membrane type-1 matrix metalloproteinase in humans but not in mice : potential implications for tumorigenesis *

Membrane type-1 matrix metalloproteinase (MT1-MMP) exhibits distinctive and important pericellular cleavage functions. Recently, we determined that MT1-MMP was trafficked to the centrosomes in the course of endocytosis. Our data suggested that the functionally important, integral, centrosomal protein, pericentrin-2, was a cleavage target of MT1-MMP in human and in canine cells and that the sequence of the cleavage sites were ALRRLLG↓LFG and ALRRLLS↓LFG, respectively. The presence of Asp-948 at the P1 position inactivated the corresponding site (ALRRLLD-LFGD) in murine pericentrin. To confirm that MT1-MMP itself cleaves pericentrin directly, rather than indirectly, we analyzed the cleavage of the peptides which span the MT1-MMP cleavage site. In addition, we analyzed glioma U251 cells, which co-expressed MT1-MMP with the wild type murine pericentrin and the D948G mutant. We determined that the D948G mutant that exhibited the cleavage sequence of human pericentrin was sensitive to MT1MMP, while unmodified murine pericentrin was resistant to proteolysis. Taken together, our results confirm that MT1-MMP cleaves pericentrin-2 in humans, but not in mice, and that mouse models of cancer probably cannot be used to critically examine MT1-MMP functionality.

the microtubular cytoskeleton, and delivered to the centrosomal compartment (16,18). The presence of MT1-MMP in the pericentrosomal space correlated with the cleavage of human pericentrin-2 (kendrin), an integral and functionally important centrosomal, 3336-amino-acid residue long, protein (19 -22), and chromosome instability in non-malignant epithelial Madin-Darby canine kidney cells (18,23). Centrosomes, spindle pole bodies, are cellular organelles that exhibit an ability to organize microtubules and to nucleate (24). The normal functionality of centrosomes is essential to the organization of the cytoskeleton and the mitotic spindle, self-duplication, and cell cycle progression (25,26). Conversely, centrosomal abnormalities, early predictors of carcinogenesis, promote mitotic spindle aberrations and chromosome instability, events which are frequently observed in neoplastic cells (27)(28)(29)(30)(31). It is well established that pericentrin supports the normal functioning of the centrosomes and the cytoskeleton and that its function is important to cell cycle progression (27,32,33). Despite the evident functional link of MT1-MMP activity with the cleavage of pericentrin observed in both human and canine cells (18), there were suspicions that MT1-MMP is indirectly, rather than directly, involved in these unorthodox, intracellular, proteolytic events. To demonstrate that MT1-MMP cleaves pericentrin directly, we used mutagenesis of murine pericentrin. The peptide sequence of murine, canine, and human pericentrin-2 is homologous. There is, however, a single amino acid substitution at the P1 position of the MT1-MMP cleavage site in murine pericentrin when compared with that of human and canine proteins. Consistent with the cleavage preferences of MT1-MMP (34), we hypothesized that Asp-948 inactivates the cleavage site in murine pericentrin.
Here, we reconstructed the MT1-MMP human cleavage site in the murine pericentrin-2 sequence. Consistent with the proteolysis of human pericentrin-2 at the ALRRLLG 1156 2L 1157 FG site, a single D948G mutation transformed murine pericentrin into the cleavage target of MT1-MMP. We suggest that these results confirm that MT1-MMP cleaves pericentrin in humans, but not in mice, and that the intracellular function of centrosomal MT1-MMP in humans cannot be fully recapitulated in the cellular and animal models in mice.

MATERIALS AND METHODS
Reagents-Rabbit polyclonal antibody AB815 to the hinge region of MT1-MMP was from Chemicon (Temecula, CA). Murine monoclonal antibody 5D1 to the MT1-MMP catalytic domain was generated jointly by our laboratory and Chemicon. Rabbit polyclonal antibodies 4b and M8 to the C-terminal and N-terminal parts of pericentrin, respectively, were characterized earlier (22,35).
The recombinant catalytic domain of human and murine MT1-MMP was each expressed in Escherichia coli and then purified from the inclusion bodies and refolded to restore the catalytic activity (36). The pep-* This work was supported by National Institutes of Health Grants CA83017, CA77470, and RR020843 (to A. Y. S.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1  tides ALRRLLGLFG, ALRRLLSLFG, and ALRRLLGLFG, which span the MT1-MMP putative cleavage site in murine, canine, and human pericentrin, respectively, were synthesized by GenScript (Piscataway, NJ). The peptides were cleaved for 2 h by the catalytic domain of MT1-MMP at the enzyme-substrate ratio of 1:1000 and the digest samples were analyzed by MALDI-TOF mass-spectrometry (18,34).
Mutagenesis and Cell Transfection-The cDNA construct of murine pericentrin was inserted in the pLPX7-blasticidin vector using routine manipulations. The oligonucleotide direct and reverse primers (5Ј-CTCCGAGATGCCCTGAGGAGACTTCTAGGCCTGTTTGGG-GACACACTGAAGGCAGC-3Ј and GCTGCCTTCAGTGTGTCCCC-AAACAGGCCTAGAAGTCTCCTCAGGGCACTCGGAG-3Ј, respectively; mutant positions are underlined) were used in PCR mutagenesis to insert the D948G mutation in the sequence of murine pericentrin. The presence of the mutation in the mutant construct was confirmed by DNA sequencing.
Mock-transfected human U251 glioma cells (mock), the cells stably overexpressing MT1-MMP (MT cells) and the cells in which MT1-MMP was stably silenced by the 5Ј-GAAGCCUGGCUACAGCAAUAU-3Ј siRNA construct (siMT cells) were constructed and partially characterized earlier (18,37,38). The 5Ј-GGUCCAUGCUGCAGAAAAACU-3Ј scrambled siRNA construct was used as a control. Both siRNA constructs were cloned in the psiLentGenepuromycin vector (Promega, Madison, WI). There was no effect of the scrambled siRNA construct on the expression of MT1-MMP in U251 cells (not shown). Cells were routinely grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum.
In this work, mock, MT, and siMT cells were additionally transiently transfected with the pLPX7 plasmid coding for the wild type and the D948G mutant murine pericentrin-2. Blasticidin-resistant cells were selected in 7 days. The expression of murine pericentrin was analyzed by Western blotting of the cell lysates, which were prepared from the total pool of blasticidin-resistant cells. To identify both the N-terminal and the C-terminal fragments of pericentrin, a mixture of 4b and M8 antibodies was used in these experiments. The expression of the murine pericentrin constructs was also analyzed by immunostaining the transfected cells.
Immunofluorescence-Cells were fixed in 4% paraformaldehyde for 10 min, permeabilized with 0.1% Triton X-100 for 5 min, and blocked with 1% bovine serum albumin. Cells were then incubated for 4 h with the primary antibody followed by incubation for 2 h with the speciesspecific secondary antibody conjugated with green Alexa Fluor 488 or red Alexa Fluor 594 (Molecular Probes, Eugene, OR). DNA was stained with 4Ј,6-diamidino-2-phenylindole. Images were acquired at a 600ϫ original magnification on a Olympus BX51 fluorescent microscope equipped with a cooled MagnaFire camera (Olympus, San Diego, CA).
General Methods-Gelatin zymography of MMP-2 from medium aliquots, cell surface biotinylation, cell lysate preparation, immunocapture of biotin-labeled MT1-MMP, and Western blotting analysis of MT1-MMP and pericentrin were performed as described in our earlier publications (18,39). The buffers used for the preparation of cell lysates were supplemented with a protease inhibitor mixture (pepstatin, leupeptin, bestatin, aprotinin, and E-64) and in addition, with phenylmethylsulfonyl fluoride and EDTA (1 mM each) to prevent additional, artificial proteolysis of the lysate samples.

Analysis of the Peptides That Span the MT1-MMP Cleavage Site in
Human, Canine, and Murine Pericentrin-Our earlier data suggested that cellular MT1-MMP cleaved human centrosomal pericentrin at the ALRRLLG 1156 2L 1157 FG cleavage site (18). Sequence alignment of human, murine, and canine pericentrin shows that the ALRRLLG 1156 2L 1157 FG and ALRRLLS 2068 2L 2069 FG sequences are present in humans and canines, respectively, whereas mice exhibit the ALRRLL D 948 L 949 FG sequence (Fig. 1). In agreement, in the cleavage tests in vitro, MT1-MMP cleaved the synthetic peptides that spun the human and canine cleavage site sequences. In contrast, the peptide ALRRLLD 948 L 949 FG that corresponded to the sequence of murine pericentrin was resistant to proteolysis by human MT1-MMP. Murine MT1-MMP also did not cleave this peptide (not shown). These data are consistent with the cleavage preferences of MT1-MMP, and they suggest that the presence of a negatively charged Asp residue at the P1 position is likely to inactivate the site and to protect the peptide substrate from MT1-MMP proteolysis (34).

MT1-MMP Cleaves the Murine Pericentrin D948G Mutant That Exhibits the MT1-MMP Cleavage
Site-To validate our in vitro cleavage data in a cell setting, we first mutated the sequence of murine pericentrin ALRRLLD 948 L 949 FG and then isolated the D948G mutant. This mutant exhibited the human MT1-MMP cleavage site ALRRLLG 948 2L 949 FG. We specifically selected robust U251 cells for these studies because they, as in other aggressive malignant cells, exhibit the required compensatory mechanisms to overcome the overexpression of pericentrin. Other cell types usually do not survive the overexpression of this multifunctional, integral, centrosomal protein, the levels of which control cell cycle progression and genetic stability (29). We next transfected mock, MT, and siMT cells with the wild type and D948G mutant murine pericentrin. Finally, we determined whether the pericentrin constructs were cleaved in the transfected cells. Fig. 2 shows that the expression of MT1-MMP was silenced in siMT cells. Low levels of MT1-MMP were observed in mock cells, which synthesized MT1-MMP naturally. The expression of MT1-MMP was up-regulated in MT cells, in which a characteristic 42-45-kDa self-degradation form of MT1-MMP was observed in addition to the 60 -64-kDa full-length forms of the protease. According to our earlier data and the observations of the others, the presence of the 42-45 self-proteolytic form in the cell samples indicates the presence of high levels of the catalytically potent MT1-MMP (11,12).
Both the wild type and the D948G mutant murine pericentrins were stable in mock and siMT cells. The cleavage of the wild type murine pericentrin was not observed in MT cells, which overexpress the protease. In contrast, there was an extensive proteolysis of the D948G mutant in MT cells. The combined size of the observed N-terminal and C-terminal cleavage fragments of pericentrin (105 and 140 kDa, respectively) correlated well with the size of intact murine pericentrin (240 -250 kDa). Because these experiments were performed with the individual total cell pools obtained via transiently transfecting cells with the pericentrin constructs, the levels of pericentrin differ insignificantly among the samples.
As shown by gelatin zymography of the medium samples, mock and siMT cells did not activate secretory MMP-2, which was produced naturally by glioma cells. Consistent with many other reports (12), transfection of the cells with MT1-MMP stimulated extensive activation of the MMP-2 zymogen and its conversion into the mature enzyme. The efficiency of MMP-2 activation in MT cells co-transfected with either unmodified murine pericentrin or the D948G mutant was similar when compared with the cells expressing MT1-MMP alone (Fig. 2). These results indicated that the intracellular, pericentrin-cleaving function of MT1-MMP and the status of pericentrin do not affect the proteolytic pericellular function of MT1-MMP. Overall, these results confirmed our in vitro peptide cleavage data and suggested that the reconstruction of the MT1-MMP human cleavage site ALRRLLG 948 2L 949 FG transformed murine pericentrin into the cleavage target of MT1-MMP. In agreement with the proteolysis of mutant pericentrin by MT1-MMP, there was an evident centrosomal co-localization of these two proteins in the transfected cells (Fig. 3).

DISCUSSION
From the 24 known human MMPs, an elevated expression of MT1-MMP is most closely associated with malignancies (5,40). There is extensive evidence that cell surface-associated MT1-MMP functions as one of the key players of pericellular proteolysis in humans and mice (3,7,8). Knock-out mice models generated volumes of highly valuable information about the proteolytic function of cellular MT1-MMP (41)(42)(43)(44)(45). There is evidence, however, that after being delivered to the plasma membrane, MT1-MMP, along with many other membrane-tethered proteins (46,47), is internalized. The clathrin-and caveolin-dependent internalization pathways are both involved in the internalization and the recycling of MT1-MMP (14,15). In the pathway of MT1-MMP through the cell compartment, the proteolytically potent MT1-MMP accumulates in the microtubulin cytoskeleton-organizing centers, the centrosomes. Our earlier work (18) suggested that both human and canine pericentrins were cleavage targets of centrosomal MT1-MMP. Pericentrin is an integral centrosomal protein, and it is essential to the normal functioning of centrosomes and to mitotic spindle formation (48). The expression of MT1-MMP in the centrosomes of either human or canine cells correlated with the presence of the proteolytic fragments of pericentrin. In addition, these events correlated with the induction of mitotic spindle aberrations and aneuploidy in non-malignant MDCK cells (18,23). The sequence alignment of the putative cleavage site in human and canine pericentrin (ALRRLLG 1156 2L 1157 FG and ALRRLLS 2068 2L 2069 FG, respectively) supported our biochemical and cellular experiments, because both Gly and Ser are compatible with the known cleavage preferences of MT1-MMP. This alignment also suggested that the corresponding sequence region of murine pericentrin (ALRRLLD 948 2L 949 FGD) is protected from MT1-MMP proteolysis because of the Asp-948 at the P1 position (underlined in the peptide) in the murine sequence. Consistent with this hypothesis, murine pericentrin was resistant to MT1-MMP. Based on this knowledge, we reconstructed the human cleavage site in the murine pericentrin sequence. As expected, the D948G mutant that exhibited the human cleavage site was cleaved by MT1-MMP. We believe that our experiments proved that MT1-MMP cleaves pericentrin directly, and we suggest that centrosomal MT1-MMP, through the cleavage of pericentrin, plays a unique role in human cells. In addition, the presence of the proteolytic fragments of pericentrin in the tumor biopsies, which express high levels of MT1-MMP, supports an important role of the MT1-MMP/pericentrin axis in cancer (18). Our data imply the pericellular function of MT1-MMP appears to be common across the species, whereas the intracellular, pericentrin-cleaving function of MT1-MMP is absent in mice. These hypotheses add another level of complexity to be overcome in our attempts to understand completely the tumorigenic functions of MT1-MMP in humans, and they warrant additional studies of the genetically redesigned animal models, so that they will fully recapitulate human tumorigenesis.