A Novel Lipoxygenase from Rice PRIMARY STRUCTURE AND SPECIFIC EXPRESSION UPON INCOMPATIBLE INFECTION WITH RICE BLAST FUNGUS*

A novel lipoxygenase cDNA (3,007 base pairs) was isolated from rice leaves (Orysa sativa cv. Aichiasahi) which had been infected with an incompatible race of the rice blast fungus, Mugnaporthe grisea. A single copy of the gene is present in the rice genome and encodes a protein of 923 residues with a molecular weight of 102,714. This gene product shares the least amino acid sequence homology among plant lipoxygenases identified to date. A novel feature of this gene product is a putative transit peptide sequence at the amino terminus, suggesting the enzyme is localized in chloroplasts. An active lipoxygenase was expressed from the cDNA in Escherichia coli and characterized. The lipoxygenase in- troduces molecular oxygen exclusively into the C-13 position of linoleic and linolenic acids. The gene is expressed at high levels 15 h after inoculation with an incompatible race of M. grisea, at a low level after inoculation with a compatible race of the pathogen, and is not expressed in mock-infected leaves. Gene expression begins at the same time that the pathogen begins to pen- etrate into leaf tissue. This novel lipoxygenase gene expression is a

lipoxygenase L-2, which is expressed in maturing seeds and seedlings after germination, was reported (13).
Involvement of lipoxygenases in plant defense mechanisms against microbial invasion has been suggested. The production of toxic compounds, phytoalexins, is a typical plant defensive response to microbial invasion. Oxygenated unsaturated fatty acids constitute one such class of phytoalexins; these compounds have been isolated from rice leaves and are effective against the rice blast fungus, Magnaporthe grisea (14, 15). Ohta et al. (16) demonstrated in vitro production of rice phytoalexins using linoleate hydroperoxide, which is generated by a rice lipoxygenase and an enzyme fraction isolated from rice seeds. The same researchers also found increased lipoxygenase activity in rice leaves inoculated with an incompatible race of the blast fungus and an increase in lipid hydroperoxide decomposing activity, which catalyzes the production of antifungal compounds and hydroxylated unsaturated fatty acids (17). Others have suggested that lipoxygenase may be involved in the mechanism of resistance of oats to Puccinia coronata avena (18) and tomato to Cladosporium fulvum (19). These results indicate that the biosynthesis of oxygenated unsaturated fatty acids via the lipoxygenase pathway may be important in plant defense strategies against pathogenic attack. Studies on the role of plant lipoxygenase genes in plant defense have recently begun (20,21).
Here we report isolation of a novel lipoxygenase cDNA from rice leaves which encodes a lipoxygenase carrying a putative transit peptide sequence, and we propose a second class of plant lipoxygenase genes based upon this putative transit peptide. We also show that this gene is expressed during early stages of an incompatible plant-pathogen interaction.
EXPERIMENTAL PROCEDURES Plant and Pathogen Growth and Inoculation-A rice cultivar, Aichiasahi, was used as the host plant. Rice seeds were immersed in running tap water for 2 days and then disinfected by immersing the seeds in a fungicide, 0.2% Benlate (DuPont), for 2 h. The seeds were germinated at 37 "C overnight. Seedlings were sown in a 1:2 mixture of commercial soil, Kumiai-Baido (Nokyo, Japan), and vermiculite. The plants were grown in a phytotron at 25 "C with 70% humidity under natural light and watered every other day.
The leaves of Aichiasahi were inoculated with two races of rice blast fungus, Magnaporthegrisea, 007 and 131, as previously described (22). Briefly, conidia produced synchronously on an agar medium were suspended in a sterile solution of 0.02% 'heen 20. The concentration of conidia was adjusted to 5 x 106/ml for inoculation. The fifth leaves of young rice plants were inoculated with the blast fungus by spraying them with the conidial suspension. The inoculated plants were kept in a plastic box (44.5 cm x 38.5 cm x 43.5 cm) at 25 "C with 100% humidity for 36 h. Successful inoculation was checked by a method described previously (23 , and the resulting plasmid was designated pRLL1. The cDNA fragment of pRLLl (isolated by EcoRI digestion) was used as a probe to isolate a longer clone, ARLLZ from library A. The cDNA fragment of ARLLZ (2,857 bp) was digested with EcoRI and subcloned into pUC118 (Takara Shuzo Co., Ltd.) to obtain a plasmid designated pRLL2. That plasmid was digested with PVuI and a fragment (280 bp) located at the 5' end of the cDNA was isolated. The fragment was used to isolate clones, ARLL3 and A R L L 4 , from library B. These clones were subcloned into an EcoRI site of Bluescript pSK+ to obtain plasmids designated pRLL3 and pRLL4, respectively. DNA Sequencing-Sequencing of the cDNA fragments was carried out by the Sanger procedure (25). Nested deletion constructs of the cDNAs were made using a kit (Erase-a-Base, Promega). Singlestranded DNA was prepared from the deleted constructs using a helper phage, R408 (Stratagene). DNA sequence data were analyzed with the aid of the Genetyx program (Software Development Co. Ltd., Tokyo).
Southern AnaZysis-Genomic DNA was prepared from leaves of the rice cultivar, Aichiasahi, as described elsewhere (26). m e r restriction enzyme digestion, 10 pg of the digested DNA was separated on a 1% (Poll) by capillary transfer using 0.5 M NaOW1.5 M NaCl. A cDNA agarose gel and then transferred to a nylon membrane, Biodyne P/N fragment (nucleotide 2,444 to the 3' end of the cDNA) was isolated by digesting pRLL2 with HincII and EcoRI and then labeled with [32P]dCTP. Hybridization with the labeled probe was carried out in 0.1 x SSC at 65 "C.
Autoradiographic Analysis-We used a computerized analyzer for autoradiography of membranes hybridized with 32P-labeled probes; the membranes were exposed to an imaging plate and the location of the HPLC, high pressure liquid chromatography, LHR, lipoxygenase ho-The abbreviations used are: kb, kilobase paids); bp, base paids); mologous region. labeled compounds was visualized using Bio-Image analyzer BAS2000 (Fuji Film Co. Ltd., Tokyo).
Expression of cDNA in E. coli--To remove the putative transit peptide coding region from pRLL2, a plasmid, pRLLW485, was used to prepare an expression construct. The plasmid was one of the plasmids prepared for sequencing of RLL2 cDNA by deleting pRLL2 from the 5' end using exonuclease I11 then religating the ends. pRLL2/485 was digested with HindIII and EcoRI, and then the resulting fragment (containing the downstream sequence of the cDNA beginning with nucleotide 485) was isolated. The fragment was filled in using Klenow enzyme, resulting in additional nucleotides, AG, at the 5' end derived from the HindIII site. The fragment was introduced into an expression vector, pET3b (27) at a filled in BarnHI site. The plasmid DNAcarrying the rice gene in the appropriate orientation was designated RLIJpET3b and purified from E. coli HB101. The protein encoded by this construct has two regions: 801 amino acids encoded by the rice lipoxygenase cDNA fragment and 15 amino acids at the amino terminus which derived from the expression vector as a result of the manipulation. RLIJ pET3b was used to transformE. coli BL21(DE3), which carries the gene for T7 RNA polymerase under control of the lac UV5 promoter (27).
An active lipoxygenase was expressed from RLUpET3b by the low temperature cultivation method which was used to express rice lipoxygenase L-2 in E. coli (28). After cultivation at 15 "C for 16 h in the presence of 0.4 IIIM isopropyl-/?-o-thiogalactopyranoside, E. coli cells carrying RLUpET3b were collected and suspended in 5 ml of 50 m~ sodium phosphate (pH 6.8) and then disrupted by sonication. The particulate material was removed by centrifugation (5,000 xg, 10 mid, and the supernatant was used as the E. coli homogenate (1 mg of proteid ml). A control E. coli homogenate was prepared using E. coli carrying the vector, pETBb, in the same manner.
The activity of lipoxygenase was determined as described elsewhere (29) using a Clark oxygen electrode. The reaction mixture (3 ml) contained E. coli homogenate (0.1 mg of protein), 1.0 n m linoleic or linolenic acid, and 0.25% (v/v) Tween 20 in 0.1 M sodium acetate (pH 5.5). The activity (1 katal) is defined as the quantity of enzyme which catalyzes the consumption of 1 mol Ods at 30 "C. Protein was determined by the Folin method (30).
To determine the regiospecificity of the enzyme expressed in E. coli, a fraction containing 0.6 nanokatal of lipoxygenase activity was added to 1 ml of 0.7 m~ linoleic or linolenic acid and 0.08% (w/v) Tween 20 in sodium acetate (pH 5.5) and incubated at 20 "C for 1 h with vigorous mixing. The reaction products were analyzed by reverse-phase HPLC as previously described (31). The retention times of the reaction products were compared with those of authentic standards.
To purify the lipoxygenase protein expressed in E. coli from RLIJ pET3b, we took advantage of the accumulation of the protein in an insoluble form in the cells at a high level when E. coli was cultivated at 37 "C. E. coli was cultivated at 37 "C in the presence of 0.4 M isopropylp-D-thiogalactopyranoside in 50 ml of the medium for 3 h. The expressed protein in the cells was isolated using lysozyme and detergent as described elsewhere (32). The isolated protein (2 mg) was dissolved in l% SDS and then used to immunize a rabbit for preparation of the lipoxygenase antibodies.
Western Blotting-Rice leaves (0.4 g) were ground with a mortar and pestle in liquid nitrogen. The resulting powder was homogenized with 2 ml of 125 m~ Tris-C1, 20% glycerol, and 2% SDS (pH 7.0). The homogenate was centrifuged for 5 min at 15,000 x g. 2-Mercaptoethanol was added to the supernatant to 5% and then heated in boiling water for 5 min. The denatured proteins (15 pl) were separated on a 10% polyacrylamide gel containing 1% SDS.
The separated proteins were blotted on a nitrocellulose membrane using a semidry blotting apparatus (Biometra). The protein band was visualized using the lipoxygenase antibodies and an immunodetection kit with an anti-rabbit IgG conjugated to alkaline phosphatase (Promega) according to the supplier's protocol. All membranes shown in Fig.  5 were subjected to calorimetric detection for the same time.  Poly(A) RNA was isolated from the leaves and used to prepare two cDNA libraries (library A and library B) which differed in the type of linkers used to prepare them. We attempted to screen the cDNA library (library A) using DNA fragments of either soybean lipoxygenase L-1 (7) or rice lipoxygenase L-2 (13) as probes under various hybridization conditions. However, no positive clones were obtained, indicating a low level of nucleotide sequence homology between the rice leaf lipoxygenase cDNA and the previously identified genes.

Isolation
We utilized a peptide sequence, HAAVNFGQY, conserved in seven plant lipoxygenases (7-13) and animal 5-lipoxygenases (33,34) to prepare a mixed oligonucleotide probe. Using the nucleotide probe, a cDNAclone (1.4 kb), designated ARLL1, was isolated from library A. Using the cDNA insert of ARLLl as a probe, ARLL2 (2.8 kb) clone was isolated from library A. Two additional clones, ARLL3 and ARLIA, were isolated using a 5' end fragment of the ARLL2 cDNA from library B (Fig. 1). The nucleotide sequence of the cDNA (3,007 bp) (Fig. 2) contains a long open reading frame beginning with the first ATG codon (nucleotides 119-2,8871, and one stop codon (nucleotide 38-40) occurs upstream of the ATG codon in the same frame. The nucleotide sequence at the first ATG codon, ACCATG, is the consensus translation initiation sequence in eucaryotes as shown by Kozak (35). The molecular weight of the gene product predicted from this open reading frame is 102,714 based on 923 amino acids. A second ATG codon (nucleotide 350-352) is present in the long open reading frame. For reasons mentioned below, it is unlikely that this codon is the initiation codon, and we assume the first ATG to be the initiation codon.
The nucleotide sequence of the cDNA shows a strong bias toward G and C in the selection of nucleotides in the open reading frame (66% G/C), especially at the third codon position (91% G/C), while the 5'and 3"noncoding regions (nucleotides 1-118 and 2888-2999) contain 51 and 31% G/C, respectively. A similarly strong G/C bias was observed in the coding region of rice lipoxygenase L-2 cDNA, but not in the noncoding regions (13). The strong G/C bias (65% G/C) in the nucleotide sequence between the first ATG and the second ATG (nucleotide 118-349) of the long open reading frame suggests that the first ATG codon is the initiation codon.
Protein Structure of the Novel Lipoxygenase-Several striking features of the lipoxygenase gene product were found by comparing the deduced amino acid sequence with those of previously identified lipoxygenases from rice (13), soybean (7-101, and pea (11,12). The gene product contains two regions: (i) a sequence with homology to known lipoxygenases (a lipoxygenase homologous region, LHR), but with the lowest identity so far seen among plant lipoxygenases; and (ii) an amino terminal sequence which is limited in composition to several amino acids. It is unlikely that the amino terminus coding region of the cDNA clone is an artifact of the cloning process since the cDNA clones ARLL2, ARLL3 and ARLL4 all contain regions which overlap one another (Fig. 1). In addition to this, ARLL3 and ARLL4 were isolated from a different cDNA library than the  one from which ARLL2 was isolated. Finally, Northern blot analysis using DNA fragments from the amino terminus coding region and from the LHR showed that both probes hybridized to the same band on a Northern blot (data not shown).

ATCGATGGCCGGAACAAGGATAGAAAGCTCAAGRRCAGGTGCGGCGCCGGCATCCTGCCG I D G R N K D R K L K N R C G A G I L P 900 2819 TACCAGCTGATGAAGCCCTTCTCCGACTCCGGCGTCACCGGCATGGGCATCCCCAACAGC Y Q L M K P F S D S G V T G M G I P N S
The gene product shares approximately 43% amino acid sequence identity in the LHR with plant lipoxygenases, including rice lipoxygenase L-2 ( Table I). The LHR is located between Gly'll and Argil2, residues which are completely conserved among the plant lipoxygenases and the carboxyl terminus. Despite the fact that the homology of this protein with other plant lipoxygenases is low, the conserved amino acids found among plant and mammalian lipoxygenases, which are thought to be
involved in lipoxygenase function (1,3), are present (Fig. 2). The carboxyl-terminal sequence, GIPNS(T)SI, is also conserved among plant lipoxygenases. The hydropathy profile of the LHR (Gly"' to Ile923) demonstrates conserved secondary structure between this protein and rice lipoxygenase L-2 (data not shown). Taken together, these results suggest that the gene product is a plant lipoxygenase. The amino-terminal amino acid sequence (Met' to Ilello) found upstream of the LHR consists of hydroxyl amino acids, Ser and "hr (33%), and small hydrophobic residues, Val and Ala (24%), positively charged amino acids, Arg, Lys, and His (13%), negatively charged amino acids, Glu and Asp (5%), and other amino acids (25%). Clusters of these hydroxyl amino acids and dispersion of hydrophobic amino acids are seen throughout the sequence. These features of the amino acid sequence are consistent with the structure of transit peptides for translocation of proteins into chloroplasts (36). It is unlikely that the protein is targeted into mitochondria because the amphiphilic structure prerequisite for such targeting (36) was not found in the sequence.
Characterization of the Leaf Lipoxygenase--To determine whether the gene product has lipoxygenase activity, we expressed the gene in E. coli from a fragment of the cDNA (nucleotides 4854,007) which encodes 99.8% of the LHR and the 3"noncoding region but not the putative transit peptide sequence (Fig. 3). We took advantage of the low temperature cultivation method previously described for expression of rice lipoxygenase L-2 in E. coli (28). Cultivation of E. coli carrying the cDNA fragment at 15 "C produced an active lipoxygenase. The E. coli homogenate prepared contained lipoxygenase activity which was most active from pH 5 to 6. Activities at pH 5.5 in the homogenate were 2 nanokataldmg of protein toward linoleic acid and 6 nanokataldmg of protein toward linolenic acid. The observed activity was heat labile, and no activity was observed using a control homogenate prepared from E. coli cells carrying the vector without the cDNA fragment.
The regiospecificity of hydroperoxidation on linolenic and linoleic acids by the lipoxygenase expressed in E. coli was analyzed using reverse-phase HPJX (Fig. 4). The lipoxygenase introduced molecular oxygen exclusively into the C-13 position of these fatty acids. The ratio of 9-hydroperoxy to 13-hydroperoxy fatty acid reaction products was 2% for linolenic acid and 12% for linoleic acid. It is noteworthy that formation of 13hydroperoxy linolenic acid is a prerequisite for biosynthesis of jasmonate, a signal molecule for a variety of physiological events, including plant defense against insects (6). Expression of the Lipoxygemse Gene after Pathogen Attack-We examined expression of the gene in the leaves of a rice cultivar, Aichiasahi which cames a resistance gene, Pi-a, for the rice blast fungus. The fifth leaves of young plants were inoculated with either an incompatible race, 131, or a compatible race, 007, of the blast fungus, M. grisea, and the amount of the lipoxygenase protein and RNA was measured at various time points. We detected the lipoxygenase protein in rice leaves using antibodies raised against the protein expressed in E. coli.
The lipoxygenase protein (approximately 100 kDa) was detectable by Western blot analysis of leaf samples by 15 h after inoculation with the incompatible race, and the amount of the protein increased until 36 h after inoculation (Fig. 5). Much lower levels of the protein bands were observed on Western blots prepared from leaves infected with the compatible race of fungus, but no visible band was seen using mock-infected leaves (Fig. 5). Interestingly, at 15 h after the inoculation with the incompatible fungus, an additional faint higher molecular weight band was observed on the blot. This band may consti- increased until 33 h aRer inoculation (Fig. 6). In contrast, the expression of the mRNA was weak in leaves infected with the compatible pathogen and very weak in mock-infected leaves 24 h after treatment (Fig. 6). The amount of RNA used in these experiments was normalized using expression of a rice actin gene as an internal control.
These results indicate that expression of the lipoxygenase gene is specific to the incompatible interaction between rice and the pathogen. It is noteworthy that the start of lipoxygenase gene expression coincides with penetration of the rice leaf by the pathogen (23), indicating that gene expression is an early event in the host plant response against the pathogen.
Genomic Southern Analysis-Genomic DNA was isolated from leaves of the rice cultivar, Aichiasahi, and digested with restriction enzymes for Southern analysis (Fig. 7). The result of this experiment indicates the presence of a single copy of the lipoxygenase gene in the rice genome.

DISCUSSION
We isolated a novel lipoxygenase cDNA from rice leaves after inoculation with the incompatible blast fungus. The gene product is a novel type of plant lipoxygenase and contains two regions: an amino-terminal sequence (Met' to Ilel'o) and an LHR-(Gly"' to IlegZ3). The LHR has the lowest amino acid identity among lipoxygenases studied to date; similarity even with other rice lipoxygenases such as L-2 is quite low (13) ( Table I). In spite of low homology, amino acids generally conserved among plant and mammalian lipoxygenases, especially histidine residues which have been suggested to be putative iron-binding sites (1,31, are present in the protein sequence (Fig. 2). More recently, from the electron density map of the crystalized soybean lipoxygenase L-1, Minor et al. (37) identified five ligands of the active site iron atom in the enzyme, His4%, His504, A d s 4 , and Iles3s, the carboxyl-terminal residue which are also found in the sequence. The hydropathy profile of the LHR is similar to those of rice lipoxygenase L-2 and dicot lipoxygenases. These results and expression of an active lipoxygenase in E. coli from part of the LHR indicate that the gene encodes an active plant iipoxygenase.
We propose to classify this gene into a new category of plant lipoxygenase gene.
The The gene isolated in this study is expressed in rice leaves during an incompatible interaction between rice and a fungal pathogen, M. grisea. Southern analysis showed that the gene is derived from the rice genome (Fig. 71, indicating that the gene expression is a host reaction against the incompatible race of the pathogen. The gene expression (mRNA and protein) was observed 15 h after inoculation with the incompatible race of the pathogen (Figs. 5 and 6), the time at which the pathogen begins to penetrate into the leaves (23). This coincidence indicates that Lox2:Os:l expression is an early response of the plant against pathogen attack. In contrast, expression of the gene was weak when the plant was inoculated with a compatible race of the pathogen, although the compatible race penetrate into the leaves at the same time as the incompatible race (23). These results suggest that Lox2:Os:l gene expression is involved in plant defense against the invading pathogen. Recently, Koch et al. (20) found an increase in lipoxygenase mRNA in tomato leaves upon inoculation with plant pathogenic bacteria. They detected the lipoxygenase mRNA using a heterologous probe derived from a pea lipoxygenase cDNA (76% identity at the nucleotide level to soybean lipoxygenase L-3) under medium stringent conditions (50 "C, 2 x SSC). It seems likely that the tomato lipoxygenase mRNA belongs to L o x l . However, this question must be investigated further by nucleotide sequencing of the tomato lipoxgenase mRNA. More recently, Melan et al. (21) reported cloning of an Arabidopsis thulium lipoxygenase cDNA and showed induction of the gene by pathogens. That gene can be classified into L o x l . It will be of considerable interest to determine whether a counterpart of the Lox2 gene studied here also exists among dicots.
Using DEAE-'Ibyopearl chromatography, Ohta et al. (17) purified three lipoxygenases, leaf lipoxygenase-1, leaf lipoxygenase-2, and leaf lipoxygenase-3, from leaves both inoculated and uninoculated with race 131. They found that the increase in overall lipoxygenase activity was attributable mainly to leaf lipoxygenase-3, although activities of leaf lipoxygenase-1 and leaf lipoxygenase-2 also increased slightly. The similar induction pattern we observed using antibodies prepared against the product of Lox2:Os:l suggests its identity as leaf lipoxygenase-3. However, further studies will be required to confirm its identity.
It seems likely that the amino-terminal sequence, Met' to Ile'lo of the pathogen inducible lipoxygenase is a transit peptide for targeting the lipoxygenase into chloroplasts. However, we have not yet confirmed its intracellular location for the mature protein. Chloroplasts contain both substrates, oxygen and unsaturated fatty acids, required for the lipoxygenase reaction. A rapid disappearance of chloroplast glycolipids accompanied by increases in other lipid species was observed in leaves subjected to rust infection (38). It is possible that lipoxygenase localized in the same compartment produces hydroperoxides in combination with the release of fatty acids during conversion of glycolipids to other lipids.
It remains to be determined whether lipoxygenase gene expression in rice leaves is involved in the production of antifungal compounds, oxygenated fatty acids, or in the production of other physiological active compounds. Li et al. (39) showed that 13-hydroperoxides and 13-hydroxides of both linoleic and linolenic acids rapidly increase in areas of rice leaves which have been inoculated with the rice blast fungus. The highest concentrations of these compounds were reached within 24 h after inoculation. These results are consistent with our observation of lipoxygenase gene expression in rice leaves. Croft et al. (40) proposed a role for lipoxygenase in initiating processes that lead to hypersensitive cell death by showing that increases in lipoxygenase activity precede increases in superoxide dismutase and peroxidase activity in Phaseolus vulgaris (L) cv. Red Mexican, in response to infection with an incompatible race of Pseudomonas syringae, pv. Phaseolicola race 1. It will be useful to introduce antisense lipoxygenase messages into rice for assessment of iipoxygenase function in plant defense mechanisms.