Metabolic, Genomic, and Biochemical Analyses of GlandularTrichomes from the Wild Tomato Species Lycopersiconhirsutum Identify a Key Enzyme in the Biosynthesisof Methylketones W

Eyal Fridman,a Jihong Wang,a,1 Yoko Iijima,a John E. Froehlich,b David R. Gang,c

John Ohlrogge,d and Eran Picherskya,2

a Department of Molecular, Cellular, and Developmental Biology, University of Michigan Ann Arbor, Michigan 48109-1048bMichigan State University, Department of Energy, Plant Research Laboratory, Michigan State University, East Lansing,

Michigan 48824c Department of Plant Biology, University of Arizona, Tuscon, Arizona 85721-0036d Department of Plant Biology, Michigan State University, East Lansing, Michigan 48824

Medium-length methylketones (C7-C15) are highly effective in protecting plants from numerous pests. We used a bio-

chemical genomics approach to elucidate the pathway leading to synthesis of methylketones in the glandular trichomes of

the wild tomato Lycopersicon hirsutum f glabratum (accession PI126449). A comparison of gland EST databases from

accession PI126449 and a second L. hirsutum accession, LA1777, whose glands do not contain methylketones, showed that

the expression of genes for fatty acid biosynthesis is elevated in PI126449 glands, suggesting de novo biosynthesis of

methylketones. A cDNA abundant in the PI126449 gland EST database but rare in the LA1777 database was similar in

sequence to plant esterases. This cDNA, designated Methylketone Synthase 1 (MKS1), was expressed in Escherichia coli

and the purified protein used to catalyze in vitro reactions in which C12, C14, and C16 b-ketoacyl–acyl-carrier-proteins

(intermediates in fatty acid biosynthesis) were hydrolyzed and decarboxylated to give C11, C13, and C15 methylketones,

respectively. Although MKS1 does not contain a classical transit peptide, in vitro import assays showed that it was targeted

to the stroma of plastids, where fatty acid biosynthesis occurs. Levels of MKS1 transcript, protein, and enzymatic activity

were correlated with levels of methylketones and gland density in a variety of tomato accessions and in different plant

organs.

INTRODUCTION

Plants synthesize a multitude of specialized compounds to help

ward off pests. Some of these classes of compounds, like

terpenes and phenolics, are widely distributed throughout the

plant kingdom (Bennet and Wallsgrove, 1994; McGarvey and

Croteau, 1995). Others occur sporadically or are limited to one or

a few taxa. Medium-length methylketones (Figure 1), found

throughout the plant kingdom, are one class of compounds

highly effective in protecting plants against pests (Williams et al.,

1980; Kennedy, 2003). Early studies on the composition of the

essential oils of many species found medium-length methyl-

ketones in lime (Citrus limetta) leaves (Watts, 1886), in clove

(Eugenia caryophyllus) and cinnamon (Cinnamomum zeylani-

cum) oil (Walbaum and Huthig, 1902), in palm kernel (Lodoicea

maldivica), peanut (Arachis hypogaea), cottonseed (Gossypium

hirsutum), and sunflower (Helianthus annuus) seed oils

(Jasperson and Jones, 1947), and in oil of hop (Humulus lupulus)

(Sorm et al., 1949). 2-Tridecanone was characterized as a crys-

talline constituent of the essential oil of matsubasa (Shizandra

nigra maxim), a plant in the magnolia family, which is used as

a bath perfume (Sengoku, 1933). In some plants, the methylke-

tone and the derived secondary alcohol are found together, for

example, 2-heptanone and 2-heptanol in the oil of cloves

(Wehmer, 1931).

Leaves of the wild tomato species Lycopersicon hirsutum

f glabratum (Williams et al., 1980) are among the most promi-

nent sources of methylketones in plants. Several accessions

of this wild species contain mainly the two methylketones

2-undecanone and 2-tridecanone, in concentration ranging

between 2700 and 5500 mg per g fresh weight (Antonious,

2001). By comparison, the cultivated tomato L. esculentum has

only minute amounts of these compounds, up to 80 mg per g

(Antonious, 2001). In one of the accessions of the L. hirsutum f

glabratum (PI134417) the methylketones were reported to com-

pose up to 90% of the tip contents of the glandular trichomes

1Current address: Department of Biological Sciences, Brock University,St. Catharines, Ontario, Canada.2 To whom correspondence should be addressed. E-mail lelx@umich.edu; fax 734-936-3522.The author responsible for distribution of materials integral to thefindings presented in this article in accordance with the policy describedin the Instructions for Authors (www.plantcell.org) is: Eran Pichersky(lelx@umich.edu).WOnline version contains Web-only data.Article, publication date, and citation information can be found atwww.plantcell.org/cgi/doi/10.1105/tpc.104.029736.

The Plant Cell, Vol. 17, 1252–1267, April 2005, www.plantcell.orgª 2005 American Society of Plant Biologists

(Dimock and Kennedy, 1983). Trichomes, both glandular and

nonglandular, are prominent features of the foliage and stems

in the genus Lycopersicon (Luckwill, 1943), with glandular tri-

chomes predominating on most surfaces and the nonglandular

trichomes predominating on leaf veins (Figures 2A and 2B).

The glandular trichomes of various Lycopersicon species have

been shown to contain several types of secondary compounds,

and when the leaf is touched or chewed by herbivores, the

contents of the glands are released onto the leaf surface.

Considerable effort has been made to link the presence and

density of glandular trichomes to levels of host-plant pest re-

sistance both within and among Lycopersicon species (Carter

and Snyder, 1985; Yu et al., 1992; Li et al., 2003). In L. hirsutum,

several studies have shown a strong positive correlation among

density of leaf glandular trichomes, levels of methylketones, and

plant resistance against various pathogens, including tobacco

hornworm (Manduca sexta), spotted spider mite (Tetranuchus

urticae), green peach aphids (Myzus persicae), maize earworm

(Heliothis zea), and Colorado potato beetle (Leptinotarsa de-

cemlineata) (Kauffman and Kennedy, 1989; Antonious, 2001).

The prevalence in plants, as well as in other organisms, of

methylketones with an odd number of carbons (including the

long-chain methylketones [C > 20] found in cuticular waxes)

suggests that they are derived from fatty acids, perhaps by

decarboxylation of the respective b-ketoacids, which are inter-

mediates in both the biosynthesis and degradation pathways of

fatty acids. Several studies in the 1950s through the 1970s

investigating the role of microorganisms in the spoilage of fats

attempted to elucidate the biosynthesis ofmethylketones in fungi

(Mukherjee, 1951; Forney andMarkovetz, 1971). In these experi-

ments, Penicillium was grown in pure culture on individual fatty

acids, and in each case the methylketone formed had one fewer

carbon than the progenitor fatty acid. This suggested that

a decarboxylation reaction was involved, possibly during an

abortive fatty acid degradation process. In fatty acid degrada-

tion, the intermediates are CoA esters, and free b-ketoacids are

known to be highly labile and to undergo spontaneous decar-

boxylation (Ege, 1989). Thus, the enzyme responsible for the

production of 2-tridecanone (in the case of myristic acid) may

simply be a thioesterase acting on b-ketomyristoyl-CoA, giving

a free b-ketomyristic acid which then decarboxylates spontane-

ously. But enzymatic decarboxylation could not be ruled out

because a purified enzyme was not obtained and, therefore, the

Figure 1. Structures of Common Methylketones Found in Plants.

Figure 2. Glandular Trichomes of Tomato.

(A) A light microscope picture of the adaxial surface of a young leaf,

showing the predominant four-celled glandular trichomes (in focus) and

the less abundant long, nonglandular trichomes.

(B) Scanning electron micrograph of the abaxial surface of a mature leaf

of L. hirsutum. The four-celled glandular trichomes (arrow and in inset)

predominate.

(C) Light micrograph of the four-celled glandular trichomes isolated from

L. hirsutum leaves after final purification.

Tomato Methylketone Synthesis 1253

nature of the reaction was not defined further in vitro (Hwang

et al., 1976).

Whereas in fungi the methylketones appear to be derived

from degradation of fatty acids through a b-ketoacyl-CoA

intermediate (Forney and Markovetz, 1971), their synthesis in

plants may be equally likely to be derived from a b-ketoacyl–

acyl-carrier-protein (ACP) intermediate formed during fatty acid

biosynthesis. Under this hypothesis, an esterase could cleave

b-ketoacyl-ACP, rather thanb-ketoacyl-CoA, to form the unstable

free b-ketoacid, which will then decarboxylate, either enzymati-

cally or nonenzymatically to form the methylketone. In this study,

we used a combination of chemical, biochemical, and genomics

approaches to show that methylketones are not only stored but

also synthesized through the biosynthetic pathway of fatty acids in

tomato glandular trichomes.Wewere further able to demonstrate

the activity of a novel enzyme, Methylketone Synthase 1 (MKS1)

that catalyzes the final step of methylketone synthesis.

RESULTS

Localization of Methylketone Storage to the Glandular

Trichomes of L. hirsutum f glabratum

Although the presence of methylketones has been detected

in several tomato species, including the cultivated tomato L.

esculentum, the highest levels have been reported in L. hirsutum

f glabratum, including accessions PI126449, PI134417, and

LA0407 (Antonious, 2001). We extracted volatiles from intact

leaves of several L. hirsutum accessions (see Methods) and

confirmed that L. hirsutum f glabratum accession PI126449 had

high levels of the methylketones 2-undecanone and 2-trideca-

none, as well as 2-pentadecanone (Table 1). We also analyzed

the volatile content in the related L. hirsutum accession LA1777

(Table 1), the only L. hirsutum accession for which a gland EST

database was available at the beginning of our investigation.

LA1777 was found not to make detectable amounts of methyl-

ketones (Table 1). Leaves from both lines were shown to contain

a variety of terpenes, mostly sesquiterpenes, in agreement with

a previous study on the biosynthesis of terpenes in these

accessions (van Der Hoeven et al., 2000).

InL.hirsutum, the predominant typeof trichome is a four-celled

gland mounted on a relatively short stalk (Figures 2A and 2B). A

previous study showed that material extracted directly from

attached glands of this type contains methylketones (Lin et al.,

1987). Therefore,we tested the correlationbetween thepresence

of these glands and methylketone content in different organs of

methylketone-producing plants. Methylketone content per unit

freshweight was highest in young leaves, whereasmature leaves

contained approximately half this amount (Figure 3A). Glandular

trichomes were observed on both sides of the leaves, and young

and mature leaves had similar weight to surface ratios, but the

density of the glandular trichomes in young leaves (58.8 6 8.0

glands/mm2) was twice that of mature leaves (31.66 3.7 glands/

mm2). Sepals were similar to young and old leaves in having

glandular trichomes on both sides and similar weight/surface

ratios, but sepal gland density was similar to young leaves, and

methylketone content per unit weight was intermediate between

that of young and mature leaves (Figure 3A). Petals contained

;20% of the methylketone content per unit weight present in

young leaves. Examination of the petal surface revealed that

glandular trichomeswere present only on the abaxial side around

the base of the petal, where cells appeared to contain chlorophyll

as well as yellow pigment. Young stems had <7% of the

methylketone content per unit weight present in young leaves,

and this correlated with the >25-fold reduction in weight/surface

ratio of the stem, aswell as a lower density of glandular trichomes

(15.2 6 0.9 glands/mm2). We also examined roots, which con-

tained no methylketones (Figure 3A).

To examine more closely the correlation between glandular

trichomes and methylketones, we compared leaves and stems

for methylketone content before and after these organs were

brushed to remove the surface glands. As was observed with

other plants containing glands (Gang et al., 2001), it was not

possible to remove all glands from leaves by brushing because

Table 1. Comparison of the Volatile Content in Mature and Young (<2 cm) Leaves of L. hirsutum (LA1777) and L. hirsutum f glabratum (PI126449)a

Genotype LA1777 PI126449

Leaf Stage Mature Young Mature Young

Parameter Average SD Average SD Average SD Average SD

Ocimene 15.6 3.3 32.9 5.8 NDb ND

Elemene 7.6 1.3 15.2 3.2 ND ND

a-Bergamotene 12.6 1.5 23.3 4.5 ND ND

a-Farnasene 24.1 8.9 42.8 5.0 ND ND

b-Caryophyllene ND ND 149.0 38.0 537.0 89.0

Germacrene D 60.8 13.0 118.6 20.0 ND ND

Dedecatriene 33.2 5.0 55.5 15.4 ND ND

2-Undecanone ND ND 599.0 196.0 1940.0 547.0

2-Tridecanone ND ND 2150.0 587.0 6050.0 1190.0

2-Pentadecanone ND ND 90.5 20.7 261.0 67.0

a Results are presented as micrograms/gram fresh weight and are the average of four replicates. The less volatile sesquiterpene carboxylic acids were

not scored.b ND, not detected.

1254 The Plant Cell

such treatment eventually led to excessive damage to the leaves.

Leaves after moderate brushing contained only 36% of the

amount of methylketones found in unbrushed leaves (Figure 3B).

Stems are much hardier than leaves, and rigorous brushing of

stems led to the removal of almost all glands and 98% of the

methylketones (Figure 3B).

We next isolated glandular trichomes from L. hirsutum acces-

sions PI126449 and LA1777 (Figure 2B) and examined them for

the presence of methylketones. After gland isolation (Figure 2C),

volatiles were extracted and analyzed by gas chromatography–

mass spectrometry (GC-MS). The glands of LA1777 contained

mostly sesquiterpenes and one monoterpenene (ocimene), but

no methylketones could be detected (Figure 3C). In PI126449,

on the other hand, 2-undecanone, 2-tridecanone, and

2-pentadecanone were prominent peaks (Figure 3C) found in

similar ratios to those found in extracts from whole leaves.

Moreover, additional peaks tentatively identified by mass spec-

trometry as modified methylketones were also observed. These

results confirm that the glandular trichomes are the site of

storage of methylketones.

A Gland EST Database of the Methylketone-Producing

Accession PI126449 Has High Levels of Sequences

Encoding Fatty Acid Biosynthetic Enzymes Compared

with an EST Database from LA1777 Glands

To examine whether the methylketones in L. hirsutum f glabra-

tum are derived from fatty acid biosynthesis or degradation, we

constructed an EST database from the glands of PI126449. The

prevalence of cDNAs encoding enzymes involved in fatty acid

biosynthesis and degradation was determined in the analyzed

sequences and compared with those found in the previously

reported EST database from trichomes of LA1777 (The Institute

for Genomic Research, http://www.tigr.org/tdb/Igi/). The

PI126449 database contains ;5500 ESTs, and the LA1777

database contains ;2500. Most of the plant genes encoding

both the enzymes of the degradative pathway (which occurs in

the peroxisomes) and the enzymes of the biosynthetic pathway

(which occurs in the plastids) have been molecularly character-

ized (Mekhedov et al., 2000; Graham and Eastmond, 2002), so

this analysis was relatively straightforward. One of the most

striking results of this analysis was the high abundance of the

cDNA for the ACP in the PI126449 EST database. ACP is a small

(;9.5 kD) acidic protein that functions as an essential cofactor

in fatty acid synthesis, acyl-ACP desaturation (McKeon and

Stumpf, 1982), and plastidic acyl-transferase reactions (Frentzen

et al., 1983). Multiple ACP isoforms have been found in evolu-

tionarily diverse species of higher and lower plants, and some are

constitutively expressed, while others show developmental and

tissue-specific expression (Suh et al., 1999). We identified two

distinct contigs in PI126449 that encode two isoforms of the

protein (in a 3:1 ratio), which together account for 0.9% of the

total transcripts in the glands. The LA1777 EST database, on

the other hand, contained only one cDNA (accession number

AW616614; 0.04% of total cDNAs) encoding ACP (Figure 4A).

With two exceptions, cDNAs for all other enzymes of the

biosynthetic pathway were highly represented in the PI126449

Figure 3. Localization of Methylketones in Glandular Trichomes of

Tomato.

(A) Relative levels of methylketone content in different tissues of the

plant, per unit fresh weight. YL, young leaf; ML, mature leaf; ST, stem;

RT, root; SE, sepal; PE, petal. Values are shown as percentages of the

amount in young leaves. Each value in the graph is an average of three

replicates.

(B) Comparison of the relative levels of methylketones in leaves and

stems with or without a brushing treatment to remove glands, per unit

fresh weight. Leaf volatiles were extracted from opposite leaflets of the

same leaf. One leaflet was extracted directly with methyl-tert-butyl ether

after weighing (Brushed �), and the other was brushed under liquid N2

(Brushed þ) before weighing and extraction. Stem samples were cut in

half and one half was extracted directly after weighing (Brushed �), and

the other was brushed under liquid N2 (Brushed þ) before extraction. For

each organ, the level of methylketones in unbrushed leaves was set at

100%. Each value in the graph is an average of three replicates.

(C) Comparison of the volatile content in the glandular trichomes of L.

hirsutum f glabratum (PI126449; top) and L. hirsutum (LA1777; bottom).

Glands were isolated from young leaves, and the volatile content was

extracted with hexane and injected into a gas chromatograph–mass

spectrometer. A portion of the gas chromatograph, from 12.5 to 25 min,

is shown. Ter, terpene; MKx, unidentified methylketone derivative; 2UD,

2-undecanone; 2TD, 2-tridecanone; 2PD, 2-pentadecanone. All the

major peaks in the LA1777 chromatograph were identified as terpenes,

mostly sesquiterpenes. The scale at the y axis is in arbitrary units.

Tomato Methylketone Synthesis 1255

database compared with their representation in the LA1777

database (Figure 4A). The two exceptions are cDNAs for the

b-hydroxyacyl-ACP dehydratase (no cDNA in PI126449; 0.04%

in LA1777) and enoyl-ACP reductase (0.05% in PI126449; 0.08%

in LA1777).

The prevalence of cDNAs encoding the degradative enzymes

in the gland EST databases of the two L. hirsutum accessions

was, on average, severalfold lower than the prevalence of the

cDNAs for the biosynthetic enzymes, and there was no statisti-

cally significant difference in their abundance between the two

accessions (Figure 4B). Thus, this analysis suggested that the

methylketones in PI126449 are synthesized during de novo fatty

acid synthesis rather than via the degradative pathway.

An ESTWith Homology to Some Esterases Is Highly

Expressed in PI126449 Glands and the Encoded Protein

Is Correlated with Methylketone Production

The most abundant unknown sequence in the PI126449 data-

base (107 ESTs, 2% of total ESTs, ranked third in abundance

after a putative secretory carrier membrane protein and

S-adenosyl-L-homocysteinase) encodes a29-kDprotein,with an

open reading frame (ORF) containing 265 amino acids (Figure 5).

The correct ORF was established by examining several inde-

pendent full-length cDNAs, two of which contain 182 nucleotides

upstream of the initiating ATG (accession number AY701574).

There are four stop codons in frame in this 59 nontranslated

region, with the 39-most being located 36 nucleotides upstream

of the initiating ATG. Two more cDNAs contain 59 nontranslated

sequences that include the stop codon proximal to the initiating

ATG. In addition, the genomic sequence of this gene is com-

pletely identical to the corresponding sequences of all these

cDNAs (E. Fridman and E. Pichersky, unpublished data).

The protein encoded by this ORF, which we designated MKS1

(for Methylketone Synthase 1, see below), has 48% se-

quence identity with tomatomethyl jasmonate esterase (LeMJE)

(Stuhlfelder et al., 2004) and 44% with the alkaloid biosynthetic

enzyme polyneuridine aldehyde esterase (PNAE) from Rauvolfia

serpentina (Dogru et al., 2000) (Figure 5). In addition, this pep-

tide is similar to several proteins with no proven function, all

Figure 4. In Silico Analysis of the L. hirsutum Leaf Glandular Trichome EST Databases.

(A) Analysis of the enzymes of the fatty acid biosynthesis pathway. The frequencies of specific cDNAs encoding each enzyme (labeled A to F) and the

ACP are in parentheses (number per 10,000 ESTs; the number to the left of the diagonal is for PI126449; the number to the right is for LA1777). A, acetyl-

CoA carboxylase; B, malonyl-CoA:ACP transacylase; CI, b-ketoacyl-ACP synthase I; CIII, b-ketoacyl-ACP synthase III; D, b-ketoacyl-ACP reductase; E,

b-hydroxyacyl-ACP dehydratase; F, enoyl-ACP reductase. The total number of ESTs in accession PI126449 is;5500, and the total number of ESTs in

accession LA1777 is ;2500.

(B) Analysis of the peroxisomal core enzymes of the b-oxidation of saturated fatty acids. The frequencies of ESTs (number per 10,000 ESTs) encoding

each enzyme (labeled A to C) are in parentheses (left number, PI126449; right number, LA1777). A, acyl-CoA oxidase; B, multifunctional protein, with

BI-2-trans-enoyl-CoA hydratase and BII-L-b-hydroxyacyl-CoA dehydrogenase; C, b-ketoacyl-CoA thiolase.

1256 The Plant Cell

considered to be part of the large a/b-hydrolase fold proteins

(Waspi et al., 1998; Zhong et al., 2001; Marshall et al., 2003). This

cDNA was represented only once (0.04%; accession number

AW617364) in the LA1777 gland EST database. The PI126449

EST database did not contain any other abundant sequences

(>0.5%) that encoded proteins with similarity to known thio-

esterases or other general esterases.

To investigate the correlation between levels of MKS1 and

methylketone content, we first sampled young leaves of 10

different accessions of Lycopersicon. We determined their

methylketone content by GC-MS and analyzed the transcript

levels ofMKS1 and the abundance of its protein product. In RNA

gel blots, a signal was observed for mRNA isolated from whole

leaves (i.e., leaves containing glands) of PI126449, LA134417,

and LA0407 (Figure 6A, lanes 8 to 10, respectively), all L. hirsutum

f glabratum accessions that were reported previously to contain

high levels of methylketones (Antonious, 2001). Furthermore,

high levels ofMKS1 transcripts were also detected in L. hirsutum

f glabratum accession LA1624 (lane 5), which had not previously

been examined for methylketone content, but which our mea-

surements showed to have similar methylketone content to the

above three genotypes (Figures 6A and 6C). MKS1 protein

content in all these accessions was concordant with MKS1

transcript levels and methylketone content (Figure 6B). On the

other hand,MKS1 transcripts orMKS1proteinwere not detected

in leaves from the cultivated tomato M82 (L. esculentum; lane 1),

from two accessions of L. pennellii (lanes 2 and 3), and from

L. hirsutum LA1777 (Figures 6A and 6B, lane 6), all lines that

did not contain detectable levels of methylketones (Figure

6C). L. hirsutum f glabratum accession PI251305 and L.

hirsutum accession PI127826 had low levels ofMKS1 transcripts

and MKS1 protein (Figures 6A and 6B, lanes 7 and 4, respec-

tively); metabolic profiling of these two accessions showed

that the former had barely detectable levels of methylketones,

whereas no methylketones could be detected in the latter

(Figure 6C).

The signal was significantly stronger when RNA from purified

PI126449 glands was used and still absent with RNA from

purified LA1777 glands (Figure 6A, lanes 11 and 12, respectively).

Moreover, the high and specific expression of MKS1 in the

PI12449 glands, but not LA1777 glands, was also reflected in the

abundance of its protein product in the PI126449 glands, but not

LA1777 glands (Figure 6B). Using a calibration curve with known

amounts of purified MKS1 produced in Escherichia coli, we

calculated the concentration of MKS1 in the PI126449 accession

to be 2 and 25 ng�mg�1 total soluble protein of the leaves and

glands, respectively, and 0.5 and 50 ng�mg fresh weight�1 of the

leaves and glands, respectively.

To further examine the correlation among MKS1, methyl-

ketones, and glands, wemeasured the concentration of MKS1 in

different parts of the plant (Figure 7A). MKS1 protein levels were

quantified by SDS-PAGE, followed by immunoblotting of the

protein samples together with known amounts of the purified

MKS1 protein. MKS1 levels were highest in both young and

mature leaves and in sepals and lower in stems and petals,

generally correlating with the methylketone content of these

organs (Figure 3A). Aswithmethylketone level, MKS1 levels were

also correlated with the presence of glands because brushed

leaves and stems contained much reduced amounts of MKS1

(75 and 90% reduction in the brushed leaves and stems, re-

spectively) (Figure 7B).

Localization of the Tomato MKS1 Protein within the

Soluble Fraction of the Chloroplast

Bioinformatic analysis of the protein encoded by MKS1

cDNA with software obtained at http://www.inra.fr/servlets/

WebPredotar predicts a plastidic localization of the protein (plas-

tid score ¼ 75.1%). No other organellar targeting sequences

and, in particular, no peroxisomal targeting sequences were

identified. However, although the subcellular localization of

neither LeMJE nor PNAE has been reported, both also lack

Figure 5. Amino Acid Sequence Alignment Comparing the L. hirsutum f glabratum Accession PI126449 MKS1 Protein with Two Functionally

Characterized Esterases Belonging to the a/b-Hydrolase Superfamily.

The MKS1 accession number is AY701574. LeMJE, methyl jasmonate esterase from L. esculentum (accession number AAS10488); PNAE,

polyneuridine aldehyde esterase from R. serpentina (accession number AAF22288). The first amino acid position of the mature PNAE and LeMJE,

determined by N-terminal sequencing, is underlined and in bold. When the same amino acid occurs in two or all three proteins in a given position, it is

shown in white on a black background.

Tomato Methylketone Synthesis 1257

a standard transit peptide because N-terminal sequencing of the

mature proteins showed that they begin from positions 3 and 7,

respectively, of the ORFs (Dogru et al., 2000; Stuhlfelder et al.,

2002) (Figure 5). The region between positions 12 and 53, which

includes amino acids known to be important for the catalytic

activity of PNAE, is highly conserved inMKS1 aswell as in LeMJE

(Figure 5). Furthermore, the protein encoded by the ORF of the

MKS1 cDNA has the same number of amino acids in the 59-end

of this conserved domain. In addition, the MKS1 protein from

glands, detected by immunoblotting, appears to have the same

or very similarMr value to that of E. coli–produced MKS1 (Figure

6B). Theseobservations strongly suggest that, atmost, only a few

amino acids are removed from the N terminus to produce the

mature L. hirsutum f glabratum MKS1; therefore, the N-terminal

part encoded by the MKS1 ORF is not a typical, cleavable tran-

sit peptide. However, since the chemical structure of methyl-

ketones and our gene expression data led us to hypothesize that

methylketones are derived from intermediates in fatty acid

biosynthesis, which occurs in the chloroplast, we tested the

ability of MKS1 to import into isolated chloroplasts.

The in vitro–translated, radiolabeled tomatoMKS1 protein was

incubated with isolated pea (Pisum sativum var Little Marvel)

chloroplasts and ATP, followed by reisolation of the chloroplasts,

incubation with proteases, lysis, fractionation into membrane

and soluble components, andSDS-PAGE. The translated protein

was found to be imported into the chloroplast (Figure 8B), as

deduced from its resistance to proteolytic digestion by both

thermolysin and trypsin, compared with the sensitivity of the

unprotected MKS1 (treatment of chloroplasts with Triton X-100

before protease treatment makes their membrane permeable to

the proteases; Figures 8A and 8B, lanes 7 to 10), and could be

recovered from the soluble fraction of the chloroplast.

The resistance of MKS1 to trypsin after import and its recovery

in the soluble fraction of the chloroplast indicate that MKS1 is

routed to the stroma compartment after crossing the outer and

inner envelopes, as trypsin is known to be able to cleave proteins

that protrude into the intermembrane space between the inner

and outer envelopes (Figure 8C), but not proteins buried in the

inner membrane or found in the stroma (Froehlich et al., 2001).

Identical import and protease resistance behavior to that of

MKS1, with the exception of a clear transit peptide cleavage, is

observed with the mature small subunit of ribulose-1,5-bisphos-

phate carboxylase/oxygenase (labeled mSS, for mature small

subunit; Figure 8D), a well-established stromal protein (Olsen

Figure 6. Correlation betweenMKS1 Transcript and Protein andMethyl-

ketone Content in Leaves and Glands of Different Tomato Accessions.

(A) RNA gel blots of MKS1 mRNA abundance in young leaves and in

glands of various Lycopersicon accessions. Lanes 1 to 10, 5 mg each of

total RNA isolated from young leaves; lanes 11 and 12, 2 mg each of total

RNA isolated from glands. Lanes 1 to 12, rRNA control.

(B) Protein immunoblot of the levels of MKS1 protein in young leaves and

glands of various Lycopersicon accessions. Lanes 1 to 10 (15% SDS-

PAGE), 2 mg each of total plant protein; lanes 11 and 12, 1 mg each; lane

13, 80 ng of MKS1 purified from E. coli lysate.

(C) Total methylketone content in young leaves of various Lycopersicon

accessions. The bars and error bars indicate mean and SE, respectively,

obtained from four measurements (each from a different plant) for each

accession. Le, L. esculentum; Lp, L. pennellii; Lh, L. hirsutum; Lhg, L.

hirsutum f glabratum.

Figure 7. Correlation between MKS1 Protein and Glands in the Whole

Plant.

(A) Relative levels of MKS1 content in different tissues of the plant. YL,

young leaf; ML, mature leaf; ST, stem; RT, root; SE, sepal; PE, petal.

MKS1 levels (per unit fresh weight) were quantified by immunoblotting.

Values are shown as percentages of the amount in young leaves. Each

value is an average of three replicates.

(B) Comparison of the relative levels of MKS1 in leaves and stems with

and without a brushing treatment to remove glands, per unit fresh weight.

Leaves and stems treated as in Figure 3B. For each organ, the level of

MKS1 in unbrushed leaves was set at 100%. Each bar represents an

average of at least three replicates.

1258 The Plant Cell

and Keegstra, 1992). Under the conditions of this assay, theMr of

MKS1 did not appear to decrease after import, consistent with

the observation that the MKS1 protein synthesized in E. coli from

the complete ORF is essentially identical in size to the one

extracted from glands (Figure 6B), indicating that MKS1 import

into the plastids is accompanied by the removal of either a few

amino acids or none at all.

The Tomato MKS1 Protein Catalyzes the Formation of

2-Undecanone, 2-Tridecanone, and 2-Pentadecanone

from b-Ketolauroyl-ACP, b-Ketomyristoyl-ACP,

and b-Ketopalmitoyl-ACP in the Glands

PNAE is amember of thea/b-hydrolase superfamily (Dogru et al.,

2000), and it catalyzes the deesterification of a methylester with

a b-keto functionality (Figure 9A). Once the methyl group is

removed, the resulting b-ketoacid decarboxylates, although it is

not yet known if this reaction is also catalyzed by PNAE or oc-

curs spontaneously either before or after the b-ketoacid leaves

the active site of the enzyme. The similarity of the PNAE ester-

ase protein sequence to MKS1 suggested that the latter may be

the hypothesizedMKS enzyme that deesterifies b-ketoacyl-ACPs

to the b-ketoacids and thus leads to the formation of methyl-

ketones after decarboxylation (Figure 9B).

To test if MKS1 catalyzes the deesterification and possibly the

decarboxylation of b-ketoacyl-ACPs, we expressed its cDNA in

E. coli and purified it to near homogeneity (see Supplemental

Figure 1 online). Because the predominant L. hirsutum f glabar-

tum accession PI126449 methylketone is 2-tridecanone, we first

synthesized its hypothesized precursor, [3-14C] b- ketomyristoyl

-ACP (because the b-ketoacyl-ACP substrates are unstable

and not commercially available, we developed a synthesis

method to produce such compounds, see Methods and Sup-

plemental Figures 2 and 3 online). The PI126449 gland protein

extract and the purified E. coli–produced MKS1 were tested in

a reaction containing the synthesized and trichloroacetic acid–

purified [3-14C] b- ketomyristoyl-ACP. A crude protein extract of

LA1777 glands and BSA were assayed as controls. The extracts

from PI126449 glands reproducibly converted this substrate to

a product that was identified as 2-tridecanone by comigration

Figure 8. MKS1 Is Targeted to the Stroma of Chloroplasts.

(A) Sensitivity of unprotected MKS1 to thermolysin and trypsin. Lane 1, 1 mg MKS1 purified from E. coli, run on SDS-PAGE, and visualized by

Coomassie bluee staining; lane 2, in vitro–translated, [35S]-labeled MKS1 product (translated protein [TP]) electrophoresed under identical conditions.

Lanes 3 and 4, radioactively labeled MKS1 (same amount as in lane 2) treated with either thermolysin or trypsin for 30 min at 48C before electrophoresis.

The gel containing lanes 2 to 4 was analyzed by fluorography after SDS-PAGE.

(B) MKS1 import assay. [35S]-labeled MKS1 was incubated with isolated pea chloroplasts in the presence of 4.0 mM Mg-ATP for 30 min. Lane labeled

TP represents 10% of translated product added to a single import assay. After import, chloroplasts were reisolated, Iysed, and fractionated into soluble

and insoluble fractions (see Methods). Both fractions were analyzed by SDS-PAGE and fluorography. P, membrane (pellet) component; S, soluble

component.

(C) Import assay of Arc6, a DnaJ-like protein that contains a cleavable transit peptide and is imported into the plastid envelope (Vitha et al., 2003).

Experiments were analyzed in the same way as in (B). The N terminus of the protein precursor (pArc6) is cleaved after import, and the mature protein

(mArc6) is partially sensitive to trypsin because it protrudes into the intermembrane space. The asterisk depicts the protected part of the protein, which

is embedded in the membrane.

(D) Import assay of the small subunit of ribulose-1,5-bisphosphate carboxylase/oxygenase that contains a cleavable transit peptide and is imported into

the stroma. The N terminus of the protein precursor small subunit (pSS) is cleaved after import, and the mature small subunit protein (mSS) is therefore

not sensitive to trypsin digestion.

Tomato Methylketone Synthesis 1259

with an authentic 2-tridecanone standard (Figure 10A, lane 1).

This product was also identified as 2-tridecanone by GC-MS

analysis (see Supplemental Figure 4 online). Extracts of LA1777

did not produce any 2-tridecanone (Figure 10A, lane 2). The

purified MKS1 protein also catalyzed the formation of 2-trideca-

none (Figure 10A, lane 3), whereas a control reaction containing

BSA produced no 2-tridecanone (Figure 10A, lane 4).

To further examine the correlation between MKS activity

and the presence of glands, we compared the levels of MKS

enzymatic activity in the glands to that in leaves and stems with

and without the brushing treatment. On average, MKS1-specific

activity (per milligram of protein) in the leaf and stem was 48 and

32%, respectively, of that of the glands (Figure 10B). Partial

removal of the glands from the leaf and stem reduced the levels

ofMKS1 activity by 79 and 90%, respectively, comparedwith the

activity with the glands (Figure 10B).

b-Ketolauroyl-ACP and b-ketopalmitoyl-ACP were syn-

thesized in a similar way, and MKS1 was able to convert

b-ketolauroyl-ACP to 2-undecanone and b-ketopalmitoyl-ACP to

2-pentadecanone. We next determined the apparent Km values

of purified MKS1 for all three substrates (Figure 11). The enzyme

had an apparent Km value for b- ketomyristoyl-ACP of 3.1 mM

(Figure 11A). The apparent Km values for b-ketolauroyl-ACP

and b-ketopalmitoyl-ACP were 1.2 and 10.7 mM, respectively

Figure 9. The Hypothesized Reaction Leading to Methylketones Is Similar to the Reaction Catalyzed by PNAE.

(A) General scheme of the reaction catalyzed by PNAE.

(B) Proposed scheme of the reaction catalyzed by MKS.

Figure 10. Radio–Thin Layer Chromatography Analysis of the Hexane-Soluble Products from Enzymatic Assays Using [3-14C] b-Ketomyristoyl-ACP as

a Substrate.

(A) Enzymatic assays with gland protein and purified MKS1. Arrows indicate the positions of the 2-tridecanone (2TD) standard, the [1-14C] lauric acid

standard (C12:0), and the origin. Lane 1, crude protein extract of PI126449 glands; lane 2, crude protein extract of LA1777 glands; lane 3, MKS1 purified

from E. coli; lane 4, BSA control. The concentration of the substrate in each reaction was 10 mM. Lanes 1 and 2 contained 5 mg total protein. Lanes 3 and

4 contained 0.5 mg of protein.

(B)Comparisons of the relative levels of MKS1 activity in stems and leaves with and without treatment to remove glands. Leaves and stems treated as in

Figure 3B. The concentration of the substrate and the amount of protein in each reaction was 3 mM and 3 mg, respectively. A sample thin layer

chromatography plate is shown above, and the bars represent averages of three replicates.

1260 The Plant Cell

(Figures 11B and 11C). The Vmax value for the reaction of MKS1

with [3-14C] b- ketomyristoyl-ACP was 10�8 mmoles�s�1�mgprotein�1, and similar values were obtained for the reaction

with b-ketolauroyl-ACP and b-ketopalmitoyl-ACP.

We also tested the esterase activity of MKS1 with C10:0-ACP,

C12:0-ACP, C14:0-ACP, and C16:0-ACP. MKS1 had no ester-

ase activity with these substrates, suggesting that the enzyme is

specific to b-ketoacyl-ACP rather than the straight acyl-ACP.

The enzyme also did not hydrolyze benzoyl-CoA or the artificial

substrates p-nitrophenyl-palmitoyl and p-nitrophenyl-decanoyl

to the free benzoic acid and fatty acids, respectively, indicating

that the enzyme is not a general esterase.

DISCUSSION

The Machinery to Accumulate and Produce Methylketones

from Fatty Acids Is Located in the Glands of

L. hirsutum f glabratum

The four-celled glandular trichomes on the leaves of the wild

tomato L. hirsutum f glabratum had previously been recognized

to be the site of accumulation of methylketones (Lin et al., 1987).

We show that methylketone content correlates well with the

presence and density of glands in leaves aswell as in other green

parts of the plant (Figure 3A). However, the correlation is not

perfect; for example, stems have 7% of the methylketone con-

tent per unit fresh weight that young leaves do, but young

leaves have almost 100-fold more glands per unit fresh weight.

This observation suggests that stem glands contain more

methylketones because, perhaps, being older, they have had

more time to accumulate them. The correlation of methylketone

content with glands was further strengthened by examining

leaves and stems from which most glands were removed by

brushing. Although it was not possible to remove all glands from

the surfaces of the plant organs, the great decrease in methyl-

ketone content in brushed leaves and stems, as well as the

isolation and purification of glands and the demonstration of their

methylketone content, support the conclusion that the glands

are the main, and perhaps exclusive, site of methylketone

storage in the plant (Figures 3B and 3C).

With the demonstration of the glandular trichomes as themajor

site of methylketone storage, we applied the single-cell EST

database approach to determine if these glands might also be

the site of synthesis of methylketones and, if so, to identify the

genes and enzymes involved in their biosynthesis. Analysis

of EST databases constructed from a single type of cell in

Figure 11. Steady State Kinetic Analysis of MKS1 Assayed with [3-14C]

b-Ketoacyl-ACPs.

Initial velocities are shown as dots, and the line represents the nonlinear

least-squares fit of the initial velocities versus [3-14C] b-ketoacyl-ACP

concentration to the hyperbolic Michaelis-Menten equation. 2UD, 2-

undecanone; 2TD, 2-tridecanone; 2PD, 2-pentadecanone.

(A) A representative measurement with [3-14C] b-ketomeristoyl-ACP as

the substrate. The apparent Km value shown is an average of three

independent assays.

(B) A measurement with [3-14C] b-ketolauroyl-ACP as the substrate. This

assay gave a similar result to the one shown in (A) and was done only

once because of the low availability of substrate (see Methods).

(C) A representative measurement with [3-14C] b-ketopalmitoyl-ACP as

the substrate. The apparent Km value shown is an average of two

independent assays.

Tomato Methylketone Synthesis 1261

conjunction with other types of data, such asmetabolic profiling,

is a powerful method to identify major biochemical pathways

operating in the cell. We have previously used this approach to

identify genes important in the phenylpropene and terpene

pathways by constructing and comparing EST databases from

peltate gland cells of several basil varieties (Gang et al., 2001,

2002; Iijima et al., 2004a, 2004b). Gene identification in these

studies was based on prevalence of specific sequences, com-

parisons with known sequences encoding enzymes of similar

function (either in terms of substrates or type of reactions),

metabolic profiling of the plant material, and enzymatic assays

of candidate proteins (Fridman and Pichersky, 2005). This ap-

proach has allowed us to determine that methylketones are

made via the de novo fatty acid biosynthetic pathway in the

chloroplast and to identify MKS1 as the enzyme responsible for

the reaction leading from C12, C14, and C16 b-ketoacyl-ACPs to

the C11, C13, and C15 methylketones, respectively. The levels of

MKS1 transcripts and protein are closely correlated with the

presence of methylketones in the Lycopersicon genus (Figure 6).

Furthermore, the levels of MKS1 transcripts, protein, and MKS

enzymatic activity were found to correlate well with the presence

of glands throughout the plant (Figures 6, 7, and 10). Taken to-

gether, these results indicate that these glands are also the

major, and perhaps exclusive, site of methylketone biosynthesis.

MKS1 Is a Plastid Protein

The results of the chloroplast import experiments indicate that

MKS1 is targeted to the soluble stroma compartment of the

chloroplast (Figure 8B), in close proximity to the site of fatty acid

biosynthesis (Walker and Harwood, 1985). The routing of a pro-

tein to the plastid stroma without the cleavage of a typical transit

peptide is rare but not unprecedented; Rathinasabapathi et al.

(1994) reported that betaine aldehyde dehydrogenase is im-

ported into the chloroplast and is ultimately localized in the

stroma without cleavage of an N-terminal peptide. In addition,

several inner and outer membrane proteins are known not to

contain a cleavable transit peptide (Froehlich et al., 2001; Miras

et al., 2002). Although our results indicate that MKS1 is not sub-

stantially processed during import (Figures 6B and 8B), we

cannot exclude the possibility that a small peptide is removed

from MKS1 during its targeting to the stroma.

The Mechanism of Action of MKS1

Our enzymatic characterization ofMKS1 indicated that it can use

several b-ketoacyl-ACPs with high affinity to produce the corre-

sponding methylketones. However, the efficiency of the reaction

was too low to account for the amount ofmethylketonesmade by

Figure 12. Unrooted Neighbor-Joining Phylogenetic Tree of Selected Plant Proteins Belonging to the a/b-Hydrolase (Esterase) Family.

The source and accession numbers for the non-Arabidopsis proteins that do not appear in the sequence alignment in Figure 5 are as follows: EIE

(Ethylene Induced Esterase) from Citrus sinensis, accession AAK58599; SABP2 (Salicylic Acid Binding Protein 2) from Nicotiana tabacum, AAR87711;

PIR7 (pathogen-induced) from Oryza sativa, CAA84024; StMJE (methyl jasmonate esterase) from Solanum tuberosum (potato), AY684102. The scale

indicates average substitutions per site for each cladogram.

1262 The Plant Cell

the plant. It can be calculated that only 1% of the observed

amount of methylketones in the leaves could be made given the

amount of MKS1 in the leaves and the Vmax value of the enzyme

determined in the in vitro assays. There may be several ex-

planations for this discrepancy between the in vitro rate of the

MKS1-catalyzed reaction and the actual in vivo synthesis of

methylketones in the cell. It is likely that the in vitro assay we

developed is not yet optimized. The enzyme may need some

cofactors that have not yet been identified. For example, we

have found that adding the soluble fraction of an E. coli

extract increases the efficiency by ;10-fold (E. Fridman and

E. Pichersky, unpublished data). In addition, we used spinach

(Spinacia oleracea) ACP in our assays, but the enzymemay work

faster with L. hirsutum ACPs. Several previous kinetic character-

izations of fatty acid biosynthetic enzymes (e.g., acetyl-CoA

carboxylase andmalonyl-CoA:ACP transacylase) and polyketide

synthases, some of which use substrates containing ACP

moieties, have reported either low in vitro Vmax values or higher

Km values (Alban et al., 1994) that could not explain the observed

rate of polyketide or fatty acid synthesis in the organism. In some

cases, the ACP moiety was directly tied to the low Vmax values

(Tang et al., 2003), but in other cases the reasons remained

undetermined or were attributed to a lack of components of

a multienzyme complex, where the product of one reaction is

generated immediately adjacent to the active site of the next

enzyme in the reaction sequence (e.g., substrate channeling)

(Roughan, 1997; Tang et al., 2003).

There are some indications that the plastidic enzymes involved

in fatty acid biosynthesis are part of a complex (Roughan and

Ohlrogge, 1996). It is also likely that MKS1 is part of this protein

complex in the L. hirsutum glands and that its kinetic parameters

are quite different when it acts as part of such a complex. The

existence of such a complex may also relate to the intriguing

question of how the substrates ofMKS1,which are intermediates

in the fatty acid biosynthetic pathway, become accessible to

MKS1 and why in L. hirsutum f glabratum the main products are

2-tridecanone, 2-undecanone, and 2-pentadecanone (and not

smaller or larger methylketones). Furthermore, although MKS1

showed similar specificity to the three different b-ketoacyl-ACPs

that lead to the formation of 2-undecanone, 2-tridecanone, and

2-pentadecanone (Figure 11), the three methylketones exist in

different quantities in the glands (Table 1, Figure 3C). It is in-

teresting that in one of the L. hirsutum f glabratum accessions,

LA0407, the ratio between 2-tridecanone and 2-undecanone

is the opposite of that found in the rest of this subspecies, with

2-undecanone being the major methylketone in the leaf

(Antonious, 2001).Yet, thesequenceof theLA0407alleleofMKS1

is identical to that of MKS1 from PI126449 (E. Fridman and E.

Pichersky, unpublished data). This suggests that the variation

in the abundance of the different methylketones in the glands

(and the real turnover rate of the enzyme) may be determined

by the availability of the substrate rather than the specificity of

MKS1 for the different substrates.

Relationship of MKS1 to Other a/b-Hydrolase Enzymes

MKS1 is part of the a/b-hydrolase superfamily that includes

several enzymes with a proven esterase function. These include,

in addition to PNAE and tomato MJE, potato (Solanum tuber-

osum) MJE (accession number AY684102) and PIR7, a protein

from rice (Oryza sativa) that was shown to have esterase activity

with artificial substrates (Waspi et al., 1998). In addition, the

tobacco (Nicotiana tabacum) SABP2 protein, which has recently

been shown to be methyl salicylate esterase (Forouhar et al.,

2005), and an uncharacterized protein encoded by an ethylene-

induced citrus EST (Zhong et al., 2001), as well as 20 proteins

encoded in the Arabidopsis genome, show similarity to MKS1

(Figure 12). PIR7, PNAE, SABP2, and MJE all appear to have

defense-related functions. For example, the defense-related rice

Pir7 transcripts were found to accumulate upon infiltration with

the resistance-inducing Pseudomonas syringae pv syringae

(Waspi et al., 1998), and PNAE is involved in the biosynthesis

of defense alkaloids in the Indian medicinal plant R. serpentina.

MJE may be involved in the jasmonate signal transduction

pathway that turns on many defense genes (Sasaki et al., 2001;

Stuhlfelder et al., 2004). The role of MKS1 in the biosynthesis of

the highly protective methylketone is therefore not surprising.

However, the origin ofMKS1 during evolution and the prevalence

of MKS1-related sequences in the plant kingdom (and else-

where) remain to be determined.

METHODS

Plant Material

Seeds for the different tomato (Lycopersicon) accessions were obtained

from the Tomato Seedstock Center at the University of California (Davis,

CA) (LA accessions) and from the USDA–Agricultural Research Service

(Ithaca, NY) (PI accessions). Seeds were germinated in the dark for 3 d at

258C, after which they were transferred to a growth chamber for 3 weeks

at 258C with 18 h of light. Three-week-old plants were transplanted to

pots and arranged in a randomized block design in the greenhouse at

258C with 18 h of light. For gland extractions, plants were propagated by

rooting cuttings, using the rooting powder Hormorex (0.10% indole-3-

butyric acid; Brooker Chemical, Chatsworth, CA). Cuttings were rooted

under a plastic bag at 258C with 18 h of light for 2 weeks, after which time

plants were exposed gradually to normal greenhouse conditions and kept

in 1-liter pots containing Sunshine Mix 1 (Sun Gro Horticulture, Bellevue,

WA) potting soil at 258C with 18 h of light.

Scanning Electron Microscopy

Young leaves of L. hirsutum f glabratum were prepared for scanning

electron microscopy as previously described (Gang et al., 2001).

Leaf Gland Trichome Isolation and Removal for Protein Extraction,

Volatile Analysis, and RNA Isolation

Leaf glandular trichomes were isolated from young leaves of different

tomato accessions following the procedure described previously (Gang

et al., 2001) with a fewmodifications. The glands were collected on a final

40-mmmesh cloth, and the yield was;1 to 1.5 mL of packed glands per

15 g of leaf sample. Glands were left on ice for 15 min to settle down and

were used either for analysis of essential oil constituents or for protein

extraction. The crude protein extract of the glands was obtained by

resuspending the packed glands in ice-cold gland lysis buffer (10 mM

NaCl, 50 mM Tris-HCl, pH 8.0, 1 mM EDTA, 10 mM b-mercaptoethanol,

and 10% glycerol) followed by sonication for 2 min and collection of the

Tomato Methylketone Synthesis 1263

supernatant after centrifugation at 14,000g at 48C for 20 min. For RNA

isolation, the glands were isolated in a similar way, except that aurin-

tricarboxylic acid (1 mM) was added to all buffers and solutions. After

letting the glands settle down on ice for 15 min, the overlaying gland

isolation buffer was removed and 450 mL of buffer RLT (Qiagen, Valencia,

CA) was added for each 100 mg of glands. The glands were lysed by a

30-s sonication pulse, and the total RNA was purified using the RNeasy

plant mini kit (Qiagen).

Leaf and stem samples were collected from five independent PI126449

plants that were randomly planted in pots with other tomato accessions.

Partial removal of the glands from leaves and stems was performed

in liquid N2 with a painter’s brush before weight measurements and

extractions. Compared leaves, with or without treatment, were opposite

leaflets on the same leaf, and compared stem samples were from the

same stem.

cDNA Library Construction, Sequencing, and EST Analysis

Adirectional cDNA librarywas constructed aspreviously described (Gang

et al., 2001), and recombinant plasmids from3840colonies (103384-well

plates) were randomly and automatically isolated and sequenced from

both 59- and 39-ends, using the T3 and T7 primers, respectively. After

removal of vector and poor quality sequences, the remaining 5496 EST

sequences were compared against the SWISS-PRO and nonredundant

databases using the tBLASTn search algorithms. The contig assemblies

and the singletons were assigned specific functions based on highest

similarity and each was BLAST searched against the L. hirsutum (acces-

sion LA1777) trichome EST database (The Institute for Genomic Re-

search, http://www.tigr.org/tdb/Igi/) to obtain the number of ESTs for

each gene.

GC-MS Analysis of Volatiles

Plant samples were weighted and then placed in 5-mL glass vials

containing 1 mL of methyl-tert-butyl ether. The vials were sealed with

rubber septa cups and mildly shaken for 2 h. Toluene was added as an

internal standard (0.03%), and the resulting extract was used for GC-MS

analysis. GC-MS was performed as described previously (Gang et al.,

2001). Three microliters of the extract, with no prior concentration, were

analyzedbyGC-MS for determination of the essential oil constituents. The

identities of the main methylketones and terpenes were verified with

commercially available standards. For determination of the volatiles in the

glands, 100 mg of packed glands (see extraction protocol above) were

extracted twicewith 200mLof hexane, followedbyconcentrationandGC-

MS analysis.

Chloroplast Import Assays

Pea plants (Pisum sativum var Little Marvel; Olds Seed, Madison, WI)

were grown under natural light in the greenhouse at 18 to 208C.

Chloroplasts were isolated from 8- to 12-d-old plants as described

previously (Perry and Keegstra, 1994) . MKS1, ARC6 (Vitha et al., 2003),

and RBCS (Olsen and Keegstra, 1992) genes were transcribed, trans-

lated, and labeled with [35S]-Met using the TNT-coupled Reticulocyte

Lysate system (Promega,Madison, WI). Import assays were performed in

450 mL of import buffer (50 mM Hepes-KOH, pH 8.0; 330 mM sorbitol)

containing 75 mL of chloroplasts, 4 mM Mg-ATP, and 5 3 106 dpm

precursor protein for 30min, at room temperature, in the presence of light.

Chloroplasts were recovered by sedimentation through a 40% Percoll

cushion and then incubated with (þ) or without (�) thermolysin or trypsin

for 30 min at 48C as previously described (Jackson et al., 1998), with or

without Triton X-100. Intact chloroplasts not treated with Triton X-100

were again recovered by centrifugation through a 40% Percoll cushion

then lysed in a solution containing 25mMHepes, pH 8.0, and 5mMEDTA

and fractionated into stroma and total membrane (including outer

envelope, inner envelope, and thylakoid). Samples treated with Triton

X-100 were centrifuged at 100,00g and separated to pellet and superna-

tant fractions. All import reactions were analyzed by SDS-PAGE. After

electrophoresis, the gels were subjected to fluorography and exposed to

x-ray film (Eastman-Kodak, Rochester, NY).

RNA Extraction and Gel Blot Analysis

RNA from glands was isolated using the RNeasy plant mini kit (Qiagen) as

described above, and leaf RNAwas isolated using TRIzol reagent (Gibco-

BRL, Grand Island, NY) according to the manufacturer’s protocol. RNA

gel blots were performed as described previously (D’Auria et al., 2002)

using 2 and 5 mg of RNA in each lane for glands and leaf samples,

respectively. An MKS1 probe was synthesized with the rediprime II kit

(Amersham Pharmacia Biotech, Piscataway, NJ) using a PCR-amplified

fragment of MKS1.

Protein Expression and Purification

The following primers were used to amplify and clone the full ORF of

MKS1 from a PI126449 leaf cDNA into the Escherichia coli expression TA

cloning vector (pCRT7/CT TOPO-TA; Invitrogen, Carlsbad, CA): forward,

59-AATGGAGAAAAGCATGTCGC-39; reverse, 59-GCTCTTTATTTATAC-

TTGTTAGC-39. PCR conditions were 30 s at 948C, followed by 25 cycles

of 20 s at 948C, 20 s at 568C, 40 s at 688C, and an additional 10 min at

688C. The full-length MKS1 construct was transformed into BL21(DE3)

Codonþ cells, and a single transformed colony was grown at 378C

overnight in 2 mL of LB media containing 100 and 34 mg/mL of ampicillin

and chloramphenicol, respectively. The culture was then used to start

a 200-mL culture (with the same antibiotics) that was grown at 378C until

an A600 value of 0.4 was obtained. After induction with 0.5 mM isopropyl

1-thio-b-galactopyranoside, the culture was grown overnight at 188C.

Cells were pelleted, harvested, and resuspended in 20 mL of lysis buf-

fer (10 mM NaCl, 50 mM Tris-HCl, pH 8.0, 1 mM EDTA, 10 mM

b-mercaptoethanol, and 10% [v/v] glycerol). After sonication for 2 min

and centrifugation at 14,000g for 20 min at 48C, the supernatant was

collected and kept at �808C for further applications.

A plasmid containing the spinach (Spinacia oleracea) ACP isoform I

gene using codons preferred byE. coli, with aC-terminal His-tag (pSACP-

1a; Broadwater and Fox, 1999), was kindly provided to us by Brian Fox

(University of Wisconsin, Madison). The plasmid was transformed into

BL21(DE3) LysS cells, and a single transformed colonywas grown at 378C

overnight in 20 mL of LB medium containing 100 and 34 mg/mL of ampi-

cillin and chloramphenicol, respectively. The culture was then used to

start a 500-mL terrific-broth culture (Tartof and Hobbs, 1987) (with the

same antibiotics) that was grown at 378Cuntil anA600 value of 1.0 was ob-

tained. After induction with 0.5 mM isopropyl 1-thio-b-galactopyranoside,

the culture was grown overnight at 188C. Cells were pelleted,

harvested, and resuspended in 50 mL of ACP lysis buffer (20 mM Tris-

HCl, pH 7.9, 250 mM NaCl, 10 mM imidazole, 1% Tween 20, 10 mM

b-meracptoethanol, and 10% [v/v] glycerol). The lysate was stirred on ice

for 1 h with 50mgmL�1 of lysozyme and DNase (Sigma-Aldrich, St. Louis,

MO). After sonication for 2 min and centrifugation at 30,000g for 20 min at

48C, the supernatant was collected and kept at �808C until purification.

MKS1 was purified from E. coli cell lysates by diethylaminoethyl-

cellulose anion exchange column chromatography (DE53; Sigma-Aldrich)

followed by chromatography on a Hi-Trap Phenyl HP (hydrophopic

interactions) column (Amersham Pharmacia Biotech) connected to

a fast protein liquid chromatography system. MKS1 did not bind to

DE53 and eluted with the flow-through, thus achieving a substantial

purification. It boundwell to aPhenyl HPcolumn, eluting at;0.3 to 0.35M

1264 The Plant Cell

(NH4)2SO4. Fractions with highest concentrations of MKS1 eluting from

the Phenyl HP column were concentrated by centrifugation as described

previously (Iijima et al., 2004a).

Spinach ACP was purified from E. coli cell lysates on a column

containing 2 mL of packed Ni2þ-NTA resin (Qiagen). The column was

equilibrated with 40 mL of double distilled water, followed by 40 mL of

ACP lysis buffer before loading all 50 mL of lysate. The column was then

washed once with 20 mL of ACP lysis buffer followed by a 40 mL wash

with the same buffer but with no Tween 20. The His-tagged ACP was

eluted with 4 mL of ACP lysis buffer with no Tween 20 and with 250 mM

imidazole. The eluted 0.5-mL fractions were analyzed on a 15% SDS-

PAGE, and fractions with high levels of ACP-His6 were pooled and

dialyzed against 100 to 200 volumes of 10 mMMes, pH 6.1, with 0.5 mM

dithioerythritol overnight at 48C and stored at �808C for further applica-

tions.

The acyl-acyl carrier protein synthase enzyme (Aas) was over-

expressed and purified from E. coli cell line BL21(DE3) LysS, using the

plasmid and procedures described previously (Shanklin, 2000).

Preparation of MKS1 Antibodies and Protein Gel Blots

One milligram of the purified protein (see Supplemental Figure 1 online,

lane 5) was electrophoresed on a 13% SDS-PAGE gel, stained with

Coomassie Brilliant Blue, and the protein excised from the gel. Antibodies

were prepared by Cocalico Biologicals (Reamstown, PA) by injecting

macerated gel fragments containing the purified MKS1 and following the

company’s protocol. Specificity of the antibody was tested with the

preinjection serum and with different purified proteins (data not shown).

Anti-MKS1 antibodies were used at a 1:2000 dilution and incubated with

the protein gel blots for 2 to 16 h. All other conditions of the protein gel

blots were performed as described previously (Dudareva et al., 1996).

Substrate Preparation, Enzyme Assays, and Product Identification

The coupled enzymatic assaywedevelopedwas done in three steps, with

the first two steps needed to produce the b-ketoacyl-ACP substrate for

the reaction (third step) catalyzed by the putativeMKS (see Supplemental

Figure 2 online). The following is a description of the synthesis of [3-14C]

b-ketomyristoyl-ACP from the starting material [1-14C] 12:0 lauric acid.

The same procedure was used for the synthesis of [3-14C] b-ketolauroyl-

ACP or [3-14C] b-ketopalmitoyl-ACP, starting from [1-14C] 10:0 capric

acid or [1-14C] 14:0 myristic acid, respectively.

In the first step, [1-14C] 12:0 lauric acid (ARC 257; American Radio-

labeled Chemicals, St. Louis, MO) is converted to an acyl-ACP in

a reaction containing spinach ACP and an E. coli acyl-ACP synthetase

(Rock andGarwin, 1979). (ThisE. coli acyl-ACP synthetase couplesC12:0

and C14:0 fatty acids efficiently, but it has low coupling efficiency with

decanoic acid [C10:0], thereby limiting the amount of [3-14C] b-lauroyl-

ACP that we were ultimately able to make). Because the ACP that was

used in these syntheses carried a His-tag, a Ni2þ-NTA column (0.5 mL;

Qiagen) was used to purify the acyl-ACP product. Before loading the

samples, the column was equilibrated with 20 volumes (10 mL) of double

distilled water, followed by 20 volumes of 10 mM Mes, pH 6.1. After

loading the synthesized acyl-ACP, the column was washed with 10 and

20 volumes of 10 mM Mes, pH 6.1, with and without 10 mM imidazole,

respectively. The purified substrate was eluted with 2 mL of 10 mMMes,

pH 6.1, with 250 mM imidazole, and the 250-mL fractions were subjected

to scintillation counting to detect the active fractions, which were kept at

�808 for further applications.

In the next step, [3-14C] b-ketomyristoyl-ACP is synthesized by enzy-

matic condensation of malonyl-ACP and the [1-14C] (12:0) lauroyl-ACP

obtained in the previous reaction. This stepmakes use of the E. coli crude

extract as the source for several enzymes in fatty acid biosynthesis. The

relevant enzymes in these extracts are the malonyl CoA:ACP transacyl-

ase, which converts the supplied malonyl-CoA and ACP to malonyl-ACP,

and the b-ketoacyl-ACP synthase I, which condenses the malonyl-ACP

and the [1-14C] lauroyl-ACP to [3-14C] b-ketomyristoyl-ACP (Figure 4A).

The success of this step is monitored by reacting an aliquot with the

reducing agent NaH4B, followed by radio–thin layer chromatography

(TLC) with standards (Garwin et al., 1980). Any [1-14C] lauroyl-ACP (the

product of step 1) that failed to elongate, as well as free [1-14C] lauric acid

(the starting material in step 1) is reduced to C12 alcohol (see Supple-

mental Figure 3online), and theproduct of step 2, [3-14C]b-ketomyristoyl-

ACP, is reduced to C14 1,3-diol (see Supplemental Figure 3 online). The

substrate was then purified by adding 2.5% trichloroacetic acid, in-

cubating on ice for 10min, and centrifuging at 14,000g for 5 min, followed

by a gradual resuspension of the pellet with 100mMTris-HCl to pH 6 to 7.

After centrifugation for 20 s, the supernatant was collected and its activity

was determined by counting 2 mL twice in a scintillation counter. The

[3-14C] b-ketomyristoyl-ACP was then supplied to the MKS1 reaction

assay solution that included the tested enzyme (MKS1, gland crude

protein extract, or other controls) with 30 mM Tris-HCl, pH 8.0, in a final

volume of 100 mL. After incubation at room temperature for 30 min, the

reaction was stopped and product was extracted by adding 200 mL of

hexane and vortexing for 10 s, followed by centrifugation for 5 min at

maximum speed. The organic phase was collected, concentrated,

spotted on a silica TLC and developed with hexane:ether:acetic acid in

a 80:20:1 ratio and comparedwith a 2-tridecanone standard that was also

applied to the plate. The TLC plates were scanned with an Instant Imager

(Packard Instrument, Meriden, CT) overnight, and the cold 2-tridecanone

standard was detected by spraying the TLC plate with 0.4% 2,4-

dinitrophenyl-hydrazine (catalog number D199303; Sigma-Adrich) in 2 N

HCl. The product was also identified by GC-MS analysis (see Supple-

mental Figure 4 online).

Steady state kinetic constants for MKS1 with the different [3-14C]

b-ketoacyl-ACP substrates were determined by fitting initial velocity

versus substrate concentrations to the hyperbolic Michaelis-Menten

equation using the software Hyper32 version 1.0.0.0 (downloaded from

the Web site of J.S. Easterby, University of Liverpool, UK; http://

homepage.ntlworld.com/john.easterby/). For the kinetics analysis, ap-

propriate enzyme concentrations and incubation times were determined

by time-course enzyme assays so that the reaction velocity was linear

during the reaction period.

Sequence Alignment and Construction of the Phylogenetic Tree

Sequence alignment and construction of the phylogenetic tree were

performed using ClustalX (version 1.81), and the neighbor-joining method

was used with 1000 bootstrap trials. The full actual alignment for the tree

is included in Supplemental Figure 5 online. TreeView was used to

visualize the resulting tree (Page, 1996). A maximum parsimony tree was

also generated using the PAUP* program (Sinauer Associates, Sunder-

land, MA), and it showed the same branches as those in the neighbor-

joining tree.

Access to the Glandular Trichome EST Database

The PI126449 glandular trichome EST database, including a BLAST

search engine, can be accessed through our Web site (http://www.

biology.lsa.umich.edu/research/labs/pichersky/).

Sequence data from this article have been deposited with the EMBL/

GenBank data libraries under the following accession numbers: MKS1

from L. hirsutum f glabratum (accession PI126449), AY701574; MJE from

S. tuberosum (potato), AY684102.

Tomato Methylketone Synthesis 1265

ACKNOWLEDGMENTS

We thank Juie Mahajan and Zachary Szpiech (University of Michigan,

Ann Arbor, MI) for their technical assistance, and Na’ama Menda

(Hebrew University, Rehovot, Israel) for her valuable assistance in the

bioinformatic analysis of the EST database. We also thank John

Shanklin (Brookhaven National Laboratory, Upton, NY) for the AasH

plasmid, Brian Fox (University of Wisconsin, Madison, WI) for the ACP

clones, and Dave Kudrna and HyeRan Kim (Arizona Genomics Institute,

University of Arizona, Tucson, AZ) for their assistance in sequencing.

This research was supported by the Michigan Life Science Corridor

Grant 085P1000466 and by the National Research Initiative of the USDA

Cooperative State Research, Education, and Extension Service Grant

2004-35318-14874 to E.P. J.E.F. was supported in part by Grant MCB

0316262 from the National Science Foundation and by the Division of

Energy Biosciences at the U.S. Department of Energy. E.F. was sup-

ported in part by a Binational Agricultural Research and Development

Fund Postdoctoral Fellowship Grant FI-328-2002.

Received November 27, 2004; accepted February 12, 2005.

REFERENCES

Alban, C., Baldet, P., and Douce, R. (1994). Localization and charac-

terization of two structurally different forms of acetyl-CoA carboxylase

in young pea leaves, of which one is sensitive to aryloxyphenoxypro-

pionate herbicides. Biochem. J. 300, 557–565.

Antonious, G.F. (2001). Production and quantification of methyl ketones

in wild tomato accessions. J. Environ. Sci. Health B 36, 835–848.

Bennet, R.N., and Wallsgrove, R.M. (1994). Secondary metabolites in

plant defense-mechanism. New Phytol. 124, 617–633.

Broadwater, J.A., and Fox, B.G. (1999). Spinach holo-acyl carrier

protein: Overproduction and phosphopantetheinylation in Escherichia

coli BL21(DE3), in vitro acylation, and enzymatic desaturation of

histidine-tagged isoform I. Protein Expr. Purif. 15, 314–326.

Carter, C.D., and Snyder, J.C. (1985). Mite responses in relation to

trichomes of Lycopersicon esculentum x Lycopersicon hirsutum F2

hybrids. Euphytica 34, 177–185.

D’Auria, J.C., Chen, F., and Pichersky, E. (2002). Characterization of

an acyltransferase capable of synthesizing benzylbenzoate and other

volatile esters in flowers and damaged leaves of Clarkia breweri. Plant

Physiol. 130, 466–476.

Dimock, M.B., and Kennedy, G.G. (1983). The role of gladular

trichomes in the resistance of Lycopersicon hirsutum f. glabaratum

to Heliothis zea. Entomol. Exp. Appl. 33, 263–268.

Dogru, E., Warzecha, H., Seibel, F., Haebel, S., Lottspeich, F., and

Stockigt, J. (2000). The gene encoding polyneuridine aldehyde

esterase of monoterpenoid indole alkaloid biosynthesis in plants is

an ortholog of the alpha/beta hydrolase super family. Eur. J. Biochem.

267, 1397–1406.

Dudareva, N., Cseke, L., Blanc, V.M., and Pichersky, E. (1996).

Evolution of floral scent in Clarkia: Novel patterns of S-linalool

synthase gene expression in the C. breweri flower. Plant Cell 8,

1137–1148.

Ege, S.N. (1989). Enols and enolate anions as nucleophiles I. Haloge-

nation, alkylation, and condesation reactions. In Organic Chemistry,

S.N. Ege, ed (Lexington, MA: Heath), pp. 628–674.

Forney, F.W., and Markovetz, A.G. (1971). The biology of methyl

ketones. J. Lipid Res. 12, 383–395.

Forouhar, F., Yang, Y., Kumar, D., Chen, Y., Fridman, E., Park, S.W.,

Chiang, Y., Acton, T.A., Montelione, G.T., Pichersky, E., Klessig,

D.F., and Tong, L. (2005). Crystal structure and biochemical studies

identifies tobacco salicylic binding protein 2 (SABP2) as a methyl

salicylate esterase and further implicate it in innate immunity. Proc.

Natl. Acad. Sci. USA 102, 1773–1778.

Frentzen, M., Heinz, E., Mckeon, T.A., and Stumpf, P.K. (1983).

Specificities and selectivities of glycerol-3-phosphate acyltransferase

and monoacylglycerol-3-phosphate acyltransferase from pea and

spinach chloroplasts. Eur. J. Biochem. 129, 629–636.

Fridman, E., and Pichersky, E. (2005). Metabolomics, genomics,

proteomics, and the identification of enzymes and their substrates

and products. Curr. Opin. Plant Biol., in press.

Froehlich, J.E., Itoh, A., and Howe, G.A. (2001). Tomato allene oxide

synthase and fatty acid hydroperoxide lyase, two cytochrome P450s

involved in oxylipin metabolism, are targeted to different membranes

of chloroplast envelope. Plant Physiol. 125, 306–317.

Gang, D.R., Lavid, N., Zubieta, C., Chen, F., Beuerle, T., Lewinsohn,

E., Noel, J.P., and Pichersky, E. (2002). Characterization of phenyl-

propene O-methyltransferases from sweet basil: Facile change

of substrate specificity and convergent evolution within a plant

O-methyltransferase family. Plant Cell 14, 505–519.

Gang, D.R., Wang, J., Dudareva, N., Nam, K.H., Simon, J.E.,

Lewinsohn, E., and Pichersky, E. (2001). An investigation of the

storage and biosynthesis of phenylpropenes in sweet basil. Plant

Physiol. 125, 539–555.

Garwin, J.L., Klages, A.L., and Cronan, J.E., Jr. (1980). Structural,

enzymatic, and genetic studies of beta-ketoacyl-acyl carrier pro-

tein synthases I and II of Escherichia coli. J. Biol. Chem. 255, 11949–

11956.

Graham, I.A., and Eastmond, P.J. (2002). Pathways of straight and

branched chain fatty acid catabolism in higher plants. Prog. Lipid Res.

41, 156–181.

Hwang, D.H., Lee, Y.J., and Kinsella, J.E. (1976). b-Ketoacyl decar-

boxylase activity in spores and mycelium of Penicillium roqueforti. Int.

J. Biochem. 7, 165–171.

Iijima, Y., Davidovich-Rikanati, R., Fridman, E., Gang, D.R., Bar, E.,

Lewinsohn, E., and Pichersky, E. (2004b). The biochemical and

molecular basis for the divergent patterns in the biosynthesis of

terpenes and phenylpropenes in the peltate glands of three cultivars

of basil. Plant Physiol. 136, 3724–3736.

Iijima, Y., Gang, D.R., Fridman, E., Lewinsohn, E., and Pichersky, E.

(2004a). Characterization of geraniol synthase from the peltate glands

of sweet basil. Plant Physiol. 134, 370–379.

Jackson, D.T., Froehlich, J.E., and Keegstra, K. (1998). The hydro-

philic domain of Tic110, an inner envelope membrane component of

the chloroplastic protein translocation apparatus, faces the stromal

compartment. J. Biol. Chem. 273, 16583–16588.

Jasperson, H., and Jones, R. (1947). Some unsaponifiable constitutes

of deodorization distillate of vegetables oils. J. Soc. Chem. Ind. 66,

13–17.

Kauffman, W.C., and Kennedy, G.G. (1989). Inhibition of Campoletis

sonorensis parasitism of Heliothis zea and of parsitoid development

by 2-tridecanone mediated insect resistance of wild tomato. J. Chem.

Ecol. 15, 1919–1930.

Kennedy, G.G. (2003). Tomato, pests, parasitoids, and predators:

Tritrophic interactions involving the genus Lycopersicon. Annu. Rev.

Entomol. 48, 51–72.

Li, L., Zhao, Y., McCaig, B.C., Wingerd, B.A., Wang, J., Whalon, M.E.,

Pichersky, E., and Howe, G.A. (2003). The tomato homolog of

CORONATINE-INSENSITIVE1 is required for the maternal control of

seed maturation, jasmonate-signaled defense responses, and glan-

dular trichome development. Plant Cell 16, 126–143.

Lin, S.Y.H., Trumble, J.T., and Kumamoto, J. (1987). Activity of volatile

compounds in glandular trichomes of Lycopersicon species aganist

two insect herbivores. J. Chem. Ecol. 13, 837–850.

1266 The Plant Cell

Luckwill, L.C. (1943). The Genus Lycopersicon: A Historical, Biological

and Taxonomic Survey of the Wild and Cultivated Tomato. (Aberdeen,

Scotland: Aberedeen University Press).

Marshall, S.D., Putterill, J.J., Plummer, K.M., and Newcomb, R.D.

(2003). The carboxylesterase gene family from Arabidopsis thaliana.

J. Mol. Evol. 57, 487–500.

McGarvey, D.J., and Croteau, R. (1995). Terpenoid metabolism. Plant

Cell 7, 1015–1026.

McKeon, T.A., and Stumpf, P.K. (1982). Purification and characteriza-

tion of the stearoyl-acyl carrier protein desaturase and the acyl-acyl

carrier protein thioesterase from maturing seeds of safflower. J. Biol.

Chem. 257, 12141–12147.

Mekhedov, S., de Ilarduya, O.M., and Ohlrogge, J. (2000). Toward

a functional catalog of the plant genome. A survey of genes for lipid

biosynthesis. Plant Physiol. 122, 389–402.

Miras, S., Salvi, D., Ferro, M., Grunwald, D., Garin, J., Joyard, J., and

Rolland, N. (2002). Non-canonical transit peptide for import into the

chloroplast. J. Biol. Chem. 277, 47770–47778.

Mukherjee, S. (1951). Studies on degradation of fats by microorgan-

isms. I. Preliminary investigations on enzyme system involve in

spoilage of fats. Arch. Biochem. Biophys. 33, 364–376.

Olsen, L.J., and Keegstra, K. (1992). The binding of precursor proteins

to chloroplasts requires nucleoside triphosphates in the intermem-

brane space. J. Biol. Chem. 267, 433–439.

Page, R.D.M. (1996). TREEVIEW: An application to display phylogenetic

trees on personal computers. Comput. Appl. Biosci. 12, 357–358.

Perry, S.E., and Keegstra, K. (1994). Envelope membrane proteins that

interact with chloroplastic precursor proteins. Plant Cell 6, 93–105.

Rathinasabapathi, B., McCue, K.F., Gage, D.A., and Hanson, A.D.

(1994). Metabolic engineering of glycine betaine synthesis: Plant

betaine aldehyde dehydrogenases lacking typical transit peptides

are targeted to tobacco chloroplasts where they confer betaine

aldehyde resistance. Planta 193, 155–162.

Rock, C.O., and Garwin, J.L. (1979). Preparative enzymatic synthesis

and hydrophobic chromatography of acyl-acyl carrier protein. J. Biol.

Chem. 254, 7123–7128.

Roughan, P.G. (1997). Stromal concentrations of coenzyme A and its

esters are insufficient to account for rates of chloroplast fatty acid

synthesis: Evidence for substrate channelling within the chloroplast

fatty acid synthase. Biochem. J. 327, 267–273.

Roughan, P.G., and Ohlrogge, J.B. (1996). Evidence that isolated

chloroplasts contain an integrated lipid-synthesizing assembly that

channels acetate into long-chain fatty acids. Plant Physiol. 110, 1239–

1247.

Sasaki, Y., et al. (2001). Monitoring of methyl jasmonate-responsive

genes in Arabidopsis by cDNA macroarray: Self-activation of jas-

monic acid biosynthesis and crosstalk with other phytohormone

signaling pathways. DNA Res. 8, 153–161.

Sengoku, T. (1933). Crystalline constituent of essential oil of matsubasa

(Shizandra nigra, maxim). J. Pharm. Soc. Jap. 53, 947–951.

Shanklin, J. (2000). Overexpression and purification of the Escherichia

coli inner membrane enzyme acyl-acyl carrier protein synthase in an

active form. Protein Expr. Purif. 18, 355–360.

Sorm, F., Mleziva, J., and Arnold, Z. (1949). On terpenes. XII. On

the composition of oil of hops. Collect. Czech. Chem. Commun. 14,

693–698.

Stuhlfelder, C., Lottspeich, F., and Mueller, M.J. (2002). Purification

and partial amino acid sequences of an esterase from tomato.

Phytochemistry 60, 233–240.

Stuhlfelder, C., Mueller, M.J., and Warzecha, H. (2004). Cloning and

expression of a tomato cDNA encoding a methyl jasmonate cleaving

esterase. Eur. J. Biochem. 271, 2976–2983.

Suh, M.C., Schultz, D.J., and Ohlrogge, J.B. (1999). Isoforms of acyl

carrier protein involved in seed-specific fatty acid synthesis. Plant J.

17, 679–688.

Tang, Y., Lee, T.S., Kobayashi, S., and Khosla, C. (2003). Ketosyn-

thases in the initiation and elongation modules of aromatic polyketide

synthases have orthogonal acyl carrier protein specificity. Biochem-

istry 42, 6588–6595.

Tartof, K.D., and Hobbs, C.A. (1987). Improved media for growing

plasmid and cosmid clones. Bethesda Res. Lab Focus 9, 12.

van Der Hoeven, R.S., Monforte, A.J., Breeden, D., Tanksley, S.D.,

and Steffens, J.C. (2000). Genetic control and evolution of ses-

quiterpene biosynthesis in Lycopersicon esculentum and L. hirsutum.

Plant Cell 12, 2283–2294.

Vitha, S., Froehlich, J.E., Koksharova, O., Pyke, K.A., van Erp, H.,

and Osteryoung, K.W. (2003). ARC6 is a J-domain plastid division

protein and an evolutionary descendant of the cyanobacterial cell

division protein Ftn2. Plant Cell 15, 1918–1933.

Walbaum, H., and Huthig, O. (1902). Ueber das Ceylon-Zimmtol.

J. Prakt. Chem. 66, 47–58.

Walker, K.A., and Harwood, J.L. (1985). Localization of chloroplastic

fatty acid synthesis de novo in the stroma. Biochem. J. 226, 551–556.

Waspi, U., Misteli, B., Hasslacher, M., Jandrositz, A., Kohlwein, S.D.,

Schwab, H., and Dudler, R. (1998). The defense-related rice gene

Pir7b encodes an alpha/beta hydrolase fold protein exhibiting ester-

ase activity towards naphthol AS-esters. Eur. J. Biochem. 254, 32–37.

Watts, F. (1886). On the essential oil of lime leaves (Citrus limetta).

Preliminary notice. J. Chem. Soc. London Trans. 49, 316–317.

Wehmer, C. (1931). Die Pflanzenstoffe. (Ann Arbor, MI: Edwards

Brothers).

Williams, W.G., Kennedy, G.G., Yamamoto, R.T., Thacker, J.D., and

Bordner, J. (1980). 2-tridecanone: A naturally occurring insecticide

from the wild tomato species Lycopersicon hirsutum f. glabratum.

Science 207, 888–889.

Yu, H., Kowalski, S.P., and Steffens, J.C. (1992). Comparison of

polyphenol oxidase expression in glandular trichomes of Sulanum and

Lycopersicon species. Plant Physiol. 100, 1885–1890.

Zhong, G.Y., Goren, R., Riov, J., Sisler, E.C., and Holland, D. (2001).

Characterization of an ethylene-induced esterase gene isolated from

Citrus sinensis by competitive hybridization. Physiol. Plant. 113,

267–274.

Tomato Methylketone Synthesis 1267