MFGE8 is down‐regulated in cardiac fibrosis and attenuates endothelial‐mesenchymal transition through Smad2/3‐Snail signalling pathway

Abstract Endothelial‐mesenchymal transition (EndMT) is a major source of transformed cardiac fibroblasts, which is reported to play a key role in cardiac fibrosis (CF), a pathogenesis of cardiovascular diseases such as heart failure, myocardial infarction and atrial fibrillation. Nonetheless, the specific mechanism underlying the progression of EndMT to CF is still largely unknown. In this study, we aimed to investigate the role of milk fat globule‐EGF factor 8 (MFGE8), a kind of soluble glycoprotein, in TGF‐β1‐induced EndMT. In animal experiments, the expression of MFGE8 was found down‐regulated in the left ventricle and aorta of rats after transverse aortic constriction (TAC) compared with the sham group, especially in endothelial cells (ECs). In in vitro cultured ECs, silencing MFGE8 with small interfering RNA (siRNA) was found to promote the process of TGF‐β1‐induced EndMT, whereas administration of recombinant human MFGE8 (rh‐MFGE8) attenuated the process. Moreover, activated Smad2/3 signalling pathway after TGF‐β1 treatment and EndMT‐related transcription factors, such as Snail, Twist and Slug, was potentiated by MFGE8 knock‐down but inhibited by rh‐MFGE8. In conclusion, our experiments indicate that MFGE8 might play a protective role in TGF‐β1‐induced EndMT and might be a potential therapeutic target for cardiac fibrosis.

Milk fat globule-EGF-factor 8 (MFGE8) is a kind of lactadherin mainly secreted by mononuclear cells. This protein has been reported to be involved in various physiological functions and pathophysiological processes of the cardiovascular and cerebrovascular systems, including atherosclerosis, 12 cardiac hypertrophy, 13 angiogenesis, 14 myocardial infarction (MI) 15 and oxidative stress following subarachnoid haemorrhage. 16 In previous studies, knockout of Mfge8 in mice exacerbates pressure overload-induced cardiac hypertrophy and leads to remarkable inflammatory responses and an enormous reduce in survival after MI. 15 In addition, in recent years, a growing number of studies have emphasized the latent role of MFGE8 in organ fibrosis, such as blunting the degree of liver fibrosis in mice 17 and decreasing tissue fibrosis in mice model with lung fibrosis. 18 In our previous study, we found the protective role of MFGE8 in attenuating Ang-II-induced atrial fibrosis and atrial fibrillation through TGF-β1-Smad2/3 pathway. 19 However, how MFGE8 functions and the inner molecular mechanisms between MFGE8 and EndMT remain to be explored.
In the present study, we identified MFGE8 as a vital regulator of cardiac fibrosis because it was found to be down-regulated in rats with CF induced by transverse aortic constriction. In vitro knockout and administration experiments of ECs indicated that MFGE8 exerts protective action against transforming growth factor beta 1 (TGF-β1)-induced EndMT via the Smad2/3-Snail signalling pathway.
Data herein demonstrates the regulatory role of MFGE8 in EndMT and highlights the utility of MFGE8 as a diagnostic biomarker and a potential therapeutic target for CF. 8-week-old male Sprague Dawley wild-type rats (220-240 g) were used and randomly assigned into different groups (n = 15 in each group). Cardiac fibrosis was induced by transverse aortic constriction (TAC). Briefly, operations were conducted after the rats were anesthetized intraperitoneally using pentobarbital sodium (50 mg/ kg). After sternotomy, we tied a 27-gauge needle along with the thoracic aorta tightly with a 7-0 thread, then removed the needle and we could see a local narrow left in the aortic. Rats in the sham-operated group went through the same operation procedure without undergoing ligation. Left ventricle (LV) and aorta tissues were obtained by the end of the 4th week after operations.

| Histological analysis and immunofluorescence staining
LV and aorta tissues from the rats were fixed in 4% paraformaldehyde, embedded in paraffin and sliced into 4 µmol/L sections. Collagen of LV was stained by picrosirius red (PSR) to show the extent of interstitial fibrosis. Quantification of the fibrosis area was analysed using Image-Pro Plus 6.0. Immunofluorescence staining of LV and aorta sections were performed using antibodies against MFGE8 (Abclonal, A12322), CD31 (Abclonal, A0378) and Vimentin (Abclonal, A2584) with a dilution factor of 1:200 according to methods described previously. 20

Peprotech) was added to stimulate the cells for 48 hours to induce
EndMT, and recombinant human MFGE8 (rh-MFGE8, 500 ng/mL, R&D) resolved in sterile phosphate buffered solution (PBS) was used.

| RNA interference
Chemically synthesized small interfering RNAs (siRNAs) targeting human MFGE8 were purchased from Dharmacon (Lafayette, CO). According to the manufacturer's instructions, cultured cells were transiently transfected with 50 nmol/L siRNA using Lipofectamine™2000 (Life Technologies). The specific siRNA that provided the most efficient inhibition (resulted in a 75%-80% de-

| Western blot analysis
Proteins from the cultured cells and LV were extracted by RIPA solution (Beyotime, P0013B) with a mixture of protease and phosphatase inhibitor (Thermo Scientific, 78441), and the concentrations were quantified using BCA protein estimation kit (Thermo Scientific, 23225). (1:1000) was incubated for 1 hour at room temperature, and signals were visualized using ECL reagents on Tanon 5200. Immunoblots were quantified by densitometry using ImageJ software.

| Cellular immunofluorescence
Cells cultured on cover slips were fixed in 4% paraformaldehyde for 15 minutes after stimulation. Following permeabilization with 0.1% Triton X-100 for 20 minutes and blockage for 40 minutes, the cells were incubated overnight with primary antibodies against CD31 (Abclonal, A0378) and/or α-SMA (Abclonal, A7248) at 4°C. Secondary antibody conjugated with Alexa Fluor 488 and/or Alexa Fluor 594, and DAPI were used for visualization.

| Scratch assay
HUVECs were seeded in 6-well plates at a density of 5.0 × 105 cells/ mL. When cells reached the confluency of 90%, cell monolayer was scratched using a sterile 200 μL pipette tip and gently washed twice with PBS. Different marks were drawn at the bottom of the wells to ensure that the same visual fields were assayed. Images were taken at 0, 6, 12 and 24 hours post-scratching and Image-Pro Plus 6.0 was used to measure the distance of migration.

| Transwell migration assay
Samples containing 1 × 105 cells were resuspended in serum-free medium and added into the upper layer of Transwell Chambers (8 μm pore size, Corning, 3413). After incubated for 24 hours with complete medium in the lower layer, the chambers were washed, fixed and stained with crystal violet. Numbers of cells in five randomly selected fields were counted to assess cell migration.

| Angiogenesis assay
Matrigel (Corning, 354248) was added into 96-well plates and allowed to facilitate solidification at 37°C for 1 hour. Then, suspension of 15 000 cells with different stimuli was gently pipetted into each well and incubated for 18 hours in a standard environment. Tube formation was observed by microscopy.

| Statistical analysis
All data were presented as means ± SEM and analysed by SPSS 19.0 statistical software and GraphPad Prism 6.0. Statistical analysis was performed using two-tailed Student's t test and one-way ANOVA.
Differences were considered statistically significant at P < .05.

| MFGE8 expression is decreased in endothelial cells of rats with cardiac fibrosis
Previous studies indicate that MFGE8 is under-express in cardiovascular and cerebrovascular diseases. To expound the role of MFGE8 in ing demonstrated cardiac fibrosis in TAC-treated rats ( Figure 1A). In order to investigate whether EndMT occurred in rats with cardiac fibrosis, immunofluorescence staining was used to detect the expression of endothelial cell markers CD31, mesenchymal cell markers Vimentin and MFGE8. As is shown, CD31 and MFGE8 were highly expressed in vascular endothelial cells of left ventricle and aorta in the sham-operated group. However, in TAC group, the fluorescence intensity of CD31 and MFGE8 in ECs decreased, while the fluorescence intensity of Vimentin increased, suggesting that EndMT occurred in rats with cardiac fibrosis. In both LV and aorta, MFGE8 is prominently colocated with CD31 in ECs, suggesting that MFGE8 may be involved in EndMT ( Figure 1B,C). Western blot experiment further confirmed the decreased MFGE8 expression in ventricular tissues of rats ( Figure 1D).

TA B L E 1 Primers sequences used for qRT-PCR
These data indicated that vascular endothelial cells of rats with cardiac fibrosis undergo endothelial-mesenchymal transformation, and MFGE8 might have a hand in this pathological process.

| TGF-β1 induced endothelial-mesenchymal transition in ECs cultured in vitro
To establish an EndMT cell model in vitro, we stimulated cultured In the presence of TGFβ1, the level of CD31 mRNA was greatly reduced while ACTA2 mRNA levels were increased ( Figure 2B,E).
Western blot analysis provided further evidence that ECs experiencing EndMT lost the molecular marker CD31 and began to express molecular markers of mesenchymal or myofibroblast, such as α-SMA ( Figure 2C,F). The double-immunofluorescence images of HMECs provided evidence not only for changes in shapes, but also for the weakened CD31 signal ( Figure 2D). All these results indicated that ECs experienced EndMT under the stimulation of TGF-β1.

| Knock-down of MFGE8 with siRNA promoted EndMT in HMECs
As expected, the mRNA and protein levels of MFGE8 was downregulated approximately to 50% in HMECs after TGF-β1 stimulation ( Figure 3A,B). Next, to explore the role of MFGE8 in TGF-β1-induced EndMT, three MFGE8-specific siRNAs were transfected into HMECs for 24 hours to down-regulate MFGE8, and their knock-down efficiency was corroborated by qRT-PCR ( Figure 3C) and Western blot ( Figure 3D) post-transfection. As we found, si-(2) had the most significant knock-down efficiency, which could reduce the expression of MFGE8 mRNA and protein by 70%-80%, and was used in the subsequent experiments. At mRNA ( Figure 3E) and protein levels ( Figure 3G,H), MFGE8 knock-down decreased the ECs marker CD31 and in contrast increased the TGF-β1-induced expression of the mesenchymal markers Vimentin, ACTA2 (α-SMA) and FSP1. Same results of changes in markers were also observed by double-immunofluorescence. In addition, from the immunofluorescence images, it could be seen that MFGE8 knock-down HMECs have a tendency to transform into spindle-like fibroblasts ( Figure 3F). Experiments above demonstrated that MFGE8 silencing could promote EndMT process induced by TGF-β1 in HMECs.

| Knock-down of MFGE8 by siRNA exacerbates EndMT in HUVECs
To further confirm the effect of MFGE8 on EndMT in different ECs, we analysed its expression at protein levels in HUVECs and found a 40% decrease in TGF-β1-treated group ( Figure 4A). Then, si-MFGE8 was transfected into HUVECs. Compared with si-NC, si-MFGE8 remarkably strengthened the effect of TGF-β1 in inducing EndMT for the decrease of CD31 and increase of α-SMA in both mRNA ( Figure 4B) and protein levels ( Figure 4D). Moreover, immunofluorescence images provided intuitive and similar results ( Figure 4C).
Generally, the data clarified that knock-down of MFGE8 gave impetus to EndMT induced by TFG-β1 in HUVECs.

| MFGE8 silence inhibits angiogenesis and promotes cell migration in HUVECs
Subsequently, we conducted angiogenesis assay to evaluate the influence of MFGE8 on angiogenesis, which is reported to be a characteristic of partial EndMT. Our data revealed that si-MFGE8 transfection led to a great decrease of angiogenesis ability compared with the group with TFG-β1 treatment alone ( Figure 5A).
Furthermore, results of the scratching assay ( Figure 5B) and transwell assay ( Figure 5C) proved the enhanced endothelial cell migration caused by MFGE8 silence in TFG-β1 group. These experiments confirmed that silence of MFGE8 further promoted the development of EndMT in the pathological environment induced by TGF-β1.

| rh-MFGE8 reverses EndMT induced by TGF-β1 in HUVECs
In order to investigate whether exogenous MFGE8 could play a protective role in EndMT, we introduced recombinant human MFGE8 (rh-MFGE8) into HUVECs. As is shown in Figure 6 in HUVECs cultured with rh-MFGE8 was observed by dual-immunofluorescence staining ( Figure 6B). The results highlighted that endogenous rh-MFGE8 could attenuate the process of TGF-β1induced EndMT in vitro.

| MFGE8 regulates EndMT induced by TGF-β1 through Smad2/3 signalling pathway and influences the expression of EndMTrelated transcription factors
Our next aim was to elucidate the underlying mechanism of MFGE8-mediated regulation of TGF-β1-induced EndMT in vitro.
Previous studies have confirmed that transcription factors Snail, Twist and Slug are closely related to EndMT. 22  These results were also confirmed in HMECs ( Figure S1). Taken together, these observations indicated that MFGE8 regulated

| D ISCUSS I ON
Cardiac fibrosis is a critical contributor to the progression of heart failure. It is due to excessive deposition of extracellular matrix caused by activation and accumulation of cardiac fibroblasts and results in a shrinkage of microvascular system and destruction of normal myocardial structure. 25 During development, fibroblasts derived from ECs undergo epithelial-mesenchymal transition (EMT), 26 while in pressure overload-induced cardiac lesions, they are mainly derive from pathologically induced EndMT. 10 Studies have verified that a variety of environmental factors including hypoxia, radiation, high blood glucose levels, inflammation and TGF-β can trigger EndMT. [27][28][29][30][31] In our previous study, high glucose could induce EndMT in HUVECs through Smad2/3 pathway. 32 Compelling evidence suggests that in cardiomyocytes, systolic dysfunction, and myocardial fibrosis was observed. 13 In our previous work, MFGE8 suppressed atrial fibrosis via TGF-β1/Smad2/3 pathway in an integrinβ3-dependent way and attenuated the vulnerability to atrial fibrillation. 19 Accordingly, in the present study, immunofluorescence images showed significantly decreased expression of MFGE8 in Vimentinlabelled fibrosis region in the rat model of TAC-induced cardiac fibrosis compared with that of the control group ( Figure 1B).
Moreover, both in the LV and aorta tissues, vascular endothelial cells showed decreased fluorescence intensity of CD31 and increased fluorescence intensity of Vimentin, indicating that ECs underwent EndMT after TAC ( Figure 1B,C). Therefore, we hypothesized that MFGE8 might participate in the occurrence and  Figure S1). In addition, MFGE8 acts as a momentous bridging molecule that binds to integrin receptors, such as α v β 5 on phagocytes and phosphatidylserine on dead cells. Nakaya

CO N FLI C T O F I NTE R E S T
The authors have read and approved the final manuscript for submission and declare no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data supporting the findings of this study are included in the article.