Lipid Utilization, Gluconeogenesis, and Seedling Growth in Arabidopsis Mutants Lacking the Glyoxylate Cycle Enzyme Malate Synthase*

The aim of this research was to test the role of the glyoxylate cycle enzyme malate synthase (MLS) in lipid utilization, gluconeogenesis, and seedling growth in Arabidopsis. We hypothesized that in the absence of MLS, succinate produced by isocitrate lyase (ICL) could still feed into the tricarboxylic acid cycle, whereas glyoxylate could be converted to sugars using enzymes of the photorespiratory pathway. To test this hypothesis we isolated knock-out mls mutants and studied their growth and metabolism in comparison to wild type and icl mutant seedlings. In contrast to icl seedlings, which grow slowly and are unable to convert lipid into sugars (Eastmond, P. J., Germain, V., Lange, P. R., Bryce, J. H., Smith, S. M. & Graham, I. A. (2000) Proc. Natl. Acad. Sci. U. S. A. 97, 5669–5674), mls seedlings grow faster, use their lipid more rapidly, and are better able to establish as plantlets. Transcriptome and metabolome analyses show that icl seedlings exhibit many features characteristic of carbohydrate starvation, whereas mls seedlings differ relatively little from wild type. In the light mls seedlings generate more sugars than icl seedlings, and when fed with [14C]acetate, 14C-labeling of sugars is three times greater than in icl seedlings and more than half that in wild type seedlings. The mls seedlings also accumulate more glycine and serine than icl or wild type seedlings, consistent with a diversion of glyoxylate into these intermediates of the photorespiratory pathway. We conclude that, in contrast to bacteria and fungi in which MLS is essential for gluconeogenesis from acetate or fatty acids, MLS is partially dispensable for lipid utilization and gluconeogenesis in Arabidopsis seedlings.

The glyoxylate cycle catalyzes the conversion of two acetate molecules into succinate, providing the means for microorganisms to grow on ethanol, acetate, or fatty acids (1). This function has been confirmed through the analysis of mutants lack-ing isocitrate lyase (ICL 1 ; EC 4.1.3.1) and malate synthase (MLS; EC 4.1. 3.2), key enzymes of the glyoxylate cycle. Bacterial and fungal mutants do not grow on acetate, ethanol, or fatty acids (2)(3)(4)(5). Loss of ICL or MLS also leads to avirulence in bacterial and fungal pathogens of plants and mammals (2)(3)(4)(5). For example Mycobacterium tuberculosis lacking ICL cannot grow on fatty acids or acetate and is unable to persist in macrophages, where lipid is the primary carbon source available (2). Similarly Candida albicans and Saccharomyces cerevisiae lacking ICL cannot grow on acetate, and the Candida mutant is less infectious in mice where macrophage lipid is also the primary carbon source (3). ICL and MLS have therefore been identified as targets for therapeutic drugs to treat some bacterial and fungal infections since the glyoxylate cycle is absent from vertebrates.
In germinating oilseeds the glyoxylate cycle also enables acetate from lipid breakdown to be converted to four-carbon gluconeogenic substrates to support seedling growth (6). In Arabidopsis, seedlings of icl mutants grow poorly because they are unable to convert acetate from fatty acid ␤-oxidation into sugars (7). Instead, the seedlings slowly respire their fatty acids by transferring either acetate or citrate from the peroxisome to the mitochondrion (8,9). Without ICL, the tricarboxylic acid cycle cannot be supplied with succinate or with malate produced from glyoxylate by MLS. Thus, oxaloacetate cannot be generated for gluconeogenesis. We considered the possibility that if instead of blocking the glyoxylate cycle at ICL, it is blocked downstream at MLS, the succinate would still feed into the tricarboxylic acid cycle to regenerate oxaloacetate, and the glyoxylate could potentially feed into the photorespiratory pathway for conversion to sugar (Fig. 1). Although seed germination (radicle emergence) is fueled by a limited reserve of sugars in Arabidopsis, the subsequent growth of the seedling is fueled largely by oil breakdown, which occurs concurrently with the acquisition of photosynthetic competence. It has been clearly established by three independent studies that glyoxylate cycle and photorespiratory enzymes co-exist in the same peroxisomes during seedling growth in oilseed species (10 -12). Thus, in mls mutants, glyoxylate produced by ICL could serve as a substrate for serine-glyoxylate aminotransferase, leading to glycine and subsequently to serine, hydroxypyruvate, glycerate, and ultimately sugars (Fig. 1). We, therefore, hypothe-sized that mls mutant seedlings would be capable of gluconeogenesis from lipid and would grow better than icl seedlings. To test this hypothesis we isolated two independent mls mutants and studied their growth and metabolism relative to icl mutants. Consistent with our hypothesis, mls mutant seedlings grow much better, break down their lipid more rapidly, and accumulate more sugars than icl mutant seedlings. In addition they are capable of gluconeogenesis from acetate, unlike icl seedlings (7).

EXPERIMENTAL PROCEDURES
Plant Material and Growth Conditions-Seeds of Arabidopsis were surface-sterilized, stratified, and germinated as described by Sherson et al. (13). Except where stated, seedlings were grown for 48 h, corresponding to principal growth stage 0.6 as defined by Boyes et al. (14) in continuous light (100 mol of photons⅐m Ϫ2 ⅐s Ϫ1 ) or dark in the presence or absence of 1% (w/v) sucrose.
Screening of T-DNA Insertion Lines and Isolation of mls Knock-out Mutants-PCR screening was used to identify T-DNA insertions the Arabidopsis MLS gene (At5g03860). The gene-specific primers MS50 (5Ј-ATG GAG CTC GAG ACC TCA GTT TAT C-3Ј), MS51 (5Ј-GCT GCT TTC GAG GAC GCT TTG TCT C-3Ј), MS30 (5Ј-GAG CCT TGA GAC ATT GAT AGG GTA G-3Ј), and MS31 (5Ј-ACA AGT ACG GAT GAG AAG ATC AGA G-3Ј) were used in combination with T-DNA left border primers (13,15,16). This screening identified an insertion in a plant in the Wassilewskija ecotype from the Versailles collection (15,16). DNA sequence analysis revealed two copies of the T-DNA in the first intron of the MLS gene (mls-1; Fig. 2A). A second mutant was identified in the SALK SIGnAL T-DNA collection, which was generated in the Col-0 ecotype (17), containing a T-DNA insertion in the third intron of the MLS gene (mls-2, SALK_060987; Fig. 2A). In both knock-out lines kanamycin resistance conferred by the T-DNA co-segregated with the interrupted MLS gene (data not shown). A wild type segregant was isolated concurrently with each mutant and named MLS-1 and MLS-2 to indicate the mutant to which each corresponds.
RT-PCR Analysis-RNA was isolated from 2-day seedlings using the Qiagen RNAeasy kit and used to generate cDNA with the Qiagen Omniscript RT-PCR kit according to manufacturers' instructions. PCR was carried out with gene-specific primers MS51 and MS30 for MLS and ICL51 and ICL30 for ICL (7). Gene-specific primers for the ACT2 gene (At3g18780) were used to normalize the amount of template in each RT-PCR reaction.
Transcriptome Analysis-Microarray analysis using Affymetrix ATH1 genome arrays was carried out by the Nottingham Arabidopsis Stock Centre (nasc.nott.ac.uk) under the auspices of the Genomic Arabidopsis Resource Network (www.york.ac.uk/res/garnet/garnet.htm). The procedures are Minimum Information about a Microarray Experiment compliant, and data are available on the NASC website.
Biochemical Analysis-For enzyme assays, tissue extracts were prepared from Arabidopsis seedlings as described in Eastmond et al. (7). MLS and ICL assays were carried out as described in Cooper and Beevers (18). A coupled assay for serine-glyoxylate aminotransferase and hydroxypyruvate reductase activity was carried out using a modified version of the method described by Nakamura and Tolbert (19), in which 1 mM glyoxylate was used to start the reaction. Protein content was determined as reported by Bradford (20) using bovine serum albumin as the standard. Total fatty acids were extracted and quantified by gas chromatography-mass spectroscopy using the method described by Browse et al. (21), with C20:1 as a marker for triacylglycerol (TAG). 1 H NMR was used to analyze major polar metabolites in 2-day-old seedlings as reported by Ward et al. (22) under the auspices of the Genomic Arabidopsis Resource Network (as above). The levels of glucose (Glc), fructose (Fru), and sucrose (Suc) were quantified in ethanol-soluble extracts as described in Nielsen et al. (23). 14 C-Labeling Experiments and Paper Chromatography-The metabolism of sodium [2-14 C]acetate, [U-14 C]glycine, and L-[U-14 C]serine by triplicate batches of 100 2-day-old Arabidopsis seedlings grown on 1% (w/v) Suc in continuous light was analyzed according to the method described previously by Eastmond et al. (7), with the following modifications. Seedlings were incubated in the dark (to prevent re-fixation of 14 CO 2 ) for 4 h, during which time released 14 CO 2 was trapped in a well containing 200 l of 5 N KOH. The soluble components of the seedlings were then extracted in three 1-ml aliquots of 80% (v/v) ethanol at 80°C followed by 1 ml of water at 40°C. The total soluble extracts were combined, and the hydrophobic components were extracted with 1.5 ml of chloroform. The ethanol-soluble components were dried, resuspended in water, and then further separated into neutral, basic, and acidic fractions by ion exchange chromatography. The amount of 14 C present in each fraction was quantified by scintillation counting. The presence of mainly [ 14 C]Suc, [ 14 C]Glc, and [ 14 C]Fru in the neutral fraction was confirmed by paper chromatography of aliquots of the neutral fraction and subsequent scintillation counting of the spots corresponding to sugar markers on the chromatogram. Between 80 and 90% of the radioactivity in the neutral fraction was found on the chromatogram in the regions corresponding to these sugar makers (data not shown).

Isolation of mls Knock-out Mutants-ICL and MLS are both encoded by single genes in Arabidopsis.
To test the hypothesis that Arabidopsis mls mutant seedlings are capable of lipid utilization and gluconeogenesis from acetate, two independent T-DNA insertion knock-out mutants of the MLS gene (At5g03860) were identified (see "Experimental Procedures"). Homozygous individuals lacking MLS (mls-1 and mls-2; Fig.  2A) and wild type segregants (designated MLS-1and MLS-2) were isolated and analyzed by RT-PCR. This confirmed that neither mls mutant contains a MLS transcript ( Fig. 2B and data not shown). RT-PCR using ICL gene-specific primers indicates that the mls mutants have no apparent change in the level of ICL transcripts. Furthermore, although icl-2 mutant seedlings have no ICL transcripts, they have normal levels of MLS transcripts (Fig. 2B). MLS enzyme activity was not detectable in mls seedlings ( Fig. 2C and data not shown), whereas the wild type segregants showed a peak of enzyme activity 1.5 days post-imbibition (Fig. 2C). mls mutant seedlings contain ICL enzyme activity at the same level as in wild type seedlings ( Fig. 2C and data  MLS enzyme activity confirms that we have isolated two null mutants. The expression patterns for genes encoding a number of key enzymes of lipid metabolism were analyzed in mls and icl-2 mutant lines: 3-keto-acyl CoA thiolase (␤-oxidation), two peroxisomal isoforms each of citrate synthase and malate dehydrogenase (glyoxylate cycle), and two isoforms of phosphoenolpyruvate carboxykinase (gluconeogenesis) showed no apparent changes in the patterns of the transcripts for these genes in the mutant lines (data not shown).
The mls Mutants Have a Less Severe Phenotype than the icl Mutant-The mls seedling phenotypes were examined alongside the icl-2 mutant line isolated previously together with wild type revertant ICL-2 (7). The icl-2 mutant has a stunted phenotype during post-germinative growth, which can be rescued by the addition of exogenous Suc or by growth in high light conditions ( Fig. 3A; Ref. 7). However, mls-1 and mls-2 mutants have only a slightly stunted phenotype when grown in the absence of exogenous sugar in the light (Fig. 3A). The mls phenotype is more obvious in the dark where hypocotyl elongation is inhibited, and little root development takes place (Fig.  3A). mls mutant seedlings can be rescued by the addition of exogenous Suc in all conditions (Fig. 3A). Furthermore, lower concentrations of Suc are able to restore the mls-2 mutant seedlings to wild type growth in the light than are required to rescue icl-2 seedlings (data not shown). A further phenotype observed in icl-2 mutants was the failure of seedlings to establish into plantlets with true leaves under conditions of limited light ( Fig. 3B; Ref. 7). In contrast, the establishment frequency of mls-2 seedlings is similar to wild type levels and only becomes compromised when seedlings are grown in short days (Fig. 3B).
Introduction of the MLS cDNA driven by the cauliflower mosaic virus 35 S promoter into the mls-1 knock-out resulted in complementation of the observed phenotypes (data not shown). Subsequent experiments were performed only with mls-2 and MLS-2 lines as they are in the same Col-0 ecotype as icl-2.
Transcriptome Analysis Reveals Major Differences in the Two Glyoxylate Cycle Mutants-To investigate differences in gene expression in the two glyoxylate cycle mutants, RNA was isolated from triplicate batches of 2-day-old, light-grown mutant and wild type seedlings. This time point represents the onset of the period of most rapid lipid breakdown (Ref. 7 and see below), so that any differences in gene expression may be most marked. RNAs were each hybridized to Affymetrix ATH1 genome chips. The average hybridization signals detected in each mutant line were compared with the signal strengths in the corresponding wild types (Fig. 4). The data reveal that there are many more differences in gene expression in icl-2 (Fig. 4A)  ( Fig. 4B) relative to their wild type controls. In icl-2, 161 genes showed a more than 2-fold increase in expression, and 236 genes showed a more than 2-fold decrease relative to wild type (Table I indicates the 20 genes with the most altered gene expression in each class). In contrast, in mls-2 only 10 genes showed increased expression, and 12 genes decreased relative to wild type (Table II).
Some of the most highly induced genes in icl-2 seedlings reflect stress responses in this mutant (e.g. pathogenesis-related protein, chitinase, and glycine-rich protein), whereas others reflect carbohydrate limitation (e.g. ␤-glucosidase, asparagine synthetase, and enzymes of branched chain amino acid catabolism). Other induced genes include those encoding enzymes of protein degradation (e.g. proteases, aminotransferases, amino acid transporters) and numerous glycohydrolases and senescence-associated proteins.
The transcript data from mls-2 indicates that these seedlings are not carbohydrate-limited. Gene transcripts expressed at low levels in the mutant include two jasmonateinducible genes but otherwise do not fall into any obvious functional groupings, and most of them are not underexpressed in icl-2 (Table II). However, the 10 genes overexpressed in mls-2 relative to wild type include several that are also overexpressed in icl-2 (Table II). Among these is that encoding Gln-dependent asparagine synthetase, which is the only induced transcript coding for an enzyme known to be associated with carbohydrate limitation (24).
Metabolome Analysis Shows That mls Mutant Seedlings Are Not Carbohydrate-limited and Accumulate Glycine and Serine-To establish the effects of the mutations on levels of sugar, organic acids, and amino acids in mutant seedlings, the metabolomes of icl-2 and mls-2 seedlings were investigated and compared with wild types using non-biased 1 H NMR (22). Principal component analysis of the data was carried out, and a six-component model explained 90% of the variance. The first two principal components reveal that icl-2 forms a distinct cluster separate from wild types and mls-2 (Fig. 5A). This clear separation is due mainly to decreased levels of glutamine (Gln) and glucose (Glc) in icl-2 compared with all other lines (Fig. 5,  B and C). There were also small increases in valine, isoleucine, and lysine in icl-2. In contrast, the glucose levels in mls-2 did not differ appreciably from MLS-2, although the Gln level in mls-2 was also appreciably reduced (Fig. 5, B and C). A further significant finding in this analysis was that mls-2 seedlings have appreciably increased levels of serine (Ser) and glycine (Gly) relative to all other lines (Fig. 5C). There were no changes in organic acids detectable using this method in any genotypes (data not shown).
The Rate of Lipid Breakdown and Sugar Accumulation in mls Seedlings Is Similar to Wild-type Seedlings, whereas icl Seedlings Are Compromised in Both-During the heterotrophic stages of seedling development, TAG stored in the embryo is converted to sugar to fuel seedling growth (25). Accordingly, in wild type seedlings TAG levels decline rapidly during the first few days post-imbibition, reaching undetectable levels after 5 days (Fig. 6A). The absence of MLS in mls-2 mutant seedlings has relatively little effect on the rate of TAG utilization, which after an initial lag period declines at a similar rate to wild type and is undetectable by day 7 (Fig. 6Ai). In contrast, the initial TAG content of icl-2 seeds is lower than that of other lines, and its rate of utilization in seedlings is much slower, such that TAG is still present 7 days post-imbibition ( Fig. 6Aii; Ref. 7).
Measurement of total soluble sugar levels shows that the sugar content of mls-2 seedlings is appreciably greater than that of icl-2 seedlings (Fig. 6, B and C). Levels of Suc decline to very low amounts in all genotypes within the first 2 days post-imbibition, as Suc is used to fuel this stage of growth (Fig.  6B). Thereafter, the Suc content remains very low in the dark (Fig. 6B, iii and iv) but increases slowly in the light in all genotypes, presumably as a result of photosynthetic activity (Fig. 6B, i and ii). However, after 8 days in the light icl-2 seedlings contain half as much Suc as mls-2 seedling and wild types (Fig. 6Bii). The hexose levels detected reveal that in the light, mls-2 seedlings are able to generate more Glc and Fru than icl-2 seedlings. Hexose content increases during the first 2 days in wild type and mls-2 seedlings in the light but not in icl-2 seedlings (Fig. 6C, i and ii). After day 2, the hexose content increases progressively in all genotypes in the light, although more slowly in icl-2. In the dark (Fig. 6C, iii and iv), the hexose contents of wild type seedlings increase rapidly up to day 2 and later decline. During this period, the hexose contents of both mutants are lower than their respective wild types. Together with the observation that TAG is utilized more rapidly in mls-2 than in icl-2 mutants (Fig. 6A), these observations support our original hypothesis that in the light mls-2 seedlings are capable of gluconeogenesis from acetate after lipid ␤-oxidation, whereas icl-2 seedlings are not (7). mls Seedlings Carry Out Gluconeogenesis from [ 14 C]Acetate, whereas icl Seedlings Do Not-To directly test the hypothesis that mls-2 seedlings are capable of gluconeogenesis from acetate, 2-day-old light-grown seedlings were fed with [2-14 C]acetate and fractionated, and the amount of 14 C was incorporated into each fraction determined (Fig. 7). mls-2 and wild type seedlings incorporated the same total amount of 14 C during the experiment (123 Ϯ 6 and 122 Ϯ 4 ϫ 10 4 dpm, respectively); the pattern of incorporation in the two genotypes was similar in all fractions except that mls-2 seedlings incorporate 60% of the amount of 14 C into sugars and release twice as much 14 CO 2 compared with wild type seedlings. In contrast, icl-2 seedlings incorporated less total 14 C during the experiment (75 Ϯ 5 ϫ 10 4 dpm), and significantly less 14 C was detected in all ethanolsoluble fractions than in wild type seedlings. Only 20% of 14 C is detected in the sugar fraction of icl-2 seedlings compared with wild type relative to total incorporation (5% in absolute terms). icl-2 seedlings release more than three times the percentage of 14 CO 2 compared with wild type seedlings (Fig. 7). These data confirm that mls-2 seedlings are capable of gluconeogenesis from acetate, whereas icl-2 seedlings are not.

C]Glycine and L-[ 14 C]Serine Can Act as Gluconeogenic Precursors in 2-Day-old
Seedlings-To investigate whether gluconeogenesis is possible from Gly and Ser, consistent with our hypothesis that glyoxylate may be incorporated into sugars via these metabolites, 14 C-labeled Gly and Ser were fed to 2-day-old seedlings. Because the pools of unlabeled Gly and Ser in these seedlings vary, comparison of the rates of transfer of 14 C to sugar is not informative. Nevertheless, 14 C was detected in sugars of seedlings of all genotypes after feeding, confirming that gluconeogenesis occurs from these amino acids (data not shown). DISCUSSION Consistent with our hypothesis, mls seedlings grow faster, use their TAG more rapidly, and are better able to become established as plantlets with true leaves than icl seedlings. The transcriptome data showed that gene expression in mls-2 seed-lings differed very little from that of wild type. In contrast, expression of 397 genes differed more than 2-fold in icl-2 seedlings compared with wild type. Genes overexpressed in icl-2 compared with wild type included many that are commonly up-regulated in carbohydrate-limited or senescent plant tissues. These include genes encoding enzymes of protein and amino acid catabolism and various glycohydrolases that might scavenge sugars from diverse carbohydrate sources. A gene encoding Gln-dependent Asn synthetase, which is highly responsive to carbohydrate limitation (24), was found to be induced in icl-2 and to a lesser extent in mls-2 seedlings.
NMR analysis of polar solutes showed that icl-2 seedlings have low levels of Glc, whereas mls-2 seedlings contain Glc at wild type levels (Fig. 5B). Both icl-2 and mls-2 seedlings have low Gln (Fig. 5C), which is consistent with both having elevated expression of a Gln-dependent asparagine synthetase gene, but Asn levels were not elevated. The elevated levels of valine, isoleucine, and lysine in icl-2 seedlings may result from in-TABLE I Differences in the transcriptomes of 2-day-old icl-2 mutant seedlings compared to ICL-2 seedlings Upper, the top 20 genes underexpressed more than 2-fold in icl-2 compared with ICL-2. Lower, the top 20 genes induced in icl-2 compared with ICL-2. The fold change in gene expression in mls-2 is also shown for comparison. NA, not applicable because the signals for the corresponding genes in MLS-2 and mls-2 are less than 100. creased protein catabolism, as suggested by transcriptome data. Changes in amino acids and increased expression of genes encoding enzymes of branched-chain amino acid catabolism in icl-2 are likely to be causally related. The greater rate of breakdown of TAG in mls relative to icl seedlings (Fig. 6A) and the accumulation of sugars in mls seedlings as they grow (Fig. 6, B and C) imply that mls seedlings are able to convert fatty acids from TAG into sugars. This was confirmed by the transfer of 14 C from acetate to sugars in mls-2 at a level of 60% that of wild type seedlings (Fig. 7). In contrast to icl-2 seedlings, the pattern of 14 C labeling in mls-2 seedlings differed from wild types only in the small increase in level of radioactivity released as CO 2 and the small decrease incorporated into sugars (Fig. 7). This is entirely consistent with gluconeogenesis from acetate in mls-2 seedlings, which is less efficient than in wild type, such that more of the [ 14 C]acetate is respired in mls-2 but is far more efficient than in icl-2 seedlings. The observation that seedlings lacking MLS are capable of gluconeogenesis from acetate is in marked contrast to the observation that ICL is required for such gluconeogenesis in Arabidopsis seedlings (7) and fulfils our original hypothesis. This observation also contrasts with the observation that both MLS and ICL are essential for gluconeogenesis and growth of microorganisms on ethanol, acetate, or fatty acids (2)(3)(4)(5).
Because icl mutant seedlings are essentially incapable of gluconeogenesis from lipid ( Fig. 7; Ref. 7), we deduce that ICL is required for gluconeogenesis in the mls mutants. The succinate produced by ICL will be oxidized in the mitochondria and oxaloacetate regenerated to maintain the cycle. ICL can only continue to produce succinate as long as the glyoxylate is metabolized. Furthermore, conversion of this glyoxylate to sugar is required for the net conversion of TAG to sugar to support plant growth; its respiration or degradation would achieve no more than is achieved by the respiration of TAG in the icl

TABLE II
Differences in the transcriptomes of mls-2 mutant seedlings compared to MLS-2 seedlings Upper, genes repressed greater than 2-fold in mls-2 compared with MLS-2. Lower, genes induced greater than 2-fold in mls-2 compared with MLS-2. The fold change in gene expression in icl-2 is also shown for comparison. NA, not applicable because the signals for the corresponding genes in MLS-2 and mls-2 are less than 100. Final confirmation of our proposed pathway of MLS-independent gluconeogenesis from acetate will depend upon detailed genetic and molecular characterization in the future. However, our hypothesis that glyoxylate will feed into the photorespiratory pathway is supported by elevated levels of Gly and Ser in mls seedlings and by the transfer of 14 C from Gly and Ser into sugars. Microarray data show that the genes encoding the enzymes of the photorespiratory pathway are expressed as highly in 2-day-old seedlings as in 7-day-old seedlings, and mature leaves (information obtained from nasc. nott.ac.uk). We have also determined that serine-glyoxylate aminotransferase activity coupled to hydroxypyruvate reductase activity in mls-2 seedlings was 20.8 Ϯ 2.1 and 39 Ϯ 4.3 mol⅐min Ϫ1 ⅐mg of protein Ϫ1 at days 2 and 5, respectively. This is similar to enzyme activity measured in MLS-2 seedlings of 17.7 Ϯ 1.2 and 32.4 Ϯ 1.6 mol⅐min Ϫ1 ⅐mg of protein Ϫ1 at days 2 and 5, respectively, and is sufficient to account for the observed flux of carbon from fatty acids to sugar.
We propose that at the transition from germinated seed to seedling, the mls mutant will be compromised by the absence of the glyoxylate cycle and inability to assimilate acetate into sugars. This is reflected in the low hexose content of such seedlings after 2 days of growth (Fig. 6C) and an accumulation of Ser and Gly (Fig. 5C). We further propose that around day 2, when the seedlings start to turn green, they acquire the ability to metabolize glyoxylate through the photorespiratory pathway, and the block caused by the lack of MLS is alleviated. At this point TAG breakdown starts to accelerate, and sugars begin to accumulate.
The observation that seeds of icl-2 accumulate less lipid than wild type while mls-2 seeds accumulate normal amounts ( Fig.  6A; Ref. 7) provides evidence that ICL has an important function at other stages of the life cycle for which MLS is not required. There are reports of ICL activity in the absence of MLS in green leaves (28,29,30), which may thus function in such a pathway.
The phenotypes of glyoxylate cycle mutants clearly indicate that there is a cost to Arabidopsis seedlings lacking ICL and MLS. However, although the absence of ICL and thus gluconeogenesis from lipid results in severely compromised seedling growth ( Fig. 3; Ref. 7), the absence of MLS results in a much less severe phenotype because the seedlings are able to employ an alternative gluconeogenic mechanism. This novel metabolic pathway from acetate to sugar, potentially employing photorespiratory enzymes in conjunction with glyoxylate cycle enzymes, in mls mutant seedlings is an example of the remarkable metabolic flexibility of Arabidopsis.
FIG. 6. Lipid and sugar levels in wild type and mutant seedlings during post-germinative growth. A, total fatty acids were extracted at daily intervals from seedlings grown in continuous light. छ, mls-2; ࡗ, MLS-2; icl-2; Ⅺ and f, ICL-2. The level of C20:1, a marker for TAG, was quantified in mls-2 and MLS-2 seedlings (i) and icl-2 and ICL-2 seedlings (ii). Data presented are the means Ϯ S.D. for three measurement each from two separate experiments. B, Sucrose levels in seedlings. Symbols are as above. Sucrose in (i) mls-2 and MLS-2 and (ii) icl-2 and ICL-2 seedlings, grown in continuous light. Sucrose in mls-2 and MLS-2 (iii) and icl-2 and ICL-2 seedlings (iv), grown in darkness. C, total hexoses (Glc plus Fru) in seedlings. Hexoses in mls-2 and MLS-2 (i) and icl-2 and ICL-2 (ii) seedlings, grown in continuous light. Hexoses in mls-2 and MLS-2 (iii) and icl-2 and ICL-2 (iv) seedlings, grown in darkness. Data presented are the means Ϯ S.D. for triplicate batches of seedlings. DPI, days post-imbibition.