Eur. J. Biochem. 250, 751-757 (1997) 0 FEBS 1997
Membrane-type matrix metalloproteinases 1 and 2 exhibit broad-spectrum proteolytic capacities comparable to many matrix metalloproteinases Marie-Pia D'ORTHO I,*, Horst WILL', Susan ATKINSON', Georgina BUTLER*,Anthea MESSENT', Jelena GAVRILOVIC2, Bryan SMITH", Rupert TIMPL', Luciano ZARD16 and Gillian MURPHY* ' INSERM U296, FacultC de Mtdecine, Avenue du GCnCral Sarrail, CrCteil, France ' Strangeways Research Laboratory, Cambridge, UK InVitek GmbH, Berlin-Buch, Germany Celltech Therapeutics Ltd, Slough, UK ' Max-Planck-Institut fur Biochemie, Martinsried, Germany ' Laboratory of Cell Biology, Istituto Nazinale per la Ricerca sul Cancro, Genoa, Italy (Received 38 September 1997) - EJB 97 1334/4
Soluble proenzyme forms of the catalytic domains of membrane-type matrix metalloproteinases 1 and 2 (MTI-MMP and MT2-MMP) and a form of MTI-MMP containing the catalytic and hemopexin domains were expressed as soluble recombinant proteins. Purified, activated forms of the MT-MMP were shown to degrade fibronectin, tenascin, nidogen, aggrecan and perlecan. Only MT2-MMP showed activity against laminin. MT1-MMP retaining the hemopexin domain was able to specifically cleave native type-I and type-I11 collagens into the 3/4- 1/4 fragments typical of the specific collagenases. The catalytic domain alone did not retain this activity. The MT-MMP did not degrade interleukin-lp, but, similarly to many other MMP, could process a pro [tumor necrosis factor (TNF) a] fusion protein to release mature TNF. However, the latter was subsequently degraded into smaller fragments. These results demonstrate that, in addition to their ability to activate other MMP, such as progelatinase A/proMMP2 and procollagenase-3/proMMPI 3, MT-MMP degrade a number of extracellular matrix macromolecules. Their location at the surface of cells implies that they could play a significant role in the modulation of cell-matrix interactions. Keywords: tissue inhibitor of matrix metalloproteinase; extracellular matrix ; cytokine; wound healing ; invasion.
Extracellular matrices (ECM) play a key role in cell biology. Not only do they contribute to tissue architecture, but they also provide immobilized ligands for cellular receptors, sequester growth factors, and in basement membrane form selectively permeable barriers between tissue compartments. Hence, ECM components influence and regulate cell growth and differentiation, cell adhesion, mobility and spreading. Therefore their turnover is an important process in physiological and pathophysiological situations. Matrix metalloproteinases (MMP) have a prominent role in the degradation of ECM, and have a wide spectrum of other substrates, cleaving for example interleukin 1p (ILIP) [I] or pro[tumor necrosis factor (TNF) a] [2]. MMP have been classified into different subclasses according to their substrate specificity and amino-acid-sequence alignment. The major subclasses identified so far are interstitial collagenases, gelatinases, stromelyCorrespondence to M.-P. d'Ortho, INSERM U296, FacultC de Mtdecine de CrCteil, 8 Avenue du GCnCral Sarrail, F-94010 CrCteil, France Fax: +33 1 48 98 17 77. E-mail:
[email protected] Abbreviations. AcHgPhNH,, aminophenylmercuric acetate; ECM, extracellular matrix ; GST, glutathione S-transferase: ILlP, Interleukin 1P; MMP, matrix metalloproteinase: MT-MMP, membrane-type matrix metalloproteinase; TIMP, tissue inhibitor of matrix metalloproteinase; TNL, large isoform of tenascin-C; TNS, small isoform of tenascin-C; TNF, tumor necrosis factor.
sins, and metallo-elastase [3]. Recently, another subclass of MMP has been described: the membrane-type MMP (MTMMP), so called because of the presence of a transmembrane domain at their C-termini. Four MT-MMP have been cloned, identified by screening cDNA libraries for similarity to conserved regions of the known MMP genes [4-71. The localization of MT-MMP as ectoenzymes, rather than soluble proteins, has led to the hypothesis that MT-MMP are the upstream activators in the progelatinase AlproMMP2 proteolytic cascade. Using soluble recombinant proteins, studies have shown that MT1MMP and MT2-MMP can process progelatinase Alpro-MMP2 into its active form [8-101, in accordance with previous data obtained by transient transfection [4, 51. The ability of MTIMMP to degrade some ECM components has been shown [9, 111. The aim of our study was to further define the substrate specificity of MT-MMP using soluble recombinant MT1 -MMP and MT2-MMP.
MATERIALS AND METHODS Expression and purification of MT-MMP. Recombinant MT-MMP were expressed as soluble proteins with a C-terminal His tag in the periplasm of Escherichia coli as described previously [lo]. The three constructs used were the proform of the catalytic domain of human MT1-MMP, des-(269-559)-MTI-
752
d'Ortho et al. ( E m J. Biochern. 250)
confirmed by treatment with six zinc metalloproteases. Each treatment produced a 17-kDa fragment (approximate apparent molecular mass on SDSPAGE) as expected for mature TNF Activation of des-(269-559)-MTl-MMP and activity of with the same N-terminal sequence as TNFa produced by a variMT-MMP using a fluorescent substrate. Conversion of pro- ety of cell lines transfected with human proTNFa. Recombinant forms of progelatinase A/proMMP2 [191, proMT-MMP to their active forms was studied by SDS/PAGE followed by silver staining, and by hydrolysis of a fluorescent sub- matrilysin/proMMP7 [20], procollagenase 3/proMMP13 [21], human TIMP-1, TIMP-2 and TIMP-3 [22-241 were prepared as strate 110, 121. The activation of des-(269-559)-MTl-MMP was effected using 5 pg/ml tosylphenylalanine-chloromethane- described previously. Activation of progelatinase A/proMMP2, treated trypsin at 25°C and stopped using 50 pglrnl soybean promatrilysin/proMMP7 or procollagenase-3/proMMP13 was performed by treatment with 1 mM 4-aminophenylmercuric acetrypsin inhibitor [lo]. Activation of des-(501-559)-MTl-MMP tate (AcHgPhNH,) as described previously [21, 251. was obtained by incubation with activated des-(269-559)-MTlMMP (1 :20 molar ratio). Activation of des-(263 -628)-MT2Proteolytic cleavage of ECM components, ILlB and GSTMMP was shown to occur spontaneously by self-processing at proTNFa. ECM substrates, I L l p or GST-proTNFa were incubated in the presence of MT-MMP in 100 mM Tris/HCl, 30 mM 37°C [S]. Hydrolysis of a quenched fluorescent peptide substrate (7- CaCI,, pH 7.9, at 37"C, for 16 h, unless otherwise stated. Enmethoxycoumarin-4-y1)acetyl-Pro-Leu-Gl y-Leu-[3-(2,4-dinitro- zyme concentrations varied from 2 pM to 12.5 nM depending phenyl)-~-2,3-diarninopropionyl]-Ala-Arg-NH,by MT-MMP on the substratelenzyme molar ratio. Negative controls always was assayed as described previously [lo, 121. The specific activ- included substrate incubated alone in the assay buffer for the ity of each recombinant enzyme was expressed as nmol syn- same time. Positive controls were carried out by incubating subthetic substrate degraded . min-' . mg enzyme-'. Assays com- strate with either AcHgPhNH,-activated gelatinase AMMP2, paring des-(269 -559)-MT1-MMP, des-(501- 559)-MT1-MMP AcHgPhNH,-activated collagenase 3/MMP13 or AcHgPhNH,and des-(263 -628)-MT2-MMP with respect to their ability to activated matrilysinlMMP7, depending on the substrate. degrade a given ECM component were performed with equivaCleavage products of fibronectin, nidogen, TNL, TNS, collalent amounts of enzyme determined by active-site titration using gens and domains of perlecan were analyzed by SDSPAGE untissue inhibitor of matrix metalloproteinase (T1MP)-2 as de- der reducing conditions followed by Coomassie blue or silver staining. scribed previously [lo]. Laininin cleavage was analyzed by western blotting using Sources of ECM proteins, glutathione S-transferase (GST)-proTNFa, other MMP and TMP 1-3. Fibronectin was an anti-(mouse laminin) Ig at a dilution recommended by the obtained from plasma and purified from its gelatinase contami- manufacturer. Immunoblotting for fragments of soluble fibronant by immobilized metal-affinity chromatography using a nectin was performed with mAbs 3E1 (clone I) and 4B2 (clone modification of the method of Smilenov et al. [13]. The column 111). The dilutions were 1 : 1000 (clone I) and 1 :400 (clones III), was washed with 10 mM imidazole in 50 mM Tris/HCI, 50 mM respectively, as reported previously [26]. TNL cleavage sites 6-aminohexanoic acid, 20 mM sodium citrate, pH 7.6, and re- were mapped by immunoblotting using different tenascin-C tained fibronectin was eluted with 100 mM EDTA. The fibro- specific mAbs [14]. Degradation of type-I collagen was tested nectin-containing fractions were dialyzed against 50 mM Tris/ using the diffuse-fibril assay at 35°C [18]. Degradation of agHCI, pH 7.4, 150 mM NaCI, 0.02% sodium azide. These frac- grecan was tested using a bead assay, as reported previously tions were analyzed by zymography and found to be free of [27], and was performed at 37°C for 16 h. GST-proTNFa was detectable gelatinolytic activity. Antibodies to fibronectin do- incubated for 1 h at 37°C with different amounts of des-(269559)-MTI-MMP or des-(263-62S)-MT2-MMP, and cleavage mains were from Gibco BRL. Tenascin-C small (TNS) and large isoforms (TNL) were ob- products were analyzed by SDWPAGE followed by silver tained from baby hamster kidney cells transfected with tenascin- staining. NH,-terminal amino acid sequencing of the cleavage C cDNA constructs, and mAb specific to their different epitopes products was performed by automated Edman degradation using were prepared as reported previously [14]. Mouse laminin and a 470 A gas-phase sequencer (Applied Biosystems, Inc) with type-IV collagen, and a polyclonal antibody directed to laminin on-line ABI 190A HPLC for analysis of phenylthiohydantoin were from Collaborative Research. Recombinant mouse nidogen derivatives. was purified from the culture medium of stably transfected human cell clones [15]. Mouse perlecan domains I1 [16] and V (unpublished data), and three fragments of domain I11 [17] were RESULTS AND DISCUSSION produced in stably transfected human embryonic kidney 293 cell MT-MMP are members of the MMP family. Given their loclones. Type-I collagen was prepared from rat skin, as described previously [18], and bovine skin type-I11 collagen was a gener- calization at the cell surface, initial studies mainly addressed ous gift from Dr M. Barnes (Strangeways Research Laboratory, the question of their potential role as progelatinase A/proMMP2 Cambridge, UK). activators. However, their similarity to other members of the GST-proTNFa was prepared by intracellular expression in E. MMP family [4-71 especially in the region of the catalytic site, coli, followed by sonication of the cells and Sepharose-glutathi- strongly suggested that they should be able to degrade ECM one affinity chromatography. The proTNFa sequence retained components. Difficulties in studying the biochemical properties the 25 residues between the membrane-spanning region of pro- of MT-MMP came in part from technical problems associated TNFa and the N-terminus of the mature TNFa. This sequence with isolating and purifying membrane proteins. Such difficulwas fused at its N-terminus to the C-terminus of GST with a C- ties have been overcome by generating soluble mutants of MT1terminal six-residue extension of Leu-Val-Pro-Arg-Gly-Ser. The MMP and MT2-MMP in E. coli as described previously [8, lo]. overall mass of this fusion protein was calculated to be These mutants have been used to examine the proteolytic prop46 357 Da. SDS/PAGE of the prepared GST-proTNFa showed erties of MT-MMP towards various substrates, including ECM the major product to have the expected size of 46 kDa, with components and the cytokines I L l p and TNFa. minor species of about 63 kDa and 33 kDa also present. The Fibronectin was degraded by des-(269-559)-MTl-MMP, suitability of this fusion protein as a substrate for proteases was des-(501-559)-MTI-MMP and des-(263 -628)-MT2-MMP MMP, the proform of the catalytic domain of human MT2-MMP, des-(263 -628)-MT2-MMP, and MT1-MMP lacking its transmembrane domain, des-(501-559)-MTl-MMP.
753
d’Ortho et al. ( E m J. Biochem. 250)
1)
A B C D E
F G H
2)
250-
250-
97-
97-
786 8-
7868-
55-
55-
0 0 . 5 2 3 4 5 6 8 1 8 2 4
Fig. 1. Degradation of fibronectin. (Panel 1) Degradation of fibronectin by des-(269-559)-MTI-MMP, des-(263 -628)-MT2-MMP and des-(501559)-MTLMMP. Degradation products were analyzed by silver-stained SDWPAGE. Molecular-mass standards are shown (in kDa). Lane A, unincubated fibronectin; lane B, fibronectin incubated alone; lanes C and D, fibronectin incubated with des-(269-559)-MTI-MMP at enzymehbstrate nioiar ratios of 1:30 (lane C) and 1 : I 5 (lane D) ; lanes E and F, fibronectin incubated with des-(S01-559)-MTI-MMP at enzymeisubstrate molar ratios of 1:7.5 (lane E) and 1 : l S (lane F); lanes G and H, fibronectin incubated with des-(263-628)-MT2-MMP at an enzymehbstrate molar ratios of 1:7.5 (lane G) and 1.15 (lane H). (Panel 2) Time course of fibronectin degradation by des-(269-559)-MTI-MMP. Fibronectin was incubated with des-(269-559)-MTI-MMP at an enzymehbstrate molar ratios of 1:30. Degradation products were analyzed by silver-stained SDS/ PAGE. Molecular-mass standards are shown (kDa). Numbers on the top of the figure indicate the duration of incubation (11).
ABC
2)
B
A
ABC
P-
138-
a{=
7560-
Fig. 2. Characterization by western blotting of fibronectin fragments generated by des-(269-55Y)-MTl-MMP or gelatinase A/MMP2. Lanes A, fibronectin incubated alone; lanes B, fibronectin incubated with des-(269-559)-MTl-MMP; lanes C, fibronectin incubated with gelatinase NMMP2. (Panel I) Western blot using a mAb directed to an epitope near the C-terminal heparin-binding domain of fibronectin (clone I). The molecular-mass (kDa) of two of the fragments generated by des(269-559)-MTl-MMP only and not by gelatinase N M M P 2 is indicated on the left; (panel 2), western blot using an mAb directed to an epitope near the gelatin-binding domain (clone 111). The molecular-mass (kDa) of two of the fragments generated by des-(269-559)-MTI-MMP but not by gelatinase A/MMP2 are indicated.
(Fig. 1). Degradation of fibronectin occurred in a time-dependent and dose-dependent manner. Several fragments of 200,160, 140, 110, 80 and 60 kDa were generated. Fragments of 200 kDa and 80 kDa were the first products to appear after 2 h incubation, and 140-, 110-, 80- and 60-kDa fragments were the main cleavage products. The 200-kDa and 80-kDa products were the same size as those generated by gelatinase NMMP2 in a previous study [28]. Degradation of fibronectin by des-(269-559)MT1-MMP or by des-(263 -628)-MT2-MMP was completely inhibited by TIMP-2 and TIMP-3 in a 1:l molar ratio (TIMPI MT-MMP), and less efficiently by TIMP-1 (5:l molar ratio; data not shown), in agreement with previous data [lo].
A B C D E
F G H
I
Fig. 3. Degradation of type-I and type-I11 fibrillar collagens by MT1MMP. Degradation products were analyzed by electrophoresis on 7 % SDSlPAGE and stained with Coomassie blue. The positions of the collagen 1and a chains are indicated. (A) Type-I collagen (10 pg) was incubated at 25°C for 20 h alone (A), with 100 ng collagenase-3/MMP13 (B), with 25 ng active des-(S01-559)-MTI-MMP (C), with 50 ng des(501-S59)-MTI-MMP (D) and with 60 ng des-(269-559)-MTl-MMP (E). (B) Type-I11 collagen (30 pg) was incubated at 25°C for 20 h alone (F), with SO ng collagenase-3IMMP13 (G), with SO ng des-(501-559)MTI-MMP (H) and with 60 ng des-(269-559)-MTI-MMP (I).
The degradation products of fibronectin generated by des(269-559)-MTl-MMP and des-(263 -628)-MT2-MMP were characterized by immunoblotting using mAbs and compared with those generated by gelatinase AlMMP2. Fig. 2 shows results obtained with des-(269-559)-MTI-MMP and gelatinase A/MMP2. The characterization of fragments obtained with des(263 -628)-MT2-MMP was identical with those for des-(269559)-MTl-MMP (data not shown). Two bands, of 75 kDa and 60 kDa, were recognized by the antibody against the C-terminal heparin-binding domain (clone I, Fig. 2) but not by the antibody raised against the N-terminal gelatin-binding domain of fibro-
d’Ortho et al. ( E M J. Biochem. 250)
754
Table 1. Reactivity of fragments of the TNL isoform with mAbs. Digests of TNL obtained with des-(269-559)-MTI-MMP, des-(263-628)MT2-MMP or gelatinase AIMMP2 were analyzed by immunoblotting with six tenascin-C-specific mAbs [14]. Positive (+) or negative (-) reactions of the two resistant fragments are reported. Des-(269-%9)-MTl -MMP and des-(263 -628)-MT2-MMP cleaved TNL in the A3 fibronectin-like repeat, as did gelatinase A/MMP2. Enzyme
Fragment
Monoclonal antibody BC-4
BC-2
a-A3
a-A2
a-B
a-D
kDa Gelatinase A/MMP2
190 120
Des-(269 - 559)-MTl-MMP
190 120
Des-(263-628)-MT2-MMP
190 120
1) A B C D E F G H I J
2 ) A B C D E
F
AB*
a
125 115
Fig. 4. Degradation of laminin and nidogen by MT-MMP. (Panel 1) Cleavage of laminin by MT-MMP, analyzed by western blotting. Arrows on the left indicate the position of the laminin chains: A, a1 chain; B, /l1 and y 1 chains. The arrow on the right indicates the approximate molecularmass of the fragment generated (180 kDa). Lane A, unincubated laminin; lane B, laminin incubated alone; lanes C-E, laminin incubated with des(269-559))MTI-MMP at molar ratios of enzyme/substrate of 1 : 1, 1 : 5 and 1 : 100, respectively; lanes F and G, laminin incubated with des-(501S59)-MTl-MMP at molar ratios of enzymelsubstrate of 1 : 1 and 1:5, respectively; lanes H and I, laminin incubated with des-(263-62X)-MT2MMP at molar ratios of enrymelsubstrate of 1 : 1 and 1:S, respectively; lane J, laminin incubated with trypsin ( 5 pg/ml) and trypsin inhibitor (50 pg/ ml). (Panel 2) Cleavage of nidogen. Lane A, unincubated nidogen; lane B, nidogen incubated alone; lane C, nidogen incubated with des-(269559)-MTI-MMP at a molar ratios of emymelsubstrate of I : 10; lane D, nidogen incubated with des-(S01-559)-MTl-MMP at a molar ratios of enzymeisubstrate of 1 : 10; lane E, nidogen incubated with des-(263-628)-MT2-MMP at a molar ratios of enzymeisubstrate of’ 1 :S; lane F, nidogen incubated with matrilysin/MMP7 at a molar ratios of enzymelsubstrate of 1 : 5. Incubation with matrilysinlMMP7 gave rise to different fragments than with des-(269-559)-MTl-MMP or des-(263-628)-MT2-MMP. Arrows and numbers on the left indicate the molecular-m fragments generated by MT-MMP; arrows and numbers on the right indicate the two extra fragments generated by matrilysinlMMP7.
nectin (clone 111, Fig. 2). Some of the bands generated by MTMMP seemed to be unique to MT-MMP compared with gelatinasc A/MMP2 (Fig. 2). Proteolytic fragments of fibronectin are expected to contribute to cell-regulatory processes, as suggested by previous studies [26]. They could block specific receptors for intact fibronectin, decrease attachment of tumour cells and lead to alteration in migration. Considering that the C-terminal heparin-binding domain is the major site for interactions with proteoglycans [29], the 70-kDa and 65-kDa fragments could interfere with fibronectin-proteoglycan interactions. Fibronectin fragments have been reported to be involved in the induction of specific genes [30],and particularly in induction of other proteases involved in the hydrolysis of the ECM [31]. With their ability to degrade fibronectin, MT-MMP may play a role in wound healing, since fibroblasts have been shown to migrate into wounds on fibronectin fragments [32], and the latter are chemotactic for rnonocytes 1331. TNL was degraded by des-(269 - 559)-MTl-MMP, giving two products of 190 kDa and 120 kDa, and was degraded similarly by des-(501-559)-MTI-MMP and des-(263 -628)-MT2MMP (data not shown). The fragments of TNL generated by
MT-MMP were of similar size to those generated by gelatinase A/MMP2 [14]. By contrast, TNS was resistant to cleavage (data not shown). TNL has been shown to be more susceptible to proteolytic cleavage than the small isoform 1141. Our results suggest that cleavage of TNL occurs within the alternatively spliced fibronectin-like repeats present in TNL and not in TNS. In an attempt to map the cleavage sites, digests obtained with des-(269-559)-MTl-MMP and des-(263-628)-MT2-MMP were analyzed by immunoblotting using specific mAbs. Results showed that the 190-kDa fragment includes the N-terminal end of the molecule (BC7), the EGF-like repeat (BC4) and the fibronectin-like repeat A1 to A2. Des-(269-559)-MTl-MMP and des-(263-628)-MT2-MMP completely digested the antibody epitope in fibronectin-like repeat A3 of TNL, in a similar manner to gelatinase A/MMP2 (Table 1). These results suggest that MT-MMP could act in concert with activated gelatinase AIMMP2 to degrade matrix components such as fibronectin and tenascin. This could be of particular importance in tumour spreading and wound healing, where these enzymes and ECM proteins are preferentially expressed [14, 34, 351.
d’Ortho et al. (EUI:J. Biochem. 250)
1)
A B C D E F G H
*
200 +
19‘76
66 -b 55 -b
36.5-b
31 + 21.5+ 14 -b
6+
2)
A B C D E F G H
200 +
‘gs
66 +
55 -&
36.5+ 31 +
21.5+
‘8 z Fig. 5. Degradation of perlecan fragments by des-(269-559)-MTlMMP and des-(263-628)-MT2-MMP. Silver-stained 8 % SDS/PAGE. Lanes A and E, unincubated substrate; lanes B and F, substrate incubated alone; lanes C and G, substrate incubated in the presence of des-(269559)-MT1-MMP (molar ratios of enzymehbstrate, 1: 5 ; lanes D and H, substrate incubated in the presence of des-(263-628)-MT2-MMP (enzyme/substrate molar ratios, 1: 5). The positions of trypsin and trypsin inhibitor used to activate des-(269-559)-MTl-MMP are indicated by white stars (lanes C and G). (Panel 1)Lanes A-D, perlacan domain 11; lanes E-H, perlacan domain 111. (Panel 2) Lanes A-D, fragment 2 of perlacan domain 111; lanes E-H, fragment 3 of perlacan domain Ill.
Fibrillar type-I and type-111 collagens were degraded by des(501 -559)-MTl-MMP, but not by des-(269-559)-MTl-MMP (Fig. 3) or des-(263-628j-MTZ-MMP (data not shown). The collagen cleavage pattern was the same as that obtained with collagenase-3NMP13, which was used as a positive control, and showed a 3/4-1/4 cleavage (Fig. 3), which differs from the results obtained by Pei and Weiss [9]. The specific activity of des-(501-559)-MTl-MMP, calculated from degradation of fibrillar [‘4C]collagen, was 77.3 pg . min ’ . nmol des-(501559)-MTl-MMP-’. This result suggests that MT1-MMP is a collagenase, with triple-helicase activity determined by the hemopexin domain, as demonstrated previously for other collagenases [21]. Cleavage of type-I11 collagen occurs in a nonhelical interruption within the triple helix, and similar results were obtained using full-length collagenase-3/MMP13 (Fig. 3). It was shown previously that the truncated catalytic domain of collagenase-3/MMP13 had activity against type-I11 collagen [21], but this could not be demonstrated for des-(269-559)MT1-MMP (Fig. 3 B). Since MT-MMP have been found to be expressed in various tumours [4, 25, 361, in which disruption of basement-membrane
755
is thought to be an important step, degradation of the basement membrane components was examined. Type-IV collagen was resistant to cleavage (data not shown) by the three forms of recombinant MT-MMP, in accordance with results obtained by Pei and Weiss [9]. Laminin degradation analyzed by western blotting showed that (11,pl and y l chains of laminin were degraded by des-(263-628)-MT2-MMP but not by des-(269-.559)-MTlMMP nor by des-(501-559)-MTI-MMP (Fig. 4) in contrast to a previous study, which has reported the cleavage of the laminin a1 chain by MT2-MMP [9]. Nidogen (Fig. 4) was degraded by the three soluble forms of MT-MMP, but the bands generated were different from those generated by matrilysin MMP7 (Fig. 4). The cleavage of nidogen by MT-MMP could be important in the disruption of basement membranes, as several studies have supported a model in which nidogen provides structural stability to basement membranes by bridging laminin to typeIV collagen and by anchoring a variety of components, such as collagen or perlecan [37]. The ability of MT-MMP to degrade proteoglycans was examined. Degradation of aggrecan was examined in the proteoglycan-bead assay [27]. Des-(269- 5 9 - M T l - M M P released 9180 pg chondroitin sulfate/nmol in 16 h. Incubation of recombinant perlecan fragments showed that domain I1 (Fig. Sj, and fragments 1-3 of domain 111 were degraded by des-(269-559)MT1-MMP and des-(263 -628)-MT2-MMP. However, perlecan fragments 2 and 3 of domain 111 were degraded only slightly by des-(263 -628)-MT2-MMP. In contrast domain V was resistant to proteolysis (data not shown). These results suggest that MTMMP cleave full-length perlecan at multiple sites within the molecule. Cleavage of perlecan could be of particular importance as this molecule is involved in growth-factor binding [38]. In this regard, it has been shown previously that cleavage of perlecan by other MMP released bound basic fibroblast growth factor [39]. TNFa and ILlP are highly expressed at inflammatory and tumour sites. In view of previous data describing the role of MMP in the degradation of ILlP [I] and activation of proTNFa [ 2 ] ,their cleavage by MT-MMP was examined. ILlP was resistant to cleavage but GST-proTNFa was cleaved yielding bands of 17 kDa and 13 kDa (data not shown). Identical results were obtained upon incubating GST-proTNFa with des-(269-559)MT1-MMP, des-(263 -628)-MT2-MMP or des-(501-559)MTI-MMP. Sequencing of the 17-kDa band demonstrated that it corresponded to mature TNFa (V77RSSSRTPSD) while the 13-kDa band sequence was consistent with cleavage within TNFa at the sequence L113LANGVELRDNQL. These results suggest that MT-MMP cleave pro-TNFa to its mature form but further degrade it into smaller fragments. Our study confirms that MTI-MMP and MT2-MMP exhibit a wide spectrum of proteolytic activities, as expected from their sequence similarity to other MMP. Sequence alignment of MT1MMP and MT2-MMP with other MMP showed the presence of a Pro C-terminal to the Met turn, Pro235 and Pro217 for MTIMMP and MT2-MMP, respectively. This residue is believed to provide a hydrophobic environment necessary for the catalytic zinc atom [40], and its importance has been discussed for human stromelysin 3/MMP11 [41]. Des-(269-559)-MTl-MMP and des-(S01-559)-MTl-MMP degraded the same substrates with comparable efficiency, except for collagens I and 111. Cleavage of the latter substrates required the presence of the C-terminal hemopexin domain in addition to the catalytic domain. The Cterminal domain of the soluble forms of MT1-MMP appeared to have no role in proteolytic specificity towards other substrates. Des-(269-559)-MTl-MMP and des-(263-628j-MT2MMP appeared to cleave the same substrates, with the exception
756
d’Ortho et al. (Eur: J. Biochem. 250)
of laminin, which was degraded only by des-(263-628)-MT2MMP. MT-MMP may therefore play a role in ECM degradation both directly and indirectly: they are associated with the cell membrane and so localize matrix digestion to the vicinity of the cell surface, and can activate progelatinase A/proMMP2 [9, 101 and procollagenase-3/proMMP13 [42], leading to a metalloproteinase-activation cascade and initiating an amplification of the breakdown of ECM. We thank J. S. Emtage, M. Wales and M. Dorning for the preparation of GST-proTNFa. M.-P. d’Ortho was the recipient of a fellowship from the lnsfitut National de la Sariti et de la Recherche Midicale (France). This work was supported by the Arthritis and Rheumatism Council, the Wellcome Trust, The Cancer Research Campaign and the Medical Research Council (United Kingdom) and by funds of the Associazione Zfuliana per la Ricerca sul Cancro (AIRC) and the Consiglio NuZionale delle Richerche, ‘Progetto Finalizzuto: Applicazioni Cliniche Della Ricerca Oncologica’ (Italy).
REFERENCES 1. Ito, A,, Mukaiyama, A., Itoh, Y., Nagase, H., Thorgesen, I. B., Enghild, J. J., Sasguri, Y. & Mori, Y. (1996) Degradation of ILlP by MMP, .I. Biol. Chem. 271, 14657-34660. 2. Gearing, A. J . H., Beckett, P., Christodoulou, M., Churchill, M., Clements, J., Davidson, A. H., Drummond, A. H., Galloway, W. A,, Gilbert, R., Gordon, J. L., Leber, T. M., Mangan, M., Miller, K., Nayee, P., Owen, K., Patel, S., Thomas, W., Wells, G., Wood, L. M. & Wooley, K. (1994) Processing of TNFa precursor by metalloproteinases, Nature 370, 555-556. 3. Murphy, G. & Reynolds, J. J. (1993) Extracellular matrix degradation, in Connective tissue and its heritable disorders. Molecular, genetic, and medical aspects (Royce. P. M. & Steinniann, B., eds) pp. 287-316, Wiley-Liss, New York. 4. Sato, H., Takino, T., Okada. Y., Cao, J., Shinagawas, A., Yamamoto, E. & Seiki, M. (1994) A matrix metalloproteinase expressed on the surface of invasive tumor cells, Nature 370, 61 -65. 5. Takino, T., Sato, H., Shinagawa, A. & Seiki, M. (1995) Identification of the second membrane-type matrix metalloproteinase (MTMMP-2) gene from a human placenta cDNA library, J. B i d . Cliem. 270, 23013-23020. 6. Will, H. & Hinzmann, B. (1995) cDNA sequence and mRNA tissue distribution of a novel human matrix metalloproteinase with a potential transmembrane segment, Euu. J. Biochem. 231, 602608. 7. Puente, X. S., Pendas, A. M., Llano, E., Velasco, G. & Lopez-Otin, C. (1996) Molecular cloning of a novel membrane-type matrix metalloproteinase from a human breast carcinoma, Cancer Res. 56, 944-949. 8. Butler, G. S., Will, H., Atkinson, S. J. & Murphy, G. (1997) Meinbrane-type-2 matrix metalloproteinase can initiate the processing of progelatinase A and is regulated by the tissue inhibitors of metalloproteinases. Eur: J. Biochrm. 244, 653-657. 9. Pei, D. & Weiss, S. J. (1996) Transmembrane-deletion mutants of the membrane type matrix metalloproteinase-I process progelatinase A and express intrinsic matrix-degradading activity. J. Biol. Chem. 271, 9135-9140. 10. Will, H., Atkinson, S., Butler, G. S., Smith, B. & Murphy, G. (1996) The soluble catalytic domain of membrane-type 1 matrix metalloproteinase cleaves the propeptide of progelatinase A and initiales autoproteolytic activation. Regulation by TIMP-2 and 3, .I. Aiol. Chem. 271. 17119-17123. 11. Ohuchi, E., Iniai, K., F u j i Y., Sato, H., Seiki, M. & Okada, Y. (1997) MTI-MMP digests interstitial collagens and other extracellular matrix macromolecules, J. Bid. Chem. 272, 2446-2451. 12. Knight, C. G., Willenbrock, F. & Murphy, G. (1992) A novel coumarin-labelled peptide for rensitive continuous asays of the matrix metalloproteinascs. FEBS Lert. 296, 263 -266. 13. Smilenov. L., Forsberg, E., Zeligman, I., Sparrman, M. & Johansson, S. (1992) Separation of fibronectin from a plasma gelatinase using
immobilized metal affinity chromatography, FEBS Lett. 302, 227-230. 14. Siri, A,, Knauper, V., Veinara, N., Caocci, F., Murphy, G. & Zardi, L. (1995) Different susceptibility of small and large human tenascin-C isoforms to degradation by matrix metalloproteinases, J. Biol. Chem. 270, 8650-8654. 15. Fox, J . W., Mayer, U., Nischt, R., Aumailley, M., Reinhart, D., Wiedeman, H., Mann, K., Timpl, R., Krieg, T., Engel, J. & Chu, M.-L. (1991) Recombinant nidogen consists of three globular domains and mediate binding of laminin to collagen, EMBO J. 10, 3137- 3146. 6. Costell, M., Sasaki, T., Mann, K., Yamada, Y. & Timpl, R. (1996) Structural characterization of recombinant domain I1 of the basement membrane proteoglycan perlecan, FEBS Lett. 396, 127131. 7. Schulze, B., Sasaki, T., Costell, M., Mann, K. & Timpl, R. (1996) Structural and cell-adhesive properties of three recombinant fragments derived from perlecan domain 111, Matrix B i d . 15, 349357. 18. Cawston, T. E. & Barrett, A. J. (1979) A rapid reproducible assay for collagenase using (’*C) acetylated collagen, Anal. Biochem. 99, 340-345. 19. Murphy, G., Willenbrock, F., Ward, R. V., Cockett, M. I., Eaton, D. & Docherty, A. J. P. (1992) The C-terminal domain of 72 kD gelatinase A is not required for catalysis, but is essential for membrane activation and modulates interactions with tissue inhibitors of metalloproteinases, Biochem. J. 283, 637-641. 20. Murphy, G., Cockett, M. I., Ward, R. V. & Docherty, A. J. P. (1991) Matrix metalloproteinase degradation of elastin, type IV collagen and proteoglycan, Biochem. J. 277, 277-279. 21. KnLiuper, V., Lhpez-Otin, C., Smith, B., Knight, G. & Murphy, G. (1996) Biochemical characterization of human collagenase-3, J. Biol. Chem. 271, 1544-1550. 22. Murphy, G. & Willenbrock, F. (1995) Tissue inhibitors of matrix metalloendopeptidases, Methods Enzymol. 248, 496-510. 23. Murphy, G., Houbrechts, A., Cockett, M. I., Williamson, R. A., O’Shea, M. & Docherty, A. J. P. (1991) The N-terminal domain of tissue inhibitor of metalloproteinases retains metalloproteinase inhibitory activity, Biochemistry 30, 8097- 8102. 24. Apte, S. S., Olsen, B. R. & Murphy, G. (1995) The gene structure of TIMP-3 and its activities defines the distinct TIMP gene family, J , B i d . Chern. 270, 14313-14318. 25. Okada, Y., Morodomi, T., Enghild, J. J., Suzuki, K., Yasui, A., Makamishi, I., Salvesen, G. & Nagase, H. (1990) Matrix metalloproteinase 2 from human rheumatoid synovial fibroblasts. Purification and activation of the precursor and enzymatic properties, Eur: J. Biochem. 194, 721 -730. 26. von Bredow, D. C., Nagle, R. B., Bowden, G. T. & Cress, A . E. (1 995) Degradation of fibronectin fibrils by matrilysin and characterization of the degradation products, E.xp. Cell Rex 221, 8391. 27. Woessner, F. & Nagase, H. (1980) An improved assay for proteases and polysaccharidases employing a cartilage proteoglycan substrate entrapped in polyacrylamide particles, Anal. Biochem. 107, 385-392. 28. Fukai, F., Ohtaki, M., Fujii, N., Yajima, H., Ishii, T., Nishizawa, Y., Miyazaki, K. & Katayama, T. (1995) Release of biological activities from quiescent fibronectin by a conformational change and limited proteolysis by matrix metalloproteinases, Biochemistry 34, 11 453 - 1 1 459. 29. Ruoslahti, E. (1989) Proteoglycans in cell regulation, J. Biol. Clzem. 264, 13 369- 13 372. 30. Werb, Z., Tremble, P. M., Behrendtsen, P., Crowly, E. & Damsky. C. H. (1989) Signal transduction through the fibronectin receptor induces collagenase and stromelysin expression, J . Cell B i d . 109. 877 889. 31. Homandberg, G. A,, Meyers, R. & Xie, D. L. (1992) Fibronectin fragments cause chondrolysis of bovine articular cartilage slices in culture, J. B i d . Chem. 267, 3597-3604. 32. Clark, R. A. F. (1990) Fibronectin matrix deposition and fibronectin receptor expression in healing and normal skin, J. Invest. Dermtrtol. 94, 128s-134s. 33. Lohr, K. M., Kurth, C. A,, Xie, D. L., Seyer, J. M. & Honiandberg, G. A. (1990) The arninoterminal 29- and 72-kD fragments of fi-
d’Ortho et al. ( E m J. Biochern. 250)
34. 35. 36.
37.
38.
bronectin mediate selective monocyte recruitment, Blood 76, 2117-2124. Deryugina, E. I. & Bourdon, M. A. (1996) Tenascin mediates human glioma cell migration and modulates cell migration on fibronectin, J. Cell Sci. 109, 643-652. Schwarzbauer, J. E. (1991) Fibronectin: from gene to protein, Cum Opin. Cell Biol. 3, 786-791. Yamamoto, M., Mohaman, S., Sawaya, R., Fuller, G. N., Seiki, M., Sato, H., Kokaslan, Z . L., Liotta, L. A,, Nicolson, G. L. & Rao, J. S. (1996) Differential expression of membrane-type matrix metalloproteinase and its correlation with gelatinase A activation in human malignant brain tumours in vivo and in vitro, Cancer Rex 56, 384-392. Timpl, R. & Brown, J. C. (1996) Supramolecular assembly of basement membranes, Bioessuys 18, 123- 132. Costell, M., Mann, K., Yamada, Y. & Timpl, R. (1997) Characterization of recombinant perlecan domain I and its substitution by glycosaminoglycans and oligosaccharides, Eul: J . Biochern. 243, 115-121.
757
39. Whitelock, J. M., Murdoch, A. D., Iozzo, R. V. & Underwood, P. A. (1996) The degradation of human endothelial cell-derived perlecan and release of bound bFGF by stromelysin, collagenase, plasmin and heparanases, J. Biol. Chern. 271, 10079-10086. 40. Bode, W., Gomis-Ruth, F. X. & Stocker, W. (1993) Astacins, serralysins, sanke venoms and matrix metalloproteinzses exhibit identical zinc-binding environment (HEXXHXXGXXH and Metturn) and topologies and should be grouped into a common family, the ‘metzincins’, FEBS Left. 331, 134-140. 41. Noel, A., Santavicca, M., Stoll, I., L‘Hoir, C., Staub, A,, Murphy, G., Rio, M.-C. & Basset, P. (1995) Identification of structural determinants controlling human and mouse stromelysin-3 proteolytic activities, J. Biol. Chern. 270, 22866-22872. 42. Knauper, V., Smith, B., Stanton, H., Hembry, R. & Murphy, G. (1996) Cellular mechanisms for human procollagenase-3 (MMP13) activation, J. Biol. Chern. 271, 17 124-17 131.
