The Acyltransferase GPAT5 Is Required for the Synthesis of Suberin in Seed Coat and Root of Arabidopsis^[W][OA]

gpat5 Mutants Are Affected in the Composition and Amount of Lipid Polyesters but Not in Membrane and Storage Lipids

Homozygous gpat5-1 and gpat5-2 mutant plants were morphologically identical throughout development and reached similar sizes compared with wild-type plants. No significant differences in root growth were observed when seeds were germinated on vertical agar plates. The fertility of gpat5 mutants was not affected (the number of seeds per silique was ∼50, similar to that in the wild type). No differences in pollen grain size and shape between the wild type and gpat5 were observed under scanning electron microscopy. GPAT5 has been shown to have glycerol-3-phosphate acyltransferase activity in vitro (Zheng et al., 2003). Thus, to evaluate whether gpat5 mutants carried a biochemical defect in some lipids, we analyzed the fatty acids of lipid compounds in organs in which GPAT5 is expressed (seed, root, flower) and, as a control, in leaves. Analyses were performed on intracellular lipids extractable in organic solvents (membrane and storage lipids) and also on lipid polymers, which are nonextractable in organic solvents.

No differences in the amount of fatty acids derived from soluble lipids were detected in seeds, roots, and leaves of the mutants compared with the wild type (see Supplemental Figure 1 online). When soluble lipids of the manually dissected seed coat/endosperm fraction of mature seeds were analyzed, the amount of total fatty acids and the fatty acid composition were found to be the same in the wild type and gpat5 (Figure 3). The soluble lipids collected at the seed surface by chloroform dipping were typical Arabidopsis stem/leaf wax components, including alkane and fatty alcohols (see Supplemental Figure 2 online). The major component identified in seed coats was 29-carbon alkane, which is also the major wax component of Arabidopsis stems. Expressed per seed surface area, the wax coverage was ∼0.3 μg/cm² for both the wild type and mutants. In contrast with soluble lipids, aliphatic monomers released from insoluble lipid polyesters were on average reduced by twofold in gpat5 seeds (Table 1). More than 90% of the total insoluble aliphatic polyester monomers found in seeds have been shown to come from the seed coat/endosperm fraction (Molina et al., 2006). The major aliphatic monomer (24:0 ω-hydroxy fatty acid), as well as 22:0 fatty acid, 22:0 ω-hydroxy fatty acid, and 22:0/24:0 dicarboxylic acids, were reduced by at least twofold in the mutants compared with the wild type (Figure 4A). All of these strongly reduced monomers were typical monomers of aliphatic suberin (Kolattukudy, 2001). In contrast with the aliphatic monomers, the amount of aromatic monomers released from the polyester by depolymerization was not different between mutants and the wild type. However, depolymerization by transesterification presumably underestimates total hydroxycinnamate derivatives, which may be cross-linked by nonester bonds (Bernards et al., 1995).

Open in a separate window

Figure 3.

Fatty Acids from the Seed Coat/Endosperm Fraction of the Wild Type and gpat5 Mutants.

Mature seeds were manually dissected, and total fatty acids of the membrane and storage lipids of the seed coat/endosperm fraction were analyzed as fatty acid methyl esters by gas chromatography. Values are means of six replicates. Error bars denote 95% confidence intervals.

Open in a separate window

Figure 4.

Lipid Polyester Monomers from Seeds, Roots, and Flowers of Wild-Type and gpat5 Plants.

(A) Polyester monomers from mature seeds.

(B) Polyester monomers from roots of 1-week-old seedlings grown on agar.

(C) Polyester monomers from opened flowers.

The insoluble dry residue obtained after grinding and delipidation of tissues with organic solvents was depolymerized with sodium methoxide, and aliphatic and aromatic monomers released were analyzed by gas chromatography–mass spectrometry. Values are means of six data points (two independent experiments using different biological samples involving triplicate assays for the depolymerization reaction). Error bars denote 95% confidence intervals. DCAs, fatty dicarboxylic acids; FAs, fatty acids; fw, fresh weight; PAs, primary alcohols. The polyol fatty acids are 10,16-hydroxy 16:0 and 9,10,18-hydroxy 18:1.

Similar to seed coats, roots of 1-week-old seedlings grown on agar (primary state of growth), showed significant changes in the aliphatic monomers of suberin in gpat5 compared with the wild type (Figure 4B). As with seeds, the change was not the same for all monomers. A 20 to 50% reduction was observed in C20-C24 fatty acid derivatives, whereas the C16-C18 fatty acid derivatives remained constant or increased. In 3-week-old roots grown on agar (beginning of the secondary state of growth), the suberin composition was different and the mutants showed a global 50% decrease in some monomers, including the 22:0 fatty acid and the major 18:1 ω-hydroxy fatty acid (see Supplemental Figure 3 online). The reduction in total aliphatic suberin monomers in 3-week-old roots was also shown by the reduced staining intensity of some parts of the gpat5 roots when stained with Sudan black B (Figure 5), a lipophilic dye used for suberin histochemical detection (Robb et al., 1991). In older wild-type roots grown on soil, in which the GPAT5 gene is weakly expressed (6 weeks old, secondary state of growth), the composition of suberin aliphatic monomers detected was similar to that reported previously (Franke et al., 2005), and no significant difference in the amount of each monomer was found between the wild type and the mutants.

Open in a separate window

Figure 5.

Staining of 3-Week-Old Roots of gpat5-1 and Wild-Type Plants with the Suberin Dye Sudan Black B.

In flowers (Figure 4C), a decrease in some lipid polyester monomers was also detected (e.g., 22:0 fatty acid and 18:2 ω-hydroxy fatty acid) but was globally less strong than in roots and seeds, consistent with the fact that the gene is expressed only in anthers. As expected, no difference in the lipid polyester monomers of leaves was detected (see Supplemental Figure 4 online).

Together, these results show that the gpat5 mutants are altered (both quantitatively and qualitatively) in the synthesis of insoluble lipid polyesters but are not impaired in the synthesis of the bulk of storage and membrane lipids. A specific role of a GPAT family member in lipid polyester synthesis is strongly supported by other observations, including the fact that two other GPATs are required for lipid polyester synthesis in leaves (see Discussion).

Seed Coats of gpat5 Mutants Show Increased Permeability to Dyes and Darker Color

The mutant plants produced seeds of the same weight as wild-type plants (17.0 ± 0.7 compared with 16.9 ± 0.1 μg/seed). Closer examination of the seeds via scanning electron microscopy showed that the mutant seeds are in the same range of size, have the same oblong shape, and show the same surface structure, with similar columella heaps, as wild-type seeds (Figure 6A). Mucilage production of the mutant seeds is also the same as that of wild-type seeds, as demonstrated via staining with ruthenium red (Figure 6B).

Open in a separate window

Figure 6.

Phenotype of gpat5 Seeds, and Permeability to Dyes.

Similar results were obtained with gpat5-2.

(A) Scanning electron microscopy of the surface of wild-type versus gpat5-1 seeds. Bars = 40 μm.

(B) Ruthenium red staining of seed mucilage in wild-type versus gpat5-1 seeds.

(C) Tetrazolium salt staining (24 h) of wild-type versus gpat5-1 seeds.

(D) Seeds from (C) dissected after staining.

(E) Tetrazolium salt staining (4 h) of wild-type versus gpat5-1 seeds.

(F) Autofluorescence of wild-type versus gpat5-1 seeds (365-nm excitation).

(G) Sudan red 7B staining of wild-type versus gpat5-1 seeds.

(H) Tetrazolium salt staining (24 h) of seeds resulting from the fertilization of wild-type plants by gpat5-1 pollen (left) and of gpat5-1 plants by wild-type pollen (right).

Permeability properties of the seed coat of the mutants were tested using tetrazolium red, a cationic dye that is normally excluded by the Arabidopsis seed coat but that is reduced to red products (formazans) by NADPH-dependent reductases after penetrating the embryo (Debeaujon et al., 2000). As shown in Figures 6C and 6D, after staining for 24 h, gpat5-1 and gpat5-2 seed coats were much more permeable to tetrazolium red than were wild-type seed coats, suggesting that the seed coat is indeed affected in the mutants. Shorter incubation times with tetrazolium red showed that the staining first appeared in the region of the hilum and diffused outward from there (Figure 6E). Control staining experiments run on embryos whose seed coats had been manually removed showed that wild-type and mutant embryos had the same kinetics and intensity of staining and that the red products appeared at the same time in all parts of the embryo surface. Therefore, these controls ruled out a possible difference in the capacity to metabolize tetrazolium red between mutant and wild-type embryos or between the parts of the embryo close to the hilum region and other parts. Differences seen in Figures 6C and 6D could be attributed to a difference in the permeability of wild-type and mutant seed coats to tetrazolium red, especially in the hilum region. The hilum is the scar left on the seed coat after detachment from the funiculus. In mature seeds of Arabidopsis, it is adjacent to the micropyle (where the radicle will emerge) and faces the chalazal pole. When excited at 365 nm (Figure 6F), seed coats of the mutant revealed a decrease in autofluorescence in the hilum region (arrows), suggesting a decrease in suberin content. Furthermore, the seed coat surface of gpat5 mutants was clearly less stained in the hilum region (Figure 6G, arrows) when using the lipophilic suberin dye Sudan red 7B (which proved to be more efficient than Sudan black B for staining polyesters of wild-type Arabidopsis seed coats). A very localized area of strong Sudan red 7B staining was visible in the wild type, whereas in the mutant the staining was weak or not visible at all in this region. The seed coats of the mutant were more fragile, as they almost always broke when mounted between slide and cover slip. The permeability of wild-type seeds remained unaffected after the removal of surface waxes by chloroform dipping, excluding the possibility that seed surface waxes contribute to seed coat impermeability to tetrazolium salts.

Genetic analysis of the seed coat permeability phenotype in gpat5 mutants indicated that this phenotype segregated as a single recessive allele and that its inheritance was consistent with the expression of GPAT5 in the seed coat (tissue of maternal origin). Indeed, F1 seeds (i.e., on the F0 mother plants) were all nonpermeable when wild-type plants were fertilized by pollen from gpat5-1 knockout plants, whereas they were all permeable when gpat5-1 knockout plants were fertilized by wild-type pollen (Figure 6H). In addition, analysis of subsequent progeny of the F1 heterozygous plants showed that all F2 seeds were nonpermeable. Segregation analysis of F2 plants demonstrated that the seed permeability phenotype segregated in F3 seeds fitted a 3:1 ratio for a Mendelian single recessive mutation (χ² = 0.17; P = 0.68). As expected, F2 plants giving permeable F3 seeds were homozygous and the F2 plants giving nonpermeable F3 seeds were either wild type or heterozygous, as determined by PCR. Furthermore, it was observed that gpat5-1 and gpat5-2 seeds had a darker appearance than wild-type seeds (Figure 7A) and that in the F1, F2, and F3 seeds this darker seed coat color was always associated with the seed coat permeability phenotype and never with the nonpermeable phenotype. The segregation analysis thus demonstrated that seed coat permeability and color cosegregated with the T-DNA insertions and that these genetic lesions in GPAT5 segregated as single recessive alleles.

Open in a separate window

Figure 7.

Brown Pigmentation of Wild-Type and gpat5 Seeds.

(A) Batch color of wild-type versus gpat5-1 and gpat5-2 seeds.

(B) Amount of soluble and insoluble PAs in wild-type versus gpat5-1 and gpat5-2 seeds. Values are means of six data points (two independent experiments using different seed batches involving triplicate assays for the depolymerization reactions). Error bars denote 95% confidence intervals.

Here, it is important to note that Arabidopsis seed coat color is conferred by the brown pigments formed during seed desiccation by the oxidation of colorless proanthocyanidins (PAs; also called condensed tannins) (Lepiniec et al., 2006). Analysis of soluble PAs and the cell wall–bound insoluble PAs by an acid hydrolysis method optimized for Arabidopsis seeds (Routaboul et al., 2006) showed that gpat5 mutant seeds have the same level of both types of PAs as wild-type seeds, ruling out an increase in PA amount (Figure 7B). This result suggests that the darker color may result indirectly from a higher degree of oxidation of PAs, although we cannot exclude other alternative explanations.

Possible Functions of GPAT5 in Lipid Metabolism

Analysis of the fatty acid derivatives found in seeds, roots, and flowers showed that GPAT5 was essential for the synthesis of lipid polyesters but not for the other lipids analyzed. Moreover, in these organs, the expression of the gene was restricted (e.g., in the root the gene was expressed in the differentiation zone but not in the division and elongation zones that must actively synthesize membrane lipids). These results indicated that GPAT5 is not likely to be a housekeeping gene of fatty acid metabolism that indirectly affects lipid polyester composition, like, for example, the FatB gene (Bonaventure et al., 2004). A specific role of a GPAT family member in lipid polyester synthesis is further supported by two other observations. First, a GPAT (GPAT6; At2g38110) is among the few genes upregulated by the overexpression of the transcription factor WIN1 that is specifically required for cuticle formation (Broun et al., 2004). Second, at least two other GPATs are required for cutin synthesis and normal cuticle impermeability in leaves but are not essential for the synthesis of membrane and storage lipids (Y. Li and F. Beisson, unpublished data). Nevertheless, at this stage, we cannot completely exclude the involvement of GPAT5 in the synthesis of intracellular soluble glycerolipids.

The pathways of the synthesis, transport, and assembly of plant extracellular lipid polymers are poorly understood. Central to polyester synthesis is fatty acid ω-oxidation to form fatty hydroxy and dicarboxylic acids. The reactions of fatty acid oxidation and some partially characterized enzymes (cytochrome P450 monooxygenases, etc.) that may be involved in these oxidation processes have been described (Kolattukudy, 2001), but the in vivo substrates of the oxidation enzymes are not known and might be acyl-CoAs, free fatty acids, etc. After the oxidation steps, the monomers are assembled by the polyester synthase(s)—that is, the enzyme(s) responsible for the synthesis of the primary ester bonds in the polyester chain. The presence of glycerol in cutin and suberin raises the possibility that acylglycerols are substrates for oxidation and/or polymerization reactions. Glycerol 3-phosphate acyltransferase (EC 2.3.1.15) catalyzes the transfer of an acyl group to the sn-1 position of glycerol 3-phosphate to form 2-lysophosphatidic acid (Murata and Tasaka, 1997). Because Arabidopsis GPATs were isolated by structural similarity to yeast glycerol-phosphate acyltransferases and because most of the members, including GPAT5, have been shown to be able to catalyze the transfer of acyl chains from acyl-CoA to glycerol-3-phosphate to form lysophosphatidic acid (Zheng et al., 2003), it is tempting to conclude that the function of GPAT5 is to provide lysophosphatidic acid for the oxidation or assembly machinery of suberin synthesis. However, because other acyl acceptors than glycerol-3-phosphate and other acyl donors than acyl-CoAs have not been tested for GPAT5, it cannot be excluded that in vivo glycerol could be an acyl acceptor and hydroxy acyl-CoA could be an acyl donor. Direct acylation of glycerol has indeed been observed in animal cells (Hanel and Gelb, 1995; Lee et al., 2001). Thus, the reaction catalyzed by GPAT5 could occur before and/or after the oxidation steps. Other possibilities are more unlikely but cannot be ruled out at this stage (e.g., that GPAT5 is a polyester synthase). Therefore, based on the previous biochemical characterization of GPAT5 and the results described here, the possibilities for the molecular function of GPAT5 can be summarized as follows: (1) formation of unsubstituted fatty acid–containing acylglycerols to feed the oxidation machinery; (2) formation of oxidized fatty acid–containing acylglycerols that will be used as acyl carriers and/or substrates for the polymerization reactions; and (3) addition of acyl chains to a growing glycerol-containing lipid polyester network. These possibilities can be fitted into the tentative metabolic pathway for suberin biosynthesis described by Bernards (2002).

Regarding the possible acyl chain specificity of GPAT5, the common feature of the polyester monomer profiles in seeds, roots, and flowers of the gpat5 mutants is the strong reduction (at least twofold) observed in 22:0/24:0 fatty acids and derivatives (Figure 4). Therefore, we propose that the role of GPAT5 is to provide the polyester synthesis pathway with acylglycerols containing 22:0 and 24:0 fatty acids (and possibly oxidized derivatives, if oxidation steps precede GPAT5 activity). The hypothetical specificity of GPAT5 for C22-C24 acyl chains is strongly supported by the fact that the ectopic overexpression of GPAT5 in Arabidopsis results in the accumulation of saturated very long chain fatty acids attached to glycerol (Y. Li and F. Beisson, unpublished data).

The increase and decrease in various C16 and C18 monomers that were also observed to different degrees in roots, seeds, and flowers of the mutants are not unexpected given the nature of lipid polymers. Indeed, the aliphatic monomers analyzed are not part of independent simple lipid molecules but are structurally linked to each other in a complex manner, possibly a network in which glycerol and other monomers such as dicarboxylic acids allow cross-linking. Therefore, the decrease in very long chain fatty acids resulting from the loss of GPAT5 activity could influence the incorporation of C16 and C18 monomers in a different way in seeds, roots, and flowers, depending on (1) the respective structures and compositions of aliphatic suberin in these organs, (2) the necessity of incorporation of GPAT5-provided long chain monomers before some C16-C18 chain monomers (sequential order requirement), and (3) the overincorporation of C16-C18 monomers wherever possible to maintain some structural features of the polymer (compensatory mechanism). Complex effects on polymer composition involving a sequential order requirement and a compensatory mechanism have been observed in xyloglucan chains of mutants affected in the incorporation of a specific sugar monomer (Madson et al., 2003).

Given the complexity of lipid polyester assembly, the unknown connectivity between monomers in suberin of different organs, and the functional redundancy in the GPAT5 family (see below), more complex interpretations of the polyester monomer profiles of the mutants cannot be excluded. Further experiments will be required to confirm our hypothesis and identify the precise in vivo substrates of GPAT5.

GPAT5 and Seed Coat Permeability

The polyesters of mature seed coats have been shown in several species to consist largely of cutin-type monomers, whereas typical suberin monomers have been found to be of lower abundance and thought to be restricted to a small region of the seed coat (Espelie et al., 1979). At the end of seed maturation, the import of nutrients from the mother plant via the funiculus ceases and the funiculus attachment region is sealed, leaving a scar on the seed coat, the hilum. In grapefruit (Citrus paradisi) seeds, it has been shown that the hilum region is indeed sealed by the deposition of aliphatic suberin in several cell layers (Espelie et al., 1980). Unlike Arabidopsis seeds, grapefruit seeds can be dissected and enough material from the hilum/chalazal region of the seed coat can be obtained for polyester analysis. The polyester monomers in this seed coat region are typical suberin monomers, with 22- and 24-carbon ω-hydroxy fatty acids and dicarboxylic acids representing ∼37 mol% of the total. The rest of the grapefruit inner seed coat is enriched in cutin monomers. Therefore, based on the following lines of evidence, we propose that a major role of GPAT5 in Arabidopsis seeds is to seal the hilum region by providing aliphatic monomers for suberin deposition: (1) the typical suberin very long chain monomers found in the grapefruit hilum are abundant in Arabidopsis seed coat polyester monomers, and these monomers are the ones that are specifically reduced severalfold in gpat5 mutants (Figure 4); (2) at the end of the desiccation stage, when suberin deposition in the hilum region is expected to occur, the expression of the GPAT5 gene is highest in this region (Figure 2B); (3) the kinetics and pattern of appearance of the red staining in the gpat5 embryo suggest that the tetrazolium salts diffuse mostly or exclusively through the hilum region to the embryo (Figure 6E); and (4) the hilum region of the mutant seed coat had a lower amount of insoluble lipophilic material (Figure 6G).

GPAT5 expression was also detected more broadly throughout the seed coat, especially during early seed desiccation (Figure 2A). This observation suggests a role of GPAT5 in the synthesis of the polyesters in other parts of the seed coat than the hilum region (the extrahilar region), mostly by providing suberin-like monomers. However, the extrahilar region of the seed coat seemed to remain impermeable to tetrazolium salts, which suggests that a defect in GPAT5 may be compensated by functionally redundant GPATs.

The location of the lipid polyesters in the extrahilar region of Arabidopsis is unknown. At maturity, the Arabidopsis seed coat consists of dead cells corresponding to the five cell layers of epidermal origin differentiating from the ovule integuments (L1 to L5 from the outermost to the innermost). L1 (epidermis) has a preserved structure with the thick cell walls of the columella and the mucilage, whereas layers 2 to 5 are largely collapsed and crushed together and contain the brown pigments (Haughn and Chaudhury, 2005). In mature Arabidopsis seeds, an electron-dense cutin-like layer of polyesters cannot be observed clearly on transmission electron micrographs in the L1 cell layer (epidermis) and may be located in the thick electron-dense crushed cell layers L2 to L5. Interestingly, in developing seeds, an electron-dense layer bordering the inner cell wall of L5 (endothelium), which is in contact with the embryo sac, has been observed. This layer reacts positively with osmium tetroxide and therefore looks like a cuticle that is present throughout seed development until the mature embryo stage (Beeckman et al., 2000).

The increased permeability of the seed coat in the gpat5 mutants had an effect on the germination rates under increased salt conditions, which could be physiologically important for germination and seedling establishment on various soils.

After-Ripening Requirement and Seed Coat Color in gpat5 Mutants

The darker color of the gpat5 mutants is not attributable to an increased amount of PAs (Figure 7B) but may result from a higher visibility of the brown pigments through the seed coat, to a fraction of the brown pigments that cannot be depolymerized or extracted (Routaboul et al., 2006), or from a higher degree of oxidation of PAs. Although, at this time, the connection between seed coat polyesters and seed color is not understood, we speculate that the crushed L2 to L5 layers of the final seed desiccating stage might be composed in part of a polyester network (the cuticle initially bordering L5) in close contact with oxidized polymerized PAs (brown complexes) and that disruption of the polyester network might cause some changes in the formation of these brown complexes or in their visibility.

The dormancy of the mutant seeds was released more slowly, as indicated by the time shift in germination rates measured under normal conditions (Figure 8). The germination rate was always scored 7 d after imbibition, so the difference observed was not attributable to a delay in germination in mutant seeds. Reciprocal crossing of gpat5 and the wild type showed that the embryo of the mutants was not more dormant. In Arabidopsis, seed dormancy has been described as seed coat–enhanced dormancy (Bewley, 1997). The seed coat exerts a germination-restrictive action by being impermeable to water and/or oxygen, by producing germination-inhibiting compounds, and/or by its mechanical resistance to radicle protrusion. The increase in the after-ripening period and the darker seed color we have observed in the gpat5 mutants are consistent with previous observations in Arabidopsis, in which a reduced dormancy phenotype has been observed in several seed coat mutants with reduced brown pigmentation (Leon-Kloosterziel et al., 1994; Debeaujon et al., 2000), and in several other plant seeds in which it has been noted that the darker the seed coat, the more dormant the seeds (Kantar et al., 1996). The underlying mechanism connecting seed coat color and dormancy, however, is unclear, and a direct effect of seed coat polyesters on seed dormancy cannot be completely excluded. Indeed, the reduction in mutant seed coat polyesters could alter the kinetics of critical compounds leaking from the embryo, such as nitric oxide, the accumulation of which is important for the breaking of dormancy (Bethke et al., 2006). Because germination and dormancy are of agronomic significance, the identification of lipid polyesters from the hilum or extrahilar region as a factor regulating seed dormancy may provide additional insights into the control of this process and will be worth further investigation.

GPAT5 and Root Suberin

The chemical analysis of 1- and 3-week-old roots grown on agar clearly shows that GPAT5 is required for suberin synthesis in roots. Thus, the expression of GPAT5 (as viewed by GUS staining) in the youngest part of the specialization zone (but not in the division and elongation zones) in 4-d-old roots and 1- to 4-week-old roots still elongating is consistent with a role in the initial deposition of Casparian and lamellar suberin, which occurs in the primary state of growth for all roots investigated (Enstone et al., 2003), including Arabidopsis (Di Laurenzio et al., 1996; Cheng et al., 2000). A transverse section of the differentiation zone of 4-d-old Pro_GPAT5:GUS roots showed that GUS staining was present in all cell layers (data not shown). This pattern of expression of GPAT5 was also observed in microarray data from root tissues (Birnbaum et al., 2003). Histochemical staining of Arabidopsis young roots (Di Laurenzio et al., 1996; Cheng et al., 2000; Franke et al., 2005) indicates an apparent restriction of suberized cell walls to the endodermis. The more widespread expression of the GPAT5 gene thus suggests that additional depositions of aliphatic lipid polyesters may occur in roots but might remain undetectable by the usual staining procedures, either because of a lower abundance than the endodermal suberin or a difference in structure (the usual dyes are thought to bind the aromatic domain of suberin, but their exact mode of action and their sensitivity/specificity for various lipophilic polymers remain unclear). Another possibility is that GPAT5 produces in the differentiation zone of roots a small pool of soluble glycerolipids.

The patchy expression of GPAT5 in some of the older parts of 1- to 4-week-old roots, in cells in which the initial suberin deposition has already taken place earlier during development, suggests that an additional deposition or a remodeling of suberin composition may occur later in root development before the formation of periderm. In this regard, the composition of aliphatic suberin in 3-week-old roots was changed compared with that in 1-week-old roots grown under the same conditions (Figure 4B; see Supplemental Figure 3 online). The later deposition/remodeling of suberin in apparently specific zones could have several nonexclusive causes. First, in 1- to 4-week-old roots grown on agar, the endodermis divides periclinally according to a complex helicoidal pattern and features of secondary growth are observed, which could also give rise to new suberin deposition (Dolan and Roberts, 1995; Baum et al., 2002). Second, there are at least six other GPATs expressed in roots (Figure 11), and they could contribute to this additional suberin deposition/remodeling in other zones than GPAT5 (see below). Third, the formation of lateral roots could require the expression of GPAT5, as suggested by Figure 2F. Fourth, the synthesis of suberin could be induced by environmental conditions. Concerning this latter point, it is not likely that the GPAT5 GUS staining pattern results from induction by pathogens (the roots are grown on sterile medium), but it could result from local gradients of water or nutrients on the agar plates. The understanding of the contribution of GPAT5 and the other GPAT isoforms expressed in roots to the building of the suberin depositions at various stages of Arabidopsis root development will require detailed studies of tissue gene expression in conjunction with histochemical studies of the deposition of Casparian and lamellar suberin.

Open in a separate window

Figure 11.

Gene Tree and Gene Expression Profiles of the Eight Putative GPATs of Arabidopsis.

(A) The cladogram shows the branching order of Arabidopsis GPATs according to a phylogenetic tree of protein sequences of plant acyltransferases (Kim and Huang, 2004). The original tree was built using the neighbor-joining method with 1000 bootstrap replicates. Bootstrap values are percentages.

(B) Microarray expression data derived from AtGenExpress (Schmid et al., 2005). Expression levels in each tissue (root, leaf, stem, flower, and seed) at different developmental stages were averaged (bars represent means ± se). The expression profile for GPAT7 was determined in this study via RT-PCR analysis because its expression profile is not available at AtGenExpress.

The fact that the transfer of 3-d-old germinated seedlings to a medium supplemented with 200 mM NaCl resulted on average in a fourfold higher rate of bleaching for gpat5 compared with the wild type (Figure 10) suggests that gpat5 roots are indeed affected in their ability to restrict ion movements and prevent a massive entry into the cortex, a role that has been ascribed to suberin depositions of Casparian bands (Sattelmacher, 2001; Enstone et al., 2003; Ma and Peterson, 2003). An effect attributable to osmotic stress could be ruled out by the absence of any difference in seed germination and seedling establishment between wild-type and mutant plants grown with up to 400 mM mannitol or polyethylene glycol 8000. The increased sensitivity to salt stress observed in a mutant affected in root suberin content thus confirms a role previously proposed for suberin in roots and shows that it can be critical for seedling survival at moderate salt concentrations.

Mutant Isolation

T-DNA insertional lines (SALK_018117 and SALK_142456) were identified using the SIGnAL T-DNA Express Arabidopsis Gene Mapping Tool (http://signal.salk.edu/cgi-bin/tdnaexpress) provided by the Salk Institute Genomic Analysis Laboratory (Alonso et al., 2003). SALK_018117 has a T-DNA insertion at the first exon, with the left border of the T-DNA pointing toward the 5′ end of the gene; SALK_142456 has a T-DNA insertion in the first and only intron, with the T-DNA left border pointing to the 3′ end of the gene. Individual seeds for these lines were obtained from the ABRC at Ohio State University. Plants were grown from these seeds, and for each line DNA was prepared and used for genotype screening. The gene-specific primers used for the screening of insertions into the GPAT5 gene were 5′-GCTATTTTTCCATTTGCAGATACGT-3′ (forward) and 5′-ACATCTCGGATTCTTGTCAATC-3′ (reverse) for SALK_018117 and 5′-CTAAGGAGCATCTTAGAGCAGATGA-3′ (forward) and 5′-TCCAGCGAGAACCCTATACTTATCT-3′ (reverse) for SALK_142456. These primers were used together with the T-DNA left border primer LBa1 (5′-TGGTTCACGTAGTGGGCCATCG-3′) to check for the presence of a wild-type or T-DNA mutant allele, respectively.

RNA Isolation and Gene Transcript Analysis by RT-PCR

Total RNA was prepared from rosette leaves, stems, open flowers, roots, and developing seeds of Arabidopsis plants by grinding these tissues in liquid nitrogen followed by RNA extraction with the Plant RNeasy Mini kit from Qiagen. Extracted RNA was quantified by spectrophotometer, and its integrity was verified by separating 1 μg of total RNA on a denaturing formaldehyde gel. RNA was stored at −80°C until use in reverse transcription reactions. To eliminate any contaminating DNA, RNA preparations were treated with DNase using the DNA-free kit from Ambion, and the treated RNA (1 to 5 μg) was subjected to reverse transcription using the SuperScript III first-strand synthesis system for RT-PCR (Invitrogen). The GPAT5 transcript was amplified using the gene-specific primers 5′-CTAAGGAGCATCTTAGAGCAGATGA-3′ (forward) and 5′-TCCAGCGAGAACCCTATACTTATCT-3′ (reverse), and the GPAT7 transcript was amplified using the gene-specific primers 5′-CCTTCGCCTACTTCATGCTC-3′ (forward) and 5′-GGTCTCGGGTTCATGAAAAA-3′ (reverse), with the initiation factor eIF4A-1 (At3g13920) as a control (forward, 5′-CCAGAAGGCACACAGTTTGATGCA-3′; reverse, 5′-TCATCATCACGGGTCACGAAATTG-3′).

Promoter–GUS Fusion Analysis

A 1-kb sequence upstream of the first ATG in the GPAT5 cDNA, and 588 bp of the GPAT5 coding region, were cloned in-frame as a HindIII-XbaI fragment into the binary vector pBI101.1 carrying the GUS gene downstream of the inserted promoter. The primers used for amplification of the fragment from genomic DNA were 5′-CACACAAGCTTAAAAAAGCGTTTTAATTAG-3′ (forward) and 5′-CACACTCTAGACTCACATAACGATAAGAA-3′ (reverse) (inserted restriction sites are underlined). The construct (Pro_GPAT5:GUS) was used to transform Arabidopsis wild-type plants by Agrobacterium tumefaciens vacuum infiltration (Bechtold et al., 1993). T1 seeds of Pro_GPAT5:GUS transformants were selected on sterile plates (50 μg/mL kanamycin) from surface-sterilized seeds. Resistant seedlings were transferred to soil for continued growth and seed collection. The T3 progeny from several individual kanamycin-resistant plants were analyzed for GUS gene expression. Briefly, fresh tissues were suspended and vacuum-infiltrated in a staining solution (0.1 M NaH₂PO₄, pH 7.0, 10 mM Na₂-EDTA, 0.5 mM K-ferricyanide, 0.5 mM K-ferrocyanide, 1.0 mM X-glucuronide prepared freshly each time in DMSO, and 0.1% Triton X-100 for 30 min to 1 h depending on the tissue) (Jefferson, 1987). The samples were then incubated at 37°C for 2 to 18 h until staining was visible. Stained tissues were fixed in a mixture of ethanol:acetic acid:formaldehyde (50:5:3.7, v/v) for 10 min at 65°C and destained in 80% (v/v) ethanol until removal of chlorophyll. Images of whole mount tissues were taken with a Leica MZ 12.5 microscope coupled with a digital camera.

Seed Germination and Dormancy Tests

Germination rates of wild-type and gpat5 seeds were compared for both soil-grown seeds and seeds grown on agar plates supplemented with increasing concentrations of salts. Plates were transferred to a controlled growth chamber (see Plant Materials and Growth Conditions) after cold treatment in the dark for 3 d at 4°C. Germination was scored at 7 d after transfer to the growth chamber (12 d in the case of added salts). Seeds that showed penetration of the radicle through the seed coat were counted as germinated seeds. Seedling establishment was scored by looking for the presence of two green cotyledons. The tests were repeated on at least three different batches of seeds harvested from plants grown at different times.

For dormancy release experiments, wild-type and gpat5 seeds harvested immediately after maturity from plants of the same age and grown under the same conditions were tested over a period of dry storage (0, 10, 17, 21, and 30 d at room temperature) (Bentsink and Koornneef, 2002). Mature seeds were sown in triplicate in Petri dishes (100 to 150 seeds per plate), incubated in a controlled growth chamber, and scored for germination as described above. The tests were repeated on three different batches of seeds harvested from plants grown at different times.

Staining Procedures

For seed coat permeability tests, tetrazolium red assays were used (Debeaujon et al., 2000). Briefly, Arabidopsis dry seeds were incubated in the dark in an aqueous solution of 1% (w/v) tetrazolium red (2,3,5-triphenyltetrazolium) at 30°C for 4 to 48 h. Mucilage of Arabidopsis mature seeds was stained in an aqueous solution of 0.03% (w/v) ruthenium red for 15 min at room temperature (Western et al., 2000). Seeds were rinsed in water before imaging.

For lipid polyester staining of roots, a solution of 1% (w/v) Sudan black in 75% ethanol was used. Roots were placed into this solution for 1 h, rinsed briefly in water, and imaged.

For lipid polyesters of seed coats, mature seeds were incubated for 24 h at room temperature in water containing 0.01% (w/v) Triton X-100 and 10% (v/v) commercial bleach to fade the seed coat pigments. After rinsing successively with distilled water and 100% ethanol, seeds were incubated for 30 min with chloroform:methanol (2:1, v/v), rinsed with 100% ethanol, and air-dried. Seeds were finally incubated at room temperature for 1 to 4 h with a solution of Sudan red 7B in polyethylene glycol 400:glycerol:water prepared as described by Bundrett et al. (1991), rinsed in water, mounted between slide and cover slip, and observed with a Leica MZ12.5 light microscope coupled to a digital camera.

Analysis of Fatty Acids and Cuticular Waxes

Total fatty acid content and composition in Arabidopsis seeds were quantified according to Li et al. (2006). The same method of direct methylation was used for total fatty acids of leaf, root, and flower. For cuticular waxes, the leaf tissue (1 g from 5-week-old plants) was dipped in chloroform (20 mL) for 30 s, with the addition of internal standards (20 μg/g n-octacosane, 10 μg/g docosanoic acid, and 10 μg/g fresh leaf 1-tricosanol). The epicuticular waxes were silylated to convert free alcohols and carboxylic acids to their trimethylsilyl ethers and esters, respectively, by heating the sample at 110°C for 10 min in 100 μL of pyridine and 100 μL of N,O-bis(trimethylsilyl)trifluoroacetamide. After cooling, the solvent was evaporated under nitrogen and the product was resuspended in heptane:toluene (1:1, v/v) for gas chromatography–mass spectrometry analysis (Bonaventure et al., 2003). For seed surface wax analysis, mature seeds (100 mg) were soaked in chloroform for 2 min, and internal standards (5 μg/g n-octacosane, 5 μg/g docosanoic acid, and 5 μg/g seed 1-tricosanol) were added and processed as described above for leaf cuticular wax analysis. Seed surface areas were calculated assuming a prolate spheroid geometry and using semiaxes estimated from scanning electron microscopy images of seeds.

Analysis of Lipid Polyesters

Polyester monomers were obtained and analyzed according to Bonaventure et al. (2004). Briefly, the sodium methoxide depolymerization method was used with slight modifications described by Suh et al. (2005). Polyester monomers were separated, identified, and quantified by gas chromatography–mass spectrometry. Splitless injection was used, and the mass spectrometer was run in scan mode over 40 to 500 atomic mass units (electron impact ionization), with peaks quantified on the basis of their total ion current. Polyester monomer amounts are expressed per surface area for seeds and per gram of dry residue depolymerized for roots. Further details of aliphatic and aromatic monomer identifications, analytical methods, and seed coat and embryo polyester monomer localization experiments are presented by Molina et al. (2006).

Analysis of PAs

Analysis was performed on samples of 15 mg of mature seeds using acid-catalyzed depolymerization and spectrophotometric quantification as described previously (Routaboul et al., 2006).

Scanning Electron Microscopy and Fluorescence Microscopy

Arabidopsis pollen and mature seeds were coated with gold particles during 3 min at a coating rate of ∼7 nm/min using an EMSCOPE SC500 sputter coater (Ashford). Scanning electron microscopy images were taken with a JEOL JSM-6400V microscope. Assuming prolate spheroid seed geometry, semiaxes were estimated from scanning electron microscopy images using Analysis PRO software version 3.2 (Soft Imaging System).

For seed coat autofluorescence, mature seeds were illuminated and observed with a Zeiss Axiophot microscope using epifluorescence interference and absorption filters (excitation filter, 365 nm; dichromatic beam splitter, 395 nm; long-pass emission filter, 435 nm) coupled to a digital camera.

Accession Numbers

Arabidopsis Genome Initiative locus codes are as follows: GPAT1, At1g06520; GPAT2, At1g02390; GPAT3, At4g01950; GPAT4, At1g01610; GPAT5, At3g11430; GPAT6, At2g38110; GPAT7, At5g06090; and GPAT8, At4g00400.

We thank Ewa Danielewicz (Center for Advanced Microscopy, Michigan State University) for performing scanning electron microscopy of Arabidopsis seeds, Abraham Koo and Alicia Pastor (Michigan State University) for their help with fluorescence and light microscopy, and Laurent Nussaume (Commissariat à l'Energie Atomique, Cadarache, France) for his help with Arabidopsis root sections and helpful comments. This work was supported in part by the Dow Chemical Company, Dow AgroSciences, and the National Research Initiative of the USDA Cooperative State Research, Education, and Extension Service (Grant 2005-35318-15419).