Arabidopsis HARMLESS TO OZONE LAYER Protein Methylates a Glucosinolate Breakdown Product and Functions in Resistance to Pseudomonas syringae pv. maculicola*

Almost all of the chlorine-containing gas emitted from natural sources is methyl chloride (CH3Cl), which contributes to the destruction of the stratospheric ozone layer. Tropical and subtropical plants emit substantial amounts of CH3Cl. A gene involved in CH3Cl emission from Arabidopsis was previously identified and designated HARMLESS TO OZONE LAYER (hereafter AtHOL1) based on the mutant phenotype. Our previous studies demonstrated that AtHOL1 and its homologs, AtHOL2 and AtHOL3, have S-adenosyl-l-methionine-dependent methyltransferase activities. However, the physiological functions of AtHOLs have yet to be elucidated. In the present study, our comparative kinetic analyses with possible physiological substrates indicated that all of the AtHOLs have low activities toward chloride. AtHOL1 was highly reactive to thiocyanate (NCS−), a pseudohalide, synthesizing methylthiocyanate (CH3SCN) with a very high kcat/Km value. We demonstrated in vivo that substantial amounts of NCS− were synthesized upon tissue damage in Arabidopsis and that NCS− was largely derived from myrosinase-mediated hydrolysis of glucosinolates. Analyses with the T-DNA insertion Arabidopsis mutants (hol1, hol2, and hol3) revealed that only hol1 showed increased sensitivity to NCS− in medium and a concomitant lack of CH3SCN synthesis upon tissue damage. Bacterial growth assays indicated that the conversion of NCS− into CH3SCN dramatically increased antibacterial activities against Arabidopsis pathogens that normally invade the wound site. Furthermore, hol1 seedlings showed an increased susceptibility toward an Arabidopsis pathogen, Pseudomonas syringae pv. maculicola. Here we propose that AtHOL1 is involved in glucosinolate metabolism and defense against phytopathogens. Moreover, CH3Cl synthesized by AtHOL1 could be considered a byproduct of NCS− metabolism.

Almost all of the chlorine-containing gas emitted from natural sources is methyl chloride (CH 3 Cl), which contributes to the destruction of the stratospheric ozone layer. Tropical and subtropical plants emit substantial amounts of CH 3 Cl. A gene involved in CH 3 Cl emission from Arabidopsis was previously identified and designated HARMLESS TO OZONE LAYER (hereafter AtHOL1) based on the mutant phenotype. Our previous studies demonstrated that AtHOL1 and its homologs, AtHOL2 and AtHOL3, have S-adenosyl-L-methionine-dependent methyltransferase activities. However, the physiological functions of AtHOLs have yet to be elucidated. In the present study, our comparative kinetic analyses with possible physiological substrates indicated that all of the AtHOLs have low activities toward chloride. AtHOL1 was highly reactive to thiocyanate (NCS ؊ ), a pseudohalide, synthesizing methylthiocyanate (CH 3 SCN) with a very high k cat /K m value. We demonstrated in vivo that substantial amounts of NCS ؊ were synthesized upon tissue damage in Arabidopsis and that NCS ؊ was largely derived from myrosinase-mediated hydrolysis of glucosinolates. Analyses with the T-DNA insertion Arabidopsis mutants (hol1, hol2, and hol3) revealed that only hol1 showed increased sensitivity to NCS ؊ in medium and a concomitant lack of CH 3 SCN synthesis upon tissue damage. Bacterial growth assays indicated that the conversion of NCS ؊ into CH 3 SCN dramatically increased antibacterial activities against Arabidopsis pathogens that normally invade the wound site. Furthermore, hol1 seedlings showed an increased susceptibility toward an Arabidopsis pathogen, Pseudomonas syringae pv. maculicola. Here we propose that AtHOL1 is involved in glucosinolate metabolism and defense against phytopathogens. Moreover, CH 3 Cl synthesized by AtHOL1 could be considered a byproduct of NCS ؊ metabolism.
Methyl chloride (CH 3 Cl) is the most abundant halohydrocarbon emitted into the atmosphere and constitutes about 17% of the chlorine currently in the stratosphere (1). CH 3 Cl is derived mainly from natural sources and contributes to the destruction of the stratospheric ozone layer. As the total abundance of ozone-depleting gases such as chlorofluorocarbons in the atmosphere has begun to decrease in recent years as a result of The Montreal Protocol on Substances That Deplete the Ozone Layer, the impact of CH 3 Cl emission from natural sources will become greater on the atmospheric chemistry. CH 3 Cl emission into the atmosphere has been estimated at 1,700 -13,600 Gg/year (1), which underscores the great uncertainty of the estimation. Oceans (2), biomass burning (3), woodrotting fungi, and coastal salt marshes (4) are the major sources of CH 3 Cl production. Recently, it was reported that large amounts of CH 3 Cl are emitted from tropical and subtropical plants, which are hence considered as the major sources of CH 3 Cl (5-7). It was estimated that the CH 3 Cl emission from tropical plants could account for 30 -50% of the global CH 3 Cl emission (8). To accomplish an accurate estimation of CH 3 Cl production in the atmosphere through "bottom-up" approaches, elucidating the mechanisms and physiological functions of CH 3 Cl emission from plants will be important.
The biological synthesis of methyl halides has been demonstrated mainly by biochemical analyses. The enzymatic activities that transfer a methyl group from S-adenosyl-L-methionine (SAM) 2 to halide ions (Cl Ϫ , Br Ϫ , I Ϫ ), which synthesize methyl halides, were first discovered in cell-free extracts of Phellinus pomaceus (a white rot fungus), Endocladia muricata (a marine red alga), and Mesembryanthemum crystallinum (ice plant, a halophytic plant) (9). Enzyme purification and cDNA cloning of the methyl chloride transferase (MCT) was first reported with Batis maritima, a halophytic plant that grows abundantly in salt marshes. As high concentrations of ions such as Cl Ϫ are often detrimental to plants, halophytic plants are considered to possess various salt tolerance mechanisms. MCT was hypothesized to control and regulate the internal concentration of Cl Ϫ , rich in the habitat in which halophytic plant grows (10,11).
In the meantime, purification of thiol methyltransferase (TMT), which methylates bisulfide (HS Ϫ ) and halide (Cl Ϫ , Br Ϫ , I Ϫ ) ions was reported with cabbage, Brassica oleracea (12). The purified and recombinant TMTs were later shown to also * This work was supported by Grant-in-aid for Scientific Research 19780249 methylate the thiocyanate ion (NCS Ϫ ), which is called pseudohalide because of its chemical properties similar to halide ions (13,14). NCS Ϫ is a hydrolysis product found in some glucosinolates, which are secondary metabolites found mainly in the order Brassicales including the model plant Arabidopsis thaliana (15). Upon tissue damage such as by insect or herbivore attack, glucosinolates are hydrolyzed by myrosinase (␤-thioglucosidase) into biologically active compounds including isothiocyanates. Isothiocyanates derived from indole glucosinolates and 4-hydroxybenzyl glucosinolates are reported to be highly unstable and yield NCS Ϫ upon reacting with various nucleophiles (15)(16)(17). Based on the enzymatic activity, the physiological role of TMT was speculated to metabolize glucosinolate breakdown products (14). However, there are no reported studies that examine these MCT and TMT hypotheses through in vivo experiments.
An Arabidopsis homolog of MCT was also identified, and its T-DNA insertion Arabidopsis mutants were analyzed (18). Because the gene disruption eliminated almost all of the methyl halide emissions from the mutants, the gene was revealed to be involved in methyl halide synthesis and was designated HOL (HARMLESS TO OZONE LAYER; denoted as AtHOL1 in our studies) based on the mutant phenotype (18). Recently, we identified AtHOL1 homologs AtHOL2 and AtHOL3 in Arabidopsis, and we demonstrated biochemically that the three recombinant AtHOLs have SAM-dependent methyltransferase activities (19). In this study, reverse genetic and biochemical analyses of all AtHOL isoforms revealed that AtHOL1 in vivo is involved in the methylation of NCS Ϫ produced by glucosinolate hydrolysis. Although there are several studies that have examined the biological activities of glucosinolate hydrolysis products, the mechanisms of NCS Ϫ synthesis and its methylation to methyl thiocyanate (CH 3 SCN) have yet to be reported in detail. The biological activity and physiological function of CH 3 SCN synthesized by AtHOL1 was also examined.

EXPERIMENTAL PROCEDURES
Plant Materials and Growth Conditions-Wild-type Arabidopsis (A. thaliana ecotype Col-0), three T-DNA insertion Arabidopsis mutants (hol1, hol2, hol3), and the three AtHOL1overexpressing Arabidopsis lines were used in this study. Seeds were sterilized and grown on half-strength Murashige-Skoog (MS) agar medium or soil under controlled conditions (12 h light/12 h dark cycle at 22 Ϯ 1°C).
Phylogenetic Analysis-The amino acid sequences of HOL homologs deduced from the completed genomic sequences, cDNA sequences, and assembled expressed sequence tag sequences from various photosynthetic organisms were extracted through BLAST search (blast.ncbi.nlm.nih.gov/Blast.cgi) of the available nucleic acid sequence data bases. The unrooted phylogenetic tree was constructed based on the obtained amino acid sequences using the ClustalW (21) and Njplot programs (22).
Overexpression of AtHOL1 in Arabidopsis-The AtHOL1 cDNA fragment containing the open reading frame was prepared by PCR (19) and cloned between the XbaI and XhoI sites in the binary vector pBI121 under the control of the cauliflower mosaic virus 35S promoter. The constructed binary vector (pBI121-AtHOL1) was introduced into Agrobacterium tumefaciens (LBA4404) by electroporation, which was then used to transform wild-type Arabidopsis by floral dipping. Transgenic T1 Arabidopsis were selected on half-strength MS agar medium containing 25 mg/liter kanamycin sulfate.
GC-ECD and GC-MS Analyses-The quantitative determination of CH 3 SCN was carried out by gas chromatography equipped with an electron capture detector (GC-ECD; GC-9A, Shimadzu, Kyoto, Japan). The samples in 2.0-ml glass vials were incubated at 70°C for 30 min, and each headspace was sampled using a gastight syringe followed by an injection into a 200 ϫ 0.3-cm inner diameter stainless column packed with Porapak Q (Waters) in the GC-ECD. The column and injection port temperatures were 180 and 250°C, respectively, and the flow rate of the carrier gas (N 2 ) was 40 ml/min. The product was identified by comparison with the retention time of pure CH 3 SCN (Fluorochem, Glossop, UK) and quantified by peak area. The structural identification of CH 3 SCN produced by Arabidopsis was performed by gas chromatography/mass spectrometry (GC-MS; 5973 MSD, Agilent Technologies, Wilmington DE) using the HP-5ms semi-volatile column (30 m ϫ 0.25 mm inner diameter, 0.25-m film thickness, Agilent Technologies). The samples in 2.0-ml glass vials were incubated at 70°C for 30 min, and each sampled 500-l head space was injected into the column. The carrier gas (helium) flow was 0.9 ml/min. The column temperature was 30°C, and the mass spectrometer was operated in electron impact mode at 70 eV. The CH 3 SCN was identified by comparison with pure CH 3 SCN in selected ion-monitoring mode at mass to charge ratios (m/z) of 73, 72, and 46 (supplemental Fig. S1).
Kinetic Analyses of Recombinant AtHOL Proteins-For kinetic analyses of AtHOL1, AtHOL2, and AtHOL3, glutathione S-transferase-tagged recombinant AtHOLs were expressed in Escherichia coli strain BL21, and tag-deleted AtHOLs were purified from soluble fractions as described (19). Each of the recombinant AtHOLs had 15 extra amino acids (GSTSLYKK-AGSEFAL) at the N terminus. The assay mixture volume was 75 l containing 0.1 M Tris acetate (pH 7.5), up to 1.58 g of each of the AtHOLs, and varying concentrations of each substrate. S-(5Ј-Adenosyl)-L-methionine chloride (Wako, Osaka, Japan), KSCN (Nacalai Tesque, Kyoto, Japan), KCl (Nacalai Tesque), and (NH 4 ) 2 S (Nacalai Tesque) were used as substrates. Because the reaction rates were linear for at least 1 h (data not shown), all of the assay mixtures were incubated for 1 h at 25°C. To stop the reaction, 1 M HClO 4 (75 l) was added to the assay mixture. After centrifugation (16,100 ϫ g for 10 min), 50 l of the supernatant was injected into the HPLC system. The methyltransferase activities of AtHOLs were measured by quantifying the SAH produced by the enzymatic reactions.
Quantitative Analyses of NCS Ϫ Synthesis in Arabidopsis-In the assay method that was established to sensitively and specifically quantify NCS Ϫ in Arabidopsis, NCS Ϫ was methylated by the recombinant AtHOL1 protein and converted to CH 3 SCN, which was then quantified using GC-ECD. 3 The quantification was not influenced by ingredients in Arabidopsis extracts. To quantify NCS Ϫ in homogenized Arabidopsis, ϳ4-week-old wild-type and T-DNA mutant lines were homogenized and centrifuged at 16,100 ϫ g for 10 min. Each supernatant (20 l) was analyzed. To investigate the effects of lower degrees of wounding on NCS Ϫ synthesis, ϳ20 mg (fresh weight) of 4-week-old wild-type Arabidopsis seedlings were nicked by scissors along the mid-vein of all the leaves. Then the seedlings were homogenized on ice with 100 l of 2.5% 5-sulfosalicylic acid to denature proteins contained in the samples, and the mixtures were neutralized by adding 0.2 N NaOH (100 l). The homogenates were centrifuged at 16,100 ϫ g for 10 min, and the supernatant (80 l) was analyzed. Following the same procedure, basal level NCS Ϫ concentration was determined with the intact Arabidopsis seedlings that were homogenized under denaturing conditions. To examine the influence of myrosinase on NCS Ϫ production, 0.025 unit of myrosinase (thioglucosidase, EC 3.2.1.147; Sigma-Aldrich) was added to the extracts of the intact Arabidopsis seedlings.
Quantitative Analyses of CH 3 SCN Synthesis in Arabidopsis-About 400 mg (fresh weight) of 4-week-old wild-type and T-DNA insertion mutant Arabidopsis were homogenized on ice. The samples were centrifuged at 16,100 ϫ g for 10 min, and each supernatant (200 l) was transferred to a 2.0-ml glass vial sealed with a screw cap fitted with a Teflon-lined septum and incubated overnight at 25°C followed by the head-space analyses of CH 3 SCN by GC-ECD. To examine the rate-limiting substrate in CH 3 SCN production, the supernatants (180 l) were mixed with 10 mM KSCN (20 l) or 10 mM SAM (20 l), incubated overnight, and analyzed by GC-ECD as indicated above. To examine the involvement of wounding in CH 3 SCN production, 3-week-old wild-type Arabidopsis grown on soil were used. Wild-type Arabidopsis (ϳ200 mg) was wounded as performed in the analyses of NCS Ϫ synthesis indicated above. The samples were incubated overnight in 2.0-ml glass vials at 25°C, and the 500-l head space was analyzed for CH 3 SCN by GC-ECD.
In Vitro Assay of Antibacterial Activity-Two bacterial pathogens, Pseudomonas syringae pv. maculicola (MAFF 302539) and Erwinia carotovora subsp. carotovora (MAFF 301879), which were obtained from the Genebank of the National Institute of Agrobiological Sciences, Japan, and a nonpathogenic bacterium, E. coli strain DH5␣, were used for antibacterial activity assay. All of the bacteria were cultured in LB medium at 28°C for P. syringae and E. carotovora and at 37°C for E. coli. The bacteria were cultured overnight for E. carotovora and E. coli and for 2 days for P. syringae. The concentrations of KSCN and CH 3 SCN causing 50% bacterial growth inhibition (IC 50 ) were determined by monitoring the cell populations at A 600 (23,24). The assay cultures (200 l) were prepared from the bacterial suspension cultures at an A 600 of 0.2, and KSCN and CH 3 SCN were dissolved in water and ethanol, respectively. To avoid volatilization of CH 3 SCN, each assay culture was incubated in a 2.0-ml glass vial sealed with a screw cap fitted with a Teflon-lined septum. The cultures were incubated at temperatures appropriate for each bacterium and cultured by shaking at 150 rpm for 15 h followed by measuring the A 600 . The corresponding amounts of solvents used to dissolve KSCN and CH 3 SCN did not affect bacterial growth (data not shown).
Pathogen Inoculation-P. syringae pv. maculicola was cultured in LB medium as described above and harvested by centrifugation at 4000 ϫ g for 10 min. The bacterial pellet was rinsed once by 10 mM MgCl 2 and resuspended in 10 mM MgCl 2 at an A 600 of 0.5 (25). The wild-type and hol1 Arabidopsis seeds (ϳ400 seeds each) grown on half-strength MS agar medium were inoculated with droplets of the bacterial suspension (3 l). For mock treatment, 10 mM MgCl 2 solution was used instead of the bacterial suspension. The survival rate of the seedlings grown for 19 days was evaluated based on the emergence of true leaves.

RESULTS
Phylogenetic Analysis of HOL Genes-BLAST searches of nucleotide sequence data bases were performed using the amino acid sequence of AtHOL1 as a query. Twenty-five HOL homologs from 19 different plant species were distributed from unicellular green algae to gymnosperms and angiosperms (supplemental Fig. S2). All of the plant species such as Arabidopsis, rice, poplar, wine grape, milo, moss (Physcomitrella patens), and red alga (Cyanidioschyzon merolae), for which the genomes have been completely sequenced, contained at least one HOL homolog, suggesting that HOL is a gene conserved in photo-synthetic organisms. A phylogenetic tree was constructed based on the amino acid sequences of HOL homologs from multicellular species only (Fig. 1). The tree reflected the gene-alogy of the derived species. One of the most remarkable aspects was that HOL homologs from Brassicales plants including Arabidopsis were grouped into two clusters, cluster I and cluster II. However, HOL homologs from dicotyledonous plants other than Brassicales plants were all grouped in a single cluster, cluster III, and those from monocotyledonous plants were also all grouped in a single cluster, cluster IV. Because AtHOL1 and AtHOL2 formed a cluster together with TMT and MCT, we hypothesized that AtHOL proteins would have enzymatic activities similar to those of TMT and MCT. However, a detailed enzymatic characterization of the AtHOL proteins had yet to be reported.
Kinetic Characterization of Recombinant AtHOL Proteins-We first examined the substrate specificities and kinetic characteristics of the three AtHOL proteins. Among the compounds that worked as in vitro substrates for MCT or TMT (11,14), three substrates (Cl Ϫ , NCS Ϫ , and HS Ϫ ) that were known to exist in Arabidopsis were analyzed. Glutathione S-transferasetagged recombinant AtHOL proteins were expressed in E. coli, and tag-deleted AtHOL proteins were   Fig. S3), which permitted more proper folding of the proteins than those prepared from insoluble fractions as done in previous studies (11,13). Enzymatic activities for the three substrates were largely distinct among AtHOL isoforms ( Table 1). The catalytic efficiencies (k cat /K m ) of all AtHOL proteins for Cl Ϫ (AtHOL1, 7.9 ϫ 10 3 ; AtHOL2, 1.5 ϫ 10 5 ; AtHOL3, not detected) were the lowest among the three analyzed substrates. On the other hand, AtHOL1 was highly reactive to NCS Ϫ with a markedly high k cat /K m value (9.4 ϫ 10 8 ). Based on the k cat /K m values, the most preferred substrate for both AtHOL2 and AtHOL3 was HS Ϫ .
Isolation of AtHOL T-DNA Insertion Mutants-To investigate the physiological functions of the three AtHOL genes through in vivo experiments, we obtained T-DNA insertion Arabidopsis mutants for AtHOL1, AtHOL2, and AtHOL3. As shown in Fig. 2A, the mutants for AtHOL1 (SALK_005204C), AtHOL2 (SALK_021226C) and AtHOL3 (SALK_014648C) possessed T-DNA inserts within introns 2, 6, and 4, respectively. Semiquantitative RT-PCR analyses with the obtained homozygous mutant plants, designated hol1, hol2, and hol3, indicated no transcript accumulation of the T-DNA-inserted AtHOL genes in each mutant (Fig. 2B). The obtained hol1 plant was the same line used to show that AtHOL1 is responsible for methyl halide production (18). There were no morphological defects in these mutant plants under normal growth conditions (data not shown).
Responses of Wild-type, AtHOL-disrupted, and AtHOL1overexpressing Arabidopsis to NCS Ϫ -Based on our prediction that the substrate preferences of AtHOLs shown in our kinetic analyses reflect the in vivo responses to the substrates, we examined the response of the wild type and the three hol mutant Arabidopsis toward NCS Ϫ , HS Ϫ , and Cl Ϫ . All of the analyzed Arabidopsis seedlings showed the same growth under various concentrations of (NH 4 ) 2 S (as HS Ϫ ), NaCl (as Cl Ϫ ), and KCl (as Cl Ϫ ) (data not shown). On the other hand, among the three mutants, only hol1 showed an increased sensitivity to 1 mM KSCN (as NCS Ϫ ) (Fig. 3A). AtHOL1-overexpressing Arabidopsis were grown on the medium containing a higher con-centration of KSCN (2.5 mM), at which concentration the growth of wild-type Arabidopsis was inhibited (26,27) (Fig. 3B). Compared with wild-type Arabidopsis, ϳ4-week-old AtHOL1overexpressing plants showed clearer resistance to KSCN on the medium (Fig. 3B) than the seedlings just after germination (28). These results provided the first evidence that, among the three isoforms, only AtHOL1 metabolized NCS Ϫ in vivo. NCS Ϫ and CH 3 SCN Synthesis in Wild-type and AtHOL-disrupted Arabidopsis-Thereafter, we focused on the functional analyses of AtHOL1, which was highly reactive to NCS Ϫ . To date, NCS Ϫ has been detected in crushed glucosinolate-containing plants such as Brassica vegetables (16). Other reports showed that in vitro hydrolysis of indole glucosinolates by myrosinase produces NCS Ϫ (16). However, to our knowledge, there have been no detailed in vivo studies demonstrating that NCS Ϫ is derived solely from myrosinase-catalyzed hydrolysis of glucosinolates. Myrosinases and glucosinolates are stored separately in intact plant tissues. Upon tissue wounding, such as caused by chewing insects, these enzymes come into contact followed by glucosinolate hydrolysis (29 -31). To investigate NCS Ϫ synthesis, we first established an assay to sensitively and  specifically quantify NCS Ϫ in the supernatant of homogenized, "thoroughly wounded" Arabidopsis seedlings. In the assay, NCS Ϫ in the reaction mixture was converted to CH 3 SCN by recombinant AtHOL1, which specifically methylates NCS Ϫ , and then CH 3 SCN was quantified by GC-ECD (see "Experimental Procedures"). By utilizing the established method, we demonstrated that NCS Ϫ accumulated to ϳ140 M in homogenized wild-type Arabidopsis seedlings, and the accumulation was unaffected by the disruption of each AtHOL gene (Fig. 3C). Because the concentration of accumulated NCS Ϫ (ca. 140 M) was much higher than the K m value (62 M) for NCS Ϫ of the recombinant AtHOL1, we hypothesized that NCS Ϫ was methylated to CH 3 SCN by AtHOL1 in homogenized Arabidopsis seedlings. Indeed, GC-MS analyses confirmed the synthesis of CH 3 SCN in the supernatant of homogenized wild-type Arabidopsis seedlings (supplemental Fig. S1). To investigate the involvement of the three AtHOL genes in CH 3 SCN production, we quantified CH 3 SCN synthesized in the supernatant of each homogenized AtHOL-disrupted mutant, hol1, hol2, and hol3. As a result, CH 3 SCN synthesis in hol1 was undetectable, whereas both hol2 and hol3 showed the same levels of CH 3 SCN synthesis as wild-type Arabidopsis (Fig. 3D).
Involvement of Myrosinase in NCS Ϫ and CH 3 SCN Synthesis in Arabidopsis-We then verified the involvement of myrosinase in NCS Ϫ and CH 3 SCN production in Arabidopsis. The amount of NCS Ϫ and CH 3 SCN in the seedlings of leaves uniformly wounded by scissors was 13 and 7.7%, respectively, that of the homogenized ones (Fig. 4, A and B). In the unwounded seedlings, NCS Ϫ accumulation was only 3.7% of the homogenized ones, and CH 3 SCN synthesis was undetectable. These results suggested that both NCS Ϫ and CH 3 SCN synthesis varied based on the degree of wounding. When purified myrosinase was added to the assay mixture of unwounded seedlings, the NCS Ϫ concentration increased to the same level as that in the homogenized seedlings. This result revealed that myrosinase was involved in the in vivo NCS Ϫ production, and almost all of the accumulated NCS Ϫ in homogenized Arabidopsis seedlings was actually derived from glucosinolates.
Rate-limiting Substrate for CH 3 SCN Synthesis in Arabidopsis-We then determined the ratelimiting substrate for CH 3 SCN synthesis. The addition of KSCN to the supernatant of homogenized Arabidopsis seedlings did not increase CH 3 SCN synthesis, indicating that all the SAM was exhausted in the supernatant. Meanwhile the addition of SAM increased the synthesis by ϳ40-fold (Fig. 4C). These results indicated that the supply of SAM may be rate-limiting for CH 3 SCN synthesis, although CH 3 SCN synthesis was triggered by wounding that first induced NCS Ϫ synthesis.
In Vitro Antibacterial Activities of NCS Ϫ and CH 3 SCN-Thus far, there have been several studies reporting on the biological activities of glucosinolate-hydrolyzed products, typically isothiocyanates and nitriles (29). As for NCS Ϫ , none of the reported studies could detect biological activities on the analyzed bacteria and fungi (15,32). On the other hand, a limited number of studies have demonstrated that bacteria belonging to Pseudomonas genus showed a negative chemotaxis to CH 3 SCN (33,34). Based on this finding, we hypothesized that CH 3 SCN synthesized by Arabidopsis also had biological activities toward bacteria. To answer this question, we examined the influence of CH 3 SCN on bacterial growth. P. syringae pv. maculicola and E. carotovora subsp. carotovora, which are pathogenic to Arabidopsis and normally invade the wound site, and non-pathogenic E. coli strain DH5␣ were selected, and their growth was investigated under various concentrations of KSCN and CH 3 SCN. The KSCN concentration that caused 50% inhibition of bacterial growth (IC 50 ) was over 50 mM for all of the bacteria analyzed (Fig. 5). On the other hand, the IC 50 of CH 3 SCN for P. syringae pv. maculicola, E. carotovora subsp. carotovora, and E. coli was dramatically lowered to 1.2, 9.4, and 3.6 mM, respectively. These results suggested, for the first time, that the conversion of NCS Ϫ into CH 3 SCN catalyzed by AtHOL1 increased the toxicity of NCS Ϫ to the bacteria.
Enhanced Susceptibility of AtHOL1-disrupted Arabidopsis to P. syringae pv. maculicola-To investigate whether the pathogens were affected by CH 3 SCN synthesis in planta, we tested the wild-type and hol1 Arabidopsis on their susceptibility to P. syringae pv. maculicola. Because younger wild-type seedlings accumulate larger amounts of AtHOL1 mRNA than 4-week-old seedlings (18), we hypothesized that the differences in suscep- tibility between wild-type and hol1 Arabidopsis might be observed with younger seedlings. Seeds of wild-type and hol1 Arabidopsis were infected with P. syringae pv. maculicola and grown for 19 days on half-strength MS agar medium (supplemental Fig. S4). Indeed, the percentage of surviving seedlings was significantly lower for hol1 (7.5%) compared with wild-type Arabidopsis (21%) (Fig. 6), suggesting that hol1 Arabidopsis increased the susceptibility toward P. syringae pv. maculicola.

DISCUSSION
AtHOL1 Was Highly Reactive to NCS Ϫ rather than Cl Ϫ in Arabidopsis-We demonstrated, for the first time, the comparative kinetic analyses of all the HOL isoforms in a single plant species and found that AtHOL1 was highly reactive to NCS Ϫ . AtHOL1 was originally reported to be involved in the produc-tion of methyl halides in Arabidopsis (18). However, the kinetic analyses of recombinant AtHOL1 showed that the k cat /K m value for Cl Ϫ was lowest among the analyzed substrates, and the K m value (280 mM) was much higher than the normal intracellular Cl Ϫ concentration ( Table 1). The activity of AtHOL3 for Cl Ϫ was undetectable, whereas that of AtHOL2 was the highest among the three isoforms. However, the contribution of AtHOL2 to CH 3 Cl production was hypothesized to be subtle because methyl halide emissions depend mostly on AtHOL1 (18), and AtHOL2 mRNA accumulation is much lower than AtHOL1 (19). Ni and Hager (11) proposed that the physiological function of MCT, an AtHOL1 homolog in a Brassicales plant, is to regulate the intracellular Cl Ϫ concentration. However, the amount of CH 3 Cl emitted from Arabidopsis was so low that intracellular Cl Ϫ concentrations could not be reduced in Arabidopsis tissues. This hypothesis is supported by the results of our kinetic analyses showing the lower activities of AtHOLs toward Cl Ϫ and by the results showing that none of the hol mutants was more sensitive than wild-type Arabidopsis toward 0 -100 mM KCl or NaCl in the half-strength MS agar medium (data not shown). These results together suggest that Cl Ϫ is not the physiological substrate for AtHOL proteins.
HS Ϫ was the most preferred substrate for AtHOL2 and AtHOL3 and the second most preferred substrate for AtHOL1 among the analyzed substrates. Therefore, the activity of AtHOL proteins for HS Ϫ may also contribute to the synthesis of CH 3 SH (35,36), which is reported also to be synthesized from methionine hydrolysis by methionine ␥-lyase in Arabidopsis (37).
On the other hand, NCS Ϫ was highly reactive for AtHOL1 with a k cat /K m that was ϳ250and 980-fold higher than that for AtHOL2 and AtHOL3, respectively. The NCS Ϫ concentration (ϳ140 M) in the homogenized Arabidopsis tissue was higher than the K m value (62 M) of AtHOL1 for NCS Ϫ (Table 1). Furthermore, CH 3 SCN synthesis was dependent on AtHOL1 (Fig. 3). These results suggest that NCS Ϫ is the physiological substrate for AtHOL1. Through reverse genetic and biochemical analyses, we have demonstrated that AtHOL1 is responsible for CH 3 SCN production in vivo (Fig. 3).
NCS Ϫ Was Derived from Glucosinolate Hydrolysis in Arabidopsis-We demonstrated that NCS Ϫ and CH 3 SCN synthesis varied based on the degree of wounding; the synthesis of almost all NCS Ϫ was derived from glucosinolates through hydrolysis by myrosinase (Fig. 4). Our in vivo studies demonstrated a novel metabolic pathway of a glucosinolate breakdown product catalyzed by AtHOL1. Indole glucosinolate breakdown products have also been detected on intact leaf surfaces (38). Glucosinolates in intact tissues of Arabidopsis  . Susceptibility of wild-type and AtHOL1-disrupted Arabidopsis toward P. syringae pv. maculicola. The survival rate of the seedlings was evaluated based on the emergence of true leaves and was normalized by the seed germination rate of wild-type (WT) and hol1 Arabidopsis. The germination rate between wild-type and hol1 seeds was not significantly different (data not shown). The survival rate was significantly different (***, p Ͻ 0.001, 2 ϭ 31.79 by chi-square test) between wild-type (n ϭ 452) and hol1 (n ϭ 428) Arabidopsis. The data are derived from three independent experiments for both seeds.
are constantly renewed in the cell (39,40). Sulfur deficiency also induces a conditional turnover of glucosinolates in Arabidopsis (41). Hence the basal level of NCS Ϫ in Arabidopsis tissues observed in Fig. 4A was possibly derived from the turnover of glucosinolates.

Role of CH 3 SCN Production in Plant
Defense-Only about 1% of NCS Ϫ could be converted to CH 3 SCN, possibly because of the exhaustion of SAM in the supernatant of homogenized Arabidopsis seedlings (Fig. 3, C and D). Thus, the primary physiological role of AtHOL1 may not be the detoxification of NCS Ϫ by volatilization (14). Rather, CH 3 SCN synthesis (by methylation of NCS Ϫ ) markedly increased its toxicity toward the bacteria (Fig. 5). To date, the biological activities of glucosinolate breakdown products such as nitriles and isothiocyanates have been studied in detail (29). For example, isothiocyanates contain antibacterial activities (15), and indole-3-acetonitrile has antifungal activity (42). Indole-3-carbinol and indole-3-acetonitrile act as a stimulant and a deterrent, respectively, for an ovipositing butterfly (38). CH 3 SCN reported herein may be considered a novel member of such biologically active compounds derived from glucosinolates. Although further studies are needed to understand how CH 3 SCN inhibits bacterial growth, CH 3 SCN indeed showed an inhibitory effect on the analyzed bacteria ranging from plant pathogens to a bacterium unrelated to plant pathogens. The observed effect on the different classes of bacteria implies that CH 3 SCN synthesis is involved in a broad spectrum antibacterial defense in Arabidopsis. Indeed, our in vivo studies have demonstrated that the disruption of AtHOL1 increases the susceptibility of the mutant seedlings toward P. syringae pv. maculicola (Fig. 6). These results imply that CH 3 SCN synthesis catalyzed by AtHOL1 is at least in part responsible for the resistance to P. syringae pv. maculicola.
To date, indole glucosinolates are the only glucosinolates reported to generate NCS Ϫ through a myrosinase-catalyzed hydrolysis that also exists in Arabidopsis. Indole glucosinolate synthesis is induced by methyl jasmonate (43), elicitors derived from the bacterial pathogen E. carotovora (32), and the fungal pathogen Alternaria brassicae (44) and by an interaction with a mycorrhizal fungus (45). PEN2 in Arabidopsis is responsible for efficient entry by the non-host fungus Blumeria graminis f. sp. hordei (46). PEN2 may preferentially hydrolyze indole glucosinolates producing defensive compounds at incipient fungal entry sites without tissue damage (40,46). These studies (40,(42)(43)(44)(45)(46) imply that CH 3 SCN production by AtHOL1 might also be induced by pathogens, in which case NCS Ϫ synthesis and a continuous supply of SAM within an intact cell would cause a higher level of CH 3 SCN synthesis at the pathogen entry site.
These observations may explain the difference between the CH 3 SCN concentration in homogenized wild-type Arabidopsis (Fig. 3D) and the CH 3 SCN concentration that inhibited bacterial growth (Fig. 5).
Methyl Halide Production by AtHOL1-In conclusion, our study suggests that AtHOL1 metabolizes glucosinolate-derived NCS Ϫ , which induces the plant defense mechanism (CH 3 SCN synthesis) (Fig. 7). Hence, methyl halide production by AtHOL1 could be considered a byproduct of NCS Ϫ metabolism in Arabidopsis. Glucosinolate-containing plants generally show a high level of methyl halide synthesis (47), which might be due to HOL genes belonging to cluster I (Fig. 1). On the other hand, we have shown that HOL homologs are widespread among photosynthetic organisms (Fig. 1). Hence, HOL homologs not found in cluster I may function differently than AtHOL1 and function commonly in photosynthetic organisms. Further analyses are needed to elucidate whether there is a common function for HOL homologs and their involvement in methyl halide emission, which is an important issue in global atmospheric chemistry.