Systematic chemical nomenclature does not give functional group positions in accord with biosynthetic pathways. For example, octadeca-cis-6, cis-9-diene-1,18-dioate is derived directly by oxidation at C(18) of linoleate (octadeca-cis-9, cis-12-dieneoate), but because of the 1,18-symmetry of the two carboxylate groups the designation of the position of the double bonds changes from 9,12 to 6,9. Another example encountered in the text is 1,7,16-trihydroxyhexadecane, which is directly derived by reduction of 10,16-dihydroxypalmitate.
Although the surface waxes from Arabidopsis thaliana leaves and stems have been thoroughly characterized, the monomer composition of the polyesters of the cuticular membrane has not been analyzed. Delipidated Arabidopsis leaves or stems, when depolymerized under conditions to cleave polyesters, produced typical ω–hydroxy fatty acid cutin monomers such as 16-hydroxy-palmitate, 10,16-dihydroxy-palmitate and 18-hydroxy-9,10-epoxy-stearate. However, the major monomer was octadeca-cis-6, cis-9-diene-1,18-dioate, with lesser amounts of octadec-cis-9-ene-1,18-dioate and hexadeca-1,16-dioate. These dicarboxylates were found predominantly in epidermal peels from Arabidopsis stems and are therefore likely to be associated with the cuticular membrane. They were also found in analyses of canola leaves but were absent in tomato and apple fruit cutins. In the fad2-1 mutant line of Arabidopsis, which has reduced levels of linoleate and linolenate and elevated oleate in cytosolic phospholipids, the amount of octadeca-cis-6, cis-9-diene-1,18-dioate was 50% reduced, with a concomitant increase in octadec-cis-9-ene-1,18-dioate. In a fatb-ko line of Arabidopsis, where the availability of cytosolic palmitate is impaired, there was an 80% loss of C16 monomers and a compensating increase in C18 monomers. The presence of substantial amounts of dicarboxylates in cuticular membranes is unexpected. High amounts of aliphatic dicarboxylates are usually considered as an indicator of suberin, and are reported only as very minor components of cutin. The high level of polyunsaturation is also unusual in cuticles; saturated fatty acid monomers usually predominate, with lesser amounts of monounsaturates. These novel findings for Arabidopsis demonstrate that a broad range of monomer compositions are possible for polyesters of the epidermis.
All aerial parts of vascular plants are protected from the environment by the cuticle, a lipophilic layer synthesized by epidermal cells (Kolattukudy, 1980). The cuticle is composed of a cutin polymer matrix with waxes embedded in the matrix and also deposited on its surface. The chemical composition of cutin and the associated waxes largely determine the physical properties of the plant cuticle. One of the main functions of the cuticle is to form a barrier for permeation of solutes, gases and water. It also provides mechanical support to plant organs (Hoffman-Benning and Kende, 1994) and affects the susceptibility of plants to pathogen attacks by providing both mechanical resistance and cellular signals for defense events (Schweizer et al., 1996). Moreover, the cuticle has a critical role in ensuring organ identity during development, preventing the fusion of cell walls from adjacent organs (Sieber et al., 2000).
Cutin is usually defined as an insoluble polymer composed mainly of aliphatic monomers derived from fatty acids and cross-linked by ester and ether bonds (Kolattukudy, 1980). Minor amounts of phenolics, such as p-coumaric acid, and carbohydrates have also been reported as structural components of cutins (Fang et al., 2001; Kolattukudy, 1977). Recently, glycerol has been detected in the cuticle of several plant species and shown to be esterified to cutin fatty acid monomers (Graça et al., 2002). The major aliphatic monomers obtained after cleavage of ester bonds are ω-hydroxy C16 and C18 fatty acids. These monomers frequently contain additional mid-chain hydroxyl, keto or epoxy functional groups. The residue left after depolymerization, often referred to as cutan, can amount to 10–60% by weight of the total isolated cuticle and is believed to contain ether-linked aliphatic components as well as carbohydrates (Nip et al., 1986; Schmidt and Schonherr, 1982). The primary hydroxyl groups in cutin are essentially all esterified but the mid-chain, secondary hydroxyl groups are found in both free and esterified form (Agullo et al., 1984). The extent to which cutin can be considered to have distinct linear, dendrimeric and anchored or crosslinked domains is unknown.
The biosynthesis of cutin is believed to be initiated by the cytochrome P450-catalyzed ω–hydroxylation of fatty acids. In this regard, labeling experiments performed on Vicia faba leaf disks demonstrated the sequential synthesis of 16-hydroxypalmitate and 10,16-dihydroxypalmitate from palmitate (Kolattukudy and Walton, 1972; Walton and Kolattukudy, 1972a). Using apple skins, oleate produced 18-hydroxyoleate, 9,10-epoxy-18-hydroxyoleate and finally 9,10,18-trihydroxystearate (Kolattukudy et al., 1973). The model for the assembly of cutin monomers into the final polymer proposes the esterification of carboxylate groups, as acyl-CoAs, with primary hydroxyl groups esterified for polymer elongation and with secondary hydroxyl groups for cross-linking (Croteau and Kolattukudy, 1974; Kolattukudy, 1977).
At present, several Arabidopsis thaliana mutants in cuticle synthesis have been characterized. These mutants show reduced thickness of the epidermal cuticular layer in leaves, increased cuticular permeability, reduced growth, induction of pollen germination on leaves, and in some cases organ fusion phenotypes. As an example, the lacerata (lcr) mutant of Arabidopsis is smaller than wild type, supports pollen germination on its leaf surfaces, has reduced fertility and shows a weak organ fusion phenotype (Wellesen et al., 2001). The genetic lesion of this mutant occurs in the LCR gene, which encodes a cytochrome P450 monooxygenase (CYP86A8) with the capacity for ω-hydroxylation of free fatty acids in vitro. A second example is the Arabidopsis wax2 mutant, which carries a mutation in a gene with high homology to the sterol desaturase and dehydrogenase/reductase family of proteins (WAX2) (Chen et al., 2003). This mutant also displays post-genital organ fusion, reduced fertility, and increased epidermal permeability. Finally, the Arabidopsis lacs2 mutant, which contains a mutation in an epidermis-specific long chain acyl-CoA synthetase, has a 40% reduction in the thickness of the cuticle layer of the abaxial leaf surface, supports pollen germination on leaves and has increased epidermis permeability (Schnurr et al., 2004). Although these mutations cause morphological changes in the structure of the cuticle no direct evidence for alterations in the composition or amount of cutin has been reported. This lack of information is caused, at least in part, by the lack of a reliable analytical method for cutin analysis in Arabidopsis. Nawrath (2002) has noted that general protocols for the analysis of cutin have not been successfully applied in Arabidopsis because the cuticle is thin and fragile. In this paper we demonstrate that bypassing the cuticle isolation step and applying the hydrogenolysis method for cutin depolymerization (Walton and Kolattukudy, 1972b) directly on delipidated Arabidopsis leaves and stems gives a reliable method for the determination of the polyester monomer composition of Arabidopsis epidermis. After depolymerization the products are analyzed by gas-chromatography mass-spectrometry (GC-MS). Epidermal polyesters of wild-type Arabidopsis and two mutants with altered fatty acid metabolism are analyzed and show alterations in the monomer composition consistent with their fatty acid phenotype. A notable feature of the analyses is a high level of dicarboxylate monomers.
Identification of polyester monomers from Arabidopsis leaves and stems
Standard methods for the analysis of cutin monomers begin with the isolation of the cuticle, and are followed by solvent extraction and then depolymerization of the extracted cuticular membrane (Holloway, 1982; Kolattukudy, 1977, 1980). However, the very thin and fragile cuticle found on Arabidopsis tissues (20–25 nm in leaves and 50–80 nm in stems) makes its isolation difficult (Nawrath, 2002). In addition, peeling the epidermis from Arabidopsis stem and leaf is a very laborious process. Because cutin monomers have very distinctive structures we chose to bypass the process of cuticular membrane preparation. Plant tissue was first delipidated by exhaustive solvent extractions and the remaining residue (containing cutin polyester) subsequently depolymerized by hydrogenolysis or transmethylation.
Depolymerization of Arabidopsis residues with LiAlH4 or LiAlD4. During hydrogenolysis with lithium aluminum hydride (LiAlH4) ester bonds are cleaved and the carboxylate groups are reduced to primary alcohols. In addition, functional groups such as aldehydes, ketones and oxiranes are reduced to primary or secondary alcohols. As a result, cutin monomers containing one or more hydroxyl groups are obtained. After derivatization the monomers are analyzed by GC-MS (Walton and Kolattukudy, 1972b). The hydrogenolysate product of Arabidopsis leaf and stem tissues was primarily composed of C16 and C18 polyhydroxy alkanes and alkenes (Figure 1a,c). The most abundant hydrogenolysis product in leaves was octadecadiene-1,18-diol, with lesser amounts of octadecene-1,18-diol, hexadecane-1,7,16-triol, hexadecane-1,16-diol and octadecene-1,9,18-triol and octadecane-1,9,18-triol. In addition, only traces of octadecane-1,9,10,18-tetraol and no long-chain (>C18) aliphatic compounds were detected in the polyol mixture. The polyester composition of stem residues showed a similar distribution of monomers to leaf residues (Figure S1a).
To further discriminate between the different oxidation states of substrate monomers that give identical products, isotopic labeling by deuteriolysis with lithium aluminum deuteride (LiAlD4) is employed (Walton and Kolattukudy, 1972b). For example, deuteriolysis of an ester would introduce two deuterium atoms into the product alcohol from the carboxylate moiety, increasing its molecular weight (MW) by two atomic mass units (amu). By contrast, the reduction of an aldehyde or keto group would introduce one D atom into the molecule at a single site, increasing by one amu its MW. Employing this method on the extracted residues from Arabidopsis leaves and stems revealed that octadecadiene-1,18-diol is primarily the product of a C18:2 dicarboxylic acid, with much smaller amounts arising from the corresponding ω-hydroxy fatty and very small amounts from the ω-oxo fatty acid and 1,ω-diol (mass spectra are discussed in detail below). Similarly, octadecene-1,18-diol derives predominantly from the dicarboxylic acid. Analysis of the mass spectra from hydrogenolysis and deuterolysis show that octadecane-1,9,18-triol and octadecene-1,9,18-triol are derived from 18-hydroxy-9,10-epoxy-stearate and -oleate, respectively; 1,16-hexadecanediol derives from both 16-hydroxypalmitate and hexadecane-1,16-dioate, while 1,7,16-hexadecanetriol derives largely from 10,16-dihydroxypalmitate (see Supplementary material).
GC-MS analysis of the hydrogenolysis products of the organic soluble fraction obtained from exhaustive extractions of leaf and stem tissues did not show C16 and C18 aliphatic polyols similar to those detected in the hydrogenolysate of leaf and stem residues (data not shown). Therefore, these molecules are derived from insoluble polyester polymers and/or are covalently linked to other cellular structures (e.g. cell wall components, proteins) found in the residue.
Depolymerization of Arabidopsis residues by transmethylation. To confirm the identity of the monomers detected by hydrogenolysis and deuteriolysis, delipidated Arabidopsis leaf and stem residues were transmethylated using sodium methoxide (NaOCH3) in methanol. Transmethylation using BF3/methanol was avoided owing to its propensity to produce artifacts and high losses of polyunsaturated fatty acids (Christie, 2003). GC-MS analysis of the silylated products is shown in Figure 1(b,d). The major product of the leaf residues is dimethyl octadecanediene-1,18-dioate, with lower amounts of dimethyl octadecene-1,18-dioate, methyl 18-hydroxy-octadecadienoate, methyl 18-hydroxy-octadecenoate, dimethyl hexadecane-1,16-dioate, and methyl 10,16-hydroxy-hexadecanoate. This analysis confirmed dicarboxylic acids as the major components of Arabidopsis leaf polyesters and also validated the occurrence of many other components determined by hydrogenolysis/deuteriolysis.
The relative distribution of dicarboxylic and ω-hydroxy-fatty acids obtained from methanolysis was in good agreement with the distribution obtained after LiAlH4/LiAlD4 treatments (Table S1). Hydrogenolysis/deuteriolysis produced very consistent monomer composition and content data, whereas methanolysis gave more variable monomer contents, with values up to 50% less than hydrogenolysis. Thus, we used hydrogenolysis/deuteriolysis of dry, delipidated tissue residues to determine the polyester monomer composition and content of Arabidopsis tissues. Transmethylation was used as a complementary method to corroborate the chemical structures identified by hydrogenolysis/deuteriolysis.
Determination of the chemical structures of the dicarboxylate monomers. Saturated and monounsaturated dicarboxylic acids are well known as components of suberin (Kolattukudy, 1980). However, because the dicarboxylic acids were an unexpected major product from Arabidopsis epidermis, and because the C18:2 dicarboxylic acid has been reported only as a very minor component (<0.1%) in apple (Eglinton and Hunneman, 1968) and cranberry (Croteau and Fagerson, 1972) cutins, and has therefore been incompletely characterized, we undertook a more thorough examination of its chemical structure. The mass spectra of the trimethylsilyl (TMS) ethers of octadecadiene- and octadecene-1,18-diols obtained by hydrogenolysis are shown in Figure 2(a,b), respectively, while the mass spectra of methyl diester derivatives obtained from transmethylation are shown in Figure 2(c,d). The TMS ethers of the diols show ions typical of primary hydroxyl groups at m/z 55, 73, 75, 89 and 103. In addition, weak molecular ion peaks were observed at m/z 426 for octadecadiene-1,18-diol and at m/z 428 for octadecene-1,18-diol. The inserts in Figure 2(a,b) show the molecular ion clusters of the hydrogenolysis and deuteriolysis products, respectively. Analysis of these ion clusters indicated that about 80% of octadecanediene-1,18-diol is derived from octadecadiene-1,18-dioate (M + 4), 8% from 18-hydroxy-octadecadienoate (M + 2) and 8% from octadecadiene-1,18-diol (M), with possibly up to 4% from 18-oxo-octadecadienoate (M + 3) in the intact polyester. The bracketed formulae indicate the increased mass of the molecular ion peak after deuteriolysis. Similarly, 60% of octadecene-1,18-diol originates from the dicarboxylic acid (M + 4), 27% from the ω-hydroxy-fatty acid (M + 2), 7% from the 1, ω-diol (M) and 6% from the ω-oxo-fatty acid (M + 3). The mass spectra of the octadecadiene-1,18-dioate and octadecene-1,18-dioate methyl diesters give similar fragmentation patterns to previous data (Eglinton and Hunneman, 1968). The m/z 74 peak is diagnostic of a methyl ester (McLafferty rearrangement). Peaks at m/z 274 and 307 for octadecadiene-1,18-dioate and at m/z 276 and 309 for octadecene-1,18-dioate correspond to (M+-64) and (M+-31) ions, respectively, and predict the correct MW for the assigned structures. In addition to the mass spectral analysis the methyl diesters were identified by their chromatographic behavior before and after hydrogenation.
Although we expected octadecadiene-1,18-dioate and octadecene-1,18-dioate to be derived directly from linoleate and oleate, respectively, and therefore to retain the double bond positions and configurations of those precursors, we determined these structural parameters by independent chemical methods. First a purified fraction of the methyl diesters was isolated by thin layer chromatography (TLC). Double bond position was determined on this mixture by oxidative cleavage (vonRudloff, 1956). Briefly, the oxidative cleavage of the double bonds in octadeca-6,9-diene-1,18-dioate and octadec-9-ene-1,18-dioate methyl diesters is expected to produce the monomethyl esters of dicarboxylic acids in an equimolar mixture of C6 and C9 and solely C9 respectively. These products were indeed the only oxidation products observed (Figure S2a). Analysis of the double bond geometric configuration was achieved by treating the diester fraction with oxides of nitrogen (Litchfield et al., 1965). With this method stereoisomerization to an equilibrium cis/trans mixture of products occurs without double bond migration, with trans products dominating. Subsequent analysis on a GC column that can resolve geometric isomers shows that the starting unsaturated diesters are all in the cis configuration (Figure S2b,c). Thus we have unambiguously determined the structure of the two major unsaturated dicarboxylic acid monomers released on the depolymerization of Arabidopsis tissues as octadeca-cis-6, cis-9-diene-1,18-dioate§ and of octadec-cis-9-ene-1,18-dioate.
Identification of the major ω-hydroxy-fatty acid monomers, namely 10,16-hydroxy-hexadecanoate (with lesser amounts of the 9,16-isomer), 9,10-epoxy-18-hydroxyoctadec-12-enoate, and 9,10-epoxy-18-hydroxystearate [or the corresponding 9(10)-oxo-compound], are described in the Supplementary material (Figure S3).
Depolymerization of epidermal peels from Arabidopsis stems
Dicarboxylic acids are frequently found as major components of suberin in tissues such as roots and elements of the vascular system whereas they are much less abundant in cutins (Kolattukudy, 2001). They can also be associated with secondary cell walls. To determine whether the C18:1 and C18:2 dicarboxylic acids and the other polyester components recovered after depolymerization of Arabidopsis residues originated from epidermis-associated polyesters, the epidermis from wild-type stems was manually dissected under a microscope and subjected to depolymerization by both LiAlH4 and LiAlD4. The remaining stem material from which the epidermis was removed, and whole stems were also analyzed by the same methods. In all cases, no solvent extractions were performed prior to depolymerization with the intention of preserving all the epidermal wax components in order to quantify the relative amounts of polyester monomers to epicuticular waxes.
GC-MS analysis of the hydrogenolyzed epidermis gave similar distributions of polyester aliphatic components (Figure 3) to that obtained with the stem and leaf residues (see Figure 1a). In this regard octadecadiene-1,18-diol was accompanied by smaller peaks for hexadecane-1,16-diol, hexadecane-1,7,16-triol and octadecene-1,18-diol (Figure 3). Their corresponding deuterated derivatives indicated that the distribution of dicarboxylic acids, ω-hydroxy-fatty acids and alcohols was similar to the distribution of these components in dry residues (data not shown). These observations indicate that C18:2 and C18:1 derivatives exist primarily as dicarboxylic acids in the polyesters of the epidermis. Small amounts of chlorophyll were detected in the epidermis peels [0.05 mg chlorophyll/gram of fresh weight (gfw) versus 0.6 mg chlorophyll/gfw in total stems] and potential contamination with non-epidermal tissues cannot be ruled out. However, if present, these contaminants are minor (see below).
Hydrogenolysis of the stem material without the epidermis did not produce any detectable C16 and C18 polyhydroxylated aliphatic compound when analyzed by GC (data not shown), indicating that the amount of these components associated with the vascular system is small compared with the epidermis. In addition, hydrogenolysis of total stems showed that the relative amounts of waxes to polyester monomers was conserved when compared with isolated epidermis, suggesting that most polyester monomers are derived from the stem epidermis. In this regard, the ratio of octadecadiene-1,18-diol to C29 alkane (the most abundant epicuticular wax component) was 0.25 in the isolated epidermis and 0.24 in total stems. Furthermore, the relative amounts of C16 to C18 polyester monomers (without considering palmitic, oleic and linoleic acids) were similar between the epidermis and total stem (0.40 versus 0.45 C16/C18 ratio, respectively).
In summary, these results suggest that the depolymerization products, particularly dicarboxylic acids, obtained from Arabidopsis leaf and stem residues originate from polyesters deposited by epidermal cells rather than underlying tissues. This conclusion is substantiated by: (i) the absence of oxygenated derivatives of C16 and C18 fatty acids in stem tissue stripped of epidermis, (ii) the constant proportion of waxes to polyester components in the epidermis and whole stem, (iii) the similar ratio of C16/C18 monomers in epidermis and whole stems, and (iv) the similar distribution of polyester monomers between stems and leaves.
Polyester composition of leaves from wild-type Arabidopsis and mutants in fatty acid and lipid metabolism
Depolymerization by LiAlH4 and LiAlD4 was used to quantify leaf polyester composition of wild-type Arabidopsis and mutants with altered fatty acid metabolism. The data are presented in Figure 4. In wild-type Arabidopsis leaves, 27% of the polyester monomers were derived from palmitic acid and 73% from C18 fatty acids (Figure 4). Hexadecane-1,16-diol (10%) and hexadecane-1,7,16-triol (14%) were the most abundant C16 hydrogenolysis products. As at least half of the C16 diol is derived from the dicarboxylate, the C16 polyester monomers are 10,16-dihydroxypalmitate > hexadecane-1,16-dioate ≥ 16-hydroxypalmitate > palmitate. Among C18 hydrogenolysis products, unsaturated components accounted for 68% of total aliphatic monomers, with octadecadiene-1,18-diol (52%) and octadecene-1,18-diol (10%) the most abundant. Correcting for multiple origins of the C18 diols the C18 polyester monomers are octadecadiene-1,18-dioate ≫ octadecene-1,18-dioate ≥ 18-hydroxylinoleate, with minor components (<5% each) including 18-hydroxyoleate, 18-hydroxy-9,10-epoxystearate, 18-hydroxy-9,10-epoxyoctadecenoate and unmodified fatty acids. The total amount of monomers detected in wild-type Arabidopsis leaves was 1.8 ± 0.2 mg per gram of dry residue. Considering that approximately 10% of leaf fresh weight is recovered as dry material after lipid extractions (data not shown), there are approximately 0.2 mg of polyester monomers per gram fresh weight recovered by hydrogenolysis in Arabidopsis. Wild-type Arabidopsis plants were also analyzed for leaf polyester composition at 4, 5 and 6 weeks after germination, and no substantial differences in the polyester composition and monomer load were observed between the different stages (data not shown).
A quantitative analysis of polyester composition was performed on leaves of two Arabidopsis mutants with altered fatty acid metabolism. The rationale was to investigate whether changes in these biochemical processes bring about alterations in polyester amounts and composition. The mutants analyzed were fatb-ko (Bonaventure et al., 2003) and fad2-1 (Miquel and Browse, 1992).
In the fatb-ko mutant the total amount of saturated fatty acids in various tissues is curtailed by 40–50% compared with wild type. This reduction occurs only in the cytosolic pool of saturated fatty acids, affecting the fatty acid composition of extraplastidial phospholipids and the wax load in leaves and stems (Bonaventure et al., 2003). Leaf polyester analysis of this mutant showed that polyester monomer load was similar to wild type; however the composition changed (Figure 4). The C16 derivatives were reduced sevenfold, from 27% in wild type to 4% in fatb-ko. The lower amounts of C16 monomers were compensated by increased deposition of unsaturated C18 monomers to give similar amounts of leaf polyester (1.7 ± 0.2 mg per gram dry residue in fatb-ko versus 1.8 ± 0.2 mg per gram dry residue in wild type). Importantly, deuteriolysis of fatb-ko leaf residues indicated that a 17% reduction in C16 hydroxy-fatty acids and a 4% reduction in C16 dicarboxylic acids compared with wild type were compensated almost entirely by increased synthesis of C18 dicarboxylic acids (18.5%) with a small compensation by C18 hydroxy-fatty acids (1.5% increase).
In wild-type Arabidopsis, cytosolic linoleate (18:2Δ9,12) is produced primarily by desaturation of oleate bound to phospholipids by a membrane-bound oleate desaturase (FAD2) (Okuley et al., 1994). Because substantial amounts of C18:2 derivatives were found in leaf polyesters of Arabidopsis, we analyzed the fad2-1 mutant, which has an 80% reduction in C18:2 in leaf PC. Using LiALH4 depolymerization, fad2-1 leaves gave a twofold reduction in the level of octadecadiene-1,18-diol and a threefold increase in the level of octadecene-1,18-diol (Figure 4). In addition, a slight increase in oleate was detected. C16 derivatives showed slight changes in composition but their total amount remained similar to wild type. Despite the compositional changes the total polyester load remained similar (1.9 ± 0.1 mg per gram dry residue in fad2-1 versus 1.8 ± 0.2 mg per gram dry residue in wild type).
The polyester composition of leaves from canola (Brassica napus) was also analyzed (Figure S1b). After hydrogenolysis the most abundant components were hexadecane-1,16-diol (22%), hexadecane-1,7,16-triol (24%) and octadecene-1,9,18-triol (13%). Deuteriolysis indicated that dicarboxylic acids accounted for 10% of total aliphatics and that ω-hydroxy fatty acids were predominant.
Methodology for the analysis of the polyester composition of the epidermis of Arabidopsis leaves and stems
Plant cutins and suberins are considered the most abundant lipid polymers in nature (Heredia, 2003). However, understanding the structure, biosynthesis, function and genetics of these extracellular polymers has been a challenge in plant biology. This arises in part because of their synthesis in very specific structures within plant tissues, and because of the difficulties in undertaking their chemical analysis and in defining details of their polymer structure. With the opportunity to use reverse genetic approaches in Arabidopsis to unravel cutin biosynthesis and function, the requirement for a suitable method for the analysis of cutin monomer composition becomes self-evident. Here, we describe a method for such analysis. Moreover, we demonstrate that this method is sensitive enough to detect compositional differences in epidermal polyesters of fatty acid mutants.
Analysis of delipidated whole tissue residues has the advantage of bypassing the cuticle isolation step and therefore simplifying the process of cutin analysis in Arabidopsis. A potential disadvantage is the presence in the residue of non-epicuticular contaminating polyester components, such as suberin from suberized cell walls of the vascular system. However, we demonstrate that most, if not all, of the polyester-derived products in the solvent-extracted residues of leaves and stems originate from the epidermis in Arabidopsis. The advantages of using LiAlH4/LiAlD4 for depolymerization are several: (i) the reaction occurs in a reducing environment and thus the oxidation of polyunsaturated components is less problematic, (ii) the reduction in functional groups gives a simple mixture of components for analysis and (iii) the use of LiAlD4 followed by GC-MS allows the more complete identification of monomers from the intact polyester. The use of NaOCH3 catalyzed methanolysis for depolymerization was largely restricted to aiding structure identification. Some differences in polyester composition and content of Arabidopsis leaves were observed when hydrogenolysis and methanolysis were compared (Table S1). That there is a certain degree of dependence between leaf polyester composition and the method used for its analysis is hardly surprising given that cutin and suberin are amorphous, cross-linked polymers that require vigorous conditions for depolymerization and that the methods use totally different chemistries. Nonetheless, both methods were consistent in detecting the relative changes in the epidermal polyester composition of wild-type and mutant plants (data not shown).
Neither hydrogenolysis nor transmethylation cleaves either ether or carbon–carbon bonds and therefore the cutan fraction of cutin remains polymerized and is not represented in our analysis (Walton and Kolattukudy, 1972b). Thus our estimation of the amounts of cutin monomers will underestimate the total amount of polyesters present in the Arabidopsis epidermis. Studies with other tissues suggest that cutan can amount to 10–60% by weight of the total isolated cuticle and contains ether-linked aliphatic components as well as carbohydrates (Nip et al., 1986; Schmidt and Schonherr, 1982). The high polyunsaturated content of Arabidopsis epidermis polyesters might lead to extensive cross-linking through oxidative processes centered on the 1,4-pentadiene structure.
Potential ramifications of the novel polyester composition found in wild-type Arabidopsis epidermis
Cutin monomers are usually classified into C16 and C18 families (Holloway, 1982). The most common and abundant C16 monomers are 16-hydroxy-hexadecanoate, 10,16-hydroxy-hexadecanoate, and 16-hydroxy-10-oxohexadecanoate. The most common and abundant C18 monomers are 18-hydroxy-9,10-epoxy-stearate and 9,10,18-trihydroxy-stearate (Kolattukudy, 2001). However, in the epidermis of Arabidopsis leaf and stem octadecadiene-1,18-dioate is the most abundant polyester monomer. Octadecene-1,18-dioate and hexadecane-1,16-dioate are also present in significant amounts. Dicarboxylic acids are commonly found in suberized tissues such as roots, vascular bundles and wounded tissue (Kolattukudy, 1980, 2001). These dicarboxylates are often saturates, but octadecene-1,18-dioate can be a major component. By contrast, octadecadiene-1,18-dioate is rarely reported, and then only as a very minor component. Dicarboxylic acids are considered minor components of leaf cutins (Kolattukudy, 1980, 2001).
From the above literature review it is apparent that the presence of octadecadiene-1,18-dioate as the major component of Arabidopsis epidermal polyesters was an unexpected finding. Throughout this paper we have been careful to describe octadecadiene-1,18-dioate as a component of ‘epidermal polyesters’ rather than ‘cutin’ or ‘suberin.’ Although the dicarboxylic acids originate predominantly from the epidermal layer rather than underlying tissues, we do not know whether they are an integral part of the cutin polyester or represent an underlying ‘suberin-like’ dicarboxylic acid network present in the secondary cell wall. The Arabidopsis leaf cuticle is reported as a 20–25 nm layer (Nawrath, 2002) and has a much reduced wax load compared with stems. Independently, the Arabidopsis adaxial and abaxial leaf cuticle thicknesses were determined to be 32 and 36 nm, respectively (Schnurr et al., 2004). These estimates allow for a cuticular layer that is about 10–15 epicuticular wax molecules/cutin monomers deep if these aliphatics are in an extended conformation perpendicular to the plane of the leaf surface. Although the average Arabidopsis leaf thickness is difficult to compute given developmental variations and air spaces, 100 μ seems a reasonable first approximation (Tsukaya, 2002). Given this value, an average adaxial and abaxial thickness of 30 nm, and a density for the cuticle of 0.85, the leaf cuticles would require about 500 μg of cuticular lipid per gram fresh weight to fill the required cuticular volume. Previously we determined that there are about 100 μg of epicuticular waxes/gfw in wild-type Arabidopsis leaves (Bonaventure et al., 2003). In this paper we have estimated the total monomers that can be depolymerized from polyesters at 200 μg/gfw. Thus, it seems more likely that the potential deficit in accounting of aliphatic material would indicate that all of the monomers we have measured are associated with the cuticle. The deficit may be made up in part or whole by the cutan fraction.
Clearly, the question of whether the dicarboxylate and ω–hydroxy fatty acid components of Arabidopsis leaf epidermis are co-deposited, either spatially or temporally, will be an important focus for future experiments. Underlying this question of biochemistry is one of function. Does the dicarboxylate polyester have the same function as the ω–hydroxy fatty acid polyester? Although at lower relative levels than in Arabidopsis, the presence of unsaturated dicarboxylic acids in canola leaves suggests that these components may be associated with some plant families such as Brassicaceae. The Arabidopsis cuticle is particularly thin (20–35 nm), so one question is whether dicarboxylates may have been overlooked in previous analyses of other species with thicker cuticles. Heredia (2003) has reviewed cutin morphology, biochemistry and function and notes that isolated cuticles range from 2 mg cm−2 for fruit cuticles to 0.45–0.8 mg cm−2 for leaf cuticles, of which 40–80% of the weight corresponds to cutin. Again, using a density of 0.85 for the aliphatics, these numbers correspond to thicknesses of 2300 and 530–940 nm, respectively, many times greater than Arabidopsis. Thus, if there is only a thin layer of dicarboxylates and the progressive cuticle thickening is due to increasingly massive deposition of ω–hydroxy fatty acid polyesters, then perhaps cutin depolymerization measurements of dicarboxylates need to detect fractions of a percentage of the total or the latter components will be missed.
The presence of high amounts of dicarboxylates in Arabidopsis epidermis also raises the question concerning the functional groups to which these dicarboxylates are esterified. Glycerol has been identified as a component of cutin (Graca et al., 2002) and suberin (Graca and Pereira, 2000a), and in a partial depolymerization of potato periderm diglycerol alkenedioates were identified (Graca and Pereira, 2000b). These types of analyses need to be applied to Arabidopsis epidermis. Esters of ω-hydroxy fatty acids produce only linear polymers. Despite textbook diagrams, the introduction of mid-chain hydroxyls and their esterification will only produce dendrimeric polymers. The linkages that cause polyesters to be cross-linked or at least insoluble must either be introduced through non-ester bonds, ester bonds to the cell wall matrix, or from the esterification of dicarboxylic acids with glycerol and/or polyhydroxy-fatty acids.
Polyester composition of Arabidopsis mutants in fatty acid metabolism
Analysis of mutants in fatty acid metabolism demonstrated that FATB and FAD2 are enzymes involved in the supply of palmitate and linoleate precursors, respectively, for epidermal polyester biosynthesis. Analysis of the fatb-ko mutant shows that C16 epidermal polyester components are reduced by sevenfold compared with wild type. An important observation is that mutant cells compensate for the reduction in C16 hydroxy-fatty acids by elevating the synthesis of C18 dicarboxylic acids rather than C18 hydroxy-fatty acids. This physiological response to maintain total polyester load in leaves suggests the possible co-existence of hydroxy-fatty acids and dicarboxylic acids in the same polyester structure.
The participation of FAD2 in providing precursors for epidermal polyester synthesis suggests that oleate is first incorporated into extraplastidial phospholipids, desaturated by FAD2 and then probably released from phospholipids for ω-oxidation and subsequent utilization in polyester synthesis. However, it is possible that the initial ω-oxidation steps may occur on linoleate still esterified to phospholipids. Candidate ω-hydroxylases involved in cutin synthesis (such as cytochrome P450s, and particularly the CYP86A and CYP94 sub-families) can catalyze the hydroxylation of free fatty acids in vitro (Le Bouquin et al., 2001; Wellesen et al., 2001). However, the in vivo substrates for the different P450s remain to be determined. Thus, whether ω-hydroxylation occurs on the intact polyester molecule, on free fatty acid monomers, or on phospholipids remains unknown. What is also uncertain is the mode of synthesis of the dicarboxylic acids. The intermediate ω–hydroxy fatty acid might be oxidized to the dicarboxylic acid by a P450 hydroxylase or by an oxidoreductase (dehydrogenase). CYP94A5 from tobacco can oxidize 9,10-epoxystearic acid to 18-hydroxy-9,10-epoxy stearic acid and then to 9,10-epoxy-octadecane-1,18-dioic acid (Le Bouquin et al., 2001). Alternatively, Agrawal and Kolattukudy (1978) have demonstrated distinct NADP-dependent ω-hydroxy-fatty acid and ω-oxo-fatty acid dehydrogenase activities in extracts from suberizing potato slices.
Although physiological and microscopic analysis of several mutants suggests involvement of P450 and long-chain acyl-CoA synthetase genes in cutin and suberin biosynthesis, no data yet exist to show that these genes have a direct effect on polyester composition or content. If our understanding of monomer synthesis is at present very limited, then our understanding of the biosynthesis of extracellular lipid polyesters is almost non-existent. It is not even clear if the polyesters, or sections of the polyesters, are synthesized within the cell and then transported to the cell wall, if synthesis proceeds at the plasmalemma, or if the monomers are transported across the plasmalemma and then assembled in the extracellular matrix. No gene that directly participates in either of these acyl transfer or transport steps has been identified. With the monomer analysis of epidermal polyesters in place, the characterization of a number of Arabidopsis mutants can begin to unravel some of these mechanisms. Such mutants include lacerata, wax2 and lacs2 (Chen et al., 2003; Schnurr et al., 2004; Wellesen et al., 2001), post-genital organ fusion mutants (Lolle et al., 1998), and cuticle mutants (Tanaka et al., 2004).
Plant material and growth conditions
Wild-type A. thaliana (ecotype Wassilewskija-2) and the different mutants were grown on a mixture of soil:vermiculite:perlite (1:1:1) under white fluorescent light (80–100 μE m−2 sec−1) in a 18-h-light/6-h-dark photoperiod. The temperature was set at 20–22°C and the relative humidity at 60–70%. Seeds were always stratified for 4 days at 4°C.
Hydrogenolysis with LiAlH4
The protocol was adapted from the method of Walton and Kolattukudy (1972b). Approximately 1–3 g fresh weight of either leaf tissue from 3-week-old Arabidopsis plants or primary stems (approximately 20 cm long) was quenched in 100 ml of hot isopropanol (10 min at 80°C). On cooling the tissue was finely ground with a Polytron and stirred overnight at room temperature in isopropanol. The extract was filtered and the insoluble residue re-extracted by stirring overnight with 100 ml of 2:1 (v:v) chloroform:methanol. After filtration, the residue was re-extracted with 100 ml of 1:2 (v:v) chloroform:methanol. Finally, the residue was air-dried for 2 days and then dried under vacuum for two more days. Prior to depolymerization, the residue was further pulverized to a fine powder in a mortar and pestle. For polyester hydrogenolysis, 0.2 g of either LiAlH4 (Aldrich, Milwaukee, WI, USA) or LiAlD4 (Aldrich, 98% min atom %D) were added to 0.1 g of residue powder in 6 ml of tetrahydrofuran. Methyl heptadecanoate (Sigma, St. Louis, MO, USA) and ω-pentadecalactone (Sigma) at 1 mg g−1 dry residue each were added as internal standards. Hydrogenolysis was carried out in a sealed screw-capped glass tube at 75°C for 48 h with periodic vortexing. At the end of the reaction excess reductant was decomposed by the careful addition of ethyl acetate. Water was added, the mixture acidified and the monomers were extracted into diethyl ether (2 × 10 ml). The ethereal solution was dried over anhydrous sodium sulfate and evaporated to dryness under nitrogen.
Methanolysis with NaOCH3
Approximately 0.2 g of thoroughly dried solvent-extracted Arabidopsis residue was heated at 60°C with stirring in 8 ml of methanol containing 7.5% (v/v) methyl acetate and 4.5% (w/v) sodium methoxide. Methyl-heptadecanoate and pentadecalactone were added as internal standards in the same amounts as indicated above. After 24 h the reaction was cooled, acidified to pH 4 with 1 m acetic acid and the monomer products extracted into methylene dichloride (10 ml). The organic phase was washed several times with dilute saline solution, dried over anhydrous sodium sulfate, and the solvent evaporated to dryness under nitrogen.
Sylilation and GC-MS conditions
The products of hydrogenolysis or methanolysis were heated at 100°C for 10 min in 0.1 ml of pyridine and 0.1 ml of BSTFA [N,O-bis(trimethylsilyl)-trifluoroacetamide]. After cooling, the solvent was evaporated under nitrogen and the product dissolved in 0.5 ml of heptane:toluene 1:1 (vol:vol) for GC or GC-MS analysis. GC analysis used a HP-5 capillary column (30 m × 0.32 mm × 0.25 μm film thickness) with helium carrier gas at 2 ml min−1 and oven temperature programmed from 90 to 300°C at 10°C min−1, and then held for 10 min at 300°C. For GC analysis, samples were injected in split mode (310°C injector temperature) and peaks quantified on the basis of their FID ion current. For GC-MS, splitless injection was used and the mass spectrometer run in scan mode over 40–500 amu (electron impact ionization), with peaks quantified on the basis of their total ion current. For analysis of the molecular ion cluster, the detector was operated in single ion monitoring mode for ions (M-1) to (M + n + 1), where M is the m/z value of the major natural isotopic abundance molecular ion and n represents the highest possible isotopic enrichment.
Purification and chemical characterization of dicarboxylic acids
Approximtely 15 g of dry residue from Arabidopsis leaves was transmethylated as described above. The products were separated by preparative TLC on K6 silica plates (Whatman Inc., Clifton, PA, USA) developed with chloroform. Lipids were detected after spraying with 0.2% (w/v) 2′-7′-dichlorofluorescein in methanol and viewing under UV light. Hexadecanedioate methyl diester was used as a standard to define Rf value. Lipids were eluted from the silica with 3/2 (v/v) hexane/isopropanol.
For double bond positional analysis, the periodate-permanganate oxidation was used (vonRudloff, 1956). Approximately 1 mg of dimethyl dicarboxylates was resuspended in 1 ml of 3.9 mm potassium carbonate in 3/2 (v/v) t-butanol/water and 1 ml of aqueous 3.3 mm potassium permanganate and 102 mm sodium periodate. The sample was vortexed and allowed to stand for 1 h at room temperature. After quenching excess oxidant the reaction products were extracted into diethyl ether (3 × 3 ml). The organic phase was washed with dilute saline solution and dried over anhydrous sodium sulfate. The half-ester dicarboxylate products were analyzed directly by GC-MS on a DB-FFAP capillary column (15 m × 0.25 mm × 0.25 μm film thickness; J & W Scientific, Folsom, CA, USA) with helium carrier gas at 2 ml min−1 and oven temperature programmed at 90°C for 3 min, followed by a 10°C min−1 ramp to 240°C, and then held for an additional 10 min at 240°C. Methyl petroselenate, oleate and eicos-11-enoate were oxidized using this procedure to produce a set of monomethyl ester dicarboxylate standards.
To determine the cis/trans configuration of double bonds the dicarboxylate diesters were first subjected to stereoisomerization (Litchfield et al., 1965). Approximately 1 mg of dimethyl dicarboxylates was resuspended in 1 ml of 2-methoxy-ethylene glycol, and 0.01 ml of 6 m aqueous nitric acid plus 0.015 ml of 2 m aqueous sodium nitrite were added. The sample was vortexed and heated at 60°C for 2 h in a screwed-cap tube. After cooling, 5 ml of water was added, the esters extracted into hexane, and the hexane phase washed three times with water and dried over sodium sulfate. After evaporation under a stream of nitrogen the samples were analyzed by GC(FID) on a DB-23 capillary column (30 m × 0.25 mm × 0.25 μm film thickness; J & W Scientific) using a temperature program from 160 to 240°C at 5°C min−1. The GC column was calibrated for the cis/trans isomer separation with methyl oleate, elaidate, linoleate and linoelaidate samples.
Analysis of Arabidopsis stem epidermal peels
Epidermis peels (as a thin transparent film) were manually dissected using thin forceps under the microscope. Samples were taken from the bottom half of 20 cm long primary stems from approximately 20 wild type Arabidopsis plants (approximately 5-week old). The epidermal peels were ground in a mortar and pestle in the presence of liquid nitrogen and directly hydrogenolyzed or deuteriolyzed as indicated above.
We would like to thank John Browse of Washington State University for fad2-1 seeds, and Lacey Samuels of the University of British Columbia for advice on preparation of epidermal peels from Arabidopsis. This work was supported by the National Science Foundation (grant no. MCB-9817882), by the Department of Energy (DE-FG02-87ER13729), and by the Michigan Agricultural Experiment Station.
Note added in proof
After this manuscript was submitted a paper appeared reporting hexadecane-1,16-dioate as the major monomer in Arabidopsis cutin [Xiao, F., Goodwin S.M., Xiao, Y., Sun, Z., Baker, D., Tang, X., Jenks, M.A., and Zhou, J.-M., (2004) Arabidopsis CYP86A2 represses Pseudomonas syringae type III genes and is required for cuticle development. EMBO J.23, 2903–2913]. We speculate that the difference from our results might originate from destruction of unsaturated monomers during the boron trifluoride catalyzed transmethylation.