Maspin Regulates Endothelial Cell Adhesion and Migration through an Integrin Signaling Pathway

Abstract

Maspin has been identified as a potent angiogenesis inhibitor. However, the molecular mechanism responsible for its anti-angiogenic property is unclear. In this study, we examined the effect of maspin on endothelial cell (EC) adhesion and migration in a cell culture system. We found that maspin was expressed in blood vessels ECs and human umbilical vein endothelial cells (HUVECs). Maspin significantly enhanced HUVEC cell adhesion to various matrix proteins. This effect was dependent on the activation of integrin β₁, which subsequently led to distribution pattern changes of vinculin and F-actin. These results indicated that maspin affects cell adhesion and cytoskeleton reorganization through an integrin signal transduction pathway. Analysis of HUVECs following maspin treatment revealed increased integrin-linked kinase activities and phosphorylated FAK levels, consistent with increased cell adhesion. Interestingly, when HUVECs were induced to migrate by migration stimulatory factor bFGF, active Rac1 and cdc42 small GTPase levels were decreased dramatically at 30 min following maspin treatment. Using phosphorylated FAK at Tyr³⁹⁷ as an indicator of focal adhesion disassembly, maspin-treated HUVECs had elevated FAK phosphorylation compared with the mock treated control. The results were a reduction in focal adhesion disassembly and the retardation in EC migration. This study uncovers a mechanism by which maspin exerts its effect on EC adhesion and migration through an integrin signal transduction pathway.

Introduction

Angiogenesis is required for tumor growth and metastasis. Blood vessels in solid tumor are composed of endothelial and tumor cells. These mosaic vessels allow for the shedding of tumor cells into the vasculature and transport of tumor cells to the second site through a circulation system. Identifying angiogenic growth factors and inhibitors and characterizing their contribution to endothelial cell migration and proliferation will not only help us to gain insight of the biology of angiogenesis network, but also provide potential candidates for developing more efficient cancer therapeutic strategies targeting angiogenesis.

Our laboratory identified maspin as a potent angiogenesis inhibitor (1). Its anti-angiogenic property was later confirmed by Cher et al. (2) in a model of prostate cancer. Maspin is a member of serine protease inhibitor (Serpin) with tumor suppressing function. It is a 42-kDa protein produced by many cell types from mammary gland, prostate, skin, and cornea (3, 4). Structurally, maspin belongs to the family of serine protease inhibitors with homology to plasminogen activator inhibitor 1 and 2 (PAI-1 and PAI-2),² and ovalbumin. However, unlike PAI-1 and PAI-2, maspin does not directly inhibit serine proteases (5, 6). As a tumor suppressor, maspin functions to inhibit tumor cell migration and invasion, and induces tumor cell apoptosis (7–10).

For many years, it has not been clear how maspin controls angiogenesis, especially how maspin regulates endothelial cell (EC) motility. ECs can be activated by exogenous stimuli such as bFGF and VEGF and produce matrix metalloproteinases to degrade matrix surrounding the ECs. ECs could escape from the vessel walls and invade surrounding tissue and also proliferate to form solid sprouts connecting neighboring vessels. Cell adhesion and migration govern the escape of ECs from the original vessel structures. Many molecules produced by ECs or surrounding tissue regulate this process. The integrin family plays an important role in the EC adhesion process. Thus, it is not surprising that molecules affecting the interaction between integrins and EC matrix will likely regulate angiogenesis and EC function. In fact, several proteins that regulate integrin-mediated cell adhesion emerge as key modulators of vascular functions (6, 11, 12).

This study identifies a new molecular mechanism by which maspin controls angiogenesis. Previously, our laboratory showed that maspin interacts with integrin β₁ in mammary epithelial cells (10). In this study, we have provided direct evidence that maspin inhibits angiogenesis through controlling EC cell adhesion, migration, and adhesion-mediated cell signaling pathway. In particular, we have shown that maspin increases endothelia cell adhesion to fibronectin (FN), laminin, collagen, and vitronectin, which in turn activates integrin β₁, integrin-linked kinase (ILK), and the FAK signal transduction pathway. Subsequent changes in focal adhesion and cytoskeleton reorganization result in the attachment and spreading of ECs on the matrix. Furthermore, we discovered that maspin also blocked EC cell migration by disrupting focal adhesion disassembly. This report identifies maspin as one of the key molecules that play a key role in EC cell adhesion and migration.

EXPERIMENTAL PROCEDURES

Cell Culture and Reagents

Human umbilical vein endothelial cells (HUVEC) were cultured in endothelial basic medium, and harvested when cells reached 70–80% confluent. Anti-Rac1 and anti-ILK antibodies and myelin basic protein (MBP) protein were obtained from Upstate, Inc., vinculin antibody and collagen were from Sigma, anti-integrin α_vβ₃ (LM609) and anti-integrin β₁ (AIIB2) were a kind gift from Dr. Karl S. Matlin. Vitronectin was obtained from Chemicon. Anti-active integrin β₁ was from Pharmingen and anti-cdc42 from Santa Cruz Biotechnology. Laminin-1 and fibronectin were purchased from Invitrogen. Other cell culture reagents were from Sigma unless otherwise noted. GST-Maspin is mouse maspin fused to GST as described previously (7). Preparations of GST-Maspin and GST followed the procedure as described in Ref. 7. Recombinant proteins were pre-treated with polymyxin B beads and dialyzed against PBS before being used in cell culture.

Immunocytochemistry

Mouse embryonic yolk sac samples at E9.5 were fixed and sectioned to 5-μm slides. Anti-maspin antibody (ABS4A, 0.25 μg/μl, 1:250) was used for immunostaining as described before (13). Maspin immunostaining in HUVECs was done using the same ABS4A antibody at the same concentration with a Texas Red-conjugated secondary anti-rabbit antibody. Cells were counterstained with DAPI. For vinculin and integrin β₁ immunostaining, HUVECs were first serum starved for 4 h and detached by 0.05% trypsin EDTA. After incubating with 0.5 μm GST-Maspin or GST protein in serum-free endothelial basic medium for 30 min, cells were seeded to FN-coated coverslips for 30 min with or without bFGF. Then cells were fixed in 4% paraformaldehyde and permeabilized with 0.5% Triton X-100. For the nocodazole-treated group, serum-starved HUVECs were first incubated with 10 nm nocodazole for 4 h and then treated with 0.5 μm GST-Maspin or GST, and fixed with 4% paraformaldehyde at 0, 15, 30, 60, and 120 min. Fixed cells were incubated with antibodies against vinculin (1 μg/μl, 1:200) or active integrin β₁ (0.5 μg/μl, 1:100) at 4 °C overnight, followed by incubation with the appropriate second antibodies labeled with FITC or Texas Red. Nuclei were stained by DAPI. Images were acquired with a ×100 objective using a Zeiss fluorescence microscope.

Immunoprecipitation and Western Blot Analysis

HUVEC cells were serum-starved for 4 h, detached with 0.05% trypsin, and then incubated with GST-Maspin or GST for the indicated time at 37 °C with 5% CO₂. After lysing cells with RIPA buffer (0.05 m Tris-HCl, pH 7.6, 0.15 m NaCl, 1% Nonidet P-40, 0.5% deoxycholate, 2 mm EDTA and protease inhibitor mixture), cell extracts were subjected to immunoprecipitation with an anti-PY20 antibody and separated by an 8% SDS-PAGE gel, followed by transferring to immunoblot the PVDF membrane (Bio-Red). Western blot analysis was performed using anti-phosphorylated FAK antibody and blots were visualized with chemiluminescence (Pierce Biotechnology, Inc.). For the quantitation of active Rac1 and cdc42, starved HUVECs were preincubated with bFGF (20 ng/ml) prior to GST-Maspin or GST treatment. Cell extracts were immunoprecipitated with PAK1 and immunoblotted with anti-Rac1 or anti-cdc42 antibodies. An anti-Tyr(P)³⁹⁷-FAK antibody was used to detect phosphorylated FAK at Tyr³⁹⁷ site. An anti-FAK antibody against total FAK was purchased from Pharmingen and used at 1:200 dilution (0.2 μg/μl). Maspin Western immunoblot analysis was done using a mouse maspin polyclonal antibody as described previously (10).

Adhesion Assay

Nunclon^TM 96-well flat-bottom polyvinyl chloride plates (Nunc^TM, Roskilde, Denmark) were covered with FN, laminin, collagen, or vitronectin at 15 μg/ml or the indicated concentrations, and then blocked with 1% BSA for 30 min. Serum-starved HUVEC cells were detached with 2 mm EDTA and re-suspended in serum-free endothelial basic medium at a density of 5 × 10⁴ cells/ml. Cells were then incubated with GST-Maspin or GST at 0.5 μm concentration at 37 °C for 30 min. In the antibody-blocking assay, cells were first incubated with anti-maspin polyclonal rabbit antibody (0.5 μg/μl, 1:200) or anti-β1 antibody (AIIB2, 0.2 μg/μl, 1:80) for 20 min before GST-Maspin or GST treatment, and then added to the matrix-coated plates, followed by incubation at 37 °C for 0.5 h. After washing the plate with PBS to remove unattached cells, the adhered HUVEC cells were fixed with 4% paraformaldehyde for 30 min and then stained with 0.2% crystal violet in 10% ethanol. The violet stain was solubilized with 1:1 (v:v) 0.1 m NaH₂PO₄ (pH 4.5) and ethanol, and absorbance was measured at 540 nm with a microplate reader.

ILK Assay

ILK kinase assay was performed using a rabbit immunoaffinity purified ILK antibody (Upstate Biotechnology, Lake Placid, NY). MBP served as the substrate and could be phosphorylated by active ILK. HUVECs were first incubated with GST-Maspin or GST for 1 h and then lysed in RIPA buffer. Cell extracts (500 μg) were immunoprecipitated by protein A-agarose beads, which were conjugated with ILK antibody. After washing the beads once with lysis buffer and twice with kinase buffer (50 mm HEPES, pH 7.0, 10 mm MgCl₂, 5 mm MgCl₂, and 1 mm DTT), beads were re-suspended in 30 μl of kinase buffer containing the MBP substrate (2 μg), 10 mm ATP, and 10 mCi of [γ-³²P]ATP. The reaction mixture was incubated at 30 °C for 30 min with occasional mixing and separated by electrophoresis in a 10% PAGE gel. Phosphorylated proteins were quantified after exposure to autoradiographic film.

Maspin Knockdown by RNA Interference

Selected maspin siRNAs (229 and 455) were cloned into pSUPER.retro.puro vector (Oligo Engine) as previously described (10) and transfected into PT67 packaging cells. Supernatants containing virus were added to HUVEC cells for 2 days in the presence of 2 μg/ml of Polybrene. Empty pSUPER.retro and siRNA negative control (sequence 1; Ambion) were used as negative controls. After infection for 2 days, cells were selected with 1 μg/ml of puromycin. Maspin knockdown efficiency was evaluated by Western blot analysis using a maspin polyclonal antibody as described previously (10). Pooled cells of siRNA stable clones were used in the adhesion assays.

Cell Motility Assay

Cell motility assay were performed with a HitKit (Cellomics, Inc.) assay as described (53). Briefly, blue fluorescent beads were added to wells of a 96-well plate pre-coated with FN (1.5 μg/well). After incubating the plate in the dark at 37 °C for 60 min, the extra beads were removed with PBS buffer. HUVECs were pre-treated with 0.5 μm GST-Maspin or GST, respectively, with or without bFGF, and plated to the wells of a 96-well plate at a density of 1.0 × 10⁴ cells/ml. After incubation for 2, 7, and 18 h at 37 °C with 5% CO₂, cells were fixed with 5.5% formaldehyde for 60 min and stained with a rhodamine-phalloidin. Images from five randomly selected areas were taken for each sample under a Zeiss fluorescent microscope and the track areas were quantitatively analyzed using a NIH Image J software.

Focal Adhesion Disassembly Assay

Focal adhesion disassembly assay was performed as described (14). Briefly, nocodazole was applied to activate Rho GTPase and stimulate the formation of focal adhesion and stress fibers in the cells. After nocodazole treatment, cells were washed with PBS and focal adhesion started to disassemble. Starved HUVECs were incubated with 10 nm nocodazole for 4 h. After washing out the drug with PBS, cells were incubated with 0.5 μm GST-Maspin or GST, and then fixed with 4% paraformaldehyde at 0-, 15-, 30-, 60-, and 120-min time points, followed by permeabilization with 0.5% Triton X-100. For Western blot analysis, starved cells were first treated with 10 nm nocodazole for 4 h and incubated with 0.5 μm GST-Maspin or GST for the indicated times as described above. Cells were lysed with RIPA buffer and extracts were separated in 10% SDS-PAGE gel to determine the level of Tyr(P)³⁹⁷-FAK.

RESULTS

Maspin Is Expressed in Endothelial Cells and Acts to Increase Endothelial Cell Adhesion

Maspin is known to be expressed in the epithelial cells of mammary and prostate glands (15). Consistently, we also detected expression of maspin in embryonic endothelial cells and stromal cells in blood islands by immunohistochemistry (Fig. 1A). In cultured HUVECs, positive maspin signals were observed in the cytoplasm in a punctate staining pattern (Fig. 1C). Additionally, Western blotting analysis revealed maspin expression in HUVECs (Fig. 1E).

FIGURE 1.

Maspin expression in endothelial cells and effect on HUVEC cell adhesion. Maspin was detected in endothelial cells and stromal cells (arrow) in E9.5 mouse embryonic blood islands (A) and cultured HUVEC cells (C). B, negative control for A. Section was stained with secondary antibody (minus primary anti-maspin ABS4A). D, negative control for C. HUVECs were stained with secondary antibody (minus ABS4A) and counterstained with DAPI. Scale bar, 10 μm. E, Western immunoblot showing expression of maspin in MCF-10A cells. F, concentration-dependent enhancement of GST-Maspin on HUVEC cells adhesion to FN. G, GST-Maspin significantly enhanced HUVEC cell adhesion on FN, laminin (LN), collagen IV (Col), and vitronectin (VN). *, p < 0.05; **, p < 0.01. H, GST-Maspin significantly increases adhesion of both wild type HUVECs and maspin knockdown HUVEC-siRNA maspin cells. *, p < 0.05; **, p < 0.01. Error bars are standard deviations from triplicates. I, Western immunoblot analysis of maspin expression in wild type HUVECs, stable HUVECs transfectants with control silencing vector, and stable HUVECs transfectants with maspin silencing vector (HUVEC-siRNA maspin). Actin serves as loading control.

Secreted maspin may act on the surface of ECs to regulate cell adhesion, in a manner similar to mammary epithelial-derived maspin that acts on the cell surface to modulate adhesion to the laminin 1-rich matrix (2, 10). To test this possibility, we seeded HUVECs in wells pre-coated with FN and treated the cells with different concentrations of recombination GST-Maspin or GST. Maspin treatment enhanced cell adhesion on FN in a concentration-dependent manner. The enhanced cell adhesion peaked with 0.5 μm GST-Maspin treatment (Fig. 1F).

To examine the effect of maspin on HUVEC adhesion on different matrices, HUVEC cells were seeded in wells pre-coated with FN (15 μg/ml), laminin (2 μg/ml), collagen I (15 μg/ml), and vitronectin (1 μg/ml) and treated with 0.5 μm GST-Maspin or GST protein. Maspin significantly enhanced HUVEC cell adhesion to these matrices with varied effects (Fig. 1G). To determine whether the enhanced cell adhesion is caused by the interaction of exogenous GST-Maspin on cell membrane not the endogenous maspin, HUVEC-derived maspin was knocked down by RNA interference. Western blot analysis showed that maspin was knocked down efficiently (80% reduction) in HUVECs (Fig. 1I), and these HUVEC-siRNA maspin cells displayed significant reduction in cell adhesion (90% of wild type HUVECs). However, when HUVEC-siRNA maspin cells were treated with GST-Maspin they responded with enhanced cell adhesion, similar to the HUVEC-siRNA control and wild type HUVECs (Fig. 1H).

Previous reports showed that different regions of maspin protein participated in the regulation of cell migration, depending on the cell types and matrix contexts (16, 17). To determine whether the reactive site loop (RSL) or other regions of maspin are involved in HUVEC cell adhesion, several maspin mutants were tested by adhesion assay (Fig. 2A). A deletion mutant Mp (1–139 amino acids (aa)) abolished the effect of maspin on HUVEC cell adhesion. Mp (140–375 amino acids) retained most of the enhancing effect of maspin on adhesion (Fig. 2B), suggesting the 140–375-aa region may play the important role in cell adhesion. Mp (1–225 aa) slightly enhanced HUVEC adhesion (Fig. 2B). Previous studies showed that the maspin RSL domain plays a role in regulating breast cancer cell migration (17). Arginine 340, located in the RSL motif in maspin, is critical for its interaction with the urokinase-type plasminogen activator and involved in regulating cell migration (18). To test the role of this region in cell adhesion, we generated maspin mutants with deletion of the C terminus containing the RSL domain and a mutant with point mutation at 340 aa (Arg to Ala) in the RSL region. The cell adhesion assay showed that both mutants could enhance HUVEC cell adhesion, but at a reduced level compared with wild type maspin (Fig. 2, A and B). Using a polyclone anti-maspin antibody (raised against the full-length of maspin protein) we found that cell adhesions on both FN and laminin were significantly blocked by this antibody (Fig. 2, C and D). These findings confirm the specific enhancement of cell adhesion by maspin.

FIGURE 2.

Effect of maspin mutants on HUVEC cell adhesion. A, schematic illustration of maspin mutants. B, adhesion assay with wild-type GST-Maspin and various maspin mutants. Cell adhesion for GST-treated HUVECs was set as a reference control (100%). Note maspin mutant (Mp 1–139) failed to enhance HUVEC cell adhesion. The mutant with the N-terminal deletion (Mp 140–375) as well as point mutation in the RSL domain (Mp* 1–375) could significantly enhance HUVEC cell adhesion on FN, but the effect was reduced compared with wild type maspin. C and D, effect of maspin and maspin antibody on HUVEC adhesion to FN (C) and laminin (LN) (D). Note, anti-maspin polyclonal antibody completely abolished the effect of maspin on cell adhesion (GST-Mp-α-M). *, p < 0.05; **, p < 0.01. Mouse IgG (GST-Mp-mIgG) served as control. CK, mock-treated control, was set as 100% of adhesion. Experiments are from triplicates.

Maspin Activates Integrin β₁ Signal Transduction Pathway in Endothelial Cells

Integrins play an important role in cell adhesion. To address whether exogenous maspin affects cell adhesion through integrin, we examined the activation status of integrin following maspin treatment. Because integrin α₅β₁ mediates cell adhesion on FN, we first examined whether integrin β₁ was involved in maspin-mediated HUVEC cell adhesion on FN. Activation of integrin β₁ can be detected with a unique β₁ antibody that recognizes the activated β₁ present within integrin clusters (19). Immunocytochemical analysis, using this specific antibody, showed that integrin β₁ was rapidly activated following the treatment of GST-Maspin. Notably, formation of the integrin cluster was observed throughout the surface of HUVECs following GST-Maspin treatment (Fig. 3A). As a control, GST treatment did not activate integrin β₁ in HUVECs (Fig. 3B). On average, there were 3 sites of clustered integrins observed on each HUVEC treated with GST (n = 8). On the contrary, the GST-Mapsin-treated HUVECs contained 22 clusters per cell (n = 10) (p < 0.05). In general, clustered integrins (indicative of activation) are presented on the protuberant edge of cells, a requirement for focal adhesion formation. However, GST-Maspin-treated HUVECs displayed a unique pattern of activated integrin β₁. The distribution of activated integrin β₁, in GST-Maspin-treated HUVECs, was in a non-polarized manner (Fig. 3A).

FIGURE 3.

Maspin increases EC cell adhesion by activation of integrin β₁. A and B, immunofluorescence staining of active integrin β₁ in HUVECs. Note the presence and lack of active integrin β₁ staining and integrin β₁ clusters in GST-Maspin-treated HUVECs (A) and GST-treated cells (B). Scale bar, 10 μm. C, pretreatment of HUVECs with an anti-integrin β₁ antibody (anti-β₁+) completely abolished the enhancement of maspin on cell adhesion in FN. Note that when GST-Maspin was added in advance, anti-integrin β₁ antibody could no longer block the effect of maspin on cell adhesion (+anti-β₁). D, anti-integrin β₁ blocked the effect of maspin on HUVEC cell adhesion to laminin matrix. *, p < 0.05; **, p < 0.01. E, Western blot analysis showing that GST-Maspin treatment in HUVEC cells resulted in a transient increase of phosphorylated FAK at 30 min, which returned to the base level at 60 min. GST treatment did not activate FAK phosphorylation at any time points. Total FAK served as a loading control. F, ILK kinase phosphorylation assay with MBP as the substrate. Note the higher activity of ILK in GST-Maspin-treated HUVECs compared with GST-treated HUVECs.

To further confirm the activation of integrin β₁ by maspin, we employed an anti-integrin β₁ blocking antibody (AIIB2) to determine whether it could abolish maspin-mediated enhancement of cell adhesion on FN. HUVECs were preincubated with AIIB2 antibody for 60 min and then treated with GST-Maspin or GST. As expected, AIIB2 antibody treatment completely blocked maspin-mediated enhancement of cell adhesion. On the contrary, when HUVECs were first treated by GST-Maspin and then incubated with the AIIB2 antibody, the antibody no longer blocked the enhanced cell adhesion by GST-Maspin (Fig. 3C). Similar results were observed in another adhesion assay using laminin 1 as the matrix (Fig. 3D).

Integrin signal is transferred from outside into the intracellular compartment through the integrin cytoplasmic tail. The cytoplasmic tail of integrin β₁ interacts with ILK, which subsequently phosphorylates a series of signaling molecules in the downstream pathway. In particular, ILK could phosphorylate a MBP, which can be used as substrate in the ILK kinase assay. If maspin activates HUVEC integrin β₁, ILK activity should be increased following GST-Maspin treatment. Using MBP as the substrate in the ILK assay, we found that extracts from GST-Maspin-treated HUVECs can phosphorylate more MBP compared with the GST-treated HUVECs group (Fig. 3E), indicating that GST-Maspin treatment increased ILK activity in HUVECs. Integrin activation leads to phosphorylation of focal adhesion kinase (FAK). FAK is a cytoplasmic non-receptor protein-tyrosine kinase, whose phosphorylation is required by downstream signaling in many cell types. To examine the effect of maspin on phosphorylated FAK in HUVECs, cell extracts harvested from GST-Maspin or GST-treated HUVECs were subjected to immunoprecipitation with an anti-FAK antibody and blotted with PY20 antibody, which recognizes all tyrosine-phosphorylated proteins. We found that the level of phosphorylated FAKs was increased at 30 min following GST-Maspin treatment, which returned to baseline at 60 min (Fig. 3F).

Maspin Modulates the Rearrangement of Cytoskeleton in HUVECs

One of the cellular responses to integrin activation is cytoskeleton reorganization. Vinculin is a cytoskeletal protein that stabilizes the interaction between talin and actin or talin and membrane. By immunofluorescence staining, we found that the vinculin pattern changed drastically in maspin-treated HUVECs. In particular, we found that in GST-Maspin-treated HUVECs, the discontinuous acicular signals were observed in 30% of cells (n = 50) and distributed along the periphery of HUVECs at 30 min following treatment (Fig. 4A). However, at 60 min, 70% of cells (n = 50) exhibited the acicular signals and the density of signals became much stronger compared with 30 min (Fig. 4B). The strength of acicular signals retained the same intensity in 70% of cells at the 120-min time point (n = 50) (Fig. 4C). On the contrary, in the GST-treated control group, less than 10% of cells were observed with very weak signals surrounding the cell periphery at 30 min (n = 50) (Fig. 4D). There was no increase in the number of positive cells nor in the signal intensity in GST-treated HUVECs at the 60- or 120-min time points (Fig. 4, E and F).

FIGURE 4.

GST-Maspin treatment reorganizes cytoskeleton in HUVECs. A–C, immunofluorescence staining against vinculin (VN) showing that acicular signals were observed in GST-Maspin-treated HUVECs at the 30-, 60-, and 120-min time points. D–F, GST-treated HUVECs displayed few and very weak staining of vinculin at the corresponding time points. Arrows, focal adhesions. G–I, phalloidin staining showed F-actin pattern in HUVECs after GST-Maspin treatment for 30, 60, and 120 min. J–L, phalloidin staining showed F-actin pattern in GST-treated HUVECs at corresponding time points. Arrows, microtubles extending into focal adhesions.

Phalloidin was used to monitor the changes of actin cytoskeleton in GST- or GST-Maspin-treated HUVECs. We found that GST-Maspin treatment in HUVECs led to rapid formation of microfilaments, which extended into focal adhesions at 30 min. Abundant thin filaments were distributed in the cytoplasm (Fig. 4G). Up to 60 min post-GST-Maspin treatment, more microfilaments were formed and extended into the focal adhesions, which were distributed in a radial-like pattern along the cell peripheral in an unpolarized manner (Fig. 4H). There were more thin filaments distributed in the cytoplasm as compared with the control cells at the same time point. Even at 120 min after GST-Maspin treatment, there were still more numbers of microfilaments and thin filaments observed as compared with the control treated HUVECs (Fig. 4I), suggesting that GST-Maspin treatment led to reorganization of the cytoskeleton, which facilitates cell adhesion and subsequent spreading on matrix. GST treatment led to different F-actin pattern in HUVECs. Linear F-actin filaments and very few microfilaments were observed scattered in the cytosol of HUVECs at 30 min (Fig. 4J). Up to 60 min, the microfilament amount in cells increased. Most were surrounding the nuclei and a small portion extended into focal adhesions that were distributed in a polarized manner. There were also some thin filaments in the cytoplasm (Fig. 4K). At 120 min of GST treatment, microfilaments in cells were decreased, whereas a small amount of thin filaments with weak staining were observed scattered in the cytoplasm (Fig. 4L).

Maspin Inhibits HUVEC Cell Migration on Extracellular Matrix

Cell migration is a complex process, including delicate control of cell adhesion, disassembly of adhesion, and changes of cytoskeleton. The effect of maspin on HUVEC cell migration was examined by monitoring the migration track of cells on the matrix following GST-Maspin or GST treatments at different time points (2, 7, and 18 h) with or without bFGF. GST-Maspin treatment significantly reduced the HUVEC cell migration rate at all time points (Fig. 5A, panels a–f). However, in the presence of bFGF, GST-Maspin treatment exhibited much stronger inhibition on cell migration with a decrease of over 50% of the migration area compared with the GST treatment (Fig. 5A, panels g–l).

FIGURE 5.

Maspin inhibited HUVEC cells migration. A, effect of maspin on HUVEC cell migration on FN. The areas of the migration track of GST-treated HUVECs (a–c) were larger than that of GSP-Maspin-treated HUVECs (d–f) at 2, 7, and 18 h, respectively. The difference of migration areas was significantly elevated when cells were provided with migration-stimulating bFGF at 2, 7, and 18 h (panel g–l). GST-treated HUVECs (g–i), and GST-Maspin-treated HUVECs (j–l). Lower panel, statistical analysis of the migration areas of HUVECs using NIH Image J software. B, Western blot analysis of active Rac1 and Cdc42 GTPase proteins in GST-Maspin and GST-treated HUVECs in the presence of bFGF (25 ng/ml). Note that both active Rac1 and Cdc42 were declined at the 30- and 60-min time points following GST-Maspin treatment. Lower panel, statistical analysis of Western blotting results using NIH Image J software.

We also examined the pattern of the focal adhesion complex in HUVECs treated with bFGF and GST-Maspin or bFGF and GST. Using vinculin immunofluorescence staining, we monitored the changes of focal adhesion in these HUVECs. In GST-treated HUVECs stimulated with bFGF, focal adhesions were present in a polarized pattern of distribution. Punctate vinculin signals were located at the migrating edge of the HUVECs. However, HUVECs treated with GST-Maspin and bFGF displayed a non-polarized, immobile pattern (data not show).

Cell migration is generally associated with the increase of active Rac1 and cdc42. We further examined Rac1 and cdc42 Rho GTPase levels in HUVEC cells following GST-Maspin or GST treatment. In accordance with the pattern of focal adhesion, we found that both active Rac1 and cdc42 levels in HUVECs were significantly reduced at 30 min with GST-Maspin treatment, but the reduction was transient. At 60 min, their levels began to ascend slightly. However, in GST-treated control cells, active Rac1 and cdc42 levels were increased starting from 15 min and peaked at 30 min, which is a typical response because cells are responding to bFGF stimulation (Fig. 5B).

Maspin Disrupts Focal Adhesion Disassembly

Cell migration is a cyclic process that involves both the adhesion at the protrusive end and the disassembly of focal adhesion at the cell rear. Maspin may inhibit cell migration by promiscuously enhancing cell adhesion at all ends and impairs focal adhesion disassembly at the cell rear. To test this hypothesis, we utilized an established method (14) to analyze the maspin-mediated effect on focal adhesion disassembly. After nocodazole treatment, cells were washed with PBS and focal adhesion started to disassemble. We found that starved HUVECs formed abundant focal adhesions when incubated with 10 nm nocodazole for 4 h (Fig. 6A). In the absence of nocodazole, only a few focal adhesion signals were observed in starved cells (Fig. 6B). After nocodazole washout, focal adhesions rapidly disassembled in HUVECs. As shown in Fig. 6, GST-treated HUVECs exhibited the loss of focal adhesion complex dramatically during the 15–60-min time period after nocodazole washout (Fig. 6, D, F, and H). Focal adhesions became apparent again 120 min after nocodazole washout (Fig. 6J). Alternatively, GST-Maspin-treated HUVECs retained a high level of focal adhesions as indicated by vinculin staining at 15 min after nocodazole washout (Fig. 6C). Compared with the GST-treated control cells, the disassembly of focal adhesion was greatly impaired at 30, 60, and 120 min following nocodazole washout (Fig. 6, E, G, and J), indicating that maspin treatment drastically disrupted the focal adhesion disassembly process.

FIGURE 6.

Maspin disrupted focal adhesion disassembly. A, immunofluorescence staining against vinculin showed abundant formation of focal adhesion in starved HUVECs following nocodazole treatment for 4 h (10 nm). Time point is set as 0 min before nocodazole washout. B, vehicle-treated HUVECs formed very few focal adhesions. Panels C, E, G, and I, vinculin staining exhibited dynamic changes of focal adhesions in GST-Maspin-treated HUVECs at the indicated time points. Panels D, F, H, and J, corresponding alteration of focal adhesions in the GST control group. K, Western blotting analysis showed the alteration of the phosphorylated FAK (Y397) level in HUVECs during focal adhesion disassembly. Total FAK served as a loading control. Note that the level of Tyr(P)³⁹⁷-FAK was relatively higher in GST-Maspin-treated HUVECs compared with GST-treated HUVECs.

To analyze the disassembly of focal adhesion at the molecular level, we examined the changes of phosphorylated FAK (Tyr(P)³⁹⁷) during focal adhesion disassembly with or without maspin treatment by Western immunoblot analysis. Previous studies demonstrated that the loss of focal adhesions results in reduced Tyr(P)³⁹⁷-FAK (15). After nocodazole washout, Tyr(P)³⁹⁷-FAK levels decreased steadily from 0 to 2 h in GST-treated HUVECs. However, GST-Maspin-treated HUVECs maintained the Tyr(P)³⁹⁷-FAK levels during these corresponding time points (Fig. 6K). Collectively, these results suggest that GST-Maspin treatment maintains FAK phosphorylation (Tyr³⁹⁷), which might contribute to the observed impairment of focal adhesion disassembly.

DISCUSSION

Maspin is widely expressed in epithelial cells of many tissues, including mammary gland (3, 9), prostate (20), intestine (3), and lung (21), and in non-epithelial corneal cells as well as prostate stromal cells (4). In breast tumors, maspin expression is down-regulated and the loss of maspin is correlated with a decrease in tumor invasiveness (8). In this study, we detected the expression of maspin in ECs and smooth muscle cells in vessels of embryonic yolk sac tissue and cultured HUVECs (Fig. 1). Impagnatiello et al. (22) reported that maspin was expressed in neovascular endothelial cells. Our laboratory first reported that maspin acted as a potent angiogenesis inhibitor (1). Cell-extracellular matrix interactions between EC and the matrix proteins play a central role in many processes involved in normal vascular development and angiogenesis, including cell signaling, proliferation, and differentiation. The significance of this study is that we have discovered a new mechanism by which maspin controls endothelial cell-extracellular matrix adhesion and angiogenesis. Thus, it is very possible that we can target maspin for blocking angiogenesis and treating other vascular diseases.

Our data suggest that maspin may be secreted from ECs or smooth muscle cells, and can affect EC property through cell membrane receptor(s) signaling. Other studies have supported this claim that secreted maspin can function at the cell surface in an autocrine and/or paracrine manner (17, 19, 23, 24). For example, in prostate tumor cells, maspin inhibits prostate cancer cell migration through increased cell adhesion to various extracellular matrices (19). In MCF-10A human mammary epithelial cells, maspin increased mammary cell adhesion to the self-deposited laminin 5 matrix (10). Others showed that maspin regulated cell motility and adhesion in aggressive breast cancer cells through different signaling pathways (17, 25). To determine the effect of maspin on EC cells, we treated wild type HUVECs or maspin-silenced HUVECs with recombinant GST-Maspin or GST control. We found that maspin increased cell adhesion on FN, laminin, collagen, and vitronectin in a concentration-dependent manner. We noticed that in HUVEC-siRNA maspin cells, maspin treatment has a similar enhancing effect on cell adhesion as that in the wild type HUVECs. Moreover, this enhancing effect could be blocked by an anti-integrin β₁ antibody. It is well known that integrins mediate cell adhesion on matrix through the RGD domain. ECs express multiple integrins, including α_vβ₃, α₆β₁ (26), α₃β₁ (27), α₅β₁ (28), and α₂β₁ (29). Thus, it is possible that maspin could enhance cell adhesion by activating the integrin signaling pathway. Our laboratory previously demonstrated that maspin interacts in a complex with integrin β₁ to enhance mammary epithelial cell adhesion (10). Additionally, integrin β₁ was activated in HUVECs following maspin treatment. More recently, Bass et al. (23) showed that maspin interacted with integrin β₁ in vascular smooth muscle cells. Together, these studies strongly suggest that maspin can control cell adhesion through activation of integrin β₁.

Integrins are transmembrane proteins with cytoplasmic tails that are associated with adaptor proteins, linking integrins with cytoskeleton, cytoplasmic kinases, and transmembrane growth factor receptors (30). Binding of integrin to extracellular ligand causes focal clustering of integrin receptors, resulting in integrin activation (19). Integrin activation triggers ILK, a serine-threonine kinase, which can interact with the cytoplasmic domain of integrin β₁ and β₃, and transfer the outside-in signal (31). In our study, we found that maspin treatment increased ILK activity of HUVEC cells within 30 min, suggesting that maspin could activate ILK through integrin signaling. ILK is a PI3K-dependent kinase, which controls PKB/Akt phosphorylation on serine 473 (32–35). It possesses a binding domain for paxillin, affixin, and PINCH, and together modulates actin cytoskeleton inside the cell (36). Reducing ILK expression and suppressing ILK activity restrains cell attachment through modulating intracellular F-actin organization and CH-ILKBP/paxillin localization (37). Here, we showed that maspin potentiated the ILK activity of HUVECs (Fig. 3D), suggesting that maspin also activates the integrin pathway. Increased ILK activity is positively correlated with enhanced HUVEC cell adhesion and spreading on matrix.

Cell movement is a complex process, involving the extension of the plasma membrane at the cell front and subsequent stabilization of nascent cell-matrix adhesion formed at the tips of membrane protrusions through the interaction between integrin and extracellular matrix. Intracellular signals from integrins induce the formation of a focal adhesion complex, and modulate the dynamics of actin filament for further membrane extension (38). Vinculin, a cytoskeleton protein localized in the focal adhesion complex, contains the binding sites for talin and α-actinin in the globular head region, as well as the binding sites for F-actin and paxillin in its rod-like tail domain. Vinvulin possesses the ability to mediate interactions between integrins and the cytoskeleton at focal adhesions. This is important for control of cytoskeleton reorganization, cell spreading, and lamellipodia formation (39, 40). In this study, we found that maspin treatment led to vinculin pattern change in HUVECs, suggesting that there was a re-arrangement of the cytoskeleton and focal adhesion formation following maspin treatment.

Focal adhesion complex assembly and disassembly are critical for cell adhesion and migration. FAK plays a central role during this process. Integrin activation results in FAK phosphorylation at the Tyr³⁹⁷ site. When FAK is activated, it is autophosphorylated, binds to Src, which in turn phosphorylates other sites on FAK and the FAK-binding proteins, such as Cas and paxillin, which modulates cytoskeleton re-arrangement (41). Abrogation of FAK phosphorylation greatly impairs cell spreading and adhesion (42). In this study, we found that increased focal adhesion following maspin treatment was accompanied by enhanced FAK phosphorylation in HUVECs, which is in accordance with evidence that the level of phosphorylated FAK increases during focal adhesion formation (43). Additionally, FAK interacts with other cytoplasmic signaling molecules (44, 45). For example, FAK phosphorylation leads to phosphorylation of paxillin kinase linker through Src and facilitates the recruitment of paxillin kinase linker to the focal adhesion complex for paxillin binding, which is required for localization of the focal adhesion complex, cell spreading, and membrane protrusion (46). It is possible that maspin may also activate the FAK signal transduction pathway and recruit those molecules to the focal adhesion complexes and facilitate cell spreading.

The Rho family of small GTPases is one of the downstream targets of integrin and FAK (47). Rho family proteins control cell migration by balancing the activities between RhoA, Rac1, and other Rho kinases, modulating cell attachment, migration, and early actin network formation (26, 48). Rac1 influences the formation of nascent focal complexes at the leading edge (49). It activates the Arp2/3 complex and regulates de novo branching of actin filaments at the leading edge, resulting in the formation of lamellipodia (50). The unidirectional lamellipodia formation is often observed when endothelial cells are treated with angiogenic growth factors or chemokines, which is accompanied with a transient increase of Rac1 and cdc42 small GTPase activities. In this study, we observed that bFGF stimulated Rac1 and cdc42 activities as early as 15 min and increased endothelial cell migration through the formation of filopodia and lamellipodia at the leading edge. However, maspin treatment resulted in transient reduction of Rac1 and cdc42 activities at 30 min (Fig. 5B) and restricted the unidirectional formation of filopodia and lamellipodia in HUVECs (data not shown), which led to the repressed cell migration (Fig. 5A). Of note, the concentration of angiogenic factor bFGF added was very high. Under this situation, the inhibitory effect of maspin was gradually reversed at 60 min. However, under physiological conditions with a lower level of angiogenic factors, the inhibitory effect of maspin may last much longer. Interesting, Bass et al. (23) also showed that maspin regulated vascular smooth muscle cell migration through integrin β₁. They found that vascular smooth muscle cells treated with maspin for 60 min caused a decrease of active β₁. However, our study of HUVECs analyzed cell adhesion at much earlier signaling stages (0, 15, and 30 min) following maspin treatment. It is possible that HUVECs treated with maspin for a longer time may also cause a decrease of integrin β₁ activation, due to a desensitization of integrin signaling.

Cell migration involves not only the adhesion at the leading edge but also the disassembly of focal adhesion at the rear end. In response to angiogenic factors, endothelial cells first attach to certain matrix proteins at the leading edge to form focal adhesion. Subsequently, cells translocate forward and form a new leading edge and new cell-matrix adhesion at the tips of membrane protrusions. Cell migration occurs when there is a turnover/disassembly of previous focal adhesions. Based on our studies, we found that maspin inhibited bFGF-induced cell migration by two mechanisms. First, maspin enhanced focal adhesion formation in HUVECs. Second, maspin blocked the turnover/disassembly of focal adhesion. In this study, we used a well established focal adhesion turnover model to test the effect of maspin on focal adhesion turnover/disassembly (Fig. 6). Nocodazole treatment activates Rho GTPase and stimulates formation of focal adhesion and stress fibers (39, 51). After nocodazole washout, focal adhesions disassemble. We found that compared with the mock-treated cells, focal adhesion disassembly was reduced in maspin-treated HUVECs. One key assay that measures focal adhesion disassembly is to examine the level of phosphorylation of FAK (Tyr³⁹⁷) in nocodazole-treated cells. In a previous report, Ezratty et al. (14) showed that the level of Tyr(P)³⁹⁷-FAK was significantly reduced during the focal adhesion disassembly process. After a 2-h washout of nocodazole, Tyr(P)³⁹⁷-FAK levels were increased in conjunction with the newly formed focal adhesion complexes (14). We observed that the new focal adhesion was formed in control group of HUVECs 2 h after nocodazole washout, which was accompanied with the increased Tyr(P)³⁹⁷-FAK levels. However, in maspin-treated HUVECs, we did not observe focal adhesion disassembly after nocodazole washout. Rather, we found that Tyr(P)³⁹⁷-FAK in HUVECs treated with maspin was maintained at a much higher level compared with the mock treated HUVECs. This high level of Tyr(P)³⁹⁷-FAK prevented focal adhesion disassembly, leading to restricted cell migration. The level of Tyr(P)³⁹⁷-FAK was reduced 2 h after nocodazole washout, which was significantly delayed compared with mock-treated HUVECs, suggesting that maspin treatment slows focal adhesion disassembly resulting in impaired cell migration (Fig. 6).

In accordance with the above results, we also observed that maspin-treated HUVECs presented more intracellular actin stress fibers than the mock-treated control cells (Fig. 4). A recent study by Katz et al. (52) indicated that the increased formation of intracellular stress fibers correlated with enhanced cell adhesion in mammary cells, whereas migrating cells had fewer stress fibers. This provides another mechanism by which maspin inhibits endothelial cell migration.

In summary, this study provides direct evidence that maspin enhances HUVEC cell adhesion and restricts cell migration through the integrin signaling pathway. It regulates the activation of ILK and FAK and causes the reorganization of the actin cytoskeleton. Maspin prevents focal adhesion disassembly by increasing FAK Tyr³⁹⁷ phosphorylation resulting in impaired endothelial cell migration.