120 mutations were identified in 99 genes, including 21 potential cancer driver genes in the Oncomine Cancer Research Panel, which was used in the National Cancer Institute Match Trial (Hovelson et al., 2015). Among the 21 mutated cancer driver genes were 8 tumor suppressor genes (ATM, MSH2, PIK3R1, PTCH1, PTEN, TET2, TP53, and TSC1) and 13 oncogenes (ALK, BCL9, CTNNB1, ERBB2, FGFR2, FLT3, HNF1A, KIT, MTOR, PDGFRA, PPP2R1A, PTPN11, and SF3B1). The results were further validated by exome sequencing and Sanger sequencing (Fig. 2). Of the 10 endometrial cancer samples, 5 harbored PTEN mutations. The spectrum of PTEN mutations included missense, nonsense, and frameshift mutations. The majority of PTEN mutations were missense mutations (30%), including p.D92Y and p.R130Q (COSM5033). Nevertheless, 10% of PTEN mutations were nonsense (rs121909219) and frameshift (p.His196Argfs) mutations. In addition, the TCGA project showed that the PIK3R1, AKT2 and FOXO1 mutation rates in endometrial cancer are 33%, 2%, and 4%, respectively. Similarly, our results showed the PIK3R1, AKT2, and FOXO1 mutations were missense mutations (p.N264T, p.E237D, and p.G655D, respectively); these were novel mutations. The products of all three genes are involved in the IL-7 signaling pathway (Clark et al., 2014). These results were further confirmed by Sanger sequencing (Fig. 3). It was unclear whether the p.N264T mutation within PIK3R1, the p.E237D mutation within AKT2, and the p.G655D mutation within FOXO1 were somatic or germline abnormalities; therefore, we performed Sanger sequencing on matched normal samples. Our data confirmed the results of somatic mutations in the development of endometrial cancer in these cancer-related genes.

In this study, we screened for mutations in 578 genes in endometrial cancer using the NimbleGen Comprehensive Cancer Panel on the Illumina HiSeq. We identified several cancer driver genes and defined the pathway involved in the development of endometrial cancer. Overall, 50% of endometrial cancers harbored mutations in the PTEN gene. We also detected molecular aberrations that led to putative activation of the IL-7 signaling pathway (n = 2 [PIK3R1, AKT2, FOXO1]), which has not previously been associated with endometrial cancer. PIK3R1 encodes the 85 kD regulatory subunit. An accumulating amount of research has identified a key role for PIK3R1 in human carcinogenesis (Jaiswal et al., 2009; Miller et al., 2011; Urick et al., 2011). A nonsense mutation within PIK3R1 that results in decreased PIK3R1 expression has been strongly linked with renal cell carcinoma (Lin et al., 2015). PIK3CA mutations occur in many types of cancers, whereas PIK3R1 mutations are restricted to a few tumor types (Anon., 2008; Jaiswal et al., 2009). PIK3R1 mutations preferentially localize to the inter-SH2 domain. In this study, we identified a novel mutation (p.N264T) in a patient with endometrial cancer in the RhoGAP domain. AKT2 is an oncogene encoding a protein belonging to a subfamily of Ser/Thr kinases. AKT2 is predominately expressed in insulin-responsive tissues, such as liver, fat, and skeletal muscle (Altomare and Testa, 2005) and plays an important role in tumorigenesis. AKT2 activation by phosphorylation and/or gene increases has been frequently studied in a broad spectrum of human cancers. However, its activation by genetic alteration is less well understood. AKT2 overexpression/activation has been found in colorectal, ovarian, pancreatic, and breast carcinomas (Bellacosa et al., 2005; Gonzalez and McGraw, 2009), while AKT2 amplification has been found in breast, ovarian and pancreatic carcinomas (Kirkegaard et al., 2010; Nakayama et al., 2006). AKT2 amplification correlates with a poor prognosis in ovarian carcinoma, as well as tumor size in soft tissue sarcoma (Dobashi et al., 2014; Nakayama et al., 2006).

Fig. 1. Confirmatory analysis by Sanger sequencing of KEGG cancer pathways (hsa05200) detected via capture-based NGS testing. (A) MSH2 (B) FGFR2 (C) mTOR.

Y.-S. Chang et al. / Experimental and Molecular Pathology 100 (2016) 8–12

11

Fig. 2. Confirmatory analysis by Sanger sequencing of Oncomine Cancer Research Panel detected via capture-based NGS testing. (A) PPP2R1A (B) SF3B1.

Fig. 3. Confirmatory analysis by Sanger sequencing of IL-7 signaling pathway detected via capture-based NGS testing. (A) PIK3R1 (B) AKT2 (C) FOXO1.

Mutations in the AKT2 gene, including p.D32H, p.R368C, and p.D399N, have been detected in endometrial cancer (Dutt et al., 2009). In the current study, we identified a novel mutation, p.E237D, in a patient with endometrial cancer located at the Ser/Thr protein kinase catalytic domain (amino acids 152–366). This mutation may influence the threedimensional (3D) structure of the catalytic domain and thus alter kinase function. FOXO1 belongs to the forkhead box O (FOXO) transcription factor subfamily. The FOXO subfamily is composed of four members: FOXO1, FOXO3, FOXO4, and FOXO6 (Greer and Brunet, 2005). The transcription factor FOXO1 modulates the expression of genes involved in cell cycle arrest, apoptosis, and DNA damage repair (Huang and Tindall, 2007). Deregulation of FOXO1 is frequently observed in tumors (Ju et al., 2014; Yu et al., 2014). FOXO1 (rs17592236, C N T) of the mir-137 target site has recently been linked to hepatocellular carcinoma (Tan et al., 2015). In this study, we identified a novel mutation (p.G655D) in a patient with endometrial cancer. However, the role of this mutation remains unclear. A limitation of this study is the small number of samples. To establish a comprehensive view of the genomic landscape of endometrial cancer, sequencing of additional samples will be necessary. Our results identified genes involved in the IL-7 signaling pathway that are mutated in endometrial cancer. These findings will provide additional information for drugs used in targeted therapy.

Acknowledgments We thank the National Center for Genome Medicine of the National Core Facility Program for Biotechnology, Ministry of Science and Technology, for the technical support. We also thank Wei-Chi

Wang from Health GeneTech Corporation, Taoyuan, Taiwan for his assistance with bioinformatics analysis. This work was supported by grants from China Medical University Hospital (DMR-103-121) and the Ministry of Science and Technology, Taiwan, R.O.C. (982320-B-037-010-MY3).

References Adzhubei, I.A., et al., 2010. A method and server for predicting damaging missense mutations. Nat. Methods 7, 248–249. Altomare, D.A., Testa, J.R., 2005. Perturbations of the AKT signaling pathway in human cancer. Oncogene 24, 7455–7464. Anon., 2008. Comprehensive genomic characterization defines human glioblastoma genes and core pathways. Nature 455, 1061–1068. Bellacosa, A., et al., 2005. Activation of AKT kinases in cancer: implications for therapeutic targeting. Adv. Cancer Res. 94, 29–86. Bokhman, J.V., 1983. Two pathogenetic types of endometrial carcinoma. Gynecol. Oncol. 15, 10–17. Clark, M.R., et al., 2014. Orchestrating B cell lymphopoiesis through interplay of IL-7 receptor and pre-B cell receptor signalling. Nat. Rev. Immunol. 14, 69–80. Dedes, K.J., et al., 2011. Emerging therapeutic targets in endometrial cancer. Nat Rev. Clin. Oncol. 8, 261–271. del Carmen, M.G., et al., 2012. Uterine papillary serous cancer: a review of the literature. Gynecol. Oncol. 127, 651–661. Dobashi, Y., et al., 2014. Significance of Akt activation and AKT gene increases in soft tissue tumors. Hum. Pathol. 45, 127–136. Dutt, A., et al., 2009. Somatic mutations are present in all members of the AKT family in endometrial carcinoma. Br. J. Cancer 101, 1218–1219 author reply 1220-1. Forbes, S.A., et al., 2015. COSMIC: exploring the world's knowledge of somatic mutations in human cancer. Nucleic Acids Res. 43, D805–D811. Gonzalez, E., McGraw, T.E., 2009. The Akt kinases: isoform specificity in metabolism and cancer. Cell Cycle 8, 2502–2508. Greer, E.L., Brunet, A., 2005. FOXO transcription factors at the interface between longevity and tumor suppression. Oncogene 24, 7410–7425. Hamilton, C.A., et al., 2006. Uterine papillary serous and clear cell carcinomas predict for poorer survival compared to grade 3 endometrioid corpus cancers. Br. J. Cancer 94, 642–646.

12

Y.-S. Chang et al. / Experimental and Molecular Pathology 100 (2016) 8–12

Hovelson, D.H., et al., 2015. Development and validation of a scalable next-generation sequencing system for assessing relevant somatic variants in solid tumors. Neoplasia 17, 385–399. Huang, H., Tindall, D.J., 2007. Dynamic FoxO transcription factors. J. Cell Sci. 120, 2479–2487. Jaiswal, B.S., et al., 2009. Somatic mutations in p85alpha promote tumorigenesis through class IA PI3K activation. Cancer Cell 16, 463–474. Ju, Y., et al., 2014. FOXO1-dependent DNA damage repair is regulated by JNK in lung cancer cells. Int. J. Oncol. 44, 1284–1292. Kandoth, C., et al., 2013. Integrated genomic characterization of endometrial carcinoma. Nature 497, 67–73. Kinde, I., et al., 2013. Evaluation of DNA from the Papanicolaou test to detect ovarian and endometrial cancers. Sci Transl. Med. 5, 167ra4. Kirkegaard, T., et al., 2010. Molecular alterations in AKT1, AKT2 and AKT3 detected in breast and prostatic cancer by FISH. Histopathology 56, 203–211. Kuhn, E., et al., 2012. Identification of molecular pathway aberrations in uterine serous carcinoma by genome-wide analyses. J. Natl. Cancer Inst. 104, 1503–1513. Landrum, M.J., et al., 2014. ClinVar: public archive of relationships among sequence variation and human phenotype. Nucleic Acids Res. 42, D980–D985. Langmead, B., et al., 2009. Ultrafast and memory-efficient alignment of short DNA sequences to the human genome. Genome Biol. 10, R25. Le Gallo, M., et al., 2012. Exome sequencing of serous endometrial tumors identifies recurrent somatic mutations in chromatin-remodeling and ubiquitin ligase complex genes. Nat. Genet. 44, 1310–1315.

Liang, H., et al., 2012. Whole-exome sequencing combined with functional genomics reveals novel candidate driver cancer genes in endometrial cancer. Genome Res. 22, 2120–2129. Lin, Y., et al., 2015. PIK3R1 negatively regulates the epithelial–mesenchymal transition and stem-like phenotype of renal cancer cells through the AKT/GSK3beta/CTNNB1 signaling pathway. Sci. Rep. 5, 8997. McKenna, A., et al., 2010. The Genome Analysis Toolkit: a MapReduce framework for analyzing next-generation DNA sequencing data. Genome Res. 20, 1297–1303. Miller, T.W., et al., 2011. Mutations in the phosphatidylinositol 3-kinase pathway: role in tumor progression and therapeutic implications in breast cancer. Breast Cancer Res. 13, 224. Nakayama, K., et al., 2006. Sequence mutations and amplification of PIK3CA and AKT2 genes in purified ovarian serous neoplasms. Cancer Biol. Ther. 5, 779–785. Sherry, S.T., et al., 2001. dbSNP: the NCBI database of genetic variation. Nucleic Acids Res. 29, 308–311. Tan, C., et al., 2015. Polymorphisms in microRNA target sites of forkhead box O genes are associated with hepatocellular carcinoma. PLoS ONE 10, e0119210. Urick, M.E., et al., 2011. PIK3R1 (p85alpha) is somatically mutated at high frequency in primary endometrial cancer. Cancer Res. 71, 4061–4067. Yu, D.A., et al., 2014. Forkhead transcription factor FOXO1 inhibits nuclear factor-kappaB in gastric cancer. APMIS 122, 848–855. Zhao, S., et al., 2013. Landscape of somatic single-nucleotide and copy-number mutations in uterine serous carcinoma. Proc. Natl. Acad. Sci. U. S. A. 110, 2916–2921.