A Kazal-like Extracellular Serine Protease Inhibitor from Phytophthora infestans Targets the Tomato Pathogenesis-related Protease P69B*

The oomycetes form one of several lineages within the eukaryotes that independently evolved a parasitic lifestyle and consequently are thought to have developed alternative mechanisms of pathogenicity. The oomycete Phytophthora infestans causes late blight, a ravaging disease of potato and tomato. Little is known about processes associated with P. infestans pathogenesis, particularly the suppression of host defense responses. We describe and functionally characterize an extracellular protease inhibitor, EPI1, from P. infestans. EPI1 contains two domains with significant similarity to the Kazal family of serine protease inhibitors. Database searches suggested that Kazal-like proteins are mainly restricted to animals and apicomplexan parasites but appear to be widespread and diverse in the oomycetes. Recombinant EPI1 specifically inhibited subtilisin A among major serine proteases and inhibited and interacted with the pathogenesis-related P69B subtilisin-like serine protease of tomato in intercellular fluids. The epi1 and P69B genes were coordinately expressed and up-regulated during infection of tomato by P. infestans. Inhibition of tomato proteases by EPI1 could form a novel type of defense-counterdefense mechanism between plants and microbial pathogens. In addition, this study points to a common virulence strategy between the oomycete plant pathogen P. infestans and several mammalian parasites, such as the apicomplexan Toxoplasma gondii.

Parasitic and pathogenic lifestyles have evolved repeatedly in eukaryotes (1). Several parasitic eukaryotes represent deep phylogenetic lineages, suggesting that they feature unique molecular processes for infecting their hosts. One such lineage is formed by the oomycetes, a group of fungus-like organisms that are distantly related to fungi but closely related to brown algae and diatoms in the Stramenopiles (1)(2)(3). One of the most notorious and destructive oomycete is the Irish famine pathogen, Phytophthora infestans. This species causes late blight, a reemerging and ravaging disease of potato and tomato (4 -7). During the early stages of infection, P. infestans requires living host cells but later causes extensive necrosis of host tissue, a lifestyle that is known as hemibiotrophy. As with other biotrophic plant pathogens, processes associated with P. infestans pathogenesis are thought to include the suppression of host defense responses (3,8,9). In P. infestans, water-soluble glucans have been reported to suppress host defenses in a plant cultivar-specific manner (10 -12). Nevertheless, the molecular basis of suppression of host defenses by Phytophthora remains poorly understood (3). It is tempting to speculate that unique classes of suppressor genes have been recruited to aid in infection and counteract host defenses during the evolution of pathogenesis in the oomycete lineage.
Parasitic eukaryotes often face inhospitable environments in their hosts. For example, parasites that colonize or transit through the mammalian digestive tract must adapt to the diverse and abundant array of proteases secreted in the gastric juices (13)(14)(15). Some of these parasites secrete inhibitors that target host proteases and may aid in survival and colonization of the host. For instance, the apicomplexan obligate parasite Toxoplasma gondii secretes TgPI-1 and TgPI-2, four-domain serine protease inhibitors of the Kazal family (15)(16)(17)(18)(19), and the intestinal hookworm Ancylostoma ceylanicum secretes an 8-kDa broad spectrum serine protease inhibitor of the Kunitz family (14). In plants, the apoplast (intercellular fluid) forms a protease-rich environment that is colonized by many pathogens, including P. infestans and the fungus Cladosporium fulvum. In tomato, apoplastic proteases are integral components of the plant defense response. Serine proteases of the P69 subtilase family have long been tied to pathogen defense, and two isoforms, P69B and P69C, are known as pathogenesisrelated proteins (PR-7 class) (20 -22). More recently, an apoplastic papain-like cysteine protease, Rcr3, was shown to be required for specific resistance to C. fulvum (23). In addition, several C. fulvum extracellular proteins are processed or degraded by host proteases in the apoplast, resulting in altered functionality (24,25).
Despite the importance of extracellular proteases in plant defense, to date no protease inhibitor has been reported from microbial plant pathogens. In this paper, we describe and functionally characterize an extracellular protease inhibitor, EPI1, from P. infestans. EPI1 contains two domains with significant similarity to the Kazal family of serine protease inhibitors, which also occurs in many animal species and in apicomplexan parasites. In vitro studies indicated that recombinant EPI1 (rEPI1) 1 specifically inhibited subtilisin A among the major serine proteases. rEPI1 was further demonstrated to inhibit and interact with tomato P69B subtilisin-like serine protease. The epi1 and P69B genes were coordinately expressed and up-regulated during infection of tomato by P. infestans. Overall these results suggest that inhibition of tomato proteases by P. infestans EPI1 could form a novel type of defense-counterdefense mechanism between plants and microbial pathogens. In addition, this study points to a common virulence strategy between the oomycete plant pathogen P. infestans and mammalian parasites, such as the apicomplexan T. gondii.
Bacterial Strains and Plasmids-Escherichia coli XL1-Blue was used in this study and was routinely grown at 37°C in LB medium (28). Plasmid pFLAG-EPI1 was constructed by cloning PCR-amplified DNA fragment corresponding to the mature sequence of EPI1 into the Hin-dIII site of pFLAG-ATS (Sigma), a vector that allows secreted expression in E. coli. The oligonucleotides EPI1-F1 (5Ј-GCGAAGCTTCAAA-GCCCGCAAGTCATCAG-3Ј) and EPI1-R1 (5Ј-GCGAAGCTTATCCCT-CCTGCGGTGTC-3Ј) were used to amplify the fragment. The introduced HindIII restriction sites are underlined. The N-terminal sequence of the processed FLAG-rEPI1 protein is DYKDDDDKVKLQS-PQVISPAP. . . . The FLAG epitope sequence is underlined, and the first 10 amino acids of mature EPI1 are shown in bold type.
Plant Growth, BTH Treatment, and Infection by P. infestans-Tomato (Lycopersicon esculentum) cultivar Ohio 7814 was used throughout the study and grown in pots at 25°C, 60% humidity, under 16 h-light/8 h-dark cycle. We used the salicylic acid analog BTH to mimic pathogen infection. For BTH treatment, 10 ml of a 25 g/ml BTH solution was applied to 3-week-old tomato plants by soil drench. Plants treated with 10 ml of water were used as controls. Leaves from BTHtreated and control plants were detached for isolation of intercellular fluids 6 days after treatment. Time courses of P. infestans infection of tomato leaves were performed exactly as described earlier (29). 10-l droplets containing 1,000 zoospores of P. infestans were used to inoculate the underside of detached tomato leaves. Leaf discs of similar sizes were dissected from the inoculated regions while making sure that the inoculation spot is in the center of sampled area. Leaf discs were frozen in liquid nitrogen and stored at Ϫ80°C for later use in RNA extraction. For isolation of intercellular fluids, tomato leaves were sprayed with zoospore suspensions at the concentration mentioned above (10 5 /ml), and the intact leaves were collected at different time points for immediate preparation of intercellular fluids.
Isolation of Intercellular Fluids-Intercellular fluids were prepared using a 0.24 M sorbitol solution according to the method of de Wit and Spikman (30). The intercellular fluids were filter-sterilized (0.22 M) and were used immediately or stored at Ϫ20°C.
Sequence Analyses-GC counting was performed as described elsewhere (31). PexFinder and signal peptide predictions were performed as described by Torto et al. (32). Similarity searches were performed locally on an Intel Linux or a Mac OSX work station or through the internet on the NCGR (www.ncgr.org) and Whitehead Institute web servers (www-genome.wi.mit.edu/resources.html). Search programs included BLAST (33), and the similarity search programs implemented in the BLOCKS (34), pfam (35), SMART (36), and InterPro (37) websites. The examined sequence databases included GenBank TM nonredundant, dBEST, and TraceDB (38), PGC (39), SPC, a proprietary database of Syngenta Inc. containing ϳ75,000 ESTs from P. infestans (courtesy of the Syngenta Phytophthora Consortium, Research Triangle Park, NC), and the genome sequences of the fungal species Aspergillus nidulans, Magnaporthe grisea, Neurospora crassa, and Fusarium graminearum available through the Whitehead Institute Fungal Genome Initiative Databases (www-genome.wi.mit.edu/resources.html). Multiple alignments of the Kazal domains were conducted using the program CLUSTAL-X (40). The P. infestans and Phytophthora brassicae sequences described in this paper were deposited in GenBank TM under accession numbers AY586273-AY586284 and AY589086-AY589087, respectively. Other sequences were obtained from the NCBI nr, dBEST, or Trace Archive data bases (www.ncbi.nlm.nih.gov) ( Table I).

SDS-PAGE and Western Blot
Analyses-Proteins were subjected to 10 -15% SDS-PAGE as previously described (28). Following electrophoresis, the gels were stained with silver nitrate following the method of Merril et al. (44) or stained with Coomassie Brilliant Blue (28), or the proteins were transferred to supported nitrocellulose membranes (Bio-Rad) using a Mini Trans-Blot apparatus (Bio-Rad). Detection of antigen-antibody complexes was carried out with a Western blot alkaline phosphatase kit (Bio-Rad). Antisera to P69 subtilases were produced by immunizing rabbits with the keyhole limpet hemocyanin-conjugated peptide, H2N-TTHTPSFLGLQQNC-amide. The sequence underlined is located at the N terminus of mature P69B and P69D proteins (21) and was chosen for its highly antigenic characteristics and conservation among P69 proteins. Selection of peptides for highly antigenic characteristics, peptide synthesis, and conjugation, as well as antisera production, was performed by Rockland Immunochemicals (Gilbertsville, PA). In Western blot analyses, the antisera to the P69 peptide reacted only with ϳ70-kDa bands from tomato intercellular fluids.
Expression and Purification of rEPI1-Expression of rEPI from pFLAG-EPI1 was conducted as described previously (45). Overnight cultures of E. coli XL1-blue containing pFLAG-EPI1 were diluted (1: 100) in LB medium containing ampicillin (50 g/ml) and incubated at 37°C. When the A 600 of the cultures reached 0.6, isopropyl-␤-D-thiogalactopyranoside was added to a final concentration of 0.4 mM. The cultures were further incubated for 5-6 h before processing. rEPI1 was recovered from the culture supernatant and was purified by immunoaffinity using gravity column packed with anti-FLAG M2 affinity gel (Sigma). The proteins were eluted with 0.1 M glycine, pH 3.5, and immediately equilibrated to neutral pH with 20 l of 1 M Tris, pH 8.0, for each 1-ml eluted fraction. The protein concentrations were determined using the Bio-Rad protein assay. To determine the purity of rEPI1, 0.5 g of the purified protein was run on a SDS-PAGE gel followed by staining with silver nitrate.
Assays of Protease Inhibition-Inhibition assays of commercial serine proteases by rEPI1 were performed by the colorimetric Quanti-Cleave TM Protease Assay Kit (Pierce). 20 pmol of rEPI1 was preincubated with 20 pmol of trypsin (Pierce), chymotrypsin (Sigma), or subtilisin A (Carlsberg) (Sigma), in a volume of 50 l for 30 min at 25°C, followed by incubation with 100 l of succinylated casein (2 mg/ml) in 50 mM Tris buffer, pH 8, containing 20 mM CaCl 2 at room temperature for 20 min. Protease activity was measured as absorbance at 405 nm using a HTS 7000 Bio Assay Reader (PerkinElmer Life Sciences) 20 min after the addition of chromogenic reagent 2,4,6-trinitrobenzene sulfonic acid, which reacts with the primary amine of digested peptide and produces a color reaction that can be quantified by absorbance reader. Detailed kinetic analysis of Subtilisin A inhibition by rEPI1 was performed as follows. 2 pmol of subtilisin A was preincubated with increasing concentrations of rEPI1 in a volume of 50 l for 15 min at 25°C and was followed by the addition of 150 l assay buffer: 50 mM Tris-Cl, pH 8.0, containing 2.5% Me 2 SO, and 500 M subtilisin chromogenic substrate Boc-Gly-Gly-Leu-pNA (Calbiochem, La Jolla, CA). The experiments were performed three times and in triplicate each time. Initial reaction velocities were measured by monitoring the absorbance change at 405 nm over reaction time using the HTS 7000 Bio Assay Reader (PerkinElmer Life Sciences). K i app was determined following the method described by Morris et al. (15). The slope of the linear plot of [V 0 /V i ] Ϫ 1 versus [I] was estimated as 1/ K i app . K i app was converted to K i according to the formula K i ϭ K i app /(1 ϩ [S]/K m ) (46). Varying concentrations of substrate were incubated with 2 pmol of subtilisin A in a total volume of 200 l under the conditions described above, and the initial velocities were measured by monitoring the absorbance at 405 nm. The K m was determined graphically by doublereciprocal Lineweaver-Burk plots of 1/ Inhibition assays of plant proteases by rEPI1 were carried out with the QuantiCleave TM Protease Assay Kit (Pierce) and in-gel protease assays using the Bio-Rad zymogram buffer system. For the first method, 50 l of intercellular fluids were preincubated with or without 10 pmol of rEPI1 at 25°C for 30 min, and the protease activities were subsequently measured. For the in-gel protease assays, 10 pmol of rEPI1 were preincubated with 8 l of intercellular fluids for 30 min at 25°C and then mixed with zymogram sample buffer and loaded on a 10% SDS-polyacrylamide gel without boiling or addition of reducing reagents. Following electrophoresis, the gel was incubated in 1ϫ zymogram renaturation buffer for 30 min. Then the gel was incubated in 1ϫ zymogram development buffer for 4 h at 37°C before staining with 0.5% Coomassie Brilliant Blue.
Coimmunoprecipitation-Coimmunoprecipitation of rEPI1 and tomato intercellular fluid proteins was performed using the FLAG-tagged protein immunoprecipitation kit (Sigma) following the manufacturer's instructions. 100 pmol of purified rEPI1 were preincubated with 200 l of tomato intercellular fluid for 30 min at 25°C. 50 l of anti-FLAG M2 resin was added and incubated at 4°C for 2 h with gentle shaking. The precipitated protein complexes were eluted in 60 l of FLAG peptide solution (150 ng/l) and were analyzed by SDS-PAGE and Western blot analyses.
Tandem Mass Spectrometric Sequencing-Tandem mass spectrometric sequencing was performed at the proteomics facility of The Cleveland Clinic Foundation (Cleveland, OH). The selected protein band was cored from the gel, and protein digestion was carried out as previously described (47). The liquid chromatography-mass spectrometry system used is a Finnigan LCQ-Deca ion trap mass spectrometer system with a Protana microelectrospray ion source interfaced to a self-packed 10 cm ϫ 75 m Phenomenex Jupiter C18 reversed-phase capillary chromatography column. 2-l volumes of the peptide extract were injected, and the peptides were eluted from the column by an acetonitrile, 0.05 M acetic acid gradient at a flow rate of 0.2 l/min. The microelectrospray ion source was operated at 2.5 kV. The digest was analyzed using the data-dependent multitask capability of the instrument resulting in ϳ1000 collision-induced dissociation spectra of ions ranging in abundance over several orders of magnitude. The data were analyzed by using all collision-induced dissociation spectra collected in the experiment to search the NCBI nonredundant data base with the search program TurboSequest. All matching spectra were verified by manual interpretation.

EPI1 Belongs to the Kazal Family of Protease Inhibitors-We
mined an EST data set generated from tomato leaves 3 days after infection with P. infestans using two methods: 1) GC counting to distinguish between Phytophthora and tomato sequences (31) and 2) PexFinder to identify cDNAs encoding extracellular proteins (32). 488 of 2808 ESTs examined showed a GC content higher than 53%. Of these 42 were predicted to encode extracellular proteins using the criteria of Torto et al. (32). These ESTs were then annotated by similarity and motif searches against public databases. One EST, PC064G6 (GC content, 57.4%), showed similarity to proteins of the Kazal serine protease inhibitor family. DNA sequencing of the full cDNA revealed an open reading frame of 450 bp corresponding to a predicted translated product of 149 amino acids (Fig. 1A). SignalP (48) analysis of the predicted protein identified a 16- amino acid signal peptide with a significant mean S value of 0.88 and hidden Markov model score of 0.97. Similarity searches of the predicted protein against the nonredundant database of GenBank TM using the BLASTP program (33) revealed significant matches to Kazal protease inhibitors with the best hit corresponding to the signal crayfish protease inhibitor PAPI-1 (E value ϭ 10 Ϫ10 ). Searches against the Inter-Pro database (37) revealed two domains similar to InterPro IPR002350 for Kazal inhibitors (Fig. 1A). Based on these analyses, we propose that the examined P. infestans cDNA is likely to encode a two-headed Kazal serine protease inhibitor, and we designated the cDNA epi1 (extracellular protease inhibitor 1).

Proteins with Kazal Domains Are Diverse and Ubiquitous in Oomycetes-We used EPI1 and other Kazal domain sequences
to search for Kazal-like motifs in sequence databases from oomycetes and other microbial plant pathogens (see "Material and Methods"). We failed to identify sequences similar to the Kazal domain in all examined fungal and bacterial databases, except for a predicted protein from the ammonia-oxidizing bacterium Nitrosomonas europaea (GenBank TM accession NP_841298) (49). On the other hand, we unraveled a total of 35 different putative proteins with 56 predicted Kazal-like domains (range, 1-4/protein) in five plant pathogenic oomycete species, P. infestans, Phytophthora sojae, Phytophthora ramorum, P. brassicae, and the downy mildew Plasmopara halstedii (Table I). These oomycete Kazal motifs were identified in ESTs from a variety of developmental stages, including host tissue infected with P. sojae and P. halstedii. We identified a putative full open reading frame sequence for 27 of the identified genes and the putative start codon for 33 of the genes. All of these 33 genes were predicted to have signal peptides based on SignalP (48,50).
We used Clustal X (40) to generate a multiple alignment of representative oomycete Kazal domains with domains from signal crayfish PAPI-1 (the best hit in BLASTP searches against GenBank TM nonredundant database) and T. gondii TgPI1 (Fig. 1B). Amino acid residues defining the Kazal family signature, including the cysteine backbone, tyrosine, and asparagine residues, were highly conserved. The oomycete domain structure was usually C-X 3,4 -C-X 7 -C-X 6 -Y-X 3 -C-X 6 -C-X 9,12,13,14 -C. The first EPI1 domain was atypical and lacked Cys 3 and Cys 6 but retained the other four cysteines (Fig. 1A). The predicted active site P1, which is central to the specificity of Kazal inhibitors (51,52), was variable with 10 different amino acids represented (Ala, Asp, Glu, His, Lys, Met, Asn, Arg, Ser, and Thr). Remarkably, half (28 of 56) the P1 residues, including those of EPI1, were aspartate (Asp), an uncommon P1 amino acid in other natural Kazal inhibitors. These results suggest that genes encoding proteins with Kazal domains are diverse and ubiquitous in plant pathogenic oomycetes.
EPI1 Inhibits the Serine Protease Subtilisin A-To determine whether EPI1 functions as a serine protease inhibitor as predicted by bioinformatic analyses, we expressed in E. coli and affinity-purified rEPI1 as a fusion protein with the FLAG epitope tag at the N terminus. Silver staining of the purified rEPI1 fraction after SDS-PAGE revealed a single band indicating high purity. Chymotrypsin, trypsin, and subtilisin A, representing three major classes of serine proteases, were selected for inhibition assays with the purified rEPI1. Protease activity was measured with or without EPI1. In repeated assays, rEPI1 was found to inhibit about 90% of the measured activity of subtilisin A but did not cause apparent inhibition of the other two proteases (Fig. 2A). Time courses of chromogenic substrate hydrolysis by subtilisin A in the presence of increasing amounts of rEPI1 were performed and indicated that rEPI1 inhibition followed a typical dose-response pattern (Fig. 2B). The inhibitory constant (K i ) for subtilisin A inhibition by rEPI1 was determined at 2.77 Ϯ 1.07 nM. These results suggest that epi1 encodes a functional protease inhibitor that specifically targets the subtilisin class of serine proteases.
EPI1 Inhibits BTH-induced Apoplastic Proteases from Tomato-In tomato, some members of the subtilisin-like family P69, namely P69B and P69C, are known to be induced by pathogens and stress treatments and are classified as PR proteins (PR-7 class) (20 -22). To test whether rEPI1 inhibits PR-like proteases in tomato, the salicylic acid analog BTH was applied to tomato plants to induce defense-related proteases. In-gel protease assays of tomato leaf intercellular fluid from both H 2 Otreated and BTH-treated plants revealed that, as expected, BTH induced the production of abundant extracellular proteases in tomato that migrated as two separate but close bands (Fig. 3A). Inhibition assays revealed that rEPI1 dramatically inhibited these BTH-induced proteases as well as partially inhibited a constitutive protease. The total endoprotease activity of tomato intercellular fluids was also measured in the absence or presence of rEPI1. Significant inhibition of endoprotease activity was observed and corresponded to 28 and 27% of total activity in control and BTH-treated tomato, respectively (Fig. 4).
EPI1 Interacts with Pathogenesis-related Subtilases of the Tomato P69 Subfamily-To identify the plant proteases targeted by rEPI1, coimmunoprecipitation was performed on tomato intercellular fluid incubated with rEPI1 using FLAG antibody covalently linked agarose beads. In addition to rEPI1, two proteins were pulled down with the FLAG antibody only in the presence of rEPI1 (Fig. 5A). These two proteins exhibited a similar molecular mass of ϳ70 kDa (Fig. 5A) and were more abundant in BTH-induced intercellular fluid (Fig. 5B). These results prompted us to test whether these proteins could be tomato P69 subtilisin-like proteases. Western blot analyses with antisera raised against a peptide specific to P69 subtilisinlike proteases strongly interacted with both bands, suggesting that rEPI1 interacts with P69 subtilases of tomato (Fig. 5B). To confirm the results obtained with the Western blot and further identify which P69 isoforms are the main targets of rEPI1, the two closely migrated protein bands (Fig. 5) were cored from the FIG. 2. rEPI1 inhibits subtilisin A. A, protease activity of chymotrypsin, subtilisin A, and trypsin in the absence (gray columns) or presence of rEPI1 (black columns). The activities were determined using the QuantiCleave TM protease assay kit as described under "Materials and Methods." Activity is expressed as a percentage of total protease activity in the absence of protease inhibitors. The bars correspond to the means of three independent replications of one representative experiment of three performed. The error bars represent the standard errors calculated from the three replications. B, time course of substrate hydrolysis by subtilisin A in the presence of varying concentrations of rEPI1. Protease activity was measured as absorbance at 405 nm based on hydrolysis of a chromogenic substrate. The final concentration of subtilisin A is 10 nM. The concentrations of rEPI1 are indicated next to the curves.

FIG. 3. rEPI1 inhibits BTH-induced tomato proteases.
A, intercellular fluids obtained from water-treated (Ϫ) and BTH-treated (ϩ) tomato plants were run on SDS-PAGE gel followed by staining with Coomassie Brilliant Blue (left panel) or were used in zymogen in-gel protease assays (right panel). The asterisks represent known pathogenesis-related proteins PR1, PR3, and PR2 (from bottom to top) and confirm the induction of defense responses by BTH. The arrows indicate BTH-induced protease activities that migrated as two close bands. B, inhibition of tomato proteases by rEPI1. Intercellular fluids from BTHtreated tomato leaves were incubated in the absence (Ϫ rEPI1) or presence of rEPI1 (ϩ rEPI1) and then analyzed using zymogen in-gel protease assays. The arrows indicate the BTH-induced protease bands.
FIG. 4. rEPI1 inhibition of total protease activity from tomato intercellular fluids. Total protease activity of Intercellular fluids obtained from water-treated (ϪBTH) and BTH-treated (ϩBTH) tomato plants was measured in the absence (gray columns) or presence (black columns) of rEPI1 using the QuantiCleave TM protease assay kit as described under "Materials and Methods." Activity is expressed as absorbance at 405 nm. The bars correspond to the means of three independent replications of one representative experiment of three performed. The error bars represent the standard errors calculated from the three replications.
FIG. 5. Coimmunoprecipitation of rEPIs and P69 subtilases using FLAG antisera. A, eluates from coimmunoprecipitation of rEPI1 with proteins in tomato intercellular fluids were run on SDS-PAGE gel followed by staining with silver nitrate. The numbers on the left indicate the molecular masses of the marker proteins in kDa. rEPI1 indicates whether or not rEPI1 was added to the reaction mix. BTH indicates whether or not the intercellular fluids were obtained from plants treated with BTH. The lower molecular mass band corresponds to rEPI1, and the high molecular mass bands correspond to the rEPI1interacting protein(s). B, the same eluate samples were run on SDS-PAGE gel followed by staining with Coomassie Brilliant Blue (Coom.) or immunoblotting with antisera raised against a peptide specific for the tomato P69 family (␣-P69), and FLAG (␣-FLAG), respectively. The top two panels (Coom. and ␣-P69) correspond to the high molecular mass bands of A, whereas the bottom panel (␣-FLAG) corresponds to the low molecular mass band.
Coomassie Blue-stained SDS-PAGE gel as one sample and analyzed by tandem mass spectrometry. A total of 21 trypsindigested peptides were sequenced and perfectly matched the subtilisin-like protease P69B (GenBank TM accession number T07184 or CAA76725) (Fig. 6). Of these 21 peptides, 13 peptides were specific to P69B and did not match any of the other five known P69 isoforms. At this stage it cannot be ruled out that the two closely migrated protein bands contain other isoforms in minor amounts, but the results from the tandem mass spectrometry clearly showed that P69B is the main target of rEPI1.
The epi1 and P69B Gene Are Concurrently Expressed during Infection of Tomato by P. infestans-Expression pattern of both epi1 and P69 genes during infection of tomato by P. infestans was studied by Northern blot and RT-PCR analyses. The epi1 gene displayed the highest mRNA levels 3 days post-inoculation and was moderately up-regulated (approximately 2ϫ based on PhosphorImager quantification) compared with in vitro grown mycelium and relative to the constitutive actA gene (Fig. 7A). Semi-quantitative RT-PCR analyses confirmed these results (Fig. 7B). The expression of P69 protease genes was induced after inoculation with P. infestans and attained the highest level 2 and 3 days after inoculation (Fig. 7A). Semiquantitative RT-PCR amplifications using primers specific for P69A, P69B, and P69D, indicated that the pathogenesis-related P69B gene is the only gene that is up-regulated during interaction with P. infestans (Fig. 7B). We could not assess the expression of P69C, the other pathogenesis-related gene of the P69 family (21), because we repeatedly failed to amplify P69C from tomato cultivar Ohio 7814 based on published sequences. Increase in P69 protein during infection of tomato by P. infestans was also noted by Western blot analyses with P69 antisera of intercellular fluids obtained from a time course infection (Fig. 7C). Altogether, these results suggest that epi1 and P69 genes are concurrently expressed during infection and support the possibility of direct interaction between P. infestans EPI1 and plant P69 proteases, particularly P69B, at the infection interface.

DISCUSSION
Plant pathogens manipulate biochemical and physiological processes in their host plants through a diverse array of viru-lence or avirulence molecules, known as effectors. In susceptible plants, biotrophic plant pathogens produce effectors that promote infection by suppressing defense responses. Here, we describe EPI1, a two-domain extracellular protease inhibitor from P. infestans that inhibits apoplastic subtilases of tomato, namely the PR proteins P69. Based on its biological activity and expression pattern, EPI1 may function as a disease effector molecule and may play an important role in P. infestans colonization of host apoplast.
Suppression of host defenses is thought to play a critical role in plant-microbe interactions, especially those involving biotrophic pathogens that require live plant cells to establish a successful infection (8,53). Nonetheless, only a few pathogen molecules that suppress host defenses have been identified. Examples include tomatinase, a saponin-detoxifying enzyme from the fungal pathogen Septoria lycopersici that was recently shown to indirectly suppress host defense responses through its degradation products (9). P. sojae secretes glucanase inhibitor proteins that inhibit a soybean endo-␤-1,3-glucanase and are thought to function as counterdefensive molecules that inhibit the degradation of ␤-1,3/1,6-glucans in the pathogen cell wall and/or the release of defense-eliciting oligosaccharides by host endo-␤-1,3 glucanases (54). P. infestans and other Phytophthora species produce water-soluble glucans that suppress induction of host defense responses (10 -12). Here, we describe a novel class of pathogen suppressors of plant defense response, FIG. 6. Tandem mass spectrometry identifies P69B as the main target of rEPI1. The amino acid sequence of P69B subtilisin-like protease precursor is shown with the signal peptide sequence in italics and the propeptide domain sequence in gray. The 21 peptides sequenced by tandem mass spectrometry are shown in bold type. Each sequenced peptide ends with an Arg or a Lys residue, which is highlighted in bold italics. Underlined sequences are specific to P69B among the known P69 isoforms.
FIG. 7. The epi1 and P69 genes are concurrently expressed during colonization of tomato by P. infestans. A, time course of expression of P. infestans epi1 and actA and tomato P69B and tubulin during colonization of tomato by P. infestans. Total RNA isolated from infected leaves of tomato, 0, 1, 2, 3, or 4 days after inoculation, from noninfected leaves (To), and from P. infestans mycelium grown in synthetic medium (My) was hybridized with probes from the four genes. The approximate sizes of the transcripts are ϳ600 nucleotides for epi1, 1600 nucleotides for actA, and 2500 nucleotides for P69B and tubulin. B, RT-PCR analysis of epi1, P69A, P69B, and P69D expression during colonization by P. infestans. Total RNA from a time course similar to the one described in A was used in RT-PCR amplifications as described in the text. Amplification of P. infestans elongation factor 2␣ (Pief2␣) and tomato elongation factor 1␣ (Toef1␣) were used as controls to determine the relative expression of epi1 and P69 genes, respectively. C, Western blot analyses of tomato P69 subtilases during colonization by P. infestans. The time course is as described for A. Equal volumes of intercellular fluids were obtained from infected tomato leaves, subjected to SDS-PAGE, and immunoblotted with P69 antisera (␣-P69).
namely extracellular protease inhibitors that directly interact with and inhibit host proteases. This interaction could form another type of defense-counterdefense mechanism between plants and microbial pathogens We scanned GenBank TM and several other sequence data bases for the occurrence of Kazal-like domains. The examined data sets included the full genome sequence of several plant pathogenic bacteria and fungi. A 235-amino acid protein from the ammonia-oxidizing bacterium N. europaea (GenBank TM accession NP_841298) was the only bacterial or fungal protein with significant similarity to the Kazal motif. In sharp contrast, 56 Kazal-like motifs were detected in 35 predicted proteins of five plant pathogenic oomycete species, P. infestans, P. sojae, P. ramorum, P. brassicae, and P. halstedii (Table I). Interestingly, oomycete Kazal motif genes are often expressed during host colonization. Five of the identified sequences were from cDNAs obtained from infected plant tissue corresponding to diverse oomycete pathosystems: P. infestans tomato/potato, P. sojae soybean, and P. halstedii sunflower. Taken together, the common occurrence of Kazal motifs in several plant pathogenic oomycetes, their in planta expression, and the functional analyses of EPI1 suggest that inhibition of host proteases could be a conserved virulence strategy among oomycete pathogens. It remains unclear whether other plant pathogenic microbes have evolved inhibitors to counteract plant proteases. If so these inhibitors apparently belong to structural classes other than the Kazal inhibitor domain.
Several plant proteases have been linked to plant defense responses. In tomato, P69 and Rcr3 are two extracellular proteases that have been implicated in the defense response (20,21,23). The precise mode of action of these proteases remains unclear. They could degrade secreted proteins from the pathogen, thereby directly contributing to defense. Alternatively, plant proteases could contribute to defense signaling by processing endogenous or pathogen proteins to generate bioactive peptides. Future experiments will focus on determining whether EPI1 contributes to virulence by protecting other secreted proteins of P. infestans from proteolytic degradation in the host apoplast or by perturbing defense signaling in host plants.
P1 is the primary specificity-determining residue of Kazal inhibitors (51,52). Remarkably, half (28 of 56) the predicted P1 residues of oomycete Kazal-like inhibitors, including two-thirds (14 of 21) of the P. infestans inhibitor domains, are aspartate. This is an uncommon P1 amino acid in natural Kazal inhibitors of animals and apicomplexans. This striking feature is remarkable in light of a recent finding that two oat proteases with caspase activity and specificity are subtilisin-like serine proteases that are involved in pathogen-induced programmed cell death (55). Coffeen and Wolpert (55) coined these enzymes saspases because their active-site residue is a serine and they require an asparate residue in the P1 position of the substrate. Saspases could be the enigmatic functional analogs of animal caspases that have been tied to multiple cases of pathogeninduced programmed cell death (56). Phytophthora EPIs that carry aspartate as the P1 residue might therefore target plant saspases and suppress host cell death. This engaging hypothesis will warrant a close examination.
Proteins with Kazal inhibitor domains have a restricted taxonomic distribution as determined by our exhaustive search of sequence and protein motif data bases. In addition to oomycetes, they are mainly found in animal species and in apicomplexans such as T. gondii, a parasite that transits through the mammalian digestive tract (15)(16)(17)(18)(19). An interesting analogy can be made between plant apoplasts and mammalian digestive tracts. Both environments are rich in proteases but neverthe-less are colonized by a variety of microbial pathogens. Apparently, P. infestans and T. gondii, even though phylogenetically unrelated, have independently recruited secreted proteins of the Kazal family to inhibit host proteases and adapt to protease-rich host environments. Interestingly, unlike the T. gondii inhibitors (15,19), EPI1 does not inhibit trypsin and chymortrypsin, suggesting that coevolution between the inhibitors and their target proteases may have shaped the inhibitor specificity. Future structural and functional characterization of Kazal protease inhibitors from animal and plant pathogens will shed some light on interesting questions on the evolution of pathogenesis in eukaryotic microbes and the coevolution of pathogen effectors with host targets.