Migration-stimulating factor displays HEXXH-dependent catalytic activity important for promoting tumor cell migration

Authors


Abstract

Like most extracellular matrix (ECM) components, fibronectin (Fn) is proteolyzed generating specific activities. Fibronectin proteinase (Fn-proteinase) represents such a cryptic activity located in the gelatin-binding domain (GBD) of Fn and displays a zinc metalloproteinase activity. The migration-stimulating factor (MSF) is a truncated Fn isoform generated by alternative mRNA splicing and corresponds to the N-terminal part of Fn that comprises the GBD. We show that several human mammary epithelial cells express MSF and constitutively produce Fn-proteinase activity. Furthermore, recombinant MSF produced by HEK-293 and MCF-7 cells possesses a constitutive Fn-proteinase activity. Mutating the putative zinc-binding motif, HEXXH, of the protein abolishes its activity thereby demonstrating its specificity. Using PCR, we showed that MSF is barely expressed in normal breast tissues, whereas its expression is significantly increased in tumors. Furthermore, an association between MSF expression and invasive capacity is observed in various breast adenocarcinoma cell lines. Indeed, when stably transfected in non-invasive MCF-7 cells, MSF promotes cell migration in a mechanism mostly dependent on its Fn-proteinase activity. In summary, our study shows that: (i) MSF displays constitutive Fn-proteinase activity; (ii) MSF expression is induced in human breast cancer; and (iii) MSF confers pro-migratory activity that depends mostly on its Fn-proteinase activity. These results suggest that MSF may be involved in tumor progression. © 2005 Wiley-Liss, Inc.

Fibronectin (Fn) is an adhesive glycoprotein, secreted as a dimer, present in body fluids and within extracellular matrices. Each monomer (∼250 kDa) is organized into proteolysis-resistant functional domains. These domains can bind fibrin, collagens, integrins and proteoglycans, accounting for the wide range of Fn functions (Fig. 1a).1

Figure 1.

Schematic representation of the “full-length” fibronectin and the truncated human fibronectin isoform, MSF. (a) Fn is a dimer consisting of 2 nearly identical polypeptide chains covalently linked together by 2 disulfide bridges. A monomer is shown. Fn is a mosaic protein composed of 3 repeating modules: FnI, FnII and FnIII. These modules are organized in functional domains responsible for binding to fibrin, gelatin, collagens, integrins and heparin. The alternative exons EdA, EdB and IIICS are indicated. (b) Schematic diagram of the truncated Fn isoform MSF. MSF is a 77 kDa protein, containing the Fibrin-I domain, the gelatin/collagen-binding domain (GBD) and a portion of the first Fn Type III module. The C-terminus of the protein is a specific decapeptide coded by a retained intronic sequence.

In addition to these well-characterized properties of “full-length” Fn, proteolytically generated Fn fragments show additional biological activities. For instance, the gelatin-binding domain (GBD) influences cell proliferation and migration,2 and induces cartilage catabolism.3 In addition, the GBD isolated after Fn proteolysis or produced in recombinant cells, referred to fibronectin proteinase (Fn-proteinase),4 displays a zinc-dependent collagenase/gelatinase activity.5, 6, 7 Fn-proteinase activity is inhibited by a specific phosphinic pseudo peptide.5 This activity is involved in the degradation of cartilage induced by the GBD.8 In addition, we showed recently that Fn-proteinase is activated by plasmin and therefore may play a role in tissue remodeling.4

Zhao et al.9 recently cloned and characterized a C-terminus truncated Fn isoform generated by alternative splicing in zebrafish. This isoform, called Fn2, is a 120-kDa monomeric protein with distinct properties compared to “full-length” Fn.9 This raised the possibility that Fn fragments and their corresponding specific biological activities might also originate from alternative splicing in humans. We hypothesized that Fn-proteinase activity could be encoded by a mRNA generated by alternative splicing of the Fn mRNA. More recently, the human and murine Fn2 counterparts were also described.10 In addition, while our study was underway, Schor et al.11 cloned a new truncated isoform of human Fn generated by alternative splicing, and called migration-stimulating factor (MSF). This truncated Fn isoform (EMBL AJ2535086) is a 77-kDa protein that corresponds to the N-terminus of Fn. It comprises the Fibrin-I, the GBD and a part of the first Fn Type III module. It is ended by a specific C-terminus decapeptide encoded by a retained intronic sequence (Fig. 1b). MSF is a soluble factor secreted by fibroblasts in fetal skin and breast cancer. It is able to stimulate fibroblast migration in a native collagen-I matrix.12, 13

We provide the first evidence that MSF displays constitutive Fn-proteinase activity. We show that several human breast adenocarcinoma cell lines constitutively express MSF and have an endogenous Fn-proteinase activity. We further demonstrate that expression of recombinant MSF in HEK-293 or MCF-7 cells leads to a constitutive Fn-proteinase activity. Site-directed mutagenesis was used to show that the Fn-proteinase activity of MSF depends on an intact HEXXH sequence, which corresponds to the putative catalytic zinc-binding site of this enzyme. We also report that this Fn-proteinase activity plays a major role in the motogenic activity of MSF on breast adenocarcinoma cells. Expression of MSF in tumors and its catalytic and pro-migratory activities suggest its involvement in tumor progression.

Material and methods

Cell culture

HEK-293, MCF-7, T47D, MDA-MB231, BT549 and Hs578T cell lines were cultured in DMEM medium (GIBCO, Cergy-Pontoise, France) supplemented with 7.5–10% FCS, 2 mM L-glutamine, 100 U/ml of penicillin and 100 μg/ml of streptomycin. HEK-293 and MCF-7 cells expressing the recombinant MSF or the Phe535,539 mutant were established by transfecting cells with either the cDNA of the wild-type MSF or the Phe535,539 mutant in pCDNA3.1mycHis (Invitrogen, Cergy-Pontoise, France) and selecting stably transfected cells with geneticin (1 mg/ml). Cells were rinsed twice with phosphate-buffered saline (PBS), and left in serum-free medium for 18 hr. After 2 more washes with PBS, fresh serum-free medium was added to the cultures for 24 or 48 hr. Conditioned media were then sampled, centrifuged for 5 min at 5,000g and concentrated by centrifugation (ultrafree-4 biomax-10 kDa membrane, Millipore). Concentrated conditioned media were stored at −20°C until use.

RNA isolation and cDNA synthesis

RNA was extracted using the RNeasy midiprep kit (Qiagen, SA, Courtaboeuf, France), according to the manufacturer's instructions. RNA was digested with RNase-free DNase RQ1 (Promega) for 30 min. Total RNA (5 μg) was subjected to reverse transcription using Superscript II RT (GIBCO), according to the manufacturer's instructions.

MSF cDNA cloning by reverse transcription-polymerase chain reaction

For each PCR, 0.5 μl cDNA was mixed with 10 pmol primers, 0.2 mM dNTPs, 3 μl 10× Taq polymerase buffer and 1 U Taq polymerase (Roche Diagnostics, Neuilly sur Seine, France), in a final volume of 30 μl. The expression of MSF and 28S messenger were determined using a 31 cycle PCR (30 sec at 94°C, 30 sec at 56°C and 1 min at 72°C) and a 13 cycle PCR (20 sec at 94°C, 20 sec at 68°C and 15 sec at 72°C), respectively. The following primers were used: MSF primers (sense primer common for human Fn and MSF, 5′-TCGAATTATGAGCAGGACCAG-3′, specific antisense primer, 5′-TTTCTGGGTGGGATACTCAC-3′), 28S primers (sense primer, 5′-GTTCACCCACTAATAGGGAACGTGA-3′, antisense primer, 5′-GATTCTGACTTAGAGGCGTTCAGT-3′). The expected sizes of MSF and 28S amplicon were 794 bp and 212 bp, respectively. For densitometric quantification of MSF expression, RT-PCR products were resolved on 10% acrylamide gels and analyzed using a Fluor-S MultiImager (Bio-Rad) after staining with Gelstar (FMC Bioproducts, Rockland, ME) dye.

MSF cDNA cloning was carried out as follows: in a first PCR, a unique EcoRI site was introduced upstream of the start codon. The sense primer for the PCR was 5′-GGAATTCATGCTTAGGGGTCCGGGGCCCGGGCTGCT-3′ (the EcoRI site is underlined) and the antisense primer was 5′-TTTCTGGGTGGGATACTCAC-3′. A 31-cycle PCR was used. Each cycle consisted of 30 sec at 94°C, 30 sec at 62°C and 90 sec at 72°C. The PCR product (1.9 kb) was subcloned into the pGEM.T cloning vector (Promega). The construct was termed pGEM.T/EcoRI. In a second PCR, the stop codon was removed by introducing a unique BamHI site into the splice variant cDNA. This enabled an in-frame fusion of the 3′ end of the MSF cDNA with the myc epitope and the polyhistidine tag contained in pcDNA3.1/myc-His. The sense primer for the PCR was 5′-TCGAATTATGAGCAGGACCAG-3′ and the antisense primer was 5′-CGGGATCCGTATCCAAGGTTTCTG-3′ (the BamHI site is underlined). A 31-cycle PCR was used. Each cycle consisted of 30 sec at 94°C, 30 sec at 56°C and 1 min at 72°C. The PCR product (811 bp) was subcloned into pGEM.T. The construct was termed pGEM.T/BamHI.

A 1.2-kb SpeI-XhoI fragment of the pGEM.T/EcoRI construct, containing the 5′ part of the splice variant cDNA was excised and ligated into the pGEM.T/BamHI digested with SpeI and XhoI. The EcoRI-BamHI fragment of the MSF cDNA (1.9 kb) was isolated and ligated into the pcDNA3.1/myc-His expression vector (Invitrogen) digested with EcoRI and BamHI. All the constructs were verified by restriction mapping and sequencing.

PCR-mediated mutagenesis of catalytic active consensus site

The histidine residues His535 and His539 were mutated to phenylalanine by PCR-mediated mutagenesis. Two 31-cycle PCR were carried out in parallel. Each cycle consisted of 30 sec at 94°C, 30 sec at 56°C and 1 min at 72°C. The following primers were used for both PCR (the mutated bases are underlined): sense primer 5′-TCGAATTATGAGCAGGACCAG-3′ and antisense primer 5′-CAT GAACCCCTCTTCAAAACGCTTGTGGAA TGTG-3′; sense primer 5′-CGTTTTGAAGAGGGGTTCATGCTGAACTGTACAT-3′ and antisense primer 5′-CGGGATCCGTATCCAAGGTTTCTG-3′. Both PCR products were isolated and used as templates for a 25-cycle PCR (30 sec at 94°C, 30 sec at 56°C and 1 min at 72°C). The sense primer was 5′-TCGAATTATGAGCAGGACCAG-3′ and the antisense primer was 5′-CGGGATCCGTATCCAAGGTTTCTG-3′. The PCR product (811 bp) was subcloned into the pGEM.T cloning vector, and analyzed by sequencing. The mutated cDNA was cloned into the pcDNA3.1/myc-His expression vector, as described above for the MSF cDNA.

Enzymatic assay

Enzymatic assays were carried out as described previously.4 Briefly, aliquots of conditioned media were incubated at 37°C in 100 mM Tris/HCl, pH 7.4, 100 mM NaCl, 10 mM CaCl2, with 25 μM Mca-Pro-Leu-Gly-Leu-Dpa-Ala-Arg-NH2 (Bachem Biochimie, France), the intramolecularly quenched fluorogenic substrate of collagenase.14 Fluorescence was measured in a Fusion microplate analyzer (Packard) at 330 nm (excitation) and 385 nm (emission). Results are expressed as amounts of the Mca-Pro-Leu liberated using a calibration curve constructed with a solution of Mca-Pro-Leu (1 mg/ml) as reported.5 For inhibition studies, samples were incubated with inhibitors for 10 min at room temperature before the addition of the substrate. Batimastat was a gift from British Biotech Pharmaceuticals Ltd. (Oxford, UK). The phosphinic peptide Z-FΨ(PO2CH2)A-R-F-OH (where Z represents a benzyloxycarbonyl group) was generously provided by Dr. V. Dive (CEA, Saclay, France).

Antibody production

An antibody against the MSF was generated by conjugating the specific C-terminus decapeptide (VSIPPRNLGY) with ovalbumin and injecting the conjugate into rabbits (NeoMPS SA, Strasbourg, France). This antibody made it possible to detect the endogenous human protein on immunoblots, but not by immunohistochemistry on paraffin sections.

Immunoblotting

Aliquots of media were loaded onto 10% SDS-PAGE gels. After electrophoresis, samples were transferred onto PVDF membrane (Immobilon P, Millipore). The membrane was probed with a monoclonal antibody directed against the Fibrin-I domain of Fn (clone 616, dilution 1:1,000) (Chemicon, Euromedex, Souffelweyersheim, France), a polyclonal antibody directed against the specific C-terminus decapeptide of MSF (dilution 1:500). Bound antibodies were revealed with the biotin-streptavidin system (Amersham Biosciences, Saclay, France).

Boyden chamber invasion assay

The influence of MSF on the invasiveness of cells was assessed by using a modified Boyden chamber assay, as described previously.15 In brief, 100 μl of MCF-7 cells (5 × 105/ml) suspended in DMEM supplemented with 0.1% BSA were seeded in the upper compartment of the chamber. The lower compartment was filled with 600 μl of DMEM 1% BSA supplemented or not with 10% FCS. The 2 compartments were separated by a porous filter (8 μm pore) coated with 100 μl of native collagen (Serva Electrophoresis Gmbtt, Heidelberg, Germany) (60 μg/ml). The chambers were incubated for 24 hr at 37°C. Filters were then fixed in paraformaldehyde (4%), permeabilized in methanol and stained with Giemsa (4%). The cells on the upper surface of the filter were scrapped with a cotton swab. Invasion was quantified by counting the number of cells on the lower surface of the filters (17 fields at 400-fold magnification).

Breast tissue collection

The breast tissue samples (i.e., 17 invasive ductal breast carcinomas and 11 normal breast samples) were obtained from the Breast Tissue Bank of the Centre Hospitalier Universitaire, Liège, Belgium (Pr. F. van den Brûle). All tissues were originally obtained from the Department of Pathology of the University of Liège, Sart Tilman, Belgium (Pr. J. Boniver). All breast samples were immediately snap frozen in liquid nitrogen immediately after surgical removal. Frozen tissue samples were pulverized into fine powder before RNA extraction. All examined cancer samples were of the invasive ductal carcinoma type, with >80% cancer cell coverage of the sample in frozen companion slide. Control samples were samples from reduction mammoplasty that were identically processed.

Results

Endogenous MSF is expressed in mammary epithelial cell lines

Truncated Fn isoforms obtained by alternative splicing of the Fn mRNA transcript were recently described in human.9, 11 Among them, the MSF is a 77-kDa protein that contains the Fibrin-I, the GBD and a part of the first Fn Type III module (Fig. 1b). MSF was constitutively expressed in the human mammary epithelial cell lines MDA-MB 231, BT549 and Hs578T, as determined by RT-PCR (Fig. 2a). In contrast, 2 other mammary epithelial cell lines, MCF-7 and T47D, did not express the MSF mRNA (Fig. 2a).

Figure 2.

Constitutive MSF expression in mammary epithelial cell lines and Fn-proteinase activity in cell supernatant. (a) cDNAs from T47D (lane 2), MCF-7 (lane 3), MDA-MB231 (lane 4), BT549 (lane 5) and Hs578T (lane 6) cells were used as a template for PCR with primers specific for MSF (upper panel) or 28S (lower panel). The band sizes of the DNA ladder (lane 1) are indicated on the left. (b) Cells were incubated for 48 hr in serum-free medium. Conditioned media were then collected and Mca-Pro-Leu-Gly-Leu-Dpa-Ala-Arg-NH2 hydrolysis was measured for 5 hr with (open bars) and without the phosphinic peptide inhibitor Z-FΨ(PO2CH2)A-R-F-OH (100 μM) (shaded bars). The mean ± SEM values for 3 experiments are shown.

We hypothesized that MSF may display Fn-proteinase activity. This activity can be measured in cell supernatant as the proteolytic activity on the Mca-Pro-Leu-Gly-Leu-Dpa-Ala-Arg-NH2 substrate that is specifically inhibited by the phosphinic peptide inhibitor Z-FΨ(PO2CH2)A-R-F-OH.4 Conditioned medium from MSF-expressing cell lines hydrolyzed the fluorogenic substrate more efficient compared to medium from MCF-7 and T47D cell lines that did not express MSF (Fig. 2b). This hydrolysis was inhibited by the Fn-proteinase inhibitor Z-FΨ(PO2CH2)A-R-F-OH. Fn-proteinase activity represented 35%, 43% and 74% of total of Mca-Pro-Leu-Gly-Leu-Dpa-Ala-Arg-NH2 hydrolysis obtained with medium from MDA-MB 231, BT549 and Hs578T cell lines, respectively (Fig. 2b). These data support the hypothesis that endogenous MSF displays Fn-proteinase activity.

Recombinant MSF displays Fn-proteinase activity

To study the putative Fn-proteinase activity of MSF, MSF cDNA was cloned and recombinant MSF was produced in HEK-293 cells as a fusion protein with C-terminal c-myc and poly histidine epitopes. A protein of the expected size (83 kDa) was detected in the cell supernatant by immunoblotting with antibodies targeting the Fibrin-I domain or the specific C-terminal decapeptide (Fig. 3a, lanes 2,4).

Figure 3.

Catalytic activity of the recombinant MSF. (a) Immunoblot analysis of MSF production in HEK-293 cells. Recombinant MSF was detected in conditioned medium from HEK-293 cells expressing MSF (lanes 2,4), whereas it was not present in conditioned medium from control HEK-293 cells (lanes 1,3). MSF was recognized by a monoclonal anti Fibrin-I domain antibody (lane 2) or by the polyclonal C-ter decapeptide antibody (lane 4). (b) Conditioned media from control HEK-293 cells and HEK-293 cells expressing either the wild-type MSF or the Phe535,539 mutant were assayed for Mca-Pro-Leu-Gly-Leu-Dpa-Ala-Arg-NH2 hydrolysis for 70 min. A low level of substrate hydrolysis was observed with medium from control HEK-293 cells (control) and from HEK-293 cells expressing the Phe535,539 mutant (Phe535,539), whereas a strong substrate hydrolyzing activity was detected in MSF-containing medium (MSF). Mean ± SEM values for 3 experiments are given.

To determine whether MSF displays Fn-proteinase activity, Mca-Pro-Leu-Gly-Leu-Dpa-Ala-Arg-NH2 hydrolysis was measured in MSF-transfected HEK-293 conditioned medium. MSF produced by transfected cells induced a strong hydrolysis of the substrate (Fig. 3b). In contrast, no activity was measured in the medium from wild-type HEK-293 cells. The catalytic activity of MSF was inhibited by EDTA and by 1,10 Phenantroline, 2 inhibitors of zinc-dependent metalloenzymes (Fig. 4). In contrast, the serine and cysteine peptidase inhibitors, aprotinin, leupeptin and E64, did not inhibit MSF catalytic activity (results not shown). In addition, the phosphinic peptide Z-FΨ(PO2CH2)A-R-F-OH, which inhibited preferentially Fn-proteinase, inhibited the hydrolysis of Mca-Pro-Leu-Gly-Leu-Dpa-Ala-Arg-NH2 potently with an IC50 of 26 nM (Fig. 4). Batimastat, a potent broad spectrum inhibitor of matrix metalloproteinases, inhibited the catalytic activity of the truncated Fn isoform, but it was 1,000-fold less effective than Z-FΨ(PO2CH2)A-R-F-OH (IC50 = 40 μM) (Fig. 4). These results show that recombinant MSF displays Fn-proteinase activity.

Figure 4.

Inhibition of the MSF catalytic activity. Conditioned medium from HEK-293 expressing MSF was incubated with increasing concentrations of the phosphinic peptide inhibitor Z-FΨ(PO2CH2)A-R-F-OH (●), Batimastat (▪), 1, 10 Phenantroline (▴) or EDTA (▵), and assayed for Mca-Pro-Leu-Gly-Leu-Dpa-Ala-Arg-NH2 hydrolysis. The data are expressed as a percentage of substrate hydrolysis in control medium without inhibitor.

Consensus zinc-binding motif of MSF is necessary for its Fn-proteinase activity

MSF contains the consensus zinc-binding motif (HEXXH) of the zincin family in the eighth Type I module located in the GBD. To determine whether this motif is involved in the Fn-proteinase activity of MSF, we generated a mutant protein in which the 2 histidine residues in this motif were replaced by phenylalanine residues. The recombinant mutant protein (Phe535,539) was stably expressed in HEK-293 cells. Like its wild-type counterpart, mutant Phe535,539 is a 83-kDa secreted protein, recognized by anti Fibrin-I and anti C-ter decapeptide antibodies (results not shown). The mutation of the histidine residues into phenylalanine totally abolished the Fn-proteinase activity of MSF, as shown by the absence of substrate hydrolysis above background level (Fig. 3b). These results show that His535 and His539 are necessary for the Fn-proteinase activity of MSF and strongly suggest that its HEXXH motif is the catalytic site of Fn-proteinase.

MSF expression is induced in breast cancer

Expression of MSF in normal breast and breast carcinoma tissues samples was then investigated (Fig. 5). RT-PCR analysis of MSF expression in breast tissue showed that MSF mRNA was expressed only in 8 of 11 normal tissues analyzed (Fig. 5a, upper panel). When expressed, MSF mRNA was barely detectable. In contrast, MSF mRNA was strongly expressed in breast carcinomas and was found in all carcinomas (n = 17) (Fig. 5b, upper panel). The average expression level of MSF was 9-fold more higher in breast carcinomas than in normal tissues (p < 0.0001) (Fig. 5c).

Figure 5.

Analysis of MSF mRNA production in breast cancer. Eleven healthy breast tissues (a) and 17 breast carcinomas (b) were analyzed by RT-PCR for the expression of MSF (upper panel). Expression of 28S is shown in the lower panel. (c) Expression of MSF mRNA in healthy breast tissues (●) and breast carcinomas (▪) was semiquantified by densitometry and normalized to the 28S signal. Comparisons were carried out by an unpaired t-test with Welch's correction. p < 0.0001 vs. normal tissue.

Stimulation of cell migration by MSF is dependent on its Fn-proteinase activity

According to their in vitro invasiveness determined in a Boyden chamber assay, MDA-MB231, BT549 and Hs578T are considered as invasive cell lines and MCF-7 and T47D as non-invasive.15 Figure 2 showed that MDA-MB231, BT549 and Hs578T cells expressed MSF, whereas MCF-7 and T47D cell lines did not. These results could suggest that the expression of MSF is associated with the invasive capacity of cells. To determine whether the migratory activity of cells is influenced by their ability to express MSF, MCF-7 cells were stably transfected to produce the recombinant MSF protein. Clones of MCF-7 cells expressing the recombinant MSF displays a constitutive Fn-proteinase activity in conditioned medium, whereas no activity was present in medium from cells transfected with vector alone (Fig. 6a). Mca-Pro-Leu-Gly-Leu-Dpa-Ala-Arg-NH2 hydrolysis was 8.5- and 7.2-fold stronger with MSF-containing medium than with medium from control MCF-7 or MCF-7 transfected with vector alone, respectively. This hydrolysis was inhibited by the phosphinic peptide inhibitor of Fn-proteinase (Fig. 6a).

Figure 6.

MSF induces migration of MCF-7 cells. (a) Control MCF-7 cells (control) and clones of MCF-7 cells expressing either the wild-type MSF (MSF), the Phe535,539 mutant (Phe535,539) or transfected with the vector alone (vector) were incubated for 48 hr in serum-free medium. Conditioned media were then collected and Mca-Pro-Leu-Gly-Leu-Dpa-Ala-Arg-NH2 hydrolysis was measured with (open bars) and without the phosphinic peptide inhibitor Z-FΨ(PO2CH2)A-R-F-OH (100 μM) (shaded bars) for 5 hr. Mean ± SEM values for 3 experiments are given. Inset: cDNAs from control MCF-7 (lane 2) and clones of MCF-7 cells expressing either the wild-type MSF (lane 4), the Phe535,539 mutant (lane 5) or transfected with the vector alone (lane 3) were used as a template for PCR with primers specific for MSF (upper panel) or 28S (lower panel). The band sizes of the DNA ladder (lane 1) are indicated on the left. (b,c) Migratory activity of MSF was assessed using control MCF-7 cells (control) and clones of MCF-7 cells expressing either MSF (MSF), the Phe535,539 mutant (Phe535,539) and transfected with the vector alone (vector) in a modified Boyden chamber assay. Experiments were carried out using membranes coated with native collagen, in the absence (b, open bars) or presence of 10% FCS (b, shaded bars) in the lower compartment of the chamber as a chemoattractant. Alternatively, conditioned media from control MCF-7 cells and MCF-7 cells expressing the recombinant wild-type MSF or Phe535,539 mutant were used as a chemoattractant for control MCF-7 cells (c). The data are expressed as the mean ± SEM number of cells/field for 3 experiments carried out in triplicate. Comparisons were carried out by a one-way ANOVA with Newman-Keuls correction. §p < 0.01 vs. control MCF-7 cells, #p < 0.01 vs. Phe535,539 mutant, $p < 0.001 vs. control MCF-7 cells, †p < 0.001 vs. Phe535,539 mutant, *p < 0.05 vs. control MCF-7 cells, ‡p < 0.05 vs. Phe535,539 mutant.

We measured cell migration through Type I collagen in the presence or in the absence of 10% FCS, a chemoattractant, in the lower compartment of the Boyden chamber. MSF expression significantly increased MCF-7 cell migration, both in the presence (5.6-fold, p < 0.001) and absence of FCS (2.3-fold, p < 0.01), compared to control MCF-7 cells (Fig. 6b). As the in vivo expression of MSF may also regulate the migration of adjacent target cells in a paracrine manner, its migratory activity was then determined on MCF-7 cells using conditioned medium from either control MCF-7 cells or MCF-7 cells producing MSF as chemoattractant (Fig. 6c). MCF-7 cell migration was 2.2-fold higher with conditioned medium from cells producing MSF than with control conditioned medium (p < 0.05) (Fig. 6c).

Cell migration requires proteolytic activity to digest the extracellular matrix components. To investigate the contribution of Fn-proteinase activity in the motogenic function of MSF, MCF-7 cells stably producing the recombinant Phe535,539 mutant were established. Expression of the mutant MSF was confirmed by RT-PCR (Fig. 6a, inset). No Fn-proteinase activity was detected in conditioned medium of clone expressing the Phe535,539 mutant, as shown by the absence of substrate hydrolysis above the background level (Fig. 6a). Expression of the Phe535,539 mutant resulted in higher levels of cell migration in the presence of chemoattractant (2.6-fold, p < 0.05), compared to control MCF-7 cells (Fig. 6b). This increase was 65% lower than the increase observed for MCF-7 cells producing the wild-type MSF, although similar amounts of the 2 recombinant proteins were produced as determined by immunoblot (results not shown). In the absence of chemoattractant, however, no increase in cell migration was observed compared to control cells (Fig. 6b). In addition, medium from clone of MCF-7 cells producing the Phe535,539 producing did not stimulate the migration of control MCF-7 cells (Fig. 6c).

These results show that MSF stimulates the invasive capacity of cells in an autocrine and paracrine manner. Furthermore, they show that Fn-proteinase activity of MSF plays a major role in the pro-migratory property of MSF.

Discussion

Fn is a multifunctional protein whose activities are due to functional domains present both in the native “full-length” protein16, 17 and its multiple sub-fragments. Indeed, when released by limited proteolysis, these functional domains display specific activities, latent in the native Fn. For instance, some Fn fragments exhibit catalytic activities.5, 7, 18, 19, 20, 21, 22 This is the case for the GBD, which displays zinc-dependent collagenase/gelatinase activity,5, 6 called Fn-proteinase.4 We hypothesized that these cryptic activities of Fn domains may also arise from alternative splicing. Our results show that MSF, a recently characterized truncated Fn isoform obtained by alternative splicing of the Fn transcript, displays Fn-proteinase activity.

This is supported by several observations. First, there is a strong correlation between endogenous MSF expression and constitutive Fn-proteinase activity. Invasive mammary epithelial cancer cell lines MDA-MB231, BT549 and Hs578T produce MSF and a constitutive Fn-proteinase activity, whereas MCF-7 and T47D cells neither express MSF nor produce Fn-proteinase activity. Second, recombinant MSF expression in 2 different cell lines, HEK-293 and MCF-7, is associated with the presence of a typical pattern of Fn-proteinase activity generated after limited proteolysis of Fn.5 Conditioned medium from HEK-293 cells or MCF-7 cells that express recombinant MSF exhibited Fn-proteinase activity inhibited by the phosphinic pseudo peptide and not found in the medium of untransfected cells. Third, like most zinc-dependent metalloenzymes, MSF contains the consensus zinc-binding motif, HEXXH, which is located within the eighth Type I module, in the GBD. We demonstrated that Fn-proteinase activity was totally abolished when the 2 histidine residues of this motif were replaced by phenylalanine in the MSF sequence. This type of mutation is known to abolish the catalytic activity of zinc-dependent metalloenzymes.23, 24 These results demonstrate that the structure of the HEXXH motif is critical for Fn-proteinase activity and suggest that this sequence is the functional zinc-binding site of Fn-proteinase.

Our results on MSF mRNA expression in normal and breast cancer tissues are in agreement with those of Schor et al.11 We found that MSF messenger is weakly detectable in normal breast, whereas its expression is strongly induced in cancer. Furthermore, in tumors MSF is expressed in fibroblasts from the stroma, in cancer and endothelial cells.11In vitro study indicates that breast carcinoma cell lines express MSF mRNA and produce Fn-proteinase activity, as well as endothelial cells (not shown).

This typical expression pattern of MSF suggested that it could play a role in cancer. Furthermore, its endogenous expression in several adenocarcinoma cell lines is associated with their invasive capacity. Because cell migration is an important feature of tumor progression, the motogenic activity of MSF was investigated. We first determined whether the expression of MSF confers pro-migratory activity to MCF-7. We next determined its capacity to induce cell migration in a paracrine manner. As shown by Schor et al.11 the expression of MSF promoted migratory property of cells. In addition, we provide evidence that MSF exerts its motogenic activity in both an autocrine and a paracrine manner.

It is generally accepted that limited proteolysis of pericellular environment by proteinases is necessary for cell migration. We show that the Fn-proteinase activity of MSF plays a major role in its migration-stimulating effect. The Phe535,539 mutant exhibited a marked reduced capacity to increase cell migration compared to wild-type MSF. MSF may affect cell migration through its Fn-proteinase activity by directly digesting extracellular matrix components or by processing cell surface molecules. Further investigations are required to identify the substrates of Fn-proteinase. The 65% inhibition of cell migration property after mutating the HEXXH sequence of MSF also imply that the migratory activity of MSF is not exclusively dependent on Fn-proteinase activity. These results suggest that MSF induces cell migration through both Fn-proteinase-dependent and -independent mechanisms. Similarly, these 2 mechanisms may both participate in the degradation of cartilage induced by the GBD.8, 25

The Fn-proteinase-independent mechanism of cell migration induced by MSF may depend on either the Fibrin-I domain or the GBD, as these 2 domains are able to stimulate cell migration.2 The motogenic activity of MSF seems to be totally dependent on the GBD, however, because the mutation of Ile-Gly-Asp sequences, located within the seventh and ninth Fn Type I modules of the GBD, totally inhibited the migration-stimulating activity of MSF.11 The Ile-Gly-Asp sequence resembles the integrin-binding sequence Arg-Gly-Asp, and it is possible that these sequences target MSF at the cell surface, thus initiating signal transduction pathways responsible for the migratory activity of MSF.26 Conversely, the complete inhibition of cell migration resulting from the mutation of the Ile-Gly-Asp sequences suggests that these sequences are also critical for the Fn-proteinase-dependent mechanism of cell migration. The localization of MSF at the cell surface may be necessary for the biological activities of Fn-proteinase, as is the case for other proteinases.27, 28 Therefore, the cell surface localization of MSF may be common to the Fn-proteinase-dependent and -independent mechanisms of MSF migratory activity.

In conclusion, we show that MSF generated via the alternative splicing of the human Fn transcript exhibits Fn-proteinase activity. Fn-proteinase is therefore a cryptic activity generated by both proteolytic4, 5 and transcriptional events. Our results indicate that MSF stimulates cell migration in a manner that primarily depends on Fn-proteinase activity. These results show that MSF may play an important role in cancer progression, due to its Fn-proteinase activity.

Acknowledgements

We thank N. Lefin, F. Olivier and G. Roland for their excellent technical assistance. X.H. is supported by grants from the Fondation Simone et Cino del Duca, the Association Claude Bernard (France) and the European Commission (FP6). S.G. is supported by a grant from La Fondation de France. S.G. belongs to the European Vascular Network (http://www.evgn.org) a Network of Excellence supported by the European Community's Sixth Framework Programme for Research Priority 1 “Life Sciences, genomics and biotechnologies for health” (Contract No. LSHM-CT-2003-503254). This work was also supported by grants from Fortis Banque Assurance and the Interuniversity Attraction Poles Program from the Belgium Science Policy (Belgium).

Ancillary