Plasminogen Activator Inhibitor 1 Is an Intracellular Inhibitor of Furin Proprotein Convertase

1224 Research Article Plasminogen activator inhibitor 1 is an intracellular inhibitor of furin proprotein convertase

Denis Bernot1, Jimmy Stalin1, Pierre Stocker2, Bernadette Bonardo1, Ilse Scroyen1, Marie-Christine Alessi1 and Franck Peiretti1,* 1Inserm, U626, Université de la Méditerranée, Faculté de Médecine, 27 Boulevard Jean Moulin, 13385 Marseilles Cedex 5, France 2Université Paul Cézanne, Equipe Biosciences iSm2 UMR 6263, FST St Jérome, case 342 13397, Marseilles Cedex 20, France *Author for correspondence ([email protected])

Summary Proprotein convertases (PCs) are a family of serine proteases that are involved in the post-translational processing and activation of a wide range of regulatory proteins. The upstream role of PCs in the control of many physiological and pathological processes generates a growing interest in understanding their regulation. Here, we demonstrate that the serine protease inhibitor plasminogen activator inhibitor 1 (PAI-1) forms an SDS-stable complex with the PC furin, which leads to the inhibition of the intra-Golgi activity of furin. It is known that elevated PAI-1 plasma levels are correlated with the occurrence of the metabolic syndrome and type 2 diabetes, and we show that PAI-1 reduces the furin-dependent maturation and activity of the insulin receptor and ADAM17: two proteins involved in the onset of these metabolic disorders. In addition to demonstrating that PAI-1 is an intracellular inhibitor of furin, this study also provides arguments in favor of an active role for PAI-1 in the development of metabolic disorders.

Key words: ADAM17, Furin, Golgi, Insulin receptor, PAI-1, Convertase

Introduction could impact upon many cellular processes. However, given the The proprotein convertase (PC) family, a class of subtilisin-kexin- explanatory potential of a hypothetical diabetogenic mechanism like Ca2+-dependent serine endoproteases, is responsible for the involving the inhibition of PCs by PAI-1 (Griffiths and Grainger, cleavage of numerous precursor proteins in mammalian cells 2006), we analyzed the effect of PAI-1 on the furin-dependent (Seidah and Chretien, 1999). Processing by PCs is a crucial step in maturation of the insulin receptor (IR) and ‘a disintegrin and the formation of the biologically active form of proteins synthesized metalloproteinase 17’ (ADAM17) (Peiretti et al., 2003; Robertson

Journal of Cell Science through the secretory pathway. Furin was the first mammalian PC et al., 1993; Srour et al., 2003; Williams et al., 1990) because both identified (Roebroek et al., 1986; van de Ven et al., 1990) and has of these proteins are crucially involved in the onset of diabetes. been the most extensively studied. Furin is expressed ubiquitously Herein, we show that PAI-1 reduces the maturation of IR and and cleaves proteins C-terminally to a typical basic amino acid ADAM17, indicating that PAI-1 could be actively involved in the sequence [Arg-Xaa-(Lys/Arg)-Arg] (Thomas, 2002). It is a development of insulin resistance. transmembrane protease predominantly localized in the trans Golgi network but it can cycle between the trans Golgi network and the Results cell surface (Molloy et al., 1999). Furin is regulated at the level of In vitro inhibition of recombinant furin by PAI-1 its expression (Guimont et al., 2007; Laprise et al., 2002) and The efficacy of PAI-1 to inhibit furin activity in vitro was compared maturation (Anderson et al., 1997; Gawlik et al., 2009). A direct with that of the potent furin inhibitor 1-antitrypsin Portland (1- inhibition of furin activity, through association with the cytosolic PDX), a genetically bioengineered form of serpin A1 (1- serine protease inhibitor (serpin) proteinase inhibitor 8, has also antitrypsin). The concentration of PAI-1 needed for half-maximal been described in human activated platelet releasates (Leblond et furin inhibition was determined to be 15-fold that of 1-PDX al., 2006). (IC50480 nM compared with IC5032 nM) (Fig. 1A). Because of its inhibitory action on plasminogen activators, It has been reported that inhibition of furin by 1-PDX and plasminogen activator inhibitor 1 (PAI-1) is considered to be a inhibition of plasminogen activators by PAI-1 both involve the major regulator of fibrinolysis. However, PAI-1 is potentially able formation of covalent SDS-stable complexes (Jean et al., 1998; to inhibit several other serine proteases; this inhibition can vary Levin, 1983; Thorsen et al., 1988). We examined whether such a from being marginal to being physiologically relevant, owing to a complex was formed between furin and PAI-1 in vitro with variable affinity of PAI-1 with its target (Keijer et al., 1991; Olson recombinant proteins. Furin and PAI-1 were detected by et al., 2001). PAI-1 is a secreted serpin and, as such, could inhibit immunoblotting at their expected molecular masses (80 kDa and serine proteases in the secretory pathway and in the extracellular 45 kDa, respectively). When PAI-1 and furin were coincubated, an milieu. additional immunoreactive band with a higher molecular mass We investigated whether PAI-1 could be a potent inhibitor of (~130 kDa) was detected with the anti-furin antibody and, less furin, and here we provide evidence that it associates with furin efficiently, with the anti-PAI-1 antibody (Fig. 1B), suggesting that and inhibits its activity in the Golgi complex of human cells. it results from the association of furin with PAI-1 in a SDS-stable Because of the wide variety of furin substrates, this regulation covalent complex. Furthermore, the molecular mass of this protein PAI-1 inhibits furin 1225

Fig. 2. Interaction of cell-expressed furin and PAI-1. (A) HeLa cells were transfected with empty plasmid (c) or PAI-1 and/or furin (Fur) expression vectors. Furin inhibitor (cmk) was added at 6 hours after cell transfection. After 48 hours, cells were lysed, glycoproteins purified and the expression of furin and PAI-1 was analyzed by immunoblotting (IB). Furin and PAI-1 were detected at 80–85 kDa and 45 kDa, respectively. The PAI-1–furin complex Fig. 1. PAI-1 interacts with furin in vitro. (A) Recombinant furin was was detected at ~130 kDa. The amount of complex was directly evaluated on total cell lysate by heterodimer PAI-1–furin immunoassay (Elisa PAI-1/Fur). incubated in vitro, with various concentrations of 1-PDX or recombinant PAI-1 (rPAI-1), at room temperature for 45 minutes, then furin activity was (B) Furin, PAI-1 and complex were analyzed in HepG2 cells as described in A, measured as described in the Materials and Methods section. Results are and left untreated (c), stimulated for 2 hours with PMA (200 nM) or the cells were transfected for 48 hours with control siRNA (siC), PAI-1 siRNA (siPAI- means+s.d. (n9). (B) Recombinant furin (rFur) and PAI-1 (rPAI-1) were 1) or furin siRNA (siFur) and then stimulated with PMA. *P<0.05 compared incubated in vitro in the absence or presence of furin inhibitor (cmk; 100 M). The sample mixture was analyzed with SDS-PAGE followed by with their respective control. Results are means+s.d. (n9). immunoblotting (IB) using the anti-furin antibody (left-hand panel) or the anti- PAI-1 antibody (right-hand panel). The upper panel shows a long exposure, The complex appeared as a doublet when total-cell lysates were and the lower panel shows a short exposure. Furin and PAI-1 were detected at 85 kDa and 45 kDa, respectively. The PAI-1–furin complex was detected at used for immunoblotting analysis (supplementary material Fig.

Journal of Cell Science 130 kDa. S2), but the slower migrating band of this doublet was the major form detected in glycoprotein preparations, suggesting that the doublet comprises differentially glycosylated forms of the two complex is in accordance with the sum of the furin and PAI-1 proteins. However, the differentially glycosylated forms of the molecular masses. Addition of the furin inhibitor decanoyl-Arg- complex can still be separated into two distinct forms, possibly Val-Lys-Arg-chloromethylketone (cmk) prevented the formation representing the complex containing cleaved and uncleaved PAI- of the complex, suggesting that furin activity is necessary for the 1, as has been described for the complex between PAI-1 and tissue complex formation. plasminogen activator (tPA) (Yamashita et al., 2006). Because hepatocytes are considered a major source of PAI-1 Interaction of cell-expressed PAI-1 and furin (Konkle et al., 1992), the association between endogenous furin PAI-1 and furin were overexpressed in HeLa cells, and the ability and PAI-1 was studied in the HepG2 hepatocyte cell line. of these two proteins to stably associate was analyzed by Stimulation of PAI-1 expression by phorbol 12-myristate 13-acetate immunoblotting. Overexpressed furin was detected as a doublet of (PMA) resulted in an increased amount of the PAI-1–furin complex, ~80–85 kDa, corresponding to the pro- and mature-form of the as judged by immunoblotting and heterodimer immunoassay enzyme (Leduc et al., 1992), and PAI-1 was detected at 45 kDa analysis (Fig. 2B). Knocking down either PAI-1 or furin expression (Fig. 2A). When both proteins were overexpressed together, a by small interfering RNA (siRNA) led to a decrease in the amount higher molecular mass (120–130 kDa) protein complex was clearly of the detected complex. These results indicate that endogenous detected with the anti-furin antibody. Detection of the complex PAI-1 and furin can indeed form a complex. with the anti-PAI-1 antibody was possible but relatively ineffective (supplementary material Fig. S1). The formation of this complex Intracellular interaction of PAI-1 and furin was partially prevented by the addition of the furin inhibitor cmk, We analyzed the intracellular interaction of PAI-1 and furin in the indicating that furin activity is involved in this interaction. To human LoVo cell line, which lacks furin (Takahashi et al., 1993) confirm unambiguously that the complex detected by and poorly expresses PAI-1. Immunoblotting and heterodimer immunoblotting comprised furin and PAI-1, and to compare immunoassay analysis confirmed the above results: concomitant precisely the amount of complex between treatments, we developed expression of PAI-1 and furin led to their association in a higher a PAI-1–furin heterodimer immunoassay. This heterodimer molecular mass complex (Fig. 3A). Considering that furin is mainly immunoassay completely confirmed the immunoblotting results. present in the Golgi network and at the plasma membrane, and that 1226 Journal of Cell Science 124 (8)

Fig. 3. The PAI-1–furin association occurs intracellularly. (A) LoVo cells were transfected with furin (Fur) and/or PAI-1 expression vectors. After 48 hours, expression of furin, PAI-1 and the PAI-1–furin complex was analyzed by immunoblotting (IB) and with the heterodimer PAI-1–furin immunoassay (Elisa PAI-1/Fur; means+s.d. for n9). (B) LoVo cells were transfected with empty (c) or furin-encoding plasmid (Fur). After 24 hours, recombinant PAI-1 (rPAI-1; 4.5 g/ml) was added. After an additional 24 hours, cells were lysed  Fig. 4. Subcellular localization of PAI-1 and furin. (A–C) HeLa cells and furin expression was analyzed by immunoblotting. (C) LoVo cells were  transfected with furin alone or (D–F) with furin and PAI-1 were analyzed by transfected with either empty plasmid (c) or the furin (Fur) or PAI-1 immunofluorescence microscopy for the detection of furin (A,D), golgin-97 expression vector. After 24 hours, cells were mixed and incubated for an (B) and PAI-1 (E). Colocalizations of furin and golgin-97 (C) and furin and additional 24 hours. As a positive control, cells co-transfected with plasmids PAI-1 (D) are represented by the yellow color in the ‘merge’ images. Nuclei encoding furin and PAI-1 (Fur+PAI-1) were co-cultured with empty plasmid- were visualized with DAPI staining (blue channel). Scale bars: 10 m. transfected cells. Cell lysates were then immunoblotted for furin detection.

PAI-1 is a secreted protein, we chose to determine the cellular furin was significantly, and comparably, inhibited by the wild-type

Journal of Cell Science compartment in which PAI-1 and furin associate. When furin- and a stabilized form of PAI-1 (–40%; P<0.05). To gauge overexpressing LoVo cells were incubated with recombinant wild- approximately the efficiency of the intracellular PAI-1 inhibition type PAI-1 (Fig. 3B) or stabilized PAI-1 (data not shown), the of furin, we compared its effect with that of 1-PDX; this PAI-1–furin complex was not detected in cell lysates. In addition, comparison can be considered reasonable because overexpressed when PAI-1-overexpressing LoVo cells were co-cultured with furin- PAI-1 and 1-PDX are necessarily present in large excess over the overexpressing cells, we did not find any PAI-1–furin complex endogenous furin. The result shows that 1-PDX was (Fig. 3C), indicating that PAI-1 released into the culture medium approximately twice as effective in inhibiting furin intracellularly does not associate with furin. Immunofluorescence microscopy compared with PAI-1 (Fig. 5A), suggesting that the efficiency of revealed that furin colocalized with the Golgi marker golgin-97 the PAI-1 inhibition of furin is substantial. Furthermore, (officially known as GOLGA1) and with PAI-1 (Fig. 4), which is overexpression of serpin C1 (antithrombin III) did not inhibit furin in accordance with previously reported observations indicating (Fig. 5A and inset), indicating that overexpression of a secreted that PAI-1 and furin are localized in the Golgi complex (Ma et al., serpin does not inevitably lead to intracellular furin inhibition. 2006; Shapiro et al., 1997). Our data suggest that interaction of In HepG2 cells, in which we identified an endogenous PAI-1– furin with PAI-1 takes place in the Golgi complex. furin complex (Fig. 2B), the furin activity was significantly increased by knocking down the expression of PAI-1 (Fig. 5B). Intra-Golgi inhibition of furin by PAI-1 This result indicates that, in this cell line, endogenous PAI-1 We next investigated whether the association of PAI-1 with furin represses furin activity, emphasizing the physiological relevance actually resulted in reduction of its enzymatic activity. To monitor of the inhibition of furin by PAI-1. intra-Golgi furin activity, we transfected HeLa cells with the GRAPfurin reporter construct, which encodes an alkaline Inhibition of furin-dependent processing by PAI-1 phosphatase (AP) linked to a Golgi retention signal by the typical To consider PAI-1 an important regulator of furin, its expression furin-cleavage sequence Arg-Gln-Lys-Arg (Coppola et al., 2007). should sufficiently reduce the furin-dependent maturation of AP activity was detected in conditioned culture medium from physiological substrates such that cellular responses are altered. GRAPfurin-transfected cells, reflecting endogenous intra-Golgi We, therefore, analyzed the effect of PAI-1 on the furin-dependent furin activity (Fig. 5A). As expected, AP activity was increased by maturation of the IR and ADAM17. furin overexpression (+49%) and almost completely abrogated by Consistent with their furin-deficient phenotype, LoVo cells the addition of the furin inhibitor cmk. The activity of endogenous exhibited poor endogenous IR maturation (Fig. 6A). Furin PAI-1 inhibits furin 1227

Fig. 6. PAI-1 inhibits furin-dependent insulin receptor signaling. (A) LoVo Fig. 5. PAI-1 inhibits intra-Golgi furin activity. (A) HeLa cells were cells were transfected with empty plasmid (c), PAI-1 and/or furin (Fur) transfected with the GRAPfurin expression vector together with empty vector expression vectors. After 48 hours, cells were stimulated for 5 minutes with (c), wild-type PAI-1, stabilized PAI-1 (sPAI-1), 1-PDX, serpin C1 (ATIII) or insulin (100 nM) and lysed. Purified glycoproteins were immunoblotted (IB) furin (Fur) expression vectors. Furin inhibitor (cmk) was added at 6 hours after for IR, Akt and phosphorylated Akt. IR phosphorylation was measured in total cell transfection. Inset: immunoblot demonstrating the overexpression of cell lysates by ELISA (Elisa pIR). (B) HeLa cells were co-transfected with IR serpin C1. (B) HepG2 cells were co-transfected with the GRAPfurin expression vector and empty plasmid (c) or PAI-1 expression vector. After 48 expression vector (Grap) and control siRNA (siC) or PAI-1 siRNA (siPAI-1). hours, cells were stimulated for 10 minutes with insulin (100 nM), and IR and AP activity was assayed in the culture medium as described in the Materials Akt were analyzed as described in A. The IR proform is detected at 190 kDa and Methods and expressed as arbitrary units (AU). *P<0.05 compared with and its mature form at 90 kDa. *P<0.05 compared with cells transfected with ¶ cells only transfected with GRAPfurin (c); P<0.05 compared with 1-PDX- furin. P<0.05 compared with cells transfected with empty plasmid stimulated transfected cells. Results are means+s.d. (n9). with insulin. Results are means+s.d. (n9). Journal of Cell Science

overexpression greatly increased the amount of endogenous mature overexpression of TNF-R2 (Fig. 7B) and this release was not IR, and concomitant expression of PAI-1 reduced this increase altered by coexpression of PAI-1. Furin overexpression drastically (Fig. 6A), confirming that PAI-1 reduces IR maturation by increased TNF-R2 release, and concomitant expression of furin inhibiting furin. As expected, the PAI-1-dependent reduction in IR and PAI-1 significantly reduced TNF-R2 shedding when compared maturation decreased the amount of phosphorylated IR and Akt with that caused by furin alone (–27%; P<0.05). To ensure that the measured after insulin stimulation. However, in furin- observed effect was due to an alteration of a shedding process, we overexpressing LoVo cells, the insulin-stimulated increase in the overexpressed a spontaneously secreted form of TNF-R2 (TNF- phosphorylation of IR was somewhat marginal (+20%). Therefore, R2mut) and observed that neither furin nor PAI-1 modified the to demonstrate clearly the ability of PAI-1 to interfere with IR ADAM17-unrelated continuous release of TNF-R2mut (Fig. 7C). phosphorylation and insulin signaling, the experiment was repeated in HeLa cells overexpressing human IR. Overexpression of PAI-1 Discussion greatly reduced the maturation of coexpressed IR, which strongly Our results demonstrate that PAI-1 inhibits furin in the secretory reduced the amount of phosphorylated IR (–73%) and the pathway. Both proteins colocalize in the Golgi, where furin is phosphorylation of endogenous Akt, following stimulation by predominantly, if not exclusively, localized (Gawlik et al., 2010) insulin (Fig. 6B). These results show that PAI-1, by inhibiting the and where PAI-1 is detected when its expression increases (Ma et furin-dependent maturation of IR, alters insulin signaling. al., 2006). Association of PAI-1 with furin leads to the formation The mature form of ADAM17 was modestly represented in of a higher molecular mass SDS-stable complex and a decrease in LoVo cells (Fig. 7A) when compared with that of its proform. intra-Golgi furin-dependent maturation processes. This is the first Furin overexpression increased the maturation of ADAM17, and demonstration of an intracellular regulation of furin activity by a PAI-1 expression diminished this effect. To assess ADAM17 physiological serpin in mammalian cells. However, serpin-mediated activity, we measured the accumulation of TNF-R2 (tumor necrosis inhibition of furin has been suggested previously; indeed, the best- factor receptor 2; also known as TNFRSF1B) in the culture characterized furin inhibitor is a genetically engineered variant of medium, as this results from an ADAM17-dependent shedding 1-antitrypsin (1-PDX). 1-PDX forms a stable complex with process (Reddy et al., 2000; Solomon et al., 1999). Soluble TNF- furin and inhibits processing of precursors within the constitutive R2 was detected in the culture medium of LoVo cells upon secretory pathway (Benjannet et al., 1997). We evaluated that the 1228 Journal of Cell Science 124 (8)

none of these serpins possess the minimal furin cleavage site represented by arginine residues at positions P1 and P4 [notation of Schechter and Berger (Schechter and Berger, 1967)] (supplementary material Table S1). However, PAI-1 and some serpins possess an arginine residue at position P1, which is mandatory, but not sufficient, for an efficient cleavage by furin. Indeed, 1-antitrypsin (serpin A1) does not inhibit furin, whereas its variant form with an arginine at position P1 (1-antitrypsin Pittsburg) inhibits furin (Oda et al., 1992; Wasley et al., 1993). However, introduction of an arginine residue into position P4 of 1-antitrypsin Pittsburg (which gives 1-PDX) tremendously increases the inhibition efficiency (Anderson et al., 1993). Among the serpins with an arginine residue at position P1, which are most similar to PAI-1, are serpin E2 (protease nexin I) and serpin C1 (antithrombin III). As PAI-1 and serpin E2 both inhibit plasminogen activators it is probable that serpin E2 also inhibits furin. However, results obtained with the GRAPfurin assay indicate that serpin C1 does not inhibit intracellular furin, which suggests that other elements, in addition to the arginine residue at position P1, have to be considered when explaining why a serpin can inhibit furin. In line with this suggestion, it was shown that serpin C1 only becomes a furin inhibitor upon activation by heparin (Brennan and Nakayama, 1994), which involves modification of the conformation of its reactive-site-loop (Langdown et al., 2004). As an acute-phase protein, PAI-1 is rapidly produced and Fig. 7. PAI-1 inhibits furin-dependent ADAM17 maturation and activity. released by cells in a number of clinical situations. Consequently, (A) LoVo cells were transfected as in Fig. 6A and the expression of ADAM17 a potential furin-dependent role for PAI-1 should be studied in was analyzed by immunoblotting (IB). The proform of ADAM17 was detected pathologic contexts in which PAI-1 synthesis is increased. The at 110 kDa, and its mature form at 90 kDa. (B) LoVo cells were transfected circulating levels of PAI-1 are strongly correlated with the metabolic with TNF-R2 and empty plasmid (c) or Furin (Fur) and/or PA-1. (C) As in B syndrome (Alessi and Juhan-Vague, 2008; Alessi et al., 2007), a except that a truncated form of TNF-R2 (TNF-R2mut) was transfected, which cluster of defined metabolic disorders predisposing individuals to generates only the released soluble form of the receptor. TNF-R2 in cell lysates and in the extracellular medium was quantified by ELISA, and its cardiovascular disease and type 2 diabetes. Recently, a hypothetical shedding (as an indicator of ADAM17 activity) is presented as the ratio of diabetogenic mechanism involving the inhibition of PCs by PAI-1 TNF-R2 in the supernatant relative to that in the cell lysate. P<0.05 compared was suggested (Griffiths and Grainger, 2006), and our present

Journal of Cell Science with cells transfected with furin. Results are means+s.d. (n9). results, showing that PAI-1 is a furin inhibitor, led us to study the effect of PAI-1 on the maturation of IR and ADAM17, two furin substrates crucially implicated in the onset of the metabolic inhibition of furin in the secretory pathway by PAI-1 is half as syndrome. We demonstrated that PAI-1 inhibits furin-dependent effective as that of 1-PDX. The intracellular inhibition of furin in processing of IR, which reduces the amount of phosphorylated IR the secretory pathway by a physiological serpin has also been and, hence, subsequent Akt phosphorylation. Consequently, through shown with the Drosophila melanogaster serpin Spn4A (Oley et furin inhibition, PAI-1 impairs the cellular insulin response, which al., 2004). Up to now, serpin proteinase inhibitor 8 has been the could contribute to development of insulin resistance. However, a only naturally occurring furin inhibitor described in mammals further characterization of the metabolic phenotype of the cell will (Dahlen et al., 1998). However, probably because of its cytoplasmic be necessary in order to make definitive conclusions about the location, proteinase inhibitor 8 does not interact with furin physiological consequences of such effects. Several studies provide intracellularly, and its ability to inhibit furin in a physiological arguments for suspecting the involvement of PAI-1 in metabolic context has only been demonstrated in activated platelet releasates disorders, such as obesity and insulin resistance. In mice, PAI-1 (Leblond et al., 2006). deficiency leads to reduced weight and better insulin sensitivity PAI-1 is a serpin capable of an unusual transition from an active (Ma et al., 2004). However, although PAI-1 expression can impair to a latent form (Dupont et al., 2009). However, it has been glucose uptake, especially in adipocytes (Liang et al., 2006), its previously observed that the active form of PAI-1 is stabilized direct participation in the development of insulin resistance has under acidic conditions, such as those in the lumen of the trans been questioned (Schafer et al., 2001) and might be linked to an Golgi network (Kvassman et al., 1995), which could explain why effect on the inflammation state. Interestingly, the furin-dependent we observed that overexpression of stabilized or wild-type PAI-1 processing of the key inflammatory enzyme ADAM17 is also leads to comparable intracellular furin inhibition. These data also reduced by PAI-1, resulting in a decrease in its activity. ADAM17 suggest that the microenvironment in the Golgi complex provides is a metalloproteinase primarily described as the enzyme optimal conditions for the inhibition of furin by PAI-1. responsible for TNF- shedding from the cell surface (Black et al., It is conceivable that PAI-1 is not the only serpin capable of 1997; Moss et al., 1997). The involvement of ADAM17 in the inhibiting furin. However, the percentage identity between the TNF-mediated inflammatory response is reinforced by its sequences of the reactive-site-loop of human extracellular serpins participation in the cleavage of the TNF receptors TNF-R1 and with known inhibition function and that of PAI-1 is relatively low; TNF-R2 (Reddy et al., 2000; Solomon et al., 1999). ADAM17 is PAI-1 inhibits furin 1229

also implicated in the development of obesity and insulin resistance spectrofluorimeter reader (TECAN Infinite 200; TECAN Austria). Data were derived from the initial linear rates of substrate hydrolysis. For the association assay, (Serino et al., 2007; Xu et al., 2002a; Xu et al., 2002b). The recombinant furin and PAI-1 were coincubated at a molar ratio of 1:10. The reaction emerging concept is that the synthesis of TNF- increases in was stopped by the addition of SDS-PAGE sample buffer, and the samples were hypertrophied adipose tissue, whereas its processing rate by immediately analyzed by immunoblotting with antibodies against furin or PAI-1. ADAM17 decreases, resulting in elevated transmembrane TNF- ,  Immunofluorescence microscopy which strongly inhibits insulin action at this site. Our results Cells cultured on microscope chambers (Labtek) were treated to enable suggest that the increased synthesis of PAI-1 by adipose tissue in immunofluorescence detection of furin, golgin-97 and PAI-1. Briefly, cells were obese individuals (Alessi et al., 1997) could be responsible for the washed, fixed on ice with 4% paraformaldehyde for 10 minutes, permeabilized with 0.1% Triton X-100 in PBS and saturated for unspecific binding with 2% BSA for 30 local decrease in TNF- processing. minutes. After 1 hour of incubation with primary antibody, cells were incubated with Considering the two furin substrates analyzed in this work, it the appropriate secondary antibody coupled to Alexa-Fluor-488 or Alexa-Fluor-546. appears that the inhibition of furin by PAI-1 could have a direct Finally, cells were mounted in medium containing DAPI and visualized using a ϫ role in the onset of metabolic disorders in which high levels of microscope (Leica DMRB with a Plan 60 objective). Images were captured using a camera (ProgRes CFcool; Jenoptik) and ProgRes Capture Pro software (version PAI-1 are observed. Taken as a whole, our results show that highly 2.6). Images were prepared for presentation using ImageJ software with minimal expressed PAI-1 can directly impair intracellular processing by processing. furin, and they provide a molecular basis for the emerging concept Immunoblotting of PAI-1 being more than just a marker of metabolic disorders such Total cell lysates or glycoproteins purified from cell lysates, isolated as previously as obesity and the metabolic syndrome (Alessi et al., 2007; Griffiths described (Peiretti et al., 2003), were boiled for denaturation. Then the proteins were and Grainger, 2006). However, considering the wide variety of separated by SDS-PAGE under reducing conditions and transferred onto polyvinylidene fluoride membranes (Millipore). Membranes were blocked for 1 furin substrates, it is evident that its regulation by PAI-1 could hour in 5% (wt/vol.) skimmed-milk powder and proteins were immunodetected impact upon a board spectrum of biological processes. Furthermore, using appropriate antibodies. the high similarity between the catalytic domains of the PCs ELISAs suggests that PAI-1 could inhibit several other PCs and not just For the PAI-1–furin heterodimer immunoassay, the wells of a microtiter plate were furin. coated overnight at 4°C with 2 g/ml anti-PAI-1 capture antibody (15H12). After two washes, the remaining protein-binding sites were blocked for 2 hours with a Materials and Methods solution of 5% BSA in PBS. The cell lysates, diluted five times, were added to each Reagents well and incubated overnight at 4°C. After two washes, biotinylated anti-furin Decanoyl-Arg-Val-Lys-Arg-chloromethylketone (cmk), a furin inhibitor, and the antibody (0.25 g/ml) was added for 2 hours. Detection was performed by incubating internally quenched fluorescent furin substrate butoxycarbonyl-Arg-Val-Arg-Arg-7- the lysates with streptavidin-conjugated horseradish peroxidase (HRP) for 1 hour amido-4-methylcoumarine (Boc-RVRR-AMC) were purchased from Bachem followed by addition of TMB solution. The reaction was stopped by adding H2SO4 Bioscience. p-Nitrophenyl phosphate was from Sigma–Aldrich. Recombinant furin (1 M final concentration), and the optical density was read at 450 nm. The levels of human TNF-R2 in the conditioned medium and cell lysate were assayed using an was obtained from R&D Systems. Recombinant human bioengineered 1-antitrypsin Portland (1-PDX) was from Calbiochem. Recombinant PAI-1 (Vleugels et al., ELISA kit from R&D Systems, according to the manufacturer’s protocol. 2000) and anti-PAI-1 monoclonal antibodies for immunofluorescence microscopy Phosphorylated insulin receptor was quantified with the human phosphorylated (33H1F7) and enzyme-linked immunosorbent assay (ELISA) coating (15H12) were insulin receptor ELISA kit (R&D Systems), according to the manufacturer’s protocol. kindly provided by Paul Declerck (University of Leuven, Leuven, Belgium). Anti- furin monoclonal (MON-152) and biotinylated polyclonal antibodies were from GRAPfurin alkaline phosphatase assay Alexis Biochemicals and R&D Systems, respectively. Anti-PAI-1 monoclonal and After transient transfection of the GRAPfurin vector (Coppola et al., 2007), cell medium was replaced with uncoloured DMEM for 5 hours. The medium was then Journal of Cell Science anti-(insulin receptor) polyclonal antibodies were from Santa Cruz Biotechnology. The anti-ADAM17 polyclonal antibody was from Chemicon International. Antibodies centrifuged at 20,000 g, mixed with the alkaline phosphatase substrate p-nitrophenyl against Akt and phosphorylated Akt were from Cell Signaling. The anti-golgin-97 phosphate in a 96-well plate and incubated for 1 hour, to allow color development, monoclonal antibody was from Molecular Probes. and absorbance was read at 405 nm. For each sample, the absorbance value was normalized to the total protein content of cell lysate [(absorbance of supernatant over Expression vectors the protein content of transfected cell)–(absorbance of supernatant over the protein ϫ The PAI-1 coding sequence was inserted into vector pcDNA3 (Invitrogen). The content of untransfected cell) 1000]. Results are expressed in arbitrary units. human furin and TNF-R2 expression vectors (pcDNA3) were as previously described (Peiretti et al., 2003). To generate the soluble N-terminal form of TNF-R2 (TNF- Statistics R2mut; amino acids 1–248 of the extracellular domain), the TNF-R2 expression All experiments were performed at least in triplicate and at least three times (unless vector was digested with ApaI and then re-ligated. The plasmid encoding the specified in figure legends). Results are expressed as means±s.d., and statistical GRAPfurin construct was generously provided by Christin Hamilton and Alnawaz analysis was performed using Student’s t-tests with GraphPad Prism software. Rehemtulla (University of Michigan Medical School, Ann Arbor, MI) (Coppola et Significance was taken as at least P<0.05. al., 2007). Expression vectors for human insulin receptor, human antithrombin III and 1-PDX were kindly provided by Tarik Isaad (Institut Cochin, Paris, France), We thank Christin Hamilton and Alnawaz Rehemtulla for providing Delphine Borgel (Université Paris Sud, Paris France) and Gary Thomas (Oregon the GRAPfurin expression vector, Paul Declerck for providing Health & Science University, Portland, OR), respectively. recombinant PAI-1 and anti-PAI-1 monoclonal antibodies, and Tarik Cell culture and transfection Isaad, Delphine Borgel and Gary Thomas for sharing expression The LoVo cell line was cultured as previously described (Peiretti et al., 2003). The vectors for human insulin receptor, human antithrombin III and 1- HeLa cell line was cultured in minimal essential medium (MEM) supplemented with PDX, respectively. This work was supported by grants from Agence 10% fetal bovine serum (FBS). Transient transfections of LoVo and HeLa cells were Nationale de la Recherche (ANR-07-PHYSIO-019-01) and from the performed with ICAFectin 441 (Eurogentec), as specified by the manufacturer. For Fondation pour la Recherche Médicale (DEQ20071210508). all co-transfections, empty plasmid was used to keep the amount of DNA constant. The HepG2 cell line was cultured in MEM and Ham’s F-12 mixed medium Supplementary material available online at supplemented with 10% FBS, 100 units/ml penicillin and 100 g/ml streptomycin. http://jcs.biologists.org/cgi/content/full/124/8/1224/DC1 HepG2 cells were transfected with siRNA (Eurogentec) using Lipofectamine 2000 (Invitrogen), according to the manufacturer’s instructions. References Alessi, M. C. and Juhan-Vague, I. (2008). Metabolic syndrome, haemostasis and thrombosis. In vitro furin activity and association assays Thromb. Haemost. 99, 995-1000. Recombinant furin (0.021 g/ml) was incubated in 96-well microplates, with various Alessi, M. C., Peiretti, F., Morange, P., Henry, M., Nalbone, G. and Juhan-Vague, I. concentrations of inhibitors, at room temperature for 45 minutes in buffer (25 mM (1997). Production of plasminogen activator inhibitor 1 by human adipose tissue: possible Tris-HCl pH 9, containing 0.5% Triton X-100 and 1 mM CaCl2). Then Boc-RVRR- link between visceral fat accumulation and vascular disease. Diabetes 46, 860-867. AMC was added (10 M final concentration) and fluorescence was measured Alessi, M. C., Poggi, M. and Juhan-Vague, I. (2007). Plasminogen activator inhibitor-1, (excitation at 380 nm and emission at 460 nm) every 30 seconds with a microplate adipose tissue and insulin resistance. Curr. Opin. Lipidol. 18, 240-245. 1230 Journal of Cell Science 124 (8)

Anderson, E. D., Thomas, L., Hayflick, J. S. and Thomas, G. (1993). Inhibition of HIV- Molloy, S. S., Anderson, E. D., Jean, F. and Thomas, G. (1999). Bi-cycling the furin 1 gp160-dependent membrane fusion by a furin-directed alpha 1-antitrypsin variant. J. pathway: from TGN localization to pathogen activation and embryogenesis. Trends Cell Biol. Chem. 268, 24887-24891. Biol. 9, 28-35. Anderson, E. D., VanSlyke, J. K., Thulin, C. D., Jean, F. and Thomas, G. (1997). Moss, M. L., Jin, S. L., Milla, M. E., Bickett, D. M., Burkhart, W., Carter, H. L., Chen, Activation of the furin endoprotease is a multiple-step process: requirements for W. J., Clay, W. C., Didsbury, J. R., Hassler, D. et al. (1997). Cloning of a disintegrin acidification and internal propeptide cleavage. EMBO J. 16, 1508-1518. metalloproteinase that processes precursor tumour-necrosis factor-alpha. Nature 385, 733- Benjannet, S., Savaria, D., Laslop, A., Munzer, J. S., Chretien, M., Marcinkiewicz, M. 736. and Seidah, N. G. (1997). Alpha1-antitrypsin Portland inhibits processing of precursors Oda, K., Misumi, Y., Ikehara, Y., Brennan, S. O., Hatsuzawa, K. and Nakayama, K. mediated by proprotein convertases primarily within the constitutive secretory pathway. (1992). Proteolytic cleavages of proalbumin and complement Pro-C3 in vitro by a J. Biol. Chem. 272, 26210-26218. truncated soluble form of furin, a mammalian homologue of the yeast Kex2 protease. Black, R. A., Rauch, C. T., Kozlosky, C. J., Peschon, J. J., Slack, J. L., Wolfson, M. F., Biochem. Biophys. Res. Commun. 189, 1353-1361. Castner, B. J., Stocking, K. L., Reddy, P., Srinivasan, S. et al. (1997). A Oley, M., Letzel, M. C. and Ragg, H. (2004). Inhibition of furin by serpin Spn4A from metalloproteinase disintegrin that releases tumour-necrosis factor-alpha from cells. Nature Drosophila melanogaster. FEBS Lett. 577, 165-169. 385, 729-733. Olson, S. T., Swanson, R., Day, D., Verhamme, I., Kvassman, J. and Shore, J. D. (2001). Brennan, S. O. and Nakayama, K. (1994). Furin has the proalbumin substrate specificity Resolution of Michaelis complex, acylation, and conformational change steps in the and serpin inhibitory properties of an in situ hepatic convertase. FEBS Lett. 338, 147- reactions of the serpin, plasminogen activator inhibitor-1, with tissue plasminogen activator 151. and trypsin. Biochemistry 40, 11742-11756. Coppola, J. M., Hamilton, C. A., Bhojani, M. S., Larsen, M. J., Ross, B. D. and Peiretti, F., Canault, M., Deprez-Beauclair, P., Berthet, V., Bonardo, B., Juhan-Vague, Rehemtulla, A. (2007). Identification of inhibitors using a cell-based assay for monitoring I. and Nalbone, G. (2003). Intracellular maturation and transport of tumor necrosis factor Golgi-resident protease activity. Anal. Biochem. 364, 19-29. alpha converting enzyme. Exp. Cell Res. 285, 278-285. Dahlen, J. R., Jean, F., Thomas, G., Foster, D. C. and Kisiel, W. (1998). Inhibition of Reddy, P., Slack, J. L., Davis, R., Cerretti, D. P., Kozlosky, C. J., Blanton, R. A., Shows, soluble recombinant furin by human proteinase inhibitor 8. J. Biol. Chem. 273, 1851- D., Peschon, J. J. and Black, R. A. (2000). Functional analysis of the domain structure 1854. of tumor necrosis factor-alpha converting enzyme. J. Biol. Chem. 275, 14608-14614. Dupont, D. M., Madsen, J. B., Kristensen, T., Bodker, J. S., Blouse, G. E., Wind, T. and Robertson, B. J., Moehring, J. M. and Moehring, T. J. (1993). Defective processing of Andreasen, P. A. (2009). Biochemical properties of plasminogen activator inhibitor-1. the insulin receptor in an endoprotease-deficient Chinese hamster cell strain is corrected Front. Biosci. 14, 1337-1361. by expression of mouse furin. J. Biol. Chem. 268, 24274-24277. Gawlik, K., Shiryaev, S. A., Zhu, W., Motamedchaboki, K., Desjardins, R., Day, R., Roebroek, A. J., Schalken, J. A., Leunissen, J. A., Onnekink, C., Bloemers, H. P. and Remacle, A. G., Stec, B. and Strongin, A. Y. (2009). Autocatalytic activation of the furin Van de Ven, W. J. (1986). Evolutionary conserved close linkage of the c-fes/fps proto- zymogen requires removal of the emerging enzyme’s N-terminus from the active site. oncogene and genetic sequences encoding a receptor-like protein. EMBO J. 5, 2197-2202. PLoS One 4, e5031. Schafer, K., Fujisawa, K., Konstantinides, S. and Loskutoff, D. J. (2001). Disruption of Gawlik, K., Remacle, A. G., Shiryaev, S. A., Golubkov, V. S., Ouyang, M., Wang, Y. and the plasminogen activator inhibitor 1 gene reduces the adiposity and improves the Strongin, A. Y. (2010). A femtomol range FRET biosensor reports exceedingly low levels metabolic profile of genetically obese and diabetic ob/ob mice. FASEB J. 15, 1840-1842. of cell surface furin: implications for the processing of anthrax protective antigen. PLoS Schechter, I. and Berger, A. (1967). On the size of the active site in proteases. I. Papain. One 5, e11305. Biochem. Biophys. Res. Commun. 27, 157-162. Griffiths, S. L. and Grainger, D. J. (2006). Proposal of a novel diabetogenic mechanism Seidah, N. G. and Chretien, M. (1999). Proprotein and prohormone convertases: a family involving the serpin PAI-1. BioEssays 28, 629-641. of subtilases generating diverse bioactive polypeptides. Brain Res. 848, 45-62. Guimont, P., Grondin, F. and Dubois, C. M. (2007). Sox9-dependent transcriptional Serino, M., Menghini, R., Fiorentino, L., Amoruso, R., Mauriello, A., Lauro, D., regulation of the proprotein convertase furin. Am. J. Physiol. Cell Physiol 293, C172- Sbraccia, P., Hribal, M. L., Lauro, R. and Federici, M. (2007). Mice heterozygous for C183. tumor necrosis factor-alpha converting enzyme are protected from obesity-induced insulin Jean, F., Stella, K., Thomas, L., Liu, G., Xiang, Y., Reason, A. J. and Thomas, G. resistance and diabetes. Diabetes 56, 2541-2546. (1998). alpha1-Antitrypsin Portland, a bioengineered serpin highly selective for Shapiro, J., Sciaky, N., Lee, J., Bosshart, H., Angeletti, R. H. and Bonifacino, J. S. furin: application as an antipathogenic agent. Proc. Natl. Acad. Sci. USA 95, 7293- (1997). Localization of endogenous furin in cultured cell lines. J. Histochem. Cytochem. 7298. 45, 3-12. Keijer, J., Linders, M., Wegman, J. J., Ehrlich, H. J., Mertens, K. and Pannekoek, H. Solomon, K. A., Pesti, N., Wu, G. and Newton, R. C. (1999). Cutting edge: a dominant (1991). On the target specificity of plasminogen activator inhibitor 1, the role of heparin, negative form of TNF-alpha converting enzyme inhibits proTNF and TNFRII secretion. vitronectin, and the reactive site. Blood 78, 1254-1261. J. Immunol. 163, 4105-4108. Konkle, B. A., Schuster, S. J., Kelly, M. D., Harjes, K., Hassett, D. E., Bohrer, M. and Srour, N., Lebel, A., McMahon, S., Fournier, I., Fugere, M., Day, R. and Dubois, C. M.

Journal of Cell Science Tavassoli, M. (1992). Plasminogen activator inhibitor-1 messenger RNA expression is (2003). TACE/ADAM-17 maturation and activation of sheddase activity require proprotein induced in rat hepatocytes in vivo by dexamethasone. Blood 79, 2636-2642. convertase activity. FEBS Lett. 554, 275-283. Kvassman, J. O., Lawrence, D. A. and Shore, J. D. (1995). The acid stabilization of Takahashi, S., Kasai, K., Hatsuzawa, K., Kitamura, N., Misumi, Y., Ikehara, Y., plasminogen activator inhibitor-1 depends on protonation of a single group that affects Murakami, K. and Nakayama, K. (1993). A mutation of furin causes the lack of loop insertion into beta-sheet A. J. Biol. Chem. 270, 27942-27947. precursor-processing activity in human colon carcinoma LoVo cells. Biochem. Biophys. Langdown, J., Johnson, D. J., Baglin, T. P. and Huntington, J. A. (2004). Allosteric Res. Commun. 195, 1019-1026. activation of antithrombin critically depends upon hinge region extension. J. Biol. Chem. Thomas, G. (2002). Furin at the cutting edge: from protein traffic to embryogenesis and 279, 47288-47297. disease. Nat. Rev. Mol. Cell Biol. 3, 753-766. Laprise, M. H., Grondin, F., Cayer, P., McDonald, P. P. and Dubois, C. M. (2002). Furin Thorsen, S., Philips, M., Selmer, J., Lecander, I. and Astedt, B. (1988). Kinetics of gene (fur) regulation in differentiating human megakaryoblastic Dami cells: involvement inhibition of tissue-type and urokinase-type plasminogen activator by plasminogen- of the proximal GATA recognition motif in the P1 promoter and impact on the maturation activator inhibitor type 1 and type 2. Eur. J. Biochem. 175, 33-39. of furin substrates. Blood 100, 3578-3587. van de Ven, W. J., Voorberg, J., Fontijn, R., Pannekoek, H., van den Ouweland, A. M., Leblond, J., Laprise, M. H., Gaudreau, S., Grondin, F., Kisiel, W. and Dubois, C. M. van Duijnhoven, H. L., Roebroek, A. J. and Siezen, R. J. (1990). Furin is a subtilisin- (2006). The serpin proteinase inhibitor 8, an endogenous furin inhibitor released from like proprotein processing enzyme in higher eukaryotes. Mol. Biol. Rep. 14, 265-275. human platelets. Thromb. Haemost. 95, 243-252. Vleugels, N., Leys, J., Knockaert, I. and Declerck, P. J. (2000). Effect of stabilizing versus Leduc, R., Molloy, S. S., Thorne, B. A. and Thomas, G. (1992). Activation of human furin destabilizing interactions on plasminogen activator inhibitor-1. Thromb. Haemost. 84, precursor processing endoprotease occurs by an intramolecular autoproteolytic cleavage. 871-875. J. Biol. Chem. 267, 14304-14308. Wasley, L. C., Rehemtulla, A., Bristol, J. A. and Kaufman, R. J. (1993). PACE/furin can Levin, E. G. (1983). Latent tissue plasminogen activator produced by human endothelial process the vitamin K-dependent pro-factor IX precursor within the secretory pathway. J. cells in culture: evidence for an enzyme-inhibitor complex. Proc. Natl. Acad. Sci. USA Biol. Chem. 268, 8458-8465. 80, 6804-6808. Williams, J. F., McClain, D. A., Dull, T. J., Ullrich, A. and Olefsky, J. M. (1990). Liang, X., Kanjanabuch, T., Mao, S. L., Hao, C. M., Tang, Y. W., Declerck, P. J., Hasty, Characterization of an insulin receptor mutant lacking the subunit processing site. J. Biol. A. H., Wasserman, D. H., Fogo, A. B. and Ma, L. J. (2006). Plasminogen activator Chem. 265, 8463-8469. inhibitor-1 modulates adipocyte differentiation. Am. J. Physiol. Endocrinol. Metab. 290, Xu, H., Hirosumi, J., Uysal, K. T., Guler, A. D. and Hotamisligil, G. S. (2002a). Exclusive E103-E113. action of transmembrane TNF alpha in adipose tissue leads to reduced adipose mass and Ma, G. M., Halayko, A. J., Stelmack, G. L., Zhu, F., Zhao, R., Hillier, C. T. and Shen, local but not systemic insulin resistance. Endocrinology 143, 1502-1511. G. X. (2006). Effects of oxidized and glycated low-density lipoproteins on transcription Xu, H., Uysal, K. T., Becherer, J. D., Arner, P. and Hotamisligil, G. S. (2002b). Altered and secretion of plasminogen activator inhibitor-1 in vascular endothelial cells. Cardiovasc. tumor necrosis factor-alpha (TNF-alpha) processing in adipocytes and increased Pathol. 15, 3-10. expression of transmembrane TNF-alpha in obesity. Diabetes 51, 1876-1883. Ma, L. J., Mao, S. L., Taylor, K. L., Kanjanabuch, T., Guan, Y., Zhang, Y., Brown, N. Yamashita, T., Dai, Q., Suzuki, Y., Ihara, H., Mogami, H., Koide, T., Castellino, F. and J., Swift, L. L., McGuinness, O. P., Wasserman, D. H. et al. (2004). Prevention of Urano, T. (2006). Plasminogen activator inhibitor 1 forms a stable non-acyl enzyme obesity and insulin resistance in mice lacking plasminogen activator inhibitor 1. Diabetes inhibitor complex with tissue plasminogen activator. J. Thromb. Haemost. 4, Supplement 53, 336-346. S1, 165-165.