Arabidopsis Isochorismate Synthase Functional in Pathogen-induced Salicylate Biosynthesis Exhibits Properties Consistent with a Role in Diverse Stress Responses*

Salicylic acid (SA) is a phytohormone best known for its role in plant defense. It is synthesized in response to diverse pathogens and responsible for the large scale transcriptional induction of defense-related genes and the establishment of systemic acquired resistance. Surprisingly, given its importance in plant defense, an understanding of the underlying enzymology is lacking. In Arabidopsis thaliana, the pathogen-induced accumulation of SA requires isochorismate synthase (AtICS1). Here, we show that AtICS1 is a plastid-localized, stromal protein using chloroplast import assays and immunolocalization. AtICS1 acts as a monofunctional isochorismate synthase (ICS), catalyzing the conversion of chorismate to isochorismate (IC) in a reaction that operates near equilibrium (Keq = 0.89). It does not convert chorismate directly to SA (via an IC intermediate) as does Yersinia enterocolitica Irp9. Using an irreversible coupled spectrophotometric assay, we found that AtICS1 exhibits an apparent Km of 41.5 μm and kcat = 38.7 min-1 for chorismate. This affinity for chorismate would allow it to successfully compete with other pathogen-induced, chorismate-utilizing enzymes. Furthermore, the biochemical properties of AtICS1 indicate its activity is not regulated by light-dependent changes in stromal pH, Mg2+, or redox and that it is remarkably active at 4 °C consistent with a role for SA in cold-tolerant growth. Finally, our analyses support plastidic synthesis of stress-induced SA with the requirement for one or more additional enzymes responsible for the conversion of IC to SA, because non-enzymatic conversion of IC to SA under physiological conditions was negligible.

thases (EC 5.4.99.6) catalyze the reversible conversion of chorismate to IC, 3 thereby controlling chorismate partitioning to IC-derived products. Although synthesis and regulation of the bacterial IC pathways have been investigated, there is limited knowledge regarding these enzymes, their products, and their functional roles in plants. In bacteria, ICS enzymes have been associated with either the synthesis of low molecular weight Fe 3ϩ chelators (siderophores) or menaquinones that function as electron acceptors. For example, Escherichia coli contains two ICS genes located in distinct operons encoding proteins with distinct biochemical properties: EntC is expressed under iron-limited conditions and associated with siderophore biosynthesis, and MenF is required for menaquinone biosynthesis and expressed under anaerobic conditions where it plays an essential role in anaerobic electron transport (2)(3)(4)(5). The initial steps in bacterial siderophore production from IC involve the synthesis of salicylic acid (2-hydroxybenzoic acid) or 2,3-dihydroxybenzoic acid; these compounds are then incorporated into siderophores such as pyochelin (Pseudomonas aeruginosa (6)) or enterobactin (E. coli (7)), respectively. In P. aeruginosa, ICS (PchA) and isochorismate pyruvate lyase (PchB) are required for the synthesis of SA from chorismate (6), whereas Y. enterocolitica Irp9, a bifunctional SA synthase (SAS) (8,9), directly converts chorismate to SA via an isochorismate intermediate. Importantly, the monofunctional ICS enzyme Eco EntC and the bifunctional SAS Yec Irp9 exhibit a high degree of structural similarity and a highly conserved active site suggesting that they are evolutionarily related (10). On the other hand, as shown in Fig. 1, bacterial synthesis of menaquinones occurs via the intermediates ortho-succinyl benzoate and 1,4-dihydroxy-2-naphthoate (NA) (11). Although bi-(or multi-) functional bacterial ICS enzymes associated with synthesis of ortho-succinyl benzoate have not been reported, eubacterial genomes contain conserved colocalized MenF (ICS) and MenD (2-succinyl-6-hydroxy-2,4-cyclohexadiene-1-carboxylate synthase) genes (12) to channel IC to the menaquinone pathway (e.g. Ref. 2).
Plant ICS enzymes have been predicted to be plastid-localized based on 1) the presence of a predicted choroplast transit sequence (e.g. Refs. 17 and 19) and 2) the fact that the substrate Chorismate-derived products include primary metabolites such as the aromatic amino acids, and numerous specialized metabolites, including siderophores and phenazines in bacteria and phenylpropanoids, indole alkaloids, and anthraquinones in plants. Known chorismate-utilizing enzymes of plant and/or bacterial origin are shown in red: AS, anthranilate synthase; ADCS, 4-amino-4deoxychorismate synthase; CM, chorismate mutase; CPL, chorismate pyruvate lyase. In Y. enterocolitica, SA is synthesized directly from chorismate by Irp9 an SAS, whereas, in P. aeruginosa, the isochorismate pyruvate lyase PchB is required to convert isochorismate to SA and pyruvate (not shown). IPL (in blue), isochorismate pyruvate lyase.
chorismate is synthesized in the plastid and largely localized to the plastid (1,20,21); however, experimental evidence supporting these predictions has been lacking. Past biochemical characterizations of plant ICS enzymes utilized ICS (partially) purified from elicited cell suspension cultures of Galium mollugo L (22,23), Rubia tinctorum (24), or Catharanthus roseus (19), because these systems provided sufficient activity, protein, and product. Elicited cell suspension cultures have also been used to investigate the regulation of IC-derived products such as anthraquinones produced by Morinda citrifolia (25). However, these species were not genetically tractable, and thus detailed knowledge of the genes and enzymes involved in the synthesis of IC-derived plant products has been limited. Using the genetic/genomic model plant A. thaliana, we provided the first genetic evidence for an operational ICS pathway in plants, SA biosynthesis via AtICS1 (17), and genetic evidence for PhQ biosynthesis from IC has now been reported (12).
Here, we focus on the enzymology of AtICS1, a gene/enzyme required for pathogen-induced SA biosynthesis (17). SA is best known for its role as a key regulator of plant defense against pathogens. It is synthesized and accumulates primarily as a glucose conjugate (SAG) in response to viral, bacterial, and fungal pathogens (26) and is required for the induction of hundreds of defense-related genes (e.g. pathogenesis-related PR1), and the establishment of local and systemic acquired resistance responses (26,27). In A. thaliana, null mutations in the AtICS1 gene abrogated induced SA and SAG accumulation and associated defensive responses (17,28,29). Over the past few years, a more general role for SA as a mediator of diverse stress responses has been emerging. For example, in Arabidopsis SA is synthesized in response to abiotic stresses such as UV-C (30), ozone (31), cold (32), and heat (33,34) and has been shown to play a role in stress-induced developmental transitions, including flowering (35) and senescence (36). Although AtICS1 has been confirmed to be involved in many of these processes, it has only been implicated in others, such as cold-tolerant growth.
Despite the importance of SA, our understanding of the underlying enzymology is very limited. Here, we have undertaken experiments to fill this knowledge gap and present the first biochemical characterization of a plant ICS enzyme involved in SA biosynthesis (AtICS1), including quantitative assessment of its kinetic parameters and subcellular localization. This is particularly critical, because past biochemical studies did not measure nor account for the fact that ICS enzymes typically operate near equilibrium. Here we address this reversibility directly and provide the first accurate and detailed kinetic and thermodynamic properties for a plant ICS. Detailed knowledge of the nature of the ICS reaction, and its affinity for chorismate, as well as its subcellular localization, allows us to 1) examine positive selection for monofunctional ICS versus bifunctional SAS in higher plants, 2) assess the influence of environmental factors such as light and temperature on AtICS1 activity, SA biosynthesis, and function, and 3) understand controls over chorismate partitioning and utilization. A. thaliana contains multiple genes encoding the chorismate-utilizing enzymes chorismate mutase (including AtCM1, AtCM2, and AtCM3 (37,38)), anthranilate synthase (e.g. ␣ subunit: AtASA1 and AtASA2; ␤ subunit AtASB1 and AtASB2 (39)), and ICS (AtICS1 and AtICS2 (17)). Although some of these genes are constitutively expressed, a subset, including AtICS1, AtCM1, and AtASA1, is induced in response to pathogen treatment (e.g. Refs. 17,[37][38][39]; therefore, knowledge of the biochemical properties of AtICS1 allows one to assess its ability to successfully compete for available chorismate. Because it is estimated that 20% of carbon fixed by plants flows through the shikimate pathway to chorismate under normal growth conditions with the bulk of this fixed carbon utilized in the synthesis of specialized metabolites (40), detailed knowledge of chorismate partitioning and utilization is essential to our understanding of plant fitness.

Materials and General Protocols
All specialty reagents and chemicals were obtained from Sigma-Aldrich unless otherwise specified. HPLC-grade solvents (EMD Biosciences) were employed in the HPLC analyses. Chorismic acid (Sigma C-1761, Ն80% purity) was used in all assays with the following exception: barium chorismate (Sigma C-1259, 60 -80% purity) was used to assess recombinant AtICS1 activity during overexpression and purification and for determining the temperature-dependence of AtICS1 activity. For selection and growth of transformed cells (described below): pBAD33 derivatives were selected with 30 g/ml chloramphenicol; pET-28 derivatives were selected with 50 g/ml kanamycin; and pME3368 was selected with 100 g/ml ampicillin. Commonly utilized protein and molecular biological reagents and protocols were prepared/performed as described in (41). Independent replicate experiments were performed for all experiments described under "Results," with similar findings.

Chloroplast Import Assay
Radiolabeled precursors were prepared using the TNT T7 Coupled Reticulocyte Lysate System (Promega) and L-[ 35 S]methionine (Amersham Biosciences) with pSM156 -29 (fulllength AtICS1 cDNA (AY056055) cloned into pCR-Blunt II-TOPO) or plasmids containing Toc75 (42) or Hsp93 (43) as controls. Chloroplasts were isolated from 12-day-old pea seedlings as described previously (44). Import reactions were performed as in a previous study (42). In brief, chloroplasts (12.5 g of chlorophyll) were incubated with radiolabeled precursor in a final volume of 50 l of import buffer (50 mM Hepes-KOH, 330 mM sorbitol, pH 8.0) with 3 mM Mg-ATP at room temperature in the light for 20 min. Intact chloroplasts were re-isolated by 40% Percoll and washed once with the import buffer. For protease treatment, the intact chloroplasts that contained the imported proteins were resuspended into 100 l of the import buffer with or without 1.25 g of trypsin and incubated on ice for 30 min in the dark; trypsin inhibitor (1.25 g/100 l of import buffer) was then added, and intact chloroplasts were re-isolated as above. The chloroplasts were then resuspended with hypotonic lysis buffer (10 mM Hepes-KOH, pH 8.0, 10 mM MgCl 2 ) and centrifuged at 16,000 ϫ g at 4°C for 10 min to obtain the soluble and membrane fractions. To precipitate protein in the soluble fraction, 4ϫ volume of cold acetone was added, followed by incubation for 1 h at Ϫ20°C, and centrifugation at 16,000 ϫ g at 4°C for 20 min. Each fraction was resus-pended in sample buffer and analyzed by SDS-PAGE, followed by fluorography.

Immunolocalization of AtICS1
Arabidopsis AtICS1-V5 ics1 Transgenic Line-The A. thaliana mutant eds16 -1 (ics1-2) was stably transformed with AtICS1C-terminalV5-hexaHis under control of the native AtICS1 promoter. Agrobacterium-mediated transformation was performed using pCAMBIA3301 with 35S promoter, and GUS reporter was removed and replaced with the AtICS1 2.5-kb native promoter and genomic sequence with a V5-hexa-His (Invitrogen) C-terminal tag. Glufosinate-tolerant selection was performed using Finale, and surviving plants were selfed and tested for resistance and presence of the insert (assessed using PCR). The AtICS1-V5 ics1 transgenic line (SM156) is homozygous, and induced SA accumulation and PR1 expression are restored.
Immunoreaction-The immunoreaction protocol was modified from a previous study (46). Slides (prepared above) were immersed in 0.05% Triton X-100 in 1ϫ PBS, incubated at room temperature for 15 min, and rinsed by filling chamber with 1ϫ PBS and incubating at room temperature for 10 min. Slides were then placed in a moisture chamber (wetted filter paper placed in a Petri dish) for blocking and immunoreaction. For blocking, each squashed cell area was covered with 40 l of TBSA (5% BSA in 1ϫ PBS with 0.01% Tween 20), and incubated at 37°C for 20 min. 10 l of 1/100 dilution (in TBSA) mouse anti-V5 antibody (final concentration 1/500, Invitrogen) was added to the blocking solution. The moisture chamber was sealed, covered with aluminum, and incubated at 4°C overnight. Slides were then rinsed with 1% BSA in PBS at room temperature for 15 min, placed in a moisture chamber, and incubated with 40 l of 5% BSA in PBS at 37°C for 30 min. 10 l of 2°Ab (anti-mouse goat antibody conjugated to Alexa 488 fluorophore (Invitrogen) diluted in TBSA) was added (final concentration, 1/1000), and slides were incubated in a moisture chamber at 37°C for 1 h. After washing with 1ϫ PBS at room temperature for 15 min, material on glass slides was counterstained with 20 l of 1 g/ml 4Ј,6-diamidino-2-phenylindole (Sigma) diluted from 200 g/ml stock in TAN buffer (0.5 mM EDTA, 20 mM Tris, pH 7.5, 1.2 mM spermidine, 0.05% v/v 2-mercaptoethanol) to visualize double-stranded DNA and mounted with a drop of ProLong Antifade (Invitrogen). Pictures were taken with Axiophot 381 (Zeiss) with: excitation band pass 450 -490 and emission long pass 520 to visualize Alexa 488 fluorophore, excitation band pass 365/20 and emission long pass 397 to observe 4Ј,6-diamidino-2-phenylindole staining, and excitation band pass 535-555 and emission long pass 590 to visualize chlorophyll autofluorescence.

Cloning, Expression, and Purification of Recombinant AtICS1
The AtICS1 coding region (without the chloroplast transit sequence) was PCR-amplified from A. thaliana ecotype Columbia-O cDNA isolated from induced leaves (forward primer: 5Ј-atcgtcgacccatatgaatggttgtgatgga-3Ј; reverse primer: 5Ј-atcgtcgactcaattaatcgcctgtagaga-3Ј) and inserted as an NdeI/ SalI fragment into the NdeI/XhoI sites of pET-28c (Novagen). The resulting construct, pSM157-16, contains an N-terminal histidine tag fused to amino acid 48 of the ICS1 coding region. Construction of the recombinant precursor protein (containing the chloroplast transit sequence) expression vector used 5Јatcgtcgacccatatggcttcacttcaattttc-3Ј as the forward primer; the resulting construct, pSM159-18, contains an N-terminal histidine tag fused to the start codon. The pSM157-16 and pSM159-18 AtICS1 coding sequences were confirmed to be identical to AtICS1 sequence AY056055 (At1g74710.1).
E. coli Rosetta2 (DE3) cells (Novagen) were transformed with these plasmids. Crude cell extracts were prepared from a 2 liter culture of transformed cells in TB media containing 0.2% glucose, 50 g/ml kanamycin, and 30 g/ml chloramphenicol. Cultures were grown at 37°C to mid-log phase, 0.2 mM isopropyl-␤-D-thiogalactopyranoside was added to induce His-ICS1 synthesis, and cells were harvested after 18 h at 21°C (ϳ30 g wet weight) and resuspended in 150 ml of buffer A (20 mM sodium phosphate buffer, pH 7.4, 500 mM sodium chloride, 10% glycerol) containing 1 mM dithiothreitol (DTT), 2 mM phenylmethanesulfonyl fluoride, 5 M leupeptin, 10 g/ml DNase, and 1% Triton X-100. The cells, which could be stored at Ϫ20°C, were lysed by two passages through a French press at 18,000 p.s.i. Following centrifugation, the supernatant was filtered by syringe through an HPF Millex-HV 0.45-m filter unit attached in series to a Millex-AP prefilter (Millipore). Filtrate was applied to a 1-ml HisTrap HP nickel affinity column (Amersham Biosciences) at 1 ml/min by use of an Á KTA fast protein liquid chromatography system (Amersham Biosciences). After washing column with 10 ml of buffer A, His-AtICS1 was eluted with 40 ml of a linear gradient from 0% to 15% at 1.0 ml/min of buffer A containing 500 mM imidazole. Fractions containing ICS activity were pooled and concentrated using an Amicon Ultra-15 (10-kDa molecular mass cutoff) ultrafiltration device (Millipore) to a final volume of 900 l. The pooled solution was applied to a HiPrep 16/60 Sephacryl S-200 High Resolution gel filtration column (Amersham Biosciences) previously equilibrated with 200 ml of buffer B (50 mM sodium phosphate buffer, pH 8.0, 150 mM sodium chloride, 10% glycerol, and 1 mM DTT). The enzyme was eluted at a flow rate of 1.0 ml/min as a broad peak with 60 ml of buffer B. Fractions containing ICS activity were pooled and concentrated as before to a final volume of 2 ml. The concentrate was dialyzed overnight into buffer C (100 mM Tris buffer, pH 7.7, 10% glycerol, and 1 mM DTT) with or without 10 mM MgCl 2 as desired. The protein was aliquoted and stored at Ϫ80°C. All further characterization was performed on the mature AtICS1 (without the chloroplast transit sequence) recombinant-purified protein.
Crude cell extracts were prepared from a 2-liter culture of E. coli Rosetta2 (DE3) cells transformed with pSM147-1 grown in TB media containing 0.2% glucose, 50 g/ml kanamycin, and 30 g/ml chloramphenicol. Cultures were grown at 37°C to mid-log phase, 1 mM isopropyl-␤-D-thiogalactopyranoside was added to induce His-PchB synthesis, and cells were harvested after 4 h (ϳ25 g wet weight) and stored overnight at Ϫ20°C. His-PchB was then purified using nickel-nitrilotriacetic acid His-Bind Resin (Novagen) according to the manufacturer's directions. Aliquots of the purified recombinant PchB protein (54 mg/ml in 100 mM Tris, pH 7.7, 10% glycerol, 1 mM DTT) were stored at Ϫ80°C.

Determination of Protein Concentration
Protein concentrations were determined by the method of Bradford modified for use in a 96-well plate format with Coomassie Blue G-250 (EM Biosciences) and analyzed using a Spectramax Plus microplate spectrophotometer (Molecular Devices) with bovine serum albumin as the standard.

Molecular Mass Estimation
The subunit molecular mass of AtICS1 was estimated by SDS-PAGE in a 10% gel with Precision Plus unstained protein standards (Bio-Rad). The native molecular mass was estimated by gel filtration chromatography on a HiPrep 16/60 Sephacryl S-200 High Resolution column in buffer B (above) at 1.0 ml/min, using thyroglobulin (670 kDa), ␥-globulin (158 kDa), ovalbumin (44 kDa), myoglobin (17 kDa), and vitamin B 12 (1.35 kDa) as markers (Bio-Rad Gel Filtration Standards). The apparent molecular mass of AtICS1 was determined from a plot of the elution volumes against the logarithm of the molecular masses.

ICS Activity Assays
HPLC ICS Activity Assay-This assay was modified from a previous study (19). 40 l of substrate solution (2 mM chorismic acid in buffer D: 100 mM Tris, pH 7.7, 10% glycerol, 10 mM MgCl 2 , 1 mM DTT) was added to 40 l of a 220 g/ml solution of AtICS1 in buffer D, incubated for 60 min at 30°C, and immediately filtered through a 0.2-m Millex-LG syringe filter (Millipore), and a 50-l aliquot was injected into a Shimadzu SCL-10AVP series HPLC system equipped with a Shimadzu SPD-10AVP photodiode array detector and a Shimadzu RF-10AXL fluorescence detector. A 5-m, 15-cm ϫ 4.6-mm inner diameter Supelcosil LC-ABZPlus column (Supelco) preceded by a LC-ABZPlus guard column was maintained at 27°C and previously equilibrated in 15% acetonitrile with 25 mM potassium phosphate buffer, pH 2.5, at a flow rate of 1.0 ml/min. The elution program began with an isocratic flow of 15% acetonitrile with 25 mM potassium phosphate buffer, pH 2.5, for 1 min, followed by a linear increase to 20% acetonitrile over 7 min. Prior to injecting subsequent samples, a linear decrease to 15% acetonitrile over 2 min was undertaken, followed by re-equilibration at 15% acetonitrile for at least 5 min. Under these conditions, isochorismic acid eluted (A 280 ) at ϳ3.4 min and chorismic acid (A 280 ) at 4.2 min. The calibration curve for chorismic acid is as follows: y ϭ 0.00386319x Ϫ 11.2600 with R 2 Ͼ 0.999, where x ϭ area units and y ϭ chorismic acid in nanograms.
Coupled HPLC Assay for ICS Activity-This assay was modified from a previous study (48). 437.5 l of a solution of AtICS1 (3.8 g/ml) and PchB in excess (218 g/ml) in buffer E (100 mM potassium phosphate buffer, pH 7.0, 15 mM MgCl 2 , 10% glycerol, and 1 mM DTT) was mixed with 62.5 l of substrate solution (4 mM chorismic acid in buffer E) and incubated for 60 min at 30°C. The reaction was filtered as above, and a 50-l aliquot was injected into the HPLC system with column described above. The elution program began with an isocratic flow of 15% acetonitrile with 25 mM potassium phosphate buffer, pH 2.5, for 1 min, followed by a linear increase to 20% acetonitrile over 5 min, isocratic flow at 20% for 10 min, a linear increase from 20 to 55% acetonitrile over 17.5 min, a linear increase from 55 to 66% over 5 min, and an isocratic flow at 66% for 1.5 min. Prior to injecting subsequent samples, a linear decrease to 15% acetonitrile over 5 min was undertaken, followed by re-equilibration at 15% acetonitrile for at least 5 min. This program allowed for the detection of isochorismic acid (A 280 , 3.7 min), chorismic acid (A 280 , 4.9 min), and salicylic acid (ex 305 nm/em 407 nm, 22.9 min). The calibration curve was: y ϭ 0.000231406x Ϫ 2.81054 with R 2 ϭ 0.999, where x ϭ area units, y ϭ ng SA. Salicylic acid (S-5922, Sigma ultrapure) was used for calibrations.
Coupled Spectrophotometric Assay For ICS Activity-The ICS reaction rate was measured by coupling excess amounts of isochorismate pyruvate lyase (recombinant PchB) and lactic dehydrogenase (Sigma L-1254), which catalyzes the NADH-dependent conversion of pyruvate to lactate. Dilution series of AtICS1 were performed to establish the linear range for this assay (to ϳ25 g/ml recombinant, purified AtICS1) and our standard assay conditions (10 g/ml recombinant purified AtICS1). Velocity was linear with time (following an initial lag) for Ͼ15 min with 10 g/ml AtICS1. Unless otherwise indicated, the 200 l per well assay volume contained 0.2 mM NADH, 0.833 g/ml L-lactic dehydrogenase, 32.0 g/ml PchB, 10.0 g/ml AtICS1, and 2 mM chorismic acid in buffer D. The reaction was initiated by addition of the chorismic acid to the reaction mixture in a 96-well plate preheated to 30°C and analyzed using a Spectramax Plus microplate spectrophotometer FEBRUARY 23, 2007 • VOLUME 282 • NUMBER 8 (Molecular Devices). The change in absorbance at 340 nm was measured in each well in increments of 30 or 60 s and monitored at least 30 min. The initial reaction rate in each well was assessed for a 5-to 10-min period ϳ7 min after initiation of the reaction based on manual confirmation of linear range and calculated using least squares fitting of each curve. An extinction coefficient for NADH of 6220 M Ϫ1 cm Ϫ1 was used for conversion of these values to units of micromolar/min.

Determination of K eq Using 1 H NMR
Our protocol is similar to that previously used with E. coli EntC (4). The spectrum of the equilibrium mixture of chorismate and IC was acquired as follows: 500 l of 220 g/ml AtICS1 exchanged into D 2 O buffer containing 50 mM potassium phosphate, pD 7.5, 5 mM MgCl 2 was incubated with 500 l of chorismic acid solution (2 mM in D 2 O buffer) at 30°C for 60 min. (HPLC analyses of successive aliquots confirmed the reaction was in equilibrium.) NMR spectra for a 900-l aliquot were recorded using a 500-MHz Bruker DRX-500 spectrometer. Spectra were scanned every 8 s for a total period of 9 h each. The ratio between chorismate and IC was determined by integration of the two peaks most downfield in the spectrum (the C-2 protons). The chorismate spectra were obtained with chorismic acid in the above D 2 O buffer.

K m Determination
The coupled spectrophotometric ICS assay was modified by using working chorismic acid concentrations of 1. Reactions without AtICS1 were used as blanks. As per standard protocol, triplicate samples were run for each condition. Initial velocity data were fitted to the equation of Hanes to determine kinetic parameters (49). To determine the effective chorismate concentrations in the above assay, we assessed the conversion of chorismate to prephenate via PchB in parallel. In brief, as modified from a previous study (38), standard coupled reactions in the 96-well plate were treated with acid to convert prephenate to phenyl pyruvate after the reaction progressed for 0, 5, 10, 15, or 20 min, followed by neutralization with base, and measurement at A 320 . Prephenate standard curves were run in parallel for quantification. Chorismate utilization via PchB was estimated as equal on a molar basis to prephenate.

Effect of Mg 2؉ on ICS Activity
The coupled spectrophotometric ICS assay was modified by using buffer C (no Mg 2ϩ ) supplemented with MgCl 2 at the fol- Note that neither isochorismate pyruvate lyase (IPL) (50) nor lactic dehydrogenase activities were impacted by Mg 2ϩ . Initial velocity data were fitted to the equation of Hanes to determine kinetic parameters (49).

Effect of Other Metals on ICS Activity
The coupled spectrophotometric assay was modified by substituting other divalent metals for Mg 2ϩ . As above, buffer C was used and a working concentration of 10 mM, 1 mM, and 0.1 mM of each of the following divalent metals was substituted: CaCl 2 , BaCl 2 , MnCl 2 , ZnCl 2 , and CdCl 2 . Each reaction was performed in triplicate. Reactions without AtICS1 were used as blanks. Reactions containing no added metal ion were also included. IPL activity is reported to be unaffected by divalent cations (48). However, lactic dehydrogenase (Sigma L-1254, rabbit muscle) activity may be inhibited by heavy metals such as Hg 2ϩ and Pb 2ϩ (Sigma-Aldrich Technical Service). Therefore, we repeated the metal experiments using 1 mM of the divalent metal ion and the HPLC ICS assay, which directly measures the conversion of chorismate to IC. For metals with significant absorbance at 340 nm, we exclusively employed the HPLC ICS assay. Buffer C was substituted for buffer D, and the reactions were supplemented with each of the following metals: FeCl 2 , FeCl 3 , CoCl 2 , NiCl 2 , or CuCl 2 . Working concentrations of 10 mM, 1 mM, and 0.1 mM of each were used in the reactions.

pH Profile of ICS Activity
The HPLC assay for ICS activity was modified by changing the pH of the reaction buffer. 40 l of 100 mM buffer solution containing 220 g/ml AtICS1 was mixed with 40 l of a 2 mM solution of chorismic acid. Buffer solutions utilized 100 mM MES at pH 5.0, 5.5, 6.0, and 6.5; MOPS at pH 7.0 and 7.5; Tris at pH 7.5, 7.7, 8.0, and 8.5; CHES at pH 9.0, 9.5, and 10.0; or CAPS at pH 11.0. Reactions were performed in triplicate with results reported for the formation of IC. Reactions without AtICS1 were used as controls (to be subtracted from results with enzyme), although we observed no non-enzymatic production of isochorismate.

Temperature Profile of ICS Activity
The HPLC assay for ICS activity was modified by changing the temperature at which the reaction was incubated. Reactions were performed in triplicate incubating for 60 min at each of the following temperatures: 4°C, 15°C, 23°C, 30°C, 37°C, 44°C, 51°C, 60°C, and 70°C. Results are reported for the formation of isochorismate. Non-enzymatic production of IC was not observed. At 51°C and above, there was some decomposition of chorismate (to products other than IC); however, [chorismate] was not limiting.

Non-enzymatic Synthesis of SA from Isochorismate
IC was produced enzymatically by conversion of chorismate to isochorismate via ICS (purified recombinant Eco EntC) and purified as in a previous study (51). IC was incubated in solution (37 M IC, pH 7.5, 3 mM Mg 2ϩ , 23°C) for 1 h, 3 h, and 6 h, and chorismate, IC, and SA were analyzed by HPLC (using method of coupled HPLC ICS assay above). Incubations were performed in triplicate with the conversion rate of IC to SA (micromolar SA/h) as the slope of SA production with time (y ϭ 0.036x ϩ 1.40; R 2 ϭ 0.999).

AtICS1 Functional in Stress-induced Salicylate Biosynthesis Homology Modeling of ICS Enzymes
The primary sequences for A. thaliana AtICS1 (AAL17715) and Y. enterocolitica Irp9 (CAB46570) were aligned with Clust-alW (52) using the default parameters. Due to poor alignment of the N-terminal portion of the proteins, residues 1-198 and 1-92 of AtICS1 and Yec Irp9, respectively, were not used in model generation. Note that AtICS1 contains an N-terminal chloroplast transit sequence of ϳ45 amino acids as predicted by ChloroP (53) and confirmed via chloroplast import assays (Fig.  2). Homology modeling was performed with Modeler 8v2 (54) using the Yec Irp9 crystal structure in complex with its reaction products SA and pyruvate (Protein Data Bank identification code 2FN1) (10). Five models were generated, and the lowest energy model was selected. The AtICS1 model and Yec Irp9 were superimposed in COOT (55). All figures were generated with PyMOL. 4 Homology modeling of additional ICS enzymes, including Mycobacterium tuberculosis MbtI (CAE55483), Pae PchA (CAA57969), Eco EntC (POAEJ2), AtICS2 (NP_173321), C. roseus ICS (CAA06837), and Capsicum annum ICS (AAW66457) was performed in a similar manner.

RESULTS
AtICS1 Is Plastid-localized-We performed chloroplast import assays to determine whether AtICS1 is capable of being imported into the chloroplast and where within the chloroplast it resides. Immunolocalization experiments using epitopetagged AtICS1 transgenics expressed under control of the AtICS1 native promoter were then undertaken to assess whether AtICS1 is plastid-localized in planta in response to pathogens. We found radiolabeled AtICS1 precursor protein (ϳ62 kDa) was imported into the chloroplast and processed to its mature form (ϳ58 kDa) ( Fig. 2A). Chloroplasts containing the imported protein were then exposed to trypsin treatment to distinguish stromal proteins from those residing in the intermembrane space. Mature AtICS1 was trypsin-resistant indicative of a stromal protein. Parallel assays were performed using Toc75 and Hsp93 as controls for sensitivity and lack of sensitivity to trypsin digestion, respectively. Consistent with previous results (42), Toc75, a chloroplast outer envelope protein, was recovered in the membrane fraction (pellet) and was susceptible to trypsin treatment (Fig. 2B), whereas Hsp93, a stromal protein, was recovered in the soluble fraction and was trypsin insensitive (data not shown).
Furthermore, AtICS1 was detected in plastids of pathogeninduced mature leaves of Arabidopsis AtICS1-V5 ics1 transgenics (SM156) by immunofluorescence (Fig. 3). For these ics1 transgenics, AtICS1-V5 expressed under control of the native AtICS1 promoter restored pathogen-induced SA biosynthesis, and induction of SA-dependent PR1 expression. 5 The size, shape, chlorophyll autofluorescence, and 4Ј,6-diamidino-2phenylindole-staining of double-stranded DNA of the plastid shown in Fig. 3 are hallmarks of mesophyll chloroplasts. Fluorescence of the Alexa 488 fluorophore (conjugated to the 2°a ntibody) was observed in chloroplasts of the transgenic line SM156 but not in wild-type Columbia-O or ics1 plants (not shown). Alexa 488 fluorescence was also absent in chloroplasts FIGURE 2. AtICS1 is targeted to the chloroplast stroma and processed to its mature form. The 35 S-radiolabeled AtICS1 (A) and Toc75 (B) precursor proteins (tl, 10% input) were incubated with intact chloroplasts under the import condition. The chloroplasts were re-isolated and analyzed directly (imp), or incubated in import buffer without (Ϫ) or with (ϩ) trypsin on ice in the dark for 30 min followed by the addition of trypsin inhibitor and hypotonic lysis before being fractionated into the supernatant (s) and the pellet (p) by centrifugation. Radiolabeled proteins were separated by SDS-PAGE and visualized by fluorography. AtICS1 precursor protein (indicated with an asterisk) was processed to its mature form (mICS1). The precursor to the outer envelope protein Toc75 (indicated with an asterisk) was processed to an intermediate (i75) and mature (m75) form.  FEBRUARY 23, 2007 • VOLUME 282 • NUMBER 8

JOURNAL OF BIOLOGICAL CHEMISTRY 5925
of SM156 transgenics when the primary anti-V5 antibody was excluded from the immunoreaction. AtICS1 Exhibits ICS Activity Not SA Synthase Activity-In bacteria, the synthesis of SA via IC occurs either via a bifunctional SAS (e.g. Yec Irp9 (9)) or an enzyme complex consisting of ICS and IPL genes coexpressed and present in cis (e.g. Pae PchBA (6,48)). ICS and SAS enzymes share similar global structure and highly conserved active sites (9, 10); therefore, it was important for us to determine whether AtICS1 exhibits monofunctional ICS activity or bifunctional SAS activity. Because AtICS1 expression and function are correlated with induced SA accumulation (see the introduction and Refs. 57 and 58) and IC-pathways likely derive from bacterial endosymbiosis (e.g. Ref. 12), we thought it quite possible that AtICS1 could act as a bifunctional SAS.
To assess AtICS1 biochemical activity, we overexpressed N-terminal hexaHis-tagged mature AtICS1 (without the chloroplast transit sequence) in E. coli and purified the soluble induced protein (see supplemental data). For a typical preparation, AtICS1 was purified 69-fold from induced cell extracts to near homogeneity. The specific activity of the recombinant purified AtICS1 was 241 nmol min Ϫ1 mg Ϫ1 (4,010 picokatals mg Ϫ1 ). All further biochemical characterization (below) was performed on the purified, recombinant mature AtICS1 enzyme.
Our HPLC analyses showed that AtICS1 converts chorismate to IC (Fig. 4A). Isochorismate was confirmed by 1 H NMR. Salicylic acid was not detected as a product of this reaction (Fig.  4B, top). The addition of recombinant Pae PchB (an IPL) to the AtICS1 reaction did result in the production of SA (Fig. 4B,  bottom). Therefore, AtICS1 functions as a traditional monofunctional ICS and not a bifunctional SAS like Irp9 (9).
AtICS1 Is an Active Monomer-To determine whether AtICS1 likely exists as a monomer, dimer, or other multimer, we estimated the molecular mass of the recombinant purified enzyme by fast protein liquid chromatography using a calibrated Sephacryl S-200 gel filtration column. Active AtICS1 was estimated to have a molecular mass of 57.5 kDa. SDS-PAGE performed on eluted fractions indicated that the ICS monomer (estimated to be ϳ59 kDa, supplemental data) was exclusively enriched in the fractions with ICS activity. Therefore, similar to the bacterial ICS proteins Pae PchA (48) and Eco EntC (4), AtICS1 appears to function as a monomer, whereas the SAS Yec Irp9 appears to function as a dimer (9).  examined the reversibility of the reaction and had not considered this reversibility when apparent K m values were determined (19,23,24). This is critical as the previously employed ICS equilibrium assays are not suitable for obtaining accurate kinetic parameters for enzymes such as ICS that operate near equilibrium (59). Therefore, for AtICS1 we experimentally determined K eq and calculated quantitative kinetic parameters using a coupled irreversible continuous assay similar to those employed in the characterization of Pae PchA (48) and Eco EntC (4).

AtICS1 Catalyzes a Reversible Reaction and Exhibits an
To determine the equilibrium constant (K eq ) for recombinant AtICS1, we followed the conversion of chorismate to IC by HPLC and obtained 1 H NMR spectra for the equilibrium mixture (Fig. 5). The ratio between chorismate and IC at equilibrium was calculated from the integration of the unique olefinic protons in each compound (i.e. the C-2 protons) as shown in Fig. 5B. We obtained a K eq for AtICS1 of 0.89 Ϯ 0.02, showing that a plant ICS (AtICS1) does operate near equilibrium, with the reaction slightly favoring chorismate.
To calculate the apparent K m of AtICS1 for chorismate, we developed a coupled assay in which IC (produced via AtICS1) is converted irreversibly to SA and pyruvate by the IPL Pae PchB (in excess), similar to the protocol used to assess the catalytic properties of PchA (48). To facilitate kinetic measurements, lactate dehydrogenase is included (in excess) to convert pyruvate to lactate in an NADH-dependent reaction, and the decrease in A 340 associated with the conversion of NADH to NAD ϩ is followed by spectroscopy. We found AtICS1 exhibited standard Michaelis-Menten kinetics for chorismate. The corresponding Hanes plot yielded an apparent K m ϭ 84.2 Ϯ 3.9 M for chorismate, with V max ϭ 5.84 Ϯ 1.14 M min Ϫ1 and k cat ϭ 34.7 Ϯ 6.8 min Ϫ1 . However, PchB can also utilize chorismate as a substrate (producing prephenate) but with lower affinity for chorismate (K m ϭ 150 M) than for isochorismate (K m ϭ 14 M) (50). Because the affinity of PchB for chorismate at optimal conditions (pH 7.0, 37°C) is similar to that of AtICS1 (above, assessed at pH 7.7, 30°C), we wanted to determine whether PchB was reducing the effective concentration of chorismate in our assay. Therefore, we assessed effective chorismate concentrations for our K m experiment by measuring prephenate accumulation in the coupled spectrophotometric ICS assay in parallel. We found significant conversion of chorismate to prephenate at concentrations below 200 M chorismate. The adjusted kinetic parameters calculated using the effective chorismate concentrations were: K m ϭ 41.5 M, V max ϭ 6.50 M min Ϫ1 , and k cat ϭ 38.7 min Ϫ1 . The K m of AtICS1 for chorismate is ϳ10-fold lower than past reports for plant ICS enzymes, which employed inaccurate equilibrium assays (19,23,24) and ϳ10-fold higher than bacterial enzymes involved in SA synthesis from chorismate via IC (Pae PchA (48) and YecIrp9 (9)).
Catalytic Properties of AtICS1-We assessed the dependence of AtICS1 activity on Mg 2ϩ , an array of other divalent cations, and Fe 3ϩ . The presence of Mg 2ϩ was found to be an absolute and specific requirement for AtICS1 activity, similar to other plant and bacterial ICS enzymes (e.g. Refs. 4,19,23,48,and 60). This requirement is consistent with the dominant proposed reaction mechanism for ICS: regiospecific 1,5-S N 2Љ addition of the nucleophile water to C2 of chorismate via a Mg 2ϩ -bound transition state with concomitant loss of the hydroxyl group at C4 (61,62). A typical saturation curve was obtained for AtICS1 activity as a function of Mg 2ϩ concentration (assessed with 2 mM chorismate) with saturation occurring at ϳ 2 mM Mg 2ϩ (Fig. 6A), and a K m of 193 M for Mg 2ϩ was obtained. Incubation with other divalent cations, including Mn 2ϩ , Co 2ϩ , Ni 2ϩ , Ca 2ϩ , Zn 2ϩ , Ba 2ϩ , Cu 2ϩ , Cd 2ϩ , and Fe 2ϩ , and with Fe 3ϩ at 0.1, 1, or 10 mM, did not result in significant AtICS1 activity.
The pH optimum for recombinant AtICS1 was determined using the range of pH 5-11. As shown in Fig. 6B, AtICS1 has a broad pH range of maximal activity from 7 to 9.5, with optimal AtICS1 activity from pH 7.5 to 8. A similar range of maximal activity was reported for other plant ICS enzymes (19,23,24), although values at each pH were not provided; pH optima for bacterial ICS enzymes of 7.0 (48) and 7.5-8.0 (5) have been reported.
The temperature dependence of AtICS1 activity was assessed from 4 to 70°C with maximal activity at room temperature (23°C) and no activity at 70°C (Fig. 6C). AtICS1 exhibited a surprisingly broad range of activity with Ͼ75% maximal activity from 4 to 44°C, and Ͼ90% of maximal activity from 4 to 37°C. Extensive temperature profiles of plant ICS enzymes had not been previously reported (19,23,24), although similar optima had been found (e.g. G. mollugo (23)). Where reported, bacterial ICS enzymes exhibited optimal activity at 37°C (e.g.

Ref. 5).
Non-enzymatic Synthesis of SA from Isochorismate Is Negligible-Non-enzymatic conversion of IC to SA (ϳ25% conversion, 100°C, 10 min) has long been reported (60,63). Therefore, using our findings (above) regarding the subcellular localization and biochemical properties of AtICS1, we directly examined the potential for non-enzymatic formation of SA from IC (produced via AtICS1) under physiological conditions. We first calculated a reasonable plastidic IC pool. We assumed that AtICS1 operates near its K m (41.5 M for chorismate); therefore, the plastidic [chorismate] is taken to be 41.5 M, and a plastidic IC pool of 37 M was estimated (K eq ϭ 0.89). We then assessed non-enzymatic conversion of IC to SA using conditions consistent with the chloroplast stroma (37 M IC, pH 7.5, 3 mM Mg 2ϩ , 23°C). Very little non-enzymatic conversion of IC to SA was observed: Ͻ0.1% IC converted to SA/h ("Experimental Procedures"). This conversion rate is consistent with a recent IC thermal decomposition study (see calculation at 30°C, pH 7, 10 mM Mg 2ϩ (64)). Furthermore, we did not detect a substantial SA signature in 1 H NMR spectra of AtICS1 in equilibrium ([IC] ϳ0.47 mM, pH 7.5, 5 mM Mg 2ϩ , 22°C) acquired over a 9-h period (Fig. 5, data not shown).

DISCUSSION
Herein, we presented the first experimental evidence that a functional plant ICS enzyme, AtICS1, is localized to the plastid. We determined K eq for AtICS1 (K eq ϭ 0.89), establishing that a plant ICS enzyme operates very near equilibrium. We then provided the first kinetic data (K m , V max , and k cat ) for a plant ICS using an irreversible coupled continuous assay similar to that used to obtain accurate kinetic data for bacterial ICS enzymes (e.g. Ref. 4). Previously, K m (but not k cat ) values for plant ICS enzymes have been reported (19,23,24); however, these studies FEBRUARY 23, 2007 • VOLUME 282 • NUMBER 8

AtICS1 Functional in Stress-induced Salicylate Biosynthesis
employed equilibrium assays, which are not suitable for obtaining accurate kinetic parameters for enzymes such as ICS that operate near equilibrium (59). AtICS1 is required for the induced synthesis of SA from chorismate (17). Therefore, below, we use our detailed biochemical characterization and subcellular localization of AtICS1 to provide insights into SA biosynthesis and function as well as chorismate partitioning and utilization.
AtICS1 Is a Monofunctional ICS, Not a Bifunctional SA Synthase-Because AtICS1 expression is highly correlated with SA accumulation (see the introduction and Refs. 57 and 58), we thought it quite possible that AtICS1 could act as a bifunctional SAS. However, we found that AtICS1 does not directly convert chorismate to SA but acts as a monofunctional ICS similar to most bacterial ICS proteins (Fig. 4). Residues essential to SAS IPL activity have yet to be determined (9, 10, 65); therefore, we performed homology modeling of known monofunctional ICS enzymes involved in SA biosynthesis (AtICS1 and Pae PchA) and the recently confirmed SAS M. tuberculosis MbtI (66) using the crystal structure of the SAS Yec Irp9 in complex with its reaction products SA and pyruvate (10). As expected, our homology modeling of AtICS1 with Yec Irp9 indicates that the active site is highly conserved (Fig. 7). Comparison of aforementioned active sites leads us to speculate that Thr-348 is critical for IPL activity with potential hydrogen bonding between Thr-348 and the 2-hydroxyl group of SA (or the similarly positioned 6-hydroxyl group of isochorismate). Yec Irp9 inhibitor studies using IC mimics also support this postulation (65). Thr-348 is conserved in Irp9 and MbtI but replaced with Ala in AtICS1 and PchA. In addition, AtICS2 (the other predicted functional Arabidopsis ICS), other plant ICS proteins, and Eco EntC and MenF also contain an Ala at this position. It will be interesting to see whether site-directed mutagenesis of AtICS1 can convert it to an SA synthase.
Given that mono-and bifunctional ICS enzymes exhibit a high degree of overall structural similarity and a highly conserved active site in which few residues are likely responsible for ICS versus SAS activity (Fig. 7) (10), we would expect to observe positive selection for either monofunctional ICS or bifunctional SAS enzymes in plants. To date, only monofunctional plant ICS activities have been reported (herein and Refs. 19, 23, and 24). It should be noted, however, that previous studies characterizing plant ICS enzymes did not examine ICS enzymes associated with SA biosynthesis and did not specifically look for a product other than IC. Very recently, PHYLLO, a fusion of a non-functional, truncated 5Ј ICS and three full-length individual eubacterial genes involved in bacterial menaquinone biosynthesis, was reported to be required for PhQ production in Arabidopsis (12). The architecture of the fused PHYLLO locus is conserved in the nuclear genomes of plants and green algae, with gene fission and inactivation of the ICS module of PHYLLO occurring in higher plants. Therefore, whereas green algae encode a multifunctional enzyme involved in the conversion of chorismate to a naphthoquinone/PhQ/anthraquinone intermediate (such as 1,4-dihydroxy-2-napthoate), higher plants require an independent ICS for initial production of IC from chorismate (12) (see Fig. 1) providing strong support for positive selection for multiple monofunctional ICS enzymes in plants.
What would be the advantage of monofunctional plant ICS enzymes? First, because monofunctional ICS enzymes catalyze the reversible conversion of chorismate to IC, expression of a monofunctional ICS would not drain chorismate from other  FEBRUARY 23, 2007 • VOLUME 282 • NUMBER 8 plastid-localized chorismate-utilizing pathways such as aromatic acid biosynthesis or phenylpropanoid production. This is a valid concern because our kinetic measurements suggest that AtICS1 could potentially compete for chorismate with constitutively expressed, chorismate-utilizing enzymes associated with aromatic amino acid biosynthesis (e.g. Ref. 38). Furthermore, constitutive expression of a PchBA SAS fusion protein in Arabidopsis resulted both in a Ͼ20-fold increase in SA levels and in severe dwarfism/infertility (20). The severity of the dwarfism/infertility phenotype compared with mutants constitutively overexpressing SA is most likely due to the channeling of chorismate to SA at the expense of other essential chorismate-and IC-derived products (e.g. aromatic amino acids and PhQ, respectively), because the affinity of PchA for chorismate (48) is 10-fold higher than that of AtICS1.

AtICS1 Functional in Stress-induced Salicylate Biosynthesis
Second, expression of multiple monofunctional ICS enzymes could allow for IC to be channeled to different products (e.g. SA, induced naphthoquinones or anthraquinones, PhQ) depending upon coexpression of downstream enzymes (see Fig. 1). For example, although AtICS1 expression and function is associated with induced SA biosynthesis, it may also contribute to the synthesis of other IC-derived products (e.g. Refs. 12 and 67). Indeed, the recent examination of PhQ biosynthesis found that only the ics1ics2 double mutant but not single ics mutants resulted in a PhQ-deficient phenotype (12) suggesting that either both Arabidopsis ICS enzymes contribute IC for PhQ biosynthesis in wild-type plants or that one ICS may compensate for a deficiency in the other. In the case of SA biosynthesis, the Arabidopsis ics1 single mutant was sufficient to abolish pathogen-induced SA accumulation; however, low level endogenous SA production was unaffected (17,28). As AtICS2 is not significantly induced in response to pathogens, its expression may be insufficient to functionally complement induced SA biosynthesis; alternatively, AtICS2 may exhibit a lower affinity for chorismate than AtICS1 and be less successful in competing with other pathogen-induced chorismate-utilizing enzymes for chorismate. To determine whether AtICS1 can channel isochorismate to different products, we are examining whether AtICS1 can form complexes with different enzyme partners depending upon the developmental stage or biotic/ abiotic stressor.
Third, multiple monofunctional ICS enzymes could provide partial functional redundancy for the robust synthesis of essential compounds such as PhQ from IC. Gene duplication is a common means of facilitating the robustness of metabolic pathways (68,69). Robustness is typically defined as a measure of the ability of a biochemical network to withstand perturbations. Indeed, as mentioned above, the ics1ics2 double mutant exclusively displayed the phyllo phenotype: no PhQ, 5-15% of wild-type photosystem I activity, bleached leaves, and seedling lethality (12).
AtICS1 Activity and SA Biosynthesis as a Function of Light and Temperature-A number of SA-dependent defense-related processes are light-dependent (70); however, a detailed mechanistic understanding of this light dependence is lacking. Using chloroplast import studies and immunolocalization, we determined that AtICS1 is localized to the chloroplast stroma (Fig. 2) and is induced in response to pathogens in mesophyll chloroplasts of infected mature leaf tissue (Fig. 3). Light-dependent changes in pH, Mg 2ϩ , and redox status may regulate the activity of enzymes localized to the plastid stroma (e.g. Ref. 71). Therefore, we were interested in determining whether AtICS1 activity would be dramatically impacted by these changes. We found no evidence for substantial regulation of AtICS1 activity associated with reported light-dependent changes in stromal pH (from 7 (dark) to 8 (light)) and Mg 2ϩ (from 1-3 mM to 3-6 mM) (14). Although we did not directly examine the potential for redox regulation of AtICS1, we believe redox regulation of AtICS1 is unlikely. First, the AtICS1 protein does not contain cysteine residues within sufficient distance to form a disulfide bridge, based on homology modeling with Irp9 (above). Second, DTT was not required to obtain active recombinant mature AtICS1. DTT (or a similar thiol reductant) is typically required throughout isolation and purification protocols to retain the activity of reduced redox-activated enzymes (e.g. Ref. 72). Taken together, our findings suggest that the light-dependent regulation of AtICS1 activity due to changes in stromal pH, Mg 2ϩ , and/or redox status do not significantly contribute to the light dependence observed for SA-dependent defense responses. This supports a previous report that the observed light dependence of PR1 expression occurs downstream of SA (70).
Initially, we were surprised by the broad temperature range for AtICS1 activity (Ͼ90% of maximal activity from 4 to 37°C, Fig. 6C). However, from a defense perspective, it could benefit the plant to synthesize SA and induce plant defense responses over a wide range of naturally occurring temperatures (73). Furthermore, a more general role for SA as a mediator of temperature-dependent stress is emerging FIGURE 7. Homology modeling of AtICS1 using the crystal structure of Y. enterocolitica Irp9. A, AtICS1 (green) overlaid on the Yec Irp9 crystal structure (gray) in complex with its reaction products SA and pyruvate colored by atom with carbons in cyan and oxygens in red (PDB code 2FN1) (10). Mg 2ϩ is shown as a magenta sphere. Insertion regions of AtICS1 are colored in red. B, enlarged view of the active site with Irp9 side chains in gray and AtICS1 side chains in green. Irp9 residues are labeled with one-letter amino acid code and number for position in protein with AtICS1 residues given in parentheses. The active site of Irp9 is completely conserved in Mtu MbtI (CAE55483). AtICS1 (AAL17715), Pae PchA (CAA57969), and Eco EntC (POAEJ2) contain an Ala at Thr-348 (marked with arrow). Thr-348 is predicted to H-bond with the 2-OH of SA. See "Experimental Procedures" for additional details.

AtICS1 Functional in Stress-induced Salicylate Biosynthesis
(e.g. Refs. 32 and 34). A. thaliana Col-O accumulates SA (ϳ30 g/g fresh weight/SAG) in response to extended growth at chilling temperatures (5°C) (32) at levels similar to strong induction by pathogens. Because AtICS1 is active at these temperatures in planta, it suggests that SA made via AtICS plays a role in cold acclimation or cold-tolerant growth. Arabidopsis is a chilling-resistant species, able to fully mature and produce seed at 4°C. It will be interesting to ascertain whether ICS enzymes isolated from chilling-sensitive plant species are also highly active at 4°C and to explore the potential selection for and mechanism of this coldadapted catalysis. For example, enzymatic rates can be dramatically enhanced over non-enzymatic rates (k cat /k non ) with decreasing temperature (e.g. Ref. 74). The extent of AtICS1 activity and SAG accumulation at 4°C (or 5°C) also suggests that SA (and SAG in particular) may play an unexplored role in cold-tolerant growth.
AtICS1 Can Compete with Other Stress-induced, Plastidlocalized Chorismate-utilizing Enzymes-Using our coupled irreversible continuous assay, we determined that mature AtICS1 has an apparent K m of 41.5 M for chorismate. This is critical as it now appears that a plant ICS, AtICS1, can compete with other plastid-localized chorismate-utilizing enzymes for chorismate (see Fig. 1). For example, AtICS1 and AtASA1 are induced in response to bacterial pathogens (17,39) and defense-related products of these pathways, camalexin (via anthranilate) and SA (via IC), are detected in parallel. 5 AtASA1 (At5g05730) has an apparent K m for chorismate in the range of 21 M (75) to 180 M (76) and thus could theoretically compete with AtICS1 for available chorismate. In contrast, past comparison of apparent K m values for chorismate of elicitor-induced C. roseus ICS isoforms (558 and 319 M (19)) with AS (67 M (77)) did not favor ICS. AtCM1 (At3g29200) is also induced in response to bacterial pathogens (37,38). AtCM1 has a reported apparent K m of 2.9 mM for chorismate (38); however, it is possible that this K m for chorismate is artificially high due to the presence of the chloroplast transit sequence (38). Alternatively, because phenylpropanoids typically dominate other chorismate-derived pathogen-induced products, this higher K m may be physiologically relevant, resulting in increased flux only when sufficient chorismate is available. In addition, both AtASA1 and AtCM1 activities are allosterically modulated by the aromatic amino acids (37,38,76); whereas, there has been no evidence for allosteric regulation of a plant or bacterial ICS enzyme (19,48). Therefore, should these enzymes be expressed in the same cells, it appears that AtICS1 could successfully direct available chorismate to the production of SA. Indeed, we found that overexpression of AtICS1 under control of its native promoter resulted in increased stressinduced SA accumulation suggesting that the amount of AtICS1 and not available chorismate limits induced SA synthesis. 5 The affinity of AtICS1 for chorismate also suggests the possibility that it can compete with a subset of chorismate-utilizing enzymes involved in "constitutive" aromatic amino acid and folate production. For example, AtICS1 could successfully compete for chorismate with all three characterized A. thaliana chorismate mutases (37,38), whereas the recently characterized Arabidopsis aminodeoxychorismate synthase exhibits Ͼ10-fold higher affinity for chorismate than does AtICS1 (78).
Stress-induced SA Biosynthesis from IC Is Likely Enzymatic and Plastidic-The localization of monofunctional AtICS1 and thus IC production to the plastid strongly suggests stress-induced SA biosynthesis is plastidic. Results from two overexpression studies support this hypothesis. First, studies with transgenic tobacco plants overexpressing bacterial monofunctional ICS and IPL enzymes targeted to either the plastid or cytosol singly or in combination found that only transgenic plants in which both ICS and IPL were targeted to the plastid exhibited an SA overexpression functional phenotype: SAG levels similar to local tobacco mosaic virus-infected leaves, constitutive expression of pathogenesis-related genes, and a reduction in tobacco mosaic virus-induced lesion size (56). Second, in Arabidopsis, an SA synthase PchBA fusion protein was constitutively expressed in either the plastid or the cytosol (20). Here too, a true SA overexpression phenotype (SAG ϳ20 g/g fresh weight, PR1 expression) was only observed when the fusion protein was targeted to the plastid, implying that SA synthesis occurs in the plastid in Arabidopsis.
How then is stress-induced SA made from IC? The evidence supports enzymatic synthesis of SA from isochorismate. We found non-enzymatic conversion of IC to SA under physiological conditions to be negligible. Furthermore, total SA accumulates to levels similar to those induced by pathogen in Arabidopsis grown at 5°C (32), a temperature at which non-enzymatic conversion of IC to SA should be insignificant. In P. aeruginosa, a direct comparison of the enzymatic (k cat Pae PchB (50)) and non-enzymatic (64) rates of SA formation from IC indicates that the enzyme accelerates SA formation ϳ4 ϫ 10 5 -fold (64). Does Arabidopsis contain a gene encoding a protein with IPL activity similar to Pae PchB? Pae PchB encodes a small 101-amino acid protein that appears to have evolved from a chorismate mutase and retains residual CM activity with 10-fold lower affinity for chorismate than for IC (50). Arabidopsis contains three genes encoding proteins with confirmed chorismate mutase activity (AtCM1 At3g29200, AtCM2 At5g10870, and AtCM3 At1g69370 (37,38)) and the putative chorismate mutase At3g07630. All but AtCM2 are predicted to be plastid-localized and are possible IPL candidates, although there are also other possible routes to SA from IC.
In conclusion, AtICS1 is required for the pathogen-induced accumulation of SA, a key phytohormone mediating plant response to pathogens, abiotic stress, and stress-induced developmental transitions. The biochemical properties of AtICS1 described herein suggest 1) its activity is not regulated by lightdependent changes in the chloroplast stroma, 2) it can effectively compete with other stress-induced and "constitutive" plastidic chorismate-utilizing enzymes, and 3) it is active at both low and high physiologically relevant temperatures supporting its proposed role mediating temperature-dependent stress. Finally, we provide evidence in support of positive selection for monofunctional ICS enzymes in plants and argue that stress-induced SA biosynthesis occurs in the plastid and requires AtICS1 and one or more additional enzymes.