Membrane Type 1-Matrix Metalloproteinase Is Regulated by Chemokines Monocyte-Chemoattractant Protein-1/CCL2 and Interleukin-8/CXCL8 in Endothelial Cells during Angiogenesis*

We have investigated the putative role and regulation of membrane type 1-matrix metalloproteinase (MT1-MMP) in angiogenesis induced by inflammatory factors of the chemokine family. The absence of MT1-MMP from null mice or derived mouse lung endothelial cells or the blockade of its activity with inhibitory antibodies resulted in the specific decrease of in vivo and in vitro angiogenesis induced by CCL2 but not CXCL12. Similarly, CCL2- and CXCL8-induced tube formation by human endothelial cells (ECs) was highly dependent on MT1-MMP activity. CCL2 and CXCL8 significantly increased MT1-MMP surface expression, clustering, activity, and function in human ECs. Investigation of the signaling pathways involved in chemokine-induced MT1-MMP activity in ECs revealed that CCL2 and CXCL8 induced cortical actin polymerization and sustained activation of phosphatidylinositol 3-kinase (PI3K) and the small GTPase Rac. Inhibition of PI3K or actin polymerization impaired CCL2-induced MT1-MMP activity. Finally, dimerization of MT1-MMP was found to be enhanced by CCL2 in ECs in a PI3K- and actin polymerization-dependent manner. In summary, we identify MT1-MMP as a molecular target preferentially involved in angiogenesis mediated by CCL2 and CXCL8, but not CXCL12, and suggest that MT1-MMP dimerization might be an important mechanism of its regulation during angiogenesis.

We have investigated the putative role and regulation of membrane type 1-matrix metalloproteinase (MT1-MMP) in angiogenesis induced by inflammatory factors of the chemokine family. The absence of MT1-MMP from null mice or derived mouse lung endothelial cells or the blockade of its activity with inhibitory antibodies resulted in the specific decrease of in vivo and in vitro angiogenesis induced by CCL2 but not CXCL12. Similarly, CCL2-and CXCL8-induced tube formation by human endothelial cells (ECs) was highly dependent on MT1-MMP activity. CCL2 and CXCL8 significantly increased MT1-MMP surface expression, clustering, activity, and function in human ECs. Investigation of the signaling pathways involved in chemokine-induced MT1-MMP activity in ECs revealed that CCL2 and CXCL8 induced cortical actin polymerization and sustained activation of phosphatidylinositol 3-kinase (PI3K) and the small GTPase Rac. Inhibition of PI3K or actin polymerization impaired CCL2-induced MT1-MMP activity. Finally, dimerization of MT1-MMP was found to be enhanced by CCL2 in ECs in a PI3K-and actin polymerization-dependent manner. In summary, we identify MT1-MMP as a molecular target preferentially involved in angiogenesis mediated by CCL2 and CXCL8, but not CXCL12, and suggest that MT1-MMP dimerization might be an important mechanism of its regulation during angiogenesis.
Angiogenesis is a key event both during physiological and pathological processes and is regulated by factors from the VEGF 1 and fibroblast growth factor families as well as by cytokines and chemokines (1)(2)(3)(4). Chemokines are a family of small polypeptides classified into different subfamilies (C, CC, CXC, and CX3C) and were first described as regulators of leukocyte homing (5). The presence of an ELR (glutamyl-leucylarginyl) motif in CXC chemokines has been correlated with their angiogenic activity (4). Thus, interleukin-8/CXCL8 (CXC, ELRϩ) is angiogenic (6), whereas monokine induced by interferon-␥ (MIG)/CXCL9 (CXC, ELRϪ) is angiostatic (7). However, other CXC/ELRϪ chemokines such as stromal-derived factor-1/CXCL12 as well as chemokines from other subfamilies such as monocyte-chemoattractant protein-1/CCL2 are also angiogenic (3). Receptors for these chemokines have been reported in ECs (8 -11), but the molecular mechanisms by which they induce or inhibit angiogenesis remain undetermined.
During the angiogenic response, ECs must degrade the basement membrane and the interstitial matrix underneath. MT1-MMP is a membrane-anchored matrix metalloproteinase (MMP) that can process several extracellular matrix components as well as cell surface receptors (12). MT1-MMP activity is regulated in ECs by modulation of its subcellular localization, internalization, and caveolae-mediated traffic, and it has been implicated in endothelial migration and angiogenesis both in vitro and in vivo (13)(14)(15)(16)(17). Because it has been shown that cytokines and chemokines can affect MMP expression and/or activity in several cell types (9,18,19), we asked whether MT1-MMP might be a molecular effector of chemokines during angiogenesis.
In this report, we identify MT1-MMP as a preferential downstream effector of CCL2 and CXCL8 but not CXCL12 during induction of angiogenesis in vivo and in vitro. We further show that these chemokines modulate MT1-MMP dimerization and activity in human ECs and that this action is dependent on the induction of cortical actin polymerization and PI3K activity.
Human EC Culture-Human ECs from umbilical vein were isolated and cultured up to the third passage as described (13). ECs were seeded on dishes coated with 1% GEL and changed to HE-SFM serum-free medium from Invitrogen before performing the functional assays.
MLEC Isolation-Mice deficient in MT1-MMP have been characterized previously (15). Lungs from C57BL/6 wild type (wt), heterozygous (het), or null mice were excised, collagenase-digested, and further disaggregated to produce a single cell suspension. The mixed population obtained was first subjected to negative selection with anti-mouse CD16 mAb and then to positive selection with anti-mouse ICAM-2 mAb and anti-IgG-coated magnetic beads. The endothelial population was Ͼ90% pure as assessed by anti-mouse CD31 staining. MLECs were grown on a mixture of fibronectin, type I collagen, and GEL-coated plates with low glucose Dulbecco's modified Eagle's medium plus Ham's F-12 and growth factors and were used up to the fourth passage. MT1-MMP expression was monitored in MLECs from different genotypes by Western blot. 2 Flow Cytometry and Analysis of Receptor Internalization-Subconfluent ECs were grown on GEL and treated with or without the indicated chemokines for 6 h. Cell surface staining and receptor internalization were analyzed by flow cytometry as described (16). Briefly, receptor internalization was assessed by incubating ECs with the primary mAb at RT for 20 min, exhaustively washing with phosphatebuffered saline, and transferring to 37°C for 6 h to allow labeledreceptor internalization. Half of the cells were then fixed and permeabilized for 10 min at 4°C with lysis buffer from BD Biosciences, and all cells were processed for flow cytometry. The fluorescence of permeabilized and non-permeabilized samples (MFI) was detected on a linear scale for optimizing quantitation and analyzed at 0 and 6 h. Internalization percentages at 6 h were estimated with the following equation. Immunofluorescence Microscopy-Immunofluorescence staining with anti-MT1-MMP LEM-2/15 mAb was performed after fixation in subconfluent ECs on GEL-coated coverslips incubated with or without different chemokines for 6 h as described (13). To visualize polymerized F-actin, cells were incubated with fluorescein isothiocyanate-phalloidin after stimulation with chemokines and fixation. Cells were examined under a Leica DMR photomicroscope with a 63ϫ oil immersion objective and photographed with a Leica CCD camera.
Rac Small GTPase Activity Assays-Glutathione S-transferase-p21activated kinase, which is recognized by active Rac, was prepared as described (20). ECs grown at subconfluence and stimulated with CCL2, CXCL12, or sphingosine 1-phosphate for 15 min or 6 h were lysed, and pull-down experiments were performed as described (20).
Zymography and Western Blotting-Subconfluent ECs were grown on GEL and incubated with or without different chemokines in serumfree medium for 6 h. Cell lysates were analyzed by fibrinogen or gelatin zymography or by Western blot with anti-MT1-MMP LEM-2/15 mAb or a mixture of anti-phospho-AKT Ser-473 and Thr-308 Abs as described (13). To test MT1-MMP dimer formation, EC membranes were extracted in the hydrophobic fraction of 1% Triton X-114 lysates and resolved on 10% SDS-PAGE under strict non-reducing conditions or under reducing conditions.
Ca 2ϩ Measurement-Subconfluent ECs cultured in HE-SFM for 24 h on GEL-coated 96-well plates (Costar Corp.) were washed twice and then incubated with 1 g/ml of FLUO-4 (Molecular Probes) in Hanks' balanced salt solution and 0.02% Pluronic F-127 for 1 h at 37°C. After washing with Hanks' balanced salt solution cells were stimulated with different chemokines, and intracellular Ca 2ϩ concentration was measured at 1-min intervals with the DeltaScan Illumination System and Felix software (Photon Technology International).
F-actin Polymerization Measurement-Subconfluent ECs were incubated with or without different chemokines in HE-SFM medium for 6 h at 37°C. Intracellular actin polymerization was stopped by the addition of an equal volume of phosphate-buffered saline containing 4% paraformaldehyde, 1% Triton X-100, and 5 g/ml fluorescein isothiocyanate-conjugated phalloidin (Molecular Probes). Cells were then incubated for 30 min at 37°C, detached with cell dissociation buffer (Invitrogen), and resuspended in phosphate-buffered saline. Intracellular polymerized actin was then measured by flow cytometry in a FAC-Scalibur® (BD Biosciences).
Cell Transmigration and in Vitro Angiogenesis Assays-Human EC or MLEC transmigration assays were performed in 8-m-pore Transwell chambers (Costar) in the presence of different chemokines in the lower chamber. Matrigel assays were performed with 4 ϫ 10 4 MLEC or human ECs added atop the Matrigel in the presence of different chemokines as described (13).
Matrigel Plug Assay for Angiogenesis in Vivo-Female 12-week-old Balb/c or 1-week-old MT1-MMP wt, het, or null mice of the C57BL/6 strain were injected subcutaneously in the ventral midline using a 30-gauge needle with 0.5 or 0.25 ml of Matrigel, respectively, combined with different factors. Five to 10 days later mice were sacrificed, and the Matrigel plugs were processed for hemoglobin determination with the 3,3Ј,5,5Ј-tetramethylbenzidine liquid substrate system TMB (Sigma). A portion of the plug was also frozen, cut into 10-m sections, and stained with hematoxylin alone or together with anti-mouse CD31 mAb to assess tube formation. Housing and procedures involving experimental animals were approved by the institutional animal care of the Cleveland Clinic Foundation.
Statistical Analysis-Test and control samples in the functional assays were compared for statistical significance by using Student's t test.

MT1-MMP Is Required for CCL2-but Not CXCL12-induced
Murine Angiogenesis in Vivo and in Vitro-Angiogenesis was induced in Balb/c mice by subcutaneous injection of Matrigel premixed with angiogenic factors and blocking or control antibodies as indicated (Fig. 1). CCL2-induced hemoglobin accumulation was significantly inhibited by the presence of the anti-MT1-MMP LEM-2/63 mAb, which recognizes the murine protease, whereas CXCL12-or VEGF-induced angiogenesis was unaffected; CXCL9 did not alter constitutive angiogenesis (Fig. 1A). In agreement with this, tube formation was induced by chemokines CCL2 and CXCL12, but only those capillarylike tubes generated by CCL2 were sensitive to inhibition by the anti-MT1-MMP mAb, as demonstrated by staining Matrigel sections with hematoxylin alone or together with anti-CD31 mAb (Fig. 1B). 2 Angiogenesis was next analyzed in MT1-MMP-deficient mice (15). CCL2-induced angiogenesis was significantly inhibited in MT1-MMP null mice compared with wt or het mice, whereas CXCL12-induced angiogenesis developed similarly in wt, het, and null mice ( Fig. 2A). Staining of frozen Matrigel sections with hematoxylin alone or with anti-CD31 mAb also revealed that no capillary-like tubes were formed in the presence of CCL2 in MT1-MMP null mice, in contrast to wt mice (Fig. 2B). Similarly, cord formation induced by CCL2 was abrogated in MT1-MMP null MLECs and diminished in het MLECs, whereas the absence of MT1-MMP did not have any effect on CXCL12-induced cord formation (Fig. 2C).
MTI-MMP Is Required for Tube Formation by Human ECs Induced by CCL2 and CXCL8, and Its Expression and Internalization Are Regulated by These Chemokines-We next showed by flow cytometry that the receptors for CCL2, CXCL8, CXCL12, and CXCL9 were expressed at the surface of human ECs. The MFI values for the corresponding receptors were, respectively: CCR2 ϭ 25; CXCR1 ϭ 22 and CXCR2 ϭ 27; CXCR4 ϭ 28; CXCR3 ϭ 31. The negative value was 12. In addition, the anti-MT1-MMP LEM-2/15 mAb significantly inhibited cord formation in human ECs, by 70% in ECs stimulated with 10 nM CCL2 or CXCL8 and by 20% in ECs stimulated with CXCL12 (Fig. 3A). To determine whether the distinct ability of the anti-MT1-MMP mAb to impair chemokine-induced cord formation was related to the angiogenic potency of the chemokines, a dose response cord formation to CCL2 and CXCL12 was analyzed in the presence or absence of the inhibitory anti-MT1-MMP mAb. As shown in Fig. 3B, the anti-MT1-MMP mAb efficiently inhibited cord formation in response to CCL2 (from 10 -100 nM), in contrast to only slight inhibition in response to CXCL12 (from 1 to 100 nM).
We next showed that both CCL2 and CXCL8 increase MT1-MMP surface expression in human ECs stimulated for 6 h, in contrast to CXCL12 or CXCL9 (Fig. 3C). However, no changes were observed in the amount of total MT1-MMP protein, as analyzed by Western blot of whole cell lysates, nor in mRNA levels assessed by quantitative reverse transcription-PCR (Fig.  3, D and E). Because MT1-MMP levels at the EC surface can be regulated by internalization (16), we next evaluated this. MT1-MMP internalization, close to 100% in untreated subconfluent ECs, was significantly decreased in ECs stimulated for 6 h with CCL2 or CXCL8 but not CXCL12 or CXCL9 (Fig. 3F).
MT1-MMP Subcellular Localization, Activity, and Function Are Modulated by CCL2 and CXCL8 in Human ECs-The effect of the distinct chemokines in modulating the subcellular localization of MT1-MMP was then investigated. The number of MT1-MMP clusters present at motility-associated membrane protrusions of subconfluent ECs was significantly increased upon stimulation for 6 h with CCL2 or CXCL8 (2-3-fold) and with CXCL12 (1.5-fold); no changes were observed with CXCL9 (Fig. 4A).
Because both the amount and clustering of MT1-MMP at the cell surface were up-regulated by CCL2 and CXCL8, we measured MT1-MMP activity under these conditions. MT1-MMP activity was significantly up-regulated by CCL2 or CXCL8 but not by CXCL12 or CXCL9 in subconfluent ECs, as assessed by fibrinogen zymography (Fig. 4B, top). To measure MT1-MMP activity in an independent manner, MMP-2 processing was estimated by gelatin zymography. As shown in Fig. 4B, bottom, the percentage of active MMP-2 was significantly increased in ECs stimulated with CCL2 or CXCL8, in contrast to CXCL12. These findings correlated with endothelial migration assays, in

FIG. 2. The absence of MT1-MMP inhibits CCL2-but not CXCL12-induced angiogenesis in mice and MLECs.
A, Matrigel was mixed with 50 ng/ml VEGF or 100 nM CCL2 or CXCL12 and injected into 1-week old MT1-MMP wt (ϩ/ϩ), het (ϩ/Ϫ), or null (Ϫ/Ϫ) C57BL/6 mice. Hemoglobin (Hb) quantification was performed on plugs removed after 5 days. The arithmetic means and S.D. of five mice for CCL2, four mice for CXCL12 and VEGF, and three mice for Matrigel alone are shown. *, p Ͻ 0.01. B, frozen Matrigel sections from control plugs or plugs containing 100 nM CCL2 in wt or null mice were stained with hematoxylin or immunostained with anti-mouse CD31 mAb as indicated. In the latter case, the L and arrowhead mark the tube lumina and surrounding endothelial monolayer, respectively. Original magnification 450ϫ and 1350ϫ. C, MLECs from MT1-MMP wt (ϩ/ϩ), het (ϩ/Ϫ), or null (Ϫ/Ϫ) mice were left untreated or were treated with 10 ng/ml VEGF or 10 nM CCL2 or CXCL12 and seeded onto Matrigel. Formation of cords was quantitated after 6 h as described under "Experimental Procedures." *, p Ͻ 0. 04; ϩ, p Ͻ 0.02. which the anti-MT1-MMP LEM-2/15 mAb inhibited migration toward CCL2 and CXCL8 by about 80 and 50%, respectively. Migration toward CXCL12 was only inhibited by 10% (Fig. 4C). In this regard, MLECs obtained from MT1-MMP null mice also migrated less efficiently toward CCL2 when compared with wt or het MLECs but migrated normally toward CXCL12 (Fig.  4D). Therefore, the dependence of CCL2-and CXCL8-mediated tube formation on MT1-MMP might primarily be mediated through an increase of MT1-MMP expression and clustering at the cell surface.
CCL2 and CXCL8 Induce Sustained PI3K and Rac GTPase Activation and Cortical Actin Polymerization in Human ECs-Next, we aimed at elucidating the signaling pathways that might mediate up-regulation of MT1-MMP activity by chemokines. We first confirmed that all chemokines were able to similarly mobilize Ca 2ϩ , indicating no defects in their signaling. 2 Because chemokines can induce a variety of intracellular signals, the effects of distinct inhibitors were tested on CCL2induced MT1-MMP activity. As shown in Fig. 5, A and B, inhibition of G i protein with Ptx, of PI3K activity with WMN, and of actin polymerization with CytD and/or LAT resulted in significant decreases in CCL2-induced MT1-MMP activity, as assessed by fibrinogen or gelatin zymography. However, JPK, which stabilizes actin filaments, the Ca 2ϩ chelator EDTA, and the RhoA inhibitor C3 had no inhibitory effect (Fig. 5, A and B). Thus, PI3K and actin polymerization seemed to positively regulate MT1-MMP activity in ECs.
To directly investigate this point we analyzed the phosphorylation of the PI3K effector AKT in subconfluent ECs; AKT was phosphorylated after 6 h of stimulation with CCL2 and CXCL8 but not with CXCL12 (Fig. 5C). Next, F-actin polymerization was investigated by fluorescence microscopy and flow cytometry. Polymerized F-actin was mainly observed in the cortical area, and its amount was significantly increased in ECs treated with CCL2 or CXCL8 for 6 h compared with untreated or CXCL12-treated ECs, in which F-actin was homogenously distributed throughout the cells (Fig. 5C). 2 Rac can regulate actin polymerization (21) and be activated by PI3K (22). Rac activation was observed in subconfluent ECs after 6 h of stimulation with CCL2 or the activator sphingosine 1-phosphate but not with CXCL12 (Fig. 5D). In other cell types shorter exposure to CXCL12 can activate Rac (23), and in agreement with this, we observed Rac activation in ECs stimulated for 15 min with either CCL2 or CXCL12 (Fig. 5D).

CCL2 and CXCL8 Enhance MT1-MMP Dimerization in Human ECs in a Mechanism Dependent on PI3K Activity and
Actin Polymerization-We next investigated putative mechanisms by which CCL2-induced actin polymerization might result in increased MT1-MMP clustering and activity. Because actin polymerization regulates mobilization of receptors at the cell membrane (24 -26), MT1-MMP oligomerization was first investigated in ECs. A 120-kDa band was detected by Western blot with the anti-MT1-MMP mAb in non-reduced membrane fractions. Upon reduction, this band migrated at 60 kDa, strongly suggesting that the former corresponds to MT1-MMP homodimers (Fig. 6A). Remarkably, stimulation of ECs for 6 h with CCL2 or CXCL8 increased the amount of MT1-MMP dimers at the cell membrane by 4 -5-fold, in contrast to a lesser or no increase with CXCL12 or CXCL9 (Fig. 6A). Neither tive reverse transcription-PCR. The arithmetic mean and S.D. of the relative units of MT1-MMP mRNA obtained in three independent experiments are shown. F, internalization of MT1-MMP was quantitated by flow cytometry in subconfluent human ECs on GEL untreated or treated with 10 nM CCL2, CXCL8, CXCL12, or CXCL9 for 6 h. The arithmetic means and S.D. of the internalization percentages of three independent experiments are shown. *, p Ͻ 0.03.

FIG. 3. Role of MT1-MMP in chemokine-induced cord formation by human ECs. Regulation of its expression and internalization.
A, human ECs (HUVEC) were left untreated or were treated with 10 nM CCL2, CXCL8, CXCL12, or CXCL9 in the presence or absence of 10 g/ml anti-MT1-MMP LEM-2/15 mAb or control isotypematched mAb and seeded on Matrigel. Formation of cords was quantitated after 6 h as described under "Experimental Procedures." *, p Ͻ 0.02; **, p Ͻ 0.03; ϩ, p Ͻ 0.05. B, a dose response of CCL2 and CXCL12 was performed in the absence or presence of 10 g/ml anti-MT1-MMP LEM-2/15 mAb in human ECs on Matrigel. Formation of cords was quantitated after 6 h as described under "Experimental Procedures." The mean of two independent experiments is shown. C, the expression of MT1-MMP at the cell surface was analyzed by flow cytometry in subconfluent ECs on GEL untreated or treated with 10 nM CCL2, CXCL8, CXCL12, or CXCL9 for 6 h. X63 was included as a negative control. MFI is also indicated. A representative experiment of four conducted is shown. D, MT1-MMP total protein was assessed by Western blot analysis of cell lysates from subconfluent ECs on GEL untreated or treated with 10 nM CCL2, CXCL8, CXCL12, or CXCL9 for 6 h. Vascular endothelial (VE)-cadherin expression was used as loading control. One of five representative experiments is shown. E, MT1-MMP mRNA levels in human ECs on GEL untreated or treated with 10 nM CCL2, CXCL8, CXCL12, or CXCL9 for 6 h were analyzed by quantita-MMP-2 nor tissue inhibitor of metalloproteinase-2 seemed to be present at the 120 kDa band. 2 To explore whether signaling pathways involved in the CCL2-induced up-regulation of MT1-MMP activity (Fig. 5, A  and B) were also implicated in MT1-MMP dimerization, signaling inhibitors were tested. Ptx, WMN, CytD, and LAT interfered with CCL2-induced MT1-MMP dimerization at the EC surface, demonstrating that PI3K activity and actin polymerization are required for this process (Fig. 6B) and suggesting that dimerization is related to MT1-MMP proteolytic activity. No effect on dimerization was observed with JPK, EDTA, or C3 (Fig. 6B).

DISCUSSION
In this report we show a preferential involvement of MT1-MMP in the angiogenic response induced by CCL2 and CXCL8, but not by CXCL12, that might be relevant to certain inflammatory diseases. In addition, we characterize the signaling pathways involved in the up-regulation of MT1-MMP activity by CCL2 in ECs, which include activation of PI3K and Rac, and actin polymerization. Finally, we show that MT1-MMP dimerization in primary ECs is regulated by CCL2-induced signaling.
Angiogenesis can be induced in chronic inflammatory disease by the prolonged exposure of ECs to cytokines and chemokines, but the molecular mechanisms underlying this response have not been elucidated yet (27)(28)(29). In this regard, CXCL12 and CXCL8 have been shown to favor angiogenesis by increasing EC expression of VEGF and MMP, respectively (9,30). MT1-MMP is implicated in endothelial migration, invasion, and in vitro formation of capillary-like tubes (13); we therefore investigated its putative role in chemokine-induced angiogenesis. Remarkably, the absence of MT1-MMP in null mice or MLECs derived from them and the blockade of MT1-MMP activity with inhibitory antibodies all resulted in the specific decrease of in vivo and in vitro angiogenesis induced by CCL2 but not CXCL12. In this regard, a defective angiogenic response to fibroblast growth factor-2 in MT1-MMP null mice has been reported (15). We also report a preferential involvement of MT1-MMP in CCL2-and CXCL8-versus CXCL12-induced angiogenesis in human ECs. Our data show that CCL2 and CXCL8, but not CXCL12, increased MT1-MMP surface expression in ECs similar to other angiogenic factors and inflammatory cytokines (31,32), in this case by partially inhibiting MT1-MMP internalization. In addition, CCL2 and CXCL8 increased MT1-MMP clustering at endothelial membrane protrusions, and this correlated with an increase in MT1-MMP activity. In contrast to previous reports linking MT1-MMP internalization with its activity (16,33,34), a partial inhibition of its internalization by CCL2 and CXCL8 resulted in an increase of its clustering and activity, suggesting that MT1-MMP internalization is not the primary regulatory event. MT1-MMP and/or MMP-2 might be important during the angiogenic response induced by CCL2, since both proteolytic activities are up-regulated, as shown by fibrinogen and gelatin zymographies.
Occupancy of G protein-coupled receptors by chemokines results in mobilization of intracellular calcium as well activation of PI3K and its effectors (protein kinase C, AKT, Ras, mitogen-activated protein kinase, and small GTPases) and the JAK/STAT (signal transducers and activators of transcription) pathway (35). So far, only CCL2-induced activation of mitogenactivated protein kinase has been shown in ECs (36). Herein, we describe that although CCL2, CXCL8, and CXCL12 quickly mobilize calcium in human ECs, after 6 h only CCL2 and CXCL8 induce the activation of PI3K and Rac and the polymerization of cortical actin. However, CXCL12 did induce Rac activation in ECs at shorter times, as reported in leukocytes (23). Moreover, CXCL12 can also regulate GTPases and MT1-MMP activity in melanoma cells (37), pointing to cell typespecific events. Distinct signal responses to CCL2 and CXCL8 versus CXCL12 in ECs might be due to differences in the maintenance of signaling related to down-modulation or compartmentalization of the different chemokine receptors. In this regard CCR2 is internalized through caveolae in brain ECs (38). CXCL12 might also trigger alternative signaling pathways in ECs, leading to the expression of other genes relevant to angiogenesis such as VEGF, as reported (30). Our data would suggest that inflammatory signals such as CCL2 or CXCL8 might trigger distinct molecular pathways during angiogenesis compared with homeostatic signals such as CXCL12.
Because  polymerization, which can modulate lateral mobility of membrane receptors (24 -26), we examined the putative oligomerization of MT1-MMP in ECs. Interestingly, CCL2 and CXCL8 significantly increased the amount of MT1-MMP dimers. MT1-MMP oligomerization in transformed cells can be mediated by the hemopexin and/or cytoplasmic domains and, in accordance with our data, can be modulated by Rac GTPase (39 -41). Here, we provide the first demonstration of a physiological mechanism for the regulation of MT1-MMP dimer formation by specific angiogenic chemokines in primary ECs. Signaling events required for MT1-MMP dimer regulation have also been analyzed. The requirement of CCL2-enhanced MT1-MMP dimerization on the activation of PI3K is in agreement with its role in chemokine-induced lateral mobility of ␤ 2 -integrins (42). Moreover, actin polymerization is also required for dimer formation. Thus, activation of PI3K by CCL2 might induce sustained cortical actin polymerization that would eventually favor MT1-MMP lateral mobility likely through caveolae. In this regard, RNA interference of caveolin-1 also impaired MT1-MMP dimer formation 2 in accordance with the reported role of caveolin-1 in MT1-MMP traffic (17). MT1-MMP oligomerization has been proposed to increase its activity in transfected cells (39 -41). In our model signaling inhibitors that interfere with MT1-MMP dimer formation also impair CCL2-induced MT1-MMP activity, which strongly suggests that dimerization of MT1-MMP positively regulates its activity in ECs. Pharmacological intervention in MT1-MMP dimerization might be a future therapeutic approach for targeting angiogenesis in chronic inflammatory diseases.