An Arabidopsis fatty acid elongase gene,KCS1, with a high degree of sequence identity toFAE1, encodes a 3-ketoacyl-CoA synthase which is involved in very long chain fatty acid synthesis in vegetative tissues, and which also plays a role in wax biosynthesis. Sequence analysis ofKCS1predicted that this synthase was anchored to a membrane by two adjacent N-terminal, membrane-spanning domains. Analysis of a T-DNA taggedkcs1–1mutant demonstrated the involvement of theKCS1in wax biosynthesis. Phenotypic changes in thekcs1–1mutant included thinner stems and less resistance to low humidity stress at a young age. Complete loss ofKCS1expression resulted in decreases of up to 80% in the levels of C26 to C30 wax alcohols and aldehydes, but much smaller effects were observed on the major wax components, i.e. the C29 alkanes and C29 ketones on leaves, stems and siliques. In no case did the loss ofKCS1expression result in complete loss of any individual wax component or significantly decrease the total wax load. This indicated that there was redundancy in the elongase KCS activities involved in wax synthesis. Furthermore, since alcohol, aldehyde, alkane and ketone levels were affected to varying degrees, involvement of theKCS1synthase in both the decarbonylation and acyl-reduction wax synthesis pathways was demonstrated.
Plant surfaces are covered by a cuticle which serves as the first line of defense against pathogens, phytophagous insects and environmental stresses such as drought, UV damage and frost. The cuticle is comprised of a cutin polymer, derived from 16 and 18 carbon hydroxylated fatty acids, which is embedded with waxes ( Kolattukudy 1980). The embedded epicuticular wax (EW), which also extends above and over the cuticle, often forms crystalline structures on the surface which give the plant a white to bluish-grey appearance. This wax layer waterproofs the cuticle, reducing non-stomatal water loss and allowing rain to be shed from the plant surfaces. Cuticular waxes most often consist of a homologous series of saturated, straight very long chain (> 18 carbons in length) fatty acids, alcohols, aldehydes, alkanes, ketones, β-diketones and wax esters (for reviews see Post-Beittenmiller 1996;von Wettstein-Knowles 1995). Waterproofing waxes are also found to be associated with suberin which is located in subterranean organs (roots and tubers), wound sites and internal structures such as the Casparian strip in roots and stems and the bundle sheaths in grasses ( Espelie & Kolattukudy 1979;Kolattukudy 1980). The suberin polymer is derived primarily from ω-hydroxy fatty acids and dioic fatty acids, and suberin waxes are primarily very long chain alkanes, alcohols, fatty acids, and occasionally wax esters ( Walton 1990). Cutin, suberin and waxes are all derived from fatty acid precursors, and their physical, functional and biosynthetic relationships strongly suggest that wax production may be co-regulated with cutin and suberin production.
The very long chain fatty acids (VLCFA) required for wax synthesis are produced by membrane-bound, multi-enzyme acyl elongase systems which catalyze a series of biochemical reactions similar to those of de novo fatty acid synthase. These steps consist of the condensation between an acyl primer and a donor malonyl-CoA, followed by a reduction, dehydration and second reduction, resulting in a fully reduced and saturated acyl-CoA which is two carbons longer. Seven sequential rounds of elongation are required to extend an 18 carbon primer to a 32 carbon fatty acid. The condensation reaction is believed to be the rate-limiting step and the component most sensitive to substrate chain length. There are clearly multiple elongase systems to carry out the several sequential rounds of chain extension ( von Wettstein-Knowles 1995). However, the number of individual elongase systems required and the extent of overlap of substrate chain length specificities of the various condensing enzyme components among sequential elongase systems is not known. The condensing enzyme component of seed-specific fatty acid elongases has been cloned from jojoba ( Lassner et al. 1996 ), Arabidopsis ( James et al. 1995 ) and Brassica ( Barret et al. 1998 ), each catalyzing only two or three rounds of elongation.
In Arabidopsis, wax biosynthesis involves modification of most VLCFA by either the acyl reduction or decarbonylation pathway ( Hannoufa et al. 1993 ;Jenks et al. 1995 ). In the acyl reduction pathway, VLCFA undergo two reductions generating first aldehydes and then primary alcohols. This two-step reduction has been shown in Euglena ( Kolattukudy 1970), jojoba seed ( Pollard et al. 1979 ) and pea leaves ( Vioque & Kolattukudy 1997) to be carried out by a single enzyme, with the aldehyde intermediate remaining bound to the enzyme. The resulting primary alcohols may become esterified to fatty acids forming wax esters. In the decarbonylation pathway, VLCFA are reduced to aldehydes which then undergo decarbonylation, resulting in odd chain alkanes. Alkanes may be further modified by hydroxylation, most often on the central carbon of the odd chain alkane, and the resulting secondary alcohol may be further oxidized to the corresponding ketone. In pea, a reductase generating aldehydes rather than primary alcohols has been described and is thought to be involved in alkane biosynthesis, separate from primary alcohol production ( Vioque & Kolattukudy 1997). In Arabidopsis, the nature of the reduction reactions for the two associated pathways is not known.
Both the reduction and the decarbonylation pathways use precursors of VLCFA, but whether the associated pathways share acyl elongases or use separate but parallel systems is not fully understood. Several studies in barley using condensing enzyme inhibitors, as well as mutants defective in epicuticular wax production, indicate that separate acyl elongation systems produce VLCFA for β-diketone synthesis and for alkane and primary alcohol syntheses ( von Wettstein-Knowles 1995). At present, direct evidence for or against separate elongation systems for the associated reduction (alcohol production) and decarbonylation (alkane production) pathways is lacking. In addition, there have been no studies demonstrating whether cutin wax biosynthesis and suberin wax biosynthesis share common acyl elongation systems.
Considerable insight into wax biosynthesis has been obtained by analysis of wax defective mutants. These mutants, which are called eceriferum (cer) or glossy (gl) mutants, have a reduced or altered epicuticular wax bloom, and have bright, shiny or glossy stems or leaves. Twenty-one cer loci have been identified in Arabidopsis ( Koorneef et al. 1989 ;Lemieux 1996;McNevin et al. 1993 ), three of which, CER1, CER2 and CER3, have been cloned ( Aarts et al. 1995 ;Negruk et al. 1996 ;Xia et al. 1996 ). These 21 loci have been divided into four classes based on the degree of glossiness and on secondary phenotypes, such as male sterility, susceptibility to wilt, and slender stems ( Koorneef et al. 1989 ). There are 15 gl mutants in maize ( Schnable et al. 1994 ), of which four, Gl1, Gl2, Gl8 and Gl15, have been cloned ( Hansen et al. 1997 ;Moose & Sisco 1996;Tacke et al. 1995 ;Xu et al. 1997 ). Mutants of CER2 or Gl2 have a wax composition which is consistent with a block in acyl elongation. However, when CER2 and Gl2 were cloned and the protein sequence deduced ( Negruk et al. 1996 ;Tacke et al. 1995 ;Xia et al. 1996 ), they were found to be homologues and the gene products had no similarity with components of fatty acid synthase, as would be expected if CER2 and Gl2 encoded acyl elongase components. Furthermore, CER2 has been shown to be a nuclear-localized protein, suggesting it may have a regulatory role in wax production ( Xia et al. 1997 ). Gl8 has sequence similarity with 3-ketoacyl reductases and thus it may be a reductase component of an acyl elongation system ( Xu et al. 1997 ). CER6 may also be involved in acyl elongation based on the mutant wax composition ( Preuss et al. 1993 ), but to date, this gene has not been cloned and its function remains to be determined.
One reason why it has been difficult to clearly show whether the associated reduction and decarbonylation pathways for cuticular wax biosynthesis share the same elongation system(s) is attributable to the non-specific nature of chemical inhibitors. Furthermore, no wax defective mutants have yet been shown to affect the condensing enzyme components of acyl elongases. Ideally, to address the question of shared elongase systems, the loss of activity of a single elongase component should be evaluated on each associated pathway. In the studies presented here we have identified an insertional mutation which effectively blocks all expression of KCS1, a gene encoding an elongase 3-ketoacyl-CoA synthase (KCS) that contributes to wax biosynthesis. Isolation of this mutant allowed us to specifically examine the role of a single elongase condensing enzyme in the production of VLCFA for both the reduction and decarbonylation associated pathways generating cuticular and suberin waxes.
Isolation and characteristics of the KCS1 clone
A BLASTN search of the Arabidopsis expressed sequence tag (EST) GenBank database using the nucleotide sequence coding for the ORF of the jojoba FAE gene ( Lassner et al. 1996 ), identified several clones having high similarity to the jojoba FAE sequence. One of these clones, ATTS1282, was requested from the Arabidopsis stock center, and once sequenced, comparison with the jojoba FAE ORF sequence indicated that ATTS1282 was a partial cDNA clone. A high stringency screening of an A. thaliana ecotype Columbia, genomic DNA library, probed with ATTS1282, resulted in the isolation of a full length genomic clone. In keeping with the terminology used by Lassner et al. (1996) , we will refer to clone ATTS1282 and its full length counterparts as KCS1, for 3-ketoacyl-CoA synthase.
Similar to FAE1, the 1560 bp ORF of the KCS1 gene lacked introns. The predicted molecular mass of the 520 aa KCS1 protein (58.3 kDa) falls within the range observed for other elongase KCSs such as FAE1 (56 kDa) and jojoba (58.6 kDa). The alignment of KCS1 with seed-specific elongase KCSs of A. thaliana, B. napus and jojoba, showed that these proteins share a high degree of identity ( Fig. 1). The KCS1 synthase was most closely related to the jojoba KCS (55%), but also shared substantial identity with the KCS from Brassica (53%) and the Arabidopsis FAE1 KCS (53%). A phylogenetic analysis of these sequences based on the Jotun Hein sequence alignment also indicated that the KCS1 from Arabidopsis and the KCS from jojoba were most similar.
The four KCSs contain six completely conserved cysteine residues. Lassner et al. (1996) tentatively identified the active site cysteine of the jojoba KCS based on sequence similarity to the active site cysteine of the related resveratrol synthase ( Lanz et al. 1991 ). The sequence bracketing Cys250 of the KCS1 synthase was identical to the putative active site of the jojoba KCS. Analysis of the KCS1 for transmembrane spanning domains using multiple alignments with other KCS genes (TMAP) ( Persson & Argos 1994) predicted that there were only two spanning domains. These were adjacent to each other and located near the N-terminus of the protein (residues 41–71 and 85–113), suggesting that the KCS1 synthase is anchored to the membrane in this region.
Expression of KCS1 in yeast
The similarity of the KCS1 synthase sequence to those of other elongase KCSs strongly suggested that the KCS1 gene product was also an elongase condensing enzyme. To obtain direct biochemical evidence that KCS1 encoded an elongase KCS, we expressed a full length KCS1 cDNA in Saccharomyces cerevisiae under the direction of a gal-inducible promoter. Microsomes were isolated from transformed yeast cells grown on galactose and assayed for elongase activity. These microsomes elongated [1–14C] 18:0-CoA to 26:0-CoA ( Table 1). The overall elongation activity of KCS1 synthase towards 18:0-CoA was 66 pmol min–1 mg–1 protein higher than the activity of the control microsomes. In contrast to 18:0-CoA, 18:1-CoA served as a less suitable substrate for elongation by the KCS1 synthase. The activity of KCS1 synthase towards 18:1-CoA was only 28 pmol min–1 mg–1 protein greater than the activity obtained with the control microsomes (52.5 pmol min–1 mg–1). These data confirmed that KCS1 encoded an elongase condensing enzyme and suggested that this enzyme was most active with saturated fatty acids.
Table 1. In vitro elongation of [14C] 18:0-CoA by the microsomes from yeast expressing either the pYEUra3 vector without insert or the KCS1λYES construct. Assays were run for 30 min and the elongated products were analyzed as described in Experimental procedures
In an effort to identify the in vivo function of the KCS1 gene product, we used a PCR-based strategy to screen a population of 13 800 T-DNA tagged plants ( Fig. 2) and were able to identify a tagged mutant (kcs1–1) in this population. This strategy was similar to that used by Krysan et al. (1996) , except that the use of nested primers enabled us to use the pooled genomic DNA from the entire population of 13 800 plants as the template in our initial screen. PCR-based analysis of the tagged (kcs1–1) allele, using the primer sets LB1/A1 and S3/A1, indicated that the T-DNA insertion site was located approximately 450 bp from the 5′ end, or about one-third of the way into the KCS1 ORF.
After narrowing the kcs1–1 to a pool of 100 T3 seeds, 540 seeds were sown and a single tagged T4 plant was identified. PCR analysis of leaf genomic DNA revealed that this plant was hemizygous for the T-DNA insert, i.e. it contained an intact KCS1 as well as kcs1–1 (see below).
Seeds from this tagged T4 plant were collected, germinated, and the segregating T5 population was screened for plants which were homozygous for the T-DNA insert in the KCS1 gene (kcs1–1 mutant), hemizygous for the insert and for those which were homozygous for KCS1 (wild-type). Genomic DNAs were isolated from 48 individual plants of the segregating population and analyzed by PCR using primers to identify KCS1 (S3 and A1, see Fig. 2) or kcs1–1 (LB1 and A1). Reactions generating a 900 bp PCR product identified template from plants containing KCS1 and those generating a 450 bp band identified template from plants containing kcs1–1 ( Fig. 3a). The template that produced both bands came from hemizygous plants. The lack of a S3/A1 PCR product in some T5 plants indicated that the KCS1 gene was only represented in the genome by a single copy. The genotypic segregation ratio of homozygous KCS1:hemizygous kcs1–1:homozygous kcs1–1 was 12:27:9. The χ2 was 0.8, indicating that the observed population did not differ significantly from the expected ratio of 1:2:1.
Although the PCR analysis of the kcs1-1 mutant suggested that KCS1 was encoded by a single gene, we confirmed this by Southern analysis ( Fig. 4). When subjected to a low stringency wash, multiple bands were detected in each of the lanes probed with KCS1 ( Fig. 4a), pointing to the existence of a family of related FAE genes in Arabidopsis. This was not surprising given the degree of sequence identity found between KCS1 synthase and the 3-ketoacyl synthases from other plant species ( Fig. 1), and by the fact that BLASTN searches of the Arabidopsis expressed sequence tag (EST) database, using the sequence of the KCS1 ORF, identified several different cDNA clones for putative elongase KCSs. Increasing the washing stringency removed all but the strongest of the signals, leaving only a single band in each lane ( Fig. 4b).
To determine that no portion of the KCS1 gene was being transcribed, total RNA was isolated from leaves of the wild-type T6 segregrants and from a mixture of stem and leaf tissue from two individual homozygous T6 kcs1–1 segregants. Northern analysis of these RNAs is shown in Fig. 3(b). Expression of the KCS1 gene was clearly evident in the RNA isolated from the wild-type plant (lane 1), but was not detected in RNA from either of the homozygous kcs1–1 plants (lanes 2 and 3). This Northern blot was reprobed with a portion of the constitutively expressed β-tubulin gene to determine the levels of RNA in each lanes. The β-tubulin signal for RNA from both of the kcs1–1 mutants was a minimum of twofold stronger than that found with RNA isolated from the wild-type plants (data not shown). All subsequent analyses of the kcs1–1 mutant involved comparisons of T6 plants that were homozygous for either kcs1–1 or KCS1.
Since approximately 42% of the plants in the original tagged population contained multiple T-DNA inserts ( Feldman et al. 1994 ), genomic DNAs from six different wild-type T6 KCS1 plants were screened for the presence of additional T-DNA inserts. PCR was used with primers specific to the T-DNA left border (LB3 and LB4, Fig. 2). Since the T6 seed used was pooled seed from the segregating T5 plants, and no signal was amplified from any of the genomic templates, the original T4 parent plant most probably contained only a single T-DNA insertion.
Wax and lipid composition of kcs1–1 plants
Very long chain fatty acids are precursors for epicuticular wax (EW), glucocerebrosides and some acyl lipids. Since KCS1 encoded an elongase KCS, we compared the compositions of each of these species from the kcs1–1 and KCS1 plants to determine if kcs1–1 had altered the composition of any of these. EW are composed of long chain hydrocarbons derived from C30 and C32 fatty acids and thus it is probable that several elongase KCSs are required for the synthesis of these very long chain fatty acids. Surface waxes were isolated from various tissues of mature tagged and wild-type plants, and the results of the GC/MS analysis are shown in Fig. 5. When compared to wild-type, EW isolated from the first 6 cm of stem of the kcs1–1 mutants ( Fig. 5a), displayed decreases of 42% and 35% in the C30 aldehydes and alcohols, respectively. There was also a 70% decrease in the amount of C30 fatty acid, but no significant changes in the levels of C26 or C28 alcohols and aldehydes or alkanes and ketones were observed. The composition of the waxes isolated from the upper stem (above 6 cm) of the tagged plants was similar to that of the lower stem in that decreases in the levels of C30 alcohol and aldehyde represented the largest changes (29% and 36%, respectively, Fig. 5b). Unlike the lower stem of kcs1–1 plants, the upper stem had significant decreases in the levels of C26 and C28 alcohols (14% and 12%, respectively), C28 aldehyde (31%), C29 alkane (13%) and C29 ketone (10%). However, the level of C30 fatty acid on the upper stem of the tagged plants was not significantly different from that observed in the wild-type plants.
Analysis of EW from siliques of tagged plants revealed their composition to be very similar to that of waxes isolated from the upper stem, although decreases observed in the levels of both alcohols and aldehydes were substantially larger ( Fig. 5c). The levels of C26, C28 and C30 alcohols in the silique EW from the kcs1–1 mutants decreased by 39%, 25% and 39%, respectively, whereas the decreases in C28 and C30 aldehyde levels were greater, at 47% and 64%, respectively.
The decreases in alcohol levels observed in EW from the stem and siliques were also observed in the wax composition from rosette and cauline leaves of the tagged plants ( Fig. 5d,e). The levels of C26, C28 and C30 alcohols in the rosette leaf wax from kcs1–1 mutants decreased by 80, 60 and 34%, respectively, but were less dramatic in cauline leaf wax, where decreases of 57, 41 and 19%, respectively, were observed. Comparable decreases of 85, 80 and 62% were measured in the levels of C26, C28 and C30 fatty acids, respectively, in the rosette leaf, but no significant decreases were observed in fatty acids of cauline leaf wax. No aldehydes or C24 and C30 fatty acids were detected in the cauline leaf wax from wild-type or kcs1–1 lines. Small decreases of 19% and 11% were measured for the C29 and C31 alkane levels of the cauline leaf wax from the tagged plants, but the alkane levels remained unchanged in the waxes of the rosette leaves.
The results presented in Fig. 5 indicated that the absence of a functional KCS1 altered the EW composition from all tissues examined, primarily showing consistent decreases in the amounts of very long chain alcohols and, to a lesser extent, in the levels of very long chain aldehydes. These relative changes differed among the tissues examined, however. In waxes isolated from stem and siliques, the levels of C30 alcohol and aldehyde were most affected in the mutant line, whereas in leaf waxes the levels of C26 and C28 alcohols were affected to the largest extent.
Glucocerebrosides are a minor, but presumably important, component of the plasma membrane and in Arabidopsis they contain low levels of very long chain fatty acids ( Uemura et al. 1995 ). Analysis of the glucocerebrosides from leaves of both the tagged and wild-type plants revealed that there were no changes in their acyl composition (data not shown). Similarly, analysis of the total fatty acid composition of both the leaf and seeds from the two plant lines resulted in identical fatty acid profiles (data not shown).
Visual characteristics of kcs1–1 plants
Our initial comparison of the two plant lines showed that growth of the kcs1–1 plants displayed minor differences to that of wild-type segregrants when grown under conditions of high relative humidity (RH). Germination rates of seeds from both plant lines were greater than 80%, with flowering and silique formation occurring at similar times. The kcs1–1 plants produced 20–30% fewer flowers than did the wild-type line, and siliques from the tagged plant were on average 60–70% as long as their wild-type counterparts. No increased surface glossiness was observed on any organ of the kcs1–1 mutants, indicating that surface wax loads were not dramatically reduced. The most notable visual difference between the two plant lines was in the thickness of their stems ( Fig. 6a). The kcs1–1 mutants had consistently thinner stems, although stems from both lines grew to similar heights. The average fresh weight of the lower 6 cm of stem from the tagged plants (5.73 ≥ 0.4, SE, n = 18) was approximately 50% that observed for stems from wild-type plants (11.3 mg ≥ 0.9, SE, n = 9).
Effect of relative humidity on growth of kcs1–1 plants
The wax analysis of kcs1–1 demonstrated an altered wax composition. Since wax has been shown to play a role in non-stomatal water loss ( Lemieux et al. 1994 ), we examined if kcs1–1 might affect the ability of this plant to respond to changes in RH. When we examined germination and growth rates under a variety of conditions, it became apparent that young kcs1–1 plants were more sensitive to sudden changes in RH than KCS1 plants were. Our first experiment involved germinating seeds on soil covered with plastic wrap at high RH. Under these conditions, the mutant and wild-type seeds had similar germination rates (24 of 30 and 26 of 30 seeds, respectively), and displayed normal growth and development. After 5 d of growth, the plastic wrap was removed, rapidly exposing the seedlings to a lower RH environment. Twelve days after removal of the plastic wrap, 30% (8 of 24) of the mutant seedlings had died, compared to only 12% (4 of 26) of the wild-type seedlings. The surviving mutant and wild-type seedlings developed normally and were identical in size and appearance. Interestingly, the higher sensitivity of the kcs1–1 seedlings to rapid changes in RH was apparent only in younger plants. When kcs1–1 plants were grown for 10 or 15 days before removing the plastic wrap, equivalent mortality rates of 11–14% were observed for both the kcs1–1 and wild-type seedlings. When germinated on soil in uncovered pots the kcs1–1 plants were more affected by the low humidity conditions than the wild-type plants grown under similar conditions. After 15 days growth, fewer than 10% of the mutant plants were still living compared to greater than 80% of the wild-type plants. Furthermore, the kcs1–1 plants were considerably smaller and less well developed than the wild-type plants ( Fig. 6b,c).
Consistent with the above data, we previously observed that when germinated on filter paper and then transplanted to soil, growth of the kcs1–1 seedlings was slower than that of the wild-type. To examine this further, seeds were germinated and grown on filter paper for 8 days, and then 30 kcs1–1 or KCS1 seedlings were transplanted to soil and grown either uncovered or covered with plastic wrap. Nine days after transplanting, only two of the original 30 kcs1–1 seedlings that were uncovered displayed new growth. The remaining 28 plants were either dormant with respect to growth or had died. The kcs1–1 seedlings that were covered initially had a slow but normal pattern of development, but eventually they also began to display visible signs of stress. After 9 days almost 70% of these covered mutants had severely chlorotic cotyledons, of which only about 50% continued to develop and survive. In contrast, the wild-type plants, whether covered and uncovered, displayed normal growth and development at this point, and were a minimum of threefold larger than the slowly developing kcs1–1 mutants.
After determining that EW production was altered in stems, siliques and leaves of the kcs1–1 mutants, a Northern analysis was carried out to determine the expression pattern of KCS1 ( Fig. 7). Expression of KCS1 occurred in all tissues examined, including both silique and leaf, but appeared to be particularly strong in stem, flower and 5 d cotyledons and roots.
The strong expression of KCS1 in young roots indicated that the KCS1 synthase may also be involved in suberin wax and/or suberin production in this tissue. Analysis of methyl esters prepared from total fatty acids extracted from 10 d root tissue indicated that the levels of dioic acids were significantly increased in the roots from the kcs1–1 mutants ( Fig. 8). Dioic acids are major components of suberin monomers and are thought to be involved in the construction of linear polymers ( von Wettstein-Knowles 1993). In comparison with the wild-type roots, the levels of hexadecanedioic (C16:0), octadecanedioic (C18:0) and octadecenedioic (C18:1) acids increased 2.3-,1.7- and twofold, respectively, in roots from the kcs1–1 lines. Other significant differences between the tagged and wild-type plants included decreases of 73% and 48% in the levels of C16:2 and C22:0 FAMEs, respectively, and increases of 1.5- and 1.4-fold in the levels of C16:0 and C18:3 FAMEs, respectively. Attempts to measure the levels of waxes present in the root tissue were unsuccessful. The amount of root material used (25–40 mg) was comparable to that used for the other tissue, indicating that the level of epicuticular wax and/or suberin-associated wax in the roots was very low.
Fatty acid elongase systems are responsible for the synthesis of VLCFA in plants, and it has been suggested that the substrate specificity of these systems resides almost exclusively in the activity of the elongase KCSs ( Millar & Kunst 1997). Determining the metabolic function of a specific elongase KCS is complicated by the potentially large number of KCS genes needed to synthesize VLCFA for multiple uses. Major uses of VLCFA in A. thaliana include storage lipids, which contain C20:1 and C22:1 fatty acids, and as precursors to surface waxes. In some plants, an additional requirement for VLCFA is in ceramide synthesis ( Lynch 1993), although in Arabidopsis, VLCFA are only a minor component of the ceramides ( Uemura et al. 1995 ).
To date, characterization of elongase KCS genes is limited to those with seed-specific expression which are involved with synthesis of VLCFA for storage lipids in A. thaliana, jojoba and B. napus ( Barret et al. 1998 ;James et al. 1995 ;Lassner et al. 1996 ). A question largely unstudied concerns the role of other elongase KCSs in other metabolic pathways. For example, in Arabidopsis, synthesis of precursor C30 fatty acids occurs in all major aerial organs, requiring at least six cycles of elongation. The number of different elongase KCSs required for C30 fatty acid synthesis, the degree of overlapping activities, and the amount of tissue-specific expression of elongase KCSs are essentially unknown. It is likely that tissue specific expression occurs since there is one report of differential expression in the flower and stem of two unidentified Arabidopsis ESTs which had high sequence similarity to the seed-specific FAE1 ( Millar & Kunst 1997).
Sequence analysis and comparison of KCS1 to other elongase KCS genes strongly suggested that this gene encoded an elongase KCS. This was confirmed by analysis of its activity after KCS1 expression in yeast. Furthermore, the KCS1 synthase appeared to be most active with saturated fatty acids, and Northern analysis demonstrated that KCS1 was expressed in a wide variety of tissues. Taken together, these data suggested that KCS1 was potentially involved in wax biosynthesis, and the availability of the tagged kcs1–1 allele provided an excellent tool to examine the in vivo function of KCS1.
Analysis of the surface wax composition of the kcs1–1 plants indicated that KCS1 did in fact play a role in epicuticular wax biosynthesis. Consequently, this kcs1–1 analysis constitutes the first study in any organism of a mutation in an elongase condensing enzyme which results in an altered wax phenotype. The number of genes associated with wax metabolism that have been previously characterized is very small. Of the seven wax genes that have been cloned, only Gl8 from maize is thought to be a component of a fatty acid elongase system and, in that case, it was a 3-ketoacyl reductase ( Xu et al. 1997 ). The wax analyses of kcs1–1 provided several useful insights into wax metabolic pathways. These analyses suggest first that there is redundancy and overlap of elongase KCS activities and that the extent of the overlap varies from tissue to tissue, and second that the acyl-reduction and decarbonylation pathways share elongase systems.
The evidence for redundancy of elongase KCSs resulting in overlapping activities was apparent from the observation that there were no complete blocks in the chain length extension of any wax component. The C26–C30 alcohols, C30 aldehydes, and C29 alkanes and ketones were most affected, but usually the reduced levels of these components caused by kcs1–1 were less than 50%. Only in the case of C26–C28 alcohols in leaves did kcs1–1 cause decreases greater than 50%. However, even in this case, longer chain alcohols were less affected, indicating ample flux through the pathway. This is consistent with the observation that the total wax load on kcs1–1 plants was only slightly lower than wild-type. The extent of overlapping activities varied between tissues, as illustrated by comparison of wax alcohols on stem with wax alcohols on siliques and leaves. On stem, only the C30 alcohol was diminished, whereas on silique and leaf, the major effect was seen on C26 and C28 alcohols. This indicated that in stem, there were sufficient alternative elongase KCS activities to maintain wild-type levels of C26 and C28 alcohols, but not of the C30 alcohol. Conversely, in silique and leaf elongase KCS activity was not sufficient to maintain wild-type levels of C26 and C28 alcohols. It should be emphasized, however, that there was substantial elongase activity for all fatty acid chain lengths in the kcs1–1 plants, since in all tissues the levels of some alcohols were only reduced but not eliminated, and the total wax load was only marginally reduced.
The sharing of elongase products between the reduction and decarbonylation pathways has been postulated ( von Wettstein-Knowles 1995), although dedicated elongase systems to each of these pathways has also been suggested ( Vioque & Kolattukudy 1997). To date, there has been little direct evidence confirming either suggestion. For the waxes of cauline leaves, upper stem and siliques, there was a significant decrease (28–46%) of the levels of the alcohols synthesized by the reduction pathway, as well as a decrease of the levels of the alkanes (7–12%) and ketones (10–21%) synthesized by the decarbonylation pathway. Thus, both the acyl reduction and decarbonylation pathways were affected by a mutation in a single elongase KCS gene. The simplest interpretation of this observation is that the KCS1 synthase produced VLCFA that were used by both pathways. In the lower stem and rosette leaves, which constitute older tissue than those listed above, the effect of the kcs1–1 mutation on the chain length of alkanes and ketones was not significant. The reason for this is not known, but one explanation is that, while the overall rate of their synthesis may have been slowed by the loss of KCS1 synthase, as evidenced by the decreased levels observed in younger tissue, with time, both alkanes and ketones accumulated to levels eventually equal that of the wild-type.
The KCS1 gene of Arabidopsis is similar to the maize GLOSSY8 which encodes a β-ketoacyl reductase involved in wax biosynthesis. Both genes encode components of fatty acid elongases and affect both the reductive and decarbonylation pathways. In addition, Glossy 8 had a broad expression pattern that included roots and seed ( Xu et al. 1997 ). Unlike kcs1–1 which has normal levels of total surface waxes, the gl8 has a 70% reduction in total epicuticular waxes, reflected primarily in losses of aldehydes and alcohols, but alkanes are also decreased ( Bianchi et al. 1979 ). Thus, the changes resulting from the kcs1–1 mutation primarily affect relative levels of individual components, whereas gl8 results in large changes in classes of components.
It is curious that when aldehydes were a significant component of the wax, as in stem and silique, the effect of kcs1–1 on the aldehyde levels closely paralleled the effect on alcohols of the same chain length in these tissues. Recent evidence from pea leaves suggests that in alcohol synthesis, there is a specific reductase that catalyzes both the reduction of the fatty acid and the aldehyde, and that the intermediate aldehydes are not released from this reductase ( Vioque & Kolattukudy 1997). This would further suggest that aldehydes which accumulate in wax are used by the decarbonylation pathway for alkane synthesis. Why then did the changes in the aldehyde levels not more closely parallel the levels of the alkanes and ketones? One possible explanation is that A. thaliana does not have a fatty acid reductase for alcohol synthesis that is analogous to the pea reductase. Alternatively, if the decarbonylation step is limiting, it is likely that the most apparent effect on this pathway will be on the aldehyde levels rather than on the end-products of the pathway.
Analysis of root tissue in the kcs1–1 plants suggested that KCS1 was not limited to only cuticular wax biosynthesis. Northern analysis indicated that KCS1 expression occurs in many tissues, including strong expression in root where wax levels were too low to detect. Fatty acid analysis of kcs1–1 roots revealed that the level of dioic acids was doubled and that the C22:0 fatty acid level was reduced by 50%. Since dioic acids are involved with suberin biosynthesis and VLCFA are involved in suberin wax biosynthesis, these changes suggest that KCS1 may play some role in the biosynthetic pathway for suberin and/or suberin wax. Although KCS1 may play a role in both wax and suberin biosynthesis, it does not appear to be involved in other pathways known to utilize VLCFA. Based on our analysis of glucocerebrosides, it did not appear that KCS1 was involved in that pathway. As expected, there was no alteration in acylglycerol composition, since the only acylglycerol in Arabidopsis containing VLCFA are the seed triacylglycerols. Those VLCFAs are synthesized by the FAE1 KCS ( James et al. 1995 ;Kunst et al. 1992 ).
The broad expression pattern of KCS1 and its effect on both cuticular wax and dioic acid and biosynthesis of VLCFA in roots suggest that this elongase KCS participates in multiple pathways. It is possible that the narrow stem phenotype we observed was a result of kcs1–1 perturbing yet another pathway. The large number of putative elongase KCS genes that have been revealed as a result of the Arabidopsis genome project invites speculation that there are multiple elongases dedicated to particular pathways. Another interpretation, however, resulting from our study is that there is a large degree of redundancy in elongase KCSs such that complete loss of a single FAE KCS does not severely compromise the plant. Thus, the seed-specific expression of FAE1, which is the only elongase KCS gene involved with synthesis of VLCFA for triacylglycerols and apparently has no other metabolic function ( James et al. 1995 ;Kunst et al. 1992 ), may be an exception to multi-purpose elongase KCS genes.
In spite of the redundancy of elongase KCSs that was revealed by the analyses of kcs1–1, individual genes do not appear to be expendable. The kcs1–1 seedlings were more susceptible to low RH stress compared to wild-type plants, especially within the first few days after germination. In addition to higher mortality rates when the kcs1–1 seedlings were exposed to a rapid decrease in RH, they also displayed significantly slower growth and development when transplanted. This phenotype was consistent with our observation that KCS1 expression and alterations in wax composition were most pronounced in young tissue. Nonetheless, it was surprising that these relatively small changes in the total wax composition were capable of altering the plant’s ability to adjust to changes in RH. It is also possible that some of the observed phenotypes were due to an undetected mutation resulting from the DNA tagging.
In summary, we have identified and characterized KCS1, the first elongase KCS gene that does not exhibit seed-specific expression. The availability of a plant with a tagged KCS1 allowed us to study the function of this gene and the role it plays in the metabolism of VLCFAs. The null allele, kcs1–1, has led to new insights in wax biosynthesis that were not previously available from mutant screens or biochemical studies, namely that this elongase KCS is one of several redundant elongases whose VLCFA products are shared by more than one wax biosynthetic pathway.
cDNA and genomic clone isolation
Clone ATTS1282 was identified and retrieved from the Arabidopsis expressed sequence tag (EST) database (Arabidopsis Genome Stock Center, Ohio State University, Columbus, OH, USA) after a BLASTN search (National Institute of Health, Bethesda, MD, USA), using the nucleotide sequence coding for the ORF of the jojoba FAE gene. A full-length genomic clone of KCS1 was obtained using ATTS1282 as a probe to screen an A. thaliana ecotype Columbia, genomic DNA library in λGEM11 (Promega, Madison, WI, USA, library kindly provided by Ron Davis, Stanford University, Stanford, CA, USA). The KCS1 genomic clone was digested with BamH1 and subcloned into pBluescript (Stratagene, La Jolla, CA, USA) following standard protocols ( Sambrook et al. 1989 ). All clones were sequenced in both directions with an ABI automated sequencer (Applied Biosystems, Inc., Foster City, CA, USA). DNA sequence data were analyzed using the Lasergene computer program (DNASTAR Inc., Madison, WI, USA).
Isolation of a T-DNA tagged mutant
A mutant having a T-DNA insertion in the KCS1 gene was identified by screening pooled genomic DNA prepared from a T-DNA tagged A. thaliana ecotype Wassilewskija, (WS) population (kindly provided by Dr Sue Gibson, Rice University, Houston, TX, USA), following the method of Krysan et al. (1996) . One hundred and thirty-eight DNA preparations of 100 plants each were combined to produce 10 submaster mixes representing about 1400 plants each, and 1 master mix of 13 800 plants. Initially, the master mix was screened for the presence of a tagged KCS1 gene using nested primers ( Fig. 2) and two amplification reactions. The first amplification used primer sets RB1/S1 or A1, and LB1/S1 or A1, and the master mix as the template. The amplified product was used as the template for a second amplification using the appropriate set of nested primers (RB2 or LB2 and S2 or A2). Once the presence of a T-DNA tagged KCS1 allele in the library was established, DNA from first the 10 submaster, and then the 14–100-plant pools were similarly analyzed. After identifying the pool of 100 plants containing the tagged allele, kcs1–1, seeds from this pool were obtained from DuPont. These T2 seeds were germinated in a grid pattern (30 rows × 18 columns) and leaves from each T4 plant were harvested and pooled by rows and columns. Genomic DNA was prepared from each of these 48 groups and analyzed for kcs1–1 as described above. This approach allowed the discovery of one T4 plant containing a T-DNA insertion in the KCS1 gene.
Southern and Northern blot analyses
Ten μg aliquots of total A. thaliana ecotype Columbia, genomic DNA were digested with the indicated restriction enzyme and fractionated by agarose gel electrophoresis. Total RNAs were isolated from various tissues of A. thaliana ecotype Columbia or WS, using the RNeasy plant total RNA extraction kit (QIAGEN, Valencia, CA, USA) or the method of Chomczynski & Sacchi (1987). For hybridization, total RNA (7–10 μg) was fractionated in a 1% agarose gel. [α-32P]dCTP labeling of the KCS1 DNA fragments, gel electrophoresis, nucleic acid blotting and hybridizations were performed following standard techniques ( Sambrook et al. 1989 ). All blots were analyzed by phosphorimaging and quantified by ImageQuant (Molecular Dynamics, Inc., Sunnyvale, CA, USA).
GC/MS analysis of epicuticular waxes, glucocerebrosides and leaf, seed and root fatty acid composition
Epicuticular waxes were extracted by shaking 30–60 mg of plant tissue in chloroform for 1 min, and triacontane was added as an internal standard on a per mg fresh weight basis. The chloroform was then evaporated under gaseous N2, the wax extracts dissolved in chloroform and analyzed directly, or following derivatization with N,O-Bis(trimethylsilyl)trifluoroacetamide (BSTFA) (Alltech, Deerfield, IL, USA) for 1 h at 70°C. BSTFA was removed by a gentle stream of gaseous N2 and the wax extracts were dissolved in chloroform. All wax samples were analyzed on a Hewlett-Packard 5890 series II gas chromatograph equipped with a 5971 mass detector and 7673 auto injector. The components were separated on an HP-5 (Hewlett-Packard) capillary column (30 m × 0.25 mm, 0.25 μm film thickness). The column was operated with helium carrier gas and splitless injection (injection temperature 250°C, detector temperature 280°C). The oven temperature was increased from 150°C to 300°C at 10°C min–1, and held for an additional 10 min.
The total fatty acid fraction from roots of 10 d seedlings grown on filter paper saturated with 0.5 × Murashige and Skoog basal salt mixture supplemented with 1 × Murashige and Skoog vitamin mix (Sigma, St. Louis, MO, USA) was obtained by saponifying the root tissue in 1 Vol. H2O and 2 Vol. 4N KOH in 100% methanol for 60 min at 80°C, and then extracting the fatty acids into petroleum ether. Methyl esters were prepared and analyzed as described previously ( Hlousek-Radojcic et al. 1995 ). For all statistical analyses, standard error was calculated with n-1. Methyl esters of the total fatty acid fraction from leaves or seeds were prepared by incubating the tissue in 2% H2SO4 in 100% methanol for 2 h at 80°C ( Browse et al. 1986 ). FAMES were extracted into petroleum ether and analyzed as described previously ( Hlousek-Radojcic et al. 1995 ).
Glucocerebrosides from leaf tissue were isolated and analyzed following the method of Cahoon & Lynch (1991). All fatty acid derivatives were analyzed by GC–MS as above and were separated on a SPB-1 (Supelco) capillary column (15 m × 0.2 mm, 0.2 μm film thickness). The column was operated with helium carrier gas at a constant flow rate of 0.75 ml min–1, and splitless injection (injection temperature 300°C, detector temperature 280°C). The oven temperature was increased from 150°C (1.5 min hold) to 250°C at 10°C min–1, then to 295°C at 3°C min–1 and then decreased at 10°C min–1–150°C (1 min hold).
Construction of yeast expression vectors
Screening of the pλYES yeast expression library with the partial cDNA sequence from ATTS1282 (( Elledge et al. 1991 ), kindly provided by Ron Davis, Stanford University, Stanford, CA, USA, yielded a cDNA clone which, upon comparison with the KCS1 genomic sequence, was found to be full length. The KCS1-pλYES construct and the yeast expression vector pYEUra3 (Clontech, Palo Alto, CA, USA), lacking insert, were used to separately transform the S. cerevisiae strain AB1380 using lithium acetate ( Gietz & Woods 1994).
Preparation of yeast microsomal membranes
Transformed yeast cells were cultured overnight in YPD at 30°C with vigorous shaking. One hundred μl of the overnight culture were used to inoculate 40 ml of complete minimal dropout medium lacking uracil (CM Ura) ( Ausubel et al. 1992 ), supplemented with either glc or gal (2% w/v). These cultures were grown at 30°C to an OD600 of approximately 1.3–1.5. Microsomes were then prepared as described previously ( Tillman & Bell 1986) using ice-cold isolation buffer (IB; 80 m m Hepes-KOH, pH 7.2, 5 m m EGTA, 5 m m EDTA, 10 m m KCl, 320 m m sucrose and 2 m m DTT). The protein concentration of the resuspended microsomes was determined ( Bradford 1976) and adjusted to 2.5 μg μl–1 (in IB with 15% glycerol). Aliquots were then frozen on dry ice, stored at –80°C until used and, once thawed, were not refrozen.
Fatty acid elongase assays
Fatty acid elongase activity was measured essentially as described previously ( Hlousek-Radojcic et al. 1995 ). The standard elongation reaction mix consisted of yeast microsomes (5 μg protein), 80 m m Hepes-KOH, pH 7.2, 20 m m MgCl2, 500 μm NADPH, 1 m m ATP, 15 μm acyl-CoA, 100 μm malonyl-CoA, and 10 μm CoA-SH in a final reaction volume of 25 μl. The radiolabel [1–14C] 18:0-CoA was synthesized using [1–14C] 18:0, 55 μCi μmol–1 (ICN, Costa Mesa, CA, USA) as described by Taylor et al. (1990) . Methyl esters of the radiolabeled acyl-CoA elongation products were prepared as described ( Hlousek-Radojcic et al. 1995 ), separated on reversed phase silica gel KC18 TLC plates (Whatman, 250 μm thick) ( Evenson & Post-Beittenmiller 1995), analyzed by phosphorimaging, and the intensity of the band for each distinct FAME was quantified by ImageQuant (Molecular Dynamics, Inc., Sunnyvale, CA, USA).
We thank Angela Scott and Ann Harris of the Molecular Analysis and Sequencing Section at the Samuel Roberts Noble Foundation for DNA sequencing and oligonucleotide synthesis. We also thank Colby Starker, Plant Biology Department of the Carnegie Institute of Washington for mapping KCS1. J.T. was supported by a Collaborative Research Award from the Noble Foundation.