Interaction of the coiled-coil domain with ... - The FASEB Journal

The FASEB Journal • Research Communication

Interaction of the coiled-coil domain with glycosaminoglycans protects angiopoietin-like 4 from proteolysis and regulates its antiangiogenic activity Cle´mence Chomel,*,† Aure´lie Cazes,*,‡,¶ Cle´ment Faye,储 Marine Bignon,*,† Elisa Gomez,*,† Corinne Ardidie-Robouant,*,† Alain Barret,*,† Sylvie Ricard-Blum,储 Laurent Muller,*,† Ste´phane Germain,*,†,§ and Catherine Monnot*,†,1 *INSERM U833, Paris, France; †College de France, Experimental Medicine Unit, Paris, France; ‡ Service d’Anatomie Pathologique and §Service d’Hematologie Biologique A, Hopital Europeen Georges Pompidou, Paris, France; 储Institut de Biologie et Chimie des Proteines, UMR CNRS 5086, Universite´ Claude Bernard Lyon 1, Lyon, France; and ¶Faculte´ de Me´decine Paris Descartes, Paris, France Angiopoietin-like 4 (ANGPTL4) is involved in angiogenesis and lipid metabolism. It is secreted by liver and adipose tissues and cleaved to generate circulating coiled-coil domain (CCD) and fibrinogen-like domain (FLD) fragments. The full-length ANGPTL4 produced by hypoxic endothelial cells interacts with the extracellular matrix (ECM). The ECMbound and soluble forms of ANGPTL4 have antiangiogenic properties. We carried out a structure-function analysis to investigate the regulation of ANGPTL4 bioactivity in endothelial cells. We found that the recombinant CCD binds to the ECM, whereas the FLD is released into the medium. The CCD, like the fulllength ANGPTL4, binds to heparan and dermatan sulfates in surface plasmon resonance assays and inhibits endothelial cell adhesion, motility, and tubule-like formation. In endothelial cells, ANGPTL4 is processed in the secretion medium after release from the ECM. This processing is altered by the proprotein convertases inhibitor ␣1-PDX and abolished by the mutation of the 161RRKR164 cleavage site without modification of the ECM binding and release. These data suggest that the full-length form, which interacts with heparan sulfate proteoglycans via its CCD, is protected from proteolysis by proprotein convertases and constitutes the major active pool of ANGPTL4 in hypoxic endothelial cells.—Chomel, C., Cazes, A., Faye, C., Bignon, M., Gomez, E., ArdidieRobouant, C., Barret, A., Ricard-Blum, S., Muller, L., Germain, S., Monnot, C. Interaction of the coiled-coil domain with glycosaminoglycans protects angiopoietinlike 4 from proteolysis and regulates its antiangiogenic activity. FASEB J. 23, 940 –949 (2009) ABSTRACT

Key Words: angiogenesis 䡠 extracellular matrix 䡠 endothelial cell 䡠 hypoxia

angiogenic process. As the vascular basement membrane is degraded, ECs migrate and proliferate in direct contact with interstitial provisional ECM (1). Several molecules, including members of the angiopoietin family, can interact with the ECM and finely tune angiogenesis in a context-dependent manner. A family of proteins structurally similar to angiopoietins, the angiopoietin-like proteins (ANGPTLs), has been identified over the past decade. We have previously shown that angiopoietin-like 4 (ANGPTL4) is induced by hypoxia in ECs (2) and in human ischemic tissues from patients with peripheral artery disease (3). ANGPTL4 has been shown to play a role in regulating lipid metabolism, but its involvement in vascular processes remains unclear (4). In vitro, ECM-bound ANGPTL4 inhibits EC adhesion, migration, and sprouting (5), and soluble ANGPTL4 promotes (6) or inhibits (7, 8) tubule-like structures. In vivo, ANGPTL4 inhibits angiogenesis and vascular leakiness (7) and prevents the metastasis of xenografted melanoma cells by inhibiting tumor cell motility and invasiveness (9), but it mediates breast cancer cell metastasis to the lung in a transforming growth factor-␤-dependent manner by altering endothelial integrity (10). Human ANGPTL4 consists of 406 amino acids. Its protein structure is common to the angiopoietins, with a signal peptide directing secretion, an amino-terminal coiled-coil domain (CCD), a linker, and a carboxyterminal fibrinogen-like domain (FLD; ref. 11). Angiopoietins differ in their ability to bind to the ECM. Angiopoietin-1 (Ang-1) is incorporated into the ECM via its linker (12); angiopoietin-3 (Ang-3) binds to heparan sulfate proteoglycans (HSPGs) via its CCD (13); and angiopoietin-2 (Ang-2) is not associated with the ECM (12). Angiopoietins regulate angiogenesis 1

Hypoxia triggers vessel growth through angiogenesis. Endothelial cells (ECs) and the extracellular matrix (ECM) play major roles in the early stages of the 940

Correspondence: INSERM U833, College de France, 11 place Marcelin Berthelot, 75005 Paris, France. E-mail: catherine. [email protected] doi: 10.1096/fj.08-115170 0892-6638/09/0023-0940 © FASEB

through multiple mechanisms, according to their ability to bind to ECM, as the binding of Ang-3 to HSPG at the cell surface is required for the antiangiogenic and antimetastasis activity of this molecule (14). Microenvironment-dependent interactions of ANGPTL4 with the ECM may therefore modulate the activity of this protein. In this context, no ANGPTL has yet been studied in detail. Characterization of the domains of ANGPTL4 required for ECM binding and responsible for antiangiogenic activities is therefore of major interest. Proteolysis is also involved in regulating the function of angiopoietin-like proteins. ANGPTL3 is cleaved at a proprotein convertase (PC) recognition sequence, 221RAPR224, and inhibition of PC thus decreases the cleavage of this molecule in vivo (15, 16). This cleavage activates the protein by releasing its amino-terminal region, which inhibits lipoprotein lipase more efficiently than the full-length form. ANGPTL4 also undergoes processing, releasing CCDand FLD-containing fragments (17, 18, 19). The soluble CCD binds to lipoprotein lipase and converts the catalytically active dimeric form of the enzyme into inactive monomers (20), whereas the soluble FLD inhibits tubule-like structures and neovascularization in mice (8). It is now necessary to determine the context-dependent availability of each endogenous form to estimate the physiological relevance of its angiogenic functions. ANGPTL4 processing is tissue and species specific. In humans, full-length ANGPTL4 is found in adipose tissue, whereas the amino-terminal fragment is produced by the liver. Both the full-length protein and the amino-terminal fragment are detected in plasma (18). A major cleavage site was recently identified between lysine 168 and leucine 169 in mouse ANGPTL4 (8), separating the CCD and Linker-FLD fragments. In the context of our study, we investigated whether similar cleavage may occur in the endogenous protein, synthethized by the human endothelium. Analyses of proteolysis and of the interaction of ANGPTL4 with the endothelial ECM are therefore required to determine the relevance of these processes in the antiangiogenic activity of ANGPTL4 in hypoxic EC.

MATERIALS AND METHODS Materials Anti-mouse ANGPTL4, monoclonal anti-myc (9E10), and polyclonal anti-myc antibodies were purchased from US Biological (Swampscott, MA, USA), Roche Diagnostics (Indianapolis, IN, USA), and Santa Cruz Biotechnology (Santa Cruz, CA, USA), respectively. Anti-human ANGPTL4 recognizes the full-length protein and the amino-terminal CCD (BioVendor Laboratory Medicine, Modrice, Czech Republic). Alexa Fluor-488 conjugated goat anti-mouse antibody and ToPro-3 were purchased from Invitrogen (Carlsbad, CA, USA). Low molecular weight heparin was obtained from Sigma (St. Louis, MO, USA), and fibronectin was from BD Biosciences (Franklin Lakes, NJ, USA). The inhibitor of CCD MEDIATES ANGIOGENIC ROLE OF ECM-BOUND ANGPTL4

human matrix metalloproteinases (MMPs), GM6001, was purchased from Chemicon International (Temecula, CA, USA). For surface plasmon resonance (SPR) binding assays, HSPG, dermatan sulfate, heparin, and chondroitin sulfate from bovine trachea were purchased from Sigma, and heparan sulfate was obtained from Celsus Laboratories (Cincinnati, OH, USA). For pulse-chase experiments, RIPA-DOC buffer was made with 50 mM Tris-HCl (pH 8.0), 150 mM NaCl, 1% TritonX100, 2 mM phenylmethylsulfonyl fluoride, 1 mM Na3Vo4, protease inhibitor cocktail (Sigma), 10 mM EDTA, 1% sodium deoxycholate, 10 mM NaF, and 0.1%SDS. The plasmid encoding furin was a gift from Dr. J. W. Creemers (Laboratory for Biochemical Neuro-endocrinology, Department for Human Genetics, University of Leuven and Flanders Interuniversity Institute for Biotechnology, Ghent, Belgium). The pCDNA3.1-LinkCCD22-185-V5-his plasmid was a gift from Dr. G. Martiny-Baron (Novartis Institutes for Biomedical Research, Basel, Switzerland). Cell culture HUVECs (primary cultures of human umbilical vein ECs, from 1 to 4 passages) and HMEC-1s (human microvascular EC line) were grown as previously reported (5). Chinese hamster ovary (CHO) cell lines lacking the dihydrofolate reductase (CHODHFR) stably transfected with the myc-tagged full-length ANGPTL4 cDNA or the myc-tagged fragments cDNA were established. CHODHFR-ANGPTL4, CHODHFR-CCD22-170, and CHODHFR-LinkFLD171-406 cells were selected in ␣MEM (Invitrogen), supplemented with geneticin (750 ␮g/ml) and 20 – 80 nM methotrexate. Construction of ANGPTL4 mutants CCD22-170 and CCD22-162 mutants Plasmids pCDNA3.1-CCD22-170-myc-his and pCDNA3.1-CCD22-162myc-his were generated from pCDNA3.1-ANGPTL4-myc-his (5) by polymerase chain reaction (PCR) using primers P1 (5⬘CTCGAGCGAGGATGGGTACCGGTGCTCCGA3⬘) and P2 (5⬘GGTACCCTGGGCCATCTCGGGCAGCCT3⬘) and primers T7 and P13 (5⬘GGTACCTCTTCGGGCAGGCTTGGC3⬘), respectively. FLD186-406 and LinkFLD171-406 mutants Plasmid pCDNA3.1-FLD186-406-myc-his was generated from pCDNA3.1-ANGPTL4-myc-his in a two-step PCR procedure: a first PCR using primers P1 and P3 (5⬘GAACAGCTCCTGGCAATCCCTGCTCAGTAG3⬘) and, in parallel, P4 (5⬘GGGGTACCGGAGGCTGCCTCTGCTGC3⬘) and P5 (5⬘GCCACCGCCGTGCTACTGAGCAGGGATTGC3⬘). The resulting fragments were then amplified in a second PCR using primers P1 and P4. Plasmid pCDNA3.1-LinkFLD171-406-myc-his was generated from pCDNA3.1-ANGPTL4-myc-his by a two-step PCR procedure using primers P1 and P6 (5⬘GTGAGCCGGGTCAACTGGGCTCAGTAGCAC3⬘), P4 and P7 (5⬘CACCGCCGTGCTACTGAGCCCAGTTGACCC3⬘) for the first step; P1 and P4 for the second step. Cleavage site mutants Plasmid pCDNA3.1-ANGPTL4-161AAAA164-myc-his was generated from pCDNA3.1-ANGPTL4-myc-his. The R161A, R162A, K163A, and R164A substitutions were established by a two-step PCR using primers P1 and P10 (5⬘CTGGGCCATCTCGGGCAGCGCCGCTGCTGCGGCAGGCTTGGCCACCTC3⬘) 941

and P4 and P11 (⬘5GAGGTGGCCAAGCCTGCCGCAGCAGCGGCGCTGCCCGAGATGGCCCAG3⬘) for the first step and using primers P1 and P4 for the second step. Plasmid pCDNA3.1-CCD161 AAAA 164 -myc-his was generated from pCDNA3.1ANGPTL4-161AAAA164-myc-his by PCR using primer T7 and primer P12 (5⬘GGTACCCTGGGCCATCTCGGGCAG3⬘). Amplified cDNAs were inserted into the pGEM-T vector (Promega, Madison, WI, USA), reexcised, and inserted into the analogous site in pCDNA3.1-myc-his. Purification of recombinant ANGPTL4 mutants CCD and FLD were purified from secretion medium from CHODHFR-CCD22-170 and CHODHFR-LinkFLD171-406 cells grown in Opti-MEM (Invitrogen) with 50 ␮g/ml of low molecular weight heparin, concentrated by a factor of 10 and loaded onto a Talon Cobalt affinity column (BD Biosciences). The bound CCD22-170 and LinkFLD171-406 proteins were eluted with 250 mM imidazole plus 500 mM NaCl and 125 mM imidazole plus 100 mM NaCl, respectively. Recombinant full-length ANGPTL4 protein was purified as described previously (5). Purity of recombinant proteins was assessed by SDS-PAGE and staining with Imperial protein (Pierce, Rockford, IL, USA). Concentrations were determined from a standard curve for BSA. Elution buffers were used as the control buffer in all experiments other than SPR assays (see below). Cell adhesion assay

sulfate over streptavidin and HSPG on the chip were performed as described previously (21). The kinetic parameters [association (Ka) and dissociation (Kd) rate constants] were analyzed simultaneously using a global data analysis program (BIAevaluation 4.1 software). This software fitted curves to the sensorgrams obtained at different analyte concentrations, constraining the kinetic rate constants to a single value for each set of curves. Rate and affinity constants were calculated using the appropriate binding models. Rmax, the maximal capacity of the surface, was floated during the fitting procedure. SPR binding assays were performed in the Protein Production and Analysis Facility of IFR 128 (Biosciences, Gerland-Lyon Sud, France). Quantificative analysis of immunoblots Lysate, secretion medium, and ECM were collected from the same culture plate. Protein contents of cell lysate were determined by BCA assay kit (Sigma), and samples were equally loaded. Akt protein was used as internal control in immunoblots using anti-Akt antibody (Cell Signaling Technology, Danvers, MA, USA). Immunoreactive proteins were revealed by fluorescence imaging using AttoPhos fluorescent substrat (Promega) and FX Molecular Imager (Bio-Rad, Richmond, CA, USA). Quantitative analysis was performed using QuantityOne software (Bio-Rad). Distribution of proteins (full-length ANGPTL4 or mutants) in the ECM or secretion medium was expressed as percentage of total secreted proteins detected in both ECM and medium.

Plates were coated by incubation overnight at 4°C with various concentrations of the purified recombinant proteins ANGPTL4, CCD22-170, and LinkFLD171-406 or fibronectin (5 ␮g/ml). HUVECs were released by trypsin digestion and plated in complete medium at a density of 65,000 cells/well. Cells were allowed to adhere to the plate for 45 min, and absorbance at 570 nm was then determined with the CyQuantproliferation assay kit (Invitrogen).

Statistical analysis

Pulse-chase experiments

RESULTS

CHODHFR-ANGPTL4 cells were plated on ECM from confluent HUVECs detached in 15 mM EDTA after 4 days of growing. Cells were pulsed with [35S]cysteine-methionine for 40 min, 24 h after seeding, and the pulse was then chased with Opti-MEM for the time periods indicated. Secretion medium, ECM, and lysates were collected in RIPA-DOC buffer and were incubated overnight at 4°C with polyclonal anti-myc antibody and then for 2 h with protein G-Sepharose 4 Fast Flow beads (GE Healthcare, Piscataway, NJ, USA).

CCD is required for the binding of ANGPTL4 to the ECM

SPR binding assays SPR was carried out on a BIAcore 3000 instrument with CM4 sensor chips (Biacore, Elkridge, MD, USA). Binding assays were performed at 25°C in 50 mM Tris buffer (pH 7.5), 0.15 M NaCl, and 0.005% (v/v) P20 surfactant at a flow rate of 30 ␮l/min. Kinetic experiments were carried out at a flow rate of 90 ␮l/min. Purified ANGPTL4, CCD22-170, and LinkFLD171-406, dialyzed against PBS, were injected over immobilized heparan sulfate and HSPG. The surface was regenerated with a pulse of 1.5 M NaCl and 2 M guanidinium hydrochloride. In inhibition experiments, ANGPTL4 (10 ␮g/ml) or CCD22-170 (2 ␮g/ml) was first incubated for 1 h at room temperature with various glycosaminoglycans or HSPG as competitors (molar ratio 1:2) before injection over heparan sulfate. The activation and blocking steps and the injection of heparan 942

Vol. 23

March 2009

The results are expressed as mean ⫾ sd of the specified number of independent experiments performed in triplicate. Differences between the means of two groups were evaluated with Student’s paired t test; values of P ⬍ 0.05 were considered significant.

We generated the recombinant CCD and FLD mutants of ANGPTL4 for identification of the domains involved in the interaction with ECM. As the linker located between the CCD and the FLD is responsible for the binding of Ang-1 to the ECM (12), we also built constructs containing this region (Fig. 1A). The mutant proteins were analyzed in transiently transfected human embryonic kidney (HEK) 293 cells. Immunoblot analysis revealed that the CCD-containing fragments (CCD22-170 and CCDLink22-185) and full-length ANGPTL4 were principally detected in the ECM, whereas the LinkFLD171-406 was present in the secretion medium (Fig. 1B). The FLD186-406 was mostly retained in the cell lysate, indicating that this truncated form was not correctly secreted. The CCD22-170 and CCDLink22-185 in the ECM accounted for 63.2 ⫾ 5.0 and 88.3 ⫾ 7.4%, respectively, of the total pool of secreted protein (Fig. 1C), whereas all of the LinkFLD171-406 was secreted into the medium (98.5⫾1.1% of secreted protein). ECM and secretion medium distribution

The FASEB Journal

CHOMEL ET AL.

Figure 1. CCD is required for the binding of ANGPTL4 to the ECM. A) Structure of mutants, containing a signal peptide (SP), CCD, FLD, and linker (bold line). Broken lines indicate deleted regions. B) Immunoblot of lysate, secretion medium (SM), and ECM from transfected HEK293 cells using anti-myc or anti-V5 antibody. Akt protein was used as internal control for the cell lysates (Supplemental Fig. 1). C) Distribution of ANGPTL4-myc or its domains in the ECM or SM, expressed as percentage of total secreted proteins (ECM and SM). Quantification was performed on immunoblot analysis of ECM and SM from transfected HEK293 cells. Results are means ⫾ sd of 3 independent experiments. D) Immunofluorescence studies of nonpermeabilized CHODHFRANGPTL4, CHODHFR-CCD22-170 or CHODHFR-LinkFLD171-406 cell lines using anti-myc antibody. Nuclei were stained by ToPro-3. Scale bars ⫽ 40 ␮m. M.W., molecular weight.

demonstrated therefore that the CCD is required for the interaction with the ECM. Moreover, the linker further favored this interaction as revealed by the higher ratio in the ECM of CCDLink22-185 than of CCD22-170 but is, however, not sufficient per se to target the FLD into the ECM. In agreement with the immunoblot results, immunofluorescence analysis also revealed the presence of ANGPTL4, CCD22-170, or CCDLink22-185 in the ECM from transfected HEK293 cells, whereas the FLD186-406 or LinkFLD171-406 was not detected in the ECM (Supplemental Fig. 2). As the CCD22-170 and LinkFLD171406 correspond to the fragments generated by the cleavage of ANGPTL4 (8), these two mutants were characterized further in stably transfected CHODHFR cell lines. Confocal immunofluorescence analysis showed that both the CCD22-170 and full-length protein were diffusely distributed without organization into fibrils, whereas the LinkFLD171-406 was not detected in the ECM (Fig. 1D). The pattern of distribution of the CCD22–170 and the full-length protein favored the hypothesis of an interaction with components homogenously and largely distributed in the ECM, like proteoglycans.

ANGPTL4 binds to heparan and dermatan sulfate chains through its CCD We characterized ECM binding properties by analyzing the effect of heparin treatment on CHODHFR cell lines. Heparin dislodged both the CCD22-170 and the full-length ANGPTL4 from the ECM, releasing these proteins into the medium (Fig. 2A). The LinkFLD171-406 was not affected by heparin. Moreover, in HEK293 transfected cells, heparin treatment affected the interaction of the CCDLink22-185 with the ECM, whereas it did not modify the distribution of FLD186-406 in the medium (Supplemental Fig. 3). Characterization of the interaction of ANGPTL4 and its domains with proteoglycans was performed further by SPR binding assays using the purified proteins. Whereas no binding was observed for the LinkFLD171-406, ANGPTL4 and the CCD22-170 bound to immobilized HSPG and heparan sulfate chains (Fig. 2B and C, respectively) and both formed stable complexes with slow dissociation rates. The refractive index of CCD22-170 was lower than that of ANGPTL4 because the SPR signal depends on the mass

Figure 2. ANGPTL4 binds to heparan sulfate and HSPG through its CCD. A) Immunoblot of SM and ECM from CHODHFR-ANGPTL4, CHODHFR-CCD22-170, or CHODHFR-LinkFLD171-406 cells cultured in the presence or absence of heparin (50 ␮g/ml), using anti-myc antibody. B, C) SPR binding assays performed by injecting 90 nM ANGPTL4, CCD22-170, or LinkFLD171-406 over 100 ␮g/ml immobilized HSPG (B) or heparan sulfate (HS) chains (C). RU, resonance units. CCD MEDIATES ANGIOGENIC ROLE OF ECM-BOUND ANGPTL4

943

of the injected molecule. SPR inhibition assays were carried out with the full-length ANGPTL4 and with the CCD22-170 to assess the binding specificity of these two proteins (Table 1). Their interaction with immobilized heparan sulfate chains was abolished by prior incubation with heparin or HSPG. Furthermore, the binding of both proteins was inhibited by heparan sulfate chains and, to a lesser extent, by dermatan sulfate chains. By contrast, chondroitin sulfate chains did not affect the binding of ANGPTL4 or CCD22-170 to heparan sulfate. Kinetic analysis was carried out by injecting ANGPTL4 or CCD22-170 over heparan sulfate chains to calculate the Ka and Kd rate constants (Table 2; Supplemental Fig. 4). The experimental data were best fitted by a bivalent analyte model, which corresponds to the presence of two heparin-binding sites on ANGPTL4. Our model is consistent with previous data indicating that the soluble ANGPTL4 forms oligomers through intermolecular disulfide bonds involving cysteines 76 and 80 within the CCD (22). We also detected endothelial ECM-bound dimers for ANGPTL4 (5) and CCD22-170 (data not shown). These results suggest that there is a single heparin-binding site in each monomer of molecule. The Ka1 of the first binding site was similar for ANGPTL4 and CCD22-170, whereas the Kd1 for CCD22-170 was 3 times higher than that for the full-length protein, indicating that interaction of ANGPTL4 with ECM was more stable. The affinity constant of the first binding site (KD1⫽Kd1/Ka1) was therefore lower for ANGPTL4 (1.24⫻10⫺7 M) than for CCD22-170 (4.5⫻10⫺7 M). No striking differences were observed in the Ka2 and Kd2 for the binding of heparan sulfate chains to the second site in ANGPTL4 and CCD22-170. CCD mediates the antiangiogenic activities of immobilized ANGPTL4 Since CCD22-170 is responsible for the interaction of ANGPTL4 to the proteoglycans, we investigated whether this immobilized mutant protein was also responsible for the antiangiogenic activities of the ECM-bound ANGPTL4. HUVEC adhesion was inhibited on immobilized CCD22-170 or full-length protein, with adhesion rates reaching only 41.0 ⫾ 3.5 and 45.0 ⫾ 3.4%, respectively, of the control level (Fig. 3A, TABLE 1. ANGPTL4 and CCD22-170 bind specifically to heparan and dermatan sulfate chains Inhibition of binding to immobilized HS (%) Competitor (5 ␮g/ml)

ANGPTL4

CCD22-170

Heparin HSPG Heparan sulfate Dermatan sulfate Chondroitin sulfate

100 97.6 85 61.4 6

100 88.8 80 73.4 6

Results are expressed as percentage of inhibition of binding detected in the absence of competitors.

944

Vol. 23

March 2009

left panel), The antiadhesive effect of CCD22-170 was dose dependent, like that of ANGPTL4 (Fig. 3A, right panel). Cell migration was investigated by tracking the motility of individual HUVECs, using various coatings mixed or not with fibronectin (Fig. 3B). CCD22-170 significantly decreased HUVEC velocity (control: 17.42⫾0.84 ␮m/h; CCD22-170: 13.03⫾0.88 ␮m/h), as did ANGPTL4 (13.60⫾0.90 ␮m/h). In addition, CCD22-170 counteracted the increase in velocity induced by fibronectin, like ANGPTL4 (fibronectin: 23.40⫾1.89 ␮m/h; fibronectin mixed with CCD22-170: 17.93⫾0.61 ␮m/h; fibronectin mixed with ANGPTL4: 17.50⫾1.79 ␮m/h). Furthermore, CCD22-170 mixed with Matrigel reduced the formation of tubule-like structures of microvascular ECs, HMEC-1s (62.4⫾3.2% of tubes formed on control Matrigel). This decrease is similar to that observed for ANGPTL4-containing Matrigel (71.7⫾4.1%; Fig. 3C). In contrast, coating with the immobilized LinkFLD171-406 did not affect EC adhesion, migration, and formation of tubule-like structures (Supplemental Fig. 5). The CCD therefore mediates the antiangiogenic properties of the full-length ECM-bound ANGPTL4. Interaction of ANGPTL4 with the endothelial ECM protects against proteolysis To further understand the regulation of the bioactivity of ANGPTL4 in the hypoxic endothelial microenvironment, proteolytic processing of the protein was investigated in HUVECs. Immunoblot of ECM and medium from ECs grown under hypoxia was performed with a human anti-ANGPTL4 antibody targeting the aminoterminal part of the molecule, including the CCD. In the absence of heparin, the CCD fragment was not detected in hypoxic conditions, whereas the endogenous full-length ANGPTL4 accumulated in the ECM (Fig. 4A, left panel). After treatment with heparin, both the CCD fragment and the full-length protein were detected in the secretion medium. Immunoblot of ECM and medium from HUVECs transfected with recombinant ANGPTL4 was carried out with anti-myc antibody, targeting the carboxy-terminal part of the molecule, including the FLD. The heparin-induced release of full-length ANGPTL4 was correlated with the presence of the FLD fragment in the secretion medium (Fig. 4A, right panel). The interaction of ANGPTL4 with the ECM therefore protected the full-length form from proteolysis. When the full-length form was released from the ECM into the medium by heparin, the soluble protein was cleaved, thereby generating both the CCD (Fig. 4A, left panel) and FLD fragments (Fig. 4A, right panel). However, neither of these forms accumulated in the ECM, because FLD could not bind to the ECM and heparin abolished the interaction of CCD with proteoglycans containing heparan sulfate chains. The kinetic regulation of extracellular distribution of ANGPTL4 was analyzed by carrying out pulse-chase experiments (Fig. 4B). Full-length ANGPTL4 was tar-

The FASEB Journal

CHOMEL ET AL.

TABLE 2.

ANGPTL4 and CCD22–170. bind to heparan sulfate

Analyte

ANGPTL4 (13.6–36 nM) CCD22–170 (11.25–227 nM)

Ligand

␹2

Ka1 (M⫺1s⫺1)

Kd1 (s⫺1)

Ka2 (RU⫺1s⫺1)

Kd2 (s⫺1)

Heparin sulfate Heparin sulfate

0.874 0.589

6.97 ⫻ 104 5.42 ⫻ 104

8.63 ⫻ 10⫺3 24.4 ⫻ 10⫺3

1.1 ⫻ 10⫺4 2.1 ⫻ 10⫺4

2.28 ⫻ 10⫺4 3.91 ⫻ 10⫺4

Kinetic analyses were performed by injecting increasing concentrations of ANGPTL4 or CCD22–170 over immobilized HS.

geted to the ECM within 30 min. After 1 h, newly synthesized ANGPTL4 was released into the medium, due to the high level of accumulation of ANGPTL4 in the ECM. Indeed, the high abundance of ECM-bound ANGPTL4 may lead to a passive secretion of the full-length form. Small amounts of the FLD fragment were subsequently detected in the medium, whereas this fragment was not detected in the cell lysate. Thus, cleavage coincided with the release of the full-length form into the secretion medium. Proprotein convertases are involved in the proteolytic processing of soluble ANGPTL4 As the cleavage of the soluble ANGPTL4 therefore occurred in the extracellular compartment of the ECs, we further investigated the molecular mechanisms involved in the processing of the protein. Various concentrations of an inhibitor of the MMPs, GM6001, were applied to HUVECs producing recombinant ANGPTL4 in the presence of heparin (Fig. 5A, left panel). GM6001 had no effect on the amount of the FLD fragment in the medium (Fig. 5A, right panel), thereby ruling out the possibility of MMP involvement in the cleavage of ANGPTL4. We then investigated the possible involvement of the subtilisin-like proprotein convertases in ANGPTL4 pro-

cessing. Members of the proprotein convertase family, including furin, cleave various proteins at RXKR, RXRR, RR, and KR sequences (23). HUVECs were cotransfected with plasmids encoding ANGPTL4 and ␣1-anti-trypsin Portland (␣1-PDX), a potent inhibitor of all proprotein convertases (24). We demonstrated that ␣1-PDX was efficiently transfected (green fluorescent protein expression from pIRES2-EGFP plasmid encoding ␣1PDX; data not shown). The inhibition of the proprotein convertases by ␣1-PDX decreased the abundance of the FLD fragment, whereas the abundance of the full-length form increased in the secretion medium (Fig. 5B). Furthermore, the production of both recombinant furin and ANGPTL4 in furin-deficient LoVo-C5 cells increased the ratio of the FLD fragment in the secretion medium over the total pool of secreted proteins from 33.7 ⫾ 5.8% without furin to 68.6 ⫾ 4.7% with furin (Fig. 5C). Thus, proprotein convertases are involved in the proteolytic processing of soluble ANGPTL4 in ECs. Analysis of the amino acid sequence of the human ANGPTL4 revealed a putative cleavage site for proprotein convertases, 161RRKR164, located at the end of the CCD. We modified this motif in the full-length protein to generate the ANGPTL4161AAAA164 mutant (Fig. 5D,

Figure 3. CCD mediates the antiangiogenic activities of immobilized ANGPTL4. A) Cell adhesion. Left panel: HUVECs plated on purified immobilized ANGPTL4, CCD22-170, or fibronectin (FN) at a concentration of 5 ␮g/ml. Basal adhesion for control buffer was set to 100% (control). Right panel: HUVECs plated on various concentrations of ANGPTL4 or CCD22-170. Results are means ⫾ sd of 3 independent experiments. Nonspecific binding sites were saturated by 1% BSA in all conditions, including controls. B) Time-lapse videomicroscopy: HUVECs plated on ANGPTL4 or CCD22-170 (5 ␮g/ml; black column) or on a mixed coating including fibronectin (5 ␮g/ml; gray column). Nonspecific binding sites were saturated by 1% BSA. Cell velocity was determined as described previously (5). Results are means ⫾ sd of 3 independent experiments. C) Tubule-like structure formation: HMEC-1s seeded on Matrigel mixed with ANGPTL4 or CCD22-170 (10 ␮g/ml) or control buffer were analyzed after 24 h. Basal tube number for control buffer was set to 100% (control). Results are means ⫾ sd of 3 independent experiments. CCD MEDIATES ANGIOGENIC ROLE OF ECM-BOUND ANGPTL4

945

Figure 4. Interaction of ANGPTL4 with the endothelial ECM protects against proteolysis. A) Left panel: immunoblot using anti-human ANGPTL4 antibody of the SM and ECM from HUVECs expressing endogenous ANGPTL4, cultured in the presence or absence of heparin (50 ␮g/ml) under normoxia (N), i.e., 21% O2, or under hypoxia (H), i.e., 1% O2. Right panel: immunoblot using anti-myc antibody of the SM and ECM from HUVECs producing recombinant ANGPTL4-myc, cultured in the presence or absence of heparin (50 ␮g/ml) under normoxia. B) Autoradiograph from pulse-chase analysis of CHODHFRANGPTL4 cells seeded on ECM from HUVECs for 24 h. Lysate, SM, and ECM were collected at each time point.

left panel). Whereas the soluble full-length ANGPTL4 was cleaved as expected in the secretion medium after treatment with heparin, the FLD fragment was not detected in the secretion medium of HUVECs producing the ANGPTL4161AAAA164 mutant protein (Fig. 5D, right panel). We also observed that this mutant protein could not be cleaved in transfected HEK293 cells (data not shown). The 161RRKR164 sequence is thus the functional cleavage site of human ANGPTL4. Cleavage did not modify the interaction of ANGPTL4 to the ECM As the cleavage site is located at the end of the CCD, we investigated whether ANGPTL4 cleavage regulated the interaction with the ECM and the accumulation of the

full-length protein. The 161RRKR164 cleavage site, rich in positively charged residues, is a good candidate heparin-binding site. CCD22–170 contains this sequence and binds to ECM; we therefore modified this sequence within the CCD mutant protein to generate the CCD161AAAA164 mutant (Fig. 6A). We also generated the CCD22-162 mutant protein corresponding to the cleaved fragment released by proprotein convertases. These mutant proteins, produced in HUVECs, bound to the endothelial ECM as efficiently as the full-length ANGPTL4 (Fig. 6B), thus demonstrating the presence of a proper heparin-binding site within the CCD distinct from the RRKR sequence. Moreover, the amount of the ECMbound full-length ANGPTL4 was not affected by the ␣1-PDX-induced inhibition of proteolysis in EC (Fig. 5B). Thus, heparin is sufficient to induce the release of

Figure 5. Soluble ANGPTL4 is cleaved by proprotein convertases at the 161RRKR164 site in HUVECs. A) Left panel: immunoblot using anti-myc antibody of SM and ECM from HUVECs producing ANGPTL4-myc after heparin (50 ␮g/ml) treatment in the presence of various concentrations of GM6001. Right panel: FLD fragment in SM was quantified and expressed as percentage of total secreted proteins (full-length and fragment in ECM and SM). B) Left panel: immunoblot of lysate, SM, and ECM from HUVECs producing ANGPTL4-myc with or without ␣1-PDX on heparin (50 ␮g/ml) treatment using anti-myc antibody. Right panel: full-length ANGPTL4 and the FLD fragment were quantified and expressed as percentage of total secreted proteins (as in A). C) Left panel: immunoblot of SM and ECM from LoVo-C5 cells producing ANGPTL4-myc with or without cotransfection with a plasmid encoding furin, in absence of heparin, using anti-myc antibody. Cell lysates were analyzed for ANGPTL4 content; Akt protein was used as internal control (Supplemental Fig. 6). Right panel: FLD fragment in SM was quantified and expressed as percentage of total secreted proteins (as in A). Results are mean ⫾ sd of 3 independent experiments. D) Left panel: schematic representation of the full-length ANGPTL4 mutated within the cleavage site. Right panel: immunoblot using anti-myc antibody of SM and ECM from HUVECs producing ANGPTL4-myc or ANGPTL4-161AAAA164-myc on heparin (50 ␮g/ml) treatment. Graphed results are means ⫾ sd of 3 independent experiments. 946

Vol. 23

March 2009

The FASEB Journal

CHOMEL ET AL.

Figure 6. Cleavage site is not involved in the interaction of ANGPTL4 to the ECM. A) Schematic representations of CCD22–170 mutated within the cleavage site and of CCD22-162. B) Distribution of ANGPTL4-myc or its cleavage-site mutants in the ECM or SM, expressed as percentage of total secreted proteins (ECM and SM). Quantification was performed on immunoblot analysis of ECM and SM from transfected HUVECs. Results are means ⫾ sd of 3 independent experiments.

ANGPTL4 from ECM, and proprotein convertase activity is not required for this process. The soluble FLD inhibited angiogenesis (8). The antiangiogenic function of ANGPTL4 could then result from an indirect effect of the FLD fragment. We therefore investigated whether the antiangiogenic activity of the ECM-bound full-length form was independent of cleavage. Adhesion assays of HUVECs were performed on immobilized ANGPTL4-161AAAA164 or ANGPTL4 from the secretion medium of transfected HEK293 cells. We observed an antiadhesive effect similar to that obtained for CCD22-170 and the full-length form (data not shown). The ECM-bound full-length form therefore displays antiangiogenic activity, independently on cleavage. These data show that the 161RRKR164 sequence is not involved in either the control of endothelial ECM interaction or the release of ANGPTL4. Moreover, proprotein convertase activity is not required for the heparin-induced release of ANGPTL4. The ECMbound full-length form displays thus a direct effect on ECs, independent of the generation of the FLD fragment.

DISCUSSION Evidence is accumulating to suggest that the angiogenic functions of ANGPTL4 depend on its microenvironment. Soluble and ECM-bound full-length ANGPTL4 display antiangiogenic activity (7, 5). Yang et al. (8) recently reported that the soluble FLD fragment inhibits angiogenesis, whereas the soluble CCD does not induce any antiangiogenic response. We also observed in our study that the soluble CCD is not active on EC (data not CCD MEDIATES ANGIOGENIC ROLE OF ECM-BOUND ANGPTL4

shown). Our data highlight the involvement of the ECM binding, release, and proteolysis in the regulation of the ANGPTL4 functions in the endothelial microenvironment. We provide evidence here that the bioavailability of full-length ANGPTL4 in hypoxic ECs is regulated by the interaction of this protein with the endothelial ECM through its CCD and, once soluble, by its cleavage at the 161RRKR164 site by proprotein convertases. We characterized the ANGPTL4 domains and glycosaminoglycans involved in ECM interaction. The recombinant LinkFLD171-406 of ANGPTL4 is secreted into the medium. The linker is not sufficient to promote the binding of the FLD to the ECM. Recombinant CCD22-170 interacts with heparan or dermatan sulfate chains in the ECM. The specific role of the CCD in the interaction with the ECM states precisely its contribution to the angiogenic functions of ANGPTL4. The mechanisms involved in the interaction of ANGPTL4 with the ECM are close to those observed for Ang-3 (13). ANGPTL4 may be tethered on the cell surface or on the ECM via heparan sulfate or dermatan sulfate proteoglycans and may then display activity correlated with its local abundance in good agreement with the context-dependent angiogenic function. ECM-bound ANGPTL4 is protected from proteolysis, suggesting that the cleavage site is not accessible to proteases when the full-length form is anchored in the ECM. The heparin-binding site may therefore be located within the carboxy-terminal part of the CCD, close to the 161RRKR164 cleavage site. We found that the cleavage site was not involved in interaction with the endothelial ECM. The putative heparin-binding motif of ANGPTL4, 59HAERTRS65, corresponding to the predicted motif, 61 VHKTKG 66 , in the CCD of ANGPTL3 (15) was mutated to 59HAINTNS65. This mutation had no effect on binding to the endothelial ECM (data not shown), indicating that this sequence is not a bona fide heparin-binding motif, consistent with the presence of a negatively charged glutamate residue at position 61. Thus, no structural heparin-binding motif has been yet identified for ANGPTL4. Consensus secondary structure predictions indicate the presence of a cluster of positively charged residues at the surface of the second ␣-helix in the carboxy-terminal part of the CCD. The linker increased CCD accumulation in the ECM, suggesting that, within the full-length form, the linker forms part of the functional anchor to ECM. The 161RRKR164 cleavage site would thus be protected from the proteolytic activity of proprotein convertases by both the ␣-helix heparin-binding site and the linker. This study provides new structural and functional data about ANGPTL4 and its proteolytic fragments. The CCD alone can interact with the ECM, whereas the FLD fragment is found only in its soluble form, released into the secretion medium. We also assessed the requirement of the glycosylation site of ANGPTL4 for ECM binding. Indeed, the N-glycosylation site at asparagine 177 in the FLD fragment of mouse ANGPTL4 is required to maximize the antiangiogenic activities of 947

this domain (8), suggesting that the N-linked glycan chains may be involved in ECM interaction. Our studies demonstrate that asparagine-177, also present in human ANGPTL4, is not sufficient to regulate interaction with the ECM. Indeed, Link-FLD171– 406, which contains this glycosylated residue, does not interact with the ECM. This N-glycosylation site therefore makes no contribution to maximazing the antiangiogenic activities of the carboxy-terminal fragment through interaction with the endothelial ECM. Previous studies (18) have reported ANGPTL4 processing to be tissue specific. We further show here that proprotein convertases are responsible for ANGPTL4 proteolysis in the endothelial microenvironment. PACE4 and PC5 are good candidate molecules for the proprotein convertases involved in this process. These convertases are activated at the cell surface, where they are tethered to HSPG via interactions involving their carboxy-terminal cysteine-rich domain (25, 23). Heparin dislodges PACE4 bound to the ECM without inhibiting its activity (26). These data suggest that ANGPTL4 and its putative processing enzymes PACE4/PC5 are bound to HSPG in close proximity. Heparin releases both molecules, allowing the cleavage of ANGPTL4 into FLD and CCD. The endothelium of angiogenic blood vessels contains many enzymes that might affect the interaction of ANGPTL4 with HSPG; these enzymes include endosulfatases (27) and heparanase (28). The active mechanism involved in release from ECM is currently unknown. The ANGPTL4 produced by ECs accumulates in the ECM, where its is protected from proteolysis. Hypoxic ECs do not provide a major pool of CCD and FLD fragments. ECM-bound ANGPTL4 is therefore the relevant active form in the endothelium, mediating the autocrine regulation of angiogenesis. Nevertheless, circulating ANGPTL4 produced by liver or adipose tissue may also be cleaved by secreted proprotein convertases in the EC microenvironment, suggesting an endocrine regulation. This proteolysis generates soluble FLD, which transduces antiangiogenic activities, possibly by inhibiting the FGF-2- or VEGF-induced phosphorylation of Raf-1 and MEK1/2 (8) and CCD, which binds to endothelial ECM and transduces antiangiogenic activities through an unknown mechanism. These data open up new possibilities for the functions of ANGPTL4, various forms of which (the ECMbound full-length form, the ECM-bound CCD, and the soluble FLD) have antiangiogenic activities the relevance of which depends on the context of interaction with the ECM and cleavage. This work was supported by the European Vascular Genomics Network (http://www.evgn.org; contract LSHMCT-2003503254) and by a grant from the Agence Nationale de la Recherche (ANR-06-JCJC-0160). We thank Eric Etienne for technical help in confocal microscopy, videomicroscopy, and cell tracking. 948

Vol. 23

March 2009

REFERENCES 1. 2.

3.

4. 5.

6.

7.

8.

9.

10.

11.

12. 13. 14. 15.

16.

17.

18.

Kalluri, R. (2003) Basement membranes: structure, assembly and role in tumour angiogenesis. Nat. Rev. Cancer 3, 422– 433 Le Jan, S., Le Meur, N., Cazes, A., Philippe, J., Le Cunff, M., Leger, J., Corvol, P., and Germain, S. (2006) Characterization of the expression of the hypoxia-induced genes neuritin, TXNIP and IGFBP3 in cancer. FEBS Lett. 580, 3395–3400 Le Jan, S., Amy, C., Cazes, A., Monnot, C., Lamande, N., Favier, J., Philippe, J., Sibony, M., Gasc, J. M., Corvol, P., and Germain, S. (2003) Angiopoietin-like 4 is a proangiogenic factor produced during ischemia and in conventional renal cell carcinoma. Am. J. Pathol. 162, 1521–1528 Hato, T., Tabata, M., and Oike, Y. (2008) The role of angiopoietin-like proteins in angiogenesis and metabolism. Trends Cardiovasc. Med. 18, 6 –14 Cazes, A., Galaup, A., Chomel, C., Bignon, M., Brechot, N., Le Jan, S., Weber, H., Corvol, P., Muller, L., Germain, S., and Monnot, C. (2006) Extracellular matrix-bound angiopoietinlike 4 inhibits endothelial cell adhesion, migration, and sprouting and alters actin cytoskeleton. Circ. Res. 99, 1207–1215 Hermann, L. M., Pinkerton, M., Jennings, K., Yang, L., Grom, A., Sowders, D., Kersten, S., Witte, D. P., Hirsch, R., and Thornton, S. (2005) Angiopoietin-like-4 is a potential angiogenic mediator in arthritis. Clin. Immunol. 115, 93–101 Ito, Y., Oike, Y., Yasunaga, K., Hamada, K., Miyata, K., Matsumoto, S., Sugano, S., Tanihara, H., Masuho, Y., and Suda, T. (2003) Inhibition of angiogenesis and vascular leakiness by angiopoietin-related protein 4. Cancer Res. 63, 6651–6657 Yang, Y. H., Wang, Y., Lam, K. S., Yau, M. H., Cheng, K. K., Zhang, J., Zhu, W., Wu, D., and Xu, A. (2008) Suppression of the Raf/MEK/ERK signaling cascade and inhibition of angiogenesis by the carboxyl terminus of angiopoietin-like protein 4. Arterioscler. Thromb. Vasc. Biol. 28, 835– 840 Galaup, A., Cazes, A., Le Jan, S., Philippe, J., Connault, E., Le Coz, E., Mekid, H., Mir, L. M., Opolon, P., Corvol, P., Monnot, C., and Germain, S. (2006) Angiopoietin-like 4 prevents metastasis through inhibition of vascular permeability and tumor cell motility and invasiveness. Proc. Natl. Acad. Sci. U. S. A. 103, 18721–18726 Padua, D., Zhang, X. H., Wang, Q., Nadal, C., Gerald, W. L., Gomis, R. R., and Massague, J. (2008) TGFbeta primes breast tumors for lung metastasis seeding through angiopoietin-like 4. Cell 133, 66 –77 Kim, I., Kim, H. G., Kim, H., Kim, H. H., Park, S. K., Uhm, C. S., Lee, Z. H., and Koh, G. Y. (2000) Hepatic expression, synthesis and secretion of a novel fibrinogen/angiopoietin-related protein that prevents endothelial-cell apoptosis. Biochem. J. 346, 603– 610 Xu, Y., and Yu, Q. (2001) Angiopoietin-1, unlike angiopoietin-2, is incorporated into the extracellular matrix via its linker peptide region. J. Biol. Chem. 276, 34990 –34998 Xu, Y., Liu, Y. J., and Yu, Q. (2004) Angiopoietin-3 is tethered on the cell surface via heparan sulfate proteoglycans. J. Biol. Chem. 279, 41179 – 41188 Xu, Y., Liu, Y. J., and Yu, Q. (2004) Angiopoietin-3 inhibits pulmonary metastasis by inhibiting tumor angiogenesis. Cancer Res. 64, 6119 – 6126 Ono, M., Shimizugawa, T., Shimamura, M., Yoshida, K., Noji-Sakikawa, C., Ando, Y., Koishi, R., and Furukawa, H. (2003) Protein region important for regulation of lipid metabolism in angiopoietin-like 3 (ANGPTL3): ANGPTL3 is cleaved and activated in vivo. J. Biol. Chem. 278, 41804 – 41809 Jin, W., Wang, X., Millar, J. S., Quertermous, T., Rothblat, G. H., Glick, J. M., and Rader, D. J. (2007) Hepatic proprotein convertases modulate HDL metabolism. Cell. Metab. 6, 129 –136 Ge, H., Yang, G., Huang, L., Motola, D. L., Pourbahrami, T., and Li, C. (2004) Oligomerization and regulated proteolytic processing of angiopoietin-like protein 4. J. Biol. Chem. 279, 2038 – 2045 Mandard, S., Zandbergen, F., Tan, N. S., Escher, P., Patsouris, D., Koenig, W., Kleemann, R., Bakker, A., Veenman, F., Wahli, W., Muller, M., and Kersten, S. (2004) The direct peroxisome proliferator-activated receptor target fasting-induced adipose factor (FIAF/PGAR/ANGPTL4) is present in blood plasma as a

The FASEB Journal

CHOMEL ET AL.

19.

20.

21.

22. 23.

24.

truncated protein that is increased by fenofibrate treatment. J. Biol. Chem. 279, 34411–34420 Xu, A., Lam, M. C., Chan, K. W., Wang, Y., Zhang, J., Hoo, R. L., Xu, J. Y., Chen, B., Chow, W. S., Tso, A. W., and Lam, K. S. (2005) Angiopoietin-like protein 4 decreases blood glucose and improves glucose tolerance but induces hyperlipidemia and hepatic steatosis in mice. Proc. Natl. Acad. Sci. U. S. A. 102, 6086 – 6091 Sukonina, V., Lookene, A., Olivecrona, T., and Olivecrona, G. (2006) Angiopoietin-like protein 4 converts lipoprotein lipase to inactive monomers and modulates lipase activity in adipose tissue. Proc. Natl. Acad. Sci. U. S. A. 103, 17450 –17455 Ricard-Blum, S., Beraud, M., Raynal, N., Farndale, R. W., and Ruggiero, F. (2006) Structural requirements for heparin/heparan sulfate binding to type V collagen. J. Biol. Chem. 281, 25195–25204 Ge, H., Yang, G., Yu, X., Pourbahrami, T., and Li, C. (2004) Oligomerization state-dependent hyperlipidemic effect of angiopoietin-like protein 4. J. Lipid Res. 45, 2071–2079 Seidah, N. G., Mayer, G., Zaid, A., Rousselet, E., Nassoury, N., Poirier, S., Essalmani, R., and Prat, A. (2008) The activation and physiological functions of the proprotein convertases. Int. J. Biochem. Cell Biol. 40, 1111–1125 Scamuffa, N., Siegfried, G., Bontemps, Y., Ma, L., Basak, A., Cherel, G., Calvo, F., Seidah, N. G., and Khatib, A. M. (2008) Selective inhibition of proprotein convertases represses the

CCD MEDIATES ANGIOGENIC ROLE OF ECM-BOUND ANGPTL4

25.

26.

27.

28.

metastatic potential of human colorectal tumor cells. J. Clin. Invest. 118, 352–363 Nour, N., Mayer, G., Mort, J. S., Salvas, A., Mbikay, M., Morrison, C. J., Overall, C. M., and Seidah, N. G. (2005) The cysteine-rich domain of the secreted proprotein convertases PC5A and PACE4 functions as a cell surface anchor and interacts with tissue inhibitors of metalloproteinases. Mol. Biol. Cell 16, 5215–5226 Tsuji, A., Sakurai, K., Kiyokage, E., Yamazaki, T., Koide, S., Toida, K., Ishimura, K., and Matsuda, Y. (2003) Secretory proprotein convertases PACE4 and PC6A are heparin-binding proteins which are localized in the extracellular matrix. Potential role of PACE4 in the activation of proproteins in the extracellular matrix. Biochim. Biophys. Acta 1645, 95–104 Morimoto-Tomita, M., Uchimura, K., Bistrup, A., Lum, D. H., Egeblad, M., Boudreau, N., Werb, Z., and Rosen, S. D. (2005) Sulf-2, a proangiogenic heparan sulfate endosulfatase, is upregulated in breast cancer. Neoplasia 7, 1001–1010 Kelly, T., Miao, H. Q., Yang, Y., Navarro, E., Kussie, P., Huang, Y., MacLeod, V., Casciano, J., Joseph, L., Zhan, F., Zangari, M., Barlogie, B., Shaughnessy, J., and Sanderson, R. D. (2003) High heparanase activity in multiple myeloma is associated with elevated microvessel density. Cancer Res. 63, 8749 – 8756 Received for publication July 9, 2008. Accepted for publication October 23, 2008.

949