Turnover of fatty acids during natural senescence of Arabidopsis, Brachypodium, and switchgrass and in Arabidopsis beta-oxidation mutants.

During leaf senescence, macromolecule breakdown occurs and nutrients are translocated to support growth of new vegetative tissues, seeds, or other storage organs. In this study, we determined the fatty acid levels and profiles in Arabidopsis (Arabidopsis thaliana), Brachypodium distachyon, and switchgrass (Panicum virgatum) leaves during natural senescence. In young leaves, fatty acids represent 4% to 5% of dry weight and approximately 10% of the chemical energy content of the leaf tissues. In all three species, fatty acid levels in leaves began to decline at the onset of leaf senescence and progressively decreased as senescence advanced, resulting in a greater than 80% decline in fatty acids on a dry weight basis. During senescence, Arabidopsis leaves lost 1.6% of fatty acids per day at a rate of 2.1 mug per leaf (0.6 mug mg(-1) dry weight). Triacylglycerol levels remained less than 1% of total lipids at all stages. In contrast to glycerolipids, aliphatic surface waxes of Arabidopsis leaves were much more stable, showing only minor reduction during senescence. We also examined three Arabidopsis mutants, acx1acx2, lacs6lacs7, and kat2, which are blocked in enzyme activities of beta-oxidation and are defective in lipid mobilization during seed germination. In each case, no major differences in the fatty acid contents of leaves were observed between these mutants and the wild type, indicating that several mutations in beta-oxidation that cause reduced breakdown of reserve oil in seeds do not substantially reduce the degradation of fatty acids during leaf senescence.

Leaf senescence is considered the last stage of leaf development that leads to leaf death (Lim et al., 2007). During senescence, the catabolism of macromolecules such as proteins, carbohydrates, lipids, and nucleic acids becomes predominant in leaves, while anabolic activities are strongly decreased, with some even lost. But leaf senescence is not just a passive degradation process, since senescence is also a way to remobilize nutrients from old and photosynthetically less efficient leaves to new vegetative tissues, developing seeds, or storage organs (Himelblau and Amasino, 2001;Lim et al., 2007). In addition to age, leaf senescence can also be induced by many biotic and abiotic factors, including darkness, extreme temperature, drought, and nutrient deficiency. Plant hormones also affect leaf senescence. For example, ethylene, abscisic acid, jasmonate, salicylic acid, and sugar promote senescence, whereas cytokinins, auxin, and gibberellins delay senescence (Gan and Amasino, 1997).
Membrane degradation is one of the early and major manifestations of leaf senescence (Thompson et al., 1998). This deterioration is generally characterized by the decline of membrane lipids and the enrichment of free fatty acids. It has been found that polar lipids, especially the chloroplast lipids monogalactosyl diglyceride, digalactosyl diglyceride, and phospholipid phosphatidyl glycerol, decrease during leaf senescence, and a preferential loss of monogalactosyl diglyceride occurs in senescing leaves (Hirayama and Oido, 1969;Ferguson and Simon, 1973;Fong and Heath, 1977;Koiwai et al., 1981;Harwood et al., 1982;Yamauchi et al., 1986;Wanner et al., 1991). The total fatty acids also decreased almost concomitantly with changes in polar lipids during leaf senescence. For example, the fatty acid content of tobacco (Nicotiana tabacum) leaves decreased more than 50%, changing from 37 mg g 21 dry weight at 10 weeks to 15 mg g 21 dry weight at 14 weeks (Koiwai et al., 1981). The total fatty acids in Roseus leaves dropped from 245 mg per leaf at day 37 to 135 mg per leaf at day 77 (Mishra et al., 2006). Drought-induced senescence of Arabidopsis (Arabidopsis thaliana) leaves resulted in fatty acids decreasing to 36% of the original 33 mg g 21 dry weight (Gigon et al., 2004).
Many lipid-degrading enzymes are involved in the degradation of membrane lipids. A transcriptome study of Arabidopsis leaves revealed that 2,491 senescence-associated genes, based on analysis of 6,200 senescence-associated ESTs, were up-regulated during senescence (Guo et al., 2004). Among this set are genes encoding 11 lipase/acyl hydrolases, six phospholipases, two lipoxygenases, and nine b-oxidation enzymes. Fatty acids released from lipid hydrolysis in senescing leaves can be further metabolized by the b-oxidation pathway, which occurs mainly in the peroxisomes of leaf tissue and germinating seeds (Gerhardt, 1992;Graham, 2008). Several Arabidopsis mutants defective in the b-oxidation pathway have been genetically and biochemically studied and indicate that b-oxidation plays an essential role in the turnover of reserve oil, seed germination, and seedling establishment. The b-oxidation pathway was also suggested to be active throughout the plant life cycle, and it has been proposed to play an important role in the catabolism of lipids during leaf senescence (Pistelli et al., 1991;Graham and Eastmond, 2002). However, relatively little attention has been paid to the influence of b-oxidation on lipid profiles in senescing leaves.
Most previous studies of lipid changes during leaf senescence have been short term or have used hormones or other treatments to induce senescence. In this study, we characterized fatty acid changes in leaves during natural senescence over the life cycle of Arabidopsis, Brachypodium, and switchgrass (Panicum virgatum). Young leaves have approximately 5% lipids by dry weight. Because a typical lipid has 2 times more energy per weight than carbohydrates or protein, lipids can represent 10% of the energy content of the leaves. If high-energy lipids could be retained in leaves while proteins, carbohydrates, and nucleic acids were mobilized, it can be estimated that the energy content of leaves could be increased approximately 20%. Therefore, an additional rationale behind this study was to test the possibility of preventing the breakdown of fatty acids during leaf senescence by mutations of b-oxidation, so that the energy content of plant leaves at harvest might be increased. To this end, we examined three Arabidopsis b-oxidation mutants, acx1acx2 (ecotype Columbia [Col-0] background; Adham et al., 2005), lacs6lacs7 (ecotype Wassilewskija [Ws] background; Fulda et al., 2004), and kat2 (Ws background; Germain et al., 2001), to determine their influence on profiles of fatty acid breakdown during leaf senescence. All of these mutants are blocked in fatty acid b-oxidation during seed germination, as demonstrated by the fact that they are defective in the breakdown of seed storage oil triacylglycerol (TAG), they can accumulate long-chain acyl CoAs, and they cannot metabolize the nontoxic compound indolebutyric acid into the toxic compound indoleacetic acid (Baker et al., 2006). These mutants also showed a Sucdependent seedling growth. The ACX genes encode acyl-CoA oxidase. Prior to b-oxidation, fatty acids are activated to acyl-CoA catalyzed by long-chain acyl-CoA synthases (LACS) to initiate b-oxidation . LACS6 and LACS7 are both located in peroxisomes. In Arabidopsis, three KAT genes were found on chromosomes 1 (KAT1), 2 (KAT2), and 5 (KAT5). It has been shown that the product of KAT2, or PED1 (for peroxisome defective), the primary thiolase of b-oxidation in Arabidopsis, is implicated in peroxisome development and is essential for efficient turnover of reserve oil in seedlings (Hayashi et al., 1998;Germain et al., 2001;Zimmermann et al., 2004). The ACX1 (At4g16760), ACX2 (At5g65110), LACS6 (At3g05970), LACS7 (At5g27600), and KAT2 (At2g33150) genes are all expressed in leaves, and their expression increases during senescence. Therefore, it is reasonable to implicate their activity in the turnover of fatty acids during senescence. A third objective was to determine whether disruption of b-oxidation delays general leaf senescence, since expression of KAT2 was found to be essential for the timely onset of leaf senescence in Arabidopsis (Castillo and Leon, 2008). Delaying onset or decreasing the progression of senescence could increase the biomass of plants and, therefore, possibly boost yields of crops harvested for aboveground biomass.

Phenotype of Arabidopsis Leaves at Different Ages
To our knowledge, there are no reports on fatty acid turnover in Arabidopsis leaves during natural senescence. In this investigation, we systemically studied the age-dependent changes in fatty acids in Arabidopsis rosette leaves over the course of senescence. For comparison with Arabidopsis, we also analyzed fatty acid changes in leaves of a model for cereal plants, Brachypodium, and one potential energy crop, switchgrass. Figure 1 shows the typical rosette leaves of Arabidopsis at varying developmental stages, in which the changing appearance of the leaves can be easily observed. Leaf yellowing is a typically visible symptom of leaf senescence, caused mainly by chlorophyll turn- over, and is considered a signal of the early disintegration of the photosynthetic machinery. Age-dependent changes in chlorophyll content of Arabidopsis leaves during growth and senescence are shown in Figure 2. Leaves from 25 to 51 d after planting retained similar high levels of chlorophyll, with an average value of 1.6 mg g 21 fresh weight. Thereafter, from day 57 onward, the amount of leaf chlorophyll began to decline, until at the end of senescence the leaves contained only 6% of the maximum chlorophyll content.
In the life cycle of the plants, the leaf weight also varied with time. During leaf development, the dry weight on a per leaf basis initially increased continually to a maximum value of 3.7 mg at 65 d. Subsequently, leaf weight decreased gradually to the end of senescence, at which time leaves abscised off the plant.

Fatty Acid Changes in Arabidopsis Leaves during Growth and Senescence
Arabidopsis leaves contain palmitic acid (16:0), hexadecatrienoic acid (16:3), linoleic acid (18:2), and linolenic acid (18:3) as the major fatty acids and hexadecenoic acid (16:1), hexadecadienoic acid (C16:2), stearic acid (18:0), and oleic acid (18:1) as minor components (Fig. 3). Although all fatty acid species were present in leaves throughout the life cycle, their relative amounts varied at different ages, as shown in Figure 3. As leaf senescence initiated, all of the fatty acid classes declined on a dry weight basis. The biggest change took place in the level of 18:3, which at the end of senescence had dropped by about 92%, while the other major fatty acids 16:0, 16:3, and 18:2 decreased by 56%, 84%, and 83%, respectively. The average rates of fatty acid degradation from 51 to 104 d ranged from 1.7% per day for 18:3 to 1.1% per day for 16:3 and 16:0.
The relative contents (mol %) of 16:0 and 18:3 are given in the inset of Figure 3, which indicates a decline in the percentage of 18:3 and a relative increase in 16:0 with increased senescence. At late senescence, 16:0 became the most abundant fatty acid in leaves, nearly reaching 40% of the total fatty acids. Accordingly, these changes resulted in a decrease in the ratio of unsaturated to saturated fatty acids of leaves during senescence (data not shown).
As shown in Figure 4, the total fatty acid content, expressed on leaf dry weight, had a maximum level of about 40 mg mg 21 dry weight in the 31-to 51-d period, the stage having the highest growth rate. After that phase, fatty acids decreased abruptly. At the end of senescence (104 d), fatty acid content was only 7 mg mg 21 dry weight. The contents of total fatty acids were reduced by 83%, and the average rate of fatty acid turnover was 0.6 mg mg 21 dry weight d 21 during senescence, corresponding to a loss of 1.6% of leaf fatty acids per day. The rate of fatty acid turnover in senescing leaf is much slower than the rates of fatty acid and TAG breakdown in geminating seeds (see inset of Fig. 4). During germination, between day 0 and day 5, seeds lost 14% of all fatty acids per day at a rate of 50 mg mg 21 dry weight and 11% of eicosenoic acid (20:1), which is considered representative of TAG in Arabidopsis seeds (Germain et al., 2001;Fulda et al., 2004), at a rate of 11 mg mg 21 dry weight of seed.
When expressed on a per leaf basis (Supplemental Fig. S1A), it appears that leaf experienced two phases of fatty acid changes. Up to day 51, fatty acid in leaves increased with growth time. The net rate of synthesis of fatty acids was 4.0 mg per leaf per day (1.5 3 10 22 mmol fatty acid per leaf per day), equal to a value of about 0.16 mmol carbon mg 21 chlorophyll h 21 . This value is severalfold lower than maximum rates obtained by short-term isotope labeling of very young Arabidopsis leaves (Bao et al., 2000;Bonaventure et al., 2004). After senescence began, the net rate of fatty acid breakdown was found to be 2.1 mg per leaf per day, or approximately half the rate of synthesis that occurred during leaf expansion. We also determined the fatty acid contents on a leaf area (cm 2 ) basis (Supplemental Fig. S1B). The results showed that during leaf expansion (days 33-53), there is a slight increase in the fatty acid per area of leaves until an inflection point at day 53, where the fatty acid level is highest with a value of 43 mg cm 22 leaf area. During this period of fatty acid gain, the mean rate of fatty acid synthesis based on leaf area is 0.11 mg fatty acid cm 22 per day. After this stage, the content of fatty acids decreased, leading to a loss of approximately 30% in fatty acids at the end of senescence. The mean rate of fatty acid breakdown is 0.33 mg fatty acid cm 22 per day.
In sharp contrast to the glycerolipids, the chloroform-soluble aliphatic surface lipids in Arabidopsis leaves remained comparatively constant during the plant life cycle (Supplemental Fig. S2). For example, the wax levels in Arabidopsis leaves at 26, 50, and 77d were found to be 1.7, 1.3, and 1.3 mg mg 21 dry weight,  respectively. Therefore, the soluble aliphatic surface lipids became an increasing proportion of total leaf lipids, increasing from about 4% at 26 d to 9% at 77 d.

Fatty Acid Changes in Brachypodium and Switchgrass Leaves during Senescence
Brachypodium is a temperate wild grass species. It has recently emerged as an attractive experimental model for the study of small-grain temperate cereals and related grasses (Opanowicz et al., 2008). In addition, it is phylogenetically close to and exhibits agricultural traits similar to economically important crops, like wheat (Triticum aestivum) and barley (Hordeum vulgare), and to several potential biofuel grasses (Draper et al., 2001). In this study, we investigated, to our knowledge for the first time, the fatty acid composition in seed and leaves of Brachypodium. Analysis of seed tissues showed that the fatty acid contents in embryo and endosperm were 64 mg mg 21 embryo and 11 mg mg 21 endosperm, respectively. Because the endosperm constituted 96% of seed weight, 82% of seed fatty acids accumulated in the endosperm. As shown in Supplemental Table S1, 16:0, 18:1, and 18:2 were the main components in these two parts of the seed, accounting for more than 80% of the total fatty acids. The contents of fatty acids in Brachypodium seeds (1.4% of seed by dry weight) are similar to those in wheat seeds, in which fatty acid levels range between 0.9% and 1.9% of dry weight (Choudhury and Rahman, 1973).
In Brachypodium leaves, 18:3 is the most abundant fatty acid, followed by 16:0 and 18:2 in this order (data not shown). At leaf maturity (days 58-105), 18:3 reached 70% of total fatty acids, higher than its counterpart in Arabidopsis leaves, in which the content of 18:3 is not more than 60%. Another notable difference between Brachypodium and Arabidopsis is that Brachy-podium has no detectable 16:3 in leaves, which reflects that the biosynthesis of plastidial lipids follows a eukaryotic pathway; consequently, Brachypodium, like other grasses, is a "C18:3" plant.
The maximal content of total fatty acids in Brachypodium leaves is around 40 mg mg 21 dry weight, very close to the counterpart level in Arabidopsis leaves. During the growth time of 37 to 96 d, the levels of total fatty acids in leaves generally increased (Fig. 5). After that, the fatty acids in leaves declined with time. At day 139, the fatty acid level dropped to only 20% of that at day 96. The degradation rate of fatty acids in Brachypodium leaves was about 0.70 mg mg 21 dry Figure 3. Age-dependent changes in the levels of major fatty acid species in Arabidopsis leaves during growth and senescence. The inset shows changes in relative contents (mol %) of C16:0 and C18:3. The data represent means of three different isolations 6 SD. Figure 4. Age-dependent changes in the contents of total fatty acids based on dry weight of Arabidopsis leaves. The inset shows changes in the contents of total fatty acids and eicosenoic acid (C20:1) during seed germination. The data represent means of three different isolations 6 SD.
weight per day on average, corresponding to 1.9%, similar to the values observed in Arabidopsis leaves (Fig. 5).
Like Brachypodium, switchgrass is also a C18:3 plant, since no C16:3 was found in leaves (data not shown). On a dry weight basis, the maximum amount of fatty acids in switchgrass leaves is around 55 mg mg 21 dry weight, higher than in Arabidopsis and Brachypodium (Fig. 5). Senescence also caused a reduction of fatty acids in switchgrass leaves. After day 60, the total fatty acids in the grass leaves began to drop. At the late senescence stage of 121 d, the leaves had lost 80% of fatty acids, to 10.6 mg mg 21 dry weight. The average degradation rate of fatty acids was about 0.69 mg mg 21 dry weight per day, close to those of Arabidopsis and Brachypodium.

TAG Levels in Plant Leaves
TAG has been reported to increase in senescent leaves, reaching 12% of lipid in Arabidopsis (Kaup et al., 2002) and more than 20% in crabapple (Malus species; Lin and Oliver, 2008). We have observed that estimations of low TAG content in leaves can sometimes be inaccurate due to other lipids that comigrate with TAG on thin-layer chromatography (TLC) plates. To reevaluate TAG content in senescent leaves, we have determined TAG contents in several plant leaves using three methods: TLC, gas chromatography (GC), and electrospray ionization-mass spectrometry (ESI-MS). Figure 6 shows the TLC of neutral lipid extracts from plant leaves of Arabidopsis (Col-0; 58 d) Brachypodium (58 d), switchgrass (58 d), and crabapple (2 and 20 weeks old). All of these leaves were freeze dried for 24 h and are of the same dry weight (10 mg). As can be seen from Figure 6, although lipids from equal dry weight of leaves were applied, the size and intensity of TAG bands on the TLC plate varied with plant species, suggesting varying concentrations of TAG in different plant leaves. Particularly in older leaves, based on iodine staining, bands with greater abundance than TAG comigrated with a wax ester standard. These were not further identified but may be sterol esters or phytol esters (Hugly et al., 1990;Dyas and Goad, 1993;Patterson et al., 1993;Gaude et al., 2007).
To provide definitive identification of TAG, neutral lipid extracts of leaves were analyzed by ESI-MS (Supplemental Fig. S3). We found that the TAG levels in dry leaves of 58-d Arabidopsis were 0.12 mg mg 21 dry weight, a little higher than Brachypodium leaves (0.08 mg mg 21 dry weight) but much lower than switchgrass leaves (0.76 mg mg 21 dry weight). TAG fatty acid equivalents in these plant leaves were only 0.35%, 0.19%, and 1.23% of the corresponding total fatty acids (determined by GC) for Arabidopsis, Brachypodium, and switchgrass, respectively. The TAG content determined by ESI-MS in dry leaves of 20week-old crabapple was about 1.3 mg mg 21 dry weight. This value is similar to that determined by GC, where the TAG content is 1.8 mg mg 21 dry weight. For 2-week-old crabapple, the TAG content determined by ESI-MS is 1.4 mg mg 21 dry weight in dry leaves. In addition, a low amount of TAG was observed to accumulate in 20-week-old Cirsium vulgare leaves, with only 0.12 mg mg 21 dry weight.
For Arabidopsis, TAG levels in leaves were very low at all stages, not more than 0.5 mg mg 21 dry weight (Supplemental Fig. S4). Even in senescent leaves, TAG contents remained less than 1% of total lipids (see inset of Supplemental Fig. S4).

Effect of Mutation of b-Oxidation on the Breakdown of Fatty Acid during Leaf Senescence
It is well established that b-oxidation is critical in the turnover of storage oil and that the supply of carbon and energy from b-oxidation of fatty acid in peroxisomes is essential for seed germination and early postgermination growth. However, although it has been suggested that b-oxidation plays a general housekeeping role in lipid catabolism throughout the plant life cycle (Graham and Eastmond, 2002), less is known about the regulation and function of b-oxidation in fatty acid degradation during plant senescence.
We used three b-oxidation mutants, acx1acx2, lacs6lacs7, and kat2, to test the influence of b-oxidation on the breakdown of fatty acids during leaf senescence. The activities of the peroxisomal enzymes longchain acyl-CoA synthetase in lacs6lacs7, acyl-CoA oxidase in acx1acx2, and ketoacyl thiolase in kat2 are defective, which results in an almost complete lack of storage TAG mobilization during seedling growth. Our results showed that blocking b-oxidation in the double mutants acx1acx2 and lacs6lacs7 and the single mutant kat2 had almost no effect on the breakdown of fatty acids during leaf senescence, as indicated in Figure 7. In leaves of various ages, the levels of fatty acids in these b-oxidation mutants were not substantially different from those in the wild type. For example, the lacs6lacs7 mutant in the Ws background showed a similar rate of fatty acid degradation to the wild type. In the 28-d period of leaf senescence, lacs6lacs7 was decreased by 45% in the amount of fatty acids, comparable to the changes in the wild type that lost 49% of fatty acids.

DISCUSSION
Fatty acid catabolism is best characterized as a major metabolic pathway during seed germination and seedling growth (Graham, 2008), but it is also active throughout the plant life cycle. Although a less predominant metabolic flux during plant senescence, fatty acid degradation is an important metabolism, since it can mobilize carbon stored in leaves for utilization during seed filling and can eliminate deteriorative effects of free fatty acids released from lipid turnover (Gerhardt, 1992).
In this investigation, we systematically studied the fatty acid profiles of Arabidopsis, Brachypodium, and switchgrass leaves during natural senescence. Although similar results with other plants have been presented in earlier studies (Hirayama and Oido, 1969;Ferguson and Simon, 1973;Fong and Heath, 1977;Koiwai et al., 1981;Harwood, et al., 1982;Yamauchi et al., 1986;Wanner et al., 1991), this is a more detailed and longer term study than previously reported. In addition, this is, to our knowledge, the first study of Brachypodium and switchgrass lipids and of how b-oxidation mutations influence lipid turnover in plant leaves during natural senescence.
Our results revealed that during their senescence (days 51-104), Arabidopsis leaves lost about 1.6% of fatty acids every day, resulting in 83% lower fatty acid contents per dry weight at the end of senescence. In the same plants, a 94% loss of chlorophyll was observed; thus, the degradation rate of chlorophyll was approximately 1.8% per day. In a previous study, it was reported that during leaf senescence, nitrogen was decreased by 85.4% over a period of 10 d (Himelblau and Amasino, 2001). That means nitrogen was lost at a rate of approximately 8.5% per day in senescing leaves. Thus, fatty acid in senescing leaves degraded at a rate similar to chlorophyll but more slowly than nitrogen. Interestingly, the rate of fatty acid loss in senescing leaves is very close to that observed in young leaves. Bonaventure et al. (2004) found that breakdown of fatty acids also occurred in young Arabidopsis leaves, and the turnover rate of fatty acids was about 2% per day determined by 14 CO 2 radiolabeling. These results may suggest that fatty acid turnover by oxidation, especially b-oxidation, is similar in young and old leaves. Therefore, the net disappearance of fatty acids is as much the result of declining synthesis of fatty acids in senescent leaves as an increase in degradation, as suggested previously (Meir and Philosoph-Hadas, 1995).
During seed germination, the rate of fatty acid breakdown in seeds is 50 mg dry weight d 21 (inset of Fig. 4). Thus, during leaf senescence, the rate of fatty acid turnover is approximately 80-fold lower on a dry weight basis than fatty acid breakdown during seed germination. These large differences in fluxes are consistent with the level of expression of genes during seedling growth and in leaves (Graham, 2008).
Among the major fatty acid classes, 18:3 showed the largest decline and 16:0 showed the least (Fig. 3). The ratio of unsaturated to saturated fatty acids at the end of senescence was only 1.5, much lower than the value of 6 during leaf expansion. Thus, there is a preferential degeneration of unsaturated fatty acids in the senescing leaves. Although in part this likely reflects the earlier degradation of the highly unsaturated plastid galactolipids (Fong and Heath, 1977;Meir and Philosoph-Hadas, 1995;Thompson et al., 1998), it may also suggest that there is additional nonenzymatic oxidation (i.e. autoxidation due to reactive oxygen species, such as superoxide anion, hydrogen peroxide, and hydroxyl radical). Lipid peroxidation usually occurs in plant senescence, and unsaturated fatty acids are considered preferential targets of attack for reactive oxygen species (Paliyath and Droillard, 1992).
The importance of fatty acid b-oxidation for seed germination and postgerminative growth has been demonstrated in many b-oxidation mutants (Graham, 2008). These mutants are almost entirely blocked in the fatty acid b-oxidation process and thus accumulate long-chain acyl-CoAs in seedlings. It was suggested that senescence and seed germination share similar metabolic processes and that fatty acid b-oxidation also occurred in senescent organs (Graham and Eastmond, 2002). Indeed, a recent investigation revealed a higher expression of the KAT2 gene in natural and dark-induced leaf senescence in Arabidopsis (Castillo and Leon, 2008). By comparing gene expression using a microarray database (http://bbc.botany. utoronto.ca/efp/cgi-bin/efpWeb.cgi or http://lipids. plantbiology.msu.edu/?q=lipids/genesurvey/), we found that the expression of the KAT2 gene in senescent leaves is 2.5-fold higher than in young leaves. Other genes involved in lipid degradation also have higher levels of expression in senescing leaves. For example, the expression of LACS6 and LACS7 increases by 2.6-and 3.1-fold, ACX1 and ACX2 by 1.8-and 3.3-fold, and AIM1 and AtMFP2 by 2.3-to 2.5-fold when comparing young and senescent leaves. All of these facts point to a role for induction of the LACS, ACX, and KAT2 genes during fatty acid b-oxidation in senescent leaves and suggested that their mutation might slow this metabolic process. However, our results have shown that the breakdown of total fatty acids as well as each fatty acid class (data not shown) still occurred in b-oxidation mutants of acx1acx2, lacs6lacs7, and kat2 during leaf senescence. There was no substantial reduction in loss of fatty acids in Arabidopsis senescent leaves compared with the wild type, although the activities of ACX1, ACX2, LACS6, LACS7, and KAT2 were inhibited in the mutants. ACX1 exhibits medium-to long-chain activity ranging from C12:0 and C16:0 (Hooks et al., 1999), ACX2 has optimum activity with long-chain saturated and unsaturated acyl-CoAs (C14:0 to C20:0; Hooks et al., 1999), and LACS6 and LACS7 exhibit long-chain activity (C16:0 to C20:0; Shockey et al., 2002). Explanations for the lack of impact of these mutations are possibly that the b-oxidation rate is slow in senescent leaves, allowing other ACX, LACS, or KAT genes to compensate for the lack of ACX1ACX2, LACS6LACS7, or KAT2 in these mutants. For example, KAT5 is able to partially complement kat2 under the promoter of 35S (Germain et al., 2001). In addition, the expression of ACX3 (At1g06290) and ACX4 (At3g51840) in senescent leaves is also more abundant than in young leaves (Hayashi et al., 1999;Froman et al., 2000). Alternatively, the possibility of additional metabolic pathways for fatty acid degradation in senescent leaves cannot be ruled out. It has also been reported that fatty acids can be transported in the phloem via lipid particles (Madey et al., 2002).
Although previous studies have reported substantial accumulation of TAG in senescent leaves (Kaup et al., 2002;Lin and Oliver, 2008), we did not observe such increases. TAG contained less than 1% of total leaf fatty acids at all times in Arabidopsis. We did observe higher levels of TAG in switchgrass and in Malus leaves. Our GC and ESI-MS analyses of 2-and 20-week-old Malus leaves indicated levels ranging from 1.3 to 1.8 mg TAG mg 21 dry weight, or less than 0.2% of leaf dry weight.
The soluble aliphatic surface lipids of Arabidopsis proved much more stable during senescence than the membrane fatty acids. To our knowledge, this is the first study to compare the turnover of both membrane and surface lipids. The greater stability of the extracellular and predominantly saturated structures is not surprising, but it highlights the fact that one strategy to increase the energy content of biofuel crops might be to enhance surface lipid production. This has been achieved in principle by different transgenic strategies (Broun et al., 2004;Zhang et al., 2005;Li et al., 2007). As noted above, increases in the lipid content of leaves will increase the energy density of biofuel crops and may have potential to improve their economics. Considering that Miscanthus yields are routinely near 25 tons ha 21 (Heaton et al., 2008), a fatty acid content of only 5% of dry weight would correspond to fatty acid yields of 1,200 kg ha 21 , which is higher than canola (Brassica napus), the highest yielding temperate oil crop. If burned to generate electricity, such a crop would yield 10% more power.
The rosette leaves of Arabidopsis are numbered from the bottom (Guo and Gan, 2006). Leaves 5 to 8 were selected from nine to 12 plants for these experiments. Approximately 30 leaves in total were harvested, and leaf samples were pooled. All plants used in a given experiment were taken from a single population growing at the same time. Leaves were frozen at 280°C, and pooled leaves were dried for approximately 24 h using a lyophilizer (Labconco Freeze Dry System). Comparison with leaves dried at 80°C indicated that both methods yielded the same dry weight.

Fatty Acid Determination via Direct Methylation
Direct transmethylation of fatty acids into fatty acid methyl esters for GC analysis was performed according to the modified method of Browse et al. (1986). In brief, leaf samples (10-50 mg) were introduced into a glass tube with a Teflon-lined screw cap that was prerinsed thoroughly with chloroform. Then, 1 mL of 5% (v/v) H 2 SO 4 in methanol, 20 mL of 0.2% butylated hydroxy toluene in methanol, 20 mg of margaric acid (C17:0) as an internal standard, and 300 mL of toluene were added as a cosolvent. After that, the tubes were heated for 1 to 1.5 h at 90°C to 95°C. When samples were cooled to room temperature, 1 mL of 0.9% (w/v) NaCl and 1 mL of hexane were added. The tubes were vortexed vigorously and centrifuged at 3,000g for 5 min. The organic phase (up phase) was transferred to a new tube, and the solvent was evaporated under nitrogen and redissolved in hexane. The fatty acid methyl ester extracts were analyzed using an Agilent 6890 N GC apparatus with a flame ionization detector on a DB23 column (30 m length 3 0.25 mm i.d., 0.25 mm film thickness; J&W Scientific).

Lipid Analysis
Lipid extractions followed the method of Hara and Radin (1978). Leaf samples (up to 400 mg) were heated in isopropanol (4 mL) at approximately 80°C for 5 to 10 min to inactivate lipases. When cooled, leaves were homogenized using a Polytron. Hexane (6 mL) was then added to keep the correct volume ratio of water:isopropanol:hexane (1:4:6, v/v/v) for a monophasic system formation. This was followed by the addition of 5 mL of 6.6% (w/v) sodium sulfate solution. After samples were mixed vigorously and centrifuged for phase separation, the upper phase was transferred to a clean tube and evaporated under nitrogen.
Individual lipid bands were scraped off the TLC plate and eluted with chloroform:methanol (2:1, v/v). The eluate was evaporated to dryness under N 2 , and the lipids were stored in toluene at 220°C prior to GC or ESI-MS analysis.

TAG Analysis by ESI-MS
Neutral lipids were separated from total lipids by liquid column chromatography. A short column (approximately 7 cm in length) of silica gel was prepared from 1 g of silica acid (100-200 mesh) in a glass disposable Pasteur pipette (22.3 cm length 3 0.6 cm i.d.) plugged with solvent-washed glass wool. The column was conditioned by elution with 5 mL of chloroform, after which the lipid samples were applied to the top of the column. Samples were eluted with 10 mL of chloroform, and neutral lipid fractions were collected and dried under nitrogen.
Before analysis by ESI-MS, tritridecanoin (triC13:0) was added as an internal standard to the resulting neutral lipid samples. Then, 1 mL of the samples was directly injected into the ESI source. ESI-MS experiments were conducted with a Micromass Quattro LCZ quadrupole mass spectrometer. In MS mode, fragmentation was achieved by introducing argon into the reaction chamber in front of the second quadrupole.

Leaf Wax Analysis
Leaves were dipped in chloroform (20 mL) for 30 s twice. n-Octacosane, docosanoic acid, and 1-tricosanol (20 mg each) were added as internal standards. After removal of chloroform under nitrogen, the resulting waxes were silylated to convert free alcohols and carboxylic acids to their trimethylsilyl ethers and esters, respectively, by heating the sample at 110°C for 10 min in medium of 100 mL of pyridine and 100 mL of bis(trimethylsilyl)trifluoroacetamide. After cooling, the solvent was evaporated under nitrogen and the product was resuspended in mixed solvent of heptane:toluene (1:1, v/v) for GC-MS analysis (Li et al., 2007).

Chlorophyll Determination
Fresh leaves were weighed and then ground in liquid nitrogen using a mortar and pestle. The pulverized leaf tissue was added to 80% acetone (0.1 mL mg 21 leaf tissue), vortexed, and incubated for 30 min in the dark at room temperature. The solution was vortexed and centrifuged (3,000g, 10 min), and the absorbance of the supernatant was measured at 663 and 645 nm to determine chlorophyll based on fresh weight (Lichtenthaler and Wellburn, 1983).

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S1. Age-dependent changes in the contents of total fatty acids based on per leaf (A) and unit area (B; cm 2 ) of Arabidopsis leaves.
Supplemental Figure S2. Levels of aliphatic wax in Arabidopsis leaves during growth and senescence.
Supplemental Figure S3. ESI-MS spectra of the TAG component in neutral lipids of different plant leaves.
Supplemental Figure S4. TAG levels in Arabidopsis leaves at different developmental times.
Supplemental Table S1. Fatty acid composition in entire seed, embryo, and endosperm of Brachypodium.