Published OnlineFirst July 9, 2019; DOI: 10.1158/0008-5472.CAN-18-4032

Cancer Molecular Cell Biology Research

SCFFBXW7/GSK3b-Mediated GFI1 Degradation Suppresses Proliferation of Gastric Cancer Cells Xiaoling Kuai1, Long Li2, Ran Chen2, Kangjunjie Wang2, Min Chen2, Binghai Cui2, Yuxue Zhang2, Junqiang Li3, Hongwen Zhu4, Hu Zhou4, Jianfei Huang5, Jun Qin6, Zhiwei Wang7, Wenyi Wei8, and Daming Gao2

Abstract

Gastric cancer is the third leading cause of cancer-related mediated ubiquitination and degradation. A nondegradable death worldwide. The regulatory mechanisms underlying gas- GFI1 S94A/S98A mutant was more potent in driving gastric tric cancer cell proliferation are largely unclear. Here, we show cancer cell proliferation and tumorigenesis than wild-type that the GFI1 is associated with advanced GFI1. Overall, this study reveals the oncogenic role of GFI1 clinical gastric cancer progression and promoted gastric cancer in gastric cancer and provides mechanistic insights into the cell proliferation partially through inhibition of gastrokine-2 tumor suppressor function of FBXW7. (GKN2) transcription. GFI1 was a degrading substrate of FBXW7, whose loss was observed in gastric cancer. Mechanis- Significance: These findings demonstrate the oncogenic tically, GSK3b-mediated GFI1 S94/S98 phosphorylation trig- role of the transcription factor GFI1 and the tumor suppressive gered its interaction with FBXW7, resulting in SCFFBXW7- function of FBXW7 in gastric cancer.

Introduction Although Helicobacter pylori infection is thought to be the main cause for sporadic distal-type gastric cancer (4), gastric cancer is As a global health threat (1), gastric cancer is the fourth most distinct from other solid tumors and has many unique pathologic common cancer and the third leading cause of cancer-related features. Therefore, a better understanding of the molecular deaths worldwide (2). In addition to the sustained high economic mechanisms underlying the occurrence and progression of gastric burden of gastric cancer in most countries, gastric cancer is one of cancer is urgently needed to improve clinical practice. the most serious public health problems in China because the Growth factor independent 1 transcriptional (GFI1) incidence and mortality rate of gastric cancer is rapidly rising (3). is located within 1p22 in the (5). As a zinc finger protein, GFI1 mainly functions as a transcriptional 1Department of Gastroenterology, Nantong University Affiliated Hospital, Nan- repressor by direct or functional interaction with other cofactors, tong, Jiangsu, China. 2State Key Laboratory of Cell Biology, CAS Key Laboratory including protein lesion simulating disease 1 (LSD1; ref. 6). It has of Systems Biology, CAS Center for Excellence in Molecular Cell Science, been shown that GFI1 plays an important role in hematopoietic Innovation Center for Cell Signaling Network, Shanghai Institute of Biochemistry stem cells. Together with Fosb, Runx1, and Spi1, GFI1 contributes and Cell Biology, University of Chinese Academy of Sciences, Chinese Academy to conversion of adult endothelium to immunocompetent 3 4 of Sciences, Shanghai, China. OrigiMed Inc., Shanghai, China. Department of hematopoietic stem cells (7). Subsequent studies have revealed Analytical Chemistry and CAS Key Laboratory of Research, Shanghai Institute of Materia Medica, University of Chinese Academy of Sciences, Chinese that GFI1 restricts proliferation and preserves functional integrity Academy of Sciences, Shanghai, China. 5Department of Clinical Biobank, Nan- in addition to regulating self-renewal and engraftment of hemato- tong University Affiliated Hospital, Nantong, Jiangsu, China. 6The Key Labora- poietic stem cells (8). For the immune system, GFI1 has been tory of Stem Cell Biology, CAS Center for Excellence in Molecular Cell Science, reported to control Th17 cell immunosuppressive activity via Institute of Health Sciences, Shanghai Institutes for Biological Sciences, Chinese regulation of ectonucleotidase expression (9), and functions in Academy of Sciences/Shanghai Jiao Tong University School of Medicine, the specification of type 2 innate lymphocytes (10). Although a University of Chinese Academy of Sciences, Shanghai, China. 7Department of General Surgery, Nantong University Affiliated Hospital, Nantong, Jiangsu, previous study indicated that GFI1 may promote medulloblas- China. 8Department of Pathology, Beth Israel Deaconess Medical Center, Har- toma (11), the functional role of GFI1 in carcinogenesis, partic- vard Medical School, Boston, Massachusetts. ularly in the alimentary tract, has not been fully investigated. Note: Supplementary data for this article are available at Cancer Research Recently, a few genetic alterations have been reported as Online (http://cancerres.aacrjournals.org/). emerging markers of gastric cancer, such as key modulators of kinase pathways (12–14). In addition, F-box and WD repeat X. Kuai and L. Li contributed equally to this article. domain-containing 7 (FBXW7), the substrate recognition com- Corresponding Author: Daming Gao, Shanghai Institute of Biochemistry and ponent of the SCFFBXW7 ubiquitin ligase complex, was shown to Cell Biology, University of Chinese Academy of Sciences, Chinese Academy of be mutated in 9.2% to 18.5% of gastric tumors (15) and low Sciences, Shanghai 200031, China. Phone: 86-21-54921281; E-mail: [email protected] FBXW7 expression was previously shown to be associated with poor prognosis in patients with gastric cancer (16) because Cancer Res 2019;79:4387–98 FBXW7 controls apoptosis, proliferation, chemoresistance, and þ doi: 10.1158/0008-5472.CAN-18-4032 metastasis of gastric cancer (17, 18). Moreover, Fbxw7 / mice are 2019 American Association for Cancer Research. more prone to developing intestinal metaplasia, dysplasia, and

www.aacrjournals.org 4387

Downloaded from cancerres.aacrjournals.org on September 25, 2021. © 2019 American Association for Cancer Research. Published OnlineFirst July 9, 2019; DOI: 10.1158/0008-5472.CAN-18-4032

Kuai et al.

gastric cancer after carcinogen exposure (19). It is well-established collected to perform HA-IP. The sample was then resolved on that FBXW7 recognizes and degrades key factors, including SDS–PAGE and stained with Gelcode Blue Safe protein staining C- (20, 21), CCNE1 (22), C-JUN (23), and MCL1 (24), after reagents (Thermo Fisher Scientific). The corresponding bands phosphorylation by GSK3b (25), a kinase phosphorylated and were excised and digested using trypsin, and the resulting peptides inhibited by the activated RTK–PI3K–AKT pathway (26). With were purified using C18 stagetips (3M Company). For LC/MS-MS, respect to functioning as a general tumor suppressor in many peptides were loaded and separated on a home-made micro-tip cancer types (27), little is known about how FBXW7 functions in C18 column [ReproSil-Pur C18-AQ (1.9 mm, 75 mm 300 mm)] gastric cancer cell proliferation and progression by modulating on a nanoflow HPLC Easy-nLC 1200 system [buffer A (0.1% FA in gastric tissue-specific factors. In contrast, gastrokine-2 (GKN2) H2O) and buffer B (0.1% FA in 80% ACN); Thermo Fisher belongs to a family of secreted proteins located on chromosome Scientific]. The LC gradient was set as follows: 2% to 5% B for 2p13.3, the expression of which is restricted to normal gastric 1 minute; 5% to 33% B for 93 minutes; 33% to 45% B for 15 mucosal cells with a high prevalence in gastric juice (28, 29). minutes; 45% to 100% B for 3 minutes; and 100% B for 8 minutes. GKN2 is downregulated in gastric cancer based on differential MS data were collected using a data-dependent acquisition in the expression studies (30, 31), and GKN2 loss is correlated with poor profile mode on a Q Exactive HF mass spectrometry. Mass survival in intestinal-type gastric cancer according to multivariate spectrometric data were analyzed using MaxQuant 1.6.2.3 against analysis (32), indicating that GKNs are putative stomach-specific the human UniProt database. The data were deposited (http:// tumor suppressor . In addition, Gkn2 knockout in the IL11/ www.ebi.ac.uk/pride) with details as follows: project name, Dis- STAT3-dependent gp130F/F gastric cancer model led to tumori- section of GFI1 Function and Regulation in Gastric Cancer, and genesis of the proximal stomach (33); however, the mechanism project accession, PXD013416. underlying the low expression of GKN2 in gastric cancer has not been established. Microarray data Herein we characterized GFI1 as a potential oncoprotein in The microarray data from gastric cancer cells with GFI1 over- promoting gastric cancer cell proliferation via transcriptionally expression is deposited as GSE115530 (https://www.ncbi.nlm. inhibiting GKN2, and we further revealed that GFI1 is subject to nih.gov/geo/query/acc.cgi?acc¼GSE115530) with an access SCFFBXW7 E3 ligase/GSK3b-mediated degradation. Our results token: ijaxyqykvlutnsl. indicated that GFI1 is highly expressed in clinical gastric cancer samples and associated with poorer prognosis of patients. We Collection of human gastric cancer specimens further found that GFI1 promotes gastric cancer cell proliferation The clinical gastric tissue samples were histopathologically and and anchorage-independent growth, which involves suppression clinically diagnosed in the Department of Gastroenterology of of the gastric-specific tumor suppressor, GKN2 (34). We contin- Nantong University Affiliated Hospital. The written informed ued to identify GFI1 as a FBXW7 binding protein in gastric cancer consent was obtained from the participants. Tissue samples were cells, the stability of which is governed by FBXW7. Moreover, a collected after receiving written informed consent from partici- mechanistic study suggested that GSK3b controls GFI1 expression pants and approval from the Institutional Research Ethics Com- via phosphorylation of S94 and S98 sites, which leads to mittee. Tissue specimens were collected from 130 patients with SCFFBXW7-mediated polyubiquitination and proteasome degra- gastric cancer who underwent gastrectomy; 13 patients had chron- dation. As a result, expression of a nondegradable form of the ic gastritis, 28 patients had intestinal metaplasia, 16 patients had GFI1-S94A/S98A mutant dramatically increased xenograft tumor- low-grade intraepithelial neoplasia, and 22 patients had high- igenesis and anchorage-independent growth of gastric cancer grade intraepithelial neoplasia. All gastric cancer cases and adja- cells. Taken together, our study elucidated the oncogenic function cent nontumor tissues were diagnosed clinically and patholog- of GFI1 and implicated the mechanism by which GFI1 may be ically. Data on clinicopathologic features and prognoses of the increased in gastric cancer tissues with deficient FBXW7/GSK3b. patients were collected and analyzed retrospectively. The disease stage of each patient was classified or reclassified according to the 2009 AJCC staging system. A total of 130 patients were followed Materials and Methods until the end of 2016. Plasmids The HA-tagged coding sequence of human GFI1 wild-type or Xenograft tumorigenesis assay the relevant GFI1-S94A, GFI1-S98A, and GFI1-S94A/S98A HS746T cells (1 107) wild type or overexpressed with GFI1, mutants were cloned into the lentiviral vector, pLex-MCS-CMV- GFI1-S94A/S98A mutant, and GKN2/GFI1 double overexpres- puro (Addgene) to generate GFI1 expression plasmids. The cDNA sion in PBS were subcutaneously inoculated into the flanks of of human GKN2 was cloned into the pCDNA3.1 vector. The nude mice (7 male mice/group; 4–6 weeks old) purchased from cDNA of TRCP, SKP2, FBXW7(a, b, g), GFI1, CULLIN1, CULLIN2, Shanghai Experimental Animal Center and maintained in CULLIN3, CULLIN4A, CULLIN4B, CULLIN5, and constitutively pathogen-free conditions. Tumor sizes were measured every active GSK3b mutant were cloned into the pCMV-Flag vector. 3 days with a caliper and tumor volumes were determined GST-GFI1 and relevant mutants were generated by subcloning using the following formula: L W2 0.52, where L is the corresponding cDNA into a pGEX-4T-1 vector. longest diameter and W is the shortest diameter. Eight weeks later, all mice were sacrificed and tumors were harvested, Mass spectrometry analysis followed by photography and weighing of the specimens. All To identify GFI1 binding protein and phosphorylation sites, animals received humane care and all animal experiments were 293T cells expressing HA-GFI1 were cultured in the presence of performed in accordance with the guidelines of the Institution- 15 mmol/L MG132 for 10 hours before harvest. Cells were pulse- al Animal Care and Use Committee of Shanghai Institutes for sonicated in EBC buffer for 30 seconds, and cell lysates were Biological Sciences.

4388 Cancer Res; 79(17) September 1, 2019 Cancer Research

Downloaded from cancerres.aacrjournals.org on September 25, 2021. © 2019 American Association for Cancer Research. Published OnlineFirst July 9, 2019; DOI: 10.1158/0008-5472.CAN-18-4032

SCFFBXW7/GSK3b Degrades GFI1 in Gastric Cancer

Results indicate that high GFI1 expression was associated with worse malignancy and poorer prognosis of gastric cancer. High GFI1 expression is associated with gastric tumorigenesis To investigate the possible role of GFI1 in gastric disease, we first determined the level of GFI1 protein expression GFI1 promotes gastric cancer cell proliferation and suppresses via IHC staining in samples from patients with chronic GKN2 expression gastritis (n ¼ 13), intestinal metaplasia (n ¼ 28), and gastric Although GFI1 has been shown to be an important transcrip- cancer (n ¼ 130). The samples were then grouped as low expres- tional factor in hematopoietic stem cells, its role in oncogenesis of sion (score 0/1) and high expression (score 2/3) according to the solid tumors was largely unknown. To investigate the function of intensity of GFI1 staining. The percentage of gastric cancer sam- GFI1 in gastric cancer, we ectopically expressed GFI1 in HS746T ples with high GFI1 staining (50%; Fig. 1A) was significantly and BGC823 gastric cancer cells, and examined the proliferation higher than chronic gastritis (23.08%) and intestinal metaplasia and anchorage-independent colony formation of the resulting (32.14%; Fig. 1B). Then, we further analyzed the association of cells. Interestingly, GFI1 overexpressing cells proliferated faster high GFI1 expression with gastric cancer samples that were and formed more colonies in both anchorage-dependent and classified by 8 clinical factors. Although there was no correlation -independent conditions (Fig. 2A; Supplementary Fig. S1A–S1B). between GFI1 expression and several factors, including sex, age, In contrast, depletion of GFI1 in HS746T and HGC27 cells and histologic type, the results suggested significant associations dramatically delayed cell growth and reduced the number of of high GFI1 expression with T grade (P ¼ 0.0009), TNM stage colonies (Fig. 2B and C; Supplementary Fig. S1C–S1D). Thus, (P ¼ 0.002), and lymph node metastasis (P ¼ 0.0109; Table 1). GFI1 may exert a pro-proliferation function in gastric cancer cells, More importantly, Kaplan–Meier survival analysis from these which potentially contributes to oncogenesis. Given that GFI1 clinical samples indicated that patients with high GFI1 expression usually functions as a transcription repressor, we examined the (n ¼ 65) had poorer overall survival than patients with low GFI1 expression of several tumor suppressor genes in gastric cancer cells expression (P < 0.001; Fig. 1C). Taken together, these results with ectopic GFI1 expression to identify possible target genes of

Figure 1. GFI1 expression is associated with gastric cancer progression. A, Representative IHC images of GFI1 staining in gastric cancer. GFI1 low, scored 0/1. GFI1 high, scored 2/3. Scale bar, 50 mm. B, Differential percentage of high GFI1 samples in various gastric disorders. GFI1 low, scored 0/1. GFI1 high, scored 2/3. Chi-square test was used for statistical analysis. , P < 0.001. C, Kaplan–Meier survival curves of gastric cancer patients based on GFI1 expression. Survival analysis was performed using log-rank test.

www.aacrjournals.org Cancer Res; 79(17) September 1, 2019 4389

Downloaded from cancerres.aacrjournals.org on September 25, 2021. © 2019 American Association for Cancer Research. Published OnlineFirst July 9, 2019; DOI: 10.1158/0008-5472.CAN-18-4032

Kuai et al.

Table 1. Expression of GFI1 correlates with gastric cancer progression tively blocked tumor growth driven by ectopic GFI1 expression in GFI1 low GFI1 high a mouse xenograft assay (Fig. 2H and I). Moreover, the inverse n Characteristic expression, % expression, % correlation between GFI1 and GKN2 levels was observed in P Total 130 65 (50.00) 65 (50.00) clinical gastric cancer samples (Fig. 2J). Therefore, GFI1 is an Gender 0.690 oncogenic factor in gastric cancer cells and the function is at least Male 96 47 (48.96) 49 (51.04) Female 34 18 (52.94) 16 (47.06) partially through transcriptional inhibition of GKN2. Age 0.212 <60 53 30 (56.60) 23 (43.40) GFI1 interacts with and is degraded by FBXW7 60 77 35 (45.45) 42 (54.55) To identify possible GFI1 modulators in gastric cancer cells, we Histological type 0.584 generated SGC7901 cells expressing HA-GFI1 to perform immu- Tubular 100 47 (47.00) 53 (53.00) noprecipitation (IP)–mass spectrometry analysis. In addition to adenocarcinoma Mucinous 7 4 (57.14) 3 (42.86) the known GFI1 binding proteins, PRMT1 (35) and HDAC2 (36), adenocarcinoma FBXW7 was detected as a potential GFI1-interacting protein Poorly cohesive 15 9 (60.00) 6 (40.00) (Fig. 3A). The interaction between endogenous GFI1 and exog- carcinoma enous or endogenous FBXW7 (the a isoform) was further vali- a Others 85 3 dated by IP (Fig. 3B–D; Supplementary Fig. S2A). Interestingly, Differentiation 0.139 minimal interaction was detected between GFI1 and other F-box Poor 80 43 (53.75) 37 (46.25) Middle 36 14 (38.89) 22 (61.11) proteins, including SKP2 and b-TRCP1 or CULLIN 2-5, thus FBXW7 Well 14 8 (57.14) 6 (42.86) supporting a specific interaction between the SCF complex T grade 0.0009d and GFI1 (Fig. 3E and F). Reciprocally, ectopic expression of wild- Tis 6 6 (100.00) 0 (0.00) type FBXW7a, but not SKP2 or b-TRCP1, caused a decrease in T1 13 10 (76.92) 3 (23.08) GFI1 and shortened the half-life (Fig. 3G and H). Moreover, the T2 28 17 (60.71) 11 (39.39) GFI1 degradation effect was only specific to wild-type FBXW7a, T3 77 29 (37.66) 48 (62.34) T4 6 3 (50.00) 3 (50.00) but not disease-derived R465C mutant (Fig. 3I; ref. 37), further Lymph node metastasis 0.0109b suggesting that GFI1 is a possible physiologic substrate of FBXW7. N0 54 36 (66.67) 18 (33.33) Moreover, cotransfection of shRNAs targeting FBXW7 reduced N1 27 12 (44.44) 15 (55.56) GFI1 ubiquitination (Supplementary Fig. S2B). Because GFI1 N2 21 8 (38.10) 13 (61.90) suppressed GKN2 expression in gastric cancer cells and is N3 28 9 (32.14) 19(67.86) governed by SCFFBXW7, we continued to investigate whether Distant metastasis 0.342 fl M0 119 61 (51.26) 58(48.74) or not GKN2 expression is in uenced by FBXW7 status. Impor- M1 11 4 (36.36) 7 (63.64) tantly, depletion of endogenous FBXW7 in BGC823 GC cells TNM stage 0.002c caused a significant increase in endogenous GFI1 protein levels, 0 6 6 (100.00) 0 (0.00) which was very similar to the well-known FBXW7 substrate, I 29 21 (72.41) 8 (27.59) C-MYC, and only had a minimal effect on GFI1 mRNA levels, II 43 19 (44.19) 24 (55.81) but dramatically decreased GKN2 mRNA levels, suggesting that III 41 15 (36.59) 26 (63.41) IV 11 4 (36.36) 7 (63.64) FBXW7 might govern GKN2 expression in a GFI1-dependent fi aPapillary adenocarcinoma, one case; neuroendocrine carcinogenesis, one case; manner (Fig. 3J and K). Consistently, a signi cant positive malignant cell carcinoma, one case; mixed (tubular and mucinous) carcinoma, 5 correlation between FBXW7 and GKN2 mRNA abundance was cases. Fisher exact test was used to test the association between two categorical observed in public data sets from gastric cancer cohorts (Fig. 3L variables. and M; Supplementary Fig. S2C–S2E), indicating inhibition of b 0.01 < P < 0.05. FBXW7 may contribute to suppression of GKN2 in gastric c P 0.001 < < 0.01. cancer. Collectively, these results support the notion that GFI1 dP < 0.001. is a degradation substrate of the SCFFBXW7 complex.

GFI1 via microarray analysis. Interestingly, the expression of GSK3b destabilizes GFI1 in gastric cancer cells GKN2 was significantly reduced upon ectopic GFI1 expression In an attempt to screen for the specific upstream signaling (Supplementary Fig. S1E). However, efficient knockdown of GFI1 that regulates GFI1 expression, we found that pazopanib, a caused a 10-fold increase in GKN2 transcripts and a 3-fold increase potent multitarget receptor tyrosine kinase (RTK) inhibitor that in secreted GKN2 protein in the supernatant, suggesting that has been approved for renal cell carcinoma and soft tissue GKN2 may be an important and sensitive downstream of sarcoma treatment by numerous agencies worldwide, effectively GFI1 (Fig. 2D and E; Supplementary Fig. S1F). To further deter- reduced ectopic GFI1 expression, which could be blocked by mine whether or not GKN2 is a direct target gene of GFI1, the proteasome inhibitor, MG132 (Supplementary Fig. S3A). chromatin immunoprecipitation (ChIP) was performed to vali- Because pazopanib may alter different kinase pathways, including date direct binding of GFI1 at the GKN2 promoter region (Fig. 2F). PI3K–AKT and MAPK–ERK, we examined the change in GFI1 To validate whether or not GKN2 suppression contributes to expression upon treatment of chemicals inhibiting AKT, GFI1-mediated gastric cancer cell proliferation, GKN2 was ectop- GSK3b, and MAPK. Interestingly, both exogenous and endoge- ically expressed in SGC7901 and HS746T cells with GFI1 over- nous GFI1 levels were increased dramatically after treatment expression. Indeed, GKN2 expression significantly reversed with the GSK3b inhibitors, CHIR99021 and LiCl (Fig. 4A; Sup- anchorage-dependent colony formation and S-phase entry that plementary Fig. S3B). Thus, we speculate that GSK3b may be was promoted by GFI1 expression (Fig. 2G; Supplementary the kinase that negatively modulates GFI1 stability. In keeping Fig. S1G–S1H). Importantly, additional GKN2 expression effec- with this notion, the half-life of ectopically-expressed GFI1 was

4390 Cancer Res; 79(17) September 1, 2019 Cancer Research

Downloaded from cancerres.aacrjournals.org on September 25, 2021. © 2019 American Association for Cancer Research. Published OnlineFirst July 9, 2019; DOI: 10.1158/0008-5472.CAN-18-4032

SCFFBXW7/GSK3b Degrades GFI1 in Gastric Cancer

A HS746T BGC823 HS746T BGC823 Control 80 Control 15 *** HS746T BGC823 10 Control *** 40 HA-GFI1 Cell line: HA-GFI1 HA-GFI1 *** - - 8 *** 60 HA-GFI1 + + 30 * * 10 6 55 IB: GFI1 40 20 4 IB: HA 20 5 55 2 10 55 IB: Tubulin 0 0 Number of colons Relative cell growth 0246 0246810 0 0 Days post plating Days post plating Soft agar Cell growth

HGC27 B 25 CD 150 ** PLKO.1 HGC27 15 Control 20 * GFI1 KD shGFI1-1 ***

* 10 * PLKO.1 shGFI1-1 shGFI1-2 shGFI1-2 *

15 * 100 * 5 2.0 10 1.5 50 *** 5 1.0 0.5 0 Number of colons Relative cell growth 012345 0 0.0 Relative mRNA level Days post plating GFI1 GKN2

E F G SGC7901 HS746T Elisa 0.3 Promoter 150 *** 250 ** 800 ** * *** *** 200 600 0.2

100 SGC7901 150 400 Exon 0.1 100

% of Input 50 200

GKN2 (pg/mL) 50 Number of colons HS746T 0 0.0 0 0 Control GFI1 GFI1/GKN2

400 Control )

H 3 I Control 300 GFI1 GFI1 GFI1/GKN2 200 * * GFI1/GKN2

100 Tumor volume (mm 0 010203040 Days after inoculation

J n = 39 26 Case1 Case2 Case3 100 GKN2 positive GKN2 negative

50 Percentage

of patient samples (%) P = 0.0034 0 GKN2 GFI1 h

GFI1 highGFI1 low Gastric cancer Adjacent normal

Figure 2. GFI1 promotes gastric cancer cell proliferation and suppresses GKN2 expression. A, Immunoblot (IB; left), cell growth curve analysis (middle), and soft agar assays (right) of GFI1 overexpression gastric cancer cell lines. B, Cell growth curve analysis of HGC27 cells with GFI1 knockdown (GFI1 KD) or empty vector (PLKO.1). C, Representative images and statistical analysis of colony formation assays of indicated cell line. Scale bar, 1 cm. D, qPCR analysis of GFI1 and GKN2 mRNA levels of indicated HS746T cells. E, ELISA analysis of GKN2 protein levels in SGC7901 control or GFI1 knockdown cells. F, ChIP-qPCR analysis of GKN2 promoter region precipitated with indicated antibodies in HS746T cells. G, Statistical analysis and representative images of colony formation assays of SGC7901 and HS746T cells with indicated . Scale bar, 1 cm. H and I, HS746T cells with ectopic GFI1 expression or GFI1/GKN2 double expression were used for xenograft tumorigenesis assay. Seven nude mice were used for each group and tumor growth curves were generated by measuring on the indicated days. Unpaired Student t test was used and data are shown as the mean SEM. J, GFI1 and GKN2 levels were inversely correlated in clinical GC samples. A chi-square test was used for statistical analysis. Scale bar, 50 mm. , P < 0.05; , 0.001 < P < 0.01; , P < 0.001, represents statistically significant.

www.aacrjournals.org Cancer Res; 79(17) September 1, 2019 4391

Downloaded from cancerres.aacrjournals.org on September 25, 2021. © 2019 American Association for Cancer Research. Published OnlineFirst July 9, 2019; DOI: 10.1158/0008-5472.CAN-18-4032

Kuai et al.

IP A B SGC7901: C SGC7901: Flag-FBXW7: - + IgG GFI1 HA-GFI1: - + WCL 55 IB: GFI1 130 170 IB: FBXW7 Binding 70 IB: C-MYC 100 130 Pepde 55 FBXW7 protein 130 IP:Flag IB: Flag 55 IB: GFI1 100 PRMT1 13 100 55 IB: GFI1 70 HDAC2 9 55 IB: IgG 70 IB: C-MYC 55 IgG, GFI1 FBXW7 7 55 WCL 130 IB: Flag 100 40

D E F F-box: HA-GFI1: ++++ SGC7901: MG132: ++++ 130 IB: FBXW7 100 Flag-Cullin: - 100 +++++++ HA-GFI1: IP:HA 70 IB: Flag 55 IB: GFI1 +++++++ MG132: 55

IP:GFI1 100 IB:Flag 55 IB: IgG 75 40 IP:HA 55 IB:HA 55 IB: HA 100 IB: FBXW7 130 130 100 IB: Flag 100 55 IB: GFI1

WCL 75 70 WCL IB: Flag IB: HA WCL 55 55 IB: TUBULIN 55 40 55 IB: HA G Flag-F-box: H I HA-GFI1: ++++ GFI1 ++++ GFP: EV FBXW7 Flag-FBXW7 : - 55 IB: HA 0 0.51234 00.51 2 3 4 CHX(h) Flag-GFI1 : +++ 130 IB: Flag-GFI1 55 100 55 IB:GFI1 130 70 IB: Flag IB: Flag-FBW7 130 100 55 IB:FBXW7 70 40 100 IB: C-MYC 55 55 IB: TUBULIN 55 IB: TUBULIN IB: TUBULIN 55 30 IB: GFP

J K L M

NC siRNA: 1.5 Control Peng et al. GSE19826 Liu et al. GSE54129 ** NC siFBW7 1 55 IB: GFI1 ** *** 8 *** siFBW7 2 3.0 1.0 130 7 IB: FBXW7 2.8 100 6 FBXW7 2.6

0.5 FBXW7 70 R = 0.6838 R = 0.4278 IB: C-MYC P 5 P 2.4 < 0.0001 < 0.0001 55 n = 27 n = 132

Relative mRNA level 0.0 0246 51015 55 IB: TUBULIN GKN2 GKN2 FBXW7 GFI1 GKN2

Figure 3. FBXW7 interacts with GFI1 and governs its stability. A, Coomassie blue staining of anti-HA-GFI1 IP. B, Immunoblot (IB) analysis of whole-cell lysates (WCL) and IP from SGC7901 cells transfected with Flag-FBXW7a or empty vector in the presence of MG132. C, Endogenous co-IP experiments in SGC7901 cells were performed with the indicated antibodies. D, Endogenous co-IP experiments in SGC7901 cells pretreated with FBXW7 siRNA or control siRNA (NC) were performed with the indicated antibodies. E, Immunoblot analysis of WCL and IP of 293T cells transfected with HA-GFI1 and different Cullin constructs and treated with MG132. F, Immunoblot analysis of whole-cell lysates and IP of 293T cells transfected with HA-GFI1 and different F-box and treated with MG132. G, Immunoblot analysis of whole-cell lysates of 293T cells transfected with HA-GFI1 and different F-box constructs. H, 293T cells were transfected with GFI1 and FBXW7a constructs as indicated and treated with 100 mg/mL of cycloheximide (CHX) before whole-cell lysates were collected at the indicated time points for immunoblot analysis. I, Immunoblot analysis of 293T cells transfected with Flag-GFI1 and wild-type FBXW7a (WT) or FBXW7a-R465C mutant. J and K, Immunoblot analysis or qPCR analysis of BGC823 cells treated with indicated siRNAs. L, The association (by Pearson) between FBXW7 and GKN2 transcripts was observed in public data sets. Unpaired Student t test was used. M, Model of FBXW7–GFI1–GKN2 axis. Data are shown as the mean SEM. , 0.01 < P < 0.05; , 0.001 < P < 0.01; , P < 0.001.

4392 Cancer Res; 79(17) September 1, 2019 Cancer Research

Downloaded from cancerres.aacrjournals.org on September 25, 2021. © 2019 American Association for Cancer Research. Published OnlineFirst July 9, 2019; DOI: 10.1158/0008-5472.CAN-18-4032

SCFFBXW7/GSK3b Degrades GFI1 in Gastric Cancer

A B C DMSO GSK690693 Inhibitor: - - 0 1 CHX(hr) 24 HA-GFI1 : + +++ 55 IB:HA 55 IB:HA-GFI1 Flag-GSK3β(CA): - + + + 70 IB:C-MYC IB: HA 55 55 55 IB:TUBULIN 70 55 IB:p-AKT (S473) CHIR-99021 IB: Flag 55 40 0124CHX(hr) IB:p-GSK3β (S9) 40 IB:TUBULIN 55 55 IB:HA-GFI1 40 IB:p-ERK1/2 35 IB: GFP 55 IB:TUBULIN 25 55 IB:TUBULIN

D E F Flag-GFI1: +++ 1.5 1.5 NC Flag-GSK3β(CA): - ++ siGSK3β LiCI: --+ siRNA: *** *** siGSK3β/ siGFI1 1.0 1.0 *** IB:Flag-GFI1 55 55 IB:GFI1 55 0.5 0.5 IB:Flag-GSK3β(CA) IB:GSK3β ** 40 40 Relative mRNA level 35 level mRNA Relative IB:GFP 55 IB:TUBULIN 0.0 0.0 25 GSK3β GFI1 GKN2 GSK3β GFI1 GKN2 IB:TUBULIN 55

G Case1 Case2 Case3 Case4 n = 89 87 100 P=0.019 GFI1 80 GFI1 High 60 GFI1 Low 40 Percentage 20 (Y216) of patient sample (%)

0 P-GSK3β P-GSK3β High Low (Y216)

H I J K

- - Flag-GSK3β(CA): _ Flag-GFI1: ++ Inhibitor: Flag-GFI1: +++ Flag-GFI1: +++ HA-FBXW7: + + HA-GFI1 : ++++ - - Flag-FBXW7: - ++ HA-FBXW7: +++ LiCI: + Flag-FBXW7: +++ LiCI: - - + MG132: +++ MG132: + + GFP: + + ++ 130 IB:Flag-GFI1 IB:Flag-GFI1 IB: HA 55 IB: HA 55 55 100 130 130 130 IB: Flag IB: Flag IB:Flag-FBXW7 IP:H A IB:HA IP: Flag 55 100 100 100 35 130 55 IB:TUBULIN IB:GFP 55 IB: HA IB: Flag 100 25 40 35 WCL

WCL IB: GFP 130 55 IB: Flag 55 IB:TUBULIN IB: HA 25 100

Figure 4. GSK3b destabilizes GFI1 and negatively regulates GFI1 expression. A and B, Immunoblot (IB) analysis of whole-cell lysates from 293T cells transfected with HA-GFI1 and treated with the indicated chemicals. GSK690693 (AKT inhibitor), 10 mmol/L; CHIR-99021 (GSK3b inhibitor), 10 mmol/L; UO126 (Erk1/2 inhibitor), 10 mmol/L in A. B, One hundred mg/mL of cycloheximide (CHX) in the presence or absence of 10 mmol/L CHIR-99021. C and D, Immunoblot analysis of whole-cell lysates from 293T cells transfected with indicated plasmids treated with MG132, CHIR-99021 (10 mmol/L), or LiCl (20 mmol/L). E, Immunoblot and qPCR analysis of BGC823 cells transfected with siRNAs targeting GSK3b or control siRNA. F, qPCR of the indicated gene mRNA levels in HS746T cells. Unpaired Student t test was used for E and F and data are shown as the mean SEM. Results are from n ¼ 3 independent experiments. G, Significant inverse correlation (P ¼ 0.019) between p-GSK3b (Y216) and GFI1 levels was determined by a chi-square test in clinical gastric cancer samples. Scale bar, 50 mm. H, Immunoblot analysis of whole-cell lysates (WCL) and IP from 293T cells transfected with the indicated plasmids. I, Immunoblot analysis of whole-cell lysates and IP from 293T cells transfected with the indicated plasmids and treated with MG132 (10 mmol/L) or LiCl (20 mmol/L). J–K, Immunoblot analysis of whole-cell lysates from 293T cells transfected with the indicated plasmids and treated with MG132 (10 mmol/L), CHIR-99021 (10 mmol/L), or LiCl (20 mmol/L). , 0.001 < P < 0.01; , P < 0.001, represents statistically significant.

dramatically prolonged after CHIR99021 treatment (Fig. 4B), and BGC823 gastric cancer cells caused marked upregulation of GFI1 cotransfection with GSK3b significantly decreased GFI1 levels, protein levels, but not GFI1 transcripts, and resulted in suppressed which was effectively blocked by MG132 and GSK3 inhibitors GKN2 mRNA levels (Fig. 4E). Knockdown of GFI1 restored GKN2 (Fig. 4C and D). However, depletion of endogenous GSK3b in mRNA levels in GSK3b-depleted cells (Fig. 4F), further supporting

www.aacrjournals.org Cancer Res; 79(17) September 1, 2019 4393

Downloaded from cancerres.aacrjournals.org on September 25, 2021. © 2019 American Association for Cancer Research. Published OnlineFirst July 9, 2019; DOI: 10.1158/0008-5472.CAN-18-4032

Kuai et al.

a master regulatory role of GSK3b in modulating the GFI1–GKN2 indicated in Fig. 6A, although mRNA levels of ectopic GFI1-WT axis. Next, we examined the association between GSK3b activity and GFI1-S94A/S98A were quite similar, a much higher level of (represented by the pY216 signal) and GFI1 expression in 176 GFI1 protein product was observed in GFI1-S94A/S98A– clinical gastric cancer tissues, and found that active GSK3b is expressing cells, which resulted in a lower level of endogenous significantly correlated with a low GFI1 level and better overall GKN2 transcription and more colonies in the anchorage- survival of patients with gastric cancer (Fig. 4G; Supplementary independent soft agar assay in comparison with GFI1-WT– Fig. S3C). In support of an essential role of GSK3b in FBXW7- expressing and parental control gastric cancer cells (Fig. 6B and mediated GFI1 regulation, coexpression of GSK3b increased C). We performed xenograft tumorigenesis experiments by inoc- GFI1–FBXW7a interaction, whereas CHIR-99021/LiCl treatment ulating HS746T gastric cancer cells expressing wild-type GFI1 drastically blocked such interaction and FBXW7a-caused GFI1 (WT) and GFI1-S94A/S98A mutant in the flanks of nude mice reduction (Fig. 4H–K). Therefore, the results outlined above and used parental HS746T gastric cancer cells as the control. suggested that GFI1 protein stability is stringently regulated by Although all 3 groups of nude mice developed tumors, mice GSK3b in gastric cancer cells. implanted cells expressing nondegradable GFI1-S94A/S98A mutant developed the largest tumors (Fig. 6D–F), implicating a GSK3b-dependent S94/S98 phosphorylation mediates GFI1– potentially important functional role of deregulated GFI1 via FBXW7 interaction inactivation of FBXW7 or GSK3b in gastric cancer (Fig. 6G). Taken SCFFBXW7 has been shown to interact and degrade multiple key together, these results suggest that blockade of GFI1 protein proteins, including C-JUN, C-MYC, MCL1, and others after phos- degradation was a tumor promoting factor of gastric cancer and phorylation by GSK3b at certain sites (namely phosphodegron; efficient degradation of GFI1 by the FBXW7/GSK3b pathway may ref. 24). To gain a mechanistic insight of how GSK3b modulates be crucial for suppressing gastric cancer progression. GFI1, we analyzed the GFI1 sequence by mass spectrometry and identified S94/S98 as the putative GSK3b phosphorylation site matching the classic FBXW7 phosphodegron sequence (Fig. 5A; Discussion Supplementary Fig. S4A). We then generated GST-fused recombi- GFI1 has been shown to be a central regulator downstream of nant GFI1 proteins (wild-type and mutants) and incubated the RUNX1 and , and by recruiting important epigenetic regulators, proteins with GSK3b in the presence of [g-32P]-ATP to perform an such as LSD1 and COREST, GFI1 plays a crucial role in the early in vitro kinase assay. Indeed, mutation of either S94 or S98 to development of the hematopoieticlineage (36). The GFI1 mutation alanine (labeled as S94A and S98A) significantly reduced the in the human genome is associated with defective hematopoiesis. incorporated [g-32P]-ATP signal, and double mutation of both Although overexpression of GFI1 has been characterized in malig- sites almost completely abolished the phosphorylation by GSK3b nancies, including AML (38) and medullblastoma (39), the func- (Fig. 5B), suggesting that S94 and S98 are indeed the major GSK3b tion and regulation of GFI1 in most solid tumors are largely phosphorylation sites of GFI1. Mutation of S94 and S98 to alanine unknown. In this study we first found a significant correlation largely reduced the polyubiquitination of GFI1, and S94A and between high GFI1 expression and advanced gastric cancer, which S98A mutants were resistant to ectopic GSK3b-mediated degrada- predicts a worse prognosis (Fig. 1). Furthermore, the expression of tion (Fig. 5C and D). Furthermore, in support of an essential role GFI1 promotes gastric cancer cell proliferation, a process that occurs for S94 and S98 in SCFFBXW7-mediated GFI1 degradation, S94A at least in part via suppressing GKN2 transcription. Functionally, and S98A mutations abolished the FBXW7a–GFI1 interaction and GKN2 has been reported as an antiproliferation factor, and results ectopic FBXW7a-mediated degradation and resulted in a pro- from a cell-based study and transgenic mice model suggested that longed GFI1 half-life (Fig. 5E–G). To further strengthen a causal GKN2 overexpression suppresses the oncogenic potential of gastric relationship between GSK3b-mediated GFI1 phosphorylation cancer cells and inhibits tumor growth in vivo. Importantly, and the FBXW7aGFI1 interaction, we first phosphorylated although GKN2 protein was absent in 85% of diffuse and 54% beads-bound recombinant GST-GFI1 (wild-type and S94A/S98A of intestinal types of gastric cancer, and was associated with poorer mutant) with GSK3b kinase in the presence of ATP, then incubated survival; somatic mutation was rarely found in the coding region of the resulting beads with Flag-FBXW7 (wild-type and R465C mutant, the GKN2 allele (34). Therefore, GKN2 is mainly suppressed by a respectively) containing cell lysate to perform a GST-pulldown assay. transcriptional/epigenetic mechanism in gastric cancer. In this AsshowninFig.5H,onlyGSK3b pretreated wild-type GST-GFI1 study, we have characterized GFI1 as a major suppressor of GKN2, protein, but not the S94A/S98A mutant, effectively pulled down the depletion of which caused an approximate 10-fold upregula- wild-type Flag-FBXW7a, but not the Flag-FBXW7-R465C mutant. tion of endogenous GKN2 transcription (Fig. 2D). More impor- Collectively, these results indicated that phosphorylation of GFI1 at tantly, we further found that depletion of GSK3b by siRNA caused a S94/S98 sites by GSK3b is both necessary and sufficient to trigger the dramatic decrease in GKN2 mRNA (Fig. 4E and F), which may be GFI1–FBXW7 interaction and subsequent protein degradation. caused by upregulation of GFI1 protein. Therefore, our finding provided a potential link between decreased expression of the Nondegradable GFI1 is more potent in promoting tumor suppressor, GKN2, and genetic lesions of the RTK–PI3K/ tumorigenesis of gastric cancer cells PTEN pathway, which are often observed in gastric cancer and may As outlined above, GSK3b-mediated GFI1 S94/S98 phosphor- cause inhibition of downstream GSK3b activity (13). ylation was crucial for GFI1 stability control, and high GFI1 FBXW7 is among the most well-studied F-box proteins that expression was correlated with a poorer clinical outcome of form the SCF E3 ligase complex to degrade numerous sub- patients with gastric cancer. To gain an understanding of how strates (27). By specifically recognizing the "phosphodegron" the blockage of GFI1 degradation promotes the oncogenic poten- motif of substrates through the WD40 domain that forms a b tial of gastric cancer cells, we generated HS746T and BGC823 cells propeller structure, FBXW7 only binds substrates after the sub- that stably express wild-type GFI1 and S94A/S98A mutants. As strates are properly phosphorylated at specific threonine or serine

4394 Cancer Res; 79(17) September 1, 2019 Cancer Research

Downloaded from cancerres.aacrjournals.org on September 25, 2021. © 2019 American Association for Cancer Research. Published OnlineFirst July 9, 2019; DOI: 10.1158/0008-5472.CAN-18-4032

SCFFBXW7/GSK3b Degrades GFI1 in Gastric Cancer

A B C GST-GFI1: - HA-GFI1: - --- FBXW7 GST: + Myc-ub: ++ S/T PPLSP consensus His-GSK3β: +++++ MG132: ++ 100 GFI1 92 PPSPSASP 99 170 70 130 IB: Myc C-MYC 56 LPTPPLSP 63 100 CCNE1 378 LLTPPQSG 385 55 * HA 70

IP: 55 C-JUN 237 GETPPLSP 244 40 100 55 IB:HA H.sapiens 88 PPSPSASPA 100 70 M.musculus 89 PPSPSVSPA 101 55 170 M.mulatta 88 PPSPSASPA 100 40 130 D.rerio 93 PPSPSASPE 105 35 100 IB:Myc X.tropicalis 85 PPSPSVSPA 97 WCL 70 25 55 55 IB: HA

D E

HA-GFI1 : HA-GFI1 : ---- GSK3β(CA): - ++++--- Flag-FBXW7: ++ + + GFP: ++++++++ GFP: ++++++ ++ IB: HA 55 55 IB: HA 55 130 IB: GSK3β 40 100 IB: Flag

IB:TUBULIN 55 55 IB:TUBULIN 35 35 IB: GFP IB: GFP 25 25

F G HA-GFI1: HA-FBXW7 Flag-FBXW7: ++++ Flag-GSK3β(CA) MG132: ++++ WT S94A/S98A HA-GFI1 0 12468 012468 CHX(h) 55 IB: HA IB:HA-GFI1 Flag 55 130 IB: Flag

IP: 130 100 IB:HA-FBXW7 100 130 IB: Flag 100 IB:Flag 40 55 IB: HA 55 IB:TUBULIN WCL 55 IB:TUBULIN

H Purifying GST-GFI1 and 293T cells transfected GST-GFI1-WT: + + + --- His-GSK3β from BL21 E.coli with Flag-FBXW7 GST-GFI1-SSAA: ---+ + + Flag-FBXW7-WT: + ++- + - Whole cell lysis Flag-FBXW7-R465C: --+ --+ (+ATP, 37°C 2h) His-GSK3β: - + + - + + 5% Gst pulldown: 130 IB:Flag input 100 incubation for 4 hours 130 Input: IB:Flag 100 GST pulldown 100 In vitro Gst-GFI1 70 kinase assay Western blot 55 His-GSK3β

Figure 5. GSK3b phosphorylates S94/S98 sites of GFI1 to recruit FBXW7. A, Sequence alignment of GFI1 with the known phosphodegron sequences recognized by FBXW7. Ser94 and Ser98 sites were conserved in several species (bottom). B, Two mg recombinant GST-GFI1 or GST proteins were incubated with purified GSK3b for in vitro kinase assays in the presence of g-32P-ATP. The reactions were resolved by SDS-PAGE and detected by autoradiography. C–F, Immunoblot (IB) analysis of whole-cell lysates (WCL) and IP from 293T cells transfected with the indicated plasmids. G, 293T cells were transfected with wild-type GFI1 or GFI1-S94A/S98A mutant and HA-FBXW7a/Flag-GSK3b constructs. Cells were treated with 100 mg/mL of cycloheximide (CHX) before whole-cell lysates were collected at the indicated time points for immunoblot analysis. H, The flow chart for in vitro kinase pulldown assays (right). In vitro purified protein GST-GFI1 (WT and mutants) was phosphorylated by His-GSK3b first, then phosphorylated GST-GFI1 was added to pull down Flag-FBXW7a prepurified form 293T cells.

residues. First discovered as CDC4 in Drosophila, FBXW7 was that FBXW7 governs the destruction of crucial oncogenic factors, characterized as the F-box protein responsible for cell-cycle reg- including C-MYC (20, 21), C-JUN (23), MCL1 (24), and many ulation by degrading CCNE1, and subsequent studies revealed others (22). Thus, FBXW7 is proposed to be a putative tumor

www.aacrjournals.org Cancer Res; 79(17) September 1, 2019 4395

Downloaded from cancerres.aacrjournals.org on September 25, 2021. © 2019 American Association for Cancer Research. Published OnlineFirst July 9, 2019; DOI: 10.1158/0008-5472.CAN-18-4032

Kuai et al.

A HS746T BGC823 B HS746T BGC823 1.5 ** *** Control 500 * *** 300 *** Control - - GFI1-WT *** GFI1-WT HA-GFI1: GFI1-S94A/S98A 400 GFI1-S94A/S98A 1.0 200 300 55 IB:HA

200 0.5 100 55 IB: GFI1 100

IB:TUBULIN Relative mRNA level Relative GFI1 mRNA 55 0 0 0.0 GKN2

C Control GFI1-WT GFI1-S94A/S98A HS746T BGC823 20 50 * * ** 15 40 **

HS746T 30 10 20 5 10

Number of colony 0 0 BGC823

D E 1,500

) Control HS746T 3 GFI1-WT wild type 1,000 GFI1-S94A/S98A HS746T GFI1-WT

500 HS746T GFI1 S94A/S98A Tumor volume (mm 0 0 20 40 60 Days after inoculation

F G Wild type GFI1 GFI1 S94A/S98A

Figure 6. Nondegradable GFI1 mutant is more potent in driving tumorigenesis. A, qPCR (left) and immunoblot (IB) analysis (right) of gastric cancer cells with ectopic GFI1 expression. B, qPCR analysis of GKN2 transcription in HS746T cells with ectopic GFI1 expression. C, Representative images and quantification of soft agar assays of gastric cancer cells with ectopic GFI1 expression. Scale bar, 20 mm. D–F, Xenograft tumorigenesis of HS746T cells with ectopic GFI1 expression. Seven nude mice were used for each group and tumor growth curves were generated by measuring every 3 days (E). Scale bar, 1 cm. For D and E, unpaired Student t test was used and data are shown as the mean SEM. G, Model graph of how deregulated GFI1 may contribute to gastric tumorigenesis. GFI1 protein is stringently governed by SCFFBXW7/GSK3b so that GKN2 expresses and helps maintain homeostasis in healthy gastric tissues. When FBXW7 is mutated/downregulated or GSK3b is inhibited by elevated RTK signaling, GFI1 degradation is blocked and suppresses GKN2 transcription to promote gastric cancer cell proliferation and disease progression. , P < 0.05; , P < 0.01; , P < 0.001, represents statistically significant.

suppressor, and indeed genetic variations, including mutations approximate 6% mutation frequency was estimated with all and deletions, have been frequently observed in FBXW7 alleles. cancer samples tested with the highest mutation rate (>30%) For example, in a comprehensive mutation analysis of observed in cholangiocarcinomas and T-cell acute lymphocytic FBXW7 genetic status in 534 human cancer samples, an overall leukemia (40). Interestingly, 43% of all mutations occurred at the

4396 Cancer Res; 79(17) September 1, 2019 Cancer Research

Downloaded from cancerres.aacrjournals.org on September 25, 2021. © 2019 American Association for Cancer Research. Published OnlineFirst July 9, 2019; DOI: 10.1158/0008-5472.CAN-18-4032

SCFFBXW7/GSK3b Degrades GFI1 in Gastric Cancer

"hotspot" Arg residues responsible for binding the phosphode- serve as an additional indicative marker for usage of such ther- gron sequence of substrates. In another analysis of The Cancer apeutic agents to treat patients with gastric cancer. Collectively, Genome Atlas data, including 219 stomach cancer samples, the our study has characterized the oncogenic function and regulatory hotspot mutation rate was approximately 5 % (41), whereas the mechanism underlying high GFI1expression in gastric cancer and overall mutation rate of FBXW7 was estimated to be 9.2% to may contribute to a better understanding and clinical treatment of 18.5% in gastric cancer (15). Therefore, FBXW7 is also an impor- gastric cancer. tant tumor suppressor in gastric cancer. In addition to genetic mutations, the expression of FBXW7 is targeted by miRNAs, such Disclosure of Potential Conflicts of Interest as miR-223 (42) and miR-25 (43), and low FBXW7 expression is J. Li has ownership Interest (including stock, patents, etc.) for OrigiMed Inc. associated with poor prognosis and chemoresistance. Although No potential conflicts of interest were disclosed by the other authors. BRG1 has been identified as a key FBXW7 substrate modulating gastric cancer metastasis recently (18), little is known about how Authors' Contributions low expression or mutation of FBXW7 contributes to disease- Conception and design: X. Kuai, L. Li, Z. Wang, D. Gao Development of methodology: X. Kuai, L. Li, D. Gao specific morphology and progressive features of gastric cancer. In Acquisition of data (provided animals, acquired and managed patients, this study, we have established GFI1 as a ubiquitination substrate provided facilities, etc.): X. Kuai, L. Li, R. Chen, K. Wang, M. Chen, B. Cui, of FBXW7, and our mechanistic study further connected FBXW7 H. Zhu, H. Zhou, J. Qin, Z. Wang loss to suppression of the GFI1 downstream target gene, GKN2, Analysis and interpretation of data (e.g., statistical analysis, biostatistics, which is a gastric-specific tumor suppressor (34). Therefore, our computational analysis): X. Kuai, L. Li, Y. Zhang, J. Huang, J. Qin, D. Gao work further supports the tumor suppressor function of FBXW7 in Writing, review, and/or revision of the manuscript: L. Li, W. Wei, D. Gao Administrative, technical, or material support (i.e., reporting or organizing gastric cancer and provides new insight into how the loss of data, constructing databases): L. Li, J. Li, J. Huang, D. Gao FBXW7 promotes gastric cancer progression via inhibition of the Study supervision: J. Li, D. Gao gastric tissue-specific tumor suppressor, GKN2. As a very lethal type of cancer, gastric cancer is associated with Acknowledgments bad outcomes and there are still very few therapeutic options for We thank the Animal Facility of Shanghai Institute of Biochemistry and Cell gastric cancer. Based on the result from a systematic study from the Biology for assistance in the animal study. This work was supported by Ministry Asian Cancer Research Group with a large cohort analyzed, of Science and Technology of China (Grant No. 2015CB964502), Strategic approximately 20% of gastric cancer cases were carrying ampli- Priority Research Program of the Chinese Academy of Sciences (Grant Nos. fications or mutations in RTK genes, and about 18% cases were XDB19020203 and XDA12020364), and National Natural Science Foundation found with PIK3CA or PTEN mutations (44). Therefore, patients of China (Grant Nos. 81790253 and 91853130). with gastric cancer with RTK mutations or impaired FBXW7 The costs of publication of this article were defrayed in part by the function may potentially suffer from subsequent accumulation payment of page charges. This article must therefore be hereby marked of the GFI1 protein, which in turn would further contribute to the advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate rapid proliferation of gastric cancer cells and promote disease this fact. progression. Because multiple RTK targeting drugs have been approved for gastric cancer treatment, such as bevacizumab and Received December 21, 2018; revised May 24, 2019; accepted July 3, 2019; apatinib, it is possible that an elevated GFI1 protein level could published first July 9, 2019.

References 1. Van Cutsem E, Sagaert X, Topal B, Haustermans K, Prenen H. Gastric cancer. 10. Spooner CJ, Lesch J, Yan D, Khan AA, Abbas A, Ramirez-Carrozzi V, et al. Lancet 2016;388:2654–64. Specification of type 2 innate lymphocytes by the transcriptional deter- 2. Torre LA, Bray F, Siegel RL, Ferlay J, Lortet-Tieulent J, Jemal A. Global cancer minant Gfi1. Nat Immunol 2013;14:1229–36. statistics, 2012. CA Cancer J Clin 2015;65:87–108. 11. Northcott PA, Lee C, Zichner T, Stutz AM, Erkek S, Kawauchi D, et al. 3. Chen W, Zheng R, Baade PD, Zhang S, Zeng H, Bray F, et al. Cancer statistics Enhancer hijacking activates GFI1 family oncogenes in medulloblastoma. in China, 2015. CA Cancer J Clin 2016;66:115–32. Nature 2014;511:428–34. 4. Uemura N, Okamoto S, Yamamoto S, Matsumura N, Yamaguchi S, 12. Wadhwa R, Song S, Lee JS, Yao Y, Wei Q, Ajani JA. Gastric cancer-molecular Yamakido M, et al. Helicobacter pylori infection and the development of and clinical dimensions. Nat Rev Clin Oncol 2013;10:643–55. gastric cancer. N Engl J Med 2001;345:784–9. 13. Byun DS, Cho K, Ryu BK, Lee MG, Park JI, Chae KS, et al. Frequent 5. Roberts T, Cowell JK. Cloning of the human Gfi-1 gene and its mapping to monoallelic deletion of PTEN and its reciprocal associatioin with chromosome region 1p22. Oncogene 1997;14:1003–5. PIK3CA amplification in gastric carcinoma. Int J Cancer 2003;104: 6. Schmidt T, Karsunky H, Gau E, Zevnik B, Elsasser HP, Moroy T, et al. Zinc 318–27. finger protein GFI-1 has low oncogenic potential but cooperates strongly 14. Franke TF, Kaplan DR, Cantley LC. PI3K: downstream AKTion blocks with pim and myc genes in T-cell lymphomagenesis. Oncogene 1998;17: apoptosis. Cell 1997;88:435–7. 2661–7. 15. Huang KK, Ramnarayanan K, Zhu F, Srivastava S, Xu C, Tan ALK, et al. 7. Lis R, Karrasch CC, Poulos MG, Kunar B, Redmond D, Duran JGB, et al. Genomic and epigenomic profiling of high-risk intestinal metaplasia Conversion of adult endothelium to immunocompetent haematopoietic reveals molecular determinants of progression to gastric cancer. stem cells. Nature 2017;545:439–45. Cancer Cell 2018;33:137–50e5. 8. Hock H, Hamblen MJ, Rooke HM, Schindler JW, Saleque S, Fujiwara Y, et al. 16. Yokobori T, Mimori K, Iwatsuki M, Ishii H, Onoyama I, Fukagawa T, et al. Gfi-1 restricts proliferation and preserves functional integrity of haemato- p53-Altered FBXW7 expression determines poor prognosis in gastric cancer poietic stem cells. Nature 2004;431:1002–7. cases. Cancer Res 2009;69:3788–94. 9. Chalmin F, Mignot G, Bruchard M, Chevriaux A, Vegran F, Hichami A, et al. 17. Li H, Wang Z, Zhang W, Qian K, Xu W, Zhang S. Fbxw7 regulates tumor Stat3 and Gfi-1 transcription factors control Th17 cell immunosuppressive apoptosis, growth arrest and the epithelial-to-mesenchymal transition in activity via the regulation of ectonucleotidase expression. Immunity 2012; part through the RhoA signaling pathway in gastric cancer. Cancer Lett 36:362–73. 2016;370:39–55.

www.aacrjournals.org Cancer Res; 79(17) September 1, 2019 4397

Downloaded from cancerres.aacrjournals.org on September 25, 2021. © 2019 American Association for Cancer Research. Published OnlineFirst July 9, 2019; DOI: 10.1158/0008-5472.CAN-18-4032

Kuai et al.

18. Huang LY, Zhao J, Chen H, Wan L, Inuzuka H, Guo J, et al. SCF(FBW7)- GKN2 in gastric cancer: influence of tumor histology and relationship to mediated degradation of Brg1 suppresses gastric cancer metastasis. prognosis. Clin Cancer Res 2008;14:4161–7. Nat Commun 2018;9:3569. 33. Menheniott TR, O'Connor L, Chionh YT, Dabritz J, Scurr M, Rollo BN, et al. 19. Jiang Y, Qi X, Liu X, Zhang J, Ji J, Zhu Z, et al. Fbxw7 haploinsufficiency loses Loss of gastrokine-2 drives premalignant gastric inflammation and tumor its protection against DNA damage and accelerates MNU-induced gastric progression. J Clin Invest 2016;126:1383–400. carcinogenesis. Oncotarget 2017;8:33444–56. 34. Ouyang J, Pan X, Lin H, Hu Z, Xiao P, Hu H. GKN2 increases apoptosis, 20. Welcker M, Orian A, Jin J, Grim JE, Harper JW, Eisenman RN, et al. The reduces the proliferation and invasion ability of gastric cancer cells through Fbw7 tumor suppressor regulates glycogen synthase kinase 3 phosphor- down-regulating the JAK/STAT signaling pathway. Am J Transl Res 2017;9: ylation-dependent c-Myc protein degradation. Proc Natl Acad Sci U S A 803–11. 2004;101:9085–90. 35. Vadnais C, Chen R, Fraszczak J, Yu Z, Boulais J, Pinder J, et al. GFI1 21. Yada M, Hatakeyama S, Kamura T, Nishiyama M, Tsunematsu R, Imaki H, facilitates efficient DNA repair by regulating PRMT1 dependent methyla- et al. Phosphorylation-dependent degradation of c-Myc is mediated by the tion of MRE11 and 53BP1. Nat Commun 2018;9:1418. F-box protein Fbw7. EMBO J 2004;23:2116–25. 36. Saleque S, Kim J, Rooke HM, Orkin SH. Epigenetic regulation of hemato- 22. Moberg KH, Bell DW, Wahrer DC, Haber DA, Hariharan IK. Archipelago poietic differentiation by Gfi-1 and Gfi-1b is mediated by the cofactors regulates cyclin E levels in Drosophila and is mutated in human cancer cell CoREST and LSD1. Mol Cell 2007;27:562–72. lines. Nature 2001;413:311–6. 37. Chang CC, Lin HH, Lin JK, Lin CC, Lan YT, Wang HS, et al. FBXW7 23. Nateri AS, Riera-Sans L, Da Costa C, Behrens A. The ubiquitin ligase mutation analysis and its correlation with clinicopathological features SCFFbw7 antagonizes apoptotic JNK signaling. Science 2004;303:1374–8. and prognosis in colorectal cancer patients. Int J Biol Markers 2015;30: 24. Inuzuka H, Shaik S, Onoyama I, Gao D, Tseng A, Maser RS, et al. SCF e88–95. (FBW7) regulates cellular apoptosis by targeting MCL1 for ubiquitylation 38. Khandanpour C, Thiede C, Valk PJ, Sharif-Askari E, Nuckel H, Lohmann D, and destruction. Nature 2011;471:104–9. et al. A variant allele of Growth Factor Independence 1 (GFI1) is associated 25. Frame S, Cohen P, Biondi RM. A common phosphate binding site explains with acute myeloid leukemia. Blood 2010;115:2462–72. the unique substrate specificity of GSK3 and its inactivation by phosphor- 39. Zarebski A, Velu CS, Baktula AM, Bourdeau T, Horman SR, Basu S, et al. ylation. Mol Cell 2001;7:1321–7. Mutations in growth factor independent-1 associated with human neu- 26. Nakayama M, Hisatsune J, Yamasaki E, Isomoto H, Kurazono H, Hata- tropenia block murine granulopoiesis through colony stimulating factor-1. keyama M, et al. Helicobacter pylori VacA-induced inhibition of GSK3 Immunity 2008;28:370–80. through the PI3K/Akt signaling pathway. J Biol Chem 2009;284:1612–9. 40. Davis H, Lewis A, Behrens A, Tomlinson I. Investigation of the atypical 27. Welcker M, Clurman BE. FBW7 ubiquitin ligase: a tumour suppressor at the FBXW7 mutation spectrum in human tumours by conditional expression crossroads of cell division, growth and differentiation. Nat Rev Cancer of a heterozygous propellor tip missense allele in the mouse intestines. Gut 2008;8:83–93. 2014;63:792–9. 28. Baus-Loncar M, Lubka M, Pusch CM, Otto WR, Poulsom R, Blin N. 41. Davis RJ, Welcker M, Clurman BE. Tumor suppression by the Fbw7 Cytokine regulation of the trefoil factor family binding protein GKN2 ubiquitin ligase: mechanisms and opportunities. Cancer Cell 2014;26: (GDDR/TFIZ1/blottin) in human gastrointestinal epithelial cells. 455–64. Cell Physiol Biochem 2007;20:193–204. 42. Mansour MR, Sanda T, Lawton LN, Li X, Kreslavsky T, Novina CD, et al. The 29. Menheniott TR, Kurklu B, Giraud AS. Gastrokines: stomach-specific pro- TAL1 complex targets the FBXW7 tumor suppressor by activating miR-223 teins with putative homeostatic and tumor suppressor roles. Am J Physiol in human T cell acute lymphoblastic leukemia. J Exp Med 2013;210: Gastrointest Liver Physiol 2013;304:G109–21. 1545–57. 30. Du JJ, Dou KF, Peng SY, Wang WZ, Wang ZH, Xiao HS, et al. [Down- 43. Li Z, Sun Y, Chen X, Squires J, Nowroozizadeh B, Liang C, et al. p53 regulated full-length novel gene GDDR and its effect on gastric cancer]. mutation directs AURKA overexpression via miR-25 and FBXW7 in pros- Zhonghua Yi Xue Za Zhi 2003;83:1166–8. tatic small cell neuroendocrine carcinoma. Mol Cancer Res 2015;13: 31. Yoshikawa Y, Mukai H, Hino F, Asada K, Kato I. Isolation of two novel 584–91. genes, down-regulated in gastric cancer. Jpn J Cancer Res 2000;91:459–63. 44. Cristescu R, Lee J, Nebozhyn M, Kim KM, Ting JC, Wong SS, et al. Molecular 32. Moss SF, Lee JW, Sabo E, Rubin AK, Rommel J, Westley BR, et al. Decreased analysis of gastric cancer identifies subtypes associated with distinct clinical expression of gastrokine 1 and the trefoil factor interacting protein TFIZ1/ outcomes. Nat Med 2015;21:449–56.

4398 Cancer Res; 79(17) September 1, 2019 Cancer Research

Downloaded from cancerres.aacrjournals.org on September 25, 2021. © 2019 American Association for Cancer Research. Published OnlineFirst July 9, 2019; DOI: 10.1158/0008-5472.CAN-18-4032

SCFFBXW7/GSK3β-Mediated GFI1 Degradation Suppresses Proliferation of Gastric Cancer Cells

Xiaoling Kuai, Long Li, Ran Chen, et al.

Cancer Res 2019;79:4387-4398. Published OnlineFirst July 9, 2019.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-18-4032

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2019/07/09/0008-5472.CAN-18-4032.DC1

Cited articles This article cites 44 articles, 10 of which you can access for free at: http://cancerres.aacrjournals.org/content/79/17/4387.full#ref-list-1

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 Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/79/17/4387. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.

Downloaded from cancerres.aacrjournals.org on September 25, 2021. © 2019 American Association for Cancer Research.