Methylseleninic Acid Provided at Nutritional Selenium Levels Inhibits Angiogenesis by Down-regulating Integrin β3 Signaling

Targeting angiogenesis has emerged as a promising strategy for cancer treatment. Methylseleninic acid (MSA) is a metabolite of selenium (Se) in animal cells that exhibits anti-oxidative and anti-cancer activities at levels exceeding Se nutritional requirements. However, it remains unclear whether MSA exerts its effects on cancer prevention by influencing angiogenesis within Se nutritional levels. Herein, we demonstrate that MSA inhibited angiogenesis at 2 µM, which falls in the range of moderate Se nutritional status. We found that MSA treatments at 2 µM increased cell adherence, while inhibiting cell migration and tube formation of HUVECs in vitro. Moreover, MSA effectively inhibited the sprouts of mouse aortic rings and neoangiogenesis in chick embryo chorioallantoic membrane. We also found that MSA down-regulated integrin β3 at the levels of mRNA and protein, and disrupted clustering of integrin β3 on the cell surface. Additionally, results showed that MSA inhibited the phosphorylation of AKT, IκBα, and NFκB. Overall, our results suggest that exogenous MSA inhibited angiogenesis at nutritional Se levels not only by down-regulating the expression of integrin β3 but also by disorganizing the clustering of integrin β3, which further inhibited the phosphorylation involving AKT, IκBα, NFκB. These findings provide novel mechanistic insight into the function of MSA for regulating angiogenesis and suggest that MSA could be a potential candidate or adjuvant for anti-tumor therapy in clinical settings.

Angiogenesis is a process of forming new blood vessels from existing vasculature that is strictly regulated by multiple factors and is vital to growth and development [1][2][3] . However, mounting evidence suggest that imbalance of angiogenesis triggers the progressions of many diseases including malignant tumors 4,5 , supporting investigations into angiogenesis inhibitors for cancer treatment. Angiogenesis initially involves cell proliferation and migration 4 , which is induced by secretion of growth factors such as vascular endothelial growth factor (VEGF) 2 . VEGF exerts its effects by activating multiple cellular events within the vascular endothelium including cell proliferation, migration, and tube formation 6,7 . Additionally, angiogenesis is also regulated by integrin family members that modulate cell polarity, junction, proliferation, migration and signal transduction [8][9][10] . The integrin family has been found to have more than two dozen unique heterodimers formed by the combination of eighteen α and eight β subunits 8 . Among these integrin subunits, integrin β3 has been found to be highly expressed in endothelial cells. During angiogenesis, conformational changes and clustering states of integrin β3 initiate intracellular signal transduction through the phosphorylation cascade of FAK, AKT, IKKα, IκBα, and NF-κB 11,12 . Thus, integrin β3 signaling regulates the downstream expression of cytokines (TNFα, IL-1β, IL-6), adhesion molecules (VCAM-1, ICAM-1, E-selectin) and enzymes (iNOS, COX-2) during angiogenesis [13][14][15][16] . Moreover, integrin β3 has a concentrated cell surface distribution at the leading edge in motile cells that regulates cell polarity and directs migration 17,18 .
Selenium (Se) is an essential micronutrient for mammals that exhibits antioxidant and anti-cancer potential for human health. Se metabolites and compounds include methylseleninic acid (MSA), selenodiglutathione (SDG), and hydrogen selenide (H 2 Se), which have been reported to exhibit chemopreventive effects against different types of cancer 19 . However, the narrow window between their effective and toxic dosages limit its use and highlight the importance of determining effective dosages and forms of Se metabolites and compounds 20,21 . The recommended dietary allowance (RDA) of Se supplementation in the U.S. is 55 μg/day for adults 22 . Many publications have reported the Se nutritional status in whole blood ranging from 0.71 to 3.24 μM of subjects with moderate Se status. There is good correlation between blood Se concentration and the activity of the selenoenzyme, glutathione peroxidase (GPx), in the plasma and most cells. Therefore, the Se content of physiological fluids may be considered a reliable index of status with respect to nutritional Se 23 .
Recent studies have focused on organic forms of dietary Se, which show higher bioactivity but lower cytotoxicity. MSA is an organic species of Se that is metabolized from animal cells 20 . It exhibits anti-oxidative and anti-cancer activity at level above moderate Se nutritional concentrations. Unique features of MSA that all contribute to its anti-cancer activities include its capacity to induce ROS accumulation, to inhibit the phosphorylation of AKT, ERK and p53, and to facilitate proteolytic activation of caspase-8 and -3 20, 24-26 .Moreover, metabolism of MSA generates the precursor of methylselenol, which is an active Se form and reported to induce the apoptosis of cancer cells 27 . However, little information is known about the effects of MSA in angiogenesis within the physiological range of nutritional Se intake, and it remains unclear whether MSA influences angiogenesis related to integrin β3 signaling.
For the current study, we utilized human umbilical vein endothelial cells (HUVECs) as an in vitro model and focused on molecular mechanisms by which MSA inhibited angiogenesis. Our results showed that exogenous MSA provided nutritional Se levels effectively inhibited cell migration, tube formation, and neoangiogenesis by down-regulating integrin β3 and interrupting the clustering of integrin β3. Additionally, MSA inhibited the phosphorylation of AKT, IκBα, and NF-κB, thereby highlighting the anti-angiogenic capacity of MSA.

Results
Uptake and metabolism of MSA by HUVECs. In order to monitor the dynamic uptake and metabolism of MSA when provided at nutritional Se concentrations, HUVECs were treated with 2 μM MSA for 0 to 48 h. Se concentrations in cell lysates and culture medium were analyzed by Atomic Fluorescence Spectroscopy (AFS). As shown in Fig. 1A, Se accumulated in HUVECs with this range of MSA exposure, and reached maximum levels of 63.75 ng/10 6 cells at a longer exposure time (12 h). There was a trend towards slightly decreased but still maintained at a higher level in intracellular Se concentration within 48 h exposure. In contrast to the changes of intracellular Se, the concentration of Se in medium was relatively stable over the first 6 h, and then it gradually decreased following the uptake by HUVECs for 48 h incubation. To quantify the endogenous Se levels, total Se and free Se (separated by ultrafiltration with 1 kDa cut-off) were detected in FBS, medium, cultured supernatant and HUVECs, in which total Se were 73.4 ± 9.6 ng/ml, 14.9 ± 2.0 ng/ml, 8.9 ± 1.6 ng/ml and 16. 8 ± 1.4 ng/10 6 cells including free Se were 23.8 ± 4.9 ng/ml, 7.8 ± 0.8 ng/ml, 3.8 ± 2.6 ng/ml and 3.7 ± 0.7 ng/10 6 cells, respectively ( Supplementary Fig. S1A,B).
The presence of minor methyl-Se metabolites at low levels (10 μg/kg Se as SeMet or SeMC) has been reported for the lymphoma cells exposed to MSA using speciation analysis of cell lysates by reversed-phase HPLC-APEX-Q-ICPMS 28 . HPLC-ICP-MS chromatograms of 1:1 diluted lysates of HUVECs exposed 2 μM MSA for 0 to 24 h are shown in Fig. 1B. The results indicated that the Se metabolites with retention time similar to that of SeMC (tR = 4 min), SeMet (tR = 12.5 min) and DMSe (tR = 34 min) were the major Se species in HUVEC after exposure to MSA. The separated retention curve of MSA standard (200 μM) was shown in Fig. 1B, in which the retention time of MSA was approximately 2.5 min. It should be noticed that the peak of MSA (2 μM) was failed to be detected in the present experiment conditions. Meanwhile, the endogenous MSA also could not be identified and detected in FBS, medium and/or cell cultures (data not shown). Together, results showed that HUVECs could uptake and transformed exogenous MSA to other Se species within 6 h, during which cellular Se reached to the maximum level and maintained such high level within 48 hours.
Effects of MSA on cell proliferation of HUVECs. First, we examined the effects of different Se compounds on cell proliferation in HUVECs. Four different Se compounds (selenocysteine, selenomethionine, sodium selenite, and MSA) at doses of 0-10 µM were tested. Results showed no effect on cell proliferation at low doses (1-2 µM), while the cells treated with MSA consistently exhibited the highest levels of cell proliferation compared with other Se compounds at higher dosage ( Fig. 2A). Even so, MSA still had an inhibitory effect on cell proliferation to HUVECs at 10 µM (82.1%). Moreover, we also found that 2 μM MSA had no significant effect on VEGF-induced cell proliferation (Fig. 3A).
To further analyze the effect of MSA on apoptosis at 2 µM, Rhodamine-123 staining and PI staining were employed individually. Compared with the positive control (cisplatin, 80 µM), HUVECs exposed to 2 µM MSA showed a much higher mitochondrial membrane potential (17.14, Fig. 2B). In addition, even though no apoptotic cells were detected (Fig. 2C), exposure to 2 µM MSA could still induce a minor G1/S arrest (49.3%, Fig. 2D).
These findings were also supported by analyses of Bip or phosphorylated His on protein levels, which represent markers of protein or DNA damage during apoptosis. As shown in Fig. 2E, relatively minor changes of Bip (0.36) and p-His (0.22) were observed in the MSA-treated groups (2 µM) versus positive control (Fig. 2E). Together, these results suggest that treating with 2 µM MSA would not induce apoptosis in HUVECs.

MSA effectively increases the adherence to collagen I and inhibits cell migration of HUVECs.
Adherence and migration are crucially regulated events for tube formation in angiogenesis. Thus, we firstly examined the adhesion of HUVECs to rat-tail collagen I using an adhesion assay. As shown in Fig. 3B, MSA-treated group showed increased adherent cells (118.53%) compared with untreated controls. Next, we performed wound healing assays to examine the effects of MSA on cell migration. As shown in Fig. 3C, cells in MSA-treated group (78.28%) showed a lower migration rate compared with untreated controls, and MSA could also inhibit SCientifiC REPORTS | 7: 9445 | DOI:10.1038/s41598-017-09568-5 VEGF-induced migration (90.97% versus 139.12%). In addition, no obvious changes of cell adhesion and migration on HUVECs were detected with treatments of the endogenous Se ( Supplementary Fig. S2A,B). Taken together, our results demonstrate that MSA treatment increases the adherence of HUVECs to collagen I and reduces the HUVEC migration.

MSA down-regulates Integrin β3 and inhibits phosphorylation of AKT. Because integrins play an
important role in cell adhesion, migration and signal transduction, we explored the potential involvement of integrins in MSA exposure. The mRNA levels of integrin subunits including αv, α1, α5, β1, β3 and β5 are expressed at relatively higher level in HUVECs, which were analyzed via qPCR as candidate targets of MSA. Results showed that MSA treatments only caused a significant downregulation of integrin β3 within 12 hours, while no significant changes of α1, α5 and β5 were observed (Fig. 4A). Western blot was further employed to evaluate the effect of MSA on αv, β1, and β3 on protein levels. Down-regulation of integrin β3 was observed in a time-and dose-dependent manner ( Fig. 4B and C), while no significant changes of αv and β1 were observed (Fig. 4B). As a critical downstream mediator of integrins, AKT and its phosphorylation were also evaluated. Results showed that MSA also inhibited AKT phosphorylation in a time-and dose-dependent manner ( Fig. 4B and C).
MSA disrupts the clustering of integrin β3 surface localization. Integrin β3 shows a clustering and polar distribution in motile cells. To determine whether MSA would interfere this effect, immunofluorescence was employed. As shown in Fig. 5, although polar distribution could not be observed, fluorescent spots (arrowheads) were fully distributed in cells exposed to VEGF. However, the fluorescence intensity and spots were significantly decreased in cells exposed to MSA compared with untreated control group, as well as in cells co-treated with MSA versus VEGF-treated group. Altogether, these data indicated that MSA treatment not only down-regulates the expression of integrin β3 but also disrupts its clustering on membrane, which may further influence the migration of HUVECs during angiogenesis.

MSA inhibits VEGF-induced angiogenesis in vitro, ex vivo and in vivo.
Although angiogenesis is a complex process, tube formation is one of the crucial steps. As MSA could inhibit cell migration and down-regulate integrin β3, we further evaluated the effect of MSA on angiogenesis in vitro through tube formation assays. As shown in Fig. 6A and D, fewer tubular structures and elongated cells were observed in MSA-treated group (40.93%) compared with vehicle control. Although HUVECs became elongated and formed obvious tubular structures in the presence of VEGF (233.34%), these effects could be inhibited by co-treating with MSA (107.04%).
To further evaluate the anti-angiogenic effect of MSA, ex vivo aortic ring assays as well as in vivo CAM assays were performed. As shown in Fig. 6B,C and E, no microvessel sprouts were observed around aortic rings exposed to MSA compared with vehicle control, but the microvessel sprouts (arrowheads) induced by VEGF were obviously observed, which could be inhibited by co-treating with MSA. Furthermore, blood vessels density (47.81%) was significantly reduced on CAMs exposed to MSA compared with PBS-treated group ( Fig. 6F and G). Total Se accumulation in CAMs reached to 173.17 ng/g after 24 h exposure to MSA at 2 µM, and a ~9% decrease within 3 days indicated a relatively high Se status in the treated CAMs. In addition, the endogenous Se could not induced obvious effects on angiogenesis (Supplementary Figs S2 and S3) under the present experimental conditions. Taken together, these results suggested that exogenous MSA supplementation was capable of suppressing angiogenesis in vitro, ex vivo, and in vivo.

MSA inhibits the phosphorylation of IκBα and NF-κB, and the nuclear translocation of NF-κB.
The phosphorylation and nuclear translocation of NF-κB are downstream steps resulting from integrin activation and AKT phosphorylation. Hence, western blot was further employed to investigate the effect of MSA on NF-κB. MSA not only inhibited the phosphorylation of IκBα (25.1%) and NF-κB (8.29%) within a short time,  Fig. 3 were determined by one way ANOVA comparison test and the error bars represented the SD (n = 3). Different letters indicate statistically significant differences between groups (p < 0.05).
but also down-regulated the nuclear localization of NF-κB (37.39%, Fig. 7A). Immunofluorescence was used to further evaluate the effect of MSA on the nuclear translocation of NF-κB. MSA treatment led to relatively lower fluorescence intensity (60.93%) and no nuclear NF-κB were observed (Fig. 7B). Additionally, we performed qPCR to measure the mRNA levels of ICAM-1 and VCAM-1, which are both target genes of NF-κB. After exposure to MSA, the mRNA levels of ICAM-1 (3.00%) and VCAM-1 (4.75%) were down-regulated within a short period (Fig. 7C). To evidence the relationship between integrin β3 downregulation induced by MSA and the sequential inhibition of AKT/NF-κB phosphorylation and angiogenesis, pcDNA3.1-beta-3 29 or β3-siRNA was transfected into HUVECs to induce overexpression or knockdown of integrin β3, and treated with or without MSA in the meanwhile. As shown in the Supplementary Fig. S4, all of the groups treated with MSA displayed obvious decreases of integrin β3 and resulted in the subsequent changes of AKT and NF-κB phosphorylation, whereas pcDNA3.1-beta-3 induced transient high expression of integrin β3 and rescued those effects to some extent. In consistent with the integrin β3 signaling, promotion of cells adhesion, suppression of cells migration and tubes  α1, α5, β1, β3, β5) in HUVECs exposed to 2 μM MSA for different hours (0, 3, 6, 12 h) were quantified by qPCR assay. A 2 −ΔΔCT relative quantification method was employed and β-actin was used as housekeeping gene. The error bars represented the SD (n = 9) and were determined by one way ANOVA comparison test. *p < 0.05, **p < 0.01. Western blot was employed to analyze the protein levels of integrin αv, β1, β3 and AKT phosphorylation in (B) a time-dependent (2 μM, 0, 3, 6, 12, 24 h) or (C) a dose dependent (0, 1, 2 μM, 24 h) manner of MSA, and β-actin was served as loading control. The relative levels of protein were quantified by ImageJ software. formation were shown in the Supplementary Fig. S5, in which also showed a compensation by overexpression of integrin β3. Together, these results preliminarily elucidated the mechanism of inhibition angiogenesis induced by MSA through downregulation of integrin β3 signaling and AKT and NF-κB phosphorylation.

Discussion
A major goal of cancer biologists over the past few decades has been to develop anti-tumor drugs that limit angiogenesis and thereby reduce oxygen and nutrient supplies required for tumor growth 1, 2 . Given the potential side effects and high costs associated with these compounds, interest has emerged in focusing on natural compounds because of their biocompatibility. As a metabolite of Se from animal cells, MSA offers potential due to its widely reported anti-tumor effects at super nutritional levels 20,24,26,30 . However, little information is available about MSA regarding anti-angiogenic effects and mechanisms of its actions. In this study, we present evidence that MSA at nutritional Se levels could inhibit VEGF-induced tube formation in vitro, and neo-angiogenesis ex vivo and in vivo without inducing apoptosis. Moreover, MSA could inhibit angiogenesis by both down-regulating the expression and disrupting the clustering of integrin β3, which further inhibit the phosphorylation of AKT, IκBα, and NF-κB, as well as the nuclear translocation of NF-κB.
Generally, the narrow margin between effective and toxic doses of Se as well as its metabolites like MSA, makes it difficult to identify optimal dosages lacking cytotoxicity to normal cells. According to a previous study 23 , Se concentrations in blood ranging from 0.71 to 3.24 μM are considered in the nutritional range. As shown in our results, addition of 2 µM MSA increased the Se concentration in media by 153.03 ng/mL (1.94 μM), which falls within nutritional Se concentrations. Meanwhile, the endogenous Se levels including total Se and free Se in FBS, medium, cultured supernatant and HUVECs were around 1 μM or less, and showed no obvious changes of angiogenesis, which indicated that their effects on angiogenesis might not be dominate in our experiment conditions. Moreover, HPLC-ICP-MS results showed that SeMC, SeMet, and DMSe seemed to be the major Se species in HUVECs exposing to MSA. However, the peak of MSA was failed to be detected in our experiment conditions, Figure 5. MSA disrupts the distribution of integrin β3 on cell membranes. HUVECs were treated with MSA (2 μM) and/or VEGF (50 ng/mL) for 12 h before fixation. Integrin β3 was detected using mouse monoclonal antibody against integrin β3 followed by Cy3-labelled goat anti-mouse-IgG (red) antibodies. Nuclei were counterstained with DAPI (blue). Photos were taken by confocal laser scanning microscope (FV10i).
it was probably because the dosage of MSA we used were relatively lower and MSA could be converted to other species rapidly even after a short time exposure according to the previous work 31 .
Moreover, an MTT assay showed that doses above 5 µM had obvious inhibitory effects on cell proliferation, while the cytotoxicity of MSA was relatively lower compared with other Se compounds. Together, with the results  . (B,C and E) The numbers of sprouts from aortic rings were counted. The error bars (D-E) represented the SD (D: n = 3; E: n = 60) and were determined by one-way ANOVA comparison test. Different letters indicate statistically significant differences between groups (p < 0.05). The blood vessels density (F,G) on CAMs were quantified by ImageJ software and the changes of Se concentrations (H) in CAMS treated with/without 2 μM MSA were detected by AFS. The error bars represented the SD (n = 6, ***p < 0.001) and were determined by nonparametric tests (G) or two way ANOVA comparison test (H).
of Rodamine-123 staining, PI staining, and western blot, these results suggested that MSA within nutritional Se concentrations had no effect on cell proliferation or induce apoptosis even in the presence of VEGF, but still could inhibit angiogenesis in vitro, ex vivo and in vivo.
Although the ex vivo and in vivo experiments beyond 48 h, culture medium supplemented with MSA or VEGF were changed every other days during two weeks in aortic ring assay and the CAM assay was just performed within 72 h. Moreover, considering not only the inhibition effects of MSA on cell migration and tube formation which are crucial for initiation of angiogenesis during the early 12-24 h, but also the downregulation effects of MSA on the protein level of integrin β3 and phosphorylated AKT within 24 h, thus supplementary MSA could maintain its activity during the whole procedure ex vivo or in vivo which was consistent with the dynamic alteration of Se concentration in cultures. (C) Relative mRNA expressions of NF-κB downstream moleculars (ICAM-1 and VCAM-1) were quantified by qPCR assay after MSA treatment (2 μM, 0, 3, 6, 12 h). A 2 −ΔΔCT relative quantification method was employed and β-actin was used as housekeeping gene. The data in Fig. 7 were determined by one way ANOVA comparison test and the error bars represented the SD (n = 3). Different letters indicate statistically significant differences between groups (p < 0.05).
The family of integrins consists of eighteen α and eight β subunits, which form more than two dozen unique integrin heterodimers that are widely expressed in human tissues including endothelial cells 8 . There are considerable data suggesting that integrin β3 is prominently expressed in endothelial cells and cancer cells 32 . Integrin β3 is a crucial connective molecular for recognizing proteins containing Arg-Gly-Asp (RGD) sequence 33 , which relates to its regulation of cell attachment, migration, and outside-in pathways during angiogenesis. Our qPCR results showed that MSA could down-regulate integrin αv, β1, and β3 at the level of mRNA. However, at the level of protein, only integrin β3 was down-regulated in a time-and dose-dependent manner. Moreover, it is reported that motile cells exhibit functional polarity with a large lamellipodia at the leading edge 34 , where integrin β3 exhibit a polar surface distribution 9, 10 and participate in forming complexes with RPTPα for rigidity sensing 35 . Our data showed that integrin β3 clustered at the surface of endothelial cells exposed to VEGF, and this process could be disrupted by MSA. Together, these findings suggested that MSA could inhibit cell migration by down-regulating and disrupting the clustering of integrin β3.
After activation, integrins transduce outside-in signal by phosphorylation of FAK, which is a key initiator of focal adhesion complex formation. The phosphorylation cascade of FAK, PI3K, AKT, IKKα, and IκBα is a crucial pathway downstream of integrin activation. Phosphorylation and nuclear translocation of NFκB are mechanisms by which several genes are actively transcribed [36][37][38] . Our results show that MSA inhibits the phosphorylation of AKT and IκBα, which subsequently inhibits the phosphorylation and nuclear localization of NFκB. The mRNA expression of NFκB responsive genes, like ICAM-1 and VCAM-1, were also found to be down-regulated.
In addition, to clarify the relationship between integrin β3 downregulation induced by MSA and the sequential inhibition on AKT/NF-κB phosphorylation and angiogenesis, overexpression or knockdown of integrin β3 were employed individually. As shown in our results, downregulation of integrin β3 could be observed in cells exposed to MSA. However, transfection with pcDNA3.1-beta-3 could obviously induced transient upexpression of integrin β3, which could rescued the downregulation of integrin β3 induced by MSA as well as the subsequent inhibitory effects of MSA on AKT and NF-κB phosphorylation or angiogenesis in some degree. Together, these results suggested that integrin β3 downregulation by MSA could sequentially inhibit AKT/NF-κB phosphorylation and angiogenesis.
Overall, our results suggest that MSA within nutritional Se concentrations effectively inhibited angiogenesis in vitro, ex vivo and in vivo by down-regulating integrin 3β levels and disrupting its clustering. MSA treatment led to an inhibition of the phosphorylation of AKT and IκBα, which further inhibited the phosphorylation and nuclear translocation of NFκB. Our results demonstrated that MSA may represent a potent therapeutic agent for anti-cancer by inhibiting tumor angiogenesis in clinical applications.

Materials and Methods
Cell lines, cell culture, and reagents. MSA (CH 3 SeO 2 H) was purchased from Sigma. A 2 mM stock solution of MSA was prepared in PBS, stored at −20 °C, and then diluted to final required concentrations in cell culture medium.

ICP-MS.
HUVECs (2 × 10 5 per well) were seeded in 6-well plates and treated with 2 µM MSA for different hours (0, 0.5, 6, 12, 24 h). Different Se standards including MSA, DL-selenomethionine (Se-Met), Se-L-methylselenocysteine (Se-MC), and DMSe were used as positive controls, cells treated without MSA were served as negative control. Cell lysates were digested by using the MARS-5 microwave system (CEM). Quantification of specific Se species in cell lysates were performed by HPLC for chromatographic separations on an Agilent Zorbax Rx-C8 column coupled directly to collision reaction cell ICP-MS (Agilent 7500ce, Palo Alto) according to our previous method 39 . Calibration was performed by the standard addition technique using peak area measurements of the chromatographic signals by monitoring the isotopes 78 Se and 77 Se. For PI staining, treated cells were collected and fixed with pre-cooled 70% ethanol at −20 °C overnight. After washing with PBS and 1% BSA in PBS, cells were stained with 10 μg/mL PI (Sigma) overnight at 4 °C in dark. After staining, cells were detected by flow cytometer (Beckman Coulter, Gallios).

Cell adhesion and Migration assays.
Adhesion assays were performed as described by Chen and colleagues 40 . In brief, 96-well plates were pre-coated with 1 mg/ml Collagen I (Coring) at 37 °C for 1 h. After treating with 2 µM MSA or 50 ng/ml VEGF (Peprotech) for 24 h, cells (2 × 10 4 per well) were reseeded in 96-well plates. Cells were allowed to adhere for 20 min, then non-adherent cells were washed away. Adherent cells were cultured for another 4 h and quantified following a MTT assay protocol described above.
For migration analyses, HUVECs (2 × 10 5 per well) were seeded in 6-well plates and serum starved. Cells were wounded by 200 µL pipette tips, washed, and incubated with 50 ng/mL VEGF or 2 µM MSA. Images were taken after 24 h. The migration distances were quantified by using the software Adobe Photoshop CS6.
Assays measuring Tube formation, Aortic rings, and CAM. To measure tube formation, HUVECs (2 × 10 4 per well) were seeded in 48-well plates pre-coated with Matrigel (Corning), starved, and treated with 50 ng/ml VEGF or 2 µM MSA. Images were taken after 6 h. The tube lengths were quantified by using the software Adobe Photoshop CS6.
Aortic ring assays were performed as previously described with minor modifications 41 . Aortas were isolated from 6-week-old C57BL/6 mice, cleaned off the fatty tissue, and cut into 1-mm-long rings. After washing, the rings were serum starved with Opti-MEM (Life Technology, Gibco) for 18 h. Then, one aortic ring was embedded per well of a 96-well plate with 1 mg/mL collagen at 37 °C for 1 h, and then 150 µL Opti-MEM medium supplemented with 2% FBS and 50 ng/ml VEGF or 2 µM MSA was added. Medium were refreshed every other day and images were taken after 14 d.
For CAM assays, groups of 6 fertilized chicken eggs were incubated at 37.5 °C in an incubator with 60% humidity. After incubating for 7 d, eggs were punched, injected with 200 µL PBS or 2 µM MSA, and incubated for another 3 d. The CAMs were then imaged. A quantitative analysis of the vascular network was performed as previously described with some modification 42 . In brief, images were converted into gray-scale and analyzed using software ImageJ. The algorithm input parameters were initially set to obtain the identification of pixels related to the blood vessels as strong and positive as to the background and weak positive and tuned to minimize non-specific pixel recognition as strong positive. The ratio between the number of strong positive pixels and the sum of weak, background and strong positive pixels is the morphometric value used to quantified the blood vessels density across the experimental groups. All the analysis were performed on images with equal area. The CAMs were collected, weighed and digested as described above and Se concentrations were detected by AFS.

Real-time PCR analysis and Western blotting.
HUVECs (1.6 × 10 5 per well) were treated with 2 µM MSA for different periods (0-12 h). Total RNA was isolated with RNAPrep pure Cell/Bacterial Kit (TianGen), and 1 µg aliquots of the isolated total RNA were reverse transcribed with FASTQuant RT Kit (TianGen). The mRNA levels of integrin subunits αv, α1,α5, β1, β3, β5 were further quantified by using TransStart Tip Green qPCR Supermix (TransGen Biotech) and Bio-Rad CFX96 system. Levels of β-actin mRNA were used as housekeeping transcript. The amounts of target gene were estimated by 2 −ΔΔCT relative quantification method. The primers for qPCR were provided in supplementary information.
RIPA (Life Technology, Invitrogen) containing protease inhibitor cocktail (Sigma) was used to lyse cell samples. The insoluble fraction was removed by centrifugation at 10000 × g for 10 min at 4 °C. The concentration of total protein in supernatant was measured by BCA assay (Pierce, Life Technology). After adjusting to equal concentration with loading buffer (Bio-Rad), protein samples were boiled, separated by SDS-PAGE, and transferred to a PVDF membrane (0.22 µm, Millipore). Membranes were blotted by 5% nonfat milk (BD) in TBST for 1 h. The membranes were then incubated with primary antibodies (Cell Signaling, 1:1000) at 4 °C overnight, and HRP-conjugated secondary antibody (Cell signaling, 1:3000) for 1 h at room temperature. Protein bands were visualized on X-ray film (Kodak) by using ECL Assay Kits (Pierce, Life Technology). The relative levels of protein were quantified according to the reference band of β-actin. Statistical analysis. Experiments were carried out at least in triplicate. The data were presented as mean ± SD. Statistical analysis was performed using SPSS 20.0 and statistical comparisons between groups were performed by different analysis which described in figure legend. P value of <0.01 or <0.05 was considered statistically significant, while different letters indicate statistically significant differences between groups (p < 0.05).