Functional Characterization of the Arabidopsis b -Ketoacyl-Coenzyme A Reductase Candidates of the Fatty Acid Elongase 1[W][OA]

In plants, very-long-chain fatty acids (VLCFAs; . 18 carbon) are precursors of sphingolipids, triacylglycerols, cuticular waxes, and suberin. VLCFAs are synthesized by a multiprotein membrane-bound fatty acid elongation system that catalyzes four successive enzymatic reactions: condensation, reduction, dehydration, and a second reduction. A bioinformatics survey of the Arabidopsis ( Arabidopsis thaliana ) genome has revealed two sequences homologous to YBR159w encoding a Saccharomyces cerevisiae b -ketoacyl reductase (KCR), which catalyzes the ﬁrst reduction during VLCFA elongation. Expression analyses showed that both AtKCR1 and AtKCR2 genes were transcribed in siliques, ﬂowers, inﬂorescence stems, leaves, as well as developing embryos, but only AtKCR1 transcript was detected in roots. Fluorescent protein-tagged AtKCR1 and AtKCR2 were localized to the endoplasmic reticulum, the site of fatty acid elongation. Complementation of the yeast ybr159 D mutant demonstrated that the two KCR proteins are divergent and that only AtKCR1 can restore heterologous elongase activity similar to the native yeast KCR gene. Analyses of insertional mutants in AtKCR1 and AtKCR2 revealed that loss of AtKCR1 function results in embryo lethality, which cannot be rescued by AtKCR2 expression using the AtKCR1 promoter. In contrast, a disruption of the AtKCR2 gene had no obvious phenotypic effect. Taken together, these results indicate that only AtKCR1 is a functional KCR isoform involved in microsomal fatty acid elongation. To investigate the roles of AtKCR1 in postembryonic development, transgenic lines expressing RNA interference and overexpression constructs targeted against AtKCR1 were generated. Morphological and biochemical characterization of these lines conﬁrmed that suppressed KCR activity results in a reduction of cuticular wax load and affects VLCFA composition of sphingolipids, seed triacylglycerols, and root glycerolipids, demonstrating in planta that KCR is involved in elongation reactions supplying VLCFA for all these diverse classes of lipids.

Very-long-chain fatty acids (VLCFAs) with chain lengths between C20 and C34 are essential and ubiquitous constituents of eukaryotic cells. They are most commonly found as building blocks of sphingolipids, but they are also important components of glycerophospholipids, triacylglycerols, sterol esters, and wax esters. Depending on their cellular and tissue locali-zation and chain length, VLCFAs perform a wide range of physiological and structural roles. For example, they are involved in the stabilization of highly curved nuclear pore membranes (Schneiter et al., 1996(Schneiter et al., , 2004 and the creation of membrane domains involved in lipid and protein trafficking (Dickson et al., 2006;Toulmay and Schneiter, 2007) and cell signaling (Leonard et al., 2002). In mammals, C 20 VLCFAs also serve as precursors of biologically active eicosanoids implicated in inflammation responses (Karimi et al., 2007) and are critical components required for myelin production in the central nervous system (Poulos et al., 1992) as well as for normal functioning of photoreceptor cells in the retina (Zhang et al., 2001;McMahon et al., 2007). Furthermore, VLCFAs play an important structural role in skin barrier function (Wertz and Downing, 1983;Westerberg et al., 2004;Vasireddy et al., 2007). Similarly, in higher plants, VLCFAs give rise to cuticular waxes, lipid molecules deposited on the primary plant surfaces that serve as a barrier preventing excessive water loss and pathogen attack (Jenks et al., 1994), and mediate plant-insect interactions (Eigenbrode and Espelie, 1995) and pollenstigma signaling required for fertilization (Preuss et al., 1993). Modified VLCFAs are also present in suberin, an extracellular plant polyester that controls water and solute fluxes in plant tissues (Bernards, 2002;Franke and Schreiber, 2007).
Biochemical studies of the VLCFA biosynthesis and analyses of mutants with defects in fatty acid elongation revealed that multiple FAEs with unique substrate chain length specificities are involved in generating the complete array of C 20 to C 34 acyl chains required by eukaryotic cells (Sprecher, 1974;von Wettstein-Knowles, 1993). The KCS of the elongase determines the substrate specificity of each elongation reaction (Millar and Kunst, 1997). Consistent with the requirement for fatty acyl chains of diverse lengths, a large family of 21 FAE1-like KCS sequences has been annotated in the Arabidopsis (Arabidopsis thaliana) genome, together with an unrelated ELONGATION DEFEC-TIVE (ELO)-like family of four putative condensing enzymes . To date, only a few of these condensing enzymes have been studied in any depth, and with the exception of FAE1, a seed-specific condensing enzyme involved in C 20 and C 22 fatty acid biosynthesis for seed storage lipids (Kunst et al., 1992), and CER6, a condensing enzyme that catalyzes the elongation of fatty acyl-CoAs longer than C 22 in the epidermal cells of Arabidopsis (Millar et al., 1999;Fiebig et al., 2000), the functions the of other condensing enzymes in VLCFA elongation have not been established in planta, although several additional FAE1-like KCSs have been functionally characterized via heterologous expression in yeast (Trenkamp et al., 2004;Paul et al., 2006).
In contrast to the KCSs that have strict substrate and tissue specificities, the other three enzymes of the fatty acyl elongase, the KCR, the HCD, and the ECR, are believed to be common to all VLCFA biosynthetic reactions, have (presumed) broad substrate specificities, and are expressed in all cells (Millar and Kunst, 1997). Although these elongase component enzymes have been identified in Saccharomyces cerevisiae (Kohlwein  Beaudoin et al., 2002;Han et al., 2002;Denic and Weissman, 2007), this hypothesis could not be tested until recently because the corresponding enzymes and genes were not available from a multicellular organism. A single ECR gene has been confirmed and characterized in Arabidopsis and shown to encode a functional ECR that physically interacts with the Elo2p and Elo3p condensing enzymes when expressed in yeast (Gable et al., 2004). The Arabidopsis ECR was shown to be identical to CER10 (Zheng et al., 2005), the protein defective in the cer10 mutant isolated by Koornneef et al. (1989). Biochemical analysis of the cer10 mutant demonstrated that the ECR gene product is involved in VLCFA elongation required for the synthesis of many VLCFA-containing lipids, including cuticular waxes, seed triacylglycerols, and sphingolipids (Zheng et al., 2005). Similarly, the newly identified HCD PASTICCINO2 (Bach et al., 2008) is encoded by a single gene in Arabidopsis. Partial loss of PAS2 function in the pas2-1 mutant is associated with a general reduction of VLCFA content in seed storage triacylglycerols and complex sphingolipids as well as reduced levels of cuticular waxes that are generated from VLCFA precursors. A complete loss of PAS2 activity is embryo lethal (Bach et al., 2008). Two KCR genes, named GL8A and GL8B, are present in the maize (Zea mays) genome (Xu et al., 1997;Dietrich et al., 2005) and are required for cuticular wax accumulation on seedling leaves. However, they are not expressed only in the epidermis, but throughout the plant. Attempts to generate double mutants by crossing gl8a 3 gl8b single mutant lines failed because embryos carrying both mutations were not viable. Thus, in addition to its wax biosynthetic role, KCR has an essential function in maize, perhaps in the production of sphingolipids (Dietrich et al., 2005), but this idea has not been experimentally verified. Two KCR orthologs have also been annotated in the Arabidopsis genome (Dietrich et al., 2005). One of these genes, At1g67730, functionally complements the KCRdeficient ybr159wD mutant of S. cerevisiae , but the specific role of the protein that it encodes has not been determined. The other KCR candidate gene, At1g24470, has not been analyzed at all. In Figure 2. Arrest of atkcr1/atkcr1 embryo development. Six siliques of different developmental ages growing on heterozygous AtKCR1/atkcr1 plants were dissected. A to F, DIC images of wild-type-looking green seeds at globular (A), heart (B), early torpedo (C), late torpedo (D), and bent cotyledon (E) stages of development and at maturity (F). G to L, DIC images of atkcr1/atkcr1 mutant embryos from white seeds developing in the same silique arrested at globular stage. Bar = 50 mm. this study, we employ Arabidopsis insertional mutants disrupted in the KCR genes and transgenic lines in which KCR genes have been silenced by introduction of RNA interference (RNAi) or overexpression constructs to investigate the in planta functions of the two KCR proteins. We demonstrate that these two proteins are divergent and that At1g67730-encoded enzyme is the functional KCR of the Arabidopsis FAE essential for embryo development and plant morphogenesis.

Arabidopsis Has Two KCR Genes
The isolation of the yeast KCR gene YBR159w resulted in the identification of a single Arabidopsis ortholog, At1g67730 , which we refer to as AtKCR1. Further exploration of the Arabi-dopsis genome revealed another KCR-like sequence, At1g24470 (Dietrich et al., 2005), which we designate AtKCR2. AtKCR1 has a 957-bp open reading frame (ORF) that encodes a protein of 318 amino acids with a predicted molecular mass of 35.7 kD. AtKCR2 has a 939-bp ORF that encodes a protein of 312 amino acids with a predicted molecular mass of 35.0 kD. Both these proteins have a putative NADH binding site and contain highly conserved Ser-Tyr-Lys residues essential for catalysis , though a dilysine motif required for ER retention is obvious only in AtKCR1 (Supplemental Fig. S1). Unlike the maize orthologs GL8A and GL8B that are 97% identical and have partially redundant biochemical functions (Dietrich et al., 2005), AtKCR1 and AtKCR2 exhibit only 45% amino acid identity (Table I), suggesting that they could play diverse roles in VLCFA production in Arabidopsis.

Isolation of Mutants Disrupted in the AtKCR Genes
To investigate the in planta function of the two predicted AtKCR enzymes, we initiated a reverse genetic approach by identifying insertional mutants disrupted in AtKCR1 and AtKCR2 genes. A single line, RATH12-5282-1-G, obtained from Plant Functional Genomics Research Group of RIKEN Genomic Sciences Center (Kuromori et al., 2004) was found to carry a Ds transposon in the second intron of the AtKCR1 gene. The exact site of insertion was determined by sequencing the PCR product generated using a combination of gene-and Ds transposon-specific primers (Fig. 1A). Surprisingly, no homozygous atkcr1/atkcr1 plants were identified by PCR genotyping of the progeny of the self-fertilized insertion line, and the segregation of wild-type to heterozygous plants was approximately 1:2 (47:103), suggesting that homozygous kcr1 loss-of-function mutation may be embryolethal. To test this hypothesis, we dissected the siliques developing on heterozygous AtKCR1/atkcr1 plants and compared them with the siliques from the wild-type plants. Siliques from the wild type contained uniformly green developing seeds, whereas siliques from the self-fertilized AtKCR1/atkcr1 heterozygotes contained approximately 25% (28/120) of aborted seeds.
Initially, aborted seeds appeared white or transparent in color at a stage when wild-type appearing siblings were green (Fig. 1C). Later on, the aborted seeds were dark brown, shrunken, and deformed ( Fig. 1D). Attempts to confirm by PCR amplification that these aborted seeds were indeed homozygous kcr1 mutants failed. However, reverse transcription (RT)-PCR analysis using AtKCR1-specific primers showed reduced AtKCR1 transcript levels in AtKCR1/atkcr1heterozygotes and no detectable transcript in the presumed homozygous white seeds (Fig. 1E).
Seeds from siliques developing on heterozygous plants were further analyzed by differential interference contrast (DIC) microscopy ( Fig. 2) to compare the progression of embryogenesis in these two seed types. In young siliques, all the seeds were white and contained embryos at the globular stage. However, in older siliques, only wild-type-appearing green seeds continued to develop normally with embryos proceeding from globular stage ( Fig We also obtained two T-DNA insertion lines in the AtKCR2 gene, SALK_096487 (Alonso et al., 2003) and SAIL _536_H04 (Sessions et al., 2002), with T-DNA insertions in the second and fourth exons, respectively (Fig. 1B). No visible developmental phenotypes were observed in homozygous atkcr2 progenies from either T-DNA mutant line.

AtKCR1 and AtKCR2 Expression Domains Overlap and They Both Reside in the ER
To determine if the two apparent Arabidopsis homologs are both constitutively expressed and to delineate their expression domains, we performed RT-PCR analyses and b-glucuronidase (GUS) assays in transgenic plants transformed with AtKCR1promoter-GUS and AtKCR2promoter-GUS constructs. As shown in Figure 3, both genes are expressed in green siliques, flowers, inflorescence stems, and leaves, with the highest expression levels detected in siliques and flowers. High AtKCR1 transcript levels were also present in roots, whereas AtKCR2 transcript was almost undetectable in root tissues. These expression results were confirmed by GUS activity assays (Fig. 4).
In addition, AtKCR1promoter-and AtKCR2promoterdirected GUS activity was also detected in embryos of different stages (Fig. 4, A-F and L-R) and young developing seedlings (Fig. 4, G and S) but was absent from mature seeds (Fig. 4, K and W) and stem bases (data not shown).
To compare the subcellular localization of the AtKCR1 and AtKCR2, yellow fluorescent protein (YFP)-AtKCR1 and cyan fluorescent protein (CFP)-AtKCR2 fusion constructs were initially transiently expressed in tobacco (Nicotiana tabacum) under the control of the 35S promoter. Visualization of tobacco leaves revealed that both proteins were localized in the ER (data not shown). We also introduced both fluorescent protein fusion constructs into Arabidopsis. To ensure that the YFP-AtKCR1 is fully functional, we expressed it in AtKCR1/atkcr1 heterozygous plants. Sixteen out of 20 recovered independent transgenic lines had siliques with only green wild-type-looking seeds and no white seeds, indicating complementation ( Fig. 5B) and demonstrating that the mutation in the Figure 6. AtKCR1 complements fatty acid elongation in yeast. Gas chromatography-flame ionization detection analysis of FAMEs prepared from wild-type (WT) or mutant (ybr159D) yeast coexpressing AtFAE1 (A) or IgELO9 (B) and either a vector control, yeast YBR159, AtKCR1, or AtKCR2 gene.
AtKCR1 gene is responsible for the embryo lethality. In all the complemented YFP-AtKCR1 transgenic lines examined by confocal microscopy, YFP fluorescence labeled a reticulate network typical of the ER (Fig. 5C). Because the disruption of AtKCR2 did not result in a detectable phenotype, we could not verify if the CFP-AtKCR2 fusion protein is active by complementation of a loss-of-function mutant, but we decided to proceed with its localization. For this purpose, we transformed the 35Spromoter-CFP-AtKCR2 fusion construct into wild-type Arabidopsis. This experiment revealed that the AtKCR2 protein also resides in the ER (Fig. 5G).
Only AtKCR1 Complements the Yeast ybr159D Mutant and Embryo Lethality of the Arabidopsis atkcr1 Mutant AtKCR1 has been previously shown to restore heterologous elongase activity of the yeast ybr159D mutant similar to the expression of native yeast KCR gene YBR159 , indicating that the Arabidopsis and yeast enzymes are functionally equivalent and that AtKCR1 also likely catalyzes the reduction of the b-ketoacyl-CoA. This result was confirmed in this study (Fig. 6). By contrast, the expression of AtKCR2 failed to restore heterologous fatty acid elongation mediated by either FAE1 or the Isochrysis galbana D9-ELO-like elongating activity (Qi et al., 2002) in the ybr159D yeast mutant (Fig. 6) and also failed to rescue the slow growth phenotype of ybr159D yeast cells associated with impaired endogenous elongase function (data not shown), suggesting that AtKCR2 has no KCR activity in yeast.
The fact that atkcr1 loss-of-function mutation is embryo lethal suggests that either AtKCR2 is not a functional KCR or that AtKCR2 is not expressed during embryogenesis. GUS activity assays in transgenic plants transformed with the AtKCR2promoter:GUS construct (Fig. 4, L-W) and global gene expression analyses during Arabidopsis embryogenesis (Schmid et al., 2005) indicate that AtKCR2 is transcribed at all times during embryo development, albeit at a much lower level than AtKCR1. To determine if the low expression level of AtKCR2 was the reason why AtKCR2 could not support embryogenesis in the absence of AtKCR1 or whether the AtKCR2 was indeed inactive, we expressed AtKCR2 using the AtKCR1 promoter in the AtKCR1/atkcr1 background. In contrast to AtKCR1, which complemented the atkcr1 mutant ( Fig. 5B), AtKCR2 could not rescue the embryo lethality of atkcr1 even when expressed in the AtKCR1 manner behind the AtKCR1 promoter (data not shown), confirming that it is not a functional KCR isoform.

RNAi-Suppressed and Cosuppressed AtKCR1 Plants Display Fused Vegetative and Reproductive Organs and an Abnormal Root Morphology
Embryo lethality prevented further phenotypic analysis on homozygous atkcr1 lines, and no developmental differences or changes in seed VLCFA composition or cuticular wax accumulation were detected in AtKCR1/atkcr1 heterozygotes in comparison with wild-type plants. We therefore generated transgenic plants in which AtKCR1 was down-regulated by RNAi to characterize their morphological and biochemical phenotypes and determine the role of AtKCR1 in the different VLCFA metabolic pathways.
To test the specificity of our RNAi construct and ensure that the phenotypes that we describe are due to suppression of AtKCR1 alone, we performed RT-PCR analysis of the RNAi suppressed plants using mRNA prepared from flowers that showed high KCR1 and KCR2 expression in the wild type in previous assays (Fig. 3). This experiment confirmed that the RNAi construct specifically down-regulates the expression of only AtKCR1, but not AtKCR2 (Supplemental Fig. S2).
Twenty T 1 RNAi transformants were chosen for further evaluation. These plants displayed a variety of phenotypes, including growth retardation and fused rosette leaves (Fig. 7, A and D). Individuals with the most severe morphological phenotype never formed rosettes, but instead produced incompletely developed leaf-like structures (Fig. 7A). They were also highly sensitive to dehydration and needed to be cultivated in a controlled environment with atmospheric humidity above 85% in wet soil at all times. Eleven T 1 plants failed to grow to maturity, developed necrotic spots, and died. From the nine remaining plants, three (designated 4, 6, and 7) required mechanical organ separations to allow distorted leaves and inflorescence stems to develop (Fig. 7, C and D).
Soil-grown T 2 progeny of these AtKCR-RNAi primary transformants exhibited two predominant phenotypes: glossy inflorescence stems (Fig. 7G) or twisted and curled fiddlehead (fdh)-like stems that were not obviously glossy (Fig. 7H). Both glossy and nonglossy plants were present in the progeny of the 4, 6, and 7 lines. These three lines, together with a fourth line (line 9, showing a less extreme phenotype), were selected for a detailed morphological and biochemical characterization. Examination of the flowers from glossy plants revealed an unusual bushy stigma morphology with longer papillae and stamen with shorter filaments unable to position anthers at the height of the stigma, similar to that reported for cer10 (Zheng et al., 2005; Fig. 7E). These plants also produced short and crooked siliques, containing mostly aborted seeds, suggesting a problem with pollen viability/fertility (Fig. 7F). Cross-pollination of glossy plants with wildtype pollen resulted in longer siliques and a considerably increased number of seeds produced (data not shown). Similar short and crooked siliques were also found on the nonglossy plants. Interestingly, mature dried seeds collected from both glossy and nonglossy AtKCR-RNAi plants were considerably larger than seeds collected from wild-type plants grown under identical conditions (Table II).
An aspect of the AtKCR-RNAi phenotype that has not previously been reported in other mutants impaired in VLCFA biosynthesis is the abnormal root morphology of the glossy plants (Fig. 7J). When grown horizontally on plates, the roots of these plants ran on the surface of the gel matrix, lacked lateral branches, and had dramat- ically reduced root hairs (Fig. 7J, inset). Closer observation revealed that lateral root primordia were formed but did not elongate. In contrast, roots of nonglossy plants with curly inflorescences appeared wild-type.
To determine if the phenotypes observed were due to AtKCR1 RNAi suppression, AtKCR1 transcript levels in transgenic lines were determined by RT-PCR in 15-d-old-seedlings and in a number of tissues from 8-week-old T 2 plants. As shown in Figure 8, in the developing seedlings and in all of the tissues analyzed, the levels of AtKCR1 transcripts were lower than in the wild type. The reduction in KCR transcripts was more pronounced in the glossy plants compared with the more limited reduction observed in the nonglossy lines. Within each line and for each plant phenotype (glossy or nonglossy), levels of AtKCR1 transcripts were similar in all tissues. These results show a clear correlation between reduced levels of AtKCR1 transcripts and increased severity of the phenotype in the T 2 generation.
We also overexpressed the AtKCR1 cDNA under the control of 35S promoter. About 50% of transgenic lines displayed abnormal phenotypes, similar to those of the AtKCR1-RNAi lines, including dwarfism, curly cauline leaves, fused flower buds, and bright green glossy stems (Supplemental Fig. S3, A-C). To investigate if these phenotypes were caused by AtKCR1 cosuppression, AtKCR1 transcript levels in a variety of tissues of transgenic plants were determined by RT-PCR. As shown in Supplemental Figure S3D, in cauline leaves, stems, and floral buds, the organs exhibiting the most pronounced phenotypes, AtKCR1 transcript levels were considerably lower than in the wild type.

RNAi-Suppressed AtKCR1 Plants Have Abnormal Trichome and Epidermal Cell Morphology
As previously reported for cer10, fdh, wbc11, and wax2/pel6, both glossy and nonglossy AtKCR-RNAi plants display clearly altered trichome morphology (Fig. 9, A-F). As shown in Figure 9B, trichomes are sparsely distributed, have fused branches in glossy plants, and have shorter or bent branches in nonglossy plants (Fig. 9C). Scanning electron microscopy examination showed that the basal cells of these trichomes are swollen and bulging out of the epidermal surface (Fig. 9, D-F). A further inspection of the surface of young leaves revealed impaired integrity of the epidermis in glossy plants. Groups of cells appeared swollen to the point were they did not adhere tightly to each other, resulting in deep intercellular grooves (Fig.  9, G-I) and probably contributing to the extreme sensitivity of glossy AtKCR1-RNAi plants to dehydration. This type of abnormal epidermal structure could not be detected in nonglossy plants.

Cuticular Wax Load Is Reduced in AtKCR1-RNAi Lines
Organ fusions and stem glossiness observed in the AtKCR1-RNAi plants are typically associated with defects in plant cuticles. Cuticle abnormalities can be detected by a toluidine blue (TB) test, which results in staining of organs with permeable cuticles (Tanaka et al., 2004). We used the TB test to evaluate cuticles of glossy line 4 and nonglossy line 9 individuals (Fig.  10A). The test resulted in complete staining of line 4 glossy seedlings, indicating severe cuticular abnormalities of this RNAi line. Nonglossy line 9 seedlings displayed a range of partially stained patterns, including patchy or random staining as observed for fdh and cer14 mutants (Fig. 10A; Tanaka et al., 2004). Because VLCFAs are precursors of all the aliphatic components of cuticular wax, we suspected that cuticle permeability of AtKCR1-RNAi lines may at least in part be due to reduced wax accumulation. We therefore examined the glossy inflorescence stem surfaces of the AtKCR1-RNAi plants by scanning electron microscopy. As shown in Figure 10B, wild-type stems are densely covered with epicuticular wax crystals. In contrast, glossy stem surfaces of AtKCR1-RNAi plants are virtually devoid of wax crystals.
Wax analyses confirmed major reductions in wax loads in AtKCR1-RNAi plants. The total wax coverage on wild-type Columbia-0 (Col-0) inflorescence stems under the conditions used here was 1086 mg per g fresh weight. In contrast, wax loads measured in glossy AtKCR1-RNAi plants reached 124.2 mg per g fresh weight (11.4% of the wild type) in line 7 (7 GL) and 42.8 mg per g fresh weight (3.9% of the wild type) in line 6 (6 GL; Fig. 11). Quantification of wax components demonstrated that reduced wax loads were due to decreased accumulation of all compound classes (Fig. 11). In nonglossy AtKCR1-RNAi plants, the decrease in wax load was far less pronounced, and it was proportional to the levels of AtKCR1 transcripts detected in the stem of those plants (Fig.  8). Wax loads measured for nonglossy plants from lines 7 and 9 were 811 and 869 mg per g fresh weight (75% and 80% of the wild type), respectively (Fig. 11, inset). However, these data indicate that while lines 7 and 9 were phenotypically classified as nonglossy, they do have a reduction in wax load, as might be expected for plants with an impaired capacity to synthesize VLCFAs.

VLCFA Content Is Decreased in Root Lipids and Seed Triacylglycerols of AtKCR1-RNAi Lines
Since Arabidopsis leaves contain very low levels of VLCFAs, we analyzed the fatty acid composition in the roots of wild-type and AtKCR-RNAi plants that exhibit abnormal root morphology (Fig. 7J). A clear reduction in the content of C 20 , C 22 , and C 24 fatty acids was detected in glossy lines 4 and 6 (Fig. 12, A and B). VLCFAs accounted for 5.6% of the total root fatty acid extracted from wild-type Col-0 plants grown in this study, predominantly reported in phospholipids, such as phosphatidylethanolamine and phosphatidylserine (Devaiah et al., 2006). The VLCFA content measured in glossy plants was reduced to 1.4% and 1.7% in lines 4 and 6 (24% and 29% of the wild type), respectively. The VLCFA content in nonglossy plants was similar to that of wild-type plants with 5.3% and 5.6% measured in lines 7 and 9, respectively. These results correlated well with the levels of VLCFAs measured in the acyl-CoA pool in roots of the same plants (Fig. 12C).
Surprisingly, in the root fatty acid methyl ester (FAME) extracts prepared from glossy plants, we detected three novel peaks (Fig. 12, A and B, peaks 1-3) that were either absent or low in wild-type extracts. These peaks were tentatively identified by mass spectrometry as C16 and C18 a,v-dicarboxylic acids and an v-hydroxy fatty acid, compounds that are abundant in the cutin and suberin in Arabidopsis (Bonaventure et al., 2004, Franke et al., 2005, Beisson et al., 2007. The presence of these usually highly polymerized constituents in the AtKCR1-RNAi FAME Figure 9. Glossy AtKCR1-RNAi plants have abnormal trichomes and epidermal structures. A to C, Primary leaves of 2-week-old wild-type (A), line 4 glossy (B), and line 9 nonglossy (C) T 2 seedlings. D to F, Trichomes from the cauline leaves of wild-type (D) and glossy plants (E and F). G to I, Abnormal epidermal structure of the abaxial surface of cauline leaves in glossy plants. In the scanning electron micrographs (D-I), bars = 50 mm, except in E, where the bar = 10 mm. samples after mild methanolic acid extraction may reflect the altered composition and/or structure of the suberin when the synthesis of longer C 20 , C 22 , and C 24 compounds is reduced. The nonglossy plant extracts did not contain the three additional compounds founds in the roots of glossy plants.
Because Arabidopsis seed oil also contains VLCFAs, we examined the fatty acid composition and content of mature dry seeds. Compositional analyses of oil produced by the AtKCR1-RNAi seeds revealed that only the oil from glossy plants was depleted in C 20 and C 22 saturated and monounsaturated fatty acids. The total VLCFAs content in these seeds was about 30% lower compared to the wild type (Table III; Fig. 13). The larger seeds collected from both glossy and nonglossy AtKCR1-RNAi plants had a higher fatty acid content than wild-type seeds (Tables II and III). Overall, the increase in total fatty acid content was largely proportional to the increase in seed size and did not result from a higher proportion of oil as a percentage of the seed weight (data not shown).

Sphingolipid Analyses
To evaluate the effect of reduced AtKCR1 expression on VLCFA incorporation into sphingolipids, we analyzed the complex sphingolipid composition in the stems of glossy and nonglossy AtKCR1-RNAi plants that all exhibit abnormal shoot morphology. All the examined lines accumulated free long-chain bases (LCBs), which are characteristic of impaired sphingolipid metabolism as reported for the yeast FAE mutants (Kohlwein et al., 2001; Fig. 14A). In addition, a moderate but statistically significant decrease in almost all the sphingolipid species was detected (Fig.  14B). Conversely, neither the accumulation of free LCBs nor the reduction in any of the sphingolipid species could be detected in the stems of morphologically normal plants (data not shown). This suggests that the twisted shoot phenotype may also be caused by altered sphingolipid structure and not only the defective cuticular lipid composition as proposed for the fdh mutant.

DISCUSSION
This study describes the functional characterization of the Arabidopsis orthologs of the yeast KCR YBR159w and confirms the central role of this enzyme activity in the synthesis of VLCFAs in plants, as well as the essential nature of these fatty acids. Like maize, Arabidopsis contains two KCR-like sequences (KCR1 and KCR2), but they do not share the same high degree Figure 12. Fatty acid composition in the root of wild-type and AtKCR1-RNAi T 2 lines. Gas chromatographyflame ionization detection (FID) trace of FAMEs prepared from the roots of wild-type and AtKCR1-RNAi 15-d-old seedlings. In addition to the expected reduction in C20+ fatty acids, several elevated peaks were identified as C16:0 and C18:0 dicarboxylic acids (peaks 1 and 2, respectively) and an v-hydroxylated fatty acid (peak 3). The star indicates a phenolic compound, most likely ferulic acid. B, Quantitation of variation in root fatty acid composition in wild-type and AtKCR1-RNAi lines. WT, Wild type. C, Acyl-CoA profiling of the roots of wild-type and AtKCR1-RNAi 15-d-old seedlings. Each value is the mean of four measurements. Error bars indicate SD. GL, Glossy; NGL, nonglossy. of identity exhibited by the monocot genes. Our data indicate that only KCR1 is a bona fide KCR of the FAE in Arabidopsis, based on several independent criteria. First, KCR1 but not KCR2 is capable of rescuing fatty acid elongation in the yeast ybr159wD mutant. Second, the expression of the KCR2 ORF under the control of the KCR1 promoter is unable to suppress the embryo lethality of the kcr1 insertional mutant. Finally, while insertional mutations of KCR1 are embryo lethal, similar mutations of KCR2 have no obvious phenotypic effect and do not affect the composition of lipids that are known to accumulate or are derived from VLCFAs, such as seed storage lipids, cuticular waxes, and sphingolipids. It is possible that this apparently normal phenotype of the kcr2 mutant is a result of compensation by the KCR1 protein, but we have not detected increased KCR1 transcript levels in the kcr2 mutant to support this idea. The precise function of KCR2 remains to be determined, although it is still likely to be as a member of the short-chain dehydro-genases/reductase superfamily. Thus, Arabidopsis differs from maize, in which two closely related KCRs (GL8A and GL8B) both contribute to the activity of the microsomal elongase.
Since KCR1 represents the sole microsomal elongase KCR, it is not surprising that disruption of the KCR1 gene is embryo lethal. This also indicates that the capacity to synthesize VLCFAs is crucial at early stages of embryo development and is in agreement with the recent data of Bach et al. (2008) showing that the dehydratase PAS2, another core microsomal elongase component, is essential during embryogenesis. It is interesting to note that while PAS2 and KCR1 are both essential single genes in Arabidopsis, the CER10 gene is not, even though it represents the sole ortholog of the yeast TSC13 ECR. This very likely indicates the presence of structurally unrelated but functionally equivalent forms of the ECR in Arabidopsis.
To evaluate the requirement for VLCFAs during the growth and development of Arabidopsis after embryo-  genesis, we generated a number of transgenic lines with reduced KCR1 transcript levels via RNAimediated silencing. Such plants displayed a number of developmental and biochemical phenotypes that correlated with the relative reduction in transcript levels for KCR1. In particular, lines with more pronounced reduction in KCR1 transcripts showed an almost complete loss of cuticular waxes and a strong fiddlehead-like phenotype, whereas less severe reduction in KCR1 transcripts still generated an fdh phenotype, but only a small reduction in waxes. Given that the KCSs required for wax synthesis (predominantly CER6) is expressed at significantly higher levels than the FDH KCS (Joubès et al., 2008), the phenotypes that result from the knockdown of KCR1 may reflect not only the reduction in this elongase component activity, but a perturbation to the normal stoichiometry of different KCS activities. Thus, in the absence of sufficient KCR to interact with all the different KCSs present in a cell, the most abundant KCS could titrate out other less abundant KCSs. Such a model could explain the occurrence of an fdh phenotype in the absence of a glossy phenotype. Alternatively, loss of KCR activity could result in reduced activity of all KCSs, mediated through a currently undefined homeostatic process leading to a more uniform reduction in the synthesis of VLCFAs. However, in both scenarios, the outcome is a perturbation (reduction) in the normal total synthesis of VLCFAs as mediated by the activity of several different KCSs (which could include both FAE-like KCSs and ELO-like forms). Given our observations of discrete phenotypes, such as fdh, in the absence of significant reduction in wax loads, we favor the titration model. Our data confirm the essential nature of VLCFAs in Arabidopsis development, though given the evidence that these fatty acids accumulate in virtually all classes of lipids (sphingolipids, waxes, triacylglycerols, suberin, and some phospholipids), the precise reason(s) for the observed embryo lethality in insertion mutants and gross phenotypic perturbations in the KCR1-RNAi knockdowns requires further investigation. For example, there is clear evidence for an absolute requirement for VLCFAs in Arabidopsis for the synthesis of sphingolipids (Chen et al., 2006;Dietrich et al., 2008;Teng et al., 2008), similar to the situation observed in yeast (Dickson et al., 2006). However, recent studies have shown that it is possible to generate viable yeast mutants with sphingolipids entirely devoid of VLCFAs   (Cerantola et al., 2007), even though the yeast double KCS mutant elo2D/elo3D, which is unable to synthesize VLCFAs, is not viable (Oh et al., 1997). Thus, the assumption that the lack of VLCFAs critically impairs the formation of essential sphingolipids may not hold true in all cases. It is also interesting to note that a suppressor screen for yeast mutants capable of overcoming an absence of sphingolipids recovered a mutant allele (slc1-1) of a phospholipid acyltransferase SLC1, resulting in the abnormal accumulation of VLCFAs in yeast phospholipids (Nagiec et al., 1993). These studies indicate that the cellular function of VLCFAs, even in extensively studied single-cell model systems, is still only poorly understood.
The results obtained in our study do provide some new insights into the role of VLCFAs in Arabidopsis. First, it is clear that the ability to make waxes is intimately linked to the capacity to synthesize VLCFAs, with some of the KCR1-RNAi knockdown lines showing an almost complete absence of waxes. As previously reported, such wax-deficient mutants are still viable, but they have to be grown at high humidity. Thus, the essential nature of VLCFAs is unlikely to relate to defects in wax biosynthesis. Our study also shows that the more severe KCR1-RNAi knockdown lines are not able to form lateral roots and root hairs, which correlates with a strong reduction in C20+ saturated fatty acids. In Arabidopsis roots, saturated VLCFAs are present in phospholipids, predominantly phosphatidylethanolamine, and are also components of suberin. Strikingly, in lines 4 and 6, which show the strongest defect in lateral root formation, we detected the presence of several suberinderived dicarboxyclic acids in simple fatty acid methyl-ester extractions. Dicarboxyclic acids are not normally extracted under such mild conditions and may indicate a perturbation to the VLCFA-containing suberin polymer in the root. Further work is needed to determine if impaired root development is linked to phospholipid-or suberin-related VLCFA deficiency.
One final role of VLCFAs that needs to be considered relates to their potential involvement in the establishment of cellular polarity, most likely through the action of VLCFA-containing sphingolipids. Sphingolipids, in conjunction with sterols, have been shown to be enriched in Arabidopsis detergent-insoluble membrane fractions (Borner et al., 2005), consistent with their presence in so-called lipid rafts. Auxin carrier proteins, such as the PINs, have promiscuous (apolar) targeting in plants that are disrupted in sterol metabolism, and this has been hypothesized to relate to defects in endocytosis (Feraru and Friml, 2008). PIN1 has also recently been shown to be present in detergent-insoluble membranes (Titapiwatanakun et al., 2009). One scenario that cannot be excluded is that the KCR1 knockdown phenotypes result from perturbations to cell polarity of some critical cell types. This may be due to disruption of membrane composition, either the formation of plasma membrane microdomains (e.g. lipid rafts) or membrane trafficking as described for the ECR mutant (Zheng et al., 2005). Currently almost nothing is known in plants about the role of VLCFA-containing lipids in these processes, and as such, this represents a topic ripe for further research.

Plant Material and Growth Conditions
The Ds transposon insertional line (RATH12-5282-1-G) was obtained from Plant Functional Genomics Research Group of RIKEN Genomic Sciences Center (Kuromori et al., 2004), whereas SALK_096487 and SAIL_536_H04 T-DNA insertional lines were ordered from the Arabidopsis Biological Resource Center (www.arabidopsis.org). The seeds were germinated on AT-agar plates (Somerville and Ogren, 1982) and grown in soil (Sunshine Mix 5; SunGro) at 20°C under continuous light (90-120 mE m 22 s 21 photosynthetically active radiation).
After transformation with the KCR1-RNAi construct, T 1 seeds were surface sterilized, spread evenly onto agar plates containing kanamycin, stratified for 3 d in a cold room, and germinated at 22°C under continuous light (150 mE m 22 s 21 ). After 2 weeks, kanamycin-resistant plant were transferred to soil and grown in a controlled environment cabinet with a photoperiod of 16 h light (150 mE m 22 s 21 ) at 22°C and 8 h dark at 18°C, keeping the soil damp and atmospheric humidity above 80%.

Wax and Fatty Acid Extraction and Analysis
Cuticular wax was extracted and analyzed as described by Rowland et al. (2006). For the determination of the seed fatty acid composition, FAMEs were prepared by incubation with methanolic-HCl (Supelco) at 80°C for 2 h, followed by extraction with hexane and analysis by gas-liquid chromatography (Rossak et al., 2001).

Extraction and Analysis of Fatty Acids, Acyl-CoAs, and Sphingolipids
Total fatty acids from leaves and root tissues and from yeast cultures were extracted and methylated as described before (Sayanova et al., 1997;Napier et al., 1998). Total FAMEs were analyzed by gas chromatography and identified by mass spectrometry. Acyl-CoA extraction and profiling was carried out as described by Larson and Graham (2001). Sphingolipids were extracted from root and shoot tissues and quantified as detailed by Markham and Jaworski (2007).

Total RNA Isolation and Semiquantitative RT-PCR Analysis
Total RNAs from aerial tissues of 2-to 6-week-old Arabidopsis (Arabidopsis thaliana) plants were extracted using TRIZOL reagent (Invitrogen Life Technologies) following the manufacturer's instructions. Total RNA from root tissue was isolated from 14-d-old seedlings grown on AT-agar plates placed vertically. For RT-PCR analysis of AtKCR1 mRNA levels in seeds from immature siliques, total RNA (2.5 mg) from each sample, treated with RNase-free DNase (Promega), was used for reverse transcriptase reactions. First-strand cDNA was synthesized with random hexamers using a SuperScript first-strand synthesis system according to the manufacturer's manual (Invitrogen Life Technologies), and the constitutive expression gene glyceraldehyde-3-P dehydrogenase C subunit was used as a control for RT-PCR experiments. One microliter of reverse transcription reaction mixture was used as a template in a 20-mL PCR. The primer sequences used were as follows: AtKCR1 forward primer 5#-GGGGACAAGTTTGTACAAAAAAGCAGGCTTCATGGAGAT-CTGCACTTACTTCAAAT-3# (bold sequences = directional attB sites) and AtKCR1 reverse primer 5#-GGGGACCACTTTGTACAAGAAAGCTG-GGTCTTCTTTCTTCATGGAGTCTTTTTGG-3#. The amplification conditions were 94°C (30 s), 56°C (30 s), and 72°C (30 s), and the number of cycles varied from 28 to 32 for different genes. Each RT-PCR was repeated twice. For RT-PCR analysis of AtKCR1 and AtKCR2 mRNA levels in 2-to 6-week-old plant tissues, total RNA (2.5 mg) from each sample, treated with Turbo DNA-free DNase (Ambion), was used for reverse transcriptase reactions. First-strand cDNA was synthesized using oligo(dT) 12-18 primer and SuperScript II reverse transcriptase according to the manufacturer's instructions (Invitrogen). The constitutive expression gene tubulin (a-1 tubulin) was used as a control for RT-PCR experiments. One microliter of reverse transcription reaction mixture was used as a template in a 25-mL PCR. The primer sequences used were as follows: AtKCR1 forward primer 5#-GGCCGGTACCATGGAGATCTGCACTTAC-3# and AtKCR1 reverse primer 5#-GGCCCTCGAGTCATTCTTTCTTCATGGA-3# or AtKCR2 forward primer 5#-GGCCGGTACCATGCAGGGAGCATGCATCTC-3# and AtKCR2 reverse primer 5#-GGCCCTCGAGTTAAGATAAACTTCTTCTGC-3#. The amplification conditions were 94°C (30 s), 55°C (30 s), and 72°C (60 s), and the number of cycles varied from 25 to 28 for different genes. Each RT-PCR was repeated three times.

Functional Complementation of the Yeast ybr159wD
Mutant by AtKCR1 or AtKCR2 Wild-type and ybr159wD mutant yeast strains expressing either AtFAE1 or IgELO9 (Isochrysis galbana; IgASE1) pYES2 constructs have been described before , Qi et al., 2002. The AtKCR2 ORF was amplified by PCR using primers AtKCR2-Fwd (5#-GGCCGGATCCATGCAGGGAG-CATGCATCTC-3#) and AtKCR2-Rev (5#-GGCCGGTACCTTAAGATAAA-CTTCTTCTGC-3#). The amplified sequence was restricted using BamHI and KpnI (underlined in the forward and reverse primers, respectively), purified using the Qiagen PCR purification kit, and ligated into a BamHI/KpnI cut pESC-TRP plasmid vector (Stratagene). For functional complementation studies, the ybr159wD strains described above were cotransformed with pESC-TRP constructs containing either YBR159w, AtKCR1 , or AtKCR2 (this study). The empty pESC-TRP vector was used as negative control. Yeast cells were cultivated and the transgenes induced in the presence or absence of exogenously supplied fatty acid substrate as described previously .

GUS Activity Assays
A length of 780 bp of the 5# promoter region of AtKCR1 was amplified from genomic DNA by PCR using forward primer 5#-GGGGACAAGTTTGTA-CAAAAAAGCAGGCTTCACCCTTTGGACTTACCAACG-3# and reverse primer 5#-GGGGACCACTTTGTACAAGAAAGCTGGGTCAGGTTGA-GACTTTGGAGATGAGA-3#. The amplified product was introduced into Gateway entry vector by Gateway BP Clonase enzyme mix (Invitrogen). The confirmed entry clones were transferred into destination vector pMDC163 (Curtis and Grossniklaus, 2003) carrying the GUS gene. The generated construct was introduced into wild-type Col-0 plants by Agrobacteriummediated transformation.

TB Test
For rapid visualization of leaf cuticle defects, we used the TB test described by Tanaka et al. (2004) with the following differences. After surface sterilization, seeds were sown on plate containing half-concentration Murashige and Skoog (1962) medium solidified with 0.5% (w/v) phytagel (Sigma-Aldrich).

Microscopy
For the YFP or CFP fusion protein localization analysis, leaf samples of transgenic plants were examined using confocal laser scanning microscope as described by Zheng et al. (2005). Excitation light wavelength was 458 to 514 nm, the emission filter set at 505 to 530 nm, and autofluorescence was detected using an emission filter set at 600 to 650 nm.
Siliques of different developmental stages from heterozygous plants were dissected. Ovules from individual siliques were mounted on slides in Hoyer's clearing solution (chloral hydrate, water, glycerol, 8:2:1 v/v) and cleared overnight at 4°C (Liu and Meinke, 1998). The cleared ovule was observed with a LSM 510 Meta DIC microscope (Carl Zeiss).
For electron microscopy, sections of stem and leaves were mounted on Al cryo stubs using Optimal Cutting Temperature compound (Ted Pella) and plunge frozen in slushed liquid nitrogen. The samples were then transferred to the cryo chamber stage under vacuum and etched for 2 min at 295°C. The stage temperature was returned to 2175°C, and the samples were coated for 90 s with Au Pd. The samples were then loaded on a JEOL 6700 FEG scanning electron microscope chamber stage maintained at 2160°C for imaging using the on board software.

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S1. Sequence comparisons of AtKCR1 and AtKCR2 with homologs from other species.
Supplemental Figure S3. The phenotypes of cosuppressed plants.

ACKNOWLEDGMENTS
We thank the Genomic Analysis Laboratory of the Salk Institute and RIKEN Genomic Sciences Center for providing Arabidopsis T-DNA insertion mutants, the Bioimaging Facility at the University of British Columbia for providing microscopy and technical support, and Jean Devonshire and the