Article Rapid kinetic labeling of Arabidopsis cell suspension cultures: Implications for models of lipid export from plastids

Cell cultures allow rapid kinetic labeling experiments that can provide information on precursor product relationships and intermediate pools. T-87 suspension cells are increasingly used in Arabidopsis research, but there are no reports describing their lipid composition or biosynthesis. To facilitate application of T-87 cells for analysis of glycerolipid metabolism, including tests of gene functions, we determined composition and accumulation of lipids of light and dark-grown cultures. Fatty acid synthesis in T-87 cells was 7 to 8-fold higher than in leaves. Similar to other plant tissues, phosphatidylcholine (PC) and phosphatidylethanolamine were major phospholipids, but galactolipid levels were 3-4 fold lower than Arabidopsis leaves. Triacylglycerol represented 10% of total acyl chains, a greater percentage than in most non-seed tissues. The initial steps in T-87 cell lipid assembly were evaluated by pulse labeling cultures with [ 14 C]acetate and [ 14 C]glycerol. [ 14 C]acetate was very rapidly incorporated into PC, preferentially at sn-2 , and without an apparent precursor-product relationship to diacylglycerol (DAG). In contrast, [ 14 C]glycerol most rapidly labeled DAG. These results indicate that ‘acyl editing’ of PC is the major pathway for initial incorporation of fatty acids into glycerolipids of cells derived from a 16:3-plant. A very-short lag time (5.4 s) for [ 14 C]acetate labeling of PC implied ‘channeled’ incorporation of acyl chains without mixing with the bulk acyl-CoA pool. Subcellular fractionation of pea-leaf protoplasts indicated that 30% of lysophosphatidylcholine acyltransferase activity co-localized with chloroplasts. Together these data support a model in which chloroplast PC participates in trafficking of newly synthesized acyl chains from plastids to the endoplasmic reticulum.


INTRODUCTION
Cell suspension cultures are commonly used and are important model systems for study of many aspects of plant cell biology, biochemistry and molecular biology (Razdan, 2003). In contrast to leaf or many other plant tissues that have diverse cell types growing at different rates, the cell population of suspension cultures is much less complex. In liquid medium, many plant cell lines grow rapidly and are readily transformable at high efficiency with Agrobacterium tumefaciens (An, 1985;Nagata et al., 2004b;Ogawa et al., 2008). This eases the generation and selection of large numbers of independent transgenic lines compared to whole-plant transformation. Although tobacco BY2 cells have been the most widely used cell culture system, the abundant information and molecular and genetic tools available for Arabidopsis have increased interest in T-87 cells as a model for molecular and biochemical investigations (e.g. Alonso et al., 2010). In addition, a high-throughput procedure for the testing of large numbers of transgenes in a 96-well format has been described (Ogawa et al., 2008).
Arabidopsis T-87 cells originate from seedlings of Arabidopsis (Axelos et al., 1992). To date, studies of their lipid composition or metabolism have not been reported. To evaluate the utility of T-87-cells as a model for lipid synthesis we analyzed their lipid composition and conducted a set of experiments using [ 14 C]acetate and [ 14 C]glycerol to investigate initial steps in glycerolipid synthesis. The ability to conduct rapid pulse labeling of T-87 cells provides the ability to track precursor-product relationships and to derive information on precursor pools involved in the incorporation of newly synthesized fatty acids into membrane lipids.
In plants, the incorporation of the newly synthesized acyl chains into glycerolipids occurs by two independent pathways: the prokaryotic pathway inside plastids and the eukaryotic pathway outside the plastid (Frentzen et al., 1983;Heinz and Roughan, 1983). In the eukaryotic pathway, acyl-ACP products of fatty acid synthesis are hydrolyzed by plastid acyl-ACP thioesterase reactions and exported to an outer envelope-bound acyl-CoA (acyl-CoA) synthetase. De novo assembly of glycerolipids occurs largely at the endoplasmic reticulum (ER) by two acylations of glycerol-3-phosphate (G3P) to form phosphatidic acid (PA) (Kornberg and Pricer, 1953;Somerville et al., 2000). PC, the major phospholipid of plant cells is synthesized via PA 7 2010). As indicated in Fig. 1, cultures grown with 50 µmol m -2 s -1 light or grown in the dark accumulated similar biomass, fatty acid levels and protein throughout a 7-day cultivation period.
Between day 3 and 5, T-87 cell culture dry weights increased 2.7-fold and 3.1-fold for light and dark conditions (Fig. 1A), which corresponds to a biomass doubling rate of 35 and 31 h, respectively. The net rate of fatty acid synthesis of light-grown T-87 cultures during the 3 to 5 day rapid growth period (calculated from Fig. 1 A and B) was 18 nmol C • h -1 mg -1 fresh weight (expressed on basis of atoms of carbon) ( Fig. 1A and B). For comparison, expanding leaves of 3-week old Arabidopsis plants have a rate of fatty acid synthesis of ~2.3 nmol C h -1 mg -1 fresh weight (Bao et al., 2000;Bonaventure et al., 2004). Thus, the net rate of fatty acid synthesis in T-87 cells is 7 to 8 fold higher compared to young plants with rapidly expanding leaves and 50-fold higher compared to leaves in the dark (Browse et al., 1981).
At day 5 the total fatty acid content of T-87 cells was 98-118 µg mg -1 dry weight which is 1.5 to 2-fold higher than levels observed in Arabidopsis leaves (40 µg FA mg -1 dry weight) (e.g. Yang andOhlrogge, 2009: Li-Beisson et al., 2010). The most abundant fatty acids were linoleic (18:2), linolenic (18:3) and palmitic acid (16:0) (Table I). 16:3 represented 1.0-1.5 mol% compared to 13.8 mol% for leaves (Miguel and Browse 1986). The low 16:3 fatty acid level is similar to 15 d-old Arabidopsis roots (1.5 mol%) (Beaudoin et al., 2009). One clear difference between T-87 cells grown in light or dark was the accumulation of chlorophyll. While darkgrown cultures were yellow, light-grown T-87 cultures were pale green and contained about 0.2 mg chlorophyll per gram dry weight or 2% of the levels observed in leaf tissue (10 mg/g dry weight) (Stand et al., 1999). Low levels of the largely thylakoid associated 16:3 fatty acid in T-87-light compared to leaf tissue are consistent with the low chlorophyll content.
The major phospholipids of light and dark-grown T-87 cells were PC and phosphatidylethanolamine (PE) at 45 and 12.6 mol%, respectively (Fig. 2). Other phospholipids including phosphatidylinositol (PI), phosphatidylglycerol (PG), phosphatidylserine (PS) and PA were minor components at 6.6, 4.1, 4.1 and 1.6 mol%, respectively. Lysophospholipids (lysoPC, lysoPE and lysoPG) represented <0.5 mol%. The relative proportions of phospholipids in T-87 cell cultures (PC>PE>PI, PS>PA) were similar to values reported for other Arabidopsis tissues (Miguel and Browse, 1992;Welti et al., 2002;Li-Bession et al., 2010 Goldschmitt et al., 1992). Lipid compositions were similar in light and dark, and more closely resembled non-photosynthetic Arabidopsis tissues than leaves. Overall, T-87 cells displayed no major anomalies related to their growth in suspension cultures.
The distribution and composition of pairs of O-acyl chains on polar lipids constitute the "molecular species" within a lipid class. This arrangement of acyl chains influences the physical properties of membranes which changes in response to osmotic and temperature stress (Lynch and Thompson, 1984). To further compare T-87 lipids to Arabidopsis plants, the molecular species of phospholipids and galactolipids were characterized by ESI-MS/MS (Fig. 3A).
Molecular species of PC, PE, PI and PA and DGDG were similar to those observed in leaves and roots (Li et al., 2006). However, compared to data for leaves (Miquel and Browse 1992), and consistent with the lower 16:3 levels, there was a lower abundance of 16:x/18:x (34:0-34:6) molecular species of MGDG compared to 18:x/18:x (36:1-36:6). In addition, molecular species with longer chain fatty acids (>20 carbon) which are characteristic of PS and known to induce membrane curvature (Israelachivlli et al., 1980) were more abundant in T-87 cells compared to leaf tissue (Nerlich et al., 2007;Li et al., 2006). This may relate to the more spherical shape of suspension culture cells compared to leaf tissue.

T-87 cells accumulate TAG
TAG is a major component of seeds and pollen, but is usually found at very low levels in vegetative tissues. For example, TAG is less than 1% of lipids of leaves during growth from seedlings until senescence of Arabidopsis, Brachypodium and Switchgrass (Yang and Ohlrogge, 2009). In contrast, T-87 cells accumulated TAG as a major neutral lipid, in addition to the polar lipids described above. TAG represented 10.8 mol% of total fatty acids (7.6 mol% of 9 glycerolipids) in light-grown T-87 cells (Fig. 2). Linoleic and linolenic acids were the predominant fatty acids (>50% of total) of TAG for T-87 cells grown in either light or dark (Fig 3   B). The TAG fatty acid composition was distinct from Arabidopsis seeds which include ~ 26% fatty acids with chain length >C18 (largely 20:1), (Li et al., 2006). In contrast, the fatty acid compositions in TAG derived from T-87 cells was very low in eicosenoic acid (20:1) or other >C18 fatty acids. In agreement with the content based on fatty acid levels, after [ 14 C]acetate labeling (0-60 min) TAG accounted for 11% of total [ 14 C]-label recovered in lipids at 60 min (Supplemental Fig. S1). The occurrence of TAG in T-87 cells and the convenient transformation with Agrobacterium suggests that T-87 cells could be a useful model system for evaluating TAG metabolism in non-seed tissues and to evaluate gene candidates involved in TAG accumulation.

Radiolabel incorporation into T-87 cells to probe initial kinetics of lipid biosynthesis
The rapid growth and uniform properties of T-87 cultures are particularly useful for studying precursor-product relationships through rapid pulse radioisotope labeling. Incorporation of label  (Roughan et al., 1976;Bao et al., 2000;Koo et al., 2004). In contrast, [ 14 C]glycerol is primarily incorporated into the glycerol backbone of glycerolipids (Slack et al., 1977).  4B). Further analysis of the radiolabeled acyl groups by silver TLC revealed that over 90% of the [ 14 C]-label at the sn-2 position was monoenes whereas less than 3 % was in the form of saturates and dienes. In contrast, at sn-1 there was a 45/55 distribution between saturates and monoenes, with trace amounts of dienes (Fig. 4B).
To determine the regiospecificity of acyl chain labeling in DAG, pancreatic lipase was used to selectively remove the acyl groups from the sn-1/sn-3 position but not the acyl chain at sn-2 (Christie, 2003). In contrast to PC, DAG had a much more even distribution of [ 14 C]-acyl chains on the glycerol backbone, with 52% and 48% of the radiolabel from [ 14 C]acetate associated with sn-2 and sn-1, respectively. In addition, the distribution of [ 14 C]-saturated and unsaturated fatty acids in DAG was substantially different from that of PC ( (Roughan et al., 1980). Together, the acetate and glycerol experiments clearly demonstrate that initial incorporation of acyl chains into PC and incorporation of glycerol into the glycerolipid backbone occur through different pathways. Furthermore, acyl editing is a major pathway for incorporation of newly synthesized fatty acids into glycerolipids of T-87 cells, consistent with the 'acyl editing' pathway described by Bates et al. (2007;2009). This conclusion is supported by three lines of evidence: 1) the rapid incorporation of [ 14 C]-acyl chains into PC relative to DAG (Fig. 4A); 2) the preferential incorporation of [ 14 C]acetate label into sn-2 of PC compared to sn-1. This pattern differed from DAG where radiolabelled acyl chains were almost evenly distributed between sn-1 and sn-2 (Fig. 4B); 3) the major initial product of [ 14 C]glycerol labeling was DAG whereas PC labeling lagged at least one min (Fig. 5).
Are newly synthesized acyl chains channeled into PC at the plastid envelope? Consideration of lag times and precursor pools sizes.
The initial kinetics and lag time for radiolabel incorporation into products can provide information on the pool sizes of intermediates of a metabolic pathway. Linear kinetics of incorporation of label into product will also not occur until the intermediate pools of a pathway are saturated with radioactivity (Segel, 1976 to the x-axis (time), an average lag of 5.4 ± 4.4 s (Fig. 6A) was determined.
The where k=rate of fatty acid synthesis / substrate pool (Segel, 1976). Based on the rate of fatty acid synthesis of 4.8 pmol C s -1 mg -1 fresh weight (calculated per s from Fig. 1B)  enzyme turnover time plus the precursor pools is 18.1 s (6.2 s + 11.9 s) (Fig. 6C) Substrate channeling has previously been proposed for chloroplast fatty acid synthesis (Roughan and Ohlrogge, 1996;Roughan, 1997) and for activation by LACS of the free FA released by acyl-ACP thioesterase after FA synthesis (Koo et al., 2004). Although there are uncertainties (see below), the results above imply that substrate channeling occurs not only for fatty acid and acyl-CoA synthesis, but extends to the incorporation of newly synthesized acyl chains into PC. Below we present additional data related to this hypothesis.

Substantial lysoPC acyl transferase activity is associated with chloroplasts.
Although in vivo data are lacking, based on in vitro assays, LPCAT is a strong candidate for the enzyme activity that is responsible for PC acyl editing (Bates et al., 2007;Ståhl et al., 2008).
LPCAT transfers acyl chains from acyl-CoA onto lysoPC and also catalyzes the reverse reaction (Stymne and Stobart, 1984). The forward and reverse reactions together constitute one possible acyl-exchange mechanism between acyl-CoA and PC.
LPCAT activity is generally considered to act at the ER although its subcellular localization has not been well established. Addition of [ 14 C]acyl-CoA to isolated pea, spinach or leek chloroplasts results in [ 14 C]PC as the major labeled glycerolipid (Bertrams et al., 1981;Bessoule et al., 1995;Kjellberg et al., 2000). LPCAT activity has also been directly assayed in chloroplast envelopes isolated from pea leaves (Kjellberg et al., 2000). This activity was unaffected by thermolysin treatment suggesting it resides in the inner envelope or the inner face of the outer envelope. Although ER contamination of the chloroplasts was not assessed, these results indicate there is a plastid envelope-associated LPCAT activity, which potentially is involved in acyl editing.
Kjellberg et al. (2000) did not assay other subcellular fractions and thus did not determine whether the envelope-associated LPCAT is a minor or a more substantial proportion of total cellular LPCAT.
To further investigate a possible role for acyl editing at the chloroplast envelope we determined the subcellular distribution of LPCAT activity. Because isolation of intact organelles with high yield from T-87 cells is problematic, we fractionated lysed pea-leaf protoplasts by ultra-centrifugation on a linear sucrose gradient (Fig. 7A). We observed a clear correspondence between peaks of LPCAT activity and chlorophyll distribution. Approximately 30% of all LPCAT activity was associated with chloroplasts ( Fig. 7B, shaded region). LPCAT activity was also observed in fractions enriched in endoplasmic reticulum and in plasma membranes as determined by marker enzyme activity (Fig. 7C). LPCAT activity recovered at the top of the gradient, (presumably light microsomes and soluble proteins) was recovered in the pellet after centrifugation at 100 000 g max , indicating that this LPCAT activity is also membrane bound.
Approximately 24% of the cytochrome c reductase ER marker was associated with chloroplast fractions and may represent plastid associated membranes (PLAM; Andersson et al., 2007) or other ER-plastid associations (Kaneko and Keegstra, 1996;Hanson and Köhler, 2001).
Not only is substantial LPCAT activity associated with chloroplasts, but PC, its substrate (reverse reaction) is also the major phospholipid of the chloroplast envelope (Dorne et al., 1985).
In fact, it can be estimated that 40% of total cellular PC of leaves is localized in the outer envelope (Supplement Text 2). PC is also a major phospholipid of oilseed plastids (Miernyk, 1985). Therefore, after fatty acids are exported from the plastid and esterified to CoA by LACS, the acyl-CoA would encounter both abundant PC substrate and LPCAT activity. The colocalization of the LACS and LPCAT enzymes and the substrates for acyl editing at the same site as fatty acid export supports a hypothesis that acyl groups are channeled into PC at the chloroplast envelope without mixing with the bulk acyl-CoA pool. As noted above, this scenario is also supported by the very small lag time for [ 14 C]acetate incorporation into PC ( Fig. 6A and   B). In this study we have provided data on the lipid composition and initial reactions of glycerolipid biosynthesis of T-87 cell cultures. Together with high-throughput transformation methods, these data should be useful as a baseline for design and analysis of additional experiments, for example the testing of functions of lipid biosynthesis or regulatory genes. In addition, labeling data indicated that a major flux of newly synthesized FA into glycerolipids occurs via acyl editing. Thus, this pathway is widespread, occurring in 16:3 as well as 18:3 plants.

Conclusions/Perspective
The very rapid labeling of PC acyl chains (Fig. 4A, Fig. 6), together with the subcellular distribution of LPCAT (Fig. 7B) and of PC provides insight into models of plant lipid trafficking.
The two-pathway model of plant lipid metabolism (Roughan and Slack 1984) describes major trafficking of acyl chains from the plastid to the ER and the return of acyl chains from ER to plastid. In this model, acyl chains first synthesized in the plastid are exported to the ER for incorporation into glycerolipids and for further desaturation. After desaturation, the return of acyl chains from the ER to plastids is most often considered to involve PC as a carrier (Somerville et al., 2000;Benning, 2008). In contrast, the trafficking of newly synthesized acyl chains from the plastid to the ER has generally been assumed to occur through an acyl-CoA pool in the cytosol, followed by their incorporation into ER glycerolipids. However, in vivo evidence for acyl-CoA as a carrier of acyl chains from plastid to ER is lacking. From the data presented in this study, instead of acyl-CoA movement through an acyl-CoA pool in the cytosol, we propose that newly synthesized acyl chains enter PC via substrate channeling at the plastid envelope and that PC (rather than acyl-CoA) may then serve as a carrier of acyl chains from plastids to the ER. Thus, PC may be central to acyl fluxes that occur in both directions between plastids and the ER. This hypothesis is supported by 1) association of 30% of pea leaf LPCAT. activity with chloroplasts: 2) localization of 40% of leaf PC in the chloroplast envelope; 3) very rapid incorporation of acyl groups into PC by acyl-editing with a lag-time less than predicted if flux is through the bulk acyl- chloroplast contact sites or associations (Xu et al., 2008). Finally, uncertainties in the subcellular distribution of acyl-CoA pools (and extrapolation to T87 cells) contribute to the provisional nature of this model. Nevertheless, consideration of PC as a carrier for acyl-chains from the plastid to the ER, as well as for the reverse traffic, may be useful in building a more complete understanding of acyl lipid metabolism in plants.

Plant material
Arabidopsis T-87 cells were grown either in light (40-50 µmol m -2 s -1 ) or in the dark at 120 rpm in media as described in Alonso et al., 2010. Cell cultures were maintained by a 1: 9 (culture:

Lipid Extraction and analysis
Cultured cells were harvested by centrifugation (1800g max ) and washed with distilled water three times. Cells were resuspended in boiling isopropanol, heated for 10 min and lipids were extracted according to (Hara and Radin, 1978). Neutral lipids were separated on K6 TLC plates Levels of glyco-and phospholipids were separated by solid phase extraction according to Andersson et al., (2005). Purity of glyco-and phospholipids was examined by TLC and their abundances were quantified as FAMEs by GC-FID as above. The relative proportions of individual molecular species were determined for five biological replicates. Samples were analyzed by electrospray ionization triple quadrupole mass spectrometry with internal standard for each phospholipid and galactolipid class) at the Kansas Lipidomics Research Center (www.kstate.edu/lipid/lipidomics; Welti and Wang, 2004).

Radio-labeling of lipids
For [  (lysoPC and free fatty acids) were separated on TLC plates and each band was transmethylated to FAMEs and separated by silver TLC (10% (w/v) AgNO 3 ).

Protoplast isolation and fractionation
The lower epidermis of rapidly expanding pea leaves (7-8-days old) was abraded with a nylon brush and sliced into 1mm strips. Prior to digestion, leaf-strips were incubated for 1 h in preplasmolysis media (330 mM sorbitol, 1 mM CaCl 2 , 10 mM MES pH 6.0). The leaf strips were transferred to digestion media (2% cellulase, 0.2% mazerozyme, 550 mM sorbitol, 1mM CaCl 2 , 0.25% BSA, 10 mM MES pH 6.0) and incubated at 30° C, ~30-40 µmol m -2 s -1 with gentle agitation for 3 h. Released protoplasts were separated from the leaf tissue by filtration through 100 µm nylon mesh and leaf tissue was washed twice with wash buffer (550 mM sorbitol, 1 mM CaCl 2 , 10 mM MES pH 6.0) to collect additional protoplasts. The resulting protoplasts were washed twice by spinning (70 g max for 5 min) and further purified using differential centrifugation in combination with stacked 35% and 25% Percoll gradients (3 mL each) with 5 mL wash buffer on top. After centrifugation (250 g max for10 min), intact protoplasts were collected at the 25%interphase and washed twice by centrifugation (70 g max for 5 min) in wash buffer to remove residual Percoll. Protoplasts were ruptured by three passes thorough a 20 µm nylon mesh (Nishimura et al., 1976). The ruptured protoplasts (~1 mL) were separated on a 11 mL 20-60% linear sucrose gradient (10 mM HEPES pH 7.0, 1 mM CaCl 2 ) at 100 000 g max for 4 hours. 700 µL fractions were collected and frozen in liquid nitrogen and stored at -80° C until further analysis.
Chloroplasts were identified by the presence of chlorophyll as described (Arnon, 1949). Protein content was determined using bicinchoninic acid reagent (Smith et al., 1985).

Enzyme assays
LyosPC acyl transferase activity was assayed at room temperature in a final volume of 30 µL

27
CoA (Slack et al., 1977). Our analysis indicated 1.5-4% of the radioactivity from glycerol labeling was located in the acyl chains. Similar values have been reported for [ 14 C]glycerol labeling of DAG in soybean embryos (Bates et al., 2009).