Uterine leiomyosarcoma (ULMS) is a malignancy, which arises from the uterine smooth muscle. Because of its rarity, aggressive nature, and extremely poor prognosis, the molecular mechanisms driving ULMS remain elusive. To identify candidate cancer genes (CCG) driving ULMS, we conducted an in vivo Sleeping Beauty (SB) transposon mutagenesis screen in uterine myometrium–specific, PTEN knockout, KRAS mutant (PTEN KO/KRAS) mice. ULMS quickly developed in SB PTEN KO/KRAS mice, but not in PTEN KO/KRAS mice, demonstrating the critical importance of SB mutagenesis for driving ULMS in this model. Subsequent sequencing of SB insertion sites in these tumors identified 19 ULMS CCGs that were significantly enriched in known cancer genes. Among them, Zfp217 and Sfmbt2 functioned at early stages of tumor initiation and appeared to be oncogenes. Expression of ZNF217, the human homolog of ZFP217, was shown to be elevated in human ULMS compared with paired normal uterine smooth muscle, where it negatively correlated with patient prognosis. Inhibition of ZNF217 suppressed, whereas overexpression induced, proliferation, survival, migration, and stemness of human ULMS. In a second ex vivo ULMS SB metastasis screen, three CCGs were identified that may drive ULMS metastasis to the lung. One of these CCGs, Nrd1 (NRDC in humans), showed stronger expression in human metastatic tumors compared with primary ULMS and negatively associated with patient survival. NRDC knockdown impaired migration and adhesion without affecting cell proliferation, whereas overexpression had the opposite effect. Together, these results reveal novel mechanism driving ULMS tumorigenesis and metastasis and identify ZNF217 and NRDC as potential targets for ULMS therapy.

Significance:

An in vivo Sleeping Beauty transposon mutagenesis screen identifies candidate cancer genes that drive initiation and progression of uterine leiomyosarcoma and may serve as therapeutic targets.

Uterine leiomyosarcoma (ULMS) accounts for only 1% to 2% of all uterine malignancies, but it is a very aggressive disease with an extremely poor prognosis. The 5-year survival rate is 0% to 20% for patients with extrauterine metastasis and 50% for even early-stage patients due to a high metastatic potential (1). In addition to its rarity, the difficult preoperative diagnosis of ULMS, due to its similarity to benign uterine leiomyoma, has hindered attempts to establish a large-scale comprehensive genomic dataset for ULMS. In 2017, The Cancer Genome Atlas (TCGA) network released a dataset comprised of 206 cases of adult soft tissue sarcoma (2). Eighty leiomyosarcomas (LMS) were included in this dataset, revealing that TP53, RB1, and PTEN are deeply or shallowly deleted in 69%, 92%, and 81% of LMS tumors, or mutated in 50%, 15%, and 5% of LMS tumors, respectively. Although the importance of the PIK3CA–AKT–mTOR signaling pathway in LMS was asserted, this dataset included only 27 cases of ULMS. This rarity of ULMS has made it difficult to identify driver mutations owing to the lack of statistical power.

Forward genetic screens, conducted in vitro using pooled shRNA or sgRNA libraries, have provided a powerful tool for the discovery of candidate cancer genes (CCG) related to a phenotype of interest such as tumor progression, chemoresistance, systemic lethality, or stemness (3, 4). An important advantage of forward genetic screens is that they allow for unbiased and targeted screening; hence, forward genetic screens have contributed to many new and unexpected discoveries. Although various types of in vitro pooled library screens have been conducted, in vivo screens have the added potential of being more clinically relevant because they can select for therapeutic targets with biological significance (5, 6). In addition to shRNA or sgRNA screens, Sleeping Beauty (SB) transposon mutagenesis screens performed under normal immune conditions have also provided a powerful tool for identifying CCGs in several types of cancer (7–11). In these screens, SB transposons are randomly mobilized genome-wide via transposon excision and reintegration in a cut-and-paste manner. This provides survival advantages to the cells in which the SB transposon integrates and deregulates the expression of a proto-oncogene or downregulates the expression of a tumor suppressor gene. Therefore, SB mutagenesis makes it possible to perform both loss-of-function and gain-of-function screens simultaneously and discover CCGs associated with tumor initiation and progression.

In studies described here, we report the CCGs associated with initiation and metastases of ULMS, identified in two unbiased, whole-genome, SB mutagenesis screens performed in uterine myometrium-specific PTEN knockout and KRAS-activated mutant mice. Smooth muscle cell-specific PTEN KO mice were used because they have been reported to develop leiomyosarcomagenesis in the abdomen (but not in the uterus; ref. 12), whereas constitutive KRAS activation has been reported in numerous cancers, including human ULMS (2, 13). Numerous SB mutagenesis screens have been reported for epithelial and hematopoietic cancers (7–11, 14, 15). However, only a few SB mutagenesis screens have been reported for sarcomas. The two that have been reported include screens for soft tissue sarcoma and osteosarcoma performed under systemic and osteoblast progenitor specific TP53 loss, respectively (16, 17). Thus, our screening process performed in PTEN-mutant and KRAS-activated mice is novel and represents the first reported genetic screen for ULMS CCGs performed to date. It is also novel in that it was able to identify CCGs inducing metastases using an ex vivo approach (8, 18).

Animal experiments

Rosa26-loxP-STOP-loxP-SB11 transposase knock-in mice [Gt(ROSA)26Sortm2(sb11)Njen; ref. 19], two lines of T2/Onc2 SB transposon transgenic mice [TgTn(sb-T2/Onc2)6113Njen/Nci and TgTn(sb-T2/Onc2)6070Njen/Nci] were obtained from the NCI Mouse Repository. Amhr2tm3(cre)Bhr was provided by Dr. Richard Behringer (University of Texas M.D. Anderson Cancer Center, Houston, TX). C;129S4-Ptentm1Hwu/J and B6.129S4-Krastm4Tyj/J (The Jackson Laboratory strain numbers 004597 and 008179, respectively) were obtained from The Jackson Laboratory. Mice that were triple homozygous for the floxed Pten allele, Rosa26-lsl-SB11, and T2/Onc2 (either 6113 or 6070) and heterozygous for Kras G12D/+ were crossed with mice heterozygous for Amhr2-Cre/+ and homozygous for the floxed Pten allele, to generate Amhr2-Cre/+, T2Onc2/+, Rosa26-lsl-SB11/+, Pten fl/fl, and Kras G12D/+ mice, which were used for screening. Female mice were used in these experiments. Mice were monitored twice a week for health status and tumor burden. Palpation was used to estimate abdominal tumor size. Mice were euthanized and necropsied when the tumor size was estimated to be 2 cm in diameter or when the mice showed symptoms of morbidity/moribundity, including ruffled coat, lack of activity, distended abdomen, or dyspnea, whichever occurred first. Uterine tumors seen grossly, and normal uteri were collected for histopathologic examination and molecular analysis. Mice were housed in a specific pathogen-free facility with a 12-hour light/12-hour dark cycle. All mouse procedures were approved by Institutional Animal Care and Use Committee of Houston Methodist Research Institute.

Sequencing

Transposon insertion sites were mapped using a previously reported splink HiSeq approach (14). In brief, genomic tumor DNA (gDNA) was acoustically sheared to 200 bp in size using a Covaris S220 focused ultrasonicator and end repaired. Adapters (linker+: 5-GTA ATA CGA CTC ACT ATA GGG CTC CGC TTA AGG GAC-3′ and linker–: 5-Phos-GTC CCT TAA GCG GAG-C3spacer-3′) were annealed and ligated to repaired DNA. Purified repaired DNA was PCR-amplified using linker (5′-GTA ATA CGA CTC ACT ATA GGG C-3′), IRR (5′-GGA TTA AAT GTC AGG AAT TGT GAA AA-3′), and IRL (5′-AAA TTT GTG GAG TAG TTG AAA AAC GA-3′) primers. Nested secondary PCR was performed using indexed SB-2ndR primer (5′- CAA GCA GAA GAC GGC ATA CGA GAT (8 bp index) GTG ACT GGA GTT CAG ACG TGT GCT CTT CCG ATC TTA GGG CTC CGC TTA AGG GAC-3′), SB-RF primer (AAT GAT ACG GCG ACC GAG ATC TAC ACT CTT TCC CTA CAC GAC GCT CTT CCG ATC T (1–4 bp stagger) GGC TAA GGT GTA TGT AAA CTT CCG ACT TCA ACT G-3′), and SB-LF primer (AAT GAT ACG GCG ACC GAG ATC TAC ACT CTT TCC CTA CAC GAC GCT CTT CCG ATC T (1–4 bp stagger) AAC TTA AGT GTA TGT AAA CTT CCG ACT TCA ACT G-3′). All oligos were purchased from Integrated DNA Technologies with standard desalting. DNA fragments between 175 and 450 bp in size from the amplicon libraries were excised from the agarose gel and quantified using a Qubit Fluorometer, Agilent 2100 Bioanalyzer and qRT-PCR. Excised libraries were then pooled, loaded onto two flow cell lanes, and sequenced on the Hiseq2500 platform. Sequencing reads were converted to FASTQ format and demultiplexed using bcl2fastq-1.8.4 (Illumina) at Eurofins Genomics. Reads with a dual index that did not match the barcode were discarded and reads with a matching dual index had the index and preceding sequence removed. Next, the p1 primer sequence was clipped from the reads, and the reads were mapped to the mouse genome using the SBCapSeq pipeline (15).

Common insertion site analysis

To avoid bias due to the local hopping phenomena, all insertions on chromosome 4 in tumors from 6070 and chromosome 1 in tumors from 6,113 were removed, as these insertions were located on chromosomes containing the T2/Onc2 donor concatamer. Insertions in Sfi1 were removed due to Sfi1 being annotated in multiple different genomic loci; insertions in En2 and Foxf2, known SB hotspots, were also removed. CCGs were identified by applying gCIS to insertions with >50 sequence read counts and genes with a Bonferroni adjusted P value of 3, in the same manner as in a previous work (14). For trunk driver analysis, genes with a Bonferroni adjusted P value <0.05 were defined as candidate trunk drivers.

Primary SB cell lines

Tissues from normal uterine smooth muscle, ULMS, or lung metastases, were collected at necropsies, promptly washed by PBS, and cut into 5 mm square pieces. Each piece was put on a gelatin-coated 6-well plate (Corning) and cultured in DMEM medium supplemented with 10% FBS and penicillin–streptomycin. The cells were trypsinized and seeded to gelatin-coated 100 mm plates to expand (Corning).

Human samples

Paraffin embedded blocks, which were collected at surgery or biopsy, from April 1, 1999, to December 31, 2018, in Osaka University, were used for this study. Paired samples including normal uterine smooth muscle tissue and ULMS from 14 patients were analyzed for ZNF217 expression via IHC. Normal tissue, ULMS, and metastatic tumors from 12 cases were analyzed for NRDC expression via IHC. Among them, two cases had all three types of tissues, six cases had normal tissue and ULMS, two cases had ULMS and metastatic tumors, and two cases had only metastatic tumors. Written informed consent was obtained from all patients, and this procedure was approved by our Institutional Review Board at Osaka University.

Statistical analysis

All data were analyzed with GraphPad Prism 7 (GraphPad Prism, GraphPad, Software) and are provided as mean ± SD, unless otherwise indicated. Statistical analyses were performed using an unpaired Student t test, Mann–Whitney U test, or a one-way ANOVA. For multiple comparisons, P values were adjusted by Bonferroni correction of the differences, and the differences in the mean values among the groups were examined by Dunnett multiple comparison test. P < 0.05 was considered statistically significant. Survival outcome of ZNF217 and NRDC expression in the Sarcoma dataset from TCGA were analyzed at GEPIA (20). Other methods are described in Supplementary file S1.

To identify CCGs driving ULMS, we conducted an SB mutagenesis screen in PTEN KO/KRAS mutant mice. To activate SB transposition, express mutant KRAS, and inactivate PTEN expression specifically in uterine smooth muscle, we used a Cre-recombinase transgenic mouse line that expressed Cre under the control of the anti-Mullerian hormone type II receptor (Amhr2; ref. 21). We also used two different SB transposon transgenic lines (T2Onc2;6007 and T2Onc2;6113), in which the high-copy SB transposon concatamer was located on two different chromosomes (chromosome 4 and 1, respectively), to accomplish genome-wide mutagenesis and avoid the problems caused by SB local hopping. T2Onc2 contains a transcriptional stop cassette to inactivate tumor suppressor genes, in addition to a murine stem cell virus promoter and downstream splice donor to activate oncogenes (19). We then ages 55 experimental PTEN KO/KRAS/SB/Amhr2-Cre mice and 14 PTEN KO/KRAS/Amhr2-Cre control mice and monitored them for tumor development. All experimental mice died of multiple uterine tumors by three months of age, whereas none of the control mice developed uterine tumors (Fig. 1A and B). Instead, control mice died of low-grade serous adenocarcinoma of the ovaries by 6 months of age (Fig. 1B; Supplementary Figs. S1A and S1B). These malignant ovarian cancers were not identified in PTEN KO/KRAS/SB/Amhr2-Cre mice. The very aggressive nature of the uterine tumors that developed in experimental mice were like what is observed in human UMLS. Uterine tumors were histologically diagnosed as ULMS due to their morphology, showing a disarray of spindle cells, fascicular growth pattern, necrosis, and mitoses (Fig. 1C), and immunoreactivity for desmin and α-smooth muscle actin (αSMA; Fig. 1D). As expected, SB transposase was expressed in normal murine uterine smooth muscle, but not in uterine endometrium (Supplementary Fig. S1C). ULMS tumor cells were also immunoreactive for nuclear SB transposase (Fig. 1D). Considering that control mice without SB did not develop ULMS, these results clearly show that SB transposition is essential for ULMS. Both PTEN KO/KRAS/SB(6007)/Amhr2-Cre and PTEN KO/KRAS/SB(6113)/Amhr2-Cre lines showed 100% tumor penetrance and similar histologic findings; therefore, we combined the data from these two lines for downstream analysis.

Figure 1.

SB transposon mutagenesis induces leiomyosarcomagenesis in murine uterine smooth muscle. A, Representative gross picture of uterine tumors from PTEN KO/KRAS/SB mice. B, Kaplan–Meier plot of the overall survival of PTEN KO/KRAS, PTEN KO/KRAS/SB(6070), PTEN KO/KRAS/SB(6113), and wild-type (WT) mice. Statistical analyses were performed using the log-rank test. C, Hematoxylin and eosin staining of uterine tumors. Tumor tissue was hypercellular with spindle-shaped cells (left; original magnification, ×40) and shows necrosis (middle; ×200) and high mitotic activity (right; ×400). Scale bar, 500 μm (left), 100 μm (middle), and 50 μm (right), respectively. D, IHC staining of uterine tumors. Tumor cells showed immunoactivity for αSMA (left), desmin (middle), and nuclear SB transposase (right, ×100; top left corner, ×200). Scale bars, 200 μm.

Figure 1.

SB transposon mutagenesis induces leiomyosarcomagenesis in murine uterine smooth muscle. A, Representative gross picture of uterine tumors from PTEN KO/KRAS/SB mice. B, Kaplan–Meier plot of the overall survival of PTEN KO/KRAS, PTEN KO/KRAS/SB(6070), PTEN KO/KRAS/SB(6113), and wild-type (WT) mice. Statistical analyses were performed using the log-rank test. C, Hematoxylin and eosin staining of uterine tumors. Tumor tissue was hypercellular with spindle-shaped cells (left; original magnification, ×40) and shows necrosis (middle; ×200) and high mitotic activity (right; ×400). Scale bar, 500 μm (left), 100 μm (middle), and 50 μm (right), respectively. D, IHC staining of uterine tumors. Tumor cells showed immunoactivity for αSMA (left), desmin (middle), and nuclear SB transposase (right, ×100; top left corner, ×200). Scale bars, 200 μm.

Close modal

To identify the genes driving ULMS in PTEN KO/KRAS/SB/Amhr2-Cre mice, we extracted tumor DNA and PCR amplified and sequenced the transposon insertional sites from 85 ULMS tumors [41 tumors from PTEN KO/KRAS/SB(6007)/Amhr2-Cre mice and 44 tumors from PTEN KO/KRAS/SB(6113)/Amhr2-Cre mice] using splink HiSeq methods (8). Gene-centric common insertion site (gCIS)-calling (22) was then used to calculate the observed and expected transposon insertion numbers within the coding regions of all RefSeq genes. This analysis identified 19 statistically significant CCGs from SB ULMS tumors (P < 0.05, calculated by χ2 test followed by Bonferroni correction; Fig. 2A; Table 1; Supplementary Table S1). The number of known cancer genes in CCGs identified in the SB screen and those in all human genes were analyzed using the chi-square test (P = 2.80E−8; Fig. 2B). These CCGs are significantly enriched in known cancer genes, Nf1, Aff1, Cblb, Arid1b, and Ext1, which are listed in the cancer gene census database (Supplementary Fig. S2; ref. 23), further suggesting that they are cancer genes driving UMLS.

Figure 2.

CCGs identified in SB mutagenesis screen. A, Nineteen CCGs were identified in the PTEN KO/KRAS mutagenesis screen. The percent of total tumors with an insertional mutation in a CCG is shown, with each rectangle representing an individual tumor. Red and blue rectangles indicate insertions in the sense and antisense strands, respectively. B, Venn diagram showing a comparison of CCGs detected in the PTEN KO/KRAS SB screen and known cancer genes listed in the Cancer Gene Census database. The proportion of known cancer genes in the SB screen is significantly higher than that of known cancer genes in all human genes (P = 2.80E−8). C, Median and mean sequence read counts for each CCG are shown in a scatter plot. Known CCGs are shown in bold. D, The location of the transposon insertion sites in two CCGs with high read counts, Zfp217 (top) and Sfmbt2 (bottom) are shown. Each arrow indicates the location of a single transposon insertion. Red, transposon insertions in the sense strand. E,Zfp217 mRNA levels (left) and Sfmbt2 (right) in SB tumors with and without transposon insertions were analyzed via qPCR. *, P < 0.05. Transcriptional orientation of Zfp217 and Sfmbt2 were from 3′ to 5′ and from 5′ to 3′, respectively.

Figure 2.

CCGs identified in SB mutagenesis screen. A, Nineteen CCGs were identified in the PTEN KO/KRAS mutagenesis screen. The percent of total tumors with an insertional mutation in a CCG is shown, with each rectangle representing an individual tumor. Red and blue rectangles indicate insertions in the sense and antisense strands, respectively. B, Venn diagram showing a comparison of CCGs detected in the PTEN KO/KRAS SB screen and known cancer genes listed in the Cancer Gene Census database. The proportion of known cancer genes in the SB screen is significantly higher than that of known cancer genes in all human genes (P = 2.80E−8). C, Median and mean sequence read counts for each CCG are shown in a scatter plot. Known CCGs are shown in bold. D, The location of the transposon insertion sites in two CCGs with high read counts, Zfp217 (top) and Sfmbt2 (bottom) are shown. Each arrow indicates the location of a single transposon insertion. Red, transposon insertions in the sense strand. E,Zfp217 mRNA levels (left) and Sfmbt2 (right) in SB tumors with and without transposon insertions were analyzed via qPCR. *, P < 0.05. Transcriptional orientation of Zfp217 and Sfmbt2 were from 3′ to 5′ and from 5′ to 3′, respectively.

Close modal
Table 1.

Nineteen CCGs for ULMS.

Gene nameLocusgCISInsertion frequency (%)Mean sequence read countsMedian sequence read counts
P value
Zfp217 chr2 24.1 576.2 574.4 
Nf1 chr11 1.86E−280 30.4 629.0 371.3 
AU040972 chr11 1.90E−100 5.1 887.0 817.2 
Zmiz1 chr14 4.03E−64 10.1 493.4 336.6 
Ncoa3 chr2 8.12E−27 5.1 849.7 116.3 
Aff1 chr5 6.48E−22 5.1 743.7 767.7 
Sfmbt2 chr2 9.46E−15 6.3 1036.6 856.1 
Gsk3b chr16 2.90E−08 5.1 351.9 384.9 
Phf21a ch2 1.15E−07 5.1 159.6 121.2 
Cblb chr16 2.30E−07 5.1 239.2 161.9 
Arid1b chr17 3.91E−07 6.3 261.6 67.9 
Zcchc7 chr4 1.03E−06 7.3 479.0 201.3 
Bnc2 chr4 4.19E−05 9.8 476.6 444.7 
Adamtsl3 chr7 0.000249 5.1 524.9 172.2 
Rere chr4 0.000450 7.3 99.9 68.2 
Agmo chr12 0.000709 6.3 346.9 292.8 
Ext1 chr15 0.000825 5.1 868.0 595.2 
Zbtb20 chr16 0.004800 8.9 459.2 215.8 
Ttc28 chr5 0.012447 5.1 203.4 126.9 
Gene nameLocusgCISInsertion frequency (%)Mean sequence read countsMedian sequence read counts
P value
Zfp217 chr2 24.1 576.2 574.4 
Nf1 chr11 1.86E−280 30.4 629.0 371.3 
AU040972 chr11 1.90E−100 5.1 887.0 817.2 
Zmiz1 chr14 4.03E−64 10.1 493.4 336.6 
Ncoa3 chr2 8.12E−27 5.1 849.7 116.3 
Aff1 chr5 6.48E−22 5.1 743.7 767.7 
Sfmbt2 chr2 9.46E−15 6.3 1036.6 856.1 
Gsk3b chr16 2.90E−08 5.1 351.9 384.9 
Phf21a ch2 1.15E−07 5.1 159.6 121.2 
Cblb chr16 2.30E−07 5.1 239.2 161.9 
Arid1b chr17 3.91E−07 6.3 261.6 67.9 
Zcchc7 chr4 1.03E−06 7.3 479.0 201.3 
Bnc2 chr4 4.19E−05 9.8 476.6 444.7 
Adamtsl3 chr7 0.000249 5.1 524.9 172.2 
Rere chr4 0.000450 7.3 99.9 68.2 
Agmo chr12 0.000709 6.3 346.9 292.8 
Ext1 chr15 0.000825 5.1 868.0 595.2 
Zbtb20 chr16 0.004800 8.9 459.2 215.8 
Ttc28 chr5 0.012447 5.1 203.4 126.9 

Previous studies have suggested that most CCGs identified by SB mutagenesis function as branch driver genes, acquired during late stages of tumor progression. To identify candidate trunk driver genes that are important for sarcomagenesis, we focused on genes associated with the highest number of sequencing reads, suggesting high clonality, in a similar approach to previous SB screens (Fig. 2C; refs. 10, 11, 24). Although high clonality is consistent with a truncal mutation, it is not exactly synonymous. It is also consistent with another evolutionary scenario, selective sweep, which is the process of increasing the numbers of new advantageous mutations that arise later in a population. Among them, Zfp217 and Sfmbt2 showed transposon insertion sites that were primarily located on the lateral strand, and upstream of the transcription initiation site, suggesting their oncogenic functions (Fig. 2D; ref. 25). Indeed, mRNA levels of Zfp217 and Sfmbt2 analyzed by qPCR were significantly enhanced in tumors with transposon insertions, compared with tumors without SB insertions (Fig. 2E).

To functionally validate these CCGs, we decided to focus on Zfp217 because of its high prevalence in SB tumors (Table. 1). First, to examine the clinical relevance of ZNF217 in human ULMS, ZNF217 expression levels were assessed via IHC staining in 14 cases of ULMS in which paired samples, including both normal uterine tissues and ULMS samples, were available. This analysis showed that ZNF217 is expressed more often in ULMS compared with normal uterine myometrium (Supplementary Table. S2), and at higher levels (Fig. 3A), further suggesting that ZNF217 is a candidate ULMS driver gene. A large-scale comprehensive dataset containing only ULMS samples does not exist; hence, we analyzed the TCGA dataset targeting adult soft tissue sarcoma via GEPIA (20). Overall survival tended to be shorter in cases with high ZNF217 expression (Fig. 3B) in this dataset including sarcomas other than ULMS.

Figure 3.

ZNF217 protein was highly expressed in human uterine leiomyosarcoma, and its inhibition caused an antitumor effect. A, Representative IHC image of ZNF217 expression in normal human uterine myometrium and leiomyosarcoma (left, ×40; right, ×100) are shown. The dotted line shows the border between normal myometrium and leiomyosarcoma. Scale bar, 500 μm (left) and 200 μm (right). B, Overall survival of patients with high or low ZNF217 expression from the TCGA sarcoma dataset was analyzed at GEPIA. The median was used as a cut-off value. C, SK-UT1 (left), LMS (middle), and SKN (right) cells were counted 4 days after transfection with NC-siRNA or three ZNF217-siRNAs (n = 3 for each; *, P < 0.05 vs. all). D, SK-UT1 cells transduced with NC-shRNA or two ZNF217-shRNAs were subcutaneously injected into nude mice. Tumor volumes were measured and plotted at different days after injection (n = 8 for each; *, P < 0.05 vs. two ZNF217-shRNAs; left). Right, Western blot analysis of each xenografted tumor. E, Number of colonies more than 100 μm in size in soft agar were measured 18 days after seeding LMS (left) or SKN (right) cells transfected with NC-siRNA or three ZNF217-siRNAs (n = 3 for each; *, P < 0.05 vs. all). F, Number of tumor spheres more than 100 μm in size (LMS, left) or 200 μm (SKN, right; n = 3 for each; *, P < 0.05 vs. all) were counted 2 weeks after seeding. LMS and SKN cells were transfected with NC-siRNA or three ZNF217-siRNAs 2 days before seeding (n = 4 for each; *, P < 0.05 vs. all). G, Wound distance was measured 24 hours (LMS; left) or 12 hours (SK-UT1; right) after scratch (n = 4 for each; *, P < 0.05 vs. all). Wound closure percentage is shown as wound distance at 24 (LMS) or 12 hours (SK-UT1) divided by the initial distance.

Figure 3.

ZNF217 protein was highly expressed in human uterine leiomyosarcoma, and its inhibition caused an antitumor effect. A, Representative IHC image of ZNF217 expression in normal human uterine myometrium and leiomyosarcoma (left, ×40; right, ×100) are shown. The dotted line shows the border between normal myometrium and leiomyosarcoma. Scale bar, 500 μm (left) and 200 μm (right). B, Overall survival of patients with high or low ZNF217 expression from the TCGA sarcoma dataset was analyzed at GEPIA. The median was used as a cut-off value. C, SK-UT1 (left), LMS (middle), and SKN (right) cells were counted 4 days after transfection with NC-siRNA or three ZNF217-siRNAs (n = 3 for each; *, P < 0.05 vs. all). D, SK-UT1 cells transduced with NC-shRNA or two ZNF217-shRNAs were subcutaneously injected into nude mice. Tumor volumes were measured and plotted at different days after injection (n = 8 for each; *, P < 0.05 vs. two ZNF217-shRNAs; left). Right, Western blot analysis of each xenografted tumor. E, Number of colonies more than 100 μm in size in soft agar were measured 18 days after seeding LMS (left) or SKN (right) cells transfected with NC-siRNA or three ZNF217-siRNAs (n = 3 for each; *, P < 0.05 vs. all). F, Number of tumor spheres more than 100 μm in size (LMS, left) or 200 μm (SKN, right; n = 3 for each; *, P < 0.05 vs. all) were counted 2 weeks after seeding. LMS and SKN cells were transfected with NC-siRNA or three ZNF217-siRNAs 2 days before seeding (n = 4 for each; *, P < 0.05 vs. all). G, Wound distance was measured 24 hours (LMS; left) or 12 hours (SK-UT1; right) after scratch (n = 4 for each; *, P < 0.05 vs. all). Wound closure percentage is shown as wound distance at 24 (LMS) or 12 hours (SK-UT1) divided by the initial distance.

Close modal

Considering the possible clinical significance of ZNF217 in human ULMS, we moved to assess the effects of ZNF217 inhibition on human ULMS cell lines, SK-UT1, LMS, and SKN. siRNA-directed inhibition of ZNF217 (Supplementary Fig. S3), moderately but statistically significantly, impaired cell proliferation (Fig. 3C). Similarly, ZNF217 inhibition via shRNA suppressed tumor development in mouse xenograft assays (Fig. 3D). Likewise, ZNF217 inhibition strongly suppressed anchorage-independent growth (Fig. 3E; Supplementary Figs. S4A and S4B). Finally, tumor sphere formation (stem-like properties) and cell migration were also inhibited following downregulation of ZNF217 expression (Fig. 3F and G; Supplementary Figs. S5A and S5B). Collectively, these results suggest that ZNF217 is a potential therapeutic target in ULMS.

Next, to confirm the oncogenic function of ZNF217 in human ULMS, ULMS cell lines overexpressing ZNF217 were tested. Stable overexpression of ZNF217 in SK-UT1 and LMS cells using lentiviral vectors, and transient overexpression in SKN cells with pLXSN-ZNF217 transfection, significantly promoted cell proliferation (Fig. 4A). In contrast to suppressing anchorage-independent growth, tumor sphere formation, and cell migration via ZNF217 inhibition, ZNF217 overexpression significantly increased these capacities (Fig. 4BD), suggesting the oncogenic role of ZNF217 in human ULMS.

Figure 4.

ZNF217 overexpression promoted tumor progression. A, SK-UT1 and LMS cells were lentivirally transduced with an ZNF217 expression vector or a control vector. SKN cells were transiently transfected with pLXSN or pLXSN-ZNF217. SK-UT1 (left) and LMS (middle) 4 days after seeding and SKN (right) 4 days after transfection were assessed using the MTS assay (n = 4 for each; *, P < 0.05 between the means of groups). Western blot analysis showed protein expression levels of each cell lysate. B, Number of colonies in soft agar more than 50 μm in size (LMS, left) or 100 μm (SKN, right) were counted 14 days after seeding (n = 3 for each; *, P < 0.05 between the means of groups). C, Tumor spheres more than 50 μm in size for LMS (left) and SKN (right) were counted 14 days after seeding (n = 4 for each; *, P < 0.05 between the means of groups). D, Wound distance was measured in LMS cells 12 hours after scratching (n = 3 for each; *, P < 0.05 between the means of groups). Wound closure percentage is shown as wound distance at 12 hours divided by the initial distance.

Figure 4.

ZNF217 overexpression promoted tumor progression. A, SK-UT1 and LMS cells were lentivirally transduced with an ZNF217 expression vector or a control vector. SKN cells were transiently transfected with pLXSN or pLXSN-ZNF217. SK-UT1 (left) and LMS (middle) 4 days after seeding and SKN (right) 4 days after transfection were assessed using the MTS assay (n = 4 for each; *, P < 0.05 between the means of groups). Western blot analysis showed protein expression levels of each cell lysate. B, Number of colonies in soft agar more than 50 μm in size (LMS, left) or 100 μm (SKN, right) were counted 14 days after seeding (n = 3 for each; *, P < 0.05 between the means of groups). C, Tumor spheres more than 50 μm in size for LMS (left) and SKN (right) were counted 14 days after seeding (n = 4 for each; *, P < 0.05 between the means of groups). D, Wound distance was measured in LMS cells 12 hours after scratching (n = 3 for each; *, P < 0.05 between the means of groups). Wound closure percentage is shown as wound distance at 12 hours divided by the initial distance.

Close modal

ULMS is a very aggressive disease that metastasizes to the liver and lungs. To identify genes responsible for lung metastasis, we performed a second generation ex vivo screen (Fig. 5A) because primary ULMS in SB mice grows too rapidly to observe metastasis. In this ex vivo screen, we established 1st generation cell lines from primary ULMS tumors produced in SB mice and then injected these cells into the tail vein of immunocompromised NSG mice. Lung tumors were produced in transplanted mice with a success rate of 80% and 71% from SB (6070) and SB (6113) mice, respectively. Cells from these tumors were spindle-shaped at both low and high densities and resembled human ULMS cell lines (Fig. 5B). αSMA and desmin were also expressed in the cytoplasm of these cells, and SB transposase was expressed in the nucleus (Fig. 5C). Second generation cell lines were then established from these lung metastatic tumors and reinjected into the tail vein of immunocompetent mice, to obtain tumors that developed even in the presence of the immune system. Lung metastases again developed in immunocompetent mice, within a month of injection, at a 56.7% penetrance rate (Supplementary Fig. S6). Histologically, lung tumors resembled primary ULMS and showed a disarray of spindle-shaped mitotic cells and immunoactivity for nuclear SB transposase (Fig. 5D and E). Fifty lung metastases were then sequenced using splink HiSeq (8) and analyzed via gCIS for CCGs in the same manner as the primary-ULMS screen. Three candidate CCGs that were present in metastatic tumors but not primary tumors. Nrd1, Nsf, and Utrn were identified as potential drivers of lung metastases (Supplementary Tables S3 and S4). The most frequently mutated CCG (Nrd1 in mouse and NRDC in humans) encodes for nardilysin, a zinc-dependent metalloendopeptidase belonging to the M16 peptidase family.

Figure 5.

Ex vivo SB mutagenesis screen identified CCGs driving lung metastases of ULMS. A, First-generation cell lines were established from primary SB tumors [SB(6070) and SB(6113)]. Second-generation cell lines were then established from the lung metastases that developed in NSG mice following tail vein injection. Lung metastases (n = 50) that developed in immunocompetent mice were sequenced. B, Representative images of the 1st generation cell line, low density (top) and high density (bottom; ×100). Scale bars, 200 μm. C, Immunofluorescent staining for DAPI (blue), αSMA (green; top), desmin (green; middle), and SB, transposase (green; bottom) in a first generation cell line (×200). Scale bars, 100 μm. D, Macro image of lung metastasis in immunocompetent mouse. E, Hematoxylin and eosin staining of lung (left, ×40; middle, ×200) and IHC staining for SB transposase in a lung metastasis (right, ×100). Scale bars, 500, 100, and 200 μm, respectively.

Figure 5.

Ex vivo SB mutagenesis screen identified CCGs driving lung metastases of ULMS. A, First-generation cell lines were established from primary SB tumors [SB(6070) and SB(6113)]. Second-generation cell lines were then established from the lung metastases that developed in NSG mice following tail vein injection. Lung metastases (n = 50) that developed in immunocompetent mice were sequenced. B, Representative images of the 1st generation cell line, low density (top) and high density (bottom; ×100). Scale bars, 200 μm. C, Immunofluorescent staining for DAPI (blue), αSMA (green; top), desmin (green; middle), and SB, transposase (green; bottom) in a first generation cell line (×200). Scale bars, 100 μm. D, Macro image of lung metastasis in immunocompetent mouse. E, Hematoxylin and eosin staining of lung (left, ×40; middle, ×200) and IHC staining for SB transposase in a lung metastasis (right, ×100). Scale bars, 500, 100, and 200 μm, respectively.

Close modal

To confirm the clinical relevance of NRDC in human ULMS, NRDC protein expression was assessed in human ULMS tissue at different time points, including tissues taken from primary surgeries and biopsies, or surgeries to diagnose or resect metastatic tumors. Normal uterine tissue, ULMS, and metastatic tumors from 12 cases were analyzed for NRDC expression (Supplementary Table S5). In one case, where all three types of tissues were available, NRDC expression was not detected in normal uterine myometrium or ULMS at primary surgery but did show immunoactivity for NRDC in metastatic tissue (Fig. 6A). We also analyzed survival outcomes via GEPIA and found that high NRDC expression was significantly associated with shorter overall survival (Fig. 6B; ref. 20).

Figure 6.

Evidence indicating that NRDC is important for lung metastasis of human ULMS. A, IHC staining for NRDC in normal uterine myometrium (upper left, ×200), primary ULMS (top right, ×200), metastatic ULMS (bottom left, ×100; bottom right, ×200). Scale bars, 100 um (top left and right), 200 μm (bottom left), and 100 μm (bottom right). B, Overall survival of patients with high or low NRDC expression from TCGA sarcoma dataset analyzed at GEPIA. The median was used as cut-off. C–E, SK-UT1 and LMS cells 2 days after transfection with NC-siRNA or two NRDC-siRNAs for analysis. C, Cell numbers and representative images (×100) of SK-UT1 (left) and LMS (right) cells that migrated through a Boyden chamber (n = 3 for each; *, P < 0.05 vs. all). Scale bar, 200 μm. D, LMS cells transfected with NC-siRNA or two NRDC-siRNAs were seeded onto fibronectin or 1% BSA coated plates. Numbers and representative images (×100) of cells that remained after washing each plate with PBS are shown (n = 4 for each; *, P < 0.05 vs. all). Scale bar, 200 μm. E, Western blot analysis of protein expression of FAK, p-FAK Tyr397, and β-actin in cell lysates of LMS cells transfected with NC-siRNA or two NRDC-siRNAs. Quantified blot values via ImageJ are shown below the blot images. F and G, SK-UT1 and SKN cells 2 days after transfection with pCMV or pCMV-NRDC for analysis. F, SKN cells were seeded onto fibronectin or 1% BSA coated plates. Number of cells that remained after washing each plate with PBS are shown (n = 4 for each; *, P < 0.05 between the means of groups). G, The number of SK-UT1 (left) and SKN (right) cells that migrated through a Boyden chamber (n = 3 for each; *, P < 0.05 between the means of groups).

Figure 6.

Evidence indicating that NRDC is important for lung metastasis of human ULMS. A, IHC staining for NRDC in normal uterine myometrium (upper left, ×200), primary ULMS (top right, ×200), metastatic ULMS (bottom left, ×100; bottom right, ×200). Scale bars, 100 um (top left and right), 200 μm (bottom left), and 100 μm (bottom right). B, Overall survival of patients with high or low NRDC expression from TCGA sarcoma dataset analyzed at GEPIA. The median was used as cut-off. C–E, SK-UT1 and LMS cells 2 days after transfection with NC-siRNA or two NRDC-siRNAs for analysis. C, Cell numbers and representative images (×100) of SK-UT1 (left) and LMS (right) cells that migrated through a Boyden chamber (n = 3 for each; *, P < 0.05 vs. all). Scale bar, 200 μm. D, LMS cells transfected with NC-siRNA or two NRDC-siRNAs were seeded onto fibronectin or 1% BSA coated plates. Numbers and representative images (×100) of cells that remained after washing each plate with PBS are shown (n = 4 for each; *, P < 0.05 vs. all). Scale bar, 200 μm. E, Western blot analysis of protein expression of FAK, p-FAK Tyr397, and β-actin in cell lysates of LMS cells transfected with NC-siRNA or two NRDC-siRNAs. Quantified blot values via ImageJ are shown below the blot images. F and G, SK-UT1 and SKN cells 2 days after transfection with pCMV or pCMV-NRDC for analysis. F, SKN cells were seeded onto fibronectin or 1% BSA coated plates. Number of cells that remained after washing each plate with PBS are shown (n = 4 for each; *, P < 0.05 between the means of groups). G, The number of SK-UT1 (left) and SKN (right) cells that migrated through a Boyden chamber (n = 3 for each; *, P < 0.05 between the means of groups).

Close modal

To assess the potential of NRDC as a therapeutic target, we examined the phenotype of human ULMS cell lines following inhibition of NRDC expression. It has been reported that NRDC knockdown impairs cell proliferation/survival in gastric and breast cancer (26, 27), and invasion/migration in esophageal cancer (28). Unexpectedly, siRNA inhibition of NRDC expression in SK-UT1 and LMS cells did not affect cell proliferation (Supplementary Fig. S7). Therefore, we focused on cell migration and adhesion, because these two properties are essential for cancer metastasis. NRDC inhibition significantly suppressed the migration of SK-UT1 and LMS cells (Fig. 6C). Cell adhesion to fibronectin was also inhibited by NRDC knockdown (Fig. 6D) through reduced phosphorylation of FAK at Tyr397 (Fig. 6E). Next, to examine the oncogenic effect of NRDC in metastasis, SK-UT1 and LMS cells were transiently transfected with pCMV or pCMV-NRDC (Supplementary Fig. S8). NRDC overexpression significantly enhanced cell adhesion to fibronectin and migration capacity of SK-UT1 and LMS cells (Fig. 6F and G). Although it will be necessary to further confirm that NRDC drives metastasis in ULMS, these results suggest that NRDC inhibition could serve as a useful therapeutic target for suppressing metastasis in ULMS.

Here, we describe a unique mouse model that uses SB transposons to mutagenize the uterine smooth muscle of mice carrying a conditional PtenKO allele and an activated Kras allele, which develops ULMS at a high frequency within 3 months of age. In a previous human dataset for ULMS, loss of function of PTEN was a very common gene alteration (2, 29), while a few KRAS mutations were reported (2, 13). Therefore, our model does not fully mimic human ULMS. However, the uterine tumors that developed in our model exactly showed the pathological features of ULMS, and the CCGs identified in this screen were validated in human ULMS. Considering that mouse models of ULMS are quite rare, our PTEN/KRAS SB model is very meaningful and has the added advantage that the SB transposon insertion sites in these tumors provide molecular tags for identifying ULMS driver genes. Contrary to previous SB mutagenesis screens, including those for osteosarcoma (7, 8, 15, 16), a relatively small number of CCGs were identified in this screen. This is likely caused by rapid tumor growth, which shortens the time for transposon mobilization.

Among the 19 CCGs driving primary ULMS identified in the screen, five genes including Nf1, Aff1, Cblb, Arid1b, and Ext1 are known cancer genes, although their role in ULMS remains largely obscure. Among these five CCGs, Nf1 and Arid1b had SB insertions in both the sense and antisense strands, suggesting their function as tumor suppressor genes. The function of the other three genes is considered as indeterminate based on SB insertion pattern, probably due to fewer insertions. Neurofibromatosis type 1 (NF1) is a well-known tumor suppressor gene that functions as a negative regulator of Ras signaling. Mutations in NF1 are present at high frequency in neurofibromatosis type 1 and juvenile myelomonocytic leukemia. Among sarcomas, loss of NF1 has been reported in malignant peripheral nerve sheath tumors (30), myxofibrosarcoma and pleomorphic liposarcomas (31). Given that the mice used in this screen already contain an activating KRAS mutation, it is possible that mutations in NF1 enhance tumorigenesis by further deregulating Ras signaling. AFF1 (AF4/FMR2 family member 1) is a KMT2A fusion partner in various hematologic malignancies, and is reported to suppress CDKN1B transcription (32). Cbl proto-oncogene C (CBLC), encodes a member of the Cbl family of E3 ubiquitin ligases, involved in EGFR signaling. EGFR-isoform 1 is ubiquitinated and degraded by CBLC upon EGF activation, indicating that CBLC is a negative regulator of activated EGFR (33). However, mutations in CBLC leading to loss of E3 activity do not affect its adaptor protein functions or recruitment of signaling molecules to activated EGFR, suggesting an oncogenic role for CBLC in cancer (34). AT-Rich interaction domain 1B (ARID1B) is a member of the SWI/SNF chromatin remodeling complex. Mutations in ARID1B are associated with several types of cancer (35, 36), including synovial sarcoma (37). Exostosin glycosyltransferase 1 (EXT1) is a tumor suppressor involved in the biosynthesis of heparan sulfate, which is abundantly found as heparan sulfate proteoglycans (HSPG) in cell surfaces and the extracellular matrix (38). Mutations in EXT1, mostly resulting in loss of function, lead to hereditary multiple osteochondroma and osteosarcoma (39). Still, the function of EXT1 in cancer has been controversial. HSPGs widely effect several signaling pathways, including Wnt/Wg, TGFβ, and FGF pathways, and influence cellular proliferation and invasion (40). A tumor promoting role of EXT1 has been reported for a breast cancer cell line (41).

Among the CCGs that are not already known cancer genes, Zfp217 and Sfmbt2 were identified as candidate trunk drivers in human ULMS. Kruppel-type fingers protein 217 (ZNF217) contains C2H2-type zinc-fingers and is located at 20q13.2, a region that is frequently amplificated in human cancer, including breast cancer (42), ovarian clear cell adenocarcinoma (43), and other tumors (44, 45). ZNF217 is a component of the human histone deacetylase complex CoREST-HDAC, a transcriptional corepressor of C-terminal binding protein (CtBP1) complex 1, histone demethylases LSD1 and KDM5B/JARID1B/PLU-1, and G9a methyltranserases. ZNF217 is also associated with aggressive tumor behavior and poorer prognosis (44, 45), in addition to upregulating ERBB3 expression, leading to activation of AKT and MAPK pathways (46).

Scm-like with four Mbt domains 2 (SFMBT2) is a polycomb group protein that functions as a transcriptional repressor by binding to methylated lysines in histone tails and inducing formation of transcription-resistant higher-order chromatin structures at target genes. Sfmbt2 may promote proliferation by silencing tumor suppressors (47); however, its function in human cancers is mostly unknown. In prostate cancer, low expression levels of SFMBT2 might suppress metastatic capacity (48); however, high expression levels have been reported in thyroid cancer (49). In our screen, Sfmbt2 expression was significantly elevated in tumors with SB transposon insertions; however, its role in human ULMS is unclear.

Smooth muscle-specific PTEN-null mice develop smooth muscle hyperplasia and leiomyosarcoma in soft tissues with rapid onset and high incidence (12). PTEN loss induces the activation of AKT through mTOR activation, contributing to sarcomagenesis. An SB osteosarcoma transposon screen using Trp53 as a sensitizing mutation has also shown that loss of TP53 and PTEN is essential for driving osteosarcoma, and identified CCGs that are enriched in ERBB, PI3K–AKT–mTOR and MAPK signaling pathways (16). Similarly, recent genomic profiling of 80 ULMS cases revealed that the most common gene alterations were loss of function TP53, RB1, and ATRX mutations, and PTEN alterations were significantly found in metastatic tumors (29). Although TCGA studies for adult soft tissue sarcoma suggested that ULMS and soft tissue LMS are molecularly distinct, it did support the importance of the PI3K–AKT–mTOR pathway in leiomyosarcoma, and suggested the efficacy of its inhibitors for leiomyosarcomas (2). Consistent with these findings, STRING (50) analysis showed that nine of the 19 CCGs identified in our study connected to a network that included AKT1 and AKT2 (Supplementary Fig. S9). Our study also revealed a high expression of ZNF217 in human ULMS compared with normal myometrium, the oncogenic role of ZNF217 in human LMS cell lines, and an antitumor effect of ZNF217 inhibition in human ULMS cell lines via suppression of migration and stemness capacity. However, high ZNF217 expression did not significantly correlate with poor prognosis in a public dataset, probably because ZNF217 is not involved in the development of all human ULMS, considering that not all human ULMS show immunopositivity of ZNF217. Therefore, to establish novel treatments for human ULMS in which ZNF217 is involved, further research will be necessary to understand the exact mechanism of action of ZNF217 in the pathogenesis and initiation of ULMS.

In a second ex vivo SB screen for genes involved in lung metastasis of ULMS, we identified NRDC as a candidate lung metastasis gene. NRDC (nardilysin) encodes a zinc-dependent endopeptidase of the M16 family, which promotes ectodomain shedding of the precursor forms of various membrane proteins, including heparin-binding epidermal growth factor-like growth factor and TNFα (51). Following cleavage, the ectodomain domain is released from the cell surface as a soluble protein and regulates various signaling pathways involved in inflammation, neurodegeneration, and cancer. High expression of NRDC is correlated with poorer prognosis of esophageal, renal, and liver cancer (28). In gastric cancer (49), NRDC is expressed in the epithelium of cancer tissues, and its inhibition in gastric cancer cell lines impairs cell growth in vitro and in vivo (26). Similarly, NRDC is expressed in ductal carcinoma in situ of invasive ductal carcinoma, which is a common type of breast cancer, and its silencing suppresses cell growth/proliferation in breast cancer cell lines via cyclin D1 inhibition (27), suggesting that NRDC may function as an oncogene in gastric and breast cancer. In contrast, in esophageal squamous carcinoma, NRDC inhibition did not stimulate cell growth/proliferation, but instead impaired invasion in vitro via reduction of MMP2 and NMP3 (28). In this study, although the number of paired samples was small, some metastatic tumors shoed increased NRDC expression compared with paired primary tumors in the same patient. NRDC inhibition led to suppression of migration and adhesion capacity, suggesting that NRDC could be a therapeutic target in human ULMS. The future experiment will be necessary to investigate the mechanism of NRDC in ULMS metastasis and confirm that NRDC promotes metastasis.

In summary, SB mutagenesis has for the first time successfully generated an unbiased catalogue of CCGs driving primary and metastatic ULMS. Subsequent studies validated two of these CCGs, ZNF217 and NRDC, as potential therapeutic targets in ULMS, demonstrating the power of SB mutagenesis for defining new therapeutic targets for the treatment of ULMS.

T. Kodama reports personal fees from Eisai Co., Eli Lilly Co., Bayer, Chugai Pharmaceutical Co, and MSD outside the submitted work. T. Kimura reports personal fees from GE Healthcare Japan, Novel Pharma Co. Ltd., AstraZeneca Co. Ltd., Chugai Pharma Co. Ltd., Women's Health Japan Co., Ferring Pharma Co. Ltd., Nihon Kayaku Co. Ltd., Bayer Pharma Co. Ltd., Otska Pharma Co. Ltd., and Mochida Pharma Co. Ltd. outside the submitted work. N.A. Jenkins reports grants from CPRIT during the conduct of the study. N.G. Copeland reports grants from CPRIT during the conduct of the study. No disclosures were reported by the other authors.

M. Kodama: Data curation, formal analysis, funding acquisition, validation, investigation, visualization, writing–original draft. H. Shimura: Investigation, visualization, writing–review and editing. J.C. Tien: Investigation, methodology. J.Y. Newberg: Software, formal analysis. T. Kodama: software, supervision, writing–review and editing. Z. Wei: Investigation, visualization, writing-review and editing. R. Rangel: Investigation, writing–review and editing. K. Yoshihara: Software, writing–review and editing. A. Kuruma: Validation, investigation. A. Nakae: Investigation. K. Hashimoto: Resources, supervision. K. Sawada: Resources, supervision. T. Kimura: Resources, supervision. N.A. Jenkins: Conceptualization, resources, supervision, project administration. N.G. Copeland: Conceptualization, resources, writing–original draft, project administration.

The authors thank E. Freiter and H. Lee for monitoring mice and animal technical assistance, J. Zhang and A. Okamura for experimental assistance, S. Rico-Solitz and M. Matsui for secretarial work. This research was supported by the Cancer Prevention Research Institute of Texas (to N.G. Copeland and N.A. Jenkins) and the Astellas Foundation for Research on Metabolic Disorders (to M. Kodama) and JSPS KAKENHI [Grant Nos. JP 17K11277, JP20K09617, and 16H06279 (PAGS) for M. Kodama].

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
D'Angelo
E
,
Prat
J
. 
Uterine sarcomas: a review
.
Gynecol Oncol
2010
;
116
:
131
9
.
2.
Cancer Genome Atlas Research Network
. 
Comprehensive and integrated genomic characterization of adult soft tissue sarcomas
.
Cell
2017
;
171
:
950
65
3.
Shalem
O
,
Sanjana
NE
,
Hartenian
E
,
Shi
X
,
Scott
DA
,
Mikkelson
T
, et al
Genome-scale CRISPR-Cas9 knockout screening in human cells
.
Science
2014
;
343
:
84
7
.
4.
Luo
B
,
Cheung
HW
,
Subramanian
A
,
Sharifnia
T
,
Okamoto
M
,
Yang
X
, et al
Highly parallel identification of essential genes in cancer cells
.
Proc Natl Acad Sci U S A
2008
;
105
:
20380
5
.
5.
Chen
S
,
Sanjana
NE
,
Zheng
K
,
Shalem
O
,
Lee
K
,
Shi
X
, et al
Genome-wide CRISPR screen in a mouse model of tumor growth and metastasis
.
Cell
2015
;
160
:
1246
60
.
6.
Kodama
M
,
Kodama
T
,
Newberg
JY
,
Katayama
H
,
Kobayashi
M
,
Hanash
SM
, et al
In vivo loss-of-function screens identify KPNB1 as a new druggable oncogene in epithelial ovarian cancer
.
Proc Natl Acad Sci U S A
2017
;
114
:
E7301
E10
.
7.
Rangel
R
,
Lee
SC
,
Hon-Kim Ban
K
,
Guzman-Rojas
L
,
Mann
MB
,
Newberg
JY
, et al
Transposon mutagenesis identifies genes that cooperate with mutant Pten in breast cancer progression
.
Proc Natl Acad Sci U S A
2016
;
113
:
E7749
E58
.
8.
Kodama
T
,
Newberg
JY
,
Kodama
M
,
Rangel
R
,
Yoshihara
K
,
Tien
JC
, et al
Transposon mutagenesis identifies genes and cellular processes driving epithelial-mesenchymal transition in hepatocellular carcinoma
.
Proc Natl Acad Sci U S A
2016
;
113
:
E3384
93
.
9.
Kodama
T
,
Bard-Chapeau
EA
,
Newberg
JY
,
Kodama
M
,
Rangel
R
,
Yoshihara
K
, et al
Two-step forward genetic screen in mice identifies Ral GTPase-activating proteins as suppressors of hepatocellular carcinoma
.
Gastroenterology
2016
;
151
:
324
37
.
10.
Takeda
H
,
Wei
Z
,
Koso
H
,
Rust
AG
,
Yew
CC
,
Mann
MB
, et al
Transposon mutagenesis identifies genes and evolutionary forces driving gastrointestinal tract tumor progression
.
Nat Genet
2015
;
47
:
142
50
.
11.
Bard-Chapeau
EA
,
Nguyen
AT
,
Rust
AG
,
Sayadi
A
,
Lee
P
,
Chua
BQ
, et al
Transposon mutagenesis identifies genes driving hepatocellular carcinoma in a chronic hepatitis B mouse model
.
Nat Genet
2014
;
46
:
24
32
.
12.
Hernando
E
,
Charytonowicz
E
,
Dudas
ME
,
Menendez
S
,
Matushansky
I
,
Mills
J
, et al
The AKT-mTOR pathway plays a critical role in the development of leiomyosarcomas
.
Nat Med
2007
;
13
:
748
53
.
13.
Hill
MA
,
Gong
C
,
Casey
TJ
,
Menon
AG
,
Mera
R
,
Gillespie
AT
, et al
Detection of K-ras mutations in resected primary leiomyosarcoma
.
Cancer Epidemiol Biomarkers Prev
1997
;
6
:
1095
100
.
14.
Kodama
T
,
Yi
J
,
Newberg
JY
,
Tien
JC
,
Wu
H
,
Finegold
MJ
, et al
Molecular profiling of nonalcoholic fatty liver disease-associated hepatocellular carcinoma using SB transposon mutagenesis
.
Proc Natl Acad Sci U S A
2018
;
115
:
E10417
-E26.
15.
Mann
KM
,
Newberg
JY
,
Black
MA
,
Jones
DJ
,
Amaya-Manzanares
F
,
Guzman-Rojas
L
, et al
Analyzing tumor heterogeneity and driver genes in single myeloid leukemia cells with SBCapSeq
.
Nat Biotechnol
2016
;
34
:
962
72
.
16.
Moriarity
BS
,
Otto
GM
,
Rahrmann
EP
,
Rathe
SK
,
Wolf
NK
,
Weg
MT
, et al
A Sleeping Beauty forward genetic screen identifies new genes and pathways driving osteosarcoma development and metastasis
.
Nat Genet
2015
;
47
:
615
24
.
17.
Collier
LS
,
Carlson
CM
,
Ravimohan
S
,
Dupuy
AJ
,
Largaespada
DA
. 
Cancer gene discovery in solid tumours using transposon-based somatic mutagenesis in the mouse
.
Nature
2005
;
436
:
272
6
.
18.
Koso
H
,
Takeda
H
,
Yew
CC
,
Ward
JM
,
Nariai
N
,
Ueno
K
, et al
Transposon mutagenesis identifies genes that transform neural stem cells into glioma-initiating cells
.
Proc Natl Acad Sci U S A
2012
;
109
:
E2998
3007
.
19.
Dupuy
AJ
,
Rogers
LM
,
Kim
J
,
Nannapaneni
K
,
Starr
TK
,
Liu
P
, et al
A modified sleeping beauty transposon system that can be used to model a wide variety of human cancers in mice
.
Cancer Res
2009
;
69
:
8150
6
.
20.
Tang
Z
,
Li
C
,
Kang
B
,
Gao
G
,
Zhang
Z
. 
GEPIA: a web server for cancer and normal gene expression profiling and interactive analyses
.
Nucleic Acids Res
2017
;
45
:
W98
W102
.
21.
Xing
D
,
Scangas
G
,
Nitta
M
,
He
L
,
Xu
X
,
Ioffe
YJ
, et al
A role for BRCA1 in uterine leiomyosarcoma
.
Cancer Res
2009
;
69
:
8231
5
.
22.
Brett
BT
,
Berquam-Vrieze
KE
,
Nannapaneni
K
,
Huang
J
,
Scheetz
TE
,
Dupuy
AJ
, et al
Novel molecular and computational methods improve the accuracy of insertion site analysis in Sleeping Beauty-induced tumors
.
PLoS One
2011
;
6
:
e24668
.
23.
Tate
JG
,
Bamford
S
,
Jubb
HC
,
Sondka
Z
,
Beare
DM
,
Bindal
N
, et al
COSMIC: the Catalogue Of Somatic Mutations In Cancer
.
Nucleic Acids Res
2019
;
47
:
D941
D7
.
24.
March
HN
,
Rust
AG
,
Wright
NA
,
ten Hoeve
J
,
de Ridder
J
,
Eldridge
M
, et al
Insertional mutagenesis identifies multiple networks of cooperating genes driving intestinal tumorigenesis
.
Nat Genet
2011
;
43
:
1202
9
.
25.
Copeland
NG
,
Jenkins
NA
. 
Harnessing transposons for cancer gene discovery
.
Nat Rev Cancer
2010
;
10
:
696
706
.
26.
Kanda
K
,
Komekado
H
,
Sawabu
T
,
Ishizu
S
,
Nakanishi
Y
,
Nakatsuji
M
, et al
Nardilysin and ADAM proteases promote gastric cancer cell growth by activating intrinsic cytokine signalling via enhanced ectodomain shedding of TNF-α
.
EMBO Mol Med
2012
;
4
:
396
411
.
27.
Choong
LY
,
Lim
SK
,
Chen
Y
,
Loh
MC
,
Toy
W
,
Wong
CY
, et al
Elevated NRD1 metalloprotease expression plays a role in breast cancer growth and proliferation
.
Genes Chromosomes Cancer
2011
;
50
:
837
47
.
28.
Uraoka
N
,
Oue
N
,
Sakamoto
N
,
Sentani
K
,
Oo
HZ
,
Naito
Y
, et al
NRD1, which encodes nardilysin protein, promotes esophageal cancer cell invasion through induction of MMP2 and MMP3 expression
.
Cancer Sci
2014
;
105
:
134
40
.
29.
Hensley
ML
,
Chavan
SS
,
Solit
DB
,
Murali
R
,
Soslow
R
,
Chiang
S
, et al
Genomic landscape of uterine sarcomas defined through prospective clinical sequencing
.
Clin Cancer Res
2020
;
26
:
3881
8
.
30.
Ferner
RE
,
Gutmann
DH
. 
International consensus statement on malignant peripheral nerve sheath tumors in neurofibromatosis
.
Cancer Res
2002
;
62
:
1573
7
.
31.
Barretina
J
,
Taylor
BS
,
Banerji
S
,
Ramos
AH
,
Lagos-Quintana
M
,
Decarolis
PL
, et al
Subtype-specific genomic alterations define new targets for soft-tissue sarcoma therapy
.
Nat Genet
2010
;
42
:
715
21
.
32.
Bursen
A
,
Schwabe
K
,
Rüster
B
,
Henschler
R
,
Ruthardt
M
,
Dingermann
T
, et al
The AF4.MLL fusion protein is capable of inducing ALL in mice without requirement of MLL.AF4
.
Blood
2010
;
115
:
3570
9
.
33.
Mohapatra
B
,
Ahmad
G
,
Nadeau
S
,
Zutshi
N
,
An
W
,
Scheffe
S
, et al
Protein tyrosine kinase regulation by ubiquitination: critical roles of Cbl-family ubiquitin ligases
.
Biochim Biophys Acta
2013
;
1833
:
122
39
.
34.
Hong
SY
,
Kao
YR
,
Lee
TC
,
Wu
CW
. 
Upregulation of E3 ubiquitin ligase CBLC enhances EGFR dysregulation and signaling in lung adenocarcinoma
.
Cancer Res
2018
;
78
:
4984
96
.
35.
Coatham
M
,
Li
X
,
Karnezis
AN
,
Hoang
LN
,
Tessier-Cloutier
B
,
Meng
B
, et al
Concurrent ARID1A and ARID1B inactivation in endometrial and ovarian dedifferentiated carcinomas
.
Mod Pathol
2016
;
29
:
1586
93
.
36.
Cajuso
T
,
Hänninen
UA
,
Kondelin
J
,
Gylfe
AE
,
Tanskanen
T
,
Katainen
R
, et al
Exome sequencing reveals frequent inactivating mutations in ARID1A, ARID1B, ARID2 and ARID4A in microsatellite unstable colorectal cancer
.
Int J Cancer
2014
;
135
:
611
23
.
37.
Xing
Z
,
Wei
L
,
Jiang
X
,
Conroy
J
,
Glenn
S
,
Bshara
W
, et al
Analysis of mutations in primary and metastatic synovial sarcoma
.
Oncotarget
2018
;
9
:
36878
88
.
38.
Kim
SH
,
Turnbull
J
,
Guimond
S
. 
Extracellular matrix and cell signalling: the dynamic cooperation of integrin, proteoglycan and growth factor receptor
.
J Endocrinol
2011
;
209
:
139
51
.
39.
Jennes
I
,
Pedrini
E
,
Zuntini
M
,
Mordenti
M
,
Balkassmi
S
,
Asteggiano
CG
, et al
Multiple osteochondromas: mutation update and description of the multiple osteochondromas mutation database (MOdb)
.
Hum Mutat
2009
;
30
:
1620
7
.
40.
Knelson
EH
,
Nee
JC
,
Blobe
GC
. 
Heparan sulfate signaling in cancer
.
Trends Biochem Sci
2014
;
39
:
277
88
.
41.
Okolicsanyi
RK
,
van Wijnen
AJ
,
Cool
SM
,
Stein
GS
,
Griffiths
LR
,
Haupt
LM
, et al
. 
Heparan sulfate proteoglycans and human breast cancer epithelial cell tumorigenicity
.
J Cell Biochem
2014
;
115
:
967
76
.
42.
Vendrell
JA
,
Thollet
A
,
Nguyen
NT
,
Ghayad
SE
,
Vinot
S
,
Bièche
I
, et al
ZNF217 is a marker of poor prognosis in breast cancer that drives epithelial-mesenchymal transition and invasion
.
Cancer Res
2012
;
72
:
3593
606
.
43.
Li
J
,
Song
L
,
Qiu
Y
,
Yin
A
,
Zhong
M
. 
ZNF217 is associated with poor prognosis and enhances proliferation and metastasis in ovarian cancer
.
Int J Clin Exp Pathol
2014
;
7
:
3038
47
.
44.
Rooney
PH
,
Boonsong
A
,
McFadyen
MC
,
McLeod
HL
,
Cassidy
J
,
Curran
S
, et al
The candidate oncogene ZNF217 is frequently amplified in colon cancer
.
J Pathol
2004
;
204
:
282
8
.
45.
Ginzinger
DG
,
Godfrey
TE
,
Nigro
J
,
Moore
DH
,
Suzuki
S
,
Pallavicini
MG
, et al
Measurement of DNA copy number at microsatellite loci using quantitative PCR analysis
.
Cancer Res
2000
;
60
:
5405
9
.
46.
Krig
SR
,
Miller
JK
,
Frietze
S
,
Beckett
LA
,
Neve
RM
,
Farnham
PJ
, et al
ZNF217, a candidate breast cancer oncogene amplified at 20q13, regulates expression of the ErbB3 receptor tyrosine kinase in breast cancer cells
.
Oncogene
2010
;
29
:
5500
10
.
47.
Inoue
K
,
Hirose
M
,
Inoue
H
,
Hatanaka
Y
,
Honda
A
,
Hasegawa
A
, et al
The rodent-specific MicroRNA cluster within the Sfmbt2 gene is imprinted and essential for placental development
.
Cell Rep
2017
;
19
:
949
56
.
48.
Gwak
J
,
Shin
JY
,
Lee
K
,
Hong
SK
,
Oh
S
,
Goh
SH
, et al
SFMBT2 (Scm-like with four mbt domains 2) negatively regulates cell migration and invasion in prostate cancer cells
.
Oncotarget
2016
;
7
:
48250
64
.
49.
Uhlen
M
,
Zhang
C
,
Lee
S
,
Sjöstedt
E
,
Fagerberg
L
,
Bidkhori
G
, et al
A pathology atlas of the human cancer transcriptome
.
Science
2017
;
357
:
eaan2507
.
50.
Szklarczyk
D
,
Gable
AL
,
Lyon
D
,
Junge
A
,
Wyder
S
,
Huerta-Cepas
J
, et al
STRING v11: protein-protein association networks with increased coverage, supporting functional discovery in genome-wide experimental datasets
.
Nucleic Acids Res
2019
;
47
:
D607
D13
.
51.
Nishi
E
,
Hiraoka
Y
,
Yoshida
K
,
Okawa
K
,
Kita
T
. 
Nardilysin enhances ectodomain shedding of heparin-binding epidermal growth factor-like growth factor through activation of tumor necrosis factor-alpha-converting enzyme
.
J Biol Chem
2006
;
281
:
31164
72
.