Vascular endothelial cells produce considerable amounts of matrix metalloproteinases (MMP), including MMP-2, MMP-9, and membrane type 1 (MT1)–MMP. However, little is known about the regulatory mechanisms of these protease activities exhibited during vascular development. A glycosylphosphatidylinositol-anchored glycoprotein, reversion-inducing cysteine-rich protein with Kazal motifs (RECK), has been shown to attenuate MMP-2 maturation by directly interacting with MT1-MMP. Here, we show that an angiogenic factor angiopoietin-1 induces RECK expression in human umbilical vein endothelial cells (HUVEC), and RECK depletion in these cells results in defective vascular tube formation and cellular senescence. We further observed that RECK depletion downregulates β1-integrin activation, which was associated with decreased autophosphorylation of focal adhesion kinase and increased expression of a cyclin-dependent kinase inhibitor p21CIP1. In agreement, significant downregulation of β1-integrin activity was observed in vascular endothelial cells in Reck−/− mouse embryos. In HUVECs, specific inhibition of MMP-2 significantly antagonized the effect of RECK depletion on β1-integrin signaling, cell proliferation, and tube elongation. Furthermore, we observed that hypervascular tumor-derived cell lines can induce high RECK expression in convoluted vascular endothelial cells, and this in turn supports tumor growth. Targeting RECK specifically in tumor-associated vascular endothelial cells resulted in tumor regression. Therefore, we propose that RECK in tumor vascular endothelial cells can be an interesting target of cancer treatment via abortion of tumor angiogenesis. Mol Cancer Res; 8(5); 665–76. ©2010 AACR.

This article is featured in Highlights of This Issue, p. 627

Matrix metalloproteinases (MMP) are Ca2+ and Zn2+-dependent endopeptidases that play crucial roles in the degradation of various extracellular matrix (ECM) components (1). In addition, recent studies have shown that these enzymes exert unique biological functions by acting on various important modulators of cellular functions (2, 3). During tumor propagation, not only tumor cells but also stromal cells, including vascular endothelial cells, produce considerable amounts of MMPs (4). In particular, during angiogenesis engaged by endothelial cells, MMPs play both positive and negative roles (5). The proangiogenic functions of MMPs include degradation of ECM components, which facilitates the invasion and migration of endothelial cells, cleavage of endothelial cell-cell adhesion mediated by cadherins, generation of cryptic integrin-binding sites from ECM components, and enhancement of the bioavailability of various growth factors, such as vascular endothelial growth factor, transforming growth factor-β, and connective tissue growth factor (5). On the other hand, the antiangiogenic functions of MMPs include ectodomain shedding of receptors for growth factors, such as fibroblast growth factor receptor 1 and urokinase-type plasminogen activator receptor, resulting in the inhibition of growth signals generated from these receptors (5). Other antiangiogenic functions of MMPs include generation of cryptic angiogenesis inhibitors from plasminogen or collagens, such as angiostatin, endostatin, tumstatin, arrestin, or canstatin; all of these are supposed to exert their antiangiogenic functions by interacting with various forms of integrins, including α5β1, αvβ3, α1β1, and α3β1 (5, 6). Moreover, soluble hemopexin domain derived from degraded MMP-2 blocks the binding of intact MMP-2 to integrin αvβ3 (7). These findings suggest that the antiangiogenic functions of MMPs profoundly involve integrin signaling.

Integrins principally serve as adhesion receptors for ECM components and provide the central system to transduce biochemical signals from outside of the cells (6). Ligand occupancy and the resultant clustering of integrin receptors induce autophosphorylation of focal adhesion kinase (FAK), which evokes cellular actions (proliferation, migration, spreading, or apoptosis) that enable cells to adjust to changes in the extracellular environment (8).

The RECK gene has been identified as a negative transcriptional target of molecules, including multiple retroviral oncogenes (9), EBV latent membrane protein 1 (10), histone deacetylase (11), and oncogenic microRNA miR-21 (1214). The product of this gene can directly bind to a series of metalloendopeptidases, including MMP-9 (9), MMP-2 (15), membrane type 1 (MT1)–MMP, CD13/APN (16), and ADAM10 (17), and attenuate their proteolytic activity competitively in most of cases. Two metalloprotease substrate–like domains recently discovered in RECK may provide a structural basis for its biochemical actions.7

7Y. Takegami and C. Takahashi, in preparation.

Yet, the most discrete structural feature of RECK compared with known primary tissue metalloprotease inhibitors, such as tissue inhibitors of metalloproteinases and α2-macroglobulins (18), is that RECK is covalently anchored to the plasma membrane surface via glycosylphosphatidylinositol anchors. Because of this structural feature, RECK controls metalloendopeptidase activities in highly limited distance from the membrane surface (perimembrane area), and moreover, it exerts unique functions by interacting with membrane-tethered metalloendopeptidases. Indeed, molecular interaction between RECK and MT1-MMP enables MT1-MMP to translocate to the lipid raft fraction, which consequently changes the route of MT1-MMP endocytosis to one that is specific to that of glycosylphosphatidylinositol-anchored proteins; this results in a shorter half-life of MT1-MMP (16). Furthermore, RECK was found to stabilize the interaction between membrane-bound Notch ligands and Notch receptors by protecting Notch ligands from ADAM10-mediated ectodomain shedding, which critically contributes to the maintenance of the self-renewal property of neural stem cells (17).

We recently observed that treatment with an angiogenic factor, namely, angiopoietin-1, or coculture with hypervascular tumor-derived cells markedly induced RECK expression in human umbilical vein endothelial cells (HUVEC), which indicates that RECK may possess proangiogenic function. Indeed, severe developmental defects in the vascular network were observed in the previous analysis of Reck−/− embryos (15). No such vascular phenotypes were observed in mice deficient in known soluble MMP inhibitors (18). We further examined Reck−/− mice and identified significant developmental defects in vascular endothelial cells. Although the vascular phenotype appearing in Reck−/− embryos can be attributed not only to the defects in endothelial cells but also to those in smooth muscle cells, pericytes, and the surrounding stromal cells, in this study, we focused on the role of RECK in the development of vascular endothelial cells under physiologic and pathologic (tumor angiogenesis) conditions, mainly by using RNA interference techniques.

Cell culture and transfection

The HUVECs (CC2517, CAMBREX) were maintained in EBM-2 (growth factor free) or EGM-2 (supplemented with endothelial cell growth factors) bullet kit (CC-3162, CAMBREX) and used between passages 3 and 6. Tumor cells were cultured in DMEM supplemented with 10% FCS. Transfection of HT1080 cells was described (16).

Antibody

The following mouse or rat primary antibodies were used: anti-RECK (9), anti-α-tubulin (CP06, Calbiochem), anti-MMP-2 (F-68, Biotechnology Products), anti-5-bromo-2′-deoxyuridine (BrdUrd) (MAB4072, Chemicon), anti-active β1-integrin (9EG7, Chemicon), anti-β1-integrin (610467, BD), anti-phosphorylated FAK (FAKpY397; 611722, BD), anti-FAK (sc-557, Santa Cruz), anti-p21 (556430, BD), anti-CD31 (01951A, Pharmingen; for nude mouse assay), anti-CD31 (ab28364, Abcam; for human tumor), and anti-human CD34 (hCD34; CBL496, Chemicon) antibodies. As secondary antibodies, we used antimouse IgG phycoerythrin (PE) (550083, BD), antimouse IgG Alexa Fluor 555 (A21424, Molecular Probes), antirat IgG APC (734820, Cell Lab), antimouse IgG Texas red (T862, Molecular Probes), and antirabbit IgG Alexa Fluor 488 (Molecular Probes, A11034) antibodies. For stimulating β1-integrin activation, we used mouse anti-β1 integrin monoclonal antibody (P4G11, Chemicon; ref. 19).

Immunoblotting

Immunoblotting was done as described previously (16).

Mice

Reck+/− mice (15) were crossed with C57BL/6 Cr Slc TgN(act-EGFP)OsbC14-Y01-FM131 (20), and resultant progenies were intercrossed to generate Reck−/− embryos transmitting enhanced green fluorescent protein (EGFP) gene.

RNA interference

The siRNAs specifically targeting human RECK (siRNA1: 29149; siRNA2: 29155) and EGFP (4626) and negative control (4611G) were purchased from Ambion, and those targeting MMP-2 were purchased from Invitrogen (siRNA1: 1178208, siRNA2: 1178210). HUVECs (1 × 105) were transfected with 50 μmol/L siRNA using Lipofectamine 2000 (Invitrogen).

MMP assay

The synthetic peptide Mca-Lys-Pro-Leu-Gly-DPA-Ala-Arg-NH2 (5 μmol/L) was mixed with culture supernatants from HUVECs transfected with siRNA and incubated at 37°C for 30 minutes. Cleavage was measured as described previously (16). For the detection of MMP-2, culture supernatant was filtered through 0.45-μm ultrafiltration membrane and incubated with Gelatin Sepharose 4B (GE Healthcare) at 4°C. Proteins bound to the beads were analyzed by immunoblotting using antibody to MMP-2. Signal intensity was measured using LAS3000 chemiluminescence imaging system (Fuji).

Two-dimensional and three-dimensional vascular formation assay

For two-dimensional culture, 4 × 104 HUVECs transduced with siRNA were grown on growth factor–reduced Matrigel (354230, BD)–coated wells and incubated at 37°C for 48 hours. For three-dimensional culture, type I collagen (Cellmatrix type I-A, Nitta Gelatin) solution was mixed on ice with 10× modified Eagle's medium and then with 1 × 106 cells/mL. The mixture was allowed to solidify for 30 minutes at 37°C and incubated with EGM-2 bullet kit supplemented with 50 ng/mL phorbol 12-myristate 13-acetate for 5 days.

Reverse transcription–PCR

RNA was extracted using Trizol (15596-026, Invitrogen) and reverse transcribed using TaKaRa RNA PCR kit (RR019A, TaKaRa). PCR was done with specific primer sets: RECK (30 cycles), sense: 5′-ACTCCCTCCTCCTTCCCCTCAGC-3′, antisense: 5′-ATTTAATCAGCTTGCTTTTGCAT-3′; GAPDH (28 cycles), sense: 5′-ACCACAGTCCATGCCATCAC-3′, antisense: 5′-TCCACCACCCTGTTGCTGTA-3′; β1-integrin, sense: 5′-TTATCCTTCTATTGCTCACCTTGTC-3′, antisense: 5′-ATAACCTCTACTTCCTCCGTAAAGC-3′.

Proliferation assay

HUVECs were grown on 96-well plates (2 × 103 per well). After siRNA transfection, the cell numbers were subsequently quantified using Cell Count Reagent SF (07553-15, Nacalai Tesque).

BrdUrd uptake and cell cycle

Cultured cells were incubated with 3 μg/mL of BrdUrd for 90 minutes, trypsinized, and fixed with 70% ethanol. E9.5 embryos were labeled in utero for 2 hours by a single pulse of BrdUrd (3 mg). Samples were reacted with anti-BrdUrd and then with anti-mouse IgG PE antibody and subjected to fluorescence-activated cell sorting (FACS) analysis. For measuring propidium iodide incorporation, cells were fixed with 4% paraformaldehyde (PFA) in PBS and suspended in PBS with 50 μg/mL propidium iodide and 100 units/mL RNase A. At least 10,000 events were acquired.

Senescence-associated β-galactosidase assay

Senescence-associated β-galactosidase activity in cells was detected as described previously (21).

FACS analysis

HUVECs were trypsinized, fixed with 4% PFA, suspended in PBS with 0.5% bovine serum albumin, and stained with anti-β1-integrin antibodies (active form, 9EG7; total β1 integrin, P4G11) and secondary with antirat IgG-APC or antimouse IgG-PE, respectively. Data were collected using FACS Aria (BD Biosciences).

Immunohistochemistry

The specimens were fixed with 1% PFA, embedded in optimal cutting temperature compound, and stained as described previously (17).

Luciferase assay

HUVECs (4 × 105) were plated onto 60-mm dishes and transfected with mouse Reck promoter luciferase reporter pGL3-4110-luc (22) and pCMV-β-galactosidase expression vector (21). Twenty-four hours after transfection, cells were trypsinized, mixed with 4 × 105 tumor cells, and cultured for 24 hours. Luciferase and β-galactosidase activities in cells were measured as described (23). β-Galactosidase activity was used to normalize luciferase activity in each transfection.

Human tumors

Biopsy samples of adrenal mass were collected from two patients with neuroblastoma who had been treated at Kyoto University Hospital and analyzed according to the guideline of Kyoto University Graduate School of Medicine. Informed consent was obtained from the guardians of patients.

Severe combined immunodeficiency mouse model of tumor angiogenesis

Porous poly-l-lactic acid (PLLA; Sigma) scaffolds were prepared as described (24). siRNA-transfected HUVECs and HT1080 cells (5 × 105) were each mixed with Matrigel, loaded into scaffolds, and implanted s.c. into 5-week-old female severe combined immunodeficiency (SCID) mice (CLEA). Three weeks after transplantation, recovered scaffolds were weighed, fixed with 1% PFA in PBS, and embedded in optimal cutting temperature compound.

Angiopoietin-1 upregulates RECK in vascular endothelial cells

To address whether RECK is involved in the development of vascular endothelial cells, we first examined the effects of various angiogenic factors on the transcriptional regulation of RECK in HUVECs cultured in EBM-2 (growth factor–free culture medium). This study revealed that RECK expression in HUVECs was induced by angiopoietin-1, but not by vascular endothelial growth factor, fibroblast growth factor, or epidermal growth factor (Fig. 1A), suggesting that RECK expression is regulated in vascular endothelial cells by a specific angiogenic factor.

FIGURE 1.

Roles of RECK in vascular development. A, HUVECs grown in the growth factor–free culture medium (EBM-2) and then treated with the indicated concentration of angiopoietin-1 or other growth factors (10 ng/mL) were analyzed by immunoblotting for the indicated proteins. NT, no treatment. B, vascular phenotypes in mice of the indicated genotype at E10.0. Scale bar, 10 μm. Arrow, an endothelial cell detached from the subendothelial layer. C, HUVECs transfected with the indicated siRNA and grown in the culture medium supplemented with growth factors (EGM-2) were analyzed by immunoblotting (left). Two independent clones of siRNA targeting RECK (clones 1 and 2) were used. Endopeptidase activities to cleave Mca-Lys-Pro-Leu-Gly-DPA-Ala-Arg-NH2 in the culture supernatants were measured (middle). Active per total MMP-2 was determined by precipitation with gelatin beads and following immunoblotting using antibody to MMP-2 (right). Columns, mean from three experiments; bars, SEM. D, the reverse transcription–PCR analysis of HUVECs transfected with the indicated siRNA, trypsinized, suspended for 30 minutes, replated on Matrigel, and grown for the indicated period of time. E, Cells at 48 hours were analyzed using light microscopy. Scale bar, 100 μm.

FIGURE 1.

Roles of RECK in vascular development. A, HUVECs grown in the growth factor–free culture medium (EBM-2) and then treated with the indicated concentration of angiopoietin-1 or other growth factors (10 ng/mL) were analyzed by immunoblotting for the indicated proteins. NT, no treatment. B, vascular phenotypes in mice of the indicated genotype at E10.0. Scale bar, 10 μm. Arrow, an endothelial cell detached from the subendothelial layer. C, HUVECs transfected with the indicated siRNA and grown in the culture medium supplemented with growth factors (EGM-2) were analyzed by immunoblotting (left). Two independent clones of siRNA targeting RECK (clones 1 and 2) were used. Endopeptidase activities to cleave Mca-Lys-Pro-Leu-Gly-DPA-Ala-Arg-NH2 in the culture supernatants were measured (middle). Active per total MMP-2 was determined by precipitation with gelatin beads and following immunoblotting using antibody to MMP-2 (right). Columns, mean from three experiments; bars, SEM. D, the reverse transcription–PCR analysis of HUVECs transfected with the indicated siRNA, trypsinized, suspended for 30 minutes, replated on Matrigel, and grown for the indicated period of time. E, Cells at 48 hours were analyzed using light microscopy. Scale bar, 100 μm.

Close modal

Abnormal vascular endothelial development in Reck-null embryos

A previous study revealed that Reck−/− embryos exhibited abnormal branching of vasculatures, and overexpression of RECK in tumor cells modified tumor angiogenesis (15). We therefore speculated that RECK could be implicated in both physiologic and pathologic angiogenesis. Additional introduction of the genetic background carrying EGFP transgene into Reck−/− embryos enabled us to observe the presence of significant developmental defects in vascular endothelial cells, namely, abnormal alignment and detachment from the subendothelial layer (Fig. 1B); these were not aware in the previous study without introducing EGFP background (15). We observed such obvious morphologic disorders of vascular alignment in four of five Reck−/−;EGFP E10.0 embryos by carefully observing vasculatures in multiple slices of thin paraffin sections under fluorescence microscope, but never in five control embryos. These findings prompted us to examine the function of RECK specifically in vascular endothelial cells.

RECK depletion affects tube formation by HUVECs

To directly assess the role of RECK in angiogenesis, we did siRNA-directed depletion of RECK in HUVECs cultured in EGM-2 (culture medium supplemented with growth factors), which resulted in >95% reduction in RECK expression at 48 hours after transduction (Fig. 1C, left). This treatment increased the net metalloendopeptidase activity detected in the culture supernatant by 2-fold to 3-fold compared with that in control supernatant (Fig. 1C, middle). We also observed increased level of active form MMP-2 in RECK-depleted HUVECs (Fig. 1C, right). HUVECs transduced with control siRNA expressed RECK at 6 hours after plating onto Matrigel, and the cells were able to form tube-like structures and branches within 48 hours. However, in HUVECs transduced with siRNA targeting RECK (Fig. 1D, top), the tubes formed under the same conditions were significantly shorter and often disconnected (Fig. 1D, bottom). Shorter tube formation induced by RECK depletion was evident in type I collagen three-dimensional gel as well as in the two-dimensional assay (Supplementary Fig. S1A and B). These findings suggest that RECK plays a critical role in physiologic angiogenesis.

RECK depletion induces growth arrest and cellular senescence in HUVECs

To further characterize the primary effect of RECK depletion in HUVECs, we analyzed monolayer cells grown on noncoated plastic dish with EGM-2. HUVECs transduced with RECK siRNAs proliferated significantly slower than those transduced with control or EGFP siRNA (Fig. 2A). The G0-G1 population in cells transduced with control siRNA was 69%, and this figure increased to 85% when the cells were transduced with RECK siRNA (Fig. 2B). Similarly, BrdUrd incorporation was significantly reduced by RECK depletion (Fig. 2C). Twelve days after transduction of RECK siRNA, we observed a significant level of senescence-associated β-galactosidase activity (Fig. 2D) without any evidence of increased cell death as assessed by the terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling assay (data not shown). These findings suggest that RECK critically supports the proliferation of HUVECs.

FIGURE 2.

Effect of RECK depletion on the growth of HUVECs. A, growth curve of HUVECs transfected with the indicated siRNA. Culture medium, EGM-2. Points, mean from three experiments; bars, SEM. B, cell cycle analysis of HUVECs at 48 hours after transfection of the indicated siRNA. Data were obtained by FACS analysis. C, BrdUrd incorporation by HUVECs at 48 hours after transfection of the indicated siRNA. Columns, mean from three experiments; bars, SEM. D, detection of senescence-associated β-galactosidase activity in HUVECs at 12 days after transfection of the indicated siRNA.

FIGURE 2.

Effect of RECK depletion on the growth of HUVECs. A, growth curve of HUVECs transfected with the indicated siRNA. Culture medium, EGM-2. Points, mean from three experiments; bars, SEM. B, cell cycle analysis of HUVECs at 48 hours after transfection of the indicated siRNA. Data were obtained by FACS analysis. C, BrdUrd incorporation by HUVECs at 48 hours after transfection of the indicated siRNA. Columns, mean from three experiments; bars, SEM. D, detection of senescence-associated β-galactosidase activity in HUVECs at 12 days after transfection of the indicated siRNA.

Close modal

Effects of RECK depletion on HUVECs partially depends on MMP-2

Because it was previously shown that RECK inhibits the proteolytic activity of a series of metalloendopeptidases, including MMP-9 (9), MMP-2 (15), MT1-MMP, CD13/aminopeptidase N (16), and ADAM10 (17), we hypothesized that deregulated activation of any of these enzymes induced by RECK depletion would contribute to the appearance of phenotypes in HUVECs. siRNAs targeting individual genes encoding these enzymes were simultaneously introduced with RECK siRNAs into monolayer HUVECs cultured with EGM-2 (Fig. 3A; data not shown). Among tested siRNAs, MMP-2 siRNAs significantly antagonized the effect of RECK siRNAs on the proliferation of HUVECs(Fig. 3B). In contrast, depletion of MMP-2 without RECK depletion did not produce detectable effect on cell proliferation (Fig. 3B), suggesting that MMP-2 acquires a specific role in the absence of RECK. In the two-dimensional assay, the effect of RECK siRNA on tube elongation was significantly rescued by simultaneous inhibition of MMP-2 (Fig. 3C); however, the defect in tube branching was not rescued (Fig. 3D). The three-dimensional assay results were consistent with the two-dimensional results (Supplementary Fig. S2). These findings suggest that the effects of RECK depletion on vascular formation by HUVECs partially depends on MMP-2 and that shorter tube formation correlates withproliferation defect. Furthermore, MT1-MMP nor MMP-9 depletion failed to reverse proliferation defects in RECK-depleted HUVECs (data not shown); the implication of these findings will be discussed later (see below).

FIGURE 3.

Effect of RECK depletion on HUVECs partially depends on MMP-2. A, expression of the indicated proteins in HUVECs transfected with the indicated siRNA was analyzed by immunoblotting. Culture medium, EGM-2. B, proliferation activity of HUVECs transfected with the indicated siRNA. Columns, mean from five experiments; bars, SEM. C, HUVECs transfected with the indicated siRNA were grown on Matrigel for 48 hours and observed under light microscopy. Scale bars, 100 μm. D, tube length and number of branches were measured. Control siRNA-transfected HUVECs were set to 100. Columns, mean from five experiments; bars, SEM. *, P < 0.01.

FIGURE 3.

Effect of RECK depletion on HUVECs partially depends on MMP-2. A, expression of the indicated proteins in HUVECs transfected with the indicated siRNA was analyzed by immunoblotting. Culture medium, EGM-2. B, proliferation activity of HUVECs transfected with the indicated siRNA. Columns, mean from five experiments; bars, SEM. C, HUVECs transfected with the indicated siRNA were grown on Matrigel for 48 hours and observed under light microscopy. Scale bars, 100 μm. D, tube length and number of branches were measured. Control siRNA-transfected HUVECs were set to 100. Columns, mean from five experiments; bars, SEM. *, P < 0.01.

Close modal

RECK depletion downregulates the activation of β1-integrin and FAK in an MMP-2–dependent manner

Next, we addressed whether the involvement of MMP-2 can explain why RECK depletion suppresses the proliferation of HUVECs. Because integrins are frequent target of antiangiogenic MMP activities (5), we measured the activity of various types of integrins using antibodies that specifically recognize the activated forms of integrins by FACS. This study revealed that the activation of β1-integrin is specifically affected by RECK depletion (Fig. 4A). No quantitative differences were detected between cells transduced with RECK siRNA and the controls by both FACS and immunoblotting with antibody reacting with all forms of β1-integrin (Supplementary Fig. S3A). In addition, RECK depletion did not affect β1-integrin mRNA expression as assessed by reverse transcription–PCR (Fig. S3B). Furthermore, overexpression of RECK in HT1080 fibrosarcoma cells resulted in increased β1-integrin activation (1.7-fold induction; Fig. 4B) without detectable alteration in the total β1-integrin expression level (Supplementary Fig. S3C and D). These findings suggest that RECK stimulates β1-integrin activation in both normal and tumor cells.

FIGURE 4.

The β1-integrin–dependent function of RECK. A, the activity of β1-integrin signaling in HUVECs transfected with the indicated siRNA was measured by FACS using the monoclonal antibody (9EG7) that specifically recognizes active form β1-integrin on the cell surface (top). Culture medium, EGM-2. The FACS result obtained without adding the primary antibody is also indicated. The quantification of active (bottom left) and total (bottom right) β1-integrin is shown. The signal in cells transduced with the control siRNA was set to 1.0. Columns, mean from five experiments; bars, SEM. *, P < 0.01. B, HT1080 cells transfected with pCXN2-neo (Vector) or pCXN2-neo-RECK and selected were analyzed as in A. Columns, mean from three experiments; bars, SEM. C, the activity of β1-integrin signaling in HUVECs transfected with the indicated combinations of siRNA was measured as in A. Columns, mean from three experiments; bars, SEM. *, P < 0.01. D, the autophosphorylation status and total amount of FAK were analyzed. E, proliferation of HUVECs transfected with the indicated siRNA cultured in the presence of indicated concentrations of antibodies (P4G11). Columns, mean from five experiments; bars, SEM. *, P < 0.01. F, the expression of the indicated proteins in HUVECs treated with the indicated siRNAs and antibodies (2 μg/mL) was analyzed. G, the expression of the indicated proteins in HUVECs treated with the indicated combinations of siRNAs was analyzed.

FIGURE 4.

The β1-integrin–dependent function of RECK. A, the activity of β1-integrin signaling in HUVECs transfected with the indicated siRNA was measured by FACS using the monoclonal antibody (9EG7) that specifically recognizes active form β1-integrin on the cell surface (top). Culture medium, EGM-2. The FACS result obtained without adding the primary antibody is also indicated. The quantification of active (bottom left) and total (bottom right) β1-integrin is shown. The signal in cells transduced with the control siRNA was set to 1.0. Columns, mean from five experiments; bars, SEM. *, P < 0.01. B, HT1080 cells transfected with pCXN2-neo (Vector) or pCXN2-neo-RECK and selected were analyzed as in A. Columns, mean from three experiments; bars, SEM. C, the activity of β1-integrin signaling in HUVECs transfected with the indicated combinations of siRNA was measured as in A. Columns, mean from three experiments; bars, SEM. *, P < 0.01. D, the autophosphorylation status and total amount of FAK were analyzed. E, proliferation of HUVECs transfected with the indicated siRNA cultured in the presence of indicated concentrations of antibodies (P4G11). Columns, mean from five experiments; bars, SEM. *, P < 0.01. F, the expression of the indicated proteins in HUVECs treated with the indicated siRNAs and antibodies (2 μg/mL) was analyzed. G, the expression of the indicated proteins in HUVECs treated with the indicated combinations of siRNAs was analyzed.

Close modal

To determine whether downregulation of β1-integrin activation in HUVECs induced by RECK depletion depends on MMP-2, siRNA specifically targeting MMP-2 was simultaneously introduced into these cells cultured in EGM-2. This experiment showed that the simultaneous inhibition of MMP-2 activity significantly antagonizes the effect of RECK depletion on β1-integrin activation in HUVECs (Fig. 4C).

To validate the relationship between RECK and integrin signaling, we measured the autophosphorylation status of FAK as one of common downstream targets of integrin signaling (8). RECK depletion downregulated FAK phosphorylation, and this downregulation was significantly antagonized by the simultaneous depletion of MMP-2 (Fig. 4D).

RECK depletion upregulates p21CIP1 expression in a β1-integrin–dependent manner

To prove that downregulation of β1-integrin signaling is responsible for the growth arrest and cellular senescence induced by RECK depletion in HUVECs, we treated RECK-depleted HUVECs with monoclonal antibody that was engineered to specifically stimulate β1-integrin activation (19). This experiment resulted in the significant and dose-dependent recovery of cell proliferation in RECK-depleted HUVECs in EGM-2 (Fig. 4E). Furthermore, we observed specific induction of a cyclin-dependent kinase inhibitor p21CIP1 in HUVECs when transduced with RECK siRNA; this effect of RECK siRNA was antagonized by treatment with the β1-integrin–stimulating monoclonal antibody (Fig. 4F). In addition, simultaneous depletion of MMP-2 antagonized the effect of RECK depletion on p21CIP1 induction (Fig. 4G). These findings suggest that RECK stimulates FAK activation and suppresses p21CIP1 expression by regulating β1-integrin activation in an MMP-2–dependent manner. Because FAK has been linked to Skp2-independent control of the cellular p21CIP1 level (25), we speculated that RECK depletion induces p21CIP1 expression by downregulation of β1-integrin signaling followed by inactivation of FAK.

Reck deficiency downregulates β1-integrin activation in vivo

We next examined the activation status of β1-integrin in vascular endothelial cells in Reck−/− mice to determine the relevance of our in vitro findings in vivo. Most of Reck−/− embryos survive up to embryonic day (E) 10.5 and then die abruptly due to acute hemorrhage in the abdominal vessels (15). In E10.0 wild-type embryos, CD31+ vascular endothelial cells abundantly express both RECK and β1-integrin. Most CD31+ cells express active β1-integrin in wild-type E10.0 embryos; however, in abdominal hemorrhage-free and heart beat–positive (alive) Reck−/− E10.0 embryos, we detected active β1-integrin signals with significantly less frequency, although the total β1-integrin signals were at level similar to that of the wild-type (Fig. 5A and B). These findings suggest that Reck deficiency downregulated β1-integrin activation in the vascular endothelial cells of E10.0 mouse embryos.

FIGURE 5.

Reck deficiency downregulates β1-integrin signaling in mouse embryos. A, representative results of immunofluorescence analysis for the indicated proteins in the aorta of the indicated genotype of mice at E10.0 are shown. Total β1-integrin+/active β1-integrin endothelial cells are indicated by arrows. Scale bar, 10 μm. B, the activity of β1-integrin signaling is quantified from observation of three embryos of each genotype by assessing the frequency of active per total β1-integrin signals. Columns, mean; bars, SEM. *, P < 0.01.

FIGURE 5.

Reck deficiency downregulates β1-integrin signaling in mouse embryos. A, representative results of immunofluorescence analysis for the indicated proteins in the aorta of the indicated genotype of mice at E10.0 are shown. Total β1-integrin+/active β1-integrin endothelial cells are indicated by arrows. Scale bar, 10 μm. B, the activity of β1-integrin signaling is quantified from observation of three embryos of each genotype by assessing the frequency of active per total β1-integrin signals. Columns, mean; bars, SEM. *, P < 0.01.

Close modal

Hypervascular tumor cells induce RECK expression in vascular endothelial cells

To investigate the role of RECK in pathologic angiogenesis, we observed RECK expression in mouse-derived CD31+ endothelial cells convoluted in human-derived tumor cells (HT1080) that were s.c. inoculated into nude mice and grown for 14 days. We observed a marked induction of RECK in mouse-derived CD31+ cells in the tumors (Fig. 6A). This phenomenon was recapitulated in vitro by coculturing HUVECs with tumor cells on a noncoated dish. Significant induction of RECK under this condition was confirmed by immunoblotting and the reporter assay for the Reck promoter (Fig. 6B, top; Supplementary Fig. S4A). Hypoxic condition did not induce RECK expression in HUVECs (data not shown); thus, we speculated that tumor cells directly induce the transcription of RECK in vascular endothelial cells convoluted in the tumor mass.

FIGURE 6.

RECK induction in vascular endothelial cells supports tumor growth. A, immunofluorescence analysis of the host mouse-derived blood vessels developed in a tumor generated by s.c. introducing 1 × 106 HT1080 cells into a 4-week-old BALB/c nu/nu mouse. Antibodies to the indicated proteins were used. Arrows, RECK+/CD31+ endothelial cells. Scale bars, 20 μm. B, induction of RECK in HUVECs cocultured with HT1080 cells. HUVECs (5 × 104) were cocultured with 2.5 (+), 5.0 (++), or 10 (+++) × 104 HT1080 cells for 24 hours (upper top). Culture medium, DMEM supplemented with 10% FCS. Endogenous RECK expression in the same number of HT1080 cells is also examined (upper bottom). Induction of RECK in 5 × 104 HUVECs cocultured with the indicated tumor cell lines (5 × 104; lower top). Endogenous RECK expression in the same number of tumor cells is also examined (lower bottom). C, a section from a human neuroblastoma obtained by biopsy was stained with H&E (top). Scale bar, 100 μm. RECK expression in a CD31+ blood vessel found in the tumor mass was detected by immunofluorescence (bottom). Scale bar, 10 μm. D, HT1080 cells and HUVECs were loaded into PLLA scaffolds and grown in SCID mice for 3 weeks. Sections of different parts of a tumor (marginal and center) were stained for the indicated proteins (top). Arrow heads, CD31+/hCD34 cells; arrows, CD31+/hCD34+ cells. HT1080 cells and HUVECs transfected with the indicated siRNA were loaded into PLLA scaffolds and grown in SCID mice for 3 weeks. Tumor sections were stained for the indicated protein and 4′,6-diamidino-2-phenylindole (DAPI) or H&E (bottom). Scale bar, 100 μm. Density of hCD34+ endothelial cells (per ×40 magnification field, 10 fields were observed; top right) and tumor weights (bottom right) were measured. Columns. mean from five independent tumors; bars, SEM. *, P < 0.01; **, P < 0.05.

FIGURE 6.

RECK induction in vascular endothelial cells supports tumor growth. A, immunofluorescence analysis of the host mouse-derived blood vessels developed in a tumor generated by s.c. introducing 1 × 106 HT1080 cells into a 4-week-old BALB/c nu/nu mouse. Antibodies to the indicated proteins were used. Arrows, RECK+/CD31+ endothelial cells. Scale bars, 20 μm. B, induction of RECK in HUVECs cocultured with HT1080 cells. HUVECs (5 × 104) were cocultured with 2.5 (+), 5.0 (++), or 10 (+++) × 104 HT1080 cells for 24 hours (upper top). Culture medium, DMEM supplemented with 10% FCS. Endogenous RECK expression in the same number of HT1080 cells is also examined (upper bottom). Induction of RECK in 5 × 104 HUVECs cocultured with the indicated tumor cell lines (5 × 104; lower top). Endogenous RECK expression in the same number of tumor cells is also examined (lower bottom). C, a section from a human neuroblastoma obtained by biopsy was stained with H&E (top). Scale bar, 100 μm. RECK expression in a CD31+ blood vessel found in the tumor mass was detected by immunofluorescence (bottom). Scale bar, 10 μm. D, HT1080 cells and HUVECs were loaded into PLLA scaffolds and grown in SCID mice for 3 weeks. Sections of different parts of a tumor (marginal and center) were stained for the indicated proteins (top). Arrow heads, CD31+/hCD34 cells; arrows, CD31+/hCD34+ cells. HT1080 cells and HUVECs transfected with the indicated siRNA were loaded into PLLA scaffolds and grown in SCID mice for 3 weeks. Tumor sections were stained for the indicated protein and 4′,6-diamidino-2-phenylindole (DAPI) or H&E (bottom). Scale bar, 100 μm. Density of hCD34+ endothelial cells (per ×40 magnification field, 10 fields were observed; top right) and tumor weights (bottom right) were measured. Columns. mean from five independent tumors; bars, SEM. *, P < 0.01; **, P < 0.05.

Close modal

A comparison of the RECK-inducing activity exerted by various human tumor-derived cell lines revealed that RECK induction well correlates with the known angiogenic activity of each cell line; cell lines derived from hypervascular tumors (HT1080, fibrosarcoma; SKNSH, neuroblastoma; PC-12, pheochromocytoma) tend to induce higher RECK expression in cocultured HUVECs (Fig. 6B, bottom). Moreover, two of these three tumor cell lines induced a significant level of Reck promoter activation in coclutured HUVECs (Supplementary Fig. S4), suggesting that RECK induction by tumor cells is at least partially dependent on transcriptional control. MCF-7 was exceptional; this line significantly transactivated RECK promoter reporter (Fig. 4B) but only slightly upregulated RECK protein expression (Fig. 6B). Interestingly, cell lines with higher RECK-inducing activity also tend to express a detectable level of endogenous RECK (Fig. 6B, bottom), implicating that RECK could be induced in such tumors by an autocrine mechanism. However, depletion of endogenously expressed RECK in one of these RECK-positive tumor cell lines (HT1080) did not result in detectable changes in cell behavior, including proliferation, cell death, migration, and invasion (data not shown).

To examine the clinical relevance of RECK induction in vascular endothelial cells convoluted in tumors in nude mice, we analyzed tumor biopsy tissues from human neuroblastomas. As described, these tumors exhibited a high density of vasculature (Fig. 6C, top). We observed weak but significant level of RECK expression in CD31+ vascular endothelial cells found within the tumor mass (Fig. 6C, bottom). Similarly, the endothelial cells of the larger arteries located close to the tumor mass showed significant level of RECK expression. In these arteries, smooth muscle cells also express RECK as described previously (ref. 15; Supplementary Fig. S5). Although not directly addressing whether RECK is induced during neuroblastoma development, this study shows that vascular endothelial cells, at least in human neuroblastomas, express a detectable level of RECK.

RECK induction is required for vascular endothelial cells to support tumor growth

To determine the biological significance of RECK induction in vascular endothelial cells by tumor cells, we used a tumor angiogenesis model in which HUVECs and human fibrosarcoma cells (HT1080) were mixed with Matrigel and allowed to adhere to porous PLLA scaffolds. The complex was s.c. implanted into SCID mice and grown for 3 weeks.

To distinguish HUVECs from host mouse-derived vascular endothelial cells, we used an antibody that specifically reacts with hCD34. HUVECs (CD31+/hCD34+) were detected more frequently at the center of the tumors than in the marginal areas (Fig. 6D, top). The marginal areas were frequently occupied by host mouse-derived CD31+/hCD34 cells (Fig. 6D, top). The complex containing HUVECs transduced with RECK siRNA produced significantly smaller tumors compared with the controls (Fig. 6D, right). Histologic analysis revealed a poorer vasculature and a higher degree of necrosis at the center of tumors containing HUVECs transduced with RECK siRNA than those in the case of tumors containing HUVECs transduced with control siRNA (Fig. 6D, bottom right). These findings suggest that RECK induction in vascular endothelial cells by tumor cells is required for supporting tumor growth.

In this study, we showed that RECK is induced in vascular endothelial cells by a specific angiogenic factor during physiologic angiogenesis and in hypervascular tumors during pathologic angiogenesis. In addition, we characterized the proangiogenic function of RECK, which seemed to at least partially depend on MMP-2 and β1-integrin. A similar functional interaction between MMP-2 and β1-integrin signaling has been reported in previous analyses of cardiac myocytes (26) and chondrogenic cells (27). Thus far, no evidence of a direct interaction between MMP-2 and β1-integrin is available. Therefore, we currently speculate that, when RECK expression is attenuated in vascular endothelial cells, substrates of MMP-2 may participate in the negative regulation of β1-integrin activation as suggested previously (26, 27).

However, in the current study, it was not clarified why simultaneous MMP-2 depletion specifically antagonized the effects of RECK depletion on proliferation of HUVECs. It was reported that a selective MT1-MMP inhibitor almost completely blocked proMMP-2 processing in HUVECs (28). However, MT1-MMP depletion was not relevant to MMP-2 depletion in RECK-depleted HUVECs in our experiment. Because proMMP-2 is not the sole substrate of MT1-MMP (29), we speculate that MT1-MMP can exert distinct proangiogenic and antiangiogenic functions even in an MMP-2–independent manner. Thus, we further speculate that suppression of the MMP-2–independent proangiogenic (growth-stimulating) function of MT1-MMP masked the rescue effect of MT1-MMP depletion on RECK depletion–induced proliferation defects. MMP-9 depletion may have failed to rescue the RECK depletion–induced proliferation defects due to similar reason. Overall, these findings indicate bivalent roles of MMP activities in angiogenesis.

Another new insight provided by this study is that proliferation of vascular endothelial cells during physiologic and pathologic angiogenesis is critically controlled by RECK. As discussed above, MMP activities provided by vascular endothelial cells can be both proangiogenic and antiangiogenic to the cells themselves. RECK may function to protect proliferating vascular endothelial cells from self-produced antiangiogenic MMP activities as well as from those generated by tumor stromal cells or tumor cells. Specific downregulation of RECK in vascular endothelial cells decreased their contribution to tumor growth. Conversely, RECK overexpression in experimentally transplanted tumor cells reduced tumor volume by attenuating angiogenesis from host nude mice (15). These findings suggest that the effect of RECK on tumor angiogenesis varies depending on the location of expression.

RECK expression was undetectable in most of commonly available tumor cell lines by conventional immunoblotting techniques (9). By applying highly sensitive immunoblotting methods, we established that PC-12 and HT1080 show the highest endogenous RECK expression among the tumor cell lines; however, the level of RECK expressed in these cell lines is far less than that in proliferating vascular endothelial cells. In addition, depletion of endogenous RECK in HT1080 cells did not show detectable changes in cell behavior. Thus, we propose the possibility that RECK principally functions in vascular endothelial cells rather than in tumor cells and supports tumor growth by enhancing angiogenesis rather than suppressing it. This statement seems to be contradictory to studies showing positive correlation between RECK expression level in tumor tissues and favorable prognosis of cancer patients (30, 31). However, virtually none of such studies discriminated RECK expression in tumor cells, stromal cells, tumor vessels, or other cell lineages composing tumors. Furthermore, we still do not know whether RECK actively improves prognosis or the elevated RECK expression is a consequence of less malignant property of tumor cells. We recently obtained evidence that RECK promotes self-renewing activity of neural stem cells through modulation of Notch signaling (17) and inversely suppresses proliferation of colorectal cancer cells and mouse embryonic fibroblasts through another cellular signaling pathway.8

8Kitajima et al., unpublished data.

These findings indicate that, as MMPs are multifunctional, RECK is multifunctional and bivalent especially in growth control.

Finally, this study proposed that tumor cells may actively induce RECK expression in the surrounding vascular endothelial cells. Thus far, we did not detect angiopoietin-1 in HT1080 cells, suggesting that other factors may mediate RECK induction. Several previous reports indicated candidates of mediator, such as tissue inhibitor of metalloproteinase-2 or transforming growth factor-β (32, 33). However, we, thus far, obtained no evidence that these soluble factors could affect RECK expression particularly in HUVECs (data not shown). Thus, we are currently examining other possible factors controlling RECK expression in non–cell autonomous manner.

Taken together, our results indicate that tumor propagation via enhanced angiogenesis depends on the expression of RECK in proliferating vascular endothelial cells. We thereby propose that targeting RECK expressed in tumor vascular endothelial cells could be an interesting approach aiming to induce tumor regression.

No potential conflicts of interest were disclosed.

We thank M. Okabe for reagents, A. Nishimoto and H. Gu for technical assistance, and A. Miyazaki and M. Suzuki for secretarial assistance.

Grant Support: Princess Takamatsu Cancer Research Fund research grant 05-23706, Takeda Science Foundation, and Japan Ministry of Education, Culture, Sports, Science and Technology.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1
Nagase
H
,
Woessner
JF
 Jr
. 
Matrix metalloproteinases
.
J Biol Chem
1999
;
274
:
21491
4
.
2
Egeblad
M
,
Werb
Z
. 
New functions for the matrix metalloproteinases in cancer progression
.
Nat Rev Cancer
2002
;
2
:
161
74
.
3
Visse
R
,
Nagase
H
. 
Matrix metalloproteinases and tissue inhibitors of metalloproteinases: structure, function, and biochemistry
.
Circ Res
2003
;
92
:
827
39
.
4
Jodele
S
,
Blavier
L
,
Yoon
JM
,
DeClerck
YA
. 
Modifying the soil to affect the seed: role of stromal-derived matrix metalloproteinases in cancer progression
.
Cancer Metastasis Rev
2006
;
25
:
35
43
.
5
Rundhaug
JE
. 
Matrix metalloproteinases and angiogenesis
.
J Cell Mol Med
2005
;
9
:
267
85
.
6
Hynes
RO
. 
Cell-matrix adhesion in vascular development
.
J Thromb Haemost
2007
;
1
:
32
40
.
7
Brooks
PC
,
Silletti
S
,
von Schalscha
TL
,
Friedlander
M
,
Cheresh
DA
. 
Disruption of angiogenesis by PEX, a noncatalytic metalloproteinase fragment with integrin binding activity
.
Cell
1998
;
92
:
391
400
.
8
van Nimwegen
MJ
,
van de Water
B
. 
Focal adhesion kinase: a potential target in cancer therapy
.
Biochem Pharmacol
2007
;
73
:
597
609
.
9
Takahashi
C
,
Sheng
Z
,
Horan
TP
, et al
. 
Regulation of matrix metalloproteinase-9 and inhibition of tumor invasion by the membrane-anchored glycoprotein RECK
.
Proc Natl Acad Sci U S A
1998
;
95
:
13221
6
.
10
Liu
LT
,
Peng
JP
,
Chang
HC
,
Hung
WC
. 
RECK is a target of Epstein-Barr virus latent membrane protein 1
.
Oncogene
2003
;
22
:
8263
70
.
11
Liu
LT
,
Chang
HC
,
Chiang
LC
,
Hung
WC
. 
Histone deacetylase inhibitor up-regulates RECK to inhibit MMP-2 activation and cancer cell invasion
.
Cancer Res
2003
;
63
:
3069
72
.
12
Hu
SJ
,
Ren
G
,
Liu
JL
, et al
. 
MicroRNA expression and regulation in mouse uterus during embryo implantation
.
J Biol Chem
2008
;
283
:
23473
84
.
13
Gabriely
G
,
Wurdinger
T
,
Kesari
S
, et al
. 
MicroRNA 21 promotes gliomainvasion by targeting matrix metalloproteinase regulators
.
Mol Cell Biol
2008
;
28
:
5369
80
.
14
Zhang
Z
,
Li
Z
,
Gao
C
, et al
. 
miR-21 plays a pivotal role in gastric cancer pathogenesis and progression
.
Lab Invest
2008
;
88
:
1358
66
.
15
Oh
J
,
Takahashi
R
,
Kondo
S
, et al
. 
The membrane-anchored MMP inhibitor RECK is a key regulator of extracellular matrix integrity and angiogenesis
.
Cell
2001
;
107
:
789
800
.
16
Miki
T
,
Takegami
Y
,
Okawa
K
,
Muraguchi
T
,
Noda
M
,
Takahashi
C
. 
The reversion-inducing cysteine-rich protein with Kazal motifs (RECK) interacts with membrane type 1 matrix metalloproteinase and CD13/aminopeptidase N and modulates their endocytic pathways
.
J Biol Chem
2007
;
282
:
12341
52
.
17
Muraguchi
T
,
Takegami
Y
,
Ohtsuka
T
, et al
. 
RECK modulates Notch signaling during cortical neurogenesis by regulating ADAM10 activity
.
Nat Neurosci
2007
;
10
:
838
45
.
18
Baker
AH
,
Edwards
DR
,
Murphy
G
. 
Metalloproteinase inhibitors: biological actions and therapeutic opportunities
.
J Cell Sci
2002
;
115
:
3719
27
.
19
Wayner
EA
,
Gil
SG
,
Murphy
GF
,
Wilke
MS
,
Carter
WG
. 
Epiligrin, a component of epithelial basement membranes, is an adhesive ligand for α 3 β 1 positive T lymphocytes
.
J Cell Biol
1993
;
121
:
1141
52
.
20
Okabe
M
,
Ikawa
M
,
Kominami
K
,
Nakanishi
T
,
Nishimune
Y
. 
“Green mice” as a source of ubiquitous green cells
.
FEBS Lett
1997
;
407
:
313
9
.
21
Shamma
A
,
Takegami
Y
,
Miki
T
, et al
. 
Rb Regulates DNA damage response and cellular senescence through E2F-dependent suppression of N-ras isoprenylation
.
Cancer Cell
2009
;
15
:
255
69
.
22
Sasahara
RM
,
Takahashi
C
,
Noda
M
. 
Involvement of the Sp1 site in ras-mediated downregulation of the RECK metastasis suppressor gene
.
Biochem Biophys Res Commun
1999
;
264
:
668
75
.
23
Takahashi
C
,
Bronson
RT
,
Socolovsky
M
, et al
. 
Rb and N-ras function together to control differentiation in the mouse
.
Mol Cell Biol
2003
;
23
:
5256
68
.
24
Nor
JE
,
Peters
MC
,
Christensen
JB
, et al
. 
Engineering and characterization of functional human microvessels in immunodeficient mice
.
Lab Invest
2001
;
81
:
453
63
.
25
Bryant
P
,
Zheng
Q
,
Pumiglia
K
. 
Focal adhesion kinase controls cellularlevels of p27/Kip1 and p21/Cip1 through Skp2-dependent and -independent mechanisms
.
Mol Cell Biol
2006
;
26
:
4201
13
.
26
Menon
B
,
Singh
M
,
Ross
RS
,
Johnson
JN
,
Singh
K
. 
β-Adrenergic receptor-stimulated apoptosis in adult cardiac myocytes involves MMP-2-mediated disruption of β1 integrin signaling and mitochondrial pathway
.
Am J Physiol Cell Physiol
2006
;
290
:
254
61
.
27
Jin
EJ
,
Choi
YA
,
Kyun Park
E
,
Bang
OS
,
Kang
SS
. 
MMP-2 functions as a negative regulator of chondrogenic cell condensation via down-regulation of the FAK-integrin β1 interaction
.
Dev Biol
2007
;
308
:
474
84
.
28
Devy
L
,
Huang
L
,
Naa
L
, et al
. 
Selective inhibition of matrix metalloproteinase-14 blocks tumor growth, invasion, and angiogenesis
.
Cancer Res
2009
;
69
:
1517
26
.
29
Itoh
Y
,
Seiki
M
. 
MT1-MMP: a potent modifier of pericellular microenvironment
.
J Cell Physiol
2006
;
206
:
1
8
.
30
Noda
M
,
Takahashi
C
. 
Recklessness as a hallmark of aggressive cancer
.
Cancer Sci
2007
;
98
:
1659
65
.
31
Clark
JC
,
Thomas
DM
,
Choong
PF
, et al
. 
RECK-a newly discovered inhibitor of metastasis with prognostic significance in multiple forms of cancer
.
Cancer Metastasis Rev
2007
;
26
:
675
83
.
32
Lee
H
,
Lim
C
,
Lee
J
, et al
. 
TGF-β signaling preserves RECK expression in activated pancreatic stellate cells
.
J Cell Biochem
2008
;
104
:
1065
74
.
33
Oh
J
,
Diaz
T
,
Wei
B
, et al
. 
TIMP-2 upregulates RECK expression via dephosphorylation of paxillin tyrosine residues 31 and 118
.
Oncogene
2006
;
25
:
4230
4
.

Supplementary data