FLI1 Exonic Circular Rnas As a Novel Oncogenic Driver to Promote Tumor

Published OnlineFirst November 14, 2018; DOI: 10.1158/1078-0432.CCR-18-1447

Translational Cancer Mechanisms and Therapy Clinical Cancer Research FLI1 Exonic Circular RNAs as a Novel Oncogenic Driver to Promote Tumor Metastasis in Small Cell Lung Cancer Lingyu Li1, Wei Li1, Naifei Chen1,2, Haixin Zhao3, Guang Xu3, Yijing Zhao1,2, Xin Pan3, Xiaoying Zhang1, Lei Zhou1, Dehai Yu1, Ailing Li3, Ji-Fan Hu1,2, and Jiuwei Cui1

Abstract

Purpose: The aberrantly upregulated Friend leukemia Results: Therapeutic comparison of CRISPR Cas9 knockout virus integration 1 (FLI1) is closely correlated with the and shRNA knockdown of FLI1 identified FECRs as a new malignant phenotype of small cell lung cancer (SCLC). It noncanonical malignant driver in SCLC. Using RNA FISH and is interesting to note that the CRISPR gene knockout by quantitative PCR, we found that FECR1 (exons 4-2-3) and Cas9 gRNAs that target the FLI1 coding region and the FECR2 (exons 5-2-3-4) were aberrantly upregulated in SCLC posttranscriptional knockdown by shRNAs that target the tissues (P < 0.0001), and was positively associated with lymph 30 region of FLI1 mRNA yielded distinct antimetastasis node metastasis (P < 0.01). Notably, serum exosomal FECR1 effects in SCLC cells. This study attempts to examine if was associated with poor survival (P ¼ 0.038) and clinical FLI1 exonic circular RNAs (FECR) function as a new response to chemotherapy. Silencing of FECRs significantly malignant driver that determines the metastatic pheno- inhibited the migration in two highly aggressive SCLC cell type in SCLC. lines and reduced tumor metastasis in vivo. Mechanistically, we Experimental Design: The clinical relevance of FECRs uncovered that FECRs sequestered and subsequently inacti- was examined in 56 primary SCLC tissues and 50 non– vated tumor suppressor miR584-3p, leading to the activation small cell lung cancer (NSCLC) tissues. The prognostic value of the Rho Associated Coiled-Coil Containing Protein Kinase 1 of FECRs was examined by measuring serum exosomal gene (ROCK1). FECRs in a longitudinal cohort of patients with SCLC. The Conclusions: This study identifies FLI1 exonic circular oncogenic activity of FECRs was investigated in both SCLC RNAs as a new oncogenic driver that promotes tumor metas- cell lines and animal xenograft studies. Finally, we explored tasis through the miR584–ROCK1 pathway. Importantly, the molecular mechanisms underlying these noncoding serum exosomal FECR1 may serve as a promising biomarker RNAs as a malignant driver. to track disease progression of SCLC.

Introduction Mortality of SCLC is high, particularly when patients are diag- nosed in limited disease stage, with a 5-year survival of only Primary lung cancer is one of the leading cause of cancer-related 10% to 26% (3). Molecular therapies, although being success- death in the world (1). Small cell lung cancer (SCLC), accounting ful in the treatment of various solid malignancies, often fail in for approximately 15% of all lung cancers, is a rapidly progressive SCLC (4–6). Furthermore, lack of relatively specificbiomarkers disease (PD; refs. 1, 2). Despite its initial response to the first-line makes it more hardly to manage the treatment strategy and chemotherapy, the recurrence arises rapidly in majority of cases. evaluate prognosis of patients with SCLC. Thus, cancer research with focus on pathologic factors and specificprognosticbio- 1Cancer Center, The First Hospital of Jilin University, Changchun, China. markers involved in SCLC is urgently needed to bring break- 2Stanford University Medical School, Palo Alto Veterans Institute for Research, throughs for developing novel therapies and monitoring the Palo Alto, California. 3Institute of Basic Medical Sciences, National Center of progression of SCLC. Biomedical Analysis, Beijing, China. Bi-allelic inactivation of TP53 and RB1 occurs in nearly all the Note: Supplementary data for this article are available at Clinical Cancer SCLC tumors, and loss of these two tumor suppressors are Research Online (http://clincancerres.aacrjournals.org/). obligatory in the tumorigenesis and development of SCLC (7). fi A. Li, J.-F. Hu, and J. Cui are senior authors for this communication. However, these genetic lesions alone are not suf cient to initiate tumor formation as demonstrated in the model of mouse ery- Corresponding Authors: Jiuwei Cui, Cancer Center, First Hospital, Jilin Univer- sity, 71 Xinmin Street, Changchun 130021, China. Phone: 86-431-8878-2178; throleukemia (8). Thus, in addition to the loss of TP53 and RB1, E-mail: [email protected]; Ji-Fan Hu, Stanford University Medical School, 3801 there must be some molecular regulators that play a vital role in Miranda Avenue, Palo Alto, CA 94304. Phone: 650-493-5000 x63175; Fax: 650- the tumorigenesis of SCLC. Quite recently, we demonstrated that 849-1213; E-mail: [email protected]; and Ailing Li, Institute of Basic Medical Friend leukemia virus integration 1 (FLI1) was aberrantly Sciences, National Center of Biomedical Analysis, Beijing, China. Phone: 86-10- expressed in TP53 and RB1-deficient SCLC cells and tissues, and 66930169; Fax: 86-10-68246161; E-mail: [email protected] was closely related to the malignant phenotype (9). FLI1, an ETS doi: 10.1158/1078-0432.CCR-18-1447 transcription factor family member, was initially identified as an 2018 American Association for Cancer Research. oncogene in F-MuLV-induced mouse erythroleukemia (10). FLI1

www.aacrjournals.org OF1

Li et al.

Translational Relevance Materials and Methods Small cell lung cancer (SCLC) is the most devastative type of Parents and samples human lung malignancies. Despite advances in chemothera- The study protocol was conducted in accordance with the py, the prognosis of SCLC remains poor. In this study, we principles of the Declaration of Helsinki and was approved identified FLI1 exonic circRNAs (FECR) as an oncogenic player by the Research Ethics Board of the First Hospital of Jilin that promotes tumor metastasis. FECRs were aberrantly University. Informed consent was obtained from each partic- expressed in SCLC tissues and were positively related with ipant in the study. Clinical data related to disease stage and tumor metastasis. Using SCLC in vitro and in vivo studies, we histological grade were available for all the patients. The revealed that FECRs promoted tumor metastasis through the pathologic diagnosis was made in accordance with the histo- fi miR584–ROCK1 pathway. Our study suggests that the FLI1 logic classi cation of tumors from the World Health Organi- fi gene may play a dual role in the pathogenesis of SCLC by zation. Tumor stage was de ned according to American Joint encoding FLI1 oncoprotein and generating FECRs. Important- Committee on Cancer/International Union Against Cancer fi ly, we found that serum exosomal FECR1 was associated with classi cation system. fi fi poor tumor survival and clinical response to chemotherapy, Formalin- xed, paraf n-embedded tissues of lung cancer sam- serving as a biomarker to track disease progression of SCLC. ples were obtained from surgical resection in First Hospital of Jilin University between January 2014 and November 2016. Tumors were histologically graded according to the Elston and Ellis method. The clinical characteristics of patients with SCLC and NSCLC were shown in Supplementary Table S1. Serum samples is preferentially expressed in hematopoietic cells and tissues, were collected from cancer center of the First Hospital of Jilin endothelial cells and fibroblasts (11, 12) and acts as a major University between January 2017 and September 2017 for driver of hematological malignancies (13). Recently, we and patients with SCLC, NSCLC, breast cancer, colon cancer, gastric others found that FLI1 was also highly expressed in some solid cancer, liver cancer, pancreatic cancer, and prostate cancer. To tumors (14–16), including SCLC. Using immunohistochemical determine the role of FECR1 expression in SCLC clinical stages, we staining, we recently found that FLI1 was significantly upregulated increased the sample size of patients with extensive SCLC. A total in SCLC tissues. The expression score of FLI1 oncoprotein was of 35 patients with limited SCLC and 26 patients with extensive associated with the extensive stage of SCLC and the overexpressed SCLC were included in the study. As the control, we also collected Ki67. Knockdown of FLI1 promoted apoptosis and induced serum samples from normal subjects who had no definite tumor repression of cell proliferation, tumor colony formation, and in diseases and severe lung diseases. Clinical characteristics of vivo tumorigenicity in highly aggressive SCLC cell lines (9). patients with SCLC and normal subjects were shown in Supple- However, little is known about the mechanisms underlying the mentary Table S2. aberrant activation of FLI1 in SCLC. To correlate serum exosomal FECRs levels with therapy The covalently closed circular RNAs (circRNA), as a new class of response, we also collected blood samples from a cohort of noncoding RNA, are produced from pre-mRNA splicing in thou- SCLC patients who received etoposide (VP-16), in combination sands of genes in our genome (17, 18). These circRNAs are mainly with cis-platinum or carboplatin, from the start of treatment to exonic or intronic, and are generated by back splicing or lariat documented progression based on CT imaging. Blood was drawn introns (19). In a study to search factors that upregulate FLI1 in at every 3 courses during treatment. breast cancer, Chen and colleagues (20) in our group utilized a Cas9-guided immunoprecipitation approach to pull down the Cell culture FLI1 promoter chromatin complex and identified the presence of All lung cancer cell lines were obtained from ATCC, includ- a FLI1 exonic circular RNA (FECR1) derived by back splicing from ing NCI-H446, NCI-H1688, NCI-H460, NCI-H1299, A549, FLI1 exons 4-2-3. Notably, FECR1 bound to the FLI1 promoter in PC-9, Calu-1, HCC-827, NCI-H2170, and SK-MES-1). Cell lines cis and recruited TET1, a demethylase that is actively involved in NCI-H446, NCI-H460, NCI-H1688, NCI-H2170, NCI-H1299, DNA demethylation. In breast cancer, FECR1 functions as a and HCC-827 were cultured in RPMI1640. The remaining positive feedback regulator to activate FLI1 by coordinating DNA cell lines, including A549, PC-9, Calu-1, SK-MES-1, and methylation. However, the role of FLI1 circular RNAs in SCLC HEK-293, were maintained in DMEM supplemented with remains unknown. 10% FBS (Hyclone) and 1% penicillin/streptomycin at 37C In this study, we were surprised to note that gene knockout with 5% CO2. of FLI1 by CRISPR Cas9 gRNAs inhibited metastasis, but shRNA knockdown of FLI1 failedtodoso,eventhoughthe Knockdown of FLI1 by shRNA production of FLI1 oncoproteins was equally blocked by both To study the role of FLI1 in SCLC cells, we used two shRNAs technologies in SCLC cells. Coincidently, the CRISPR gRNAs to knockdown FLI1 following the method as previously were designed to target the coding region where FECRs are described (9). Briefly, two short hairpin RNAs (shRNA) against formed. We thus hypothesized to examine if FECRs may be a the 30 region of FLI1 mRNA were inserted into the lentiviral previously unidentified metastatic driver in SCLC. The relation vector (Addgene, plasmid #8453; Supplementary Fig. S1). The with the pathogenesis of SCLC was explored by monitoring the shFLI1 1# sequence was 50-CGTCATGTTCTGGTTTGAGAT-30 expression in clinical tumor tissues and in serum exosomes. andshFLI12#sequencewas50 GCACAAACGATCAGTAA- Finally, using SCLC in vitro and in vivo models, we studied the GAAT-30.Afterconfirmation by DNA sequencing, the lenti- molecular mechanisms underlying the role of FECRs as a viruses were packaged in HEK-293 cells using lipofectamine tumor metastasis driver. 2000 (Invitrogen). The virus-containing supernatants were

OF2 Clin Cancer Res; 2018 Clinical Cancer Research

Oncogenic FLI1 circRNAs in SCLC

collected and concentrated with Centrifugal Filter Units RNA preparation and RT-qPCR (Amicon Ultra-15; Millipore). SCLC cells in six-well plates were The nuclear and cytoplasmic fractions were extracted using infected with lentiviruses using polybrene (8 mg/mL). Three NE-PER Nuclear and Cytoplasmic Extraction Reagents (Thermo days after infection, SCLC cells were selected by puromycin. Scientific) with RNase inhibitor (Thermo Scientific). Total RNA After puromycin selection, mixed stable cells were collected for from whole-cell lysates or the nuclear and cytoplasmic fractions each shRNA group and were used for gene analysis by Western were isolated using Trizol (Life Technologies). For RNase R blot analysis. treatment, 1 mg of total RNA was incubated 1 hour at 37C with or without 2U/mg of RNase R (Epicentre Technologies), to hydro- FLI1 knockout by CRISPR Cas9 gene editing lyze linear RNAs. For comparison, we also used the CRISPR/Cas9 gene editing To quantify the amount of mature miRNA, we used miRcute system to knockout the FLI1 gene using the method as previously miRNA Isolation Kit (Tiangen Biotech) and small nuclear U6B described (21, 22). Two single-guide RNAs (sgRNA), designed (RNU6B) RNA as an internal standard. To quantify the amount of to target the exon 2 and exon 3 region, were inserted into the mRNA, cDNA was synthesized with the Prime Script RT Master lentiCRISPR V2 vector (Addgene, plasmid #52961; Supplemen- Mix (Takara) from 500 ng RNA. The real-time PCR analyses were tary Fig. S2; ref. 23). Lentiviruses were packaged in HEK-293 cells performed using SYBR Premix Ex Taq II (Takara). The relative with packaging plasmids VSVG and Pax2. After lentiviral expression of each gene was quantitated using Biosystems 7300 infection, SCLC cells were selected by puromycin for 4 weeks. Real-Time PCR system and calculated on the basis of CT values Single-cell clones were collected to determine the efficiency of against an internal standard curve for each specific set of primers. FLI1 knockout by Western blot analysis. We repeated the assay The data were normalized over the value of b-actin control. The twice and two representative cell clones (Cas9 FLI1 #1 and primers are listed in Supplementary Table S4. Cas9 FLI1 #2), one from each independent assay, were used for To confirm the presence of FECRs as circular RNAs, the RNA subsequent studies. samples were first treated with RNase R to remove the linear RNA, converted into cDNA by reverse transcription, amplified by FECR- Specific targeting of FECRs by shRNA knockdown specific PCR primers, cloned in P-JET vector, and sequenced for To specifically target FECRs, we designed two shRNAs that the presence of the exon 4-exon 2 back splicing site in FECR1 and target the back-splicing sites of FECR1 and FECR2, while saving the exon 5-exon 2 back splicing site in FECR2. the FLI1 linear mRNA. The shFECR1 sequence was 50-CCTCA- GGGAAAGAGGCTCTGTC-30 and shFECR2 sequence was 50- MiRNA mimics and transfection TGTCAAAGAAGGAGGCTCTGTC-30. Two shRNAs were ligated HEK-293 cells were plated in 12-well plates at 5 104 per well. in a lentiviral vector (Supplementary Fig. S3) and lentiviruses were Twenty-four hours after plating, 40 pmol of miRNA mimics packaged in HEK-293 cells. After lentiviral infection, SCLC cells (Ribobio) were transfected to the cells with Lipofectamine RNAi- were selected by puromycin. Mixed stable cells were collected for Max (Invitrogen) following the manufacturer protocol. A random gene analysis. miRNA (Ribobio) was also transfected into the cells as a negative control (24). Overexpression of FECRs To recapitulate circRNA, Ad-FECR1 and Ad-FECR2 were RNA FISH obtained by cloning PCR-amplified Exon 2 to Exon 4 and Exon RNA in situ hybridization was performed using specific probes 2 to Exon 5 of the human FLI1 gene, respectively, and then for the back-splice region of FECR1 and FECR2 sequences. The inserted into the pLCDH-ciR vector (Geneseed Biotech). The specific fluorescent probes of FECR1, FECR2, and miR584-3p vector pLCDH-ciR was double digested with EcoRI and BamH1. were obtained from Ribobio Biotech. NCI-H446 and NCI- Luciferase reporters with the inclusion of FECR1 and FECR2 H1688 cells were grown to the exponential phase and were sequence were constructed by subcloning human FECR1 (exon 50% to 60% confluent at the time of fixation (4% paraformal- 3-exon 4-exon 2) and FECR2 (exon 3-exon 4-exon 5-exon 2) dehyde). After prehybridization (1 PBS/0.5% Triton X-100), region directly into the downstream of a cytomegalovirus cells were hybridized in hybridization buffer with 20 nmol/L promoter-driven firefly luciferase (Luc-FECR1 and Luc-FECR2) probes specific to FECRs at 37C overnight. The double FISH assay cassette in the pCDNA3.0 vector. was performed in NCI-H446 cells with FECR1 (or FECR2) and miR584-3p. Cy3-labeled probes specific to FECR1 (or FECR2) and Isolation and purification of serum exosomes Dig-labeled locked nucleic acid miR584-3p probes (Ribobio Whole blood was collected in red-topped tubes (BD). Blood Biotech) were used in the hybridization. The signals of was clotted by leaving it undisturbed at room temperature for Dig-labeled locked nucleic acid miR584-3p probe were detected 30 minutes. The clot was removed by centrifugation at 1,000 g using tyramide-conjugated Alexa Fluor 488 fluorochrome TSA Kit for 10 minutes in a refrigerated centrifuge. The serum in the upper (Life Technologies). For nuclear counterstaining, cells were supernatant was transferred immediately into a clean tube for incubated with 4,6-diamidino-2-phenylindole (DAPI; 1:100) for circulating exosomes assays. Exosomes were purified by affinity 15 minutes. The images were acquired on a Leica SP5 confocal chromatography using the ExoEasy Maxi Kit (Qiagen). The iso- microscope (Leica Micosystems). lated exosomes were dropped onto a piece of copper grid and stood for 1 minute. Then, a drop of 2% uranyl acetate was Western immunoblotting deposited on to the copper grid and stood for 1 minute. After As previously described (9), cells were lysed in radioimmuno- being dried for 10 minutes, exosomes were examined under the precipitation assay buffer (5 mmol/L Tris pH 7.4, 1% NP-40, transmission electron microscope (FEI TecnaiG2 Spirit; Thermo 0.15 M NaCl, 0.1% SDS, plus protease inhibitor cocktail, and Fisher Scientific). 1 mmol/L phenylmethylsulfonyl fluoride). Equal amounts of

www.aacrjournals.org Clin Cancer Res; 2018 OF3

Li et al.

protein were resolved by SDS-PAGE and subjected to Western blot toxylin and eosin (H&E) staining and histopathologic analyses, analysis using enhanced chemiluminescence (Pierce). Antibodies and the number of metastatic foci in the lung of nude mice was to FLI1 and b-actin were obtained from Abcam. calculated. The other half of tumor-bearing mice (n ¼ 5 per group) were followed until the mice died, and the overall survival of each Cell proliferation assay individual mouse was recorded. All experimental procedures Cell proliferation assay was performed following the method as involving animals were in accordance with the Guide for the Care previously described (9). NCI-H446 and NCI-H1688 cells were and Use of Laboratory Animals and approved by the Ethics seeded in 96-well plates with density 20% per well. The number of Committee of the First Hospital of Jilin University. cell proliferation was measured by MTT assay from day 1 to day 4. For staining, 100 mL of 0.5 mg/mL MTT was added to each well Luciferase reporter assay and cultured in 37C incubator for 4 hours. Then, 100 mL DMSO Cells were seeded in 48-well plates at a density of 1 104 was added to each well after removing the media carefully. The HEK-293 cells per well 24 hours before transfection. The cells were absorbance was read at 490 nm. The experiment was repeated cotransfected with a mixture of 100 ng FL reporter vectors, 10 ng three times. Renilla luciferase (RL) reporter vectors (pRL-TK), and 6 hours later, miRNA mimics were also transfected into cells at the Cell cycle and apoptosis analysis by flow cytometry indicated concentration. The 11 miRNA mimics were obtained NCI-H446 and NCI-H1688 cells were fixed in absolute ethyl from Ribobio Biotech. After 48 hours, the luciferase activity was alcohol at 20C overnight, washed twice with PBS, and resus- measured with a dual luciferase reporter assay system (Promega). pended in propidium iodide (PI) staining solution containing In the luciferase screening assay, we used one internal control 0.1 mg/mL RNase A, 50 mg/mL PI, and 0.2 % Triton. Cell-cycle (RL reporter), one negative control (control microRNA, NC) and distribution was analyzed using bivariate flow cytometry on a one positive control (siRNA targeted the exon 2 of FECR, siFECR). FACSCalibur (BD). FlowJo software (TreeStar) was used for cell- Each miRNA mimic, NC or siFECR was cotransfected with RL cycle position using the cell-cycle algorithm (9). reporter and FL reporter. For comparison, the FL activity was first To detect cell apoptosis, NCI-H446 and NCI-H1688 cells were normalized with RL activity. Finally, the fold change was calcu- seeded in 12-well plates with density 20% per well, resuspended lated by each miRNA mimics compared with NC. in 500 mL Annexin V binding buffer, and incubated with 5 mL Annexin V-FITC for 10 minutes and 5 mL PI for 15 minutes, following the manufacture's instruction by flow cytometry (9). Results FLI1 shRNA knockdown and CRISPR Cas9 knockout yield Colony formation assay distinct antitumor activities in SCLC cells Cells were seeded in six-well culture plates at a density of 3 FLI1 is aberrantly upregulated in many solid malignancies, 102 cells/2 mL/well and incubated for 15 days when colonies were including SCLC. To study the role of FLI1 in SCLC, we used both visible. Crystal violet staining was performed and the number of shRNA and CRISPR Cas9-gRNA technologies to knockdown FLI1 colonies was counted. (Fig. 1A). Using the Western blot analysis, we found that both approaches were able to efficiently block the production of FLI1 Cell migration oncoprotein in treated cells (Fig. 1B). Transwell assays were used to measure cell migration capability Surprisingly, we found that both technologies yielded quite using a modified 24-well Boyden chamber with a membrane. distinct impacts on the malignant phenotype of SCLC cells. About 2 104 cells in 200 mL 1,640 without FBS were placed into Knockdown of FLI1 by both technologies was able to inhibit cell the upper chambers. The lower chambers were filled with 600 mL proliferation in NCI-H446 SCLC cells (Fig. 1C and D). However, complete medium with 10% FBS as a chemo-attractant stimulus. they had dramatically distinct effects on cell migration. FLI1 Cas9 After incubation for 24 hours at 37C, noninvading cells were knockout induced a significant inhibition of cell migration, removed from the top of the chamber with a cotton swab. whereas shRNA knockdown showed very little effect in treated Migrated cells on the bottom surface of the filter were fixed, cells (Fig. 1E and F). Clearly, in addition to FLI1 oncoprotein, stained with 0.5% crystal violet, and counted in five random there was a previously-unidentified oncogenic component that fields under a microscope and the average number of five fields was present in the shRNA knockdown cells, but was absent in the was calculated. All assays were performed in triplicate and repeat- Cas9 knockout cells. ed three times. Identification of FLI1 circular RNAs as new oncogenic driver in Animal studies SCLC cells Male BALB/c nude mice were obtained from Charles River To solve this puzzle, we compared the regions targeted by these Laboratories (Beijing, China) and maintained in sterile laminar two approaches (Fig. 1A; Supplementary Figs. S1 and S2). The 0 flow cabinets. A total of 2 106 of NCI-H446 cells (shNT, shRNA was designed to target the 3 -region of FLI1 mRNA, shFECR1, and shFECR2) were tail vein-injected into the mice whereas the Cas9-gRNA targeted the coding region downstream (n ¼ 10 per group, 4–6 weeks of age), respectively. Eight to ten of the translation initiation codon (ATG), including exons 2 and weeks after injection, whole-body fluorescence imaging using a 3. Thus, the unidentified oncogenic component, if any, should be fluorescence scanning system (Fluorimager; Molecular Dynam- present in the regions that are differentially targeted by these two ics). Then, scanned images were analyzed using computerized approaches. image analysis system Image Quant (Molecular Dynamics). Half In a recent study to search factors that upregulate FLI1 in breast of tumor-bearing mice (n ¼ 5 per group) were sacrificed 10 weeks cancer, Chen and colleagues (20) in our lab utilized a Cas9 after injection. Lungs and livers were collected for further hema- immunoprecipitation assay to pull down the FLI1 promoter

OF4 Clin Cancer Res; 2018 Clinical Cancer Research

Oncogenic FLI1 circRNAs in SCLC

Figure 1. FLI1 shRNA-knockdown and CRISPR Cas9-knockout have distinct impacts on phenotypes of SCLC cells. A, Schematic location of two CRISPR Cas9 gRNAs and shRNAs that target FLI1.30-UTR, untranslated region; Cas9, CRISPR Cas9; gRNA, Cas9 guiding RNAs that target the FLI1 exons 2-3 (downstream of the transcription initiation site); shFLI1, shRNAs that target the 30 region of FLI1 mRNA. B, Western blot analysis of FLI1 oncoprotein in NCI-H446 cells. FLI1 was efficiently inhibited by two shRNAs (shFLI1 #1, shFLI1 #2) in mixed cells and by CRISPR Cas9 gRNAs in two cell clones (Cas9-FLI1 #1, Cas9-FLI1 #2) collected from two independent assays. C, Cell proliferation in FLI1 shRNA-knockdown NCI-H446 cells. Cell proliferation is measured by the MTT assay. , P < 0.05 as compared with the control group. D, Cell proliferation in Cas9-gRNA knockout NCI-H446 cells. , P < 0.05 as compared with the control group. E, Cell migration in FLI1 shRNA-knockdown NCI-H446 cells. Cell migration was examined by Transwell assay. Note that knockdown FLI1 using 30-UTR shRNAs did not affect migration of NCI-H446 cells (P > 0.05). F, Cell migration in Cas9-gRNA knockout NCI-H446 cells. Note that knockout FLI1 using CRISPR Cas9 significantly inhibited migration of NCI-H446 cells (, P < 0.01). chromatin complex. DNA sequencing identified a FECR1 as a activities derived from the shRNA and Cas9 approaches. We component in the FLI1 promoter complex (Supplementary reasoned that gene knockout of FLI1 by Cas9 gRNAs not only Fig. S4A). FECR1 functions as a positive feedback regulator and interrupted the coding of FLI1 oncoprotein, but also affected the activates FLI1 in breast cancer by coordinating DNA methylating formation of FECRs. However, the posttranscriptional knock- and demethylating enzymes. Coincidently, FECR1 is produced by down by shRNAs that target the FLI1 30 region did not affect the back splicing from FLI1 exons 4-2-3 (Supplementary Fig. S4B), formation of FECRs in SCLC cells, as FECRs did not contain the where the two FLI1 knockout Cas9 gRNAs are located (Supple- FLI1 30 region. mentary Fig. S2). We thus hypothesized that FECR1 and its family To test this hypothesis, we used primers to examine FLI1 circular RNAs might contribute to the distinct antimetastasis circRNA candidates from the region where the Cas9 gRNAs are

www.aacrjournals.org Clin Cancer Res; 2018 OF5

Li et al.

Figure 2. Identification of FECR1 and FECR2 in SCLC cell lines. A, Schematic illustration of FLI1 circRNA candidates. B, Expression of FLI1 circRNA family candidates in a panel of SCLC and NSCLC cell lines. C, Expression of FLI1 circRNA candidates in NCI-H446 and NCI-H1688 cells. D, Validation of FECR1 and FECR2 as circRNAs by RNA sequencing. RNA samples were treated with RNase R to remove the linear RNA. After RNase R digestion, FECRs were amplified by FECR-specific PCR primers and cloned in P-JET vector for sequencing. Note the presence of the exon 4-exon 2 splicing sites in FECR1 and the exon 5-exon 2 splicing site in FECR2. E, RNA fluorescence in situ hybridization (FISH) for FECR1 and FECR2. Nuclei were stained with DAPI. Scale bar, 5 mm. F, Cellular location of FECR1 and FECR2 in NCI-H446 cells. The qPCR data of FECRs in the nucleus were normalized to the value in the cytoplasm.

targeting. Based on the gene structure, it was estimated that FLI1 considerably high levels of FECR candidates. Non–small cell lung might generate a total of six circRNA candidates, including FLI1 cancer (NSCLC) cell lines, however, expressed these FECR candi- exons 4-2-3 (FECR1), exons 5-2-3-4 (FECR2), exons 3-2 (FECR3), dates at low levels, including NCI-H460, A549, PC-9, NCI-H1299, exons 6-2-3-4-5 (FECR4), exons 5-3-4 (FECR5), and exons 4-3 NCI-H2170, HCC-827, Calu-1, and SK-MES-1. Among the FECR (FECR6; Fig. 2A). We used qPCR to examine the expression of families, we found that the expression of FECR1 and FECR2 these FECR candidates in lung cancer cell lines. Among the cell was signiﬁcantly higher than that of other FECR candidates in lines tested, NCI-H446 and NCI-H1688 cell lines expressed NCI-H446 and NCI-H1688 SCLC cell lines (Fig. 2B and C). Thus,

OF6 Clin Cancer Res; 2018 Clinical Cancer Research

Oncogenic FLI1 circRNAs in SCLC

Figure 3. FECRs are correlated with the metastasis status in patients with SCLC. A, Representative images of FECR RNA-FISH and FLI1 protein IHC staining in primitive SCLC, lung squamous cancer, and lung adenocarcinoma, respectively. B, Quantitation of FECR1 FISH in NSCLC and SCLC tissues. C, Quantitation of FECR2 FISH in NSCLC and SCLC tissues. D and E, Mean fluorescence intensity of FECR1 (D) and FECR2 (E) in patients with SCLC with no metastasis and lymph node metastasis. , P < 0.01. these two FECRs might contribute to the distinct antitumor FLI1 oncoprotein encoded by the linear mRNA product was also activities from the shRNA and Cas9 approaches. In support to examined by histoimmunochemical staining. this hypothesis, we found that these two FECRs were down- Interestingly, we found that FECR1 and FECR2 were highly regulated by Cas9-gRNAs that target FLI1 exons 2-3, but were expressed in tissue samples of SCLC tumor samples (Fig. 3A). not be affected by the shRNAs that target the 30-region of FLI1 Yet, they were either negative or expressed at negligible (Supplementary Fig. S5). levels in NSCLC tissues (squamous carcinoma and adenocar- To confirm the presence of FECRs as circular RNAs, we treated cinoma). There was a good correlation between these two RNA samples with RNase R to remove the linear RNA. After RNase circRNAs and FLI1 oncoprotein. The FISH assay showed R digestion, the expression of the circular RNA was amplified by that both FECRs were generally located in the cytoplasm of FECR-specific PCR primers and further confirmed by sequencing. SCLC tissues with variable intensities. Quantitation of the Using RNase R-sequencing, we validated the exon 4-2 back fluorescence intensity also showed a significant difference in splicing site in FECR1 and the exon 5-2 back splicing site in expression between SCLC and NSCLC tissues (Fig. 3B and C, FECR2 (Fig. 2D). P < 0.0001). We then used RNA FISH to examine the location of FECRs in To further examine the role of FECRs, we grouped SCLC SCLC cell lines. Notably, we found that both two FECRs were patients based on the status of metastasis to lymph node. We localized preferentially in the cytoplasm (Fig. 2E). Using found that patients with SCLC with positive lymph node metas- cytoplasm-nuclear separation, we further confirmed that tasis had significantly higher intensity of FECR1 and FECR2 than FECRs were mostly localized in the cytoplasm of SCLC cells those with negative lymph node metastasis (P < 0.01, Fig. 3D and (Fig. 2F). E). These data suggest that FECRs are correlated with the metastatic characteristics of SCLC. FECRs are correlated with the metastatic status in patients with SCLC Serum exo-FECR1 as an indicator to track disease progression The role of FECRs in SCLC remains unknown. We used RNA of SCLC FISH staining to compare the expression pattern of FECRs in 56 To examine if the tumor-derived exosomal FECRs (exo-FECR) SCLC and 50 NSCLC (Supplementary Table S1). For comparison, may serve as useful prognostic factors for cancer prediction, we

www.aacrjournals.org Clin Cancer Res; 2018 OF7

Li et al.

Figure 4. Serum exo-FECR1 level correlates with therapy response. A, Image of electronic microscope and Western blot analysis detection of serum exosome. B, Quantitative PCR of serum exo-FECR1 in patients with SCLC, NSCLC, liver cancer, breast cancer, gastric cancer, colon cancer, ovary cancer, pancreatic cancer, and prostate cancer (ﬁve cases for each tumor). C, Comparison of serum exo-FECR1 in patients with SCLC and normal subjects. D, Serum exo-FECR1 in normal subjects and patients with SCLC with extensive and limited stage. E, FECR expression and disease survival. Patients were grouped as the high (median level) and the low exo-FECR1 (

extracted serum exosomes (Fig. 4A) and used qPCR to quantitate compared with other cancers (Fig. 4B). The level of FECR2 was exo-FECRs. Interestingly, we found that exo-FECR1 was detected considerably low in exosomes. Thus, we focused on exo-FECR1 as at relatively high levels in the serum of patients with SCLC as the monitoring indicator.

OF8 Clin Cancer Res; 2018 Clinical Cancer Research

Oncogenic FLI1 circRNAs in SCLC

We extended the measurement of serum exo-FECR1 in clinical We then lunched a series of studies to investigate the tumor samples from patients with SCLC (N ¼ 61) and normal subjects phenotype in FECR-knockdown SCLC cells, including cellular (N ¼ 55). Using qPCR, we found that serum exo-FECR1 was proliferation, apoptosis, cell cycle, migration, and ability of significantly higher in patients with SCLC than that in normal colony formation. Knockdown of the FLI1 linear mRNA slowed subjects (P < 0.001, Fig. 4C). In addition, the exo-FECR1 level in down cell growth in both NCI-H446 (Fig. 5C) and NCI-H1688 patients with extensive stage SCLC was significantly higher than (Fig. 5D). However, silencing of both FECR1 and FECR2 did not that in limited stage patients with SCLC (P ¼ 0.0014, Fig. 4D), significantly affect cell proliferation in these two cells. Similarly, especially for those extensive patients with SCLC who had distant silencing of FLI1 linear mRNA, but not FECR1 and FECR2, metastasis. significantly inhibited cell cycle and colony formation, and pro- We then investigated if exo-FECR1 could be used to predict the moted apoptosis in these two highly aggressive SCLC cell lines prognosis of SCLC. For this, we divided the patients into two (Supplementary Fig. S6; Fig. 3D). Clearly, both FECRs and FLI1 subgroups: high level of exo-FECR1 (median level) and low linear mRNA have distinct functions in the pathogenesis of SCLC. level of exo-FECR1 (microRNAs, the control (shNT). Additionally, we also used shFLI1 to target which have been shown to be the cytoplasmic targets of long FLI1 linear mRNA as the control. In NCI-H446 SCLC cells, noncoding RNAs. Using TargetScan 7.1 (http://www.targetscan. we showed that two shRNAs specifically knocked down FECR1 org), we screened the microRNA candidates that may potentially and FECR2, respectively, but did not affect the linear mRNA be sequestered by FECRs. (Fig. 5A). As expected, treatment with shFLI1 knocked down To validate the potential of the microRNA candidates, we first FLI1 mRNA but saved the circRNAs. Similar data were also examined their expression in two SCLC tumor cells lines (Fig. 7A). obtained for a second SCLC cell line (NCI-1688; Fig. 5B). Several candidates were expressed in considerable amounts in Together,thesedatasuggestthatitispossibletodesignshRNAs both SCLC cell lines, including miR584-3p, let7b-5p, miR17-3p, that specifically knocked down the circular transcript by target- miR937-5p, and miR92a-5p. We then constructed a luciferase ing the back-splice sequence, whereas they did not interfere the reporter vector by linking FECR1 (exons 4-2-3) and FECR2 (exons linear FLI1 mRNA. 5-2-3-4) to the end of the luciferase gene as its 30-UTR. We

www.aacrjournals.org Clin Cancer Res; 2018 OF9

Li et al.

Figure 5. Knockdown of FECRs inhibits the migration of SCLC cells. A, Knockdown of FECRs and FLI1 mRNA in NCI-H446 cells. Data are the mean SEM of three experiments. B, Knockdown of FECRs and FLI1 mRNA in NCI-H1688 cells. C, Cell proliferation in NCI-H446 cells transfected with three shRNAs assessed. Cell numbers were assessed using MTT assay at different time points. D, Cell proliferation in NCI-H1688 cells by MTT assay at the indicated hours. E, Migration of NCI-H446 cells transfected with three shRNAs. Cell migration was evaluated using Transwell assay. Data are the mean SEM of three experiments. F, Migration of NCI-H1688 cells in Transwell assay after 24 hours. , P < 0.001; , P < 0.01; , P < 0.05.

OF10 Clin Cancer Res; 2018 Clinical Cancer Research

Oncogenic FLI1 circRNAs in SCLC

Figure 6. Targeting FECRs suppresses tumor metastasis and improves animal survival. A, Real-time images of tumor growth in animals at weeks 8 and 12. B, Xenograft tumor growth of luciferase-labeled NCI-H446 cells that express shNT, shFECR1, or shFECR2. Tumor growth was tracked by bioluminescence at 8 and 12 weeks. C, Tumor survival. Mice implanted with NCI-H446 cells expressing shNT, shFECR1, or shFECR2 were maintained until death or for 112 days, whichever came ﬁrst. D, Representative images of sections (H&E stained) of mouse lungs and livers 8 weeks after transplantation. Arrow indicates a tumor formed from NCI-H446 cells. E and F, The quantitative data of tumor metastasis in the lung (E) and liver (F).

www.aacrjournals.org Clin Cancer Res; 2018 OF11

Li et al.

Figure 7. FECRs target miR584-3p and inhibit its activity. A, Expression of miRNAs in NCI-H446 and NCI-H1688 SCLC cells. B, Luciferase assay. The FECR1 (exon 4-2-3) and FECR2 (exon 5-2-3-4) sequence were cloned in the downstream of the 30-UTR of luciferase gene (Luc-FECR1 and Luc-FECR2 vectors). Luciferase reporter assay was performed in HEK-293T cells by cotransfecting Luc-FECR1 or Luc-FECR2 with siFECR or vector control. Data were adjusted over the negative control (NC) and were represented as mean SEM of three experiments (, P < 0.01). C and D, Screening of miRNAs using dual luciferase reporter assay. Luciferase activity of Luc-FECR1 (C) and Luc-FECR2 (D) was assayed in HEK-293Tcells that were treated with 11 miRNA mimics. miRNAs as the target of FECR1 (C) or FECR2 (D) were screened by luciferase activity. NC, control RNA as the negative control; siFECR, positive controls. E, Colocalization between FECRs and miR584-3p in NCI-H446 cells. FISH was used to localize FECRs and miR584-3p. Nuclei were stained with DAPI. Scale bar, 5 mm. F, Rescue assay in shFECR-treated SCLC cells. Treatment of miR584-3p inhibitor rescues the inhibition of cell migration induced by FECR1 and FECR2 knockdown. NC, control RNA. Data are the mean SEM of three experiments (, P < 0.01).

cotransfected reporter vector with siFECR that targets the splicing site activity of luciferase. However, cotransfection with siFECR signiﬁ- or microRNA control (NC). As expected, cotransfection of the cantly inhibited the activity of luciferase (Fig. 7B). By further reporter vector with the NC control did not signiﬁcantly affect the screening miRNA candidates, we notably found that co-transfection

OF12 Clin Cancer Res; 2018 Clinical Cancer Research

Oncogenic FLI1 circRNAs in SCLC

NC 2.0 A B shNT 2.0 miR-584-3p inhibitor shFECR1 * 1.5 * shFECR2 1.5 ** ** ** ad-vector 1.0 1.0 ad-FECR1 ad-FECR2 0.5 0.5

0.0 ROCK1 relative expression 0.0 ROCK1 relative expression H446 H1688 H446 H1688 C lamroN recnaC

Figure 8. FECRs function through the miR584-3p/ROCK1 pathway. A, Quantitation of ROCK1 in NCI-H446 and NCI-H1688 cells treated with miRNA control RNA (NC) or miR584-3p inhibitor. ROCK1 was measured by qPCR. , P < 0.05 as compared with the NC group. B, ROCK1 in NCI-H446 and NCI-H1688 cells. Cells were transfected with shNT, shFECR1, and shFECR2 that knockdown FECRs, or with ad-vector, ad-FECR1, and ad-FECR2 that overexpress FECRs. , P < 0.05 and , P < 0.01 as compared with the controls. C, The proposed model for the function of FECRs and FLI1 protein in SCLC. FLI1-encoding protein and its FECRs may synergistically regulate the tumor phenotypes of SCLC cells. FECRs sequester the activity of miR584-3p to ensure smooth regulation of RhoA–ROCK1 pathway, inducing tumor metastasis. with miR584-3p consistently reduced the activity of the cell–cell adhesion (25). Rho-associated coiled-coil kinase 1 luciferase reporter in both FECR1 and FECR2 constructs (ROCK1), a major effector of the small GTPase RhoA, is (Fig. 7C and D). responsible for many important cancer-associated phenotypes, We then examined if miR584-3p is physically colocalized with including metastasis. It is now clear that the oncogenic activity FECRs in the cells. Using RNA FISH, we confirmed that two FECRs of ROCK-1 is controlled by miR584-3p (26). We thus inves- were colocalized with their downstream target miR584-3p in the tigated whether FECRs may play a role in metastasis through cytoplasm (Fig. 7E). Notably, we were able to block cell invasion this pathway. by an inhibitor that inhibits miR584-3p, the downstream target of We transfected two SCLC cells with the miR584-3p inhibitor, two FECRs. and found that inhibition of this microRNA was able to To further validate the role of miR584-3p, we transfected SCLC upregulated ROCK1 (Fig. 8A). We then compared the activity cells that carry shNT, shFECR1, and shFECR2 with miRNA control of ROCK1 in cells where the two FECRs were knocked down. As (NC) and miR584-3p inhibitor respectively, and compared cell seen in Fig. 8B, knockdown of both FECRs downregulated invasion using a Transwell assay. As compared with the NC ROCK1 in both NCI-H446 and NCI-H1688 SCLC cells. control, the miR584-3p inhibitor significantly blocked cell inva- In contrast, ectopic expression of FECRs using adenoviruses sion induced by FECR knockdown (Fig. 7F). significantly upregulated ROCK1 in NCI-H446, although this effect was not significant in the second SCLC cells FECRs regulate the miR584–ROCK1 pathway (NCI-H1688). Collectively, these data suggest that FLI1 Rho GTPases family proteins play a key role in cell migra- circRNAs may control tumor metastasis in SCLC cells through tion and tumor metastasis by regulating actin dynamics and the miR584–ROCK1 pathway.

www.aacrjournals.org Clin Cancer Res; 2018 OF13

Li et al.

Discussion miR584-3p is physically colocalized with FECRs in the cytoplasm. Finally, application of the miR584-3p inhibitor is able to rescue In the past two decades, almost all studies have been focused shFECR-induced inhibition of SCLC cell invasion. etiologically on genomic alterations, such as somatic copy num- The RhoA-signaling pathway is activated in response to signals ber alterations or mutations, in the onset of many types of cancers, from extracellular matrix and adhesion molecules, hormones, including SCLC (7). It is conventionally believed that these and nutrients (29). Active GTP-bound RhoA binds to a wide series genomic alterations lead to tumorigenesis and development of of downstream effectors, including ROCK1, and induces allosteric cancer by affecting the oncoproteins or tumor suppressors activation of the serine/threonine kinase domain and autopho- encoded by these mutated genes. In this study, however, we sphorylation (30). ROCK-1 functions as an oncogene and is uncover a noncanonical function for FLI1 proto-oncogene in the overexpressed in many tumors (29), including lung cancers. Its development of SCLC. We demonstrate a novel pathway by which expression is not only associated with migration, invasion, and FLI1 controls the malignant behavior of SCLC cells through its metastasis (31, 32), but is also closely related with pulmonary exonic circular RNAs (FECR). Notably, we reveal that FLI1 linear fibrosis (33, 34) and lung diseases (35). The oncogenic activity of mRNA and FECRs function differently in SCLC cells. FLI1 linear ROCK-1 is controlled by miR584 (26). Consisting with these mRNA mainly affected cell proliferation through encoding FLI1 findings, we also demonstrate in this study that inhibition of oncoprotein, whereas FECRs tended to affect tumor metastasis miR584-3p is able to upregulate ROCK1 in SCLC cells. Similarly, through the miR584–3p/ROCK1 pathway. Thus, the FLI1 knockdown or ectopic expression of FECRs that sequester oncogene may play a duel role in the development of SCLC miR584-3p affects the activity of ROCK1. ROCK inhibitors, such through its canonical FLI1 proto-oncogene and its noncanonical as Y-27632, have been used in the clinical treatment of glaucoma circular RNAs. in Japan (36). Both animal studies and early clinical trials also CircRNAs are derived from precursor mRNA via exon circular- indicate that ROCK inhibitors may be a highly anticipated emerg- 0 ization, a back-splicing reaction in which a downstream 5 splice ing treatment option for glaucoma (37). Importantly, ROCK 0 site interacts with an upstream 3 splice site. Almost all FLI1 inhibitors have the ability to inhibit proliferation and invasion circular RNAs arise from exon 2 and exon 3 near the translation of tumor cells (38, 39). Therefore, it would be interesting to initiation site. By using CRISPR Cas9 to target the exon 2-exon 3 investigate whether ROCK inhibitors could clinically benefit region, we not only knocked down FLI1 protein, but also inter- patients with SCLC. fered FLI1 circular RNAs. As a result, we detected a significant Lack of reliable clinical biomarkers is a bottleneck in monitor- reduction of FECRs (FECR1 and FECR2) in the Cas9 knockdown ing the progression and therapeutic response of SCLC. In this cells (Supplementary Fig. S3). In contrast, shRNAs that target the study, we proved that serum exo-FECR1 level was aberrantly 0 3 -UTR of FLI1 was able to knockdown FLI1 linear mRNA, but increased in patients with SCLC compared with normal subjects. showed little effect on circular RNAs that lack the target region. We also uncovered that the serum exosomal FECR1 level was Although both technologies were able to efficiently block the closely associated with the tumor burden of SCLC and extensive production of FLI1 oncoprotein, they had the distinct effect on cell SCLC patients, especially for those with distant metastasis phenotypes. FLI1 knockout using CRISPR Cas9 significantly sup- expressed significantly higher level of FECR1. Importantly, we pressed SCLC cell migration, but this effect could be observed in demonstrated that the level of serum exosomal FECR1 was closely FLI1 shRNA knockdown cells. This leads us to investigate if FECRs related to chemotherapy response in SCLC patients. Our findings function as a new metastatic driver of SCLC. thus suggest that serum exosomal FECR1 might be clinically Recently, we reported that FLI1 promoted tumor growth of useful as a biomarker to track disease progression. SCLC cells by affecting cell apoptosis, cell proliferation, and It should be noted that some findings in this study, although tumor colony formation (9). In this study, however, we demon- statistically significant, exhibit only the modest effect. For exam- strate that FECRs play a distinct role in SCLC cells. Unlike FLI1 ple, knockdown by either shRNA or CRISPR Cas9 targeting only oncoprotein, FECRs significantly inhibited migration and metas- induced about 50% reduction in FECRs. Consequently, a statis- tasis of highly aggressive SCLC cell lines both in vitro and in animal tically significant but modest therapeutic effect was observed in studies. In contrast, they have little effect on cell proliferation, metastasis and rumor xenograft studies. It will be interesting to cell cycle, apoptosis, and colony formation. Taken together, these examine if a better inhibition of FECRs by optimizing Cas9 gRNAs data suggest that the linear coding mRNA and circular RNAs, or shRNAs could produce a more robust therapeutic effect. although derived from the same FLI1 oncogene, may function Similarly, the effect of FECR knockdown on ROCK1, whereas through different pathways to simultaneously affect the progres- statistically significant, was quantitatively modest. Future studies sion of SCLC. are needed to examine if knockdown of both FECR1 and FECR2 CircRNAs have been suggested to function as miRNA sponges. will improve the therapeutic effect. In addition, the metastasis However, only a few of such circRNAs contain multiple binding of SCLC cells is determined by multiple factors. Herein, the sites to trap a particular miRNA. In our study, based on genome- FECR–miR584-3p–ROCK1 pathway identified in this study may wide bioinformatic analyses of all circRNA candidates and lucif- represent just one of those undetermined factors that control the erase reporter assay, we identified miR584-3p as a specific miRNA progression of SCLC. Finally, as shown in Fig. 4C, the differences target for FECRs. miR584-3p is a potential tumor suppressor of serum exosomal FECR1, whereas statistically significant, show microRNA in several tumors, including clear cell renal cell carci- some overlap between normal subjects and patients with limited noma (26), anaplastic thyroid carcinoma (27), and breast cancer SCLC and extensive SCLC. Future studies are needed to validate (28). In this study, we reveal that FECRs directly target miR584-3p. these findings by recruiting large cohort of patients with SCLC. First, cotransfection with miR584-3p consistently inhibits the In summary, this study demonstrates FECRs as a previously activity of the luciferase reporter in both FECR1 and FECR2 unidentified oncogenic driver that promotes tumor metastasis reporter vectors. Second, using RNA FISH, we demonstrate that through the miR584–ROCK1 pathway. Thus, FLI1-encoding

OF14 Clin Cancer Res; 2018 Clinical Cancer Research

Oncogenic FLI1 circRNAs in SCLC

protein and its exonic noncoding RNA may synergistically regu- (2016YFC1303804), National Natural Science Foundation of China grant late the tumor phenotypes of SCLC cells. Importantly, measure- (81672275, 81874052, 3A214DJ63428), Jilin Provincial Key Laboratory of ment of serum exo-FECR1 may serve as a promising biomarker to biological therapy (20170622011JC), Provincial Science Fund of Jilin Province Development and Reform Commission (2014N147 and 2017C022), Oncology track disease progression of SCLC. Prevention and Control Program of Action for Major Diseases Prevention and Control of the State Health Planning Commission (2016ZX-07-002), Jilin Disclosure of Potential Conflicts of Interest Provincial Finance Department Project (2018SCZWSZX-010) and Research and No potential conflicts of interest were disclosed. Cultivation Fund of the First Hospital of Jilin University (LCPYJJ2017003) to J. Cui; the National Natural Science Foundation of China (31430021, 31871297), Authors' Contributions the National Key R&D Program of China (2018YFA0106902), the National Basic Research Program of China (973 Program; 2015CB943303), Natural Conception and design: L. Li, A. Li, J.-F. Hu, J. Cui Science Foundation of Jilin Science and Technique (20180101117JC), and Development of methodology: L. Li, N. Chen, L. Zhou, A. Li California Institute of Regenerative Medicine (CIRM) grant (RT2-01942) to J.-F. Acquisition of data (provided animals, acquired and managed patients, Hui; the National Natural Science Foundation of China grant (81372835) provided facilities, etc.): L. Li, N. Chen, X. Zhang, A. Li and Jilin Science and Technique International Collaboration grant Analysis and interpretation of data (e.g., statistical analysis, biostatistics, (20130413010GH) to W. Li; the Youth Fund of the National Natural Science computational analysis): L. Li, X. Pan, D. Yu, A. Li, J.-F. Hu Foundation of China (81802487) and Youth Development Foundation of the Writing, review, and/or revision of the manuscript: L. Li, N. Chen, A. Li, First Hospital of Jilin university (JDYY92018028) to L. Li. J.-F. Hu, J. Cui Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): L. Li, H. Zhao, G. Xu, Y. Zhao, X. Pan, A. Li, J. Cui The costs of publication of this article were defrayed in part by the payment of Study supervision: W. Li, A. Li, J. Cui page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Acknowledgments This work was supported by Research on Chronic Noncommunicable Dis- Received May 15, 2018; revised September 28, 2018; accepted November 9, eases Prevention and Control of National Ministry of Science and Technology 2018; published first November 14, 2018.

References 1. Ferlay J, Soerjomataram I, Dikshit R, Eser S, Mathers C, Rebelo M, et al. 13. Cui JW, Vecchiarelli-Federico LM, Li YJ, Wang GJ, Ben-David Y. Continuous Cancer incidence and mortality worldwide: sources, methods and major Fli-1 expression plays an essential role in the proliferation and survival of patterns in GLOBOCAN 2012. Int J Cancer 2015;136:E359–86. F-MuLV-induced erythroleukemia and human erythroleukemia. Leukemia 2. Siegel R, Ma J, Zou Z, Jemal A. Cancer statistics, 2014. CA Cancer J Clin 2009;23:1311–9. 2014;64:9–29. 14. Ramani N, Aung PP, Hwu WJ, Nagarajan P, Tetzlaff MT, Curry JL, et al. 3. Simon GR, Turrisi A, American College of Chest P. Management of small Aberrant expression of FLI-1 in melanoma. J Cutan Pathol 2017;44:790–3. cell lung cancer: ACCP evidence-based clinical practice guidelines 15. Liang X, Shi D, Yun J, Mao Y, Ouyang P, Su Z, et al. Friend leukemia virus (2nd edition). Chest 2007;132:324S–39S. integration 1 expression has prognostic significance in nasopharyngeal 4. Jotte R, Conkling P, Reynolds C, Galsky MD, Klein L, Fitzgibbons JF, et al. carcinoma. Transl Oncol 2014;7:493–502. Randomized phase II trial of single-agent amrubicin or topotecan as 16. Gorthi A, Romero JC, Loranc E, Cao L, Lawrence LA, Goodale E, et al. second-line treatment in patients with small-cell lung cancer sensitive to EWS-FLI1 increases transcription to cause R-loops and block BRCA1 repair first-line platinum-based chemotherapy. J Clin Oncol 2011;29:287–93. in Ewing sarcoma. Nature 2018;555:387–91. 5. Schmittel A, Sebastian M, Fischer von Weikersthal L, Martus P, Gauler TC, 17. Jeck WR, Sharpless NE. Detecting and characterizing circular RNAs. Kaufmann C, et al. A German multicenter, randomized phase III trial Nat Biotechnol 2014;32:453–61. comparing irinotecan-carboplatin with etoposide-carboplatin as first-line 18. Memczak S, Jens M, Elefsinioti A, Torti F, Krueger J, Rybak A, et al. therapy for extensive-disease small-cell lung cancer. Ann Oncol 2011;22: Circular RNAs are a large class of animal RNAs with regulatory potency. 1798–804. Nature 2013;495:333–8. 6. Spigel DR, Townley PM, Waterhouse DM, Fang L, Adiguzel I, Huang JE, 19. Li Z, Huang C, Bao C, Chen L, Lin M, Wang X, et al. Exon-intron circular et al. Randomized phase II study of bevacizumab in combination with RNAs regulate transcription in the nucleus. Nat Struct Mol Biol 2015; chemotherapy in previously untreated extensive-stage small-cell lung 22:256–64. cancer: results from the SALUTE trial. J Clin Oncol 2011;29:2215–22. 20. Chen N, Zhao G, Yan X, Lv Z, Yin H, Zhang S, et al. A novel FLI1 exonic 7. George J, Lim JS, Jang SJ, Cun Y, Ozretic L, Kong G, et al. Comprehensive circular RNA promotes metastasis in breast cancer by coordinately regu- genomic profiles of small cell lung cancer. Nature 2015;524:47–53. lating TET1 and DNMT1. Genome Biol 2018;19:218. 8. Wong KS, Li YJ, Howard J, Ben-David Y. Loss of p53 in F-MuLV induced- 21. Pian L, Wen X, Kang L, Li Z, Nie Y, Du Z, et al. Targeting the IGF1R pathway erythroleukemias accelerates the acquisition of mutational events that in breast cancer using antisense lncRNA-Mediated Promoter cis Compe- confers immortality and growth factor independence. Oncogene 1999;18: tition. Mol Ther Nucleic Acids 2018;12:105–17. 5525–34. 22. Wang H, Guo R, Du Z, Bai L, Li L, Cui J, et al. Epigenetic targeting of granulin 9. Li L, Song W, Yan X, Li A, Zhang X, Li W, et al. Friend leukemia virus in hepatoma cells by synthetic CRISPR dCas9 Epi-suppressors. Mol Ther integration 1 promotes tumorigenesis of small cell lung cancer cells by Nucleic Acids 2018;11:23–33. activating the miR-17-92 pathway. Oncotarget 2017;8:41975–87. 23. Sanjana NE, Shalem O, Zhang F. Improved vectors and genome-wide 10. Ben-David Y, Giddens EB, Bernstein A. Identification and mapping of a libraries for CRISPR screening. Nat Methods 2014;11:783–4. common proviral integration site Fli-1 in erythroleukemia cells induced by 24.WangH,ChaoK,NgSC,BaiAH,YuQ,YuJ,etal.Pro-inflammatory Friend murine leukemia virus. Proc Natl Acad Sci U S A 1990;87:1332–6. miR-223 mediates the cross-talk between the IL23 pathway and the 11. Hart A, Melet F, Grossfeld P, Chien K, Jones C, Tunnacliffe A, et al. Fli-1 is intestinal barrier in inflammatory bowel disease. Genome Biol 2016; required for murine vascular and megakaryocytic development and is 17:58. hemizygously deleted in patients with thrombocytopenia. Immunity 25. Fukata M, Kaibuchi K. Rho-family GTPases in cadherin-mediated cell-cell 2000;13:167–77. adhesion. Nat Rev Mol Cell Biol 2001;2:887–97. 12. Spyropoulos DD, Pharr PN, Lavenburg KR, Jackers P, Papas TS, Ogawa M, 26. Ueno K, Hirata H, Shahryari V, Chen Y, Zaman MS, Singh K, et al. Tumour et al. Hemorrhage, impaired hematopoiesis, and lethality in mouse suppressor microRNA-584 directly targets oncogene Rock-1 and decreases embryos carrying a targeted disruption of the FLI1 transcription factor. invasion ability in human clear cell renal cell carcinoma. Br J Cancer Mol Cell Biol 2000;20:5643–52. 2011;104:308–15.

www.aacrjournals.org Clin Cancer Res; 2018 OF15

Li et al.

27. Orlandella FM, Di Maro G, Ugolini C, Basolo F, Salvatore G. TWIST1/ 33. Mrass P, Oruganti SR, Fricke GM, Tafoya J, Byrum JR, Yang L, et al. ROCK miR-584/TUSC2 pathway induces resistance to apoptosis in thyroid regulates the intermittent mode of interstitial T cell migration in inflamed cancer cells. Oncotarget 2016;7:70575–88. lungs. Nat Commun 2017;8:1010. 28.Fils-AimeN,DaiM,GuoJ,El-MousawiM,KahramangilB,NeelJC, 34. Knipe RS, Probst CK, Lagares D, Franklin A, Spinney JJ, Brazee PL, et al. et al. MicroRNA-584 and the protein phosphatase and actin The rho kinase isoforms ROCK1 and ROCK2 each contribute to the regulator 1 (PHACTR1), a new signaling route through which development of experimental pulmonary fibrosis. Am J Resp Cell Mol transforming growth factor-beta Mediates the migration and Biol 2017;58:471–81. actin dynamics of breast cancer cells. J Biol Chem 2013;288: 35. Knipe RS, Tager AM, Liao JK. The Rho kinases: critical mediators of multiple 11807–23. profibrotic processes and rational targets for new therapies for pulmonary 29. Schofield AVBernard O. Rho-associated coiled-coil kinase (ROCK) fibrosis. Pharmacol Rev 2015;67:103–17. signaling and disease. Crit Rev Biochem Mol Biol 2013;48: 36. Honjo M, Tanihara H, Kameda T, Kawaji T, Yoshimura N, Araie M. 301–16. Potential role of Rho-associated protein kinase inhibitor Y-27632 in 30. Loirand G. Rho kinases in health and disease: from basic science to glaucoma filtration surgery. Invest Ophthalmol Vis Sci 2007;48:5549–57. translational research. Pharmacol Rev 2015;67:1074–95. 37. Wang SK, Chang RT. An emerging treatment option for glaucoma: Rho 31.LuoD,ChenH,LiX,LuP,LongM,PengX,etal.Activationofthe kinase inhibitors. Clin Ophthalmol 2014;8:883–90. ROCK1/MMP-9 pathway is associated with the invasion and poor 38. Liu X, Bi Y. Y-27632 increases sensitivity of PANC-1 Cells to EGCG prognosis in papillary thyroid carcinoma. Int J Oncol 2017;51: in regulating cell proliferation and migration. Med Sci Monit 2016;22: 1209–18. 3529–34. 32. Hinsenkamp I, Schulz S, Roscher M, Suhr AM, Meyer B, Munteanu B, et al. 39. Jiang L, Wen J, Luo W. Rhoassociated kinase inhibitor, Y27632, inhibits the Inhibition of Rho-Associated Kinase 1/2 attenuates tumor growth in invasion and proliferation of T24 and 5367 bladder cancer cells. Mol Med murine gastric cancer. Neoplasia 2016;18:500–11. Rep 2015;12:7526–30.

OF16 Clin Cancer Res; 2018 Clinical Cancer Research

FLI1 Exonic Circular RNAs as a Novel Oncogenic Driver to Promote Tumor Metastasis in Small Cell Lung Cancer

Lingyu Li, Wei Li, Naifei Chen, et al.

Clin Cancer Res Published OnlineFirst November 14, 2018.

Updated version Access the most recent version of this article at: doi:10.1158/1078-0432.CCR-18-1447

Supplementary Access the most recent supplemental material at: Material http://clincancerres.aacrjournals.org/content/suppl/2018/11/14/1078-0432.CCR-18-1447.DC1

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/early/2019/01/06/1078-0432.CCR-18-1447. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.