Am J Cancer Res 2020;10(4):1238-1254 www.ajcr.us /ISSN:2156-6976/ajcr0108129

Original Article SH3BGRL2 exerts a dual function in breast cancer growth and metastasis and is regulated by TGF-β1

Dou-Dou Li1,2,3, Ling Deng1, Shu-Yuan Hu1, Fang-Lin Zhang1,2,3, Da-Qiang Li1,2,3,4

1Shanghai Cancer Center and Institutes of Biomedical Sciences, Fudan University, Shanghai 200032, China; 2Cancer Institute, 3Department of Oncology, 4Shanghai Key Laboratory of Breast Cancer, Shanghai Medical College, Fudan University, Shanghai 200032, China Received January 19, 2020; Accepted March 9, 2020; Epub April 1, 2020; Published April 15, 2020

Abstract: SH3 domain-binding glutamic acid-rich-like 2 (SH3BGRL2) is a poorly defined member of the SH3BGR family with potential roles in cell differentiation and tissue development. Here, we report for the first time that SH3BGRL2 exerts a dual function in breast tumor growth and metastasis. SH3BGRL2 was downregulated in a subset of primary breast tumors, and suppressed breast cancer cell proliferation and colony formation in vitro and xenograft tumor growth in vivo. Strikingly, SH3BGRL2 enhanced breast cancer cell migratory, invasive, and lung metastatic capacity. Mechanistic investigations revealed that SH3BGRL2 interacted with and transcriptionally repressed spectrin alpha, non-erythrocytic 1 (SPTAN1) and spectrin beta, non-erythrocytic 1 (SPTBN1), two impor- tant cytoskeletal . Functional rescue assays further demonstrated that depletion of SH3BGRL2 reduced breast cancer cell invasive potential, which was partially rescued by knockdown of SPTAN1 and SPTBN1 using spe- cific small interfering RNA. Moreover, transforming growth factor-β1 (TGF-β1) transcriptionally activated SH3BGRL2 expression in breast cancer cells through the canonical TGF-β receptor-Smad pathway. Collectively, these results establish a dual function of SH3BGRL2 in breast cancer growth and metastasis and uncover SH3BGRL2 as a down- stream target of the TGF-β1 signaling pathway in breast cancer cells.

Keywords: Breast cancer, SH3BGRL2, SPTAN1, SPTBN1, TGF-β signaling

Introduction to regulate TGF-β target gene expression [11- 13]. An upregulation of TGF-β signaling is close- Breast cancer is the most frequently diagno- ly associated with breast cancer progression sed malignancy among women worldwide and and poor prognosis [10]. Consequently, block- is a highly heterogeneous disease with diverse ade of TGF-β signaling suppresses breast can- molecular profiles [1-3]. Clinical evidence sh- cer cell motility, invasion, and metastasis [14- ows that the majority of breast cancer-related 18]. However, the mechanisms underlying the deaths occur as a result of distant metastasis TGF-β mediated breast cancer metastasis rather than primary tumor itself [4-6]. Thus, a remain incompletely understood. better understanding of the molecular mecha- nisms behind breast cancer metastasis is des- SH3 domain-binding glutamic acid-rich-like pro- perately needed. Emerging evidence shows tein 2 (SH3BGRL2) is a member of the SH3 that transforming growth factor-β (TGF-β) is an domain-binding glutamic acid-rich (SH3BGR) important player in breast cancer metastasis protein family, in addition to SH3BGR, SH3B- through regulating its downstream target gene GRL, and SH3BGRL3 [19-23]. With the excep- expression [7-10]. Canonically, TGF-β binds to tion of SH3BGRL3, this family of proteins con- its type 2 receptor (TGFBR2) leading to activat- tains a highly conserved proline-rich domain ing its type 1 receptor (TGFBR1), which then [21], which is involved in an interaction with phosphorylates intracellular effectors Smad2 proteins containing specific binding modules, and Smad3. The phosphorylated Smad2/3 pro- such as the Src homology 3 (SH3), WW, and teins form a complex with common mediator Enabled/VASP homology-1 (EVH1) domains Smad4 and then translocate into the nucleus [24]. To date, the biological functions of SH3- Dual role of SH3BGRL2 in breast cancer

BGRL2 are largely unknown. Very limited evi- Materials and methods dence shows that SH3BGRL2 may be involved in erythroid differentiation [25], diabetes [26], Cell culture and reagents dietary fat intake [27], and zebrafish develop- ment [20]. Most recently, SH3BGRL2 was sh- The human breast cell lines (MCF-7, T47D, own to act as a tumor suppressor in clear cell SK-BR-3, BT474, ZR-75-1, BT20, MDA-MB-231, renal cell carcinoma through regulating Hippo/ BT549, and Hs578T), human mammary epi- TEAD1-Twist1 signaling [28]. We recently found thelial cell lines (MCF10A and HMEC), and hu- by analysis of publicly available databases that man embryonic kidney 293T (HEK293T) cell line were obtained from the Cell Bank of Chi- SH3BGRL2 is aberrantly expressed in breast nese Academy of Sciences (Shanghai, China). tumors, but its functional and mechanistic role All cell lines were authenticated by monitoring in breast cancer remains unexplored. cell vitality, mycoplasma contamination, and Spectrin is a widely expressed cytoskeletal short tandem repeat profiling. MCF10A cells protein in various cell types and is comprised were cultured in DMEM/F12 supplemented of α- and β-subunits that interact in an antipar- with 5% donor horse serum (Thermofisher), 10 mg/mL insulin, 20 ng/mL epidermal gr- allel manner to form an heterodimer [29]. owth factor (EGF), 0.5 mg/mL hydrocortisone, According to its distribution, spectrin is tradi- and 100 ng/mL cholera toxin. HMEC cells tionally divided into erythroid and nonerythroid were cultured in DMEM containing 5% fetal forms [30]. Spectrin alpha, non-erythrocytic 1 bovine serum (FBS, ExCell Bio), 20 ng/μL EGF, (SPTAN1) and spectrin beta, non-erythrocytic 1 0.5 μg/mL hydrocortisone, 10 μg/mL insulin, (SPTBN1) are two most common members of and 1% penicillin/streptomycin. SK-BR-3 cells non-erythrocytic spectrins and contain an SH3 were cultured in McCoy’s 5A medium and domain [31]. As important cytoskeletal pro- other cell lines were maintained in DMEM me- teins and signaling molecules, SPTAN1 and dium supplemented with 10% FBS and 1% SPTBN1 are involved in a number of fundamen- penicillin/streptomycin. MG-132, TGF-β recep- tal cellular processes including cell adhesion, tor inhibitor SB431542, and Smad3 inhibitor migration, and intercellular signaling transduc- SIS3 were obtained from Selleck Chemicals. tion, and their deregulation has significant Cycloheximide (CHX) was purchased from Cell impacts on cancer progression [30, 32, 33]. In Signaling Technology. Other chemicals and this context, SPTAN1 has been shown to influ- regents were purchased from Sigma-Aldrich ence cancer progression in both positive and unless otherwise noted. The detailed infor- negative ways depending on its localization and mation for chemical inhibitors used in this regulation [32]. Similarly, SPTBN1 expression study is provided in Table S1. and function differ between different tumor states or types [33]. However, the function and Tissue samples related regulatory mechanism for SPTAN1 and SPTBN1 in breast cancer remain largely un- A total of 28 pairs of primary breast tumor tis- known. sues and matched adjacent normal breast tis- sues were collected from breast cancer pati- In this study, we provide evidence for the first ents who underwent surgery at Fudan Univer- time that SH3BGRL2 is downregulated in a sity Shanghai Cancer Center. No patients re- subset of primary breast tumors and exerts a ceived chemotherapy, radiotherapy, or endo- dual function in breast cancer growth and crine therapy before operation. This study was approved by the Research Ethics Committee metastasis. Moreover, SH3BGRL2 manifests of Fudan University Shanghai Cancer Center. its function on breast cancer metastasis th- Written informed consents were obtained from rough regulating SPTAN1 and SPTBN1 expres- all the participants. All procedures were con- sion. In addition, SH3BGRL2 is regulated by ducted in accordance with the ethical guide- the TGF-β signaling pathway in breast cancer lines of the Declaration of Helsinki. cells. These findings highlight a paradoxical role of SH3BGRL2 in breast cancer and may DNA constructs, shRNAs, and siRNAs facilitate development of new targeted thera- pies to improve the outcome of breast cancer SH3BGRL2 cDNA in pEnter vector with C ter- patients. minal Flag and His tag was purchased from

1239 Am J Cancer Res 2020;10(4):1238-1254 Dual role of SH3BGRL2 in breast cancer

Vigene Bioscience. Short hairpin RNAs (shR- reagents (Tengyi Biotech). After 24 h of trans- NAs) targeting human SH3BGRL2 (shSH3B- fection, the activity of SPTAN1 and SPTBN1 GRL2) in GIPZ lentiviral vector and correspond- promoter was determined using the dual-lucif- ing control constructs were purchased from erase reporter assay system (Promega) accord- Dharmacon. To generate Flag-SH3BGRL2 con- ing to the manufacturer’s protocol. All assays struct, SH3BGRL2 cDNAs were amplified by were repeated at least three times. PCR and subcloned into the lentiviral vector pCDH-CMV-MCSEF1-Puro (System Bioscien- Antibodies, immunoblotting and immunopre- ces). The detailed information of expression cipitation vectors and primers for molecular cloning is provided in Tables S2 and S3. Small interfering The information for primary antibodies used in RNAs (siRNAs) targeting SPTAN1, SPTBN1, this study is listed in Table S6. For immuno- SMAD2, SMAD3, SMAD4, TGFBR1, TGFBR2, blotting, cells and tissues were lysed in RIPA and negative control siRNA (siNC) were pur- buffer with 1 × protease inhibitors and phos- chased from GenePharma, and their target phatase inhibitors (Bimake). The bicinchoninic sequences are listed in Table S4. acid assay kit (Yeasen) was used to determine protein concentrations. Equal amounts of pro- Plasmid transfection and lentiviral infection teins were separated by SDS-PAGE and trans- ferred onto a PVDF membrane (Millipore). The Plasmid transfection was performed using membranes were incubated with indicated pri- Neofect DNA transfection reagent (Tengyi Bio- mary antibodies and detected with enhanced tech) following the manufacturer’s protocol. To chemiluminescence detection kit (Yeasen). Qu- generate stable cell lines, plasmid constructs antitation of immunoblotting bands was per- or shRNAs in lentiviral expression vectors were formed using ImageJ software, and the relative transfected into HEK293T together with pack- expression levels of proteins were normalized aging plasmid mix using Neofect DNA trans- to those of internal control vinculin. For immu- fection reagent. Supernatants were collected noprecipitation (IP), cell extracts were incubat- after 48 h of transfection and used for infect- ed with indicated antibodies overnight at 4°C, ing cells in the presence of 8 mg/mL of poly- followed by incubation with protein A/G mag- brene. After 24 h of infection, cells were se- netic beads (Bimake) for another 3 h. The lected with 2 mg/mL of puromycin (Cayman immunoprecipitates were washed with lysis Chemicals) for 1 week. buffer three times and then analyzed by immunoblotting. RNA isolation and quantitative reverse tran- scription-PCR Proteomics analysis

Total RNA was isolated using Trizol reagent To detect SH3BGRL2 interacting proteins, cel- (Invitrogen) according to the manufacturer’s lular lysates from HEK293T cells stably ex- protocol. Reverse transcription was performed pressing pCDH and Flag-SH3BGRL2 were sub- using a PrimeScript RT reagent Kit (TaKara). jected to IP assays with anti-Flag magnetic Quantitative real-time PCR (qPCR) was per- beads (Bimake). The immunoprecipated pro- formed using SYBR green II (Takara) to com- teins were resolved by SDS-PAGE, visualized pare the relative expression levels of specific by Coomassie Blue staining, and then subje- mRNAs on an Eppendorf Mastercycleer ep re- cted to liquid chromatography-tandem mass alplex4 instrument (Eppendorf). All oligonucle- spectrometry (LC-MS/MS) analysis as describ- otide primers were synthesized in HuaGene ed previously [34]. All raw data was search- Biotech. Primer sequences for qPCR are listed ed against Swiss-Prot database by SEQUEST. in Table S5. All reactions were performed in Trans Proteomic Pipeline software (Institute of triplicate. The data is present as mean ± SD. Systems Biology, Seattle) was used to identify proteins based on the corresponding peptide Dual-luciferase reporter assays sequences with 95% confidence.

The promoter sequences of SPTAN1 and SP- Cell viability and colony formation assays TBN1 were cloned into the pLG3-basic vector. Then, 10 ng of pGL3-SPTAN1, -SPTBN1, or con- Cell viability was determined using Cell Coun- trol vectors were transfected into the corre- ting Kit-8 (CCK-8) (Dojindo Laboratories) follow- sponding cells using Neofect DNA transfection ing the manufacturer’s protocol. The absor-

1240 Am J Cancer Res 2020;10(4):1238-1254 Dual role of SH3BGRL2 in breast cancer bance was measured at a wavelength of 450 injection. The lungs were removed and meta- nm (A450). For colony formation assays, cells static nodules were counted. Hematoxylin and were plated in a 6-well plate at a density of eosin staining was performed on sections of 2000 cells/well and cultured for 14 days. Cells mouse lung tissues to verify metastatic lung were stained with crystal violet and the number tumors. of survival colonies was counted. Statistical analysis Wound-healing and transwell invasion assays All experiments were replicated at least three For wound-healing assays, the wound was gen- times and the results are presented as mean ± erated using a 100 μl tip when cells reach con- standard error. The differences between two fluence. The floated cells were removed by wa- groups were compared by the unpaired two- shing with PBS and cells were cultured in medi- tailed Student’s t-test. All statistical analyses um containing 0.1% FBS. Images were taken at were conducted with SPSS version 22.0 soft- the indicated times, and the wound closure ware. A p-value less than 0.05 was considered ratios were calculated. For transwell invasion statistically significant difference. assays, cells were resuspended in DMEM con- Results taining 1% FBS and seeded at a density of 5 × 4 10 cells per well into the upper Matrigel inva- SH3BGRL2 is downregulated in a subset of sion chambers (Corning BioCoat). Growth medi- primary breast tumors um containing 10% FBS was placed at the lower chamber. After incubation of the indicat- To examine the expression pattern of SH3- ed times, cells on the upper surface were gen- BGRL2 in breast tumors, we first analyzed tly removed with a cotton swab. Cells on the SH3BGRL2 mRNA levels in The Cancer Ge- lower surface were fixed with methanol and nome Atlas (TCGA) database via UALCAN [35]. then stained with crystal violet. The number of Results showed that SH3BGRL2 mRNA levels cells was counted under a light microscope were reduced in primary breast tumors rela- with a magnification of 100. tive to normal breast tissues (Figure 1A). Moreover, this phenomenon was observed in Tumor xenografts and lung metastasis in nude all of three major molecular subtypes, includ- mice ing luminal, HER2-positive (HER2+), and triple- negative breast cancer (TNBC) (Figure 1B). All experiments involving animals were in accor- Analysis of our recently published RNA-se- dance with the Guides for the Care and Use of quencing data of 360 primary TNBC tissues Laboratory Animals and were performed ac- and 88 adjacent normal breast tissues [36] cording to the institutional ethical guidelines. also showed that SH3BGRL2 mRNA levels The study was also approved by the Institution- were downregulated in the majority of TNBC al Animal Care and Use Committee of Fudan tumors compared to normal controls (Figure University. For subcutaneous inoculation, MDA- 1C). Moreover, analysis of The Clinical Proteo- MB-231 cells stably expressing pCDH or Flag- mic Tumor Analysis Consortium (CPTAC) data- SH3BGRL2 were injected into the mammary base [37] revealed that the protein levels of fat pad of 6-week-old female BALB/c athymic SH3BGRL2 were also reduced in breast tu- nude mice (4 × 106 cells/mouse, State Key mors, irrelevant to its molecular subtypes, Laboratory of Oncogenes and Related , compared to normal breast tissues (Figure 1D Shanghai, China). The tumors were measured and 1E). twice a week after tumor formation and the tumor volume was calculated by the formula To further validate these results, we collected of (length × width2)/2. Mice were killed after 8 28 pairs of primary breast tumor specimens weeks of the inoculation. The removed tumors and matched adjacent normal breast tissues were weighed. For experimental lung metasta- to detect the mRNA and protein levels of SH3- sis assays, MDA-MB-231 cells stably express- BGRL2 by qPCR and immunoblotting analysis, ing pCDH and Flag-SH3BGRL2 were injected respectively. As shown in Figure 1F, 92.9% into the tail vein of each mouse (3 × 106 cells/ (26/28) of breast tumor tissues showed lower mouse). Mice were sacrificed 8 weeks after SH3BGRL2 mRNA levels than adjacent nor-

1241 Am J Cancer Res 2020;10(4):1238-1254 Dual role of SH3BGRL2 in breast cancer

1242 Am J Cancer Res 2020;10(4):1238-1254 Dual role of SH3BGRL2 in breast cancer

Figure 1. SH3BGRL2 is downregulated in a subset of primary breast tumors. (A, B) Analysis of SH3BGRL2 mRNA levels in TCGA database via UALCAN [35]. (C) SH3BGRL2 mRNA levels in 360 primary TNBC tissues and 88 adjacent normal breast tissues [36]. (D, E) Analysis of SH3BGRL2 protein levels in CPTAC database [37]. (F) qPCR analysis of SH3BGRL2 mRNA levels in 28 pairs of breast tumor tissues and matched normal breast tissues. (G, H) Immu- noblotting analysis of SH3BGRL2 expression levels in 28 pairs of breast tumor tissues and matched normal breast tissues. The protein gray intensity was quantified using ImageJ software. Relative expression levels of SH3BGRL2 (SH3BGRL2/vinculin) are shown in (H). N, Normal; T, tumor. mal tissues. In addition, the protein levels of ously injected into mammary fat pads of 6- SH3BGRL2 were downregulated in 67.9% (19/ week-old female BALB/c nude mice. In support 28) of primary breast tumor samples com- of in vitro findings, tumors from SH3BGRL2 pared to adjacent normal tissues (Figure 1G overexpressing MDA-MB-231 cells grew much and 1H). Together, these results suggest that slower at the implantation sites than their con- SH3BGRL2 is downregulated in a subset of trol cells (Figure 2J and 2K). Collectively, these primary breast tumor tissues. results suggest that SH3BGRL2 suppresses breast cancer cell proliferation and colony for- SH3BGRL2 suppresses breast cancer cell mation in vitro and xenograft tumor growth in proliferation and colony formation in vitro and vivo. xenograft tumor growth in vivo SH3BGRL2 enhances breast cancer cell mi- To examine the biological function of SH3BG- gratory, invasive, and metastatic potential RL2 in breast cancer, we first examined the pro- tein levels of SH3BGRL2 in two normal human As breast cancer cells possess the charac- mammary epithelial cell lines and nine repre- teristics of invasion and metastasis [4-6], we sentative breast cancer cell lines by immunob- next investigated the impact of SH3BGRL2 on lotting analysis. As shown in Figure 2A, the invasive and metastatic phenotype of breast expression levels of SH3BGRL2 in breast can- cancer cells. Wound-healing assays showed cer MCF-7, BT474, ZR-75-1, BT20, and Hs578T that ectopic expression of SH3BGRL2 in MDA- cell lines were relatively lower than normal MB-231 and Hs578T cells increased wound cl- mammary epithelial MCF10A and HMEC cell osure rate compared to empty vector-express- lines. Based on SH3BGRL2 expression levels ing control cells (Figures 3A and S1A). Trans- and malignant biological behaviors of those well invasion assays showed that MDA-MB- breast cancer cell lines, we next stably overex- 231 and Hs578T cells expressing SH3BGRL2 pressed Flag-SH3BGRL2 in MDA-MB-231 and had increased invasive potential compared to Hs578T cells (Figure 2B) or depleted endoge- empty vector expressing control cells (Figure nous SH3BGRL2 in BT549 and MDA-MB-231 3B and 3C). In contrast, knockdown of endoge- cells (Figure 2C) by infection of cells with lenti- nous SH3BGRL2 in BT549 and MDA-MB-231 viral vectors expressing Flag-SH3BGRL2 or sh- cells significantly inhibited their migratory and SH3BGRL2, respectively. Expression status of invasive capacity (Figures 3D-F and S1B). To SH3BGRL2 in these established stable cell examine whether SH3BGRL2 affects the me- lines was validated by immunoblotting analy- tastatic capacity of breast cancer cells in vivo, sis (Figure 2B and 2C). CCK-8 and colony for- MDA-MB-231 cells stably expressing empty mation assays showed that overexpression of vector and SH3BGRL2 were injected into nude SH3BGRL2 in MDA-MB-231 and Hs578T cells mice through the tail vein. As shown in Figure suppressed cell proliferation (Figure 2D) and 3G and 3H, SH3BGRL2-overexpressing cells colony formation (Figure 2E and 2F) compar- significantly increased the number of metas- ed to empty vector control. In contrast, knock- tatic tumors in the lungs of nude mice com- down of SH3BGRL2 enhanced the proliferat- pared to empty vector-expressing cells. These ion and colony formation capability of BT549 results were also confirmed by hematoxylin- and MDA-MB-231 cells (Figure 2G-I). To exam- eosin staining of lung sections of these mice ine whether SH3BGRL2 affects tumorigenic (Figure 3I). Collectively, these results suggest capacity of breast cancer cells in vivo, MDA- that SH3BGRL2 promotes breast cancer cell MB-231 cells stably expressing empty vector migration and invasion in vitro and metastatic pCDH and Flag-SH3BGRL2 were subcutane- capacity in vivo.

1243 Am J Cancer Res 2020;10(4):1238-1254 Dual role of SH3BGRL2 in breast cancer

1244 Am J Cancer Res 2020;10(4):1238-1254 Dual role of SH3BGRL2 in breast cancer

Figure 2. SH3BGRL2 suppresses breast cancer cell proliferation and colony formation in vitro and tumorigenesis in vivo. (A) Immunoblotting analysis of SH3BGRL2 protein levels in two normal human mammary epithelial cell lines and nine breast cancer cell lines. (B) MDA-MB-231 and Hs578T cells stably expressing pCDH and Flag-SH3BGRL2 were subjected to immunoblotting analysis with the indicated antibodies. (C) BT549 and MDA-MB-231 cells stably expressing shNC and shSH3BGRL2 were analyzed by immunoblotting. (D-F) MDA-MB-231 and Hs578T cells stably expressing pCDH and Flag-SH3BGRL2 were subjected to cell proliferation assays using CCK-8 kit (D) and colony formation assays (E, F). Representative images of survival colonies (E) and quantitative results (F) are shown. (G-I) BT549 and MDA-MB-231 cells stably expressing shNC and shSH3BGRL2 were subjected to cell proliferation assays using CCK-8 kit (G) and colony formation assays (H, I). Representative images of survival colonies (H) and quantita- tive results (I) are shown. (J, K) MDA-MB-231 cells stably expressing pCDH and Flag-SH3BGRL2 were subcutane- ously injected into mammary fat pads of 6-week-old female BALB/c nude mice (n=5). Images of the xenografted tumors (J) and tumor weight (K) are shown.

SH3BGRL2 interacts with SPTAN1 and inhibitor MG-132 for 12 h. Results showed SPTBN1 and transcriptionally represses their that the noted downregulation of SPTAN1 and expression SPTBN1 in SH3BGRL2 overexpressing cells was not significantly restored following MG- As SH3BGRL2 protein contains a highly con- 132 treatment (Figure S2A). CHX chase as- served proline-rich domain that is involved in says also demonstrated that SH3BGRL2 did protein-protein interactions [21, 24], we next not significantly affect the half-life of SPTAN1 carried out proteomic analysis to identify SH3- and SPTBN1 proteins (Figure S2B). These re- BGRL2-interacting proteins. Toward this aim, sults suggest that SH3BGRL2 does not affect lysates from HEK293T cells stably expressing the stability of SPTAN1 and SPTBN1 proteins. pCDH and Flag-SH3BGRL2 were subjected to qPCR analysis showed that the mRNA levels IP assays with anti-Flag magnetic beads. After of both SPTAN1 and SPTBN1 were negatively being resolved by SDS-PAGE, the precipitated regulated by SH3BGRL2 (Figure 4F and 4G). In proteins were visualized by Coomassie Blue agreement with these data, luciferase reporter staining (Figure 4A) and then subjected to assays demonstrated that SH3BGRL2 nega- LC-MS/MS based proteomic analysis [34]. tively affected the promoter activities of SPT- According to the number of unique peptides, AN1 and SPTBN1 (Figure 4H and 4I). Together, the top 10 potential SH3BGRL2-interacting pro- these results suggest that SH3BGRL2 intera- teins are shown in Figure 4B. Among them, the cts with SPTAN1 and SPTBN1 and transcrip- top 4 proteins, including SPTAN1, SPTBN1, tionally represses their expression. myosin-1c (MYO1C), and flightless-1 (FLII), were selected for further validation. IP and immu- SH3BGRL2 promotes breast cancer cell migra- noblotting analysis demonstrated that SH3BG- tion and invasion partially through negative RL2 indeed interacted with those 4 proteins regulation of SPTAN1 and SPTBN1 (Figure 4C). Following these observations, we next examined whether SH3BGRL2 affects Accumulating evidence shows that SPTAN1 their expression levels. As showed in Figure and SPTBN1 can influence cancer progression 4D and 4E, immunoblotting analysis showed in a context-dependent manner [32, 33], but that overexpression of SH3BGRL2 led to a their contribution to breast cancer progression reduction in the protein levels of SPTAN1 and is unknown. To address whether SH3BGRL2 SPTBN1 in MDA-MB-231 and Hs578T cells, promotes breast cancer migration and inva- whereas knockdown of SH3BGRL2 upregulated sion through regulating SPTAN1 and SPTBN1, their expression in BT549 and MDA-MB-231 we knocked down SPTAN1 and SPTBN1 in cells. In contrast, SH3BGRL2 had no significant MDA-MB-231 cells stably expressing shNC impact on the expression levels of FLII and and shSH3BGRL2. The knockdown efficiency MYO1C under the condition of overexpression of siRNAs targeting SPTAN1 (siSPTAN1) and or knockdown of SH3BGRL2 (Figure 4D and SPTBN1 (siSPTBN1) was verified by qPCR as- 4E). says (Figure 5A and 5B). Transwell invasion assays also showed that reduced cell invasion To test whether SH3BGRL2 affects SPTAN1 in cells expressing shSH3BGRL2 was partially and SPTBN1 protein stability, we treated MDA- rescued following knockdown of SPTAN1 or MB-231 stably expressing pCDH and Flag- SPTBN1 (Figure 5C-E). These results indicate SH3BGRL2 with or without 10 μM proteasome that SH3BGRL2 promotes breast cancer cell

1245 Am J Cancer Res 2020;10(4):1238-1254 Dual role of SH3BGRL2 in breast cancer

1246 Am J Cancer Res 2020;10(4):1238-1254 Dual role of SH3BGRL2 in breast cancer

Figure 3. SH3BGRL2 enhances breast cancer cell migratory, invasive, and metastatic potential. (A-C) MDA-MB-231 and Hs578T cells stably expressing pCDH and Flag-SH3BGRL2 were subjected to wound-healing assays (A) and Transwell invasion assays (B, C). Representative images of wound-healing assays are shown in Figure S1A and the corresponding quantitative results are shown in (A). Representative images of Transwell invasion assays are shown in (B) and corresponding quantitative results are shown in (C). (D-F) BT549 and MDA-MB-231cells stably expressing shNC and shSH3BGRL2 were subjected to wound-healing assays (D) and Matrigel invasion assays (E, F). Repre- sentative images of wound-healing assays are shown in Figure S1B and the corresponding quantitative results are shown in (D). Representative images of Transwell invasion assays are shown in (E) and corresponding quantitative results are shown in (F). (G-I) MDA-MB-231 cells stably expressing pCDH and Flag-SH3BGRL2 were injected into nude mice (n=6) through tail vein. The lungs were harvested after 8 weeks of injection. Representative images of lung metastasis (G), quantitative results of lung nodules (H), and HE-stained sections of lung tissues (I) are shown. invasion through, at least in part, negative regu- stream target gene of TGF-β signaling, which lation of SPTAN1 and SPTBN1. regulates SH3BGRL2 expression through the canonical TGFBR/SMAD pathway. SH3BGRL2 is regulated by TGF-β1 Discussion As TGF-β has a dual function in breast cancer progression at its different stages [38, 39], we In this study, we report a dual function of next evaluated whether SH3BGRL2 is regulat- SH3BGRL2 in tumorigenesis and metastasis ed by TGF-β. Toward this aim, we treated MDA- of breast cancer. SH3BGRL2 is a member of MB-231 and BT549 cells with TGF-β1 for dif- SH3BGR protein family, which encodes a clus- ferent times and then measured the mRNA ter of small thioredoxin-like proteins [19-23]. and protein levels of SH3BGRL2, SPTAN1, and SH3BGR was initially cloned in an effort to SPTBN1. Interestingly, we found that treatment identify genes potentially involved in Down of MDA-MB-231 and BT549 cells with TGF-β1 syndrome-associated congenital heart defects resulted in an upregulation of SH3BGRL2 [21, 42, 43]. Subsequent studies demonstrat- mRNA and protein levels in a time- dependent ed that SH3BGR members play predominant manner (Figure 6A-D). Meanwhile, SPTAN1 and roles in tissue development and cell differen- SPTBN1 expression levels showed an opposite tiation [20, 44]. Emerging evidence shows that pattern (Figure 6A-D). The noted effects of the aberrant expression of the members of TGF-β1 on SH3BGRL2, SPTAN1, and SPTBN1 the SH3BGR gene family is involved in human expression were blocked by pretreatment of cancer development and progression. In this cells with SB431542, an inhibitor for TGFBR1 context, SH3BGRL is downregulated in cells [40], and SIS3, a specific inhibitor for SMAD3 expressing the v-rel oncogene and suppress- [41] (Figure 7A and 7B). To further verify these es v-Rel-mediated cell transformation [45]. In findings, we next knocked down TGFBR1 and addition, ectopic expression of murine SH- TGFBR2 by specific siRNAs and then examined 3BGRL facilitates tumor cell invasion and lung the effects of TGF-β1 on SH3BGRL2, SPTAN1, metastasis, whereas human SH3BGRL sup- and SPTBN1 expression levels. Results show- presses tumorigenesis and metastasis [46]. A ed that knockdown of TGFBR1 and TGFBR2 recent study reported that SH3BGRL2 func- (Figure S3A and S3B) compromised the effect tions as a tumor suppressor in clear cell renal of TGF-β1 on SH3BGRL2, SPTAN1, and SPT- cell carcinoma [28]. However, the functional BN1 expression levels (Figure 7C and 7D). To and mechanistic role for SH3BGRL2 in breast examine which SMAD protein is involved in TGF- cancer remains unexplored. Analysis of publicly β1 induced SH3BGRL2 expression, we next available databases and 28 pairs of matched knocked down SMAD2, SMAD3, and SMAD4 primary breast tumor specimens and normal using specific siRNA, and the knockdown effi- breast tissues by immunoblotting and qPCR cacy was verified by qPCR and immunoblotting assays demonstrated that SH3BGRL2 is analysis (Figure S4A-C). As shown in Figure 7E downregulated in a subset of primary breast and 7F, knockdown of SMAD3 and SMAD4, tumors (Figure 1). Gain- and loss-of function but not SMAD2, compromised TGF-β1 induced assays showed that SH3BGRL2 suppresses upregulation of SH3BGRL2. Together, these breast tumor growth but enhances breast can- findings suggest that SH3BGRL2 is a down- cer metastatic potential (Figures 2 and 3). In

1247 Am J Cancer Res 2020;10(4):1238-1254 Dual role of SH3BGRL2 in breast cancer

Figure 4. SH3BGRL2 interacts with SPTAN1 and SPTBN1 and transcriptionally represses their expression. (A) HEK293T Cells stably expressing pCDH and Flag-SH- 3BGRL2 were subjected to IP assays with anti-Flag magnetic beads. SDS-PAGE gel was stained using Coomassie blue solution. (B) Proteomic analysis of SH3BGRL2 interacting proteins. The top 10 SH3BGRL2 interacting proteins according to the number of unique peptides are shown. (C) HEK293T Cells stably expressing pCDH and Flag-SH3BGRL2 were subjected to sequential IP and immunoblotting analyses with the indicated antibodies. (D) Hs578T and MDA-MB-231 cells stably express- ing pCDH and Flag-SH3BGRL2 were subjected to immunoblotting analysis with the indicated antibodies. (E) BT549 and MDA-MB-231 cells stably expressing shNC

1248 Am J Cancer Res 2020;10(4):1238-1254 Dual role of SH3BGRL2 in breast cancer and shSH3BGRL2 were subjected to immunoblotting analysis with the indicated antibodies. (F) qPCR analysis of SH3BGRL2, SPTAN1, and SPTBN1 mRNA levels in MDA-MB-231 cells stably expressing pCDH and Flag-SH3BGRL2. (G) qPCR analysis of SH3BGRL2, SPTAN1, and SPTBN1 mRNA levels in MDA-MB-231 cells stably expressing shNC and shSH3BGRL2. (H) HEK293T cells were transfected with pGL3-SPTAN1 or pGL3-SPTBN1 plasmid DNA in com- bination with pCDH or Flag-SH3BGRL2. After 48 h of transfection, promoter activities of SPTAN1 and SPTBN1 were analyzed using the dual-luciferase reporter assay system. (I) HEK293T cells were transfected with pGL3-SPTAN1 or pGL3-SPTBN1 plasmid DNA in combination with shNC or shSH3BGRL2. After 48 h of transfection, promoter activi- ties of SPTAN1 and SPTBN1 were analyzed using the dual-luciferase reporter assay system.

Figure 5. SH3BGRL2 abrogates the ability of SPTAN1 and SPTBN1 to suppress breast cancer cell invasion. (A, B) MDA-MB-231 cells were transfected with siRNAs targeting SPTAN1 (A) or SPTBN1 (B). After 48 h of transfection, qPCR analysis was carried out to examine the SPTAN1 and SPTBN1 mRNA levels in these cells. (C-E) MDA-MB-231 cells stably expressing shNC and shSH3BGRL2 were transfected with siNC, siSPTAN1#3 or siSPTBN1#3 and then subjected to Transwell invasion assays. Representative images are shown in (C) and the corresponding quantitative results are shown in (D) and (E). support of our results, the dual activities of ling, plays both pro-tumorigenic and antitumori- several cancer-related molecules in tumor genic roles at different stages of mammalian development and progression have been docu- malignant progression [47]. Similarly, we also mented previously. For instance, SnoN, an recently demonstrated that F-box only protein important negative regulator of TGF-β signa- 22 (FBXO22) possesses both protumorigenic

1249 Am J Cancer Res 2020;10(4):1238-1254 Dual role of SH3BGRL2 in breast cancer

Figure 6. SH3BGRL2 is regulated by TGF-β1. (A-C) MDA-MB-231 cells were treated with or without 10 ng/ml TGF-β1 for the indicated times and then subjected to qPCR analysis of SH3BGRL2 (A), SPTAN1 (B), and SPTBN1 (C) mRNA levels. (D) MDA-MB-231 and BT549 cells treated with 10 ng/ml TGF-β1 for the indicated times and then subjected to immunoblotting analysis with the indicated antibodies. Slug is shown as a positive control. and antimetastatic roles in breast cancer pro- their expression (Figure 4). SPTAN1 has been gression [48]. Based on these findings, we shown to be involved in colorectal cancer inva- speculated that SH3BGRL2 may have differ- sion and metastasis [52, 53]. In addition, ent effects on breast cancer depending on its SPTBN1 has been documented to suppress stage. progression of hepatocellular carcinoma [54], and may be associated with metastases of the The TGF-β pathway is well characterized for its uveal melanoma [55]. In this study, we discov- pleiotropic roles in human cancer development ered that depletion of SPTAN1 and SPTBN1 and progression [8]. Ever-growing studies have enhances breast cancer cell migratory and unraveled that TGF-β plays a tumor-suppres- invasive potential. Moreover, reduced migra- sive role via inhibiting cell proliferation at the tion and invasion in cells expressing shSH3B- early malignant stage. While at the late stage of GRL2 was partially rescued by knockdown of breast cancer, the cytostatic effects of TGF-β is SPTAN1 or SPTBN1 (Figure 5). These results attenuated along with the enhanced migration, indicate that SH3BGRL2 promotes breast can- invasion and EMT, which foster tumor progres- cer cell migration and invasion through, at least sion [49, 50]. In this study, we found that TGF- β1, a potent regulator in promoting breast can- in part, negative regulation of SPTAN1 and cer cell invasion [51], regulates SH3BGRL2 SPTBN1. However, the detailed mechanism for expression through the canonical pathway regulation of SPTAN1 and SPTBN1 by SH3BG- (Figures 6 and 7). These results further support RL2 remains to be further investigated in the a prometastastic role of SH3BGRL2 in breast future. cancer. In conclusion, findings presented here suggest To address the underlying mechanism by which SH3BGRL2 exerts a dual function in breast SH3BGRL2 promotes breast cancer metasta- cancer growth and metastasis and is a down- sis, we further demonstrated that SH3BGRL2 stream target of TGF-β. Further investigation interacts with SPTAN1 and SPTBN1, two key into the underlying mechanism behind these cytoskeletal proteins, and negatively regulates observations may provide a new target for

1250 Am J Cancer Res 2020;10(4):1238-1254 Dual role of SH3BGRL2 in breast cancer

Figure 7. TGF-β1 regulates SH3BGRL2 through the canonical pathway. (A, B) MDA-MB-231 cells were pretreated with 5 μM SB431542 or 3 μM SIS3 for 2 h and then treated with 10 ng/ml TGF-β1 for another 24 h. qPCR (A) and immunoblotting (B) analysis was carried out to examine the expression levels of SH3BGRL2, SPTAN1, and SPTBN1. (C, D) MDA-MB-231 cells were transfected with siRNAs targeting TGFBR1 or TGFBR2. After 48 h of transfection, cells were treated with or without 10 ng/ml TGF-β1 for another 24 h. qPCR (C) and immunoblotting (D) analysis was carried out to examine the expression levels of SH3BGRL2, SPTAN1, and SPTBN1. (E, F) MDA-MB-231 cells were transfected with siNC, siSMAD2, siSMAD3, or siSMAD4. After 48 h of transfection, cells were treated with or without 10 ng/ml TGF-β1 for another 24 h, and then subjected to qPCR (E) or immunoblotting (F) analysis. developing tailored therapy of breast cancer in Address correspondence to: Da-Qiang Li, the future. Shanghai Cancer Center and Institutes of Bio- medical Sciences, Fudan University, Shanghai Disclosure of conflict of interest 200032, China. Tel: +86-21-38197025; Fax: +86-21-64434556; E-mail: daqiangli1974@ None. fudan.edu.cn

1251 Am J Cancer Res 2020;10(4):1238-1254 Dual role of SH3BGRL2 in breast cancer

References [15] Fang Y, Chen Y, Yu L, Zheng C, Qi Y, Li Z, Yang Z, Zhang Y, Shi T, Luo J and Liu M. Inhibition of [1] Bray F, Ferlay J, Soerjomataram I, Siegel RL, breast cancer metastases by a novel inhibitor Torre LA and Jemal A. Global cancer statistics of TGFbeta receptor 1. J Natl Cancer Inst 2013; 2018: GLOBOCAN estimates of incidence and 105: 47-58. mortality worldwide for 36 cancers in 185 [16] Ganapathy V, Ge R, Grazioli A, Xie W, Banach- countries. CA Cancer J Clin 2018; 68: 394- Petrosky W, Kang Y, Lonning S, McPherson J, 424. Yingling JM, Biswas S, Mundy GR and Reiss M. [2] Zardavas D, Irrthum A, Swanton C and Piccart Targeting the transforming growth factor-beta M. Clinical management of breast cancer het- pathway inhibits human basal-like breast can- erogeneity. Nat Rev Clin Oncol 2015; 12: 381- cer metastasis. Mol Cancer 2010; 9: 122. 394. [17] Park CY, Min KN, Son JY, Park SY, Nam JS, Kim [3] Stingl J and Caldas C. Molecular heterogeneity DK and Sheen YY. An novel inhibitor of TGF- of breast carcinomas and the cancer stem cell beta type I receptor, IN-1130, blocks breast hypothesis. Nat Rev Cancer 2007; 7: 791-799. cancer lung metastasis through inhibition of [4] Weigelt B, Peterse JL and van’t Veer LJ. Breast epithelial-mesenchymal transition. Cancer Lett cancer metastasis: markers and models. Nat 2014; 351: 72-80. Rev Cancer 2005; 5: 591-602. [18] Son JY, Park SY, Kim SJ, Lee SJ, Park SA, Kim [5] Kast K, Link T, Friedrich K, Petzold A, MJ, Kim SW, Kim DK, Nam JS and Sheen YY. Niedostatek A, Schoffer O, Werner C, Klug SJ, EW-7197, a novel ALK-5 kinase inhibitor, po- Werner A, Gatzweiler A, Richter B, Baretton G tently inhibits breast to lung metastasis. Mol and Wimberger P. Impact of breast cancer sub- Cancer Ther 2014; 13: 1704-1716. types and patterns of metastasis on outcome. [19] Mazzocco M, Maffei M, Egeo A, Vergano A, Breast Cancer Res Treat 2015; 150: 621-629. Arrigo P, Di Lisi R, Ghiotto F and Scartezzini P. [6] Kennecke H, Yerushalmi R, Woods R, Cheang The identification of a novel human homologue MC, Voduc D, Speers CH, Nielsen TO and of the SH3 binding glutamic acid-rich (SH3BGR) Gelmon K. Metastatic behavior of breast can- gene establishes a new family of highly con- cer subtypes. J Clin Oncol 2010; 28: 3271- served small proteins related to thioredoxin 3277. superfamily. Gene 2002; 291: 233-239. [7] Derynck R and Zhang YE. Smad-dependent [20] Tong F, Zhang M, Guo X, Shi H, Li L, Guan W, and Smad-independent pathways in TGF-beta Wang H and Yang S. Expression patterns of family signalling. Nature 2003; 425: 577-584. SH3BGR family members in zebrafish develop- [8] Massague J. TGFbeta in Cancer. Cell 2008; ment. Dev Genes Evol 2016; 226: 287-295. 134: 215-230. [21] Scartezzini P, Egeo A, Colella S, Fumagalli P, [9] Lamouille S, Xu J and Derynck R. Molecular Arrigo P, Nizetic D, Taramelli R and Rasore- mechanisms of epithelial-mesenchymal transi- Quartino A. Cloning a new human gene from tion. Nat Rev Mol Cell Biol 2014; 15: 178-196. 21q22.3 encoding a glutamic [10] Deckers M, van Dinther M, Buijs J, Que I, Lowik acid-rich protein expressed in heart and skele- C, van der Pluijm G and ten Dijke P. The tumor tal muscle. Hum Genet 1997; 99: 387-392. suppressor Smad4 is required for transform- [22] Yin L, Xiang Y, Zhu DY, Yan N, Huang RH, Zhang ing growth factor beta-induced epithelial to Y and Wang DC. Crystal structure of human mesenchymal transition and bone metastasis SH3BGRL protein: the first structure of the of breast cancer cells. Cancer Res 2006; 66: human SH3BGR family representing a novel 2202-2209. class of thioredoxin fold proteins. Proteins [11] Schmierer B and Hill CS. TGFbeta-SMAD signal 2005; 61: 213-216. transduction: molecular specificity and func- [23] Egeo A, Mazzocco M, Arrigo P, Vidal-Taboada tional flexibility. Nat Rev Mol Cell Biol 2007; 8: JM, Oliva R, Pirola B, Giglio S, Rasore-Quartino 970-982. A and Scartezzini P. Identification and charac- [12] Massague J, Seoane J and Wotton D. Smad terization of a new human gene encoding a transcription factors. Genes Dev 2005; 19: small protein with high homology to the pro- 2783-2810. line-rich region of the SH3BGR gene. Biochem [13] Feng XH and Derynck R. Specificity and versa- Biophys Res Commun 1998; 247: 302-306. tility in tgf-beta signaling through Smads. Annu [24] Zarrinpar A, Bhattacharyya RP and Lim WA. Rev Cell Dev Biol 2005; 21: 659-693. The structure and function of proline recogni- [14] Muraoka RS, Dumont N, Ritter CA, Dugger TC, tion domains. Sci STKE 2003; 2003: RE8. Brantley DM, Chen J, Easterly E, Roebuck LR, [25] De Andrade T, Moreira L, Duarte A, Lanaro C, Ryan S, Gotwals PJ, Koteliansky V and Arteaga De Albuquerque D, Saad S and Costa F. Ex- CL. Blockade of TGF-beta inhibits mammary pression of new red cell-related genes in ery- tumor cell viability, migration, and metastases. throid differentiation. Biochem Genet 2010; J Clin Invest 2002; 109: 1551-1559. 48: 164-171.

1252 Am J Cancer Res 2020;10(4):1238-1254 Dual role of SH3BGRL2 in breast cancer

[26] Collares CV, Evangelista AF, Xavier DJ, Taka- Daemen A, Benson JR, Takabe K, Bai F, Li DQ, hashi P, Almeida R, Macedo C, Manoel-Caeta- Wang P, Shi L, Huang W and Shao ZM. Genom- no F, Foss MC, Foss-Freitas MC, Rassi DM, ic and transcriptomic landscape of triple-nega- Sakamoto-Hojo ET, Passos GA and Donadi EA. tive breast cancers: subtypes and treatment Transcriptome meta-analysis of peripheral lym- strategies. Cancer Cell 2019; 35: 428-440, phomononuclear cells indicates that gesta- e425. tional diabetes is closer to type 1 diabetes [37] Edwards NJ, Oberti M, Thangudu RR, Cai S, than to type 2 diabetes mellitus. Mol Biol Rep McGarvey PB, Jacob S, Madhavan S and Ket- 2013; 40: 5351-5358. chum KA. The CPTAC data portal: a resource [27] Rudkowska I, Perusse L, Bellis C, Blangero J, for cancer proteomics research. J Proteome Despres JP, Bouchard C and Vohl MC. Interac- Res 2015; 14: 2707-2713. tion between common genetic variants and [38] Baxley SE and Serra R. Inhibiting breast cancer total fat intake on low-density lipoprotein peak progression by exploiting TGFbeta signaling. particle diameter: a genome-wide association Curr Drug Targets 2010; 11: 1089-1102. study. J Nutrigenet Nutrigenomics 2015; 8: 44- [39] Zhao Y, Ma J, Fan Y, Wang Z, Tian R, Ji W, Zhang 53. F and Niu R. TGF-beta transactivates EGFR and [28] Yin L, Li W, Xu A, Shi H, Wang K, Yang H, Wang facilitates breast cancer migration and inva- R and Peng B. SH3BGRL2 inhibits growth and sion through canonical Smad3 and ERK/Sp1 metastasis in clear cell renal cell carcinoma signaling pathways. Mol Oncol 2018; 12: 305- via activating hippo/TEAD1-Twist1 pathway. 321. EBioMedicine 2020; 51: 102596. [40] Koo BH, Kim Y, Cho YJ and Kim DS. Distinct [29] Liem RK. Cytoskeletal integrators: the spectrin roles of transforming growth factor-beta signal- superfamily. Cold Spring Harb Perspect Biol ing and transforming growth factor-beta recep- 2016; 8. tor inhibitor SB431542 in the regulation of [30] Machnicka B, Grochowalska R, Boguslawska p21 expression. Eur J Pharmacol 2015; 764: DM, Sikorski AF and Lecomte MC. Spectrin- 413-423. based skeleton as an actor in cell signaling. [41] Jinnin M, Ihn H and Tamaki K. Characterization Cell Mol Life Sci 2012; 69: 191-201. of SIS3, a novel specific inhibitor of Smad3, [31] Ziemnicka-Kotula D, Xu J, Gu H, Potempska A, and its effect on transforming growth factor- Kim KS, Jenkins EC, Trenkner E and Kotula L. beta1-induced extracellular matrix expression. Identification of a candidate human spectrin Mol Pharmacol 2006; 69: 597-607. Src homology 3 domain-binding protein sug- [42] Vidal-Taboada JM, Bergonon S, Scartezzini P, gests a general mechanism of association of Egeo A, Nizetic D and Oliva R. High-resolution tyrosine kinases with the spectrin-based me- physical map and identification of potentially mbrane skeleton. J Biol Chem 1998; 273: regulatory sequences of the human SH3BGR 13681-13692. located in the Down syndrome chromosomal [32] Ackermann A and Brieger A. The role of nonery- region. Biochem Biophys Res Commun 1997; throid spectrin alphaii in cancer. J Oncol 2019; 241: 321-326. 2019: 7079604. [43] Egeo A, Di Lisi1 R, Sandri C, Mazzocco M, [33] Chen S, Li J, Zhou P and Zhi X. SPTBN1 and Lapide M, Schiaffino S and Scartezzini P. De- cancer, which links? J Cell Physiol 2020; 235: velopmental expression of the SH3BGR gene, 17-25. mapping to the down syndrome heart critical [34] Zhao J, Yu H, Lin L, Tu J, Cai L, Chen Y, Zhong F, region. Mech Dev 2000; 90: 313-316. Lin C, He F and Yang P. Interactome study sug- [44] Jang DG, Sim HJ, Song EK, Medina-Ruiz S, Seo gests multiple cellular functions of hepatoma- JK and Park TJ. A thioredoxin fold protein Sh- derived growth factor (HDGF). J Proteomics 3bgr regulates Enah and is necessary for prop- 2011; 75: 588-602. er sarcomere formation. Dev Biol 2015; 405: [35] Chandrashekar DS, Bashel B, Balasubra- 1-9. manya SAH, Creighton CJ, Ponce-Rodriguez I, [45] Majid SM, Liss AS, You M and Bose HR. The Chakravarthi BVSK and Varambally S. UALCAN: suppression of SH3BGRL is important for v- a portal for facilitating tumor subgroup gene Rel-mediated transformation. Oncogene 2006; expression and survival analyses. Neoplasia 25: 756-768. 2017; 19: 649-658. [46] Wang H, Liu B, Al-Aidaroos AQ, Shi H, Li L, Guo [36] Jiang YZ, Ma D, Suo C, Shi J, Xue M, Hu X, Xiao K, Li J, Tan BC, Loo JM, Tang JP, Thura M and Y, Yu KD, Liu YR, Yu Y, Zheng Y, Li X, Zhang C, Zeng Q. Dual-faced SH3BGRL: oncogenic in Hu P, Zhang J, Hua Q, Zhang J, Hou W, Ren L, mice, tumor suppressive in humans. Onco- Bao D, Li B, Yang J, Yao L, Zuo WJ, Zhao S, gene 2016; 35: 3303-3313. Gong Y, Ren YX, Zhao YX, Yang YS, Niu Z, Cao [47] Zhu Q, Krakowski AR, Dunham EE, Wang L, ZG, Stover DG, Verschraegen C, Kaklamani V, Bandyopadhyay A, Berdeaux R, Martin GS, Sun

1253 Am J Cancer Res 2020;10(4):1238-1254 Dual role of SH3BGRL2 in breast cancer

L and Luo K. Dual role of SnoN in mammalian [53] Hinrichsen I, Ernst BP, Nuber F, Passmann S, tumorigenesis. Mol Cell Biol 2007; 27: 324- Schafer D, Steinke V, Friedrichs N, Plotz G, Zeu- 339. zem S and Brieger A. Reduced migration of [48] Sun R, Xie HY, Qian JX, Huang YN, Yang F, MLH1 deficient colon cancer cells depends on Zhang FL, Shao ZM and Li DQ. FBXO22 pos- SPTAN1. Mol Cancer 2014; 13: 11. sesses both protumorigenic and antimetastat- [54] Zhi X, Lin L, Yang S, Bhuvaneshwar K, Wang H, ic roles in breast cancer progression. Cancer Gusev Y, Lee MH, Kallakury B, Shivapurkar N, Res 2018; 78: 5274-5286. Cahn K, Tian X, Marshall JL, Byers SW and He [49] Akhurst RJ and Derynck R. TGF-beta signaling AR. βII-Spectrin (SPTBN1) suppresses progres- in cancer--a double-edged sword. Trends Cell sion of hepatocellular carcinoma and Wnt sig- Biol 2001; 11: S44-51. naling by regulation of Wnt inhibitor kallistatin. [50] Huang F, Shi Q, Li Y, Xu L, Xu C, Chen F, Wang Hepatology 2015; 61: 598-612. H, Liao H, Chang Z, Liu F, Zhang XH, Feng XH, [55] Xu Y, Han W, Xu WH, Wang Y, Yang XL, Nie HL, Han JJ, Luo S and Chen YG. HER2/EGFR-AKT Yao J, Shen GL and Zhang XF. Identification of signaling switches TGFbeta from inhibiting cell differentially expressed genes and functional proliferation to promoting cell migration in annotations associated with metastases of the breast cancer. Cancer Res 2018; 78: 6073- uveal melanoma. J Cell Biochem 2019; 120: 6085. 19202-19214. [51] Wang Y and Lui WY. Transforming growth fac- tor-beta1 attenuates junctional adhesion mol- ecule-A and contributes to breast cancer cell invasion. Eur J Cancer 2012; 48: 3475-3487. [52] Ackermann A, Schrecker C, Bon D, Friedrichs N, Bankov K, Wild P, Plotz G, Zeuzem S, Her- rmann E, Hansmann ML and Brieger A. Down- regulation of SPTAN1 is related to MLH1 defi- ciency and metastasis in colorectal cancer. PLoS One 2019; 14: e0213411.

1254 Am J Cancer Res 2020;10(4):1238-1254 Dual role of SH3BGRL2 in breast cancer

Table S1. Chemical inhibitors used inthis study Inhibitors Vendors Cat# Working concentration SB431542 Selleck S1067 5 μM SIS3 Selleck S7959 3 μM MG132 Selleck S2619 10 μM Cycloheximide CST S2112 100 μg/ml Protease inhibitor cocktail Bimake B14002 1 × Phosphatase inhibitor cocktail Bimake B15001 1 ×

Table S2. Information for the expression vectors used in this study Plasmids Sources Vectors Flag-SH3BGRL2 Vigene pEnter shSH3BGRL2#1 Dharmacon pGIPZ-shLenti shSH3BGRL2#2 Dharmacon pGIPZ-shLenti shSH3BGRL2#3 Dharmacon pGIPZ-shLenti shSH3BGRL2#4 Dharmacon pGIPZ-shLenti shSH3BGRL2#5 Dharmacon pGIPZ-shLenti shSH3BGRL2#6 Dharmacon pGIPZ-shLenti

Table S3. Primers used for molecular cloning of expression vectors Plasmid Primers Sequences Flag-SH3BGRL2 Forward CGCAAATGGGCGGTAGGCGTG Reverse CCTCTACAAATGTGGTATGGC

Table S4. siRNA target sequences Primers Sequences siSPTAN1#1 Forward GCAGAAGUUCAGCGCUUUATT Reverse UAAAGCGCUGAACUUCUGCTT siSPTAN1#2 Forward GCUGAAGCCCUGCUAGAUATT Reverse UAUCUAGCAGGGCUUCAGCTT siSPTAN1#3 Forward CCUCAUGUCUUGGAUCAAUTT Reverse AUUGAUCCAAGACAUGAGGTT siSPTBN1#1 Forward CCCGCAGAUUUGAUCGCAATT Reverse UUGCGAUCAAAUCUGCGGGTT siSPTBN1#2 Forward GCACAGAUUUGAGAGCCUATT Reverse UAGGCUCUCAAAUCUGUGCTT siSPTBN1#3 Forward GGACAUGUCUUACGAUGAATT Reverse UUCAUCGUAAGACAUGUCCTT siSMAD2 Forward CCCUGCAACAGUGUGUAAATT Reverse UUUACACACUGUUGCAGGGTT siSMAD3 Forward GAGUUCGCCUUCAAUAUGATT Reverse UCAUAUUGAAGGCGAACUCTT siSMAD4 Forward GCCAUCGUUGUCCACUGAATT Reverse UUCAGUGGACAACGAUGGCTT siTGFBR1#1 Forward CACAACCGCACUGUCAUUCTT Reverse GAAUGACAGUGCGGUUGUGGC

1 Dual role of SH3BGRL2 in breast cancer

siTGFBR1#2 Forward UUGUUCAGAGAACAAUUGCTT Reverse GCAAUUGUUCUCUGAACAAGC siTGFBR1#3 Forward AGAUUGUUGGUACCCAAGGTT Reverse CCUUGGGUACCAACAAUCUCC siTGFBR2#1 Forward AGAAAGAAUGACGAGAACATT Reverse UGUUCUCGUCAUUCUUUCUCC siTGFBR1#2 Forward AAGACGCGGAAGCUCAUGGTT Reverse CCAUGAGCUUCCGCGUCUUGC siTGFBR1#3 Forward GAGCACUGUGCCAUCAUCCTT Reverse GGAUGAUGGCACAGUGCUCGC

Table S5. Primers for qPCR analysis Primers Sequences SH3BGRL2 Forward CTCTTCCTCGGGCTTCGT Reverse GCATCTCCGTCATCTTCCAC SPTAN1 Forward CCAGAGCCGAATTGATGCCTT Reverse TCATCCGAAGCCAAGTGACCT SPTBN1 Forward TCTGGGCTCACTTTTGCAT Reverse GCTGCCTTTGTATTACGCAAC SMAD2 Forward GAGGGAGTCTCGCTCTGTTG Reverse GGTGGCTCATGCCTGTAATC SMAD3 Forward TCGTCCATCCTGCCTTTCAC Reverse GCCCCGTCTTCTTGAGTTTC SMAD4 Forward ATTCAAACCATCCAGCATCC Reverse GCAGTCCTACTTCCAGTCCAG TGFBR1 Forward TCCTCCTTAAAAGGTTCTGC Reverse AGAAAGTCCTCAGCCCAG TGFBR2 Forward CATGATTGGCAGCTACGAGAG Reverse CGACTGTCAAGCGCAGC

Table S6. Antibodies used in this study Antibodies Vendors Cat# Hosts Working concentration SH3BGRL2 Proteintech 21944-1-AP R 1:1000 (WB) SPTAN1 Abcam ab75755 R 1:1000 (WB) SPTBN1 Abcam ab124888 R 1:1000 (WB) Slug CST 9585T R 1:1000 (WB) Vinculin Sigma V9131 M 1:10000 (WB) FLII Proteintech 67039-1-Ig M 1:1000 (WB) MYO1C Abcam ab194828 R 1:1000 (WB) SMAD2 Abclonal A19114 R 1:1000 (WB) SMAD3 Abclonal A19115 R 1:1000 (WB) SMAD4 Abclonal A19116 R 1:1000 (WB) Notes: M, mouse; R, rabbit.

2 Dual role of SH3BGRL2 in breast cancer

Figure S1. SH3BGRL2 enhances breast cancer cell migratory potential. A. MDA-MB-231 and Hs578T cells stably expressing pCDH and Flag-SH3BGRL2 were sub- jected to wound-healing assays. Representative images are shown. The corresponding quantitative results are shown in Figure 3A. B. BT549 and MDA-MB-231 cells stably expressing shNC and shSH3BGRL2 were subjected to wound-healing assays. Representative images are shown. The corresponding quantitative results are shown in Figure 3D.

3 Dual role of SH3BGRL2 in breast cancer

Figure S2. SH3BGRL2 does not affect the stability of SPTAN1 and SPTBN1 proteins. A. MDA-MB-231 cells stably expressing pCDH and Flag-SH3BGRL2 were treated with or without 10 μM MG-132 for 6 h and then subjected to immunoblotting analysis with the indicated antibodies. B. MDA-MB-231 cells stably expressing shNC and shSH3B- GRL2 were treated with 100 μg/ml of CHX for the indicated times and then subjected to immunoblotting with the indicated antibodies.

Figure S3. Validation of the knockdown efficacy of siRNAs targeting TGFBR1 or TGFBR2. (A, B) MDA-MB-231 cells were transfected with siRNAs targeting TGFBR1 (A) or TGFBR2 (B). After 48 h of transfection, qPCR analysis was carried out to examine TGFBR1 or TGFBR2 mRNA levels in these cells.

Figure S4. Validation of the knockdown efficacy of siRNAs targeting SMAD2, SMAD3, and SMAD4. (A-C) MDA- MB-231 cells were transfected with siNC, siSMAD2, siSMAD3, or siSMAD4. After 48 h of transfection, qPCR and immunoblotting analysis was performed.

4