Follistatin-Like Protein 1 Inhibits Lung Cancer Metastasis by Preventing Proteolytic Activation of Osteopontin

Author Manuscript Published OnlineFirst on October 25, 2019; DOI: 10.1158/0008-5472.CAN-19-0842 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

1 Follistatin-like protein 1 inhibits lung cancer metastasis by

2 preventing proteolytic activation of osteopontin

4 Jean Chiou1#, Yu-Chan Chang1#, Hsing-Fang Tsai1, Yuan-Feng Lin2, Ming-Shyan

5 Huang3, Chih-Jen Yang4,5 and Michael Hsiao1,6

7 1Genomic Research Center, Academia Sinica, Taipei, Taiwan

8 2Graduate Institute of Clinical Medicine, College of Medicine, Taipei Medical

9 University, Taipei, Taiwan

10 3Department of Internal Medicine, E-DA Cancer Hospital, School of Medicine, I-

11 Shou University, Kaohsiung, Taiwan

12 4Department of Internal Medicine, Kaohsiung Municipal Ta-Tung Hospital,

13 Kaohsiung Medical University, Kaohsiung, Taiwan.

14 5Faculty of Medicine, College of Medicine, Kaohsiung Medical University, Taiwan.

15 6Department of Biochemistry, College of Medicine, Kaohsiung Medical University,

16 Kaohsiung, Taiwan

18 # These authors contributed equally to the research.

19 To whom correspondence should be addressed:

20 Michael Hsiao, Genomics Research Center, Academia Sinica, 128 Academia Road,

21 Taipei 11529, Taiwan. Tel: +886-2-2787-1243; Fax: +886-2-2789-9931; E-mail:

22 [email protected]

23 Or to

24 Chih-Jen Yang, Kaohsiung Municipal Ta-Tung Hospital, Kaohsiung Medical

25 University, No. 68 Chunghwa 3rd Road, Cianjin District, Kaohsiung 80145, Taiwan. 1

Downloaded from cancerres.aacrjournals.org on September 29, 2021. © 2019 American Association for Cancer Research. Author Manuscript Published OnlineFirst on October 25, 2019; DOI: 10.1158/0008-5472.CAN-19-0842 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

1 Phone: +886-7-320-8159; E-mail: [email protected]

3 Disclosure of Potential Conflicts of Interest

4 The authors have no competing financial interests to declare.

6 Running title: FSTL1 interacts with SPP1 to inhibit lung cancer metastasis

8 Abbreviations list Abbreviations Full-name FST Follistatin FSTL1 Follistatin-like protein 1 SPP1 Secreted phosphoprotein 1 PPI Protein-protein interaction TGFβ Transforming growth factor β BMP4 Bone morphogenetic protein 4 TCGA The Cancer Genome Atlas NSCLC Non-small cell lung cancer LUAD Lung Adenocarcinoma rhFSTL1 Recombinant FSTL1 protein IPA Ingenuity Pathway Analysis OE Overexpress KD Knockdown HR Hazard ratio Ab Antibody OS Overall survival DMFS Distant metastasis free survival 9

1 Abstract

2 Follistatin-like protein 1 (FSTL1) plays a critical role in lung organogenesis, but

3 is downregulated during lung cancer development and progression. The prognostic

4 significance and functional consequences of FSTL1 downregulation in lung cancer

5 are unclear. Here, reduced levels of FSTL1 were detected in various tumors compared

6 to normal tissues and were associated with poor clinical outcome in patients with non-

7 small cell lung cancer, particularly those with lung adenocarcinoma. FSTL1

8 expression negatively correlated with the metastatic potential of lung cancer cells.

9 Antibody-based neutralization of extracellular FSTL1 increased cellular

10 migration/invasion while addition of recombinant FSTL1 protein diminished the

11 metastatic capacity of lung cancer cells in vitro and in vivo. Notably, treatment with

12 FSTL1 effectively prevented the metastatic progression of lung cancer cells in an

13 orthotopic animal model. Mechanistically, FSTL1 directly bound to the pro-form of

14 secreted phosphoprotein 1 (SPP1)/osteopontin restraining proteolytic activation of

15 SPP1 which led to inactivation of integrin/CD44-associated signaling and

16 rearrangement of the actin cytoskeleton. Combined low expression of FSTL1 and

17 high expression of SPP1 predicted a poorer prognosis for patients with lung cancer.

18 This study highlights the novel interaction between FSTL1 and SPP1 and new

19 opportunities to effectively target SPP1-driven metastatic cancers characterized by

20 FSTL1 downregulation.

22 Significance

23 Findings describe the novel interaction between FSTL1 and SPP1 and its role in the

24 metastatic progression of lung adenocarcinoma

2 Introduction

3 Follistatin-like protein 1 (FSTL1) is a secreted glycoprotein that was originally

4 identified as a transforming growth factor  (TGF)-inducible protein. Recently,

5 FSTL1 has increasingly been recognized as a critical developmental regulator of

6 organogenesis, particularly lung development (1). FSTL1 appears to antagonize bone

7 morphogenetic protein 4 (BMP4) signaling during lung development (2). Fstl1-

8 deficient mice have impaired alveolar maturation due to dysfunctional distal alveolar

9 differentiation and insufficient surfactant production (3). Since many developmental

10 pathways involved in embryogenesis play critical roles in oncogenesis and cancer

11 progression (4,5)and the dysregulation of BMP pathways has been detected during

12 tumorigenesis and the malignant evolution of lung cancer (6-9), perturbations in the

13 FSTL1/BMP4 pathway during lung cancer development are worth investigating.

14 Follistatin (FST) has been reported to be a prognostic biomarker for lung

15 adenocarcinoma (10). Similar proteins in the follistatin-like (FSTL) protein family,

16 including FSTL1, IGFBP7, FSTL3, FSTL4, and FSTL5, play different roles in

17 cancers (11-13). FSTL1 has been reported to be downregulated or undetectable in

18 various human cancer cells, including lung cancer cells, but it is expressed at high

19 levels in normal lung tissues (14). Similarly, FSTL1 mRNA levels are substantially

20 reduced in a panel of lung cancer cell lines compared to nontumor lung cell lines (15).

21 The neutralization of extracellular FSTL1 by its specific antibody promotes lung

22 cancer cell invasion and metastasis (16), confirming a pivotal role for FSTL1 in

23 preventing lung cancer progression. We presented the clinical relevance of FSTL1

24 downregulation in lung cancer (17). However, the practical application of the FSTL1

25 protein in combating malignant tumors has not yet been explored.

1 Lung cancer is the leading cause of cancer-related deaths worldwide (18) and is

2 classified into nonsmall cell lung cancer (NSCLC, ~85% of all lung cancers) and

3 small cell lung cancer (~15%). Moreover, NSCLC is divided into three major

4 histological subtypes: adenocarcinoma, squamous cell carcinoma, and large cell lung

5 cancer. Lung adenocarcinoma is currently the most common type of lung cancer in

6 patients who have never smoked (19) and accounts for approximately 40% of all lung

7 cancers. Here, FSTL1 downregulation strongly predicts a poorer prognosis in patients

8 with lung adenocarcinoma, but not squamous cell carcinoma, and strongly correlates

9 with an increased metastatic potential of lung cancer cells in vitro and in vivo.

10 Importantly, treatment with FSTL1 inhibits the metastatic progression of lung cancer

11 cells by directly binding to pro-SPP1 and thereby restraining the proteolytic activation

12 of SPP1. Since SPP1 is a critical extracellular molecule involved in eliciting cancer

13 metastasis, the identification of protein-protein interactions between FSTL1 and SPP1

14 might provide a new opportunity for developing novel anti-cancer agents to combat

15 metastatic cancers.

1 Materials and Methods

2 Chemicals and Antibodies

3 The recombinant FSTL1 protein was purchased from Sino Biological Inc., Beijing,

4 P.R. C. (10924H08H50). Recombinant pro- and mature SPP1 proteins were obtained

5 from R&D (1433-OP-050) and Abnova, Taipei, Taiwan (H00006696-Q01,

6 H00006696-Q02), respectively. The FSTL1 antibody was purchased from

7 Proteintech, IL, USA (20182-1-AP, 1:2000). The SPP1 antibody was purchased from

8 Abcam, MA, USA (ab8848, 1:1000). Antibodies against Akt (9272, 1:1000), p-AkT

9 (9275S, 1:1000), Src (2109S, 1:1000) and p-Src (2101S, 1:1000) antibodies were

10 purchased from Cell Signaling Technology, MA, USA. The antibody against β-actin

11 (A2228, 1:5000) was purchased from Sigma-Aldrich, MO, USA.

12 Cell lines and cell culture conditions

13 The lung cancer cell lines A549, H1299, PC13 and PC14 were maintained in DMEM

14 supplemented with 10% FBS and 1% penicillin-streptomycin-glutamine (PSG)

15 (Invitrogen, CA, USA). H1355, H928, CL1-0 and CL1-5 lung cancer cells were

16 maintained in RPMI-1640 medium supplemented with 10% FBS and 1% PSG. The

17 Beas2B normal lung epithelial cell line was maintained in keratinocyte-SFM medium

18 (Invitrogen, CA, USA, cat: 17005042) supplemented with 1% PSG. CL1-0 and CL1-5

19 cells were derived by Chu et al. and exhibit progressively increasing invasiveness

20 (20). The PC13 and PC14 cell lines were generated by Lee et al. at the National

21 Cancer Center Hospital, Tokyo, Japan (21). Other lung cancer cell lines and the

22 Beas2B cell line were acquired from the American Type Culture Collection, USA.

23 Transwell migration and invasion assays

24 For transwell migration assays, 5 × 104 cells were plated in the top chamber on the

25 noncoated membrane (Corning Costar). For the invasion assay, each well was freshly

1 coated with Matrigel (BD Bioscience) before the invasion assay. Then, 5 × 104 cells

2 were plated in the top chamber on the Matrigel-coated membrane. In both assays,

3 cells were plated in serum-free medium and medium supplemented with 10% FBS,

4 which was used as a chemoattractant, was placed in the lower chamber. The rhFSTL1

5 protein was added to the upper layer. The cells were then incubated for another 16 or

6 24 hours for the migration assay or 24 or 48 hours for the invasion assay. The cells

7 that did not migrate or invade through the pores were removed with a cotton swab.

8 Cells on the lower surface of the membrane were fixed with methanol. After 10

9 minutes, membranes were stained with crystal violet. The number of cells migrating

10 or invading through the membrane were counted under a light microscope (40×, three

11 randomly selected fields per well).

12 Western blot analysis

13 Proteins (20-50 μg) were electrophoretically separated on 10% SDS-polyacrylamide

14 gels. After electrophoresis, the proteins were transferred to a 0.45 m poly-vinylidene

15 fluoride (PVDF) membrane that was blocked with 5% nonfat dry milk in PBS with

16 Tween 20. Immunoblotting was performed using the primary antibodies with

17 overnight incubations at 4°C. Following washes and an incubation with the

18 appropriate horseradish peroxidase-conjugated secondary antibodies, signals were

19 visualized using an enhanced chemiluminescence kit (Amersham ECL Plus™). All

20 Western blot data are representative of at least three independent replicates.

21 Lentiviral infections and shRNA sequences

22 Lentiviral FSTL1 shRNA constructs were purchased from the National RNAi Core

23 Facility Platform (Academia Sinica, Taiwan). Lentiviruses were produced by

24 cotransfecting the shRNA-expressing vector with the pMDG and p△8.91 constructs

25 into 293T cells using calcium phosphate. Viral supernatants were then harvested and

1 used to infect CL1-0 or A549 cells in the presence of 8 μg/mL polybrene (Santa Cruz,

2 TX, USA). Cells were then selected using 2 g/mL puromycin (Santa Cruz, TX,

3 USA). FSTL1-expressing cells were established by infecting cells with the pLenti6.2-

4 FSTL1 virus, and viral supernatants were harvested and used to infect CL1-5 cells in

5 the presence of 8 μg/mL polybrene. Cells were selected using 5 μg/mL blasticidin

6 (Sigma-Aldrich, MO, USA).

7 Surface plasmon resonance (SPR) experiment

8 All experiments were performed using ProteOn XPR36 instruments developed by

9 Bio-Rad Haifa (Haifa). The GLC sensor chip and amine coupling kit were also

10 purchased from Bio-Rad. For the immobilization of the recombinant FSTL1 protein,

11 FSTL1 was coupled to the carboxymethylated alginate surface of a GLC capacity chip

12 (Bio-Rad) according to the protocol described in the Bio-Rad ProteOn One-Shot

13 Kinetics Kit Instruction Manual, with slight modifications. Direct binding

14 experiments were performed on the Bio-Rad ProteOn™ XPR 36 protein interaction

15 array system (Bio-Rad). Briefly, the surface was activated with 0.1 M N-

16 hydroxysuccinimide and 0.25 M N-ethyl-N'-(3-dimethylaminopropyl)carbodiimide at

17 a flow rate of 25 L/min. FSTL1 was diluted in 10 mM sodium acetate (pH 5.5) and

18 immobilized at 25°C using a flow rate of 25 L/min for 288 sec (120 L). Activated

19 carboxylic groups were quenched with an injection of 1 M ethanolamine (pH 8.0).

20 Analyte solutions were prepared at designated concentrations in filtered and degassed

21 PBS buffer (20 mM Na-phosphate and 150 mM NaCl, pH 7.4). All binding

22 experiments were conducted at 25°C with a constant flow rate of 100 L/min of PBS

23 buffer. Sensograms for all binding interactions were recorded in real time and

24 analyzed after subtracting the blank channel. After each measurement, the surface was

25 regenerated with 1.0 M NaCl. The equilibrium dissociation constants (KD) for

1 evaluating the protein-analyte binding affinity were determined with a steady-state

2 affinity fitting analysis using the results from ProteOn Manager 2.0 (Bio-Rad).

3 Microarray analysis

4 Total RNA was extracted from CL1-0 cells that had been treated with the FSTL1

5 antibody for 0 and 6 hours using a TRIzol RNA extraction kit (Invitrogen, CA, USA).

6 The synthesis of cDNAs from total RNA and microarray hybridization/scanning were

7 performed with Affymetrix GeneChip products (HG-U133A) by the GRC Microarray

8 Core Facility (Academia Sinica, Taiwan). Data files (.CEL) were converted into probe

9 set values (log2) by RMA normalization using GeneSpring (Agilent) and submitted to

10 the Gene Expression Omnibus (GEO) database with accession number GSE75961.

11 Data files (.CEL) from public microarray databases were also converted into probe set

12 values (log2) by RMA normalization using GeneSpring prior to performing further

13 analyses. Normalized RNA data from TCGA were downloaded from Cancer Browser.

14 Animal studies

15 All animal studies were performed in accordance with a protocol approved by the

16 Academia Sinica Institutional Animal Care and Use Committee (No. 11-12-277 to

17 Michael Hsiao). Female and male age-matched NOD-SCID mice (supplied

18 by LASCO, Taiwan) between 6 and 8 weeks old were used to assess tumor growth in

19 xenograft and lung colonization metastasis models. For primary tumor growth assays,

20 viable cells [5 × 106 cells per 100 µL of PBS] were subcutaneously injected into the

21 backs of mice. Primary tumor growth rates were analyzed by measuring the tumor

22 length (L) and width (W) and calculating the volume with the formula LW2/2. For

23 experimental metastasis assays, 1 × 106 viable cells were resuspended in 0.1 ml of

24 PBS and introduced into the circulation via a tail vein injection. Lung metastasis was

25 quantified 14 days after the injection and the metastatic tumors were traced every two

1 weeks using luciferase-based, noninvasive bioluminescent imaging. The analysis was

2 performed using the Xenogen IVIS-200 system (Xenogen). The orthotopic lung

3 metastasis model required 1 × 106 viable cells resuspended in 0.01 ml of

4 Matrigel:PBS at 1:1 ratio. The mixture was introduced into the left lungs of mice via

5 an orthotopic injection. Lung metastases were traced with bioluminescent imaging

6 every 2 days after treatment with rhFSTL1 (10 mg/kg).

7 Patient selection

8 Ninety-six patients diagnosed with NSCLC at the Kaohsiung Medical University

9 Hospital of Taiwan from 1991 to 2007 (KMUH-IRB-E(I)-20160099), including 58

10 patients with adenocarcinoma, 32 patients with squamous cell carcinoma, and 6

11 patients with large cell carcinoma, were included in this study (17). All patients

12 received standard treatment protocols according to hospital guidelines. Requirement

13 for informed consent was waived by the Institutional Review Boards of Kaohsiung

14 Medical University Hospital of Taiwan. Patients with operable stage I-III NSCLC

15 underwent lobectomy or pneumonectomy with mediastinal lymphadenectomy. No

16 adjuvant chemotherapy was administered to patients with completely resected stage I

17 NSCLC. Patients with resectable stage II and III NSCLC were treated with

18 postoperative adjuvant platinum-based chemotherapy. Patients with unresectable

19 locally advanced or metastatic disease received chemotherapy with or without

20 radiotherapy. Follow-up data were available for all patients, and the longest clinical

21 follow-up time was 190 months. All tumors were staged according to the cancer

22 staging manual of the American Joint Committee on Cancer and the histological

23 cancer type was classified according to the World Health Organization classification.

24 Immunohistochemical staining

25 Three representative 1 mm diameter cores from each tumor, which were taken from

1 formalin-fixed paraffin embedded tissues, were selected based on the morphology

2 typical of the diagnosis. Immunohistochemical (IHC) staining was performed on

3 serial 5 micrometer thick tissue sections cut from the tissue microarray (TMA). IHC

4 staining for FSTL1 was performed using an automated immunostainer (Ventana

5 Discovery XT autostainer). The antigens were retrieved by heat-induced antigen

6 retrieval in TRIS-EDTA buffer for 30 minutes. The slides were stained with a

7 polyclonal rabbit FSTL1 antibody (1:250; GeneTex). After an overnight incubation at

8 4℃, the slides were visualized using the 3,3’-diaminobenzidine (DAB) peroxidase

9 substrate kit (Vector Laboratories). The IHC staining assessment was independently

10 conducted by 2 pathologists who were blinded to patient outcomes. For FSTL1 IHC,

11 cytoplasmic expressions in tumor cells in the cores was evaluated. The intensity and

12 percentage of immunoreactive cells were recorded. The intensity of staining was

13 scored using a four-tier scale and defined as follows: 0, no staining; 1+, weak

14 staining; 2+, moderate staining; and 3+, strong staining. The extent of staining was

15 scored by determining the percentage of positive cells: 0, 0-25%; 1+, 26-50%; 2+, 51-

16 75%; and 3+, 75-100%. The final IHC scores were obtained by multiplying the scores

17 for the intensity and extent of staining. All patients were divided into two groups

18 according to the final IHC scores. A low IHC expression level was defined as a score

19 less than 4, and a score equal to or greater than 5 was defined as high expression.

20 Statistical analysis

21 All observations were confirmed in at least three independent experiments, and the

22 data are presented as means ± SD. Bioluminescence intensity data are presented as

23 means ± SE. Survival curves were analyzed using the Kaplan-Meier method, and the

24 Cox proportional hazards regression analysis was used to test the prognostic

25 significance of factors included in the univariate and multivariate models. All

1 statistical tests were two-sided. P < 0.05 was considered significant. Analyses were

2 performed using SPSS (Statistical Package for the Social Sciences, version 13.0)

3 software.

4 Results

5 Decreased expression of FSTL1 is significantly correlated with a poor prognosis

6 for patients with lung adenocarcinoma

7 We analyzed the public TCGA database to examine the clinical association

8 between the expression of follistatin and the follistatin-like protein family in patients

9 with lung cancer (Fig. 1A). The expression of the FST RNA significantly correlated

10 with the tumor status. In contrast, the RNA expression of FSTL family members

11 inversely correlated with the tumor status. The results of the survival analysis using

12 data from public microarray databases also showed significant correlations between

13 low FSTL1 RNA expression (P = 0.043) and poor first-progression survival (Fig. 1B).

14 We further compared the roles of FST family in different histological subtypes of

15 lung cancer (Supplementary Figure 1). Low FSTL1 expression retained its prognostic

16 significance in lung adenocarcinoma (P = 0.00021). Forest plots of FST and FSTL

17 protein family members and their corresponding hazard ratios and Cox-P values were

18 generated for the KM-plotter database cohort (Fig. 1C), and the results also showed

19 that FSTL1 was the prognostic marker with the lowest hazard ratio in patients with

20 lung cancer (HR=0.78, P = 9 x 10-5) and LUAD (HR=0.64, P = 2.1 x 10-4). Although

21 IGFBP7 has a low hazard ratio, the IGFBP7 probe does not have a significant Cox-P

22 values. Based on these data, the downregulation of FSTL1 is significantly associated

23 with disease-related progression in patients with LUAD.

24 FSTL1 expression negatively regulates the migration/invasion of LUAD cells in

25 vitro

1 Next, we analyzed the level of extracellular FSTL1 (Fig. 2A) and cellular migration

2 in the nontumor cell line Beas2B and LUAD cell lines A549, CL1-0, CL1-5, H928,

3 H1355, PC13 and PC14. Furthermore, we examined the endogenous protein

4 expression level FSTL1 with Western blot (Supplementary Figure 2). As shown in

5 Fig. 2B, the endogenous FSTL1 expression level negatively correlated with cellular

6 migration in the detected cells. The CL1-0 and CL1-5 cell lines were established with

7 low and high invasive abilities, respectively (20). The profiles of their mRNA

8 expression, protein expression, and secretomic analysis were well studied (22-27). We

9 provided additional background information, such as microarray data, NGS data, and

10 miRNA expression array data, in the revised supplementary information to ensure that

11 readers would clearly understand the differences in these two cell lines

12 (Supplementary Table 1). Furthermore, we artificially knocked down the expression

13 of FSTL1 using its specific shRNAs in CL1-0 cells with a high endogenous FSTL1

14 expression level but reduced migratory ability. The knockdown of FSTL1

15 substantially decreased the levels of the FSTL1 protein (Fig. 2C and Supplementary

16 Figure 3A), but increased the migration (Fig. 2D and Supplementary Figure 3B) and

17 invasion (Fig. 2E) of CL1-0 cells. We also established PC14 shFSTL1 clones which

18 presented similar phenomena with CL1-0 cells (Supplementary Figure 3C and 3D).

19 Conversely, we overexpressed the exogenous FSTL1 gene in CL1-5 cells with a low

20 endogenous FSTL1 expression level but strong migratory ability. The overexpression

21 of FSTL1 noticeably increased the levels of the FSTL1 protein (Fig. 2F and

22 Supplementary Figure 4A), but suppressed the migration (Fig. 2G) and invasion (Fig.

23 2H) of CL1-5 cells. We also established A549 FSTL1 overexpressed cells

24 (Supplementary Figure 4B and 4C). We further established an animal model to

25 evaluate the in vivo metastatic potential of CL1-0 and CL1-5 cells after FSTL1

1 knockdown and overexpression, respectively. The knockdown of FSTL1 robustly

2 enhanced lung colonization, as shown by an increased number of lung nodules and

3 enhanced colony formation (Fig. 2I) of CL1-0 cells. In contrast, the overexpression of

4 exogenous FSTL1 reduced the number of lung colonies formed (Fig. 2J) by CL1-5

5 cells.

6 FSTL1 downregulation does not affect cell proliferation in vitro but promotes

7 tumor growth of LUAD cells in vivo

8 We next determined the proliferation rate of the less metastatic CL1-0 cells after

9 FSTL1 knockdown to exclude the possibility that the enhanced migratory activity was

10 due to increased cell proliferation. As shown in Supplementary Figure 5A, FSTL1

11 knockdown did not alter the proliferation rates of CL1-0 cells. Similarly, FSTL1

12 overexpression did not alter the proliferation of CL1-5 cells (Supplementary Figure

13 5B). Thus, FSTL1 downregulation promotes metastatic progression but does not

14 increase the proliferation of LUAD cells.

15 Intriguingly, FSTL1 expression negatively correlated with the growth of tumors

16 composed of LUAD cells in vivo (Supplementary Figure 5C). Based on these

17 findings, an FSTL1 deficiency fosters the tumorigenesis through a cell proliferation-

18 independent mechanism, e.g., angiogenesis, in LUAD.

19 Reduced FSTL1 expression inhibits its antagonistic effect on the extracellular

20 matrix and ultimately promotes the migration/invasion of LUAD cells

21 Since FSTL1 functions as a biological antagonist for the transforming growth

22 factor- (TGF-) superfamily and thereby modulates cellular migration, we next

23 determined the effects of neutralizing the extracellular FSTL1 with its specific

24 antibody or replenishing it with the recombinant FSTL1 protein on the migration and

25 invasion of LUAD cells. Treatment with an FSTL1-specific antibody dose-

1 dependently increased the migration and invasion of CL1-0 cells (Fig. 3A and 3B);

2 conversely, the replenishment with the recombinant FSTL1 protein inhibited the

3 migration and invasion of CL1-5 cells in a dose-dependent manner (Fig. 3C and 3D).

4 Moreover, treatment with the recombinant FSTL1 protein compromised the increase

5 in the migration/invasion of CL1-0 induced by FSTL1 knockdown (Fig. 3E). In

6 contrast, the addition of the FSTL1-specific antibody reversed the suppression of the

7 migration/invasion of CL1-5 cells induced by FSTL1 overexpression (Fig. 3F).

8 Therefore, FSTL1 downregulation leads to the dysregulation of FSTL1-modulated

9 cellular functions, likely through a protein-protein or ligand-receptor interaction in the

10 extracellular matrix, and thereby drives metastatic evolution in LUAD.

11 FSTL1 treatment effectively inhibits the tumor growth and metastatic

12 progression of malignant LUAD cells with FSTL1 downregulation in vivo

13 Next, we evaluated the therapeutic potential of FSTL1 in inhibiting the tumor

14 growth and metastasis of malignant LUAD cells with FSTL1 downregulation.

15 Treatment with the recombinant FSTL1 protein (rhFSTL1) substantially inhibited the

16 tumor growth of highly malignant CL1-5 cells that exhibit reduced endogenous

17 FSTL1 levels (Fig. 4A). Moreover, compared to the untreated group, the treatment

18 with rhFSTL1 significantly suppressed the colony-forming ability of highly invasive

19 CL1-5 cells in the lung of the animal model (Fig. 4B). Importantly, the treatment with

20 rhFSTL1 restrained the metastatic progression and tumor growth, of CL1-5 (Fig. 4C-

21 4F and Supplementary Figure 6) and A549 (Supplementary Figure 7A-7C) cells in the

22 orthotopic lung cancer model.

23 The binding of FSTL1 to pro-SPP1 prevents the proteolytic activation of SPP1

24 and thereby inhibits the metastatic progression of LUAD cells

25 We next overexpressed exogenous His-tagged FSTL1 and performed Ni-NTA-

1 based affinity chromatography (Fig. 5A) to isolate proteins that bind to His-tagged

2 FSTL1 in the extracellular matrix of CL1-5 cells and to identify the PPI network of

3 FSTL1 in the highly metastatic CL1-5 cells. As shown in Figure 5B, several proteins

4 from the extracellular matrix or exosomal compartment were identified by a mass

5 spectrometry analysis of the FSTL1-interacting protein mixture (Supplementary Table

6 2). Since SPP1 has been considered an oncogenic driver that tumorigenesis and

7 metastasis via regulating v3 integrin and CD44-mediated cellular functions in

8 various cancer types, the next experiments were thus designed to validate the protein-

9 protein interaction (PPI) between FSTL1 and SPP1 and the functional consequence of

10 this PPI in governing metastatic evolution in LUAD cells. Surface plasmon resonance

11 analysis clearly revealed an interaction between FSTL1 and the pro-form, but not the

-8 12 cleaved forms, of SPP1, with a strong binding affinity (KD = 3.48 x 10 M) (Fig. 5C).

13 The addition of active SPP1 fragment, but not inactive pro-SPP1 protein, robustly

14 restored the FSTL1-induced decrease in the migration/invasion of CL1-5 cells (Fig.

15 5D). In contrast, the inclusion of v3 integrin and CD44-specific antibodies

16 dramatically inhibited the increase in the migration/invasion of CL1-0 cells after

17 FSLT1 neutralization using its specific antibody (Fig. 5E). In addition, treatment with

18 the FSTL1 antibody induced the phosphorylation of Src and Akt (Fig. 5F), which are

19 downstream effectors of v3 integrin and CD44. These findings confirm that the

20 direct binding of FSTL1 to pro-SPP1, thereby restraining SPP1 proteolytic activation,

21 is a key step in preventing metastatic progression in LUAD.

22 The ILK signaling pathway is involved in FSTL1-mediated LUAD cell migration

23 We performed 3 microarray analyses using CL1-0 cells transfected with or without

24 shFSTL1, CL1-0 cells treated with or without the FSTL1 antibody (GSE79682), and

25 CL1-5 cells treated with or without rhFSTL1 to determine whether another novel

1 pathway regulated FSTL1-mediated LUAD cell migration. The genes displaying 1.5-

2 fold changes in expression levels compared to the control group were subsequently

3 subjected to the computational simulation with Ingenuity Pathway Analysis (IPA)

4 software to identify canonical pathways that were altered upon FSTL1 modulation

5 (Fig. 6A). The simulation of canonical pathway activity showed that HMGB1

6 signaling, the SUMOylation pathway, and ILK signaling were strongly predicted to

7 be activated (Fig. 6A, right) after FSTL1 is blocked with its specific antibody in CL1-

8 0 cells and in CL1-0/shFSTL1 cells. Based on these results, we compared the data

9 obtained from rhFSTL1-treated CL1-5 cells. HMGB1 signaling, the SUMOylation

10 pathway, and ILK signaling pathway were predicted to be downregulated. In Figure

11 6B, the IPA results revealed the activation of most genes in the ILK signaling

12 pathway after the antibody treatment. Furthermore, F-actin, G-actin, and fibronectin

13 mRNAs, which are related to cell migration, were upregulated after FSTL1 blockade.

14 Compared with the microarray data obtained from cells treated with the antibody and

15 recombinant protein, the expression of genes in the ILK signaling pathway was

16 significantly contrast. (Fig. 6C). In our microarray data, FSTL1 did not affect the

17 expression of the integrin 3 mRNA, the trigger of ILK signaling. We next examined

18 whether FSTL1 affected the integrin 3 promoter (Fig. 6D), but not identify any

19 activated transcription factors after the FSTL1 antibody treatment (Fig. 6E).

20 Furthermore, we presented migration assay with 2D08 (SUMOylation pathway

21 inhibitor), BAY117082 (NF-kB pathway inhibitor) and Quercetin (HMGB1 signaling

22 pathway inhibitor) in FSTL1 antibody treated CL1-0 cells. The results showed that

23 only BAY117082 reversed the migration ability induced by FSTL1 antibody

24 (Supplementary Figure 8). Thus, FSTL1 may mediate LUAD cell migration by

25 activating the ILK signaling pathway through different mechanisms such as protein-

1 protein interactions (PPIs).

2 Blocking peptides mimic the interaction between FSTL1 and pro-SPP1 to inhibit

3 LUAD cell migration

4 Based on the sequence covered by FSTL1 binding to SPP1, we designed 4 short

5 peptides (15 a.a.) to mimic the behaviors of FSTL1 (Fig. 7A). When CL1-0 cells were

6 treated with pro-SPP1, migration was induced. Migration was reduced after treatment

7 with the combination of the P1 and P2 peptides, but not the P3 and P4 peptides. Thus,

8 blocking the exact binding site of SPP1 potentially interrupted the subsequent

9 activation of the integrin-triggered ILK signaling pathway (Fig. 7B). Furthermore,

10 following phalloidin staining of FSTL1 knockdown CL1-0 cells and FSTL1-

11 overexpressing CL1-5 cells, FSTL1 inhibited intracellular F-actin polymerization

12 (Fig. 7C). On the other hand, the signature of combining a low level of FSTL1 and

13 high level a SPP1 predicted a substantially worse prognosis for patients with LUAD

14 (Fig. 7D and Supplementary Table 3). These findings implicate a broad therapeutic

15 value of FSTL1 in combating malignancies with a low ratio of FSTL1/SPP1 levels.

16 We thus provide a novel model of FSTL1-mediated pro-SPP1 maturation to inhibit

17 LUAD progression and metastasis.

1 Discussion

2 As shown in the present study, FSTL1 plays an important role in preventing the

3 metastatic progression of lung adenocarcinoma, but not squamous cell carcinoma. By

4 mining data from the TCGA database, we emphasize the role of FSTL1

5 downregulation in lung adenocarcinoma. Higher serum FST levels have been

6 observed in patients with lung adenocarcinoma than in healthy people (10). The

7 results convinced us because we the overexpression of the FST mRNA correlated

8 with a poor prognosis for patients with lung cancer. In our previous studies, low

9 FSTL1 expression significantly correlated with poor prognosis in LUAD but not in

10 SCC (17). In addition, FSTL1 was reported to suppress tumor cell proliferation,

11 invasion and survival in NSCLC (28). Thus, the role of FSTL1 appeared to differ

12 from the role of FST in specific subtypes of lung adenocarcinoma. Although a loss of

13 FSTL1 expression predicted a poor prognosis for patients with lung adenocarcinoma

14 in the present study, it has been reported to exert the opposite effects on patients with

15 different types of tumors. FSTL1 expression inhibits the growth and invasion of lung

16 cancer cells, and an FSTL1 antibody treatment blocks invasion. In endometrial and

17 ovarian cancer, FSTL1 also possesses a tumor suppressor function that regulates cell

18 proliferation, apoptosis, invasion and migration (29). Downregulation of FSTL1

19 expression has also been reported in various tumor cell lines, including clear cell renal

20 cell carcinoma, colon cancer and gastric cancer (14,30). In contrast, a recent study of

21 breast cancer showed that FSTL1 increases chemoresistance by regulating tumor

22 stemness (31). Upregulation of FSTL1 is also observed in glioblastoma and is

23 associated with a poor prognosis for patients with glioblastoma (32). These findings

24 confirm the hypothesis that the function of FSTL1 varies among different cancer

25 types. In addition, treatment with the recombinant FSTL1 protein significantly

1 inhibited tumor growth in highly malignant CL1-5 cells with lower endogenous

2 FSTL1 expression. Moreover, treatment with FSTL1 significantly inhibited the

3 formation of lung cancer nodules by highly invasive CL1-5 cells in animal models

4 compared to the untreated group. Importantly, treatment with the recombinant FSTL1

5 protein inhibited the metastatic progression of CL1-5 and A549 cells in an orthotopic

6 lung tumor metastasis model.

7 SPP1 is a pleiotropic chemokine that is overexpressed in various types of cancer.

8 Elevated serum SPP1 levels are frequently detected in patients with metastatic cancer

9 (33-35). Therefore, therapeutic strategies targeting osteopontin by repressing gene

10 expression or blocking its binding to v3 integrin and CD44 have recently been

11 considered useful approaches to overcome cancer metastasis (36-38). Here, FSTL1

12 directly bound to nascent SPP1 and thereby inhibited its proteolytic processing by

13 matrix metalloproteinases-3/7 or thrombin protease into active SPP1 to prevent the

14 SPP1-induced activation of αvβ3 integrin and CD44 in metastatic cancer cells.

15 Importantly, the signature of a high level of SPP1 and low level of FSTL1

16 predominantly correlates with a poor distant metastasis-free survival rate in clinical

17 patients with basal-like breast cancer, a highly metastatic breast cancer

18 (Supplementary Figure 9A). These findings imply a broad therapeutic value of FSTL1

19 in combating SPP1-driven cancer metastasis.

20 In addition to SPP1, a mass spectrometry analysis of an FSTL1-interacting protein

21 mixture obtained from highly metastatic cancer cells showed that FSTL1 may also

22 interact with other extracellular proteins. We also designed 4 short peptides (15 a.a.)

23 to mimic the behaviors of FSTL1 and block SPP1-mediated migration. However,

24 further studies are required to delineate these protein-protein interactions (PPIs) of

25 FSTL1 in metastatic cancers. On the other hand, the molecular mechanism underlying

1 FSTL1 downregulation during metastatic progression in cancer remains to be

2 elucidated. In conclusion, this study is the first to document the PPI between FSTL1

3 and SPP1, as well as the potential usefulness of this PPI in preventing cancer

4 metastasis in the clinic.

1 Acknowledgements

2 This study was supported by Academia Sinica and Ministry of Science and

3 Technology grants (AS-SUMMIT-108) awarded to Michael Hsiao.

1 References

2 1. Sylva M, Li VSW, Buffing AAA, van Es JH, van den Born M, van der Velden 3 S, et al. The BMP Antagonist Follistatin-Like 1 Is Required for Skeletal and 4 Lung Organogenesis. PloS One 2011;6:10 5 2. Geng Y, Dong YY, Yu MY, Zhang L, Yan XH, Sun JX, et al. Follistatin-like 6 1 (Fstl1) is a bone morphogenetic protein (BMP) 4 signaling antagonist in 7 controlling mouse lung development. Proceedings of the National Academy of 8 Sciences of the United States of America 2011;108:7058-63 9 3. Geng Y, Li L, Dong Y, Liu X, Li XH, Ning W. Impaired elastin deposition in 10 Fstl1-/- lung allograft under the renal capsule. PloS one 2013;8:e81368 11 4. Bailey JM, Singh PK, Hollingsworth MA. Cancer metastasis facilitated by 12 developmental pathways: Sonic hedgehog, Notch, and bone morphogenic 13 proteins. J Cell Biochem 2007;102:829-39 14 5. Kelleher FC, Fennelly D, Rafferty M. Common critical pathways in 15 embryogenesis and cancer. 2006;45:375-88 16 6. Fei ZH, Yao,C.Y., Yang,X.L., Huang,X.E. and Ma,S.L. Serum BMP-2 up- 17 regulation as an indicator of poor survival in advanced non-small cell lung 18 cancer patients. Asian PacJCancer Prev 2013;14:5293-9 19 7. Hao J, Lee,R., Chang,A., Fan,J., Labib,C., Parsa,C., Orlando,R., Andresen,B. 20 and Huang,Y. DMH1, a small molecule inhibitor of BMP type i receptors, 21 suppresses growth and invasion of lung cancer,. PLoSOne 2014;9:e90748 22 8. Langenfeld E, Hong,C.C., Lanke,G. and Langenfeld,J. Bone morphogenetic 23 protein type I receptor antagonists decrease growth and induce cell death of 24 lung cancer cell lines. PloS One 2013;8:e61256 25 9. Ye XY, Niu,X.M., Tang,N.W., Xu,Y.H., Li,Z.M., Yu,Y.F., Lu,S. and 26 Chen,S.W. Adenovirus mediated knockdown of bone morphogenetic protein 2 27 inhibits human lung cancer growth and invasion in vitro and in vivo. 28 IntJImmunopatholPharmacol 2012;25:967-76 29 10. Chen F RP, Feng Y, Liu H, Sun Y, Liu Z, Ge J, Cui X. Follistatin is a novel 30 biomarker for lung adenocarcinoma in humans. PLoS One 2014;9:e111398 31 11. Couto HL BM, Toppa NH, Bloise E, Wainstein AJ, Reis FM. Prognostic value 32 of follistatin-like 3 in human invasive breast cancer. Oncotarget 33 2017;8:42189–97 34 12. Okamura J HY, Moon D, Brait M, Chang X, Kim MS. Downregulation of 35 insulin-like growth factor-binding protein 7 in cisplatin-resistant non-small 36 cell lung cancer. Cancer Biol Ther 2012;13:148-55 37 13. Zhang D MX, Sun W, Cui P, Lu Z. Down-regulated FSTL5 promotes cell 38 proliferation and survival by affecting Wnt/β-catenin signaling in 39 hepatocellular carcinoma. Int J Clin Exp Pathol 2015; 8:3386-94 40 14. Mashimo J, Maniwa R, Sugino H, Nose K. Decrease in the expression of a 41 novel TGF beta1-inducible and ras-recision gene, TSC-36, in human cancer 42 cells. Cancer Lett 1997;113:213-9 43 15. Sumitomo K, Kurisaki A, Yamakawa N, Tsuchida K, Shimizu E, Sone S, et 44 al. Expression of a TGF-beta 1 inducible gene, TSC-36, causes growth 45 inhibition in human lung cancer cell lines. Cancer Lett 2000;155:37-46 46 16. Zhao W, Han HB, Zhang ZQ. Suppression of lung cancer cell invasion and 47 metastasis by connexin43 involves the secretion of follistatin-like 1 mediated 48 via histone acetylation. Int J Biochem Cell Biol 2011;43:1459-68 49 17. Chiou J SC, Jan YH, Yang CJ, Huang MS, Yu YL, Hsiao M. Decrease of 23

1 FSTL1-BMP4-Smad signaling predicts poor prognosis in lung 2 adenocarcinoma but not in squamous cell carcinoma. Scientific Reports 3 2017;7:9830 4 18. Reck M, Heigener DF, Mok T, Soria JC, Rabe KF. Management of non-small- 5 cell lung cancer: recent developments. Lancet 2013;382:709-19 6 19. Simon GRaT, A. Management of small cell lung cancer: ACCP evidence- 7 based clinical practice guidelines (2nd edition), . Chest 2007;132:324S-39S 8 20. Chu YW YP, Yang SC, Shyu YC, Hendrix MJ, Wu R, Wu CW. Selection of 9 invasive and metastatic subpopulations from a human lung adenocarcinoma 10 cell line. Am J Respir Cell Mol Biol 1997;17:353-60 11 21. Lee YC SN, Sasaki Y, Takahashi H, Sakurai M, Ishihara J, Hoshi A, Chen 12 KM and Hamburger AW. Clonogenic patterns of human pulmonary 13 adenocarcinoma cell lines (PC-9, PC-13 and PC-14) and how they influence 14 the results of test for chemosensitivity to cisplatin in the human tumor 15 clonogenic assay. JpnJClinOncol 1985;15:637-44 16 22. Chang YH LS, Chang HC, Tseng YL, Lai WW, Liao CC, Tsay YG, Liao PC. 17 Comparative Secretome Analyses Using a Hollow Fiber Culture System with 18 Label-Free Quantitative Proteomics Indicates the Influence of PARK7 on Cell 19 Proliferation and Migration/Invasion in Lung Adenocarcinoma. J Proteome 20 Res 2012;11:5167-85 21 23. Su LJ CC, Wu YC, Chen KC, Lin CJ, Liang SC, Lin CH, Whang-Peng J, Hsu 22 SL, Chen CH, Huang CY. Selection of DDX5 as a novel internal control for 23 Q-RT-PCR from microarray data using a block bootstrap re-sampling scheme. 24 BMC Genomics 2007;1:140 25 24. Chen CY JY, Juan YH, Yang CJ, Huang MS, Yu CJ, Yang PC, Hsiao M, Hsu 26 TL, Wong CH. Fucosyltransferase 8 as a functional regulator of nonsmall cell 27 lung cancer. Proc Natl Acad Sci U S A 2013;110:630-5 28 25. Tian T HJ, Xu A, Hao J, Luo C, Liu C, Huang L, Xiao X, He D. 29 Determination of metastasis-associated proteins in non-small cell lung cancer 30 by comparative proteomic analysis. Cancer Sci 2007;98:1265-74 31 26. Chang YH LS, Liao IC, Huang SH, Cheng HC, Liao PC. Secretomic analysis 32 identifies alpha-1 antitrypsin (A1AT) as a required protein in cancer cell 33 migration, invasion, and pericellular fibronectin assembly for facilitating lung 34 colonization of lung adenocarcinoma cells. Mol Cell Proteomics 35 2012;11:1320-29 36 27. Chiu KH CY, Wu YS, Lee SH, Liao PC. Quantitative secretome analysis 37 reveals that COL6A1 is a metastasis-associated protein using stacking gel- 38 aided purification combined with iTRAQ labeling. J Proteome Res 39 2011;10:1110-25 40 28. Ni XL CX, Wu YQ, Wu J. FSTL1 suppresses tumor cell proliferation, 41 invasion and survival in non-small cell lung cancer. Oncol Rep 2018;39:13-20 42 29. Chan Q.K. NHY, Ip P.P., et al. Tumor suppressor effect of follistatin-like 1 in 43 ovarian and endometrial carcinogenesis-025EFa differential expression and 44 functional analysis. Carcinogenesis 2009;30:114-21 45 30. Tan XJ ZY, Chang WJ, et al. Global analysis of metastasis-associated gene 46 expression in primary cultures from clinical specimens of clear-cell renal-cell 47 carcinoma. . International Journal of Cancer 2008;123:1080-8 48 31. Cheng S HY, Lou C, He Y, Zhang Y, Zhang Q. FSTL1 enhances 49 chemoresistance and maintains stemness in breast cancer cells via integrin 50 β3/Wnt signaling under miR-137 regulation. Cancer Biol Ther 2018;18:1-10 24

1 32. Reddy SP BR, Vinnakota K, et al. Novel glioblastoma markers with diagnostic 2 and prognostic value identified through transcriptome analysis. Clin Cancer 3 Res 2008;14:2978-87 4 33. Qin L. Osteopontin is a promoter for hepatocellular carcinoma metastasis: a 5 summary of 10 years of studies. Front Med 2014;8:24-32 6 34. Shevde LA, Das,S., Clark,D.W. and Samant,R.S. Osteopontin: an effector and 7 an effect of tumor metastasis. CurrMolMed 2010;10:71-81 8 35. Wai PYaK, P.C. Osteopontin: regulation in tumor metastasis. Cancer 9 Metastasis Rev 2008;27:103-18 10 36. Rao G, Wang,H., Li,B., Huang,L., Xue,D., Wang,X., Jin,H., Wang,J., Zhu,Y., 11 Lu,Y., Du,L. and Chen,Q. Reciprocal interactions between tumor-associated 12 macrophages and CD44-positive cancer cells via osteopontin/CD44 promote 13 tumorigenicity in colorectal cancer. ClinCancer Res 2013;19:785-97 14 37. Zhang H, Guo,M., Chen,J.H., Wang,Z., Du,X.F., Liu,P.X. and Li,W.H. 15 Osteopontin knockdown inhibits alphav,beta3 integrin-induced cell migration 16 and invasion and promotes apoptosis of breast cancer cells by inducing 17 autophagy and inactivating the PI3K/Akt/mTOR pathway. Cell Physiol 18 Biochem 2014;33:991-1002 19 38. Pietras A, Katz,A.M., Ekstrom,E.J., Wee,B., Halliday,J.J., Pitter,K.L., 20 Werbeck,J.L., Amankulor,N.M., Huse,J.T. and Holland,E.C. Osteopontin- 21 CD44 signaling in the glioma perivascular niche enhances cancer stem cell 22 phenotypes and promotes aggressive tumor growth. Cell Stem Cell 23 2014;14:357-69 24

1 Figure legends

3 Figure 1. FSTL1 down-regulation significantly correlates with poor prognosis in

4 LUAD patients. (A) The TCGA database heatmap indicates the correlation between

5 mRNA expression of FST and FSTL family and tumor status. (B) Low FSTL1

6 expression was correlated with a poor first-progression survival rate in patients in

7 KM-plotter public database. (C) The hazard ratios of FST and FSTL protein family.

8 FSTL1 was the only prognositic biomarker that have a lowest hazard ratio in both

9 lung cancer and LUAD patients cohort of KM-plotter public database (n=1926 and

10 720).

12 Figure 2. FSTL1 down-regulation correlates with the enhanced

13 migration/invasion abilities of LUAD cells in vitro. (A) FSTL1 protein expression

14 in lung adenocarcinoma cell lines. ELISA assay of the cultured medium. (B) The

15 correlation between FSTL1 expression level and migration ability of LUAD cell lines.

16 (C) FSTL1 knockdown in CL1-0 cells. ELISA analysis of FSTL1. Bar, means ± SD.

17 (D) Migration activity of FSTL1 knockdown cells. comparative xCELLigence

18 analysis of the effect of migration of CL1-0 cell lines. CI values for all cell lines are

19 plotted on the graph. The experiment was repeated 3 times and was consistent. Bar,

20 Standard deviation. (E) Invasion activity of FSTL1 knockdown cells. The columns

21 represent the mean values from three independent experiments. Bars indicate means ±

22 SD. (F) FSTL1 overexpression in CL1-5 cells. ELISA analysis of FSTL1. Bar, means

23 ± SD. (G) Migration activity of FSTL1 overexpression cells. comparative

24 xCELLigence analysis of the effect of migration of CL1-5 cell lines. CI values for all

25 cell lines are plotted on the graph. The experiment was repeated 3 times and was

1 consistent. Bar, Standard deviation. (H) Invasion activity of FSTL1 overexpression

2 cells. The columns represent the mean values from three independent experiments.

3 Bars indicate means ± SD. (I) Knockdown of FSTL1 expression promoted metastasis

4 in vivo. Up: representative lung images of mice injected with a non-silencing shRNA

5 and FSTL1 shRNA-expressing CL1- 0 cells. Arrowheads indicate lung metastatic

6 nodules. Bottom, total numbers of lung metastatic nodules in individual mice 10

7 weeks after tail vein injection with CL1-0 cells infected with a non-silencing shRNA

8 or FSTL1 shRNA. (J) FSTL1 expression inhibited metastasis in vivo. Up:

9 representative lung images of mice injected with a vector control and FSTL1-

10 expressing CL1- 5 cells. Arrowheads indicate lung metastatic nodules. Bottom, total

11 numbers of lung metastatic nodules in individual mice 4 weeks after tail vein injection

12 with CL1-5 cells infected with a vector control or FSTL1.

14 Figure 3. FSTL1 stimuli inversely promotes the migration/invasion abilities of

15 LUAD cells in vitro. (A) FSTL1 antibody treatment induced CL1-0 cells migration.

16 Cells with FSTL1 antibodies (1ug/mL and 10 ug/mL) were seeded on Transwell CIM-

17 Plate 16 plates and subjected to real-time migration assays (xCELLigence) over a

18 time period of 70 h. Bars indicate means ± SD. (B) FSTL1 recombinant protein

19 treatment inhibited CL1-5 cells migration. Cells with FSTL1 recombinant protein (10

20 ng/mL, 30 ng/mL and 100 ng/mL) were seeded on Transwell CIM-Plate 16 plates and

21 subjected to real-time migration assays (xCELLigence) over a time period of 25

22 hours. Bars indicate means ± SD. (C) FSTL1 antibody treatment induced CL1-0 cells

23 invasion ability. Cells with FSTL1 antibodies (1ug/mL and 10 ug/mL) were seeded on

24 Transwell pre-coated with matrigel in the upper well and subjected to invasion assays

25 for 36 hrs. The columns represent the mean values from three independent

1 experiments. Bars indicate means ± SD. (D) FSTL1 recombinant protein treatment

2 inhibited CL1-5 cells invasion. Cells with FSTL1 recombinant protein (10 ng/mL, 30

3 ng/mL and 100 ng/mL) were seeded on Transwell pre-coated with matrigel in the

4 upper well and subjected to invasion assays for 16 hrs. The columns represent the

5 mean values from three independent experiments. Bars indicate means ± SD.

6 (E) Recombinant FSTL1 protein restored shFSTL1-induced cell migration and

7 invasion. Cell migration/invasive abilities were measured by Transwell assay in CL1-

8 0/KD cells. The columns represent the mean values from three independent

9 experiments. Bars indicate means ± SD. (F) FSTL1 antibody restored FSTL1-

10 inhibited cell migration and invasion. Cell migration/invasive abilities were measured

11 by Transwell assay in CL1-5/OE cells. The columns represent the mean values from

12 three independent experiments. Bars indicate means ± SD.

14 Figure 4. FSTL1 Treatment suppresses tumor growth and lung metastatic ability

15 of LUAD cells in vivo. (A) CL1-5-GL cells were s.c. injected into NOD-SCID mice

16 that were treated over an interval of 2 days with saline only (PBS) and rhFSTL1 10

17 mg/kg. Left: Photos of tumors after 3 weeks of treatment. Right: Quantification of

18 tumor weight. (B) CL1-5-GL cells were i.v. injected into NOD-SCID mice that were

19 treated over an interval of 2 days with saline only (PBS) and rhFSTL1 10 mg/kg.

20 Upper: Luminescence image measured using a noninvasive, bioluminescence system

21 (IVIS spectrum) at days 20. The tumor growth is expressed as the bioluminescence

22 intensity (BLI) change (5 mice/group). Quantification of bioluminescence intensity

23 (BLI) in whole mice(left) at day 20. Lower: total numbers of lung metastatic nodules

24 in individual mice 20 days after tail vein injection with CL1-5 cells and treated over

25 an interval of 2 days with saline only (PBS) and rhFSTL1 10 mg/kg. (C) CL1-5-GL

1 cells were orthotopic injected into NOD-SCID mice that were treated over an interval

2 of 2 days with saline only (PBS) and rhFSTL1 10 mg/kg. Luminescence image

3 measured using a non-invasive, bioluminescence system (IVIS spectrum) at days 0

4 (left) and 16 (right) (D) Luminescence in the lung of mice treated with PBS or

5 rhFSTL1 10mg/kg at day 16. (E) rhFSTL1 inhibited metastasis of lung cancer in

6 orthotopic model. Green fluorescence in the lungs of mice treated with PBS or

7 rhFSTL1 10mg/kg at day 16 after iv injection. (field-40X). Columns: the activity of

8 GFP in the fields. Bars indicate means ± SD. (F) H&E staining of mice lung. Upper

9 panel: 12.5X, red square showed the 200X area. Lower panel: 200X

11 Figure 5. FSTL1 inhibits cellular migration/invasion abilities via prevents the

12 proteolytic activation of SPP1 in LUAD cells. (A)FPLC elution of CL1-5/FSTL1

13 conditioned medium binding with His-Trap column. (B) Surface plasmon resonance

14 (SPR) experiment for FSTL1 and SPP1 binding affinity. (C) Ingenuity Pathway

15 Analysis (IPA) of microarray results of CL1-0 cells treated with FSTL1 antibody for

16 6h. (D) Migration assay of CL1-5 treated with recombinant FSTL1 and SPP1 protein.

17 Bars indicate means ± SD. *: P <0.05. (E) Migration assay of CL1-0 treated with

18 FSTL1 and integrin alphav, integrin beta3 and CD44 antibody. Bars indicate means ±

19 SD. *: P <0.05. (F) Western blot of phosphor-Src and PKB/Akt in CL1-0 cells treated

20 with FSTL1 and SPP1 antibody.

22 Figure 6. ILK signaling involved in FSTL1-inhibited LUAD cell migration. (A)

23 Left: The 1.5 fold changing gene numbers were shown in the circle. Right: The tables

24 showed the ranking of the predicted canonical pathways in 3 microarray analysis. (B)

25 The ILK signaling network was predicted based on the common signature from the

1 IPA database overlaid with microarray data from FSTL1 antibody treatment CL1-0

2 cells with a 1.5 fold-change cutoff compared with CL1-0 control cells. The intensity

3 of the color indicates the degree of activating (red and pink). (C) The heatmap

4 indicates the intensity of mRNA expression in CL1-0+Anti-FSTL1 cells and CL1-5+

5 rhFSTL1 cells. Red: activated; Blue: downregulated; White: no significant difference.

6 (D) Prediction of integrin b3 promoter binding site (E) IPA prediction of transcription

7 factor of integrin b3 promoter

9 Figure 7. Peptides interrupt the SPP1 maturation to mimic FSTL1-inhibited

10 LUAD cell migration. (A) The brief diagram of the peptides (B) The peptides

11 interrupted the CL1-0 invasion induced by SPP1. Cells with SPP1 recombinant

12 protein (500 ng/mL) and combined peptides treatment (500 ng/mL) were seeded on

13 Transwell pre-coated with matrigel in the upper well and subjected to invasion assays

14 for 16 hrs. The columns represent the mean values from three independent

15 experiments. Bars indicate means ± SD. (C) Left: Actin polymerization increased

16 after downregulation of FSTL1 in CL1-0 cells. Right: FSTL1 overexpression

17 inhibited intracellular tubular forming in CL1-5 cells. (D) Kaplan-Meier survival

18 curve analysis of lung cancer and LUAD patients with FSTL1 and SPP1 levels as

19 determined by KM-plotter database analysis at the endpoint of overall survival.

Follistatin-like protein 1 inhibits lung cancer metastasis by preventing proteolytic activation of osteopontin

Jean Chiou, Yu-Chan Chang, Hsing-Fang Tsai, et al.

Cancer Res Published OnlineFirst October 25, 2019.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-19-0842

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2019/10/25/0008-5472.CAN-19-0842.DC1

Author Author manuscripts have been peer reviewed and accepted for publication but have not yet Manuscript been edited.

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

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

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/early/2019/10/25/0008-5472.CAN-19-0842. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.