MFAP2 promotes epithelial-mesenchymal transition in gastric cancer cells by activating TGF-β/SMAD2/3 signaling pathway.

INTRODUCTION: Microfibril-associated protein 2 (MFAP2) is an extracellular matrix protein that interacts with fibrillin to modulate the function of microfibrils. MFAP2 has been reported to play a significant role in obesity, diabetes, and osteopenia, and has been shown to be upregulated in head and neck squamous cell carcinoma. However, the molecular function and prognostic value of MFAP2 have never been reported in gastric cancer (GC) or any other tumors.
METHODS: The current study investigated the expression patterns, prognostic significance, functional role, and possible mechanisms of MFAP2 in GC.
RESULTS: We demonstrated that MFAP2 was overexpressed in GC tissues, and its overexpression was significantly correlated with poor overall and disease-free survival in patients with GC. Moreover, we found that MFAP2 promoted the proliferation, migration, invasion, and epithelial-mesenchymal transition (EMT) phenotype in GC cells. MFAP2 might modulate EMT of GC cells by activating the TGF-β/SMAD2/3 signaling pathway.
CONCLUSION: These findings provide novel evidence that MFAP2 plays a crucial role in the progression of GC. Therefore, MFAP2 may be a promising prognostic marker and a potent anticancer agent.

Introduction

Gastric cancer (GC) is the fourth most common cancer and the second most common cause of cancer mortality worldwide.1,2 Surgical treatment can cure 90% of patients with early GC,3 but most patients are diagnosed with GC in its late stages.4 Approximately 60% of patients with GC have locally advanced and metastatic tumors at the time of surgery, resulting in a relatively low therapeutic efficacy.5 Although the clinical treatment of GC has been significantly enhanced in the recent years due to improvements in surgical techniques and chemotherapy, the long-term survival rate of patients with GC remains poor.6 At present, the valid therapeutic methods for advanced GC with invasion and metastasis remain poor and limited.7 As a result, investigation of the molecular mechanisms underlying GC invasion and metastasis will provide a strong theoretical basis for its diagnosis and treatment.

Microfibril-associated protein 2 (MFAP2), also known as microfibril-associated glycoprotein 1 (MAGP1), is an abundant component of microfibrils.8,9 There is a fibrillin-binding domain in the carboxyl-terminal region of MFAP2, which contains 13 cysteine residues and binds MFAP2 to the extracellular matrix and an acidic amino-terminal region containing a growth factor interaction motif capable of interacting with the active form of transforming growth factor beta (TGF-β), but not with the latent form of the growth factor.10,11 MFAP2 plays an important role in the extracellular deposition of fibrillin-1 during development of the human ciliary zonule,12 and an in vivo study found that MFAP2 plays a key role in hemostasis and thrombosis because MFAP2-deficient mice showed delayed thrombotic occlusion of the carotid artery following injury, prolonged bleeding, and lower platelet number.13

Thus far, studies of MFAP2 have mainly focused on its function in modulating tropoelastin deposition onto micro-fibrils for formation of elastic fibers,14 and few studies have been conducted concerning the role of MFAP2 in cancers. MFAP2 was identified as one of the most co-expressed genes of the NF-κB/Snail/YY1/RKIP circuitry in multiple myeloma.15 In head and neck squamous cell carcinoma, MFAP2 has been shown to be significantly upregulated in tumor tissues, especially in lymph node metastasis.16 However, the function and prognostic role of MFAP2 were not investigated in these two studies. To the best of our knowledge, no studies have reported the molecular role and clinicopathological significance of MFAP2 in GC or any other type of malignancy. As a result, the elucidation of the expression and function of MFAP2 in GC will promote the knowledge of the function of MFAP2 in the progression of cancer.

In the current study, we report that MFAP2 is overexpressed in human GC tissues and its overexpression correlates with tumor invasion depth, lymph node metastasis, distant metastasis, and tumor node metastasis (TNM) stage. Patients with GC with high MFAP2 expression are associated with poor overall survival (OS) and disease-free survival (DFS). Knockdown of MFAP2 significantly abrogated the proliferative, migratory, and invasive properties as well as epithelial–mesenchymal transition (EMT) phenotype of GC cells. Moreover, our results demonstrated that MFAP2 functions upstream of the TGF-β/SMAD2/3 signaling pathway as an activator. Our results suggest that MFAP2 may be considered as a promising prognostic marker and a potential therapeutic target for human GC.

Materials and methods

Tissue specimens from patients with GC

Paraffin-embedded sections of GC tissues and corresponding adjacent normal tissues were obtained from 168 patients who underwent surgical treatment in Gansu Provincial Hospital between August 2009 and December 2010. The samples collected from all the patients were clinically and histopathologically diagnosed with primary GC by the Department of Pathology, and the patients had not been treated with chemotherapy or radiotherapy prior to surgery. All cases were followed up for at least 5 years. All patients were treated with radical gastrectomy and D2 lymphadenectomy followed with postoperative chemotherapy. The cases were staged and graded according to the Cancer Staging Manual (Seventh Edition) of the American Joint Committee on Cancer.17 The normal tissues adjacent to the tumors were used as controls. This study was approved by the Ethics Committee of Gansu Provincial Hospital. Written informed consent was obtained from all study subjects. The clinical parameters of patients involved in the study are summarized in Table S1.

Cell lines and culture

Human GC cell lines BGC823, MKN-45, MGC803, SGC7901, and AGS were purchased from the Shanghai Institute of Biochemistry and Cell Biology, Chinese Academy of Sciences (Shanghai, China). All cells were maintained in Roswell Park Memorial Institute (RPMI)-1640 medium supplemented with 10% fetal bovine serum (FBS; Gibco, Grand Island, NY, USA), 100 µg/mL streptomycin, and 100 U/mL penicillin (Invitrogen, Carlsbad, CA, USA). Cells were cultured at 37°C in a humidified atmosphere with 5% CO2. For transforming growth factor-beta 1 (TGF-β1) (Sigma-Aldrich Co., St Louis, MO, USA) treatment, cells were cultured in RPMI-1640 medium supplemented with 10% FBS containing 5 ng/mL TGF-β1 for 24 h.

Immunohistochemistry (IHC)

For the immunohistochemical detection of MFAP2, 4 µm-thick sections from the formalin-fixed, paraffin-embedded tissues were used. The samples were deparaffinized in xylene and rehydrated through a graded series of ethanol washes. After the endogenous peroxidase was inhibited by exposure to 3% hydrogen peroxide for 10 min and the antigen was retrieved, the sections were incubated with primary antibody against MFAP2 (Sigma-Aldrich Co.; 1:100) at 4°C overnight. Sections were then incubated with horseradish peroxidase (HRP)-conjugated secondary antibodies (Dako, Glostrup, Denmark) according to the instructions of the Dako REAL EnVision Detection System. For the evaluation of MFAP2 staining, a staining score value was calculated as the intensity staining (0: negative, 1: weak, 2: moderate, 3: strong) multiplied by the percentage of positive tumor cells (1: 1%–25%; 2: 26%–50%; 3: 51%–75%; 4: >75%). An IHC score of 3 was determined as the optimal cutoff value for the 168 patients with GC as calculated by X-tile software version 3.6.1 (Yale University School of Medicine, New Haven, CT, USA)18 and SPSS 19.0 software (IBM Corporation, Armonk, NY, USA) (Figure S1). As a result, the final staining scores ≤3 were regarded as low expression and the scores ≥4 were regarded as high expression.

Bioinformatics analysis

The Gene Expression Omnibus (GEO) database (www.ncbi.nlm.nih.gov/geo) was explored to analyze the mRNA expression of MFAP2 and survival in patients with GC. The accession numbers of the GEO database involved in the study were GSE29272, GSE27342, and GSE54129. Expression data were log2 transformed and quantile normalized using the R software (version 3.2.5; R Foundation for Statistical Computing, Vienna, Austria) with packages of the BioConductor project (http://www.bioconductor.org/)19 as previously described. For The Cancer Genome Atlas (TCGA) database analysis, an online tool Gene Expression Profiling Interactive Analysis (GEPIA) (http://gepia.cancer-pku.cn/)20 was explored to compare the mRNA levels of MFAP2 in 408 GC tissues and 211 normal tissues (TCGA-STAD). In addition, the OS and DFS analyses of patients with GC in the TCGA-STAD database were conducted using the online tool GEPIA and results were downloaded from the website.

Another online tool KM plotter database (http://www.kmplot.com) was explored to conduct the survival analyses of patients with GC, which contains updated gene expression data and survival information from a total of 876 patients with GC.21 The association between OS/DFS and MFAP2 mRNA expression levels was calculated by allocating the patients to high- and low-expression groups, according to the best performing threshold. Processed data from the KM plotter were then downloaded and survival curves were plotted using GraphPad Prism 5.0 (GraphPad Software, La Jolla, CA, USA).

Lentivirus construction and transfection

The construction of short-hairpin RNA (shRNA) targeting MFAP2 (shMFAP2) and control shRNA (mock) were designed and synthesized by GeneChem (Shanghai, China; Table S2). Transfections were performed in accordance with the manufacturer’s protocol of Lipofectamine 2000 (Invitrogen) in BGC823 and MKN-45 GC cells due to their relatively high MFAP2 expression. Stably transfected cells were selected in media with 5 µg/mL puromycin (Sigma-Aldrich Co.) for 2 weeks, and the knockdown of MFAP2 in GC cells was confirmed by quantitative real-time polymerase chain reaction (qRT-PCR) and Western blotting analysis.

Western blotting analysis

Total proteins from cell samples were prepared using radioimmunoprecipitation lysis buffer (Beyotime Biotechnology Co., Shanghai, China) containing phenylmethanesulfonyl fluoride and phosphatase inhibitors. The protein lysates were quantified using a BCA (bicinchoninic acid) protein assay kit (Thermo Fisher Scientific, San Jose, CA, USA). Equal amounts of protein lysates were separated using a 10% sodium dodecyl sulfate-polyacrylamide electrophoresis gel and transferred onto polyvinylidene difluoride membranes (Millipore, Bedford, MA, USA) for 2 h. The membranes were blocked with tris-buffered saline with 0.05% Tween containing 5% non-fat milk for 2 h at room temperature and then incubated overnight at 4°C with targeted primary antibodies: anti-MFAP2 (1:1,000; Abcam, Cambridge, UK) anti-Vimentin (1:1,000; Cell Signaling Technology (CST), Beverly, MA, USA), anti-E-cadherin (1:1,000; CST), anti-Snail (1:1,000; CST), anti-SMAD3 (1:1,000; CST), anti-p-SMAD3 (1:1,000; CST), anti-SMAD2 (1:1,000; CST), anti-p-SMAD2 (1:1,000; CST), anti-TGF-β1 (1:1,500; Abcam), and anti-β-actin (1:1,000; CST). The membranes were incubated with HRP-conjugated secondary antibody (1:2,000; CST) at room temperature for 1 h. The bands were then visualized using an enhanced chemiluminescence kit (Bio-Rad, Hercules, CA, USA) and analyzed using Image Lab 4.0 (Bio-Rad) imaging software. β-actin was used as an internal control.

qRT-PCR assay

Total cellular RNA was extracted using TRIzol reagent (Invitrogen) according to the manufacturer’s instructions. Total RNA (500 ng) was reverse-transcribed into complementary DNA using the PrimeScript RT Master Mix (Takara Bio, Shiga, Japan). qRT-PCR was then performed in a 10 µL reaction volume using the SYBR PremixEx Taq™ (TaKaRa) with appropriate primers (Table S3) according to the manufacturer’s instructions. β-actin was used to normalize the relative expression levels of other target genes. The n-fold change in mRNA expression was analyzed according to 2−∆∆Ct method to calculate the relative expression of the targeted genes. Each experiment was performed in triplicates.

Colony-formation assay

Cells were suspended and seeded into six-well plates (500 cells/well) and incubated at 37°C in RPMI-1640 medium containing 10% FBS for 14 days. The medium was changed every 3 days. When most colonies contained more than 50 cells, the cells were fixed with 4% paraformaldehyde and stained using 0.1% crystal violet. Only colonies containing more than 50 cells were counted under light microscopy. Experiments were performed in triplicates.

Cell proliferation assay

Cell proliferation was examined using a Cell Counting Kit-8 (CCK-8; Dojindo, Kumamoto, Japan) according to the manufacturer’s instructions, and GC cells were seeded in 96-well plates at a density of 3×103 cells per well. After 1, 2, 3, and 4 days of culture at 37°C with 5% CO2, CCK-8 solution was added to each well and incubated for 1 h. The absorbance at 450 nm was measured with a plate reader. The experiments were performed in triplicates.

Wound healing assay

BGC823 and MKN-45 GC cells were seeded into six-well plates and cultured in RPMI-1640 containing 10% FBS until they reached 80% confluence and then serum starved for 12 h before wounding. Then wounds were made across the monolayer using a 200-µL plastic pipette tip. The remaining cells were washed twice with PBS buffer to remove cell debris and incubated at 37°C in RPMI-1640 without FBS. Subsequently, the wound closures were visualized at 0 and 24 h, and images were then captured by inverted microscopy (Olympus IX50; Olympus, Tokyo, Japan) to detect the migratory ability. The migration rate was determined by the percentage of the recovered gap distance at each time point compared with the gap distance at 0 h. The experiments were performed in triplicates.

Transwell invasion assay

BGC823 and MKN-45 cells were cultured in serum-free RPMI-1640 medium overnight and then suspended in serum-free RPMI-1640 medium. Cells (4×104/well) were seeded into the upper chamber of transwell inserts (pore size, 8 µm; Costar; Corning Incorporated, Corning, NY, USA) pre-coated with Matrigel in 24-well plates. The lower chambers were filled with 500 µL RPMI-1640 supplemented with 10% FBS as a chemoattractant. The cells were then treated with or without TGF-β1 (5 ng/mL) for 24 h. After incubation for 24 h, the cells invading into the lower chambers were fixed in 4% paraformaldehyde for 30 min and stained with 1% crystal violet for 10 min, and non-migrating cells remaining in the top surface of the insert were removed using cotton swabs. The invaded cells in the bottom chamber were counted and photographed under the inverted microscope. The mean number of cells was determined by selecting five random mid-power fields and the mean was calculated. All experiments were performed in triplicates.

Statistical analysis

Data were analyzed using the SPSS 19.0 software (SPSS Inc., Chicago, IL, USA). Comparisons between two groups were performed using Student’s t-test (two-sided). The relationship between MFAP2 expression and the clinicopathological characteristics of the patients with GC was evaluated using the chi-squared test. Ranked data were displayed by the rank sum test. In addition, OS and DFS were analyzed using the Kaplan–Meier method and the P-value was assessed by the log-rank test. A Cox proportional hazards regression analysis was used for univariate and multivariate analyses of prognostic values to investigate the differences in all possible death risk factors. Values were presented as mean ± standard deviation. P<0.05 was considered to indicate a statistically significant difference.

Results

MFAP2 expression is significantly upregulated in human GC tissues

The expression of MFAP2 protein was examined by immunohistochemical staining in 168 paired tumor tissues and corresponding normal (non-tumorous) tissues. As shown in Table 1 and Figure 1A, 56.55% (95/168) of GC specimens demonstrated high expression of MFAP2, which was significantly higher than that of normal tissues (39.88%, 67/168) (P=0.003, Table 1). We then analyzed the IHC score of MFAP2 in 168 paired samples. The IHC score of MFAP2 in GC tissues was significantly higher than that in the corresponding normal tissues (P<0.001, Figure 1B). To further investigate the mRNA expression of MFAP2 in larger cohorts of GC samples, we analyzed the mRNA level of MFAP2 in National Center for Biotechnology Information Gene Expression Omnibus (NCBI GEO) and TCGA-STAD databases. The datasets GSE29272 and GSE27342 included the expression data of 134 and 80 pairs of GC tissues and normal tissues, respectively. The results showed that the mRNA levels of MFAP2 were significantly upregulated in GSE29272 and GSE27342 (P<0.001, Figure 1C and D). In GSE54129, the MFAP2 mRNA levels were significantly higher in 111 GC samples compared with the 21 normal samples (P<0.001, Figure 1E). Moreover, in the TCGA-STAD database, MFAP2 expression was also overexpressed in 408 GC tissues compared with that in 211 normal tissues (P<0.05, Figure 1F). Collectively, these data suggested that the expression of MFAP2 was aberrantly increased in GC at both mRNA and protein levels.

Table 1

MFAP2 expression in gastric cancer and adjacent normal mucosa

Group	MFAP2 expression		χ2	P-value
	Low	High
Carcinoma	73	95	9.345	0.003
Adjacent normal mucosa	101	67

Abbreviation: MFAP2, microfibril-associated protein 2.

Figure 1

MFAP2 is highly expressed in gastric cancer (GC) tissues. (A) Immunohistochemical staining of MFAP2 protein expression in 168 gastric cancer tissues and their corresponding non-cancerous tissues; two representative cases are shown (original magnification: 200×; scale bar, 50 µm). (B) The IHC score of MFAP2 was significantly higher in gastric cancer tissues than in adjacent normal tissues. Analysis of GSE29272 database (C), GSE27342 database (D), and GSE54129 database (E) showing a significantly higher LTBP2 expression in GC tissues than in adjacent normal tissues. (F) Analysis of TCGA-STAD database by an online tool GEPIA indicating that MFAP2 expression was higher in gastric cancer tissues than in adjacent normal tissues; *P<0.05.

Abbreviations: IHC, immunohistochemistry; MFAP2, microfibril-associated protein 2; TCGA, The Cancer Genome Atlas.

High MFAP2 expression is associated with clinical features and predicts poor prognosis in patients with GC

In order to investigate the clinical role of MFAP2 in GC, we analyzed the relationship between MFAP2 expression and the clinicopathological factors of the 168 patients with GC. As shown in Table 2, MFAP2 expression was positively correlated with tumor (T) stage (P=0.019, Table 2), node (N) stage (P=0.031, Table 2), metastasis (M) stage (P=0.036, Table 2), and TNM stage (P=0.001, Table 2) but not with sex (P=0.609), age (P=0.106), tumor location (P=0.184), or histological differentiation (P=0.110) in patients with GC.

Table 2

The relationship between MFAP2 expression and clinicopathological features of patients with gastric cancer

Clinicopathological features	N	MFAP2 expression
		Low	High	χ2	P-value
Age (years)		73	95	2.612	0.106
≥60	74	27	47
<60	94	46	48
Gender				0.262	0.609
Male	114	48	66
Female	54	25	29
Histological grade				2.561	0.110
G1+G2	45	15	30
G3	123	58	65
T stage				5.548	0.019
T1–T2	55	31	24
T3–T4	113	42	71
N stage				4.661	0.031
N0	65	35	30
N1–N3	103	38	65
M stage				4.375	0.036
M0	149	69	80
M1	19	4	15
TNM stage				10.334	0.001
I/II	101	54	47
III/IV	67	19	48
Tumor site				1.766	0.184
Distal gastric	120	56	64
Proximal gastric	48	17	31

Abbreviations: M, metastasis; MFAP2, microfibril-associated protein 2; N, node; T, tumor; TNM, tumor node metastasis.

To further elucidate the importance of the role of MFAP2 in the survival of patients with GC, we examined the correlation between MFAP2 and OS and DFS of patients with GC. The results from our cohort of 168 patients with GC showed that higher expression level of MFAP2 was associated with a significantly lower OS (P=0.0007, Figure 2A). Univariate analysis showed that T stage (P<0.001), N stage (P<0.001), M stage (P=0.001), TNM stage (P<0.001), and high MFAP2 expression (P=0.001) correlated with the OS of patients with GC patients (Table 3). Then, we incorporated the significantly associated variables in the univariate log-rank test into a multivariate Cox regression analysis. The results of the multivariate analysis demonstrated that N stage (P=0.035) and MFAP2 (P=0.044, Table 3) were independent prognostic factors for OS. We then examined the correlation between the mRNA levels of MFAP2 and the survival rate in the NCBI GEO and TCGA-STAD databases by using online tools KM plotter (available at http://www.kmplot.com) and GEPIA (available at http://www.gepia.cancer-pku.cn). A total of 875 patients with GC were involved in the analysis of OS. The results demonstrated that patients with MFAP2high tumors had significantly lower OS (P<0.0001, Figure 2B) rate than those with MFAP2low tumors. In the TCGA-STAD database, we also found that high MFAP2 expression was correlated with markedly lower OS (P=0.0056, Figure 2C).

Figure 2

High MFAP2 expression is associated with poor overall survival and disease-free survival of patients with gastric cancer. (A) Kaplan–Meier estimation of 168 patients with gastric cancer indicating lower overall survival rate in patients with MFAP2high tumors. (B) Kaplan–Meier analysis of 875 patients with gastric cancer in KM plotter database indicating that patients with MFAP2high tumors were associated with a lower overall survival rate. (C) Kaplan–Meier survival curve for overall survival of 424 patients with gastric cancer from TCGA-STAD database generated using an online tool GEPIA. (D) Kaplan–Meier analysis of 168 patients with gastric cancer indicating a lower disease-free survival rate in MFAP2high patients. (E) Kaplan–Meier analysis of disease-free survival in 641 patients with gastric cancer generated using KM plotter database. (F) Kaplan–Meier analysis of disease-free survival in 424 patients with gastric cancer from the TCGA-STAD database generated using an online tool GEPIA.

Note: n(high) and n(low) refer to patient numbers.

Abbreviations: HR, hazard ratio; MFAP2, microfibril-associated protein 2; TCGA, The Cancer Genome Atlas; TPM, transcripts per million.

Table 3

Univariate and multivariate analyses of overall survival in patients with gastric cancer

Prognostic variables	Univariate analysis		Multivariate analysis
	HR (95% CI)	P-value	HR (95% CI)	P-value
Age (years)	1.275 (0.874–1.859)	0.207	–	–
Gender	0.932 (0.624–1.394)	0.733	–	–
Histological grade	1.061 (0.690–1.630)	0.788	–	–
T stage	2.425 (1.558–3.775)	0.000	1.536 (0.915–2.577)	0.104
N stage	2.523 (1.659–3.837)	0.000	1.717 (1.039–2.838)	0.035
M stage	2.521 (1.477–4.306)	0.001	1.481 (0.822–2.672)	0.191
TNM stage	2.745 (1.876–4.016)	0.000	1.326 (0.770–2.285)	0.309
Tumor site	1.031 (0.680–1.564)	0.885	–	–
High MFAP2 expression	1.943 (1.308–2.887)	0.001	1.528 (1.012–2.307)	0.044

Notes: The endashes indicate that only the significant variables in the univariate log-rank test were incorporated into a multivariate Cox regression analysis. If the P-value of a variable in univariate analysis was >0.05, then the variable was not calculated in the multivariate analysis.

Abbreviations: CI, confidence interval; HR, hazard ratio; M, metastasis; MFAP2, microfibril-associated protein 2; N, node; T, tumor; TNM, tumor node metastasis.

The results from our cohort of 168 patients with GC demonstrated that high MFAP2 expression was correlated with a significantly lower DFS (P=0.0004, Figure 2D). T stage (P<0.001), N stage (P<0.001), TNM stage (P<0.001), and high MFAP2 expression (P=0.001) were the significant univariate variables incorporated into multivariate analysis. The multivariate Cox regression analysis for DFS showed that T stage (P=0.007) and MFAP2 expression (P=0.028, Table 4) were independent prognostic factors. Moreover, a total of 641 patients were included into the analysis of DFS in the KM plotter database and the results demonstrated that patients with MFAP2high tumors had significantly lower DFS (P<0.0001, Figure 2E) rate than those with MFAP2low tumors. The results from TCGA-STAD database also revealed the same trend (P=0.0042, Figure 2F). Taken together, our results indicated that MFAP2 contributes to poor OS and DFS of patients with GC and might be used as a promising prognostic predictor.

Table 4

Univariate and multivariate analyses of disease-free survival in patients with gastric cancer

Prognostic variables	Univariate analysis		Multivariate analysis
	HR (95% CI)	P-value	HR (95% CI)	P-value
Age (years)	1.487 (0.971–2.278)	0.068	–	–
Gender	0.793 (0.497–1.265)	0.330	–	–
Histological grade	1.319 (0.798–2.179)	0.280	–	–
T stage	3.248 (1.952–5.404)	0.000	2.202 (1.238–3.917)	0.007
N stage	2.459 (1.560–3.874)	0.000	1.673 (0.972–2.879)	0.063
M stage	1.729 (0.830–3.602)	0.144	–	–
TNM stage	3.075 (1.967–4.807)	0.000	1.336 (0.748–2.387)	0.328
Tumor site	0.981 (0.615–1.566)	0.937	–	–
High MFAP2 expression	2.180 (1.397–3.404)	0.001	1.391 (1.059–2.702)	0.028

Notes: The endashes indicate that only the significant variables in the univariate log-rank test were incorporated into a multivariate Cox regression analysis. If the P-value of a variable in univariate analysis was >0.05, then the variable was not calculated in the multivariate analysis.

Abbreviations: CI, confidence interval; HR, hazard ratio; M, metastasis; MFAP2, microfibril-associated protein 2; N, node; T, tumor; TNM, tumor node metastasis.

MFAP2 promotes the proliferation of GC cells

To explore the function of MFAP2 in GC cells, we first detected the expression of MFAP2 in SGC7901, AGS, MGC803, BGC823, and MKN-45 by qRT-PCR and Western blotting analyses. Our results demonstrated that BGC823 and MKN-45 had the highest expression of MFAP2 (Figure S2A). As a result, these two GC cell lines were used in our further work. Knockdown of MFAP2 was verified by qRT-PCR and Western blotting analysis in BGC823 and MKN-45 (Figure S2B and C) GC cells.

First, we examined the functional roles of MFAP2 on the proliferation of GC cells. The colony formation assay was conducted, and the results demonstrated that knockdown of MFAP2 significantly decreased the colony number in both BGC823 and MKN-45 cells (P<0.001, Figure 3A and B). Moreover, the results of our CCK-8 assay indicated that the proliferation ability was significantly inhibited by knockdown of MFAP2 in BGC823 (P<0.01, Figure 3C) and MKN-45 (P<0.05, Figure 3C) cells. These results indicated that MFAP2 played an important role in promoting the proliferation of GC cells.

Figure 3

MFAP2 promotes the proliferation ability of gastric cancer cells. (A) Representative images of colony formation test in mock, shMFAP2-1, and shMFAP2-2 gastric cancer cells (BGC823 and MKN-45). (B) Quantification of the colony formation assay results showing decreased colony formation capability of gastric cancer cells with knockdown of MFAP2; ***P<0.001. (C) CCK-8 assays showing decreased proliferation ability of gastric cancer cells with knockdown of MFAP2. Mock: cells infected with mock lentivirus.

Abbreviations: CCK-8, Cell Counting Kit-8; MFAP2, microfibril-associated protein 2; shMFAP2-1, #1 short-hairpin RNA targeting MFAP2; shMFAP2-2, #2 short-hairpin RNA targeting MFAP2.

MFAP2 enhances the migratory and invasive abilities of GC cells by promoting TGF-β-induced EMT

We further explored the functional role of MFAP2 in regulating the migration and invasion of GC cells. The scratch-wound healing assay demonstrated that MFAP2-knockdown BGC823 and MKN-45 cells exhibited significantly decreased migratory capabilities compared with mock cells (Figure 4A and C). In addition, our transwell assay showed that knockdown of MFAP2 in BGC823 and MKN-45 cells significantly inhibited their invasive capabilities (Figure 4B and D). Because EMT has been elucidated as a crucial mechanism in GC progression, especially during tumor cell invasion and metastasis,22 we detected EMT markers in MFAP2-knockdown BGC823 and MKN-45 cells. E-cadherin, Vimentin, and Snail were selected as the EMT markers and were then detected by Western blotting and qRT-PCR analysis. The results demonstrated that the mRNA and protein levels of mesenchymal markers Snail and Vimentin were downregulated in MFAP2-knockdown GC cells, while the epithelial marker E-cadherin was significantly upregulated (Figure 4E and F). Interestingly, we observed a markedly significant downregulation of Snail compared with the downregulation of Vimentin at both mRNA and protein levels (Figure 4E and F).

Figure 4

MFAP2 promotes the migration, invasion, and epithelial–mesenchymal transition (EMT) phenotype in gastric cancer cells. (A) Representative images of the wound-healing assay in mock and MFAP2-knockdown BGC823 and MKN-45 cells. (B) Representative images of transwell assay in mock and MFAP2-knockdown BGC823 and MKN-45 cells (original magnifications: 200×). (C) Quantification of the wound-healing assay results; ***P<0.001. (D) Quantification of the transwell invasion assay results; ***P<0.001. (E) Western blotting analyses of the protein levels of EMT-related markers (E-cadherin, Snail, and Vimentin) and TGF-β1 in mock and MFAP2-knockdown BGC823 and MKN-45 cells. (F) qRT-PCR analyses of E-cadherin, Snail, and Vimentin in mock and MFAP2-knockdown BGC823 and MKN-45 cells; *P<0.05, **P<0.01, ***P<0.001. Mock: cells infected with mock lentivirus.

Abbreviations: MFAP2, microfibril-associated protein 2; qRT-PCR, quantitative real-time polymerase chain reaction; shMFAP2-1, #1 short-hairpin RNA targeting MFAP2; shMFAP2-2, #2 short-hairpin RNA targeting MFAP2; TGF-β, transforming growth factor beta.

Previous reports have suggested that MFAP2 inhibits latent transforming growth factor beta binding protein 1 (LTBP1) binding to fibrillin microfibrils, resulting in elevated TGF-β signaling.23 To investigate whether MFAP2 influences the expression of TGF-β in GC cells, Western blotting analysis was conducted to detect TGF-β expression in MFAP2-knockdown and mock cells. Interestingly, our Western blotting analysis found that the protein level of TGF-β1 was markedly decreased in MFAP2-knockdown BGC823 and MKN-45 cells (Figure 4E). Considering that TGF-β signaling pathway plays an important role in promoting EMT in GC cells, our data indicate that MFAP2 promotes the migration and invasion of GC cells by modulating TGF-β-induced EMT.

MFAP2 promotes EMT of GC cells by activating the TGF-β/SMAD2/3 signaling pathway

The key event in EMT is the switch of E-cadherin to N-cadherin, which renders the single cell more motile and invasive.24,25 TGF-β is a major inducer of EMT, and TGF-β-induced EMT has been suggested to be associated with the development and progression of GC.26 Considering that knockdown of MFAP2 decreased the expression of TGF-β1, we decided to explore whether the TGF-β/SMAD2/3 signaling pathway was involved in MFAP2-modulated EMT in GC cells. First, we investigated whether TGF-β1 treatment could restore the impaired invasive abilities due to knockdown of MFAP2. Results of our transwell assays showed that the impaired invasive abilities of MFAP2-knockdown GC cells were significantly restored after TGF-β1 treatment (Figure 5A). Then, we explored whether exogenous TGF-β1 treatment could change the expression of MFAP2 and EMT markers in MFAP2-knockdown BGC823 and MKN-45 cells. Interestingly, the Western blotting analysis showed that the expression of MFAP2 was not significantly changed after TGF-β1 treatment (Figure 5B). However, the expression of E-cadherin was significantly downregulated together with upregulation of Vimentin and Snail after TGF-β1 treatment (Figure 5B). The qRT-PCR also showed the same results (Figure 5C). Considering that MFAP2 knockdown decreased the expression of TGF-β1 with inhibition of EMT, and that exogenous TGF-β1 treatment restored the expression of EMT markers and invasive capabilities without changing the expression of MFAP2 in GC cells, our data suggested that MFAP2 might be an important upstream regulator of TGF-β1. Because SMAD2 and SMAD3 are critical downstream regulators in the TGF-β pathway,27 we further examined the expression of SMAD2 and SMAD3 and their phosphorylation levels. The results of Western blotting analysis showed that the total expression levels of SMAD2 and SMAD3 were not significantly changed in MFAP2-knockdown GC cells. However, the levels of phosphorylated-SMAD2 (p-SMAD2) and p-SMAD3 were all significantly decreased in MFAP2-knockdown BGC823 and MKN-45 cells (Figure 5D). Collectively, our data suggest that MFAP2 promotes the EMT of GC cells partially by activating the TGF-β/SMAD2/3 signaling pathway as an upstream regulator.

Figure 5

MFAP2 enhances epithelial–mesenchymal transition (EMT) of GC cells by activating the TGF-β/SMAD2/3 signaling pathway. (A) Representative images and quantification of transwell assay in mock and MFAP2-knockdown BGC823 and MKN-45 cells with or without TGF-β treatment (original magnifications: 200×); **P<0.01, ***P<0.001. (B) Western blotting analyses of the protein levels of EMT-related markers (E-cadherin, Snail, and Vimentin) and MFAP2 in MFAP2-knockdown BGC823 and MKN-45 cells with or without TGF-β treatment. (C) qRT-PCR analyses of EMT-related markers (E-cadherin, Snail, and Vimentin) and MFAP2 in MFAP2-knockdown BGC823 and MKN-45 cells with or without TGF-β treatment; **P<0.01. (D) Knockdown of MFAP2 attenuated phosphorylation of SMAD2 and SMAD3 without changing the total levels of the proteins as evaluated by Western blotting analysis. Mock: cells infected with mock lentivirus.

Abbreviations: GC, gastric cancer; MFAP2, microfibril-associated protein 2; qRT-PCR, quantitative real-time polymerase chain reaction; shMFAP2-1, #1 short-hairpin RNA targeting MFAP2; shMFAP2-2, #2 short-hairpin RNA targeting MFAP2; TGF-β, transforming growth factor beta.

Discussion

The tumorigenesis of GC is complex and is associated with multiple disorders including dysregulation of cancer-related signaling pathways.28,29 Among the known pathways, TGF-β signaling plays a central role in the development and progression of GC.30,31 There are three isoforms of TGF-β in mammalian cells, including the most abundant TGF-β1, and the uncommon isoforms: TGF-β2 and TGF-β3.32 The TGF-β/SMAD pathway is initiated by TGF-β1 binding to its type II receptor (TβRII). Then TβRII forms a heterodimer with TGF-β type I receptor (TβRI or ALK5), thereby activating the intracellular kinase domain of ALK5 to phosphorylate SMAD2/3 complex.33 The p-SMAD2/3 and p-SMAD4 form a transcription complex, which translocates to the nucleus and binds to SMAD binding elements in the promoter region. This complex then regulates the transcription of TGF-β target genes.34 Previous findings indicated that TGF-β signaling has both tumor-promoting and tumor-suppressive roles in different kinds of tumors.35,36 Activation of the TGF-β signaling pathway results in potent cell-cycle arrest in healthy cells and inhibits tumor formation in early stage tumor cells.37,38 However, the upregulation of TGF-β and overactivation of TGF-β receptor-initiated intracellular signaling to promote the progression in advanced-stage tumors have been observed in many cancers including GC.39,40 As a result, understanding of the regulation mechanism of TGF-β in GC is extremely important for understanding the role of TGF-β as well as the discovery of new drugs targeting TGF-β.

MFAP2 was found to regulate the function of TGF-β in several kinds of healthy cells by binding the active form of TGF-β instead of the latent form.41 In MFAP2-deficient C57BL/6 mice, loss of MAGP1 significantly increased the activity of TGF-β in epididymal white adipose tissue, as evaluated by SMAD-2 phosphorylation.41 In rat lung fibroblast line-6 (RFL-6) cells, which express no MFAP2 naturally, MFAP2 expression was induced by stable transfection with expression vectors and the expression of MFAP2 reduced SMAD-2 phosphorylation.42 Although the results of tissues or cell lines from rat and mice indicated that MFAP2 suppressed the activity of TGF-β, we believe that the regulation of TGF-β is very sophisticated and is significantly different in species and tissue types such as healthy and tumor tissues. Moreover, in contrast to proteins such as MFAP2 that had an inhibitory function for TGF-β activity, many other molecules are involved in releasing TGF-β from the extracellular matrix.43,44 A study of human dermal fibroblasts revealed that MFAP2 strongly inhibited the interaction between LTBP1 and fibrillin-1 because their binding affinities to fibrillin-1 were mutually exclusive. As a result, LTBP1 was released from assembled microfibrils, thereby strongly increasing the activity of TGF-β.23 Considering that paradoxical roles have been proposed for TGF-β in early stage and advanced-stage tumors, it is not surprising that different modulatory roles of MFAP2 in the regulation of TGF-β have been reported. As a result, the elucidation of the role of MFAP2 in regulating TGF-β in cancer will help us to understand TGF-β signaling in malignancy. Intriguingly, we found that knockdown of MFAP2 significantly decreased the expression of TGF-β and its activity, as evaluated by SMAD-2 and SMAD-3 phosphorylation, and that TGF-β treatment reversed the impaired migratory and invasive abilities in MFAP2-knockdown GC cells without changing the expression of MFAP2. Our results were comparable with the study in human dermal fibroblasts that MFAP2 promotes the activity of TGF-β. Additionally, we demonstrated that MFAP2 was significantly upregulated in GC tissues and its overexpression correlated with lower OS and DFS rates. Knockdown of MFAP2 impaired the migratory and invasive capabilities of GC cells. To the best of our knowledge, the current study of MFAP2 in GC is the first one to illustrate its expression, prognostic significance, and function as well as its role in the regulation of TGF-β in tumors.

Accumulating studies have indicated that EMT can be initiated and maintained by growth factors or inflammatory cytokines.45 Among these factors, TGF-β has been reported to effectively induce EMT in gastric cells, increasing their migratory and invasive properties and resulting in enhanced metastasis.45 It is known that EMT plays an important role in the invasion and metastasis of GC cells and is one of the earliest processes during tumor cell metastasis.46 To increase motility and invasiveness, cancer cells lose apical–basal polarity and cell adhesion, remodel the cytoskeleton, and develop a mesenchymal phenotype; then the cancer cells detach from one another and migrate to new locations.47 In addition, an EMT phenotype is often accompanied by the downregulation of epithelial markers E-cadherin and cytokeratin, and upregulation of mesenchymal markers such as N-cadherin, snail, Vimentin and fibronectin.48,49 In the current study, we found that knockdown of MFAP2 significantly decreased the expression of mesenchymal markers Vimentin and Snail, along with the upregulation of the epithelial marker E-cadherin. Moreover, we found that TGF-β treatment effectively induced the expression of Vimentin and Snail and suppressed the expression of E-cadherin in MFAP2-knockdown GC cells without changing the expression of MFAP2, indicating that MFAP2 might be an important upstream regulator of TGF-β/SMAD2/3 signaling.

In summary, our results demonstrate that the overexpression of MFAP2 predicts poor outcome in patients with GC. In addition, MFAP2 functions upstream of TGF-β and promotes the proliferation, migration, invasion, and EMT phenotype by activating the TGF-β/SMAD2/3 signaling. These data highlight the importance of MFAP2 in the progression of GC and revealed a novel MFAP2-mediated mechanism that may be used as a target for TGF-β.

Conclusion

Our results indicated that MFAP2 was significantly upregulated in human GC tissues and its overexpression was correlated with poor OS and DFS rates in patients with GC. MFAP2 promotes the proliferation, migration, invasion, and EMT phenotype in GC cells by activating the TGF-β/SMAD2/3 signaling pathway as an upstream regulator. Thus, MFAP2 might be a promising prognostic marker, and targeting the MFAP2/TGF-β/SMAD2/3 axis might be a potential therapeutic strategy for GC treatment.

The determination of the optimal cutoff value for 168 patients with gastric cancer. (A) The cutoff value of 4 as calculated by X-tile software. (B) The ROC curve generated using SPSS 19.0 software (the blue line shows sensitivity and the green line shows specificity). (C) The area under the curve was 0.626 as calculated by SPSS 19.0 software. aUnder the nonparametric assumption. bNull hypothesis: true area =0.5. (D) The Youden index was calculated by SPSS 19.0 software. The red box highlights that the best Youden index is 0.245 when the cutoff value is 3.5. As the maximum value of Youden index was 0.245 when the immunohistochemical score was 3.5, a score ≥4 was defined as MFAP2high and a score ≤3 as MFAP2low. The test result variable(s): MFAP2 has at least one tie between the positive actual state group and the negative actual state group. Statistics may be biased.

Abbreviations: CI, confidence interval; MFAP2, microfibril-associated protein 2; Pt no, patient number; ROC, receiver operating characteristic; sig, significance; std, standard.

The selection of gastric cancer cell lines and knockdown of MFAP2. (A) Five gastric cancer cells lines analyzed using Western blotting and qRT-PCR analyses indicating that BGC823 and MKN-45 had the highest expression of MFAP2. Knockdown efficiency of MFAP2 was confirmed by Western blotting and qRT-PCR analyses in BGC823 (B) and MKN-45 (C) cells.

Abbreviations: MFAP2, microfibril-associated protein 2; qRT-PCR, quantitative real-time polymerase chain reaction; shMFAP2, short-hairpin RNA targeting MFAP2; WT, wild type.

Table S1

Clinical parameters of GC patients involved in the study

Clinical characteristics	Number	Percentage
Age (years)
≥60	74	44.05
<60	94	55.95
Gender
Male	114	67.86
Female	54	32.14
Histology grade
G1	6	3.57
G2
G2	39	23.21
G3	123	73.21
T stage
T1	31	18.45
T2	24	14.29
T3	82	48.81
T4	31	18.45
N stage
N0	65	38.69
N1	48	28.57
N2	44	26.19
N3	11	6.55
M stage
M0	149	88.69
M1	19	11.31
TNM stage
I	42	25.00
II	59	35.12
III	48	28.57
IV	19	11.31
Tumor site
Proximal gastric	48	28.57
Distal gastric	120	71.43

Abbreviations: GC, gastric cancer; M, metastasis; N, node; T, tumor; TNM, tumor node metastasis.

Table S2

Sequences of MFAP2 knockdown and control shRNAs

Name	Sequence
shMFAP2-1	Forward: 5′-CACCGGAACAGTTCCAGTTCCAGTCCGAAGACTGGAACTGGAACTGTTCC-3′
	Reverse: 5′-AAAAGGAACAGTTCCAGTTCCAGTCTTCGGACTGGAACTGGAACTGTTCC-3′
shMFAP2-2	Forward: 5′-CACCGGCCTTGCAAACAGTGTCTCACGAATGAGACACTGTTTGCAAGGCC-3′
	Reverse: 5′-AAAAGGCCTTGCAAACAGTGTCTCATTCGTGAGACACTGTTTGCAAGGCC-3′
shMFAP2-3	Forward: 5′-CACCGCCGTGTGTACGTCATTAACACGAATGTTAATGACGTACACACGGC-3′
	Reverse: 5′-AAAAGCCGTGTGTACGTCATTAACATTCGTGTTAATGACGTACACACGGC-3′
Control	Forward: 5′-CCGGTTCTCCGAACGTGTCACGTTTCAAGAGAACGTGACACGTTCGGAGAATTTTTG-3′
	Reverse: 5′-AATTCAAAAATTCTCCGAACGTGTCACGTTCTCTTGAAACGTGACACGTTCGGAGAA-3′

Abbreviations: MFAP2, microfibril-associated protein 2; shMFAP2-1, #1 short-hairpin RNA targeting MFAP2; shMFAP2-2, #2 short-hairpin RNA targeting MFAP2; shMFAP2-3, #3 short-hairpin RNA targeting MFAP2; shRNAs, short-hairpin RNAs.

Table S3

Sequences of primers used for qRT-PCR in this study

Gene	Primer	Sequence
MFAP2	Forward	5′-CCCAAGCTTGTGAGGAACAGTACCCGT-3′
	Reverse	5′-CGGAATTCGATACTCCCCCAACCCGA-3′
E-cadherin	Forward	5′-CGAGAGCTACACGTTCACGG-3′
	Reverse	5′-GGGTGTCGAGGGAAAAATAGG-3′
Snail	Forward	5′-AGATGAGGACAGTGGGAAAGG-3′
	Reverse	5′-TGAAGTAGAGGAGAAGGACGAAGGAG-3′
Vimentin	Forward	5′-GACGCCATCAACACCGAGTT-3′
	Reverse	5′-CTTTGTCGTTGGTTAGCTGGT-3′
TGF-β	Forward	5′-CTTTGGTATCGTGGAAGGACTC-3′
	Reverse	5′-AGCTGTACCAGAAATACAGCAACA-3′
β-actin	Forward	5′-ACTGGAACGGTGAAGGTGAC-3′
	Reverse	5′-AGAGAAGTGGGGTGGCTTTT-3′

Abbreviations: MFAP2, microfibril-associated protein 2; qRT-PCR, quantitative real-time polymerase chain reaction; TGF-β, transforming growth factor beta.