Pro-cathepsin D interacts with the extracellular domain of the β chain of LRP1 and promotes LRP1-dependent fibroblast outgrowth

Tumor progression has been recognized as the product of evolving crosstalk between tumor cells and the surrounding supportive tissue, known as the tumor stroma (Mueller and Fusenig, 2004). Cancer cells interact dynamically with several normal cell types within the extracellular matrix, such as fibroblasts, infiltrating immune cells, endothelial cells and adipocytes (Mueller and Fusenig, 2004). Stromal and tumor cells exchange enzymes, growth factors and cytokines that modify the local extracellular matrix, stimulate migration and invasion, and promote the proliferation and survival of stromal and tumor cells (Liotta and Kohn, 2001; Masson et al., 1998).

The aspartic protease cathepsin D (cath-D) is overexpressed and highly secreted by human epithelial breast cancer cells (Capony et al., 1989; Liaudet-Coopman et al., 2006; Rochefort and Liaudet-Coopman, 1999; Vignon et al., 1986). Its overexpression in breast cancer is correlated with poor prognosis (Ferrandina et al., 1997; Foekens et al., 1999; Rodriguez et al., 2005). Human cath-D is synthesized as a 52-kDa precursor that is rapidly converted in the endosomes to form an active, 48-kDa, single-chain intermediate, and then in the lysosomes into the fully active mature protease, composed of a 34-kDa heavy chain and a 14-kDa light chain. The human cath-D catalytic site includes two crucial aspartic residues (amino acids 33 and 231) located on the 14-kDa and 34-kDa chains, respectively (Metcalf and Fusek, 1993). The overexpression of cath-D in breast cancer cells leads to the hypersecretion of the 52-kDa pro-cath-D into the extracellular environment (Heylen et al., 2002; Laurent-Matha et al., 1998; Vignon et al., 1986). Secreted pro-cath-D can be endocytosed by both cancer cells and fibroblasts (Heylen et al., 2002; Laurent-Matha et al., 1998; Vignon et al., 1986), mainly via the mannose-6-phosphate receptors (M6PR) or by alternative cath-D receptors (Capony et al., 1994; Laurent-Matha et al., 2005; Rijnboutt et al., 1991). Cath-D affects both cancer cells and stromal cells in the tumor microenvironment by increasing cancer cell proliferation (Vignon et al., 1986; Glondu et al., 2001; Glondu et al., 2002; Berchem et al., 2002), metastasis (Glondu et al., 2002) and angiogenesis (Berchem et al., 2002; Hu et al., 2008); it was also found that cath-D increases the three-dimensional (3D) growth (outgrowth) of fibroblasts embedded in a Matrigel matrix (Laurent-Matha et al., 2005). We had previously observed that a mutated catalytically inactive version of cath-D (D231N) remains mitogenic for tumor cells and mouse embryonic fibroblasts (Berchem et al., 2002; Glondu et al., 2001; Laurent-Matha et al., 2005), suggesting that cath-D acts as an extracellular messenger that interacts either directly or indirectly with an as-yet-unidentified cell surface receptor.

In the present study, we show that pro-cath-D hypersecreted by cancer cells stimulates the outgrowth of human mammary fibroblasts (HMFs) embedded in a Matrigel matrix, thus indicating for the first time the relevance of the paracrine role of secreted cath-D in a pathological context. We also identify the low-density lipoprotein (LDL) receptor-related protein-1 (LRP1) as a novel binding partner for cath-D in fibroblasts, and present data in support of a functional role for this interaction in the outgrowth of fibroblasts that is induced by epithelial cancer cells. LRP1 is composed of a 515-kDa extracellular α chain, and an 85-kDa β chain (Lillis et al., 2005; Strickland and Ranganathan, 2003). The extracellular α chain contains binding sites for many structurally unrelated ligands, including lipoprotein particles, proteases and protease–inhibitor complexes. The β chain contains an extracellular domain, a transmembrane region and a cytoplasmic tail. At present, the ligands of the extracellular domain of the LRP1β chain are still unknown. In this study, we report for the first time the interaction between pro-cath-D and the extracellular domain of the β chain of LRP1.

The communication between epithelial cancer cells and fibroblasts plays a crucial role in cancer progression. We had previously developed a 3D co-culture system showing that human breast cancer MCF-7 cells induce outgrowth of mouse embryonic, human skin, and breast fibroblasts, indicating that these epithelial cancer cells secrete paracrine factors capable of promoting fibroblast outgrowth in 3D matrices (Laurent-Matha et al., 2005). These factors included cath-D, and we demonstrated that secreted cath-D was involved in the outgrowth of mouse embryonic cath-D-deficient fibroblasts independently of its catalytic activity (Laurent-Matha et al., 2005).

In order to study the relevance of the paracrine role of secreted pro-cath-D in the context of human breast cancer, we have now performed our 3D co-culture system using HMFs (Fig. 1Aa). To find out whether secreted pro-cath-D might be one of the crucial paracrine factors affecting HMF outgrowth, 3D co-culture assays were carried out using MCF-7 cells (Fig. 1Ac) in which endogenous pro-cath-D secretion was reduced by 75% by small-RNA-mediated gene silencing (cath-D siRNA) for 2 days before the co-culture assay (Fig. 1Ad). MCF-7 cells transfected with cath-D siRNA (Fig. 1Ab, middle) were less effective in triggering HMF outgrowth than MCF-7 cells transfected with control luciferase siRNA (luc siRNA) (Fig. 1Ab, left). A 75% inhibition of pro-cath-D secretion by cells transfected with cath-D siRNA still occurred during our co-culture assay, even after 3 days of co-culture (Fig. 1Ad). By contrast, in HMFs embedded alone (Fig. 1Ab, right), pro-cath-D secretion was not detected after 3 days of co-culture (Fig. 1Ad). We conclude that the inhibition of pro-cath-D secretion in breast cancer cells rapidly reduces HMF outgrowth.

To analyze the role of the catalytic activity of secreted pro-cath-D, we then carried out 3D co-culture assays with cath-D-transfected rat 3Y1-Ad12 cancer cell lines secreting either no human cath-D (control), or hypersecreting either human wild-type cath-D (cath-D) or D231N mutated cath-D devoid of proteolytic activity (D231N) (Fig. 1Bb). Only the 3Y1-Ad12 cancer cells secreting wild-type or D231N pro-cath-D (Fig. 1Bc) stimulated HMF outgrowth, and there was no effect on the 3Y1-Ad12 control cells (Fig. 1Ba), suggesting that secreted catalytically-inactive pro-cath-D promotes 3D outgrowth of HMFs in a paracrine manner. Overall, our findings support the concept that pro-cath-D secreted by epithelial cancer cells might display a crucial paracrine effect on breast fibroblasts in a manner independent of its catalytic activity.

Because cath-D hypersecreted by cancer cells is able to stimulate fibroblast outgrowth in a paracrine manner that is independent of its catalytic activity, we postulate that pro-cath-D secreted by cancer cells might act via its interaction with potential receptors present on the fibroblast cell surface. To identify possible cath-D partners, we performed a yeast two-hybrid screen using a LexA DNA-binding domain fused to 48-kDa intermediate cath-D as bait, and a cDNA library isolated from normal breast. The clone isolated in our screen was 100% identical to amino acids 307–479 of the extracellular domain of the LRP1 β chain (Fig. 2A).

To validate the cath-D–LRP1 interaction, we next performed GST pull-down experiments. The polypeptide containing amino acids 307–479 of the extracellular domain of the LRP1 β chain, GST-LRP1β(307–479), was expressed as a GST-fusion protein, and tested for its ability to bind to pro-cath-D (Fig. 2B, left panel). Our results show that pro-cath-D bound to GST-LRP1β(307–479) (Fig. 2B, right panel). Subsequently, we performed GST pull-down experiments using GST-fusion proteins containing the 52-kDa pro-cath-D, 34-kDa cath-D heavy chain, 14-kDa cath-D light chain, or 4-kDa cath-D propeptide (Fig. 2C, left panels). The full-length extracellular LRP1β domain, LRP1β(1–476), was shown to bind effectively to 52-, 34- and 14-kDa cath-D fragments, but poorly to the 4-kDa cath-D propeptide (Fig. 2C, right panels), indicating that the interaction interface spans both the 34- and 14-kDa cath-D subunits.

The LRP1β(307–479) domain contains four juxtaposed complete EGF-like repeats (repeats 19–22) (Fig. 3Aa). In order to identify the LRP1β cath-D-binding domain, GST-fusion proteins containing various fragments of LRP1β(307–479) were tested for their ability to bind to pro-cath-D (Fig. 3Ab). We identified the F4 fragment of LRP1β(349–394), which contains the EGF-like repeat 20, as the shortest fragment able to bind pro-cath-D, when compared with F7 (307–348), F6 (394–432) or F8 (433–479) fragments of LRP1β (Fig. 3Ba). It is worth noting that LRP1β(307–479) also contains other important domains, because the F4 (349–394) fragment was partially active in pro-cath-D binding when compared with F0 (307–479). The abilities of smaller LRP1β fragments (F5, F9–F13) derived from F4, and of the F14 fragment covering the entire EGF-like repeat 20 (residues 360–396) to interact with pro-cath-D (Fig. 3B, top and bottom panels a and b) were both substantially impaired, suggesting that the intact F4 (349–394) fragment is required for the cath-D–LRP1β interaction. Because of the lack of disulfide linkage in E. coli, we verified that binding of pro-cath-D was comparable after GST-LRP1β F0 and F4 refolding (Fig. 3Bc). Our results provide independent confirmation of the yeast two-hybrid screen, and reveal direct interaction between cath-D and residues 349–394 of the extracellular domain of the LRP1β chain.

We next investigated whether pro-cath-D colocalizes with LRP1β at the cell surface of intact cells. Non-permeabilized COS cells transiently co-transfected with pro-cath-D and Myc–LRP1β were analyzed by immunocytochemistry after double staining with a monoclonal antibody recognizing only the pro-cath-D proform, and a polyclonal antibody directed against the N-terminal Myc-tag of LRP1β. Fig. 4A shows that pro-cath-D (in red; panel b) partially colocalized (panel c) with LRP1β (in green; panel a) in punctuate structures. To confirm that secreted pro-cath-D colocalizes with LRP1 at the cell surface, cellular colocalization of LRP1 and pro-cath-D was further analyzed by immunogold electron microscopy in COS cells transiently co-transfected with pro-cath-D and LRP1β after double staining with an anti-pro-cath-D monoclonal antibody and a polyclonal antibody directed against the cytoplasmic tail of LRP1β (Fig. 4B). We observed colocalization of pro-cath-D and the LRP1β chain at the cell surface (Fig. 4B). Colocalization at the cell surface was also observed in COS cells transiently transfected with LRP1β and incubated with 15-nM pro-cath-D (data not shown). To investigate the interaction of secreted cath-D with endogenous LRP1, HMFs were incubated with 15 nM of pro-cath-D, and colocalization was analyzed by immunogold electron microscopy (Fig. 4C). Our data reveal that, in HMFs, pro-cath-D and LRP1β colocalize at the cell surface (Fig. 4C). Altogether, our findings indicate that, in intact cells, pro-cath-D and LRP1β colocalize at the cell surface.

A previous report described that LRP1 is localized both at the cell surface and in early endosomes (Herz and Strickland, 2001). Interestingly, some of the cell surface LRP1 is located in lipid rafts (Wu and Gonias, 2005; Zhang et al., 2004). Therefore, we next investigated whether pro-cath-D and LRP1 might colocalize at the cell surface in these micro-domains. COS cells transfected with cath-D (Fig. 5A) or co-transfected with pro-cath-D and LRP1β (Fig. 5B) were subjected to sucrose density ultracentrifugation. The raft markers flotillin and ganglioside GM1 were found predominantly in the low density fractions 4–6, whereas the non-raft marker, transferrin receptor (TfR), was found at the bottom of the sucrose gradient (Fig. 5, bottom panels). Endogenous LRP1β was detected in both the raft-rich and non-raft fractions (Fig. 5A, top). Pro-cath-D was mainly located in the non-raft fractions 7 and 8, although both pro-cath-D and its 48-kDa intermediate form were also detected in the raft-rich fractions 5 and 6 (Fig. 5A, middle). Interestingly, overexpression of LRP1β (Fig. 5B, top) resulted in a substantial increase of pro-cath-D in the raft-rich fractions (Fig. 5B, middle). These findings reveal, for the first time, the presence of pro-cath-D at the cell surface in lipid rafts, and strongly suggest that LRP1β enrichment in lipid rafts can direct pro-cath-D to these micro-domains.

To find out whether pro-cath-D interacts with the LRP1β chain in cellulo, LRP1β was co-transfected into COS cells in the presence of wild-type pro-cath-D or of a pro-cath-D mutant devoid of proteolytic activity (^D231Ncath-D). Unfortunately, no antibody recognizing only pro-cath-D was available for immunoprecipitation. We therefore used the anti-cath-D M1G8 antibody, which recognizes the 52-, 48- and 34-kDa forms of cath-D. This meant that immunoprecipitation with M1G8 would precipitate extracellular 52-kDa pro-cath-D associated with the cell surface, and intracellular 52-, 48- and 34-kDa forms of cath-D. Our results using anti-LRP1β antibody show that wild-type and D231N pro-cath-D were both coimmunoprecipitated with LRP1β (Fig. 6Aa, top). Longer gel exposure revealed that mature 34-kDa cath-D was also coimmunoprecipitated with LRP1β (Fig. 6Ab), suggesting that pro-cath-D and the associated LRP1β chain can be directed to the lysosomes. Reciprocally, LRP1β was coimmunoprecipitated with cath-D or ^D231Ncath-D using anti-cath-D M1G8 antibody (Fig. 6Aa, bottom). ^D231Ncath-D displayed greater electrophoretic mobility, corresponding to an apparent molecular mass shift of ~1 kDa (Fig. 6Aa, top), as previously reported (Laurent-Matha et al., 2006). These findings demonstrate that the LRP1 β chain interacts preferentially with the 52-kDa pro-cath-D. We therefore conclude that co-transfected pro-cath-D and LRP1β interact in cellulo.

In order to study endogenous interaction in intact cells, we screened a variety of cell lines for LRP1 expression (supplementary material Fig. S1). Cath-D is known to be overexpressed and hypersecreted by breast cancer cells (Capony et al., 1989), but our survey revealed that breast cancer cells (MCF-7 and MDA-MB-231) express low levels of LRP1 (supplementary material Fig. S1). By contrast, LRP1 was highly expressed in all the fibroblastic cell lines tested: PEA10 (LRP1^+/− MEF), CD55^−/− cath-D (cath-D^−/− MEFs transfected with human cath-D) and CCL146 mouse fibroblasts, and HMF and CCD45K human fibroblasts (supplementary material Fig. S1). However, as previously reported, fibroblasts do not secrete detectable levels of pro-cath-D (Laurent-Matha et al., 2005).

Because the cells that express high LRP1 levels and those that secrete pro-cath-D are distinct, we next studied the interaction of cath-D with endogenous LRP1β using cath-D^−/− MEF cells stably transfected with human wild-type (CD55^−/− cath-D) or mutated D231N cath-D (CD55^−/− D231N) that had previously been shown to hypersecrete pro-cath-D (Laurent-Matha et al., 2005). Anti-cath-D M1G8 immuno-affinity purification revealed that LRP1β was co-eluted with both wild-type (Fig. 6Ba) and D231N cath-D (supplementary material Fig. S2). These results indicate that endogenous LRP1 interacts with stably transfected cath-D in fibroblasts.

We finally investigated the interaction of cath-D with endogenous LRP1β in HMFs incubated with 15 nM of pro-cath-D by immuno-affinity purification using M1G8 antibody. LRP1β was co-eluted with cath-D (Fig. 6Bb). We conclude that endogenous LRP1β interacts with cath-D in fibroblasts. Altogether, our data indicate that pro-cath-D interacts with LRP1β in cellulo.

Because we had identified LRP1 as a fibroblastic receptor for pro-cath-D, we went on to check whether cath-D would induce fibroblast outgrowth via its interaction with LRP1.

To investigate the possible implication of LRP1 in the cath-D-induced fibroblast outgrowth, we performed 3D culture assays in cath-D-transfected fibroblasts hypersecreting pro-cath-D (CD55^−/−cath-D) or mutated D231N pro-cath-D (CD55^−/− D231N), which had or had not been silenced for LRP1 (Fig. 7Aa). As previously shown (Laurent-Matha et al., 2005), cath-D^−/− MEF cells transfected with an empty vector (CD55^−/− SV40) did not grow in Matrigel, and transfecting wild-type or mutated cath-D into cath-D^−/− MEF cells led to a switch to fibroblast outgrowth (Fig. 7Ab), indicating cath-D-induced fibroblast outgrowth. We therefore analyzed the consequences of silencing LRP1 in these cells. The effect of LRP1 silencing on CD55^−/− cath-D outgrowth was shown by phase-contrast microscopy (Fig. 7Ba, top) and by cell staining (Fig. 7Ba, bottom). Our findings indicate that LRP1 silencing in CD55^−/− cath-D cells using LRP1 siRNA3 (Fig. 7Bc) significantly reduced their cath-D-dependent outgrowth compared with cells transfected with luc siRNA (P<0.0125; Fig. 7Ba,Bb). Similar results were observed using LRP1 siRNA4 (supplementary material Fig. S3). These findings indicate that cath-D requires LRP1 expression to promote fibroblast outgrowth. However, interestingly, LRP1 silencing in CD55^−/− D231N cells also significantly inhibited their outgrowth, indicating the existence of a mechanism independent of cath-D proteolytic activity (P<0.0125; Fig. 7C).

To further investigate whether LRP1 mediates the paracrine action of secreted pro-cath-D on fibroblast outgrowth, we next performed 3D co-culture assays (Fig. 8A) with HMFs transfected with luc or LRP1 siRNA (Fig. 8B) and co-cultured with 3Y1-Ad12 cancer cells secreting or not secreting wild-type or D231N cath-D (Fig. 8C). The co-culture assay revealed that only pro-cath-D-secreting 3Y1-Ad12 cells stimulated the outgrowth of HMFs transfected with luc siRNA, in contrast to 3Y1-Ad12 control cells (Fig. 8D). This outgrowth was shown by phase-contrast microscopy (Fig. 8D, top) and by cell staining (Fig. 8D, bottom). Our results further showed that silencing LRP1 in HMFs using siRNA1 (Fig. 8E) blocked their mitogenic response to secreted pro-cath-D as shown by phase-contrast microscopy (Fig. 8E, top) and by cell staining (Fig. 8E, bottom). Similar results were obtained with LRP1 siRNA2 (supplementary material Fig. S4). In control experiments, we confirmed that the viability of HMF cells remained unaffected by LRP1 silencing. Phase-contrast microscopy showed that HMFs transfected with luc siRNA adopted an elongated morphology typical of fibroblasts [supplementary material Fig. S5a,b (see insets)]. LRP1 silencing using siRNA1 or siRNA2 analyzed by western blot (supplementary material Fig. S5a,b) induced a rounder shape and fewer protrusions when compared with luc-siRNA-transfected cells [supplementary material Fig. S5a,b (see insets)]. This suggests that LRP1 expression is required for the elongated fibroblast 3D morphology. The cell viability of HMF cells embedded in Matrigel was, however, not altered by LRP1 silencing. This was shown by cell staining in Matrigel (supplementary material Fig. S5a, bottom), and by cell cycle analyses using flow cytometry of cells extracted from Matrigel (proliferating cells in S and G2M phases=17% and 19.7% for luc siRNA and LRP1 siRNA1 HMF cells, respectively). Altogether, these findings highlight that pro-cath-D secreted by epithelial cancer cells promotes fibroblast outgrowth in a paracrine, LRP1-dependent manner (Fig. 9).

In this study, we identify the LRP1 receptor as a binding partner for pro-cath-D in fibroblasts. The cath-D–LRP1β interaction discovered by a yeast two-hybrid screen was further confirmed by GST pull-down, coimmunoprecipitation, co-purification and colocalization. Our results demonstrate that, in fibroblasts, the pro-cath-D protease interacts with the extracellular domain of the β chain of LRP1, and establishes pro-cath-D as the first ligand of the extracellular LRP1 β chain to be identified. We identified the F4 LRP1β(349–394) fragment, which contains the EGF-like repeat 20 with an additional 11 amino acids at its N-terminus, as the shortest fragment able to bind pro-cath-D. No ligand has previously been described as binding to this particular region, and the biological relevance of the EGF-like repeats in the extracellular domain of LRP1β is still unknown. Our results reveal that pro-cath-D colocalized with LRP1β at the cell surface in lipid rafts. At the cell surface, LRP1 has been shown to be localized in clathrin-coated pits and in lipid rafts (Boucher et al., 2002; Wu and Gonias, 2005; Zhang et al., 2004). Here, we show that LRP1β overexpression directs pro-cath-D to the lipid rafts. Interestingly, this study reveals, for the first time, the presence of pro-cath-D in lipid rafts. The lysosomal cysteine protease cathepsin B (cath-B) is also known to localize within these lipid micro-domains in association with the annexin II heterotetramer (Mohamed and Sloane, 2006). A previous study excluded LRP1 as a possible cath-D receptor (Laurent-Matha et al., 2002). These observations were based on the use of the RAP chaperone protein (Herz, 2006), which competes with ligands that bind to the LRP1 α chain (Laurent-Matha et al., 2002). We checked that RAP protein did not bind to LRP1β, and that the pro-cath-D–LRP1β interaction was not abolished by RAP (data not shown). Consequently, our finding that cath-D binds to the LRP1 β chain explains why RAP did not compete with cath-D for binding to LRP1.

Cancer is a tissue-based disease in which malignant cells interact dynamically with many normal cell types, such as fibroblasts (Mueller and Fusenig, 2004). The fibroblast is a major cell type of the stromal compartment, and is intimately involved in orchestrating the stromal part of the dialogue in tissue homeostasis (Grinnell, 1994; Elenbaas et al., 2001). A stromal reaction immediately adjacent to transformed epithelial cells has been documented in several tumor systems (Basset et al., 1990). Whereas the role of matrix metalloproteinases and urokinase plasminogen activator in the stromal compartment has been documented in various studies (Liotta and Kohn, 2001), the potential role of cath-D in fibroblasts has not yet been fully determined. It has been proposed that cath-D that is localized at the surface of breast fibroblasts might be mitogenic (Koblinski et al., 2002), or that intracellular cath-D in fibroblasts might assist cancer cells to digest the extracellular matrix during tissue invasion (Heylen et al., 2002). We had previously observed that pro-cath-D secreted by cancer cells mediates mouse embryonic fibroblast outgrowth in a paracrine manner (Laurent-Matha et al., 2005), suggesting its key role in the tumor microenvironment. In the present study, we validate these findings in HMFs, indicating the relevance of the paracrine role of secreted pro-cath-D in the tumor microenvironment in the pathological context. We also demonstrate that LRP1 is the fibroblastic receptor responsible for stimulating the pro-cath-D-induced outgrowth. Our 3D co-culture outgrowth assays with cancer cells, which did or did not secrete human pro-cath-D, and HMFs, which had or had not been silenced for LRP1, reveal that pro-cath-D requires LRP1 expression to promote HMF outgrowth. These data support a model in which pro-cath-D hypersecreted by cancer cells stimulates the outgrowth of surrounding fibroblasts in a paracrine and LRP1-dependent manner in the breast tumor microenvironment. It is worth noting that LRP1 might be involved in the onset and progression of various human malignancies, because LRP1 is expressed by fibroblasts in breast cancer biopsies, and also at the invasive front in colon cancer biopsies (Christensen et al., 1996; Obermeyer et al., 2007). In addition, C766T LRP1 gene polymorphism is associated with an increased risk of breast cancer development (Benes et al., 2003). Moreover, recent studies have reported a stimulatory role of LRP1 in cancer cell proliferation, motility and invasion (Dedieu et al., 2008; Li et al., 2003; Song et al., 2009). Although breast cancer cells express LRP1 at a much lower level than do fibroblasts, several studies have reported that hypoxia upregulates LRP1 expression in breast cancer cells (Bando et al., 2003; Koong et al., 2000; Montel et al., 2007). It is therefore possible that cath-D-enhanced breast cancer cell proliferation might be also dependent on LRP1 expression, acting via an autocrine loop. Our outgrowth assays using cells hypersecreting pro-cath-D or not, and silenced for LRP1 or not, strongly suggest that cath-D also promotes outgrowth via LRP1 in an autocrine manner. Future studies will investigate the possible relevance of the autocrine function of cath-D via LRP1 in breast cancer cells.

The mechanism by which LRP1 controls the cath-D-dependent stimulation of fibroblast outgrowth implies that cath-D has an action independent of its catalytic activity. The serine protease tPA has also been shown to act independently of its proteolytic activity as a cytokine via LRP1 (Hu et al., 2006). LRP1 is known to modify the extracellular microenvironment by internalizing numerous ligands via its α chain (Herz and Strickland, 2001), to control signal transduction pathways via cytoplasmic LRP1β tyrosine phosphorylation (Barnes et al., 2003; Boucher and Gotthardt, 2004; Boucher et al., 2002; Hu et al., 2006; Yang et al., 2004), and to modulate gene transcription by regulated intramembrane proteolysis (RIP) of its β chain (Kinoshita et al., 2003; May et al., 2002; von Arnim et al., 2005; Zurhove et al., 2008). Future research will investigate whether the pro-cath-D–LRP1β complex is internalized, and whether the pro-cath-D–LRP1β interaction can affect LRP1β tyrosine phosphorylation or LRP1β RIP.

In conclusion, we report, for the first time, that pro-cath-D interacts with the extracellular domain of the LRP1 β chain at the surface of fibroblasts, and that it promotes an LRP1-dependent fibroblast outgrowth independently of its catalytic activity. We propose that pro-cath-D, which is hypersecreted by breast cancer cells, might be involved in creating the tumor microenvironment that can sustain breast tumor growth and progression through its interaction with LRP1.

HMFs, kindly provided by Jacques Piette (IGM, Montpellier, France), were obtained from reduction mammoplasty tissues from a patient without cancer. Cath-D-deficient CD55^−/− immortalized mouse fibroblasts transfected with empty vector (CD55^−/− SV40), human cath-D (CD55^−/− cath-D) or D231N mutated cath-D (CD55^−/− D231N) were previously described (Laurent-Matha et al., 2005). 3Y1-Ad12 rat cancer cells transfected with empty vector (3Y1-Ad12/control), human cath-D (3Y1-Ad12/cath-D) or D231N mutated cath-D (3Y1-Ad12/D231N) were previously described (Glondu et al., 2001). PEA13 (LRP1^−/− MEF) and PEA10 (LRP1^+/− MEF) were purchased from the American Type Culture Collection (ATCC). Cells were cultured in DMEM with 10% fetal calf serum (FCS, GibcoBRL). The 11H4 hybridoma directed against the C-terminal intracellular part of the LRP1 β chain was purchased from ATCC. Polyclonal anti-human LRP1 β-chain antiserum directed against the C-terminal intracellular part of LRP1 β chain was previously described (Zhang et al., 2004). The anti-human, cath-D monoclonal antibody M2E8, used for immunofluorescence, interacts only with 52-kDa pro-cath-D, and not with 48- or 34-kDa cath-D (Freiss et al., 1988). The anti-human cath-D monoclonal antibody (BD Biosciences) used for immunoblotting recognized the 52-, 48- and 34-kDa forms of cath-D. The anti-human cath-D M1G8 monoclonal antibody used for immunoprecipitation and immunoaffinity purification recognized the 52-, 48- and 34-kDa forms of cath-D (Garcia et al., 1985). Control IgG1 MOPC-21 monoclonal antibody was purchased from Abcam, anti-β actin polyclonal antibody from Sigma, anti-human flotillin-1 monoclonal antibody from BD Biosciences, and anti-human transferrin receptor monoclonal antibody from Zymed. Pro-cath-D content of culture supernatants was determined by using a cath-D immunoradiometric assay (Garcia et al., 1985).

pcDNA3.1(+) Myc-tagged LRP1 β chain (Barnes et al., 2003), pcDNA3.1(−) cath-D and pcDNA3.1(−) ^D231Ncath-D expression plasmids (Glondu et al., 2001) have previously been described. The pcDNA3.1(+) LRP1β extracellular domain (1–476) expression plasmid was created by inserting the PCR-amplified cDNA encoding LRP1β(1–476) from pcDNA3.1(+) Myc-tagged LRP1β into pcDNA3.1(+) previously digested by NheI and XbaI. pGEX-4T-1 LRP1β(307–479) was generated by inserting PCR-amplified cDNA encoding LRP1β(307–479) from pYESTrp2–LRP1β(307–479) identified by a two-yeast hybrid assay into pGEX-4T-1 previously digested by EcoRI. GEX-4T-1 LRP1β smaller fragments were generated by inserting PCR-amplified cDNA LRP1β shorter fragments from pYESTrp2–LRP1β(307–479) into pGEX-4T-1 previously digested by EcoRI and XhoI. supplementary material Table S1 shows the primers used for LRP1β cloning. pGEX-2TK–cath-D constructs were obtained by inserting PCR-amplified cDNA encoding human 52-kDa pro-cath-D, 34-kDa and 14-kDa cath-D chains, and 4-kDa cath-D pro-fragment into pGEX-4T-1 previously digested by EcoRI. supplementary material Table S1 shows the primers used for cath-D cloning.

Human cath-D (48-kDa intermediate form) was fused with the LexA DNA-binding domain in the pMW101 vector. A cDNA library derived from normal breast tissue was cloned into the galactosidase-inducible pYESTrp2 vector (Invitrogen) containing a B42 activating domain. The yeast two-hybrid screen was performed as described previously (Slentz-Kesler et al., 2000).

Approximately 30,000 HMF cells were transiently transfected with 10 μg human LRP1 or luc siRNAs using Oligofectamine (Invitrogen). Approximately 50,000 CD55^−/− cath-D or CD55^−/− D231N cells were transiently transfected with 2.5 μg mouse LRP1 or luc siRNAs. Around 15,000 MCF-7 cells were transiently transfected with 1 μg siRNA using Oligofectamine (Invitrogen). For outgrowth assays, 50,000CD55^−/− cath-D or CD55^−/− D231N cells were re-suspended at 4°C in Matrigel (BD Biosciences) 24 hours post-transfection with LRP1 or luc siRNAs, and added to a pre-set layer of Matrigel (Glondu et al., 2001). In co-culture outgrowth assays, 15,000 MCF-7 cells previously transfected with cath-D or luc siRNAs, or 100,000 cells from control-, D231N- or cath-D-transfected 3Y1-Ad12 cell lines were plated and then covered first with a layer of Matrigel, and then with a layer of Matrigel containing 50,000 HMFs 48 hours post-transfection with LRP1 or luc siRNAs (Laurent-Matha et al., 2005).

Duplexes of 21-nucleotide human LRP1 siRNA1 (target sequence 5′-AAGCAGTTTGCCTGCAGAGAT-3′, residues 666–684) (Li et al., 2003), human LRP1 siRNA2 (target sequence 5′-AAGCTATGAGTTTAAGAAGTT-3′) (Dharmacon), mouse LRP1 siRNA3 (target sequence 5′-AAGCAUCUCAGUAGACUAUCA-3′) (Fears et al., 2005), mouse LRP1 siRNA4 (target sequence 5′-AAGCAGTTTGCCTGCAGAGAC-3′), human cath-D siRNA (target sequence 5′-AAGCUGGUGGACCAGAACAUC-3′, residues 666-684) (Bidere et al., 2003) or firefly luciferase (luc) siRNA (target sequence 5′-AACGTACGCGGAATACTTCGA-3′) were synthesized by MWG Biotech S.A. (France) or Dharmacon. supplementary material Table S1 shows siRNA oligonucleotides.

[³⁵S]methionine-labeled pre-pro-cath-D, and LRP1β (1-476) were obtained by transcription and translation using the TNT^T7-coupled reticulocyte lysate system (Promega). GST, GST-LRP1β fragments and GST–cath-D fusion proteins were produced in Escherichia coli B strain BL21 using isopropyl-1-thio-β-D-galactopyranoside (1 mM) for 3 hours at 37°C. GST-fusion proteins were purified on glutathione-Sepharose beads (Amersham Biosciences). For pull-down assays, 20 μl of glutathione-Sepharose beads with immobilized GST fusion proteins were incubated overnight at 4°C with [³⁵S]methionine-labeled proteins in 500 μl PDB buffer (20 mM HEPES-KOH, pH 7.9, 10% glycerol, 100 mM KCl, 5 mM MgCl₂,0.2 mM EDTA, 1 mM DTT, 0.2 mM phenylmethylsulfonyl fluoride) containing 15 mg/ml BSA and 0.1% Tween 20. The beads were washed with 500 μl PDB buffer, and bound proteins were resolved by 15% SDS-PAGE, stained with Coomassie Blue, and exposed to autoradiographic film. The refolding of GST proteins was performed using a multi-step dialysis protocol (Protein refolding kit, Novagen) followed by disulfide-bond formation using a redox system (cysteine/cystine).

COS cells were co-transfected with 10 μg of pcDNA3–Myc–LRP1β, and 10 μg of pcDNA3.1, pcDNA3.1–cath-D or pcDNA3.1–^D231Ncath-D vectors. Transient transfection was carried out using Lipofectamine and Opti-MEM (Gibco-BRL). At 2 days post-transfection, cells were directly lysed in 50 mM HEPES (pH 7.5), 150 mM NaCl, 10% glycerol, 1% Triton X-100, 1.5 mM MgCl₂, 1 mM EGTA, 100 mM NaF, 10 mM sodium pyrophosphate, 500 μM sodium vanadate and a protease inhibitor cocktail (PLC lysis buffer). Lysates were incubated with 3 μg of anti-cath-D M1G8, or control IgG1 MOPC-21 monoclonal antibodies, or 40 μl of the anti-LRP1β 11H4 hybridoma overnight at 4°C, and subsequently with 25 μl of 10% protein G-Sepharose, for 2 hours at 4°C on a shaker. Sepharose beads were washed four times with PLC buffer, boiled for 3 minutes in SDS sample buffer, and resolved by SDS-PAGE and immunoblotting. For cath-D purification, unwashed cells were lysed in PLC buffer and were passed over an anti-cath-D M1G8-coupled agarose column. The column was washed with phosphate buffer (0.5 M NaPO₄, 150 mM NaCl, 0.01% Tween 80, 5 mM β-glycerophosphate) containing protease inhibitors, and then eluted in different fractions with 20 mM lysine, pH 11.

COS cells transfected as described above were lysed in MEB buffer (150 mM NaCl, 20 mM morpholine ethane sulfonic acid, pH 6.5) containing 1% Triton X-100, 1 mM Na₃VO₄, 50 mM NaF and protease inhibitor cocktail, and were homogenized using a Dounce homogenizer. The preparation was mixed with an equal volume of 80% sucrose solution in MEB buffer in an ultracentrifuge tube over-laid with 30% sucrose and 1.5 ml of 5% sucrose, and centrifuged at 39,000 rpm for 16 hours at 4°C in a Beckman SW40 Ti. After centrifuging, fractions were collected and subjected to immunoblot analysis. The ganglioside GM1 was detected with biotin-conjugated cholera toxin B subunit (Sigma-Aldrich) followed by incubation with horseradish-peroxidase-conjugated streptavidin and revealed by chemiluminescence (ECL, GE Healthcare).

Cell extracts (100 μg) or conditioned media (80 μl) were submitted to SDS-PAGE and anti-LRP1β, anti-cath-D or anti-β-actin immunoblotting.

Cells co-transfected with pro-cath-D and Myc-LRP1β were fixed with 4% paraformaldehyde and blocked with 2.5% goat serum (Sigma). Cells were incubated with the A-14 rabbit polyclonal antibody (10 μg/ml, Santa Cruz) recognizing the 9E10 Myc tag followed by an Alexa-Fluor-488-conjugated goat anti-rabbit IgG (1/200; Invitrogen). After washing, cells were incubated with an anti-pro-cath-D M2E8 mouse monoclonal (40 μg/ml) followed by an Alexa-Fluor-568-conjugated goat anti-mouse IgG (1/200; Invitrogen). DNA was visualized by incubation with 0.5 μg/ml cell-permeant Hoechst 33342 dye (Molecular Probes; 10 minutes). Microscopy slides were observed with a motorized Leica Microsystems (Rueil-Malmaison, France) DMRA2 microscope equipped with an oil immersion 100×/1.4 apochromatic objective and a 12-bit CoolSNAP FX CCD camera (Princeton Instruments, Roper Scientific, Evry, France), both controlled by the MetaMorph imaging software (Universal Imaging, Roper Scientific).

Cells grown in 24-well inserts were fixed with 3% paraformaldehyde and 0.1% glutaraldehyde in a phosphate buffer (pH 7.2) overnight at 4°C. Cells were cryoprotected with sucrose 3.5%, and stained for 2 hours with uranyl acetate 2% in a maleate buffer 0.05 M (pH 6.2) at room temperature. Cells were embedded in Lowicryl HM20 in a Leica EM AFS using the Progressive Lowering of Temperature method. Ultra-thin sections (85 nm) were first pre-incubated in PBS with 0.1% cold water fish gelatin, 1% BSA and 0.05% Tween 20 (incubation buffer) for 2 hours at room temperature, and then incubated with rabbit anti-LRP1β (1/20) and mouse anti-cath-D M2E8 (40 μg/ml) in the incubation buffer overnight at 4°C, and subsequently with 15 nm goat anti-rabbit-gold and 6-nm goat anti-mouse-gold 5 (Aurion) diluted 1/20 in the incubation buffer for 2 hours at 4°C. Sections were observed under a Hitachi 7100 transmission electron microscope.

Cells embedded in Matrigel were recovered after 3 days with 2.5 ml Matrisperse (Becton Dickinson), re-suspended in 1 ml of a solution containing 0.1% tri-sodium citrate dehydrate, 0.1% Triton X-100, 25 μg/ml propidium iodide and 100 μg/ml RNAse, prior to FACScan analysis (Becton Dickinson, CA) with an argon ion laser turned at 488 nm, 20 mW. Propidium iodide fluorescence was measured at 585 nm. Data were collected and analyzed with Cellquest (Becton Dickinson, CA) and ModFit (Verity Software, ME) software, respectively.

RNA extraction and reverse transcription were performed, and quantitative PCR (Q-PCR) was carried out using a LightCycler and the DNA-double-strand-specific SYBR Green I dye for detection (Roche). Q-PCR was performed using gene-specific oligonucleotides, and results were normalized to RS9 levels.

We thank Nadia Kerdjadj for secretarial assistance, Jean-Yves Cance for the photographs and Chantal Cazevieille (Centre de Ressources en Imagerie Cellulaire, Montpellier) for the immunogold microscopy. We also thank Stephan Jalaguier for helpful discussions regarding the GST pull-down experiments. This work was supported by the ‘Institut National de la Santé et de la Recherche Médicale’, University of Montpellier I, ‘ANR Jeunes Chercheuses, Jeunes Chercheurs’ and the ‘Ligue Nationale contre le Cancer’, and the ‘Association pour la Recherche sur le Cancer’, which provided a fellowship for Mélanie Beaujouin.