Matrix metalloproteinase homologues from Arabidopsis thaliana. Expression and activity.

Five genes potentially encoding novel matrix metalloproteinases (MMPs) have been identified on the Arabidopsis thaliana data base. The predicted proteins have a similar domain structure to mammalian MMP-7, with a propeptide and catalytic domain but no C-terminal hemopexin-like domain. Four of the A. thaliana MMPs (At-MMPs) have a predicted C-terminal transmembrane domain. The At-MMPs are differentially expressed in flower, leaf, root, and stem tissues from 14-day-old plants. The cDNA for one of the At-MMPs (At1-MMP) was cloned and expressed in Escherichia coli. Following refolding and purification, the proenzyme At1-MMP was shown to undergo autolytic activation in the presence of an organomercurial with a concomitant decrease in M(r). In contrast to this, trypsin-treatment led to the formation of an inactive product. The activated At1-MMP digested myelin basic protein, but was unable to digest gelatin or casein. Three peptide substrates for MMPs were also cleaved by At1-MMP. The enzyme activity of At1-MMP was inhibited by human tissue inhibitors of metalloproteinases 1 and 2 and the hydroxamate inhibitor BB-94.

The matrix metalloproteinases (MMPs) 1 are a family of at least 20 zinc-dependent endoproteinases in vertebrates, capable of degrading extracellular matrix substrates; they have been divided into subgroups according to their structure and function. The MMPs have a common domain structure with a signal peptide, a propeptide, a catalytic domain, a hinge region, and a C-terminal domain. The propeptide contains an invariant Cys residue that ligates the active site zinc ion to maintain latency; the catalytic domain contains an HEXGHXXGXXH zinc-binding sequence characteristic of the metzincin superfamily of proteinases, followed by an invariant Met that is involved in a structural feature called the "Met-turn." In all family members except matrilysin (MMP-7) a hinge region links to a hemopexin-like C-terminal domain that is thought to be involved in substrate specificity and binding of inhibitors. Individual MMPs contain variations on this theme: MT-MMPs (MMPs 14 -17) have a transmembrane domain and cytoplasmic tail at the C terminus and, in common with MMP-11, contain a potential furin-cleavage site within the propeptide; the gelatinases (MMP-2 and -9) have an insert of three fibronectin type II repeats in the catalytic domain; and MMP-9 has a collagenlike sequence at one end of the catalytic domain (1).
Invertebrates have also been shown to possess proteinases homologous to MMPs; these include the envelysins, which are involved in the hatching process in sea urchins (2). A report has described three sequences in Caenorhabditis elegans that appear to correspond to MMPs (3). In 1991, soybean leaves were shown to contain a metalloproteinase, later shown to be homologous to MMPs. This enzyme was expressed only in adult leaves, and Southern blot analysis demonstrated a single copy gene; activity was demonstrated against a synthetic MMP substrate, and this activity was inhibited by mammalian TIMP-1 (4 -6).
The plant Arabidopsis thaliana is an important model species for the study of plant biology, with many of the tools and reagents in place to manipulate genes in vivo. The relatively small size of its genome has led to a project to sequence its five chromosomes; our study began with the discovery within this project that a sequence on chromosome 4 of the A. thaliana genome codes for a protein with homology to both the soybean MMP and the vertebrate MMP family (7). During our studies, four other sequences with homology to this first one have been identified on the data base; we have named these At1-MMP to At5-MMP in order of our discovery of them on the computer (see below). Here, we examined expression of At1-to At5-MMP in tissues from A. thaliana and then cloned and expressed At1-MMP in Escherichia coli and examined its ability to cleave both protein and synthetic substrates and to be inhibited by known MMP inhibitors.

EXPERIMENTAL PROCEDURES
RT-PCR-Total RNA from 14-day-old A. thaliana plant tissues was a kind gift from Dr. M. Torres (John Innes Centre, Norwich, United Kingdom). cDNA was produced from each of flower, leaf, root, and stem RNA using Superscript II reverse transcriptase (Life Technologies) and oligo(dT) primers (Amersham Pharmacia Biotech). Because the genes for the At-MMPs have no intronic sequence, the absence of genomic DNA contamination in the RNA samples was verified by treatment with RNase-free DNase (Promega) prior to the RT reaction. Primers for RT-PCR of At-MMPs were designed using the Wisconsin Package (GCG), Madison, Wisconsin, to give differing size products for each of At1-to At5-MMP and no cross-hybridization between these cDNAs (Table I). PCR was performed as follows: cycle 1, 94°C for 2 min; cycle 2, 94°C for 1 min, 55°C for 90 s and 72°C for 90 s; cycle 3, 72°C for 10 min. Cycle 2 was repeated between 17 and 35 times, and for all primer pairs, experiments were carried out by sampling across these cycles to verify that signal from the product was within the linear range. PCR using At-MMP primers was performed in 10 mM Tris-HCl, pH 8. 8 amplified by RT-PCR from A. thaliana mRNA (source as above) using the following primers: 5Ј-ACTGGGATCCGCTAGAAACACGCCGGA-G-3Ј (forward); 5Ј-GCAAGAATTCAACCATATAGCTTAAGTACACCT-GCC-3Ј (reverse 1); 5Ј-GCAAGAATTCATAGTTTAGGATTCGGACCA-TATAGC-3Ј (reverse 2); and 5Ј-GCAAGAATTCATCTATGTGATACGG-TGCCG-3Ј (reverse 3). Using these primers and the high fidelity Pfu polymerase (Stratagene), sequence was amplified to give coding regions between Ala-28 and Gly-292 (reverse 1), Leu-297 (reverse 2), or Arg-317 (reverse 3), flanked by a 5Ј BamHI site and a 3Ј EcoRI site. These cDNAs were then subcloned into pRSETA (Invitrogen) using standard techniques, and the reading frame was verified by sequencing. All three plasmids were transformed into E. coli BL21(DE3)pLysS for expression. Pilot scale expression was performed in LB containing 100 g/ml ampicillin and 34 g/ml chloramphenicol at 37°C using 1 mM isopropyl-1-thio-␤-D-galactopyranoside (Melford Laboratories, Ipswich, United Kingdom) to induce expression for 1-24 h. These experiments demonstrated that the Ala-28 to Gly-292 construct gave the highest level of expression, expressing the recombinant At1-MMP as an insoluble protein within inclusion bodies. Hence, this construct was chosen for expression in a 3-liter volume using a 1-h induction period; bacteria were harvested by centrifugation, and the pellets were stored at Ϫ20°C until extraction and purification. A pellet from 1.5 liters of culture was resuspended in 50 ml of ice-cold 10 mM Tris-HCl, pH 8.5, containing 1 mM EDTA and Complete (Roche Molecular Biochemicals) EDTA-free proteinase inhibitors, and lysed by sonication. Insoluble material was pelleted by centrifugation at 10,000 ϫ g at 4°C for 15 min and washed three times in the same buffer with sonication. The final pellet was solubilized in 5 ml of 20 mM Tris-HCl, pH 8.5, containing 6 M urea and 25 l of 2-mercaptoethanol and clarified by centrifugation at 30,000 ϫ g at 4°C for 30 min. An aliquot of this solution was separated by SDS-PAGE against bovine serum albumin standards of known concentration in order to estimate the concentration of At1-MMP. The solubilized At1-MMP was then diluted to 20 -50 g/ml by pumping (0.5 ml/min) into 20 mM Tris-H 2 SO 4 , pH 7.5, containing 5 mM CaSO 4 , 100 mM Na 2 SO 4 , 1 M ZnSO 4 , 10% glycerol, 0.05% Brij 35, and 0.02% NaN 3 (refolding buffer) at room temperature with stirring. The diluted At1-MMP was then incubated at 4°C overnight and the refolded enzyme purified on Ni-NTA (Qiagen), washing the column with refolding buffer without glycerol and ZnSO 4 (wash buffer), and eluting with 100 mM imidazole in wash buffer. Fractions across the eluted peak were analyzed by SDS-PAGE and pooled for further analysis.
Activation of At1-MMP-At1-MMP was "activated" by incubation either alone or with 1 mM APMA at 37°C for up to 4 h. Samples were also treated with trypsin at a final concentration of 1 g/ml for up to 1 h, followed by inhibition of trypsin with at least a 5-fold excess of soybean trypsin inhibitor.
Activity Assays-The activity of At1-MMP against synthetic quenched fluorescent peptide substrates was determined at 37°C using a Perkin-Elmer fluorometer (LS50B). Samples were diluted into 50 mM Tris-HCl, pH 7.5, containing 150 mM NaCl, 10 mM CaCl 2 , 0.05% Brij 35, 0.02% NaN 3 . Substrates were added from a 100ϫ stock in Me 2 SO to a final concentration of 1 M. Measurement of k cat /K m was performed using At1-MMP at approximately 5 nM (determined by active site titration against BB-94) and the substrate across a concentration range of 0.2-2 M. Measurement of apparent K i was determined using At1-MMP at 5 nM and inhibitor concentration across a concentration range of 0.5-30 nM to give greater than 90% inhibition of enzyme activity. Gelatin and casein zymography were performed by copolymerization of 1 mg/ml of either substrate into the separating gel of SDS-PAGE (8); samples were applied without reduction or boiling and separated with 20 mA/gel at 4°C; SDS was removed by shaking in two changes of 50 mM Tris-HCl, pH 7.5, containing 5 mM CaCl 2 and 2.5% (v/v) Triton X-100 for 15 min and then overnight at room temperature. After rinsing in distilled H 2 O, activity was then revealed by incubation overnight at 37°C in 50 mM Tris-HCl, pH 7.5, containing 5 mM CaCl 2 without shaking, followed by staining in Coomassie Brilliant Blue. Solution assays of gelatinase or caseinase activity utilized radiolabeled substrates, essentially as described previously (9). Degradation of myelin basic protein (MBP) was assayed using bovine MBP (Sigma) at a final concentration of 0.25 mg/ml in 100 mM Tris-HCl, pH 7.5, containing 5 mM CaCl 2 , 0.05% Brij 35, and 0.02% NaN 3 ; the enzyme was incubated with MBP for 20 h at 37°C, and products were analyzed by SDS-PAGE on a 10 -20% gradient gel.
N-terminal Amino Acid Sequencing-N-terminal sequence determination of At1-MMP was performed by automated Edman degradation using a PE Biosystems Procise 491 protein sequencer following SDS-PAGE and transfer to polyvinylidene difluoride membrane (Millipore).
Computer Software-PCR primer design was performed using PRIME, GCG (Wisconsin); alignments were made using BESTFIT or PILEUP, GCG (Wisconsin); protein sorting signal analysis was performed using PSORT (University of Osaka, available on the Internet).

TCAAGCTTCAACTCCTTCTTT
(Glycine max, Gm1-MMP) and human MMP-7 (Fig. 1) shows regions of identity and similarity. All enzymes contain a PRCGXXD motif that is characteristic of the "cysteine switch" mechanism of activation, as well as a HEXGHXXGXXH zincbinding sequence followed by a conserved methionine residue in the so-called Met-turn. Interestingly, all of the plant MMP homologues contain an invariant DLESV sequence on the Nterminal side of the zinc-binding region; this sequence is not found in any of the human MMPs, and its function is unknown. Using the PSORT prediction program for protein sorting signals, At1-, At2-, At3-, and At5-MMPs all contain cleavable signal sequences at the N terminus, along with predicted transmembrane domains at the C terminus (Fig. 1); this structure is typical of type Ia plasma membrane proteins, and may alternatively indicate a glycosylphosphatidylinositol anchor linkage to the plasma membrane. At4-MMP contains a noncleavable N-terminal signal sequence typical of an endoplasmic reticulum membrane location. Gm1-MMP contains a cleavable Nterminal signal sequence but no C-terminal transmembrane domain, and this may suggest that further A. thaliana MMPs are still to be found. The results from the BESTFIT alignment program, as shown in Table II, indicate that At2-, At3-, and At5-MMP appear more closely related to each other than At1and At4-MMP, and the converse is also true. None of the At-MMPs show a higher degree of similarity to the soybean enzyme or to human MMP-7 than any other. Alignment with each of the other human MMP amino acid sequences does not reveal any other obvious relationships. Massova et al. (10) have performed a multiple sequence analysis on 64 MMPs that demonstrates that plant MMPs are most closely related to those from other invertebrates (sea urchin and nematode worm).
Expression of At1-to At5-MMP in Plant Tissue-Expression of the At-MMPs in flower (F), leaf (L), root (R), and stem (S) of 14-day-old A. thaliana was investigated using semiquantitative RT-PCR with PCR primers shown in Table I; the UBQ5 gene was used as a loading control. In all cases, samples from PCR were collected across a range of cycle numbers to ensure linearity of response. Fig. 2 shows representative experiments at 30 cycles for At-MMPs and 20 cycles for UBQ5. Although RT-PCR was able to detect transcripts for At1-to At5-MMP in all of the tissues examined, each At-MMP had a distinct pattern of expression, indicating a functionally different role for each. For At1-MMP, the expression level was F Ϸ R Ϸ S Ͼ L; for At2-MMP, R Ͼ Ͼ F Ͼ L Ͼ S; for At3-MMP, L Ϸ R Ͼ Ͼ F Ͼ S; for At4-MMP, F Ϸ S Ͼ L Ϸ R; and for At5-MMP, L Ϸ R Ϸ S Ͼ F. These expression patterns are all different from the Gm1-MMP in soybean, in which expression was only seen in leaves from approximately 10 days after leaf emergence until leaves became senescent, and Gm1-MMP was absent from stem and root tissue (6).
Expression, Purification, and Processing of At1-MMP in E. coli-The coding region of proAt1-MMP was amplified using RT-PCR as detailed under "Experimental Procedures." The N terminus of the proenzyme was chosen on the basis of PSORT identification of the signal peptide to start at Ala-28. The C terminus of the wild-type enzyme contains a putative transmembrane domain; hence, to allow correct refolding of the recombinant protein, three different C terminii were selected: Gly-292 as the limit of the catalytic domain of the vertebrate MMPs; Leu-297 as a short extension past the catalytic domain similar to MMP-7; and Arg-317 as the maximum length sequence without the transmembrane domain. These three cDNAs were subcloned into the expression vector pRSETA and expressed with an N-terminal His tag in E. coli as inclusion bodies. Initial experiments indicated that the construct with Gly-292 as the C terminus gave the highest level of expression, and this protein was purified on Ni-NTA. The final purified recombinant At1-MMP displayed a M r of approximately 39,000 on SDS-PAGE (Fig. 3), with a faint lower band probably representing the loss of the His tag (approximately 3000). Treatment with 1 mM APMA induced a time and temperature dependent drop in M r to approximately 27,000 (Fig. 3), which could be blocked by the addition of 2 mM o-phenanthroline (data not shown). A slower, although identical, pattern of processing was observed by incubation at 37°C with no additions. Treatment with trypsin induced a concentration (of trypsin) and time-dependent processing, via an intermediate to an M r of approximately 36,000 (Fig. 3); no further processing of this product was observed. N-terminal sequencing of the processed forms of At1-MMP demonstrated that trypsin treatment yielded a final product with an N terminus of 14-ATQII, whereas APMA treatment (or simple incubation at 37°C) pro- The predicted amino acid sequences of At1-to At5-MMP were aligned with the Gm1-MMP sequence from soybean and human MMP-7 sequence using PILEUP. The cysteine switch motif PRCGXXD and zinc-binding sequence HEXGHXXGXXH are shaded, whereas the sequence DLESV, conserved in all plant sequences, is shown in boldface. The first amino acid of the predicted mature proenzymes of the plant MMPs is boldface and underlined, whereas the predicted C-terminal transmembrane domains are italicized. * indicates that residues are conserved across all MMPs shown; ϩ indicates that residues are conserved across all plant MMPs shown. duced a major N terminus of 112-INNDF and a minor N terminus of 119-TTAHY (Fig. 4).
Activity of At1-MMP-Purified proAt1-MMP and both the APMA-and trypsin-processed forms were applied to gelatin and casein zymography, but no activity (other than residual activity from the trypsin) was seen; this was confirmed using solution assays for both of these substrates (data not shown). Myelin basic protein was degraded by active At1-MMP, but with a digestion pattern different from that of degradation of MBP by stromelysin-1 (MMP-3) (Fig. 5). Enzyme activity was then assessed against three fluorigenic peptide substrates. The trypsin-processed At1-MMP showed no activity against any of the substrates; the proAt1-MMP exhibited low activity initially compared with APMA-treated enzyme, but activity increased across the time course of the assay at 37°C. Determination of k cat /K m for each substrate was performed using a range of [S] in which initial rate of substrate cleavage shows linear dependence on [S] as described previously (11); Table III shows that At1-MMP cleaves McaPLANvaDpaARNH 2 (12) and McaP ChaGNvaHADpaNH 2 (11) 4-fold more efficiently than Mca-PLGLDpaARNH 2 (13). Although these substrates were designed to be cleaved by mammalian MMPs (11)(12)(13), At1-MMP can cleave them with similar efficiency. The role of At-MMPs in the plant is currently unknown; the plant cell wall contains a number of extracellular matrix macromolecules, including hydroxyproline-rich glycoproteins, arabinogalactan proteins, glycine-rich proteins, and proline-rich proteins, some of which have homology to vertebrate extracellular matrix molecules.   Table I. Samples from the PCR were taken across a range of cycle numbers to ensure linearity of response; data here are from cycle 30 for At1-to At5-MMP primers and cycle 20 for UBQ5 housekeeping gene. Products were separated on a 2% agarose gel in 1ϫ TBE.  Mca-Pro-Leu-Gly-Leu-Dpa-Ala-Arg-NH 2 3,850 Ϯ 150 Mca-Pro-Leu-Ala-Nva-Dpa-Ala-Arg-NH 2 17,550 Ϯ 950 Mca-Pro-Cha-Gly-Nva-His-Ala-Dpa-NH 2 17,500 Ϯ 500 The cell wall (and its remodeling) is crucial to growth and development of the plant, along with its response to environmental stresses and attack by pathogens or insects (14). It is thus tempting to speculate that the At-MMPs are involved in the remodeling of plant extracellular matrix in any or all of these situations. Inhibition of At1-MMP-Using McaPChaGNvaHADpaNH 2 as substrate, the inhibition of At1-MMP by a peptide hydroxamate inhibitor (BB-94) and human TIMP-1 and TIMP-2 was examined. Activity was measured at fixed [S] over a range of inhibitor concentrations to calculate an apparent K i for each inhibitor by fitting the data to the tight binding inhibition equation of Morrison and Walsh (15). Although the data fit the equation well, this methodology only yields approximate values for K i . Human TIMP-1 and TIMP-2 were both able to inhibit At1-MMP with apparent K i in the low nanomolar range, whereas BB-94, a broad spectrum inhibitor of matrix metalloproteinases, was approximately 10-fold more potent. The maintenance of inhibition by TIMPs invites us to question whether there is a protein homologous to TIMPs in A. thaliana, but a search of the data base reveals no such gene sequence; we are currently using PCR with degenerate primers to search for the A. thaliana TIMP. Encouragingly, the first invertebrate member of the TIMP family was recently identified in Drosophila (16).
Conclusions-At least five genes with homology to vertebrate MMPs are expressed in A. thaliana. All show a simple domain structure lacking the C-terminal hemopexin-like domain, but four are likely to be anchored in the plasma membrane. At1-MMP is a functionally active enzyme, able to cleave both protein and peptide substrates and inhibited by mammalian TIMPs. The function of the At-MMPs in vivo is unknown, but we can speculate that they have roles in events in which the plant extracellular matrix is remodeled or broken down. Our future work will explore both biochemistry and function of the At-MMPs in detail.