Lysophosphatidic Acid Acyltransferase from Coconut Endosperm Mediates the Insertion of Laurate at the <em>sn-2</em> Position of Triacylglycerols in Lauric Rapeseed Oil and Can Increase Total Laurate Levels

Plants

Canola-type oilseed rape (Brassica napus) plants transformed with BTE, a 12:0-ACP thioesterase from California bay laurel (Umbellularia californica) that is driven by a napin promoter, were derived from the original transformant pCGN3828-23, as described previously (Voelker et al., 1996). This plant had 37 mol % of laurate in pools of segregating seeds, and its genome contained approximately 11 to 14 copies of the transgene on at least four to five segregating loci (not shown). Single seeds had up to 56 mol % laurate. After several generations of self-pollination and crossing to other canola varieties, homozygous lines were obtained and contained up to 58 mol % laurate. Line DH22 (full designation [A112X3828-23-198]-14-DH22-125) contained 51 mol % laurate; line DH63 (full designation LA30056-5-DH63-14-2) contained 59 mol % laurate.

Expression of CLP in Transgenic Plants

To facilitate cloning into plant expression vectors, the CLP cDNA clone pCGN5503 (Knutzon et al., 1995) was modified by PCR to insert SalI and BamHI restriction sites immediately upstream of the start and downstream of the termination codons, respectively. The CLP-coding region was inserted as a SalI/BamHI fragment into the seed-specific napin expression cassette pCGN3223 (Kridl et al., 1991), which had been digested with SalI and BglII, creating pCGN5509. The HindIII fragment of pCGN5509 containing the napin 5′-regulatory region, the CLP-coding region, and the napin 3′-regulatory region was inserted into the HindIII site of the binary vector pCGN1578PASS (McBride and Summerfelt, 1990) to create pCGN5511. pCGN5511 was transferred to Agrobacterium tumefaciens EHA101 and used to transform B. napus as described by Radke et al. (1988). Two different canola varieties were used for transformation: LP004, a low-linolenic line, and Quantum, a line with a standard canola composition. Plant cultivation and sample harvesting have been described previously (Voelker et al., 1996). Dihaploid plants were generated by the procedure of Eickenberry (1994).

Enzyme Assays and Total Fatty Acyl Composition

For LPAAT assays, developing seed pools at the mid-maturation stage (approximately 30 d after pollination) were harvested. Crude (P2) membrane fractions were prepared and assayed using radiolabeled acyl-CoA substrates, as described previously (Davies et al., 1995). Total fatty acid composition was determined via quantitative GC analysis of methyl esters, as described previously (Browse et al., 1986).

sn-2 Analysis of Triglycerides

Procedures for the extraction of seed oil, for analysis of the sn-2 composition of oil using Rhizopus arrhizus lipase and for reversed-phase HPLC of seed oil TAGs, were modifications of Voelker et al. (1996). For sn-2 determinations, R. arrhizus lipase was diluted to approximately 600,000 units/mL and held on ice. The reaction mixture contained 2.0 mg of oil, 200 μL of 0.1 m Tris-HCl, pH 7.4, 40 μL of 2.2% (w/v) CaCl₂·2H₂O, and 100 μL of 0.05% (w/v) bile salts. After 5 min of sonication to disperse the oil, 20 μL of dilute lipase was added and the mixture was vortexed continuously for 1 min at room temperature. The reaction was stopped with 100 μL of 6 m HCl. Subsequently, 500 μL of CHCl₃:MeOH (2:1, v/v) was added. The organic extract was removed, and the aqueous phase was re-extracted with 300 μL of CHCl₃. The organic extracts were combined and held on ice to minimize acyl migration before HPLC separation. The digestion products were dried down in chloroform to approximately 200 μL, and 60 μL was used for HPLC analysis.

The HPLC system was equipped with an evaporative light-scattering detector (Varex ELSD IIA, Alltech, Deerfield, IL) with the tube temperature set at 105°C and the nitrogen gas flow at 40 mL/min, an autosampler (model 712 Wisp, Waters), three solvent-delivery modules (model 114M, Beckman), a controller (model 421A, Beckman), a pneumatically actuated stream splitter (Rheodyne L.P., Rohnert Park, CA), and a microfractionator (Gilson USA, Middleton, WI). The chromatography column was a 220- × 4.6-mm, 5-μm, normal-phase silica cartridge (Brownlee Precision Co., Santa Clara, CA). The mobile phases were hexane:toluene, (1:1, v/v) (solvent A), toluene:ethyl acetate, (3:1, v/v) (solvent B), and 5% (v/v) formic acid in ethyl acetate (solvent C). At a flow rate of 2.0 mL/min, an isocratic gradient of 10% solvent B and 2% solvent C for 2 min were applied, followed by a linear gradient of 2% to 25% solvent C over 6 min and then an isocratic gradient of 25% C for 8 min and a linear gradient of 25% to 2% solvent C over 1 min. A chromatographic standard mixture was prepared in hexane:toluene (1:1) containing the following: 0.2 mg/mL tri-16:0, 2.0 mg/mL 16:0 FFA, 0.2 mg/mL di-16:0 mixed isomers, 0.2 mg/mL 3-mono-16:0, and 0.2 mg/mL 2-mono-16:0. For each sample, the fraction containing the 2-monoacylglycerol peak was collected automatically by controlled timed events relays, the fractions were evaporated at room temperature overnight, and the fatty acyl composition was obtained as described by Browse et al. (1986).

Triglyceride Chromatography

Silver-phase HPLC resolution of TAGs was performed with the chromatograph described for the sn-2 analysis. A lipid column (250 × 0.6 mm; Chrompack, Raritan, NJ) fitted with a cation-exchange guard column was held at 30°C. The detector's drift tube temperature and nitrogen flow were maintained at 130°C and 40 mL/min, respectively. The mobile phase was composed of 1:1 hexane:toluene (solvent A), 3:1 toluene:ethyl acetate (solvent B), and 500:0.12 toluene:formic acid (solvent C). The mobile phase gradient was programmed as isocratic at 90% solvent A and 10% solvent B for 1.5 min, a linear gradient to 100% solvent B from 1.5 to 1.75 min, isocratic at 100% solvent B for 7.5 min, a step change to 100% solvent C at 9.25 min, followed by a step change to the initial conditions at 9.99 min, and an equilibration delay of 6 min. The flow rate was 2 mL/min and the stream splitter was set at 60/40 (fraction/detector). Seeds were crushed and extracted overnight with 100 μL of toluene containing tri-11:0 as an internal standard at 0.5 mg/mL. Whole-sample extracts were filtered (0.2 μm) and injected (25 μL) onto the silver-phase column. The trisaturated TAG peak fraction including the tri-11:0 was collected and dried at room temperature overnight, and methyl esters were produced by the addition of 50 μL of toluene and 105 μL of 0.5 n NaOH, incubation at 90°C for 10 min, and addition of 500 μL of acetic acid and 60 μL of heptane. After GLC analysis, the trilaurin content was calculated.

Oil Level Determination

Seed oil levels were determined by NMR. Mature seeds (1–2 g) were dried at 130°C for 2 h and equilibrated to room temperature for 12 h before NMR analysis (model 4000, Oxford Instruments, Concord, MA) with an R_F level of 20 μA, an AF gain of 500, and a gate width of 1.0 G. The time of analysis was two times for 30 s (averaged signal/mass). Samples were run in triplicate in 2.0-mL Nessler cylinders of identical weight against 0% and 100% oil calibration controls.

PCR

To detect the CLP gene in transformants, leaf DNA was isolated (Bernatzky and Tanksley, 1986) and coconut (Cocos nucifera) sequences were amplified with the primers TGTGGAACATGATCATGCTGATTTTGCTCC and ATCGAGTACCCTCTGGAAAAATGATCAGCG using standard procedures.

Expression of CLP Alters the LPAAT Substrate Specificity Profile of Transgenic Canola Seeds

Knutzon et al. (1995) identified a coconut cDNA encoding CLP, a 299-amino acid protein with LPAAT activity. Expression of this cDNA in Escherichia coli conferred upon these cells a novel, laurate-preferring LPAAT activity whose substrate specificity profile matched that of the coconut enzyme. We were interested in evaluating whether this CLP enzyme would be capable of competing with the endogenous LPAAT activity in transgenic canola and facilitate the incorporation of laurate into the sn-2 position of TAG. To obtain information on the expression of CLP in canola, the cDNA was expressed in seeds under the control of a seed-specific napin promoter. Seeds at mid-stage maturation were collected and assayed for LPAAT activity using 12:0-LPA and a variety of acyl-CoA donors. As shown in Figure ​Figure11 for one transgenic event, the activity profile of the transgenic seeds differed considerably from that of control, nontransformed seeds. 12:0-CoA was the preferred substrate in the transgenic seeds. Overall, the profile matches that of CLP expressed in E. coli or of a coconut endosperm extract (Davies et al., 1995; Knutzon et al., 1995) overlayed on the endogenous 18:1-CoA activity of the canola control.

Figure 1

LPAAT substrate specificity of CLP-expressing canola seeds. A pool of mid-maturation seeds from a control plant (white bars) and from a transgenic plant, pCGN5511-LP004–5 (black bars), were assayed for LPAAT substrate specificity using 12:0-LPA and various ^{C-labeled acyl-CoAs. PA, Phosphatitic acid.}

Expression of CLP in Canola Results in Incorporation of Laurate into the sn-2 Position of TAG

Since the activity of the CLP was highest using 12:0-LPA and 12:0-CoA, and since the activity of the endogenous LPAAT was low, those substrates were used to evaluate CLP activity in maturing seeds of all of the pCGN5511-transformed B. napus LP004 lines. All of the transgenic lines showed increased 12:0-LPAAT activity compared with control (x axis of Fig. ​Fig.2).2). However, since these transformants produce no 12:0-CoA, it was not possible to directly observe the effects of the CLP in planta on altered oil composition. To evaluate CLP in a laurate-containing background, each pCGN5511 primary transformant was crossed with DH22, a homozygous BTE line that contains 51 mol % laurate. Since the BTE parent was homozygous, each resulting F₁ seed should have a complement of BTE alleles. As predicted, all F₁ seeds from these crosses contained laurate, albeit at a lower level then the homozygous parent. Seed-to-seed laurate levels ranged from 30% to 40% (not shown). Since the CLP parents were the primary hemizygous transformants, CLP alleles were expected to segregate in the F₁ population and, depending on the genetic loci of the CLP in the parents, up to 50% of these seeds were expected not to contain any CLP allele (nulls).

Figure 2

Correlation of 12:0-LPAAT activity with 12:0 accumulation at sn-2. LPAAT activity using 12:0-CoA and 12:0-LPA was determined in membrane fractions derived from developing untransformed, control canola seeds, as well as from independent CLP LP004 transformants. In addition, the proportion of 12:0 at sn-2 was measured in 20-seed pools of mature F₁ seeds derived from crosses of the same set of CLP plants with a homozygous BTE-containing line, DH22. The figure correlates these two determinations. ♦, Control; ×, individual CLP × BTE F₁ seed lots. PA, Phosphatitic acid.

Twenty F₁ seeds from each of these crosses were pooled and the sn-2 fatty acid composition determined. These values are shown plotted against the 12:0 LPAAT activity, as previously determined with the respective primary CLP transformants (Fig. ​(Fig.2).2). In all cases, laurate levels at sn-2 were greater than 5%, with maximum levels at 30%. Canola oil from BTE-only transformants with 30% to 40% laurate contains less than 5% laurate at the sn-2 position (Voelker et al., 1996). Clearly, the combination of BTE and CLP increased this value, in some cases drastically, demonstrating the efficient competition of CLP with the canola LPAAT in planta. The plot indicates only a weak correlation between 12:0-LPAAT activity of the CLP parent and the resulting invasion of 12:0 into sn-2 in F₁ seeds. We attribute this imperfect correlation to several factors. The LPAAT assay is only semiquantitative, and a variability of up to 2-fold has been observed when repeatedly assaying developing seeds from the same transformants. This was compounded by the fact that the respective maturing seed pools were segregating for the respective LPAAT alleles (homozygous, heterozygous, and null seeds), and we could not ensure that we always had a representative sample.

Coconut LPAAT Induces the Accumulation of Trilaurin

The large amounts of laurate produced by the BTE in developing canola seeds are deposited almost exclusively in the sn1 and sn3 positions of TAGs. Even at 47 mol % laurate, trilaurin represents only 2.68 weight % of all TAGs (Voelker et al., 1996). We wanted to study whether the laurate invasion into sn-2 catalyzed by CLP boosted the trilaurin fraction.

In a first experiment, F₁ plants from crosses of several LPAAT transformants with the 50 mol % laurate-containing BTE homozygous line DH22 were grown and CLP-homozygous F₂ lines selected in the next generation. Since the laurate trait was carried by several loci in the BTE parent, we observed a wide range of laurate levels in F₂ plants. Seeds from canola lines harboring the BTE alone served as controls. Extracted oil was analyzed by reversed-phase HPLC. Under these conditions, TAGs with an equal total number of unsaturations in their fatty acyl moieties migrated with almost the same mobility, which was nearly independent of the total carbon number (Hammond, 1993; Voelker et al., 1996).

In Figure ​Figure33 the overall laurate proportion of the respective oil samples is plotted against their trilaurin fraction. As reported earlier, a BTE transformant line with 47 mol % of laurate accumulated only a few percent of TAGs as trilaurin (Voelker et al., 1996). This was not unexpected, since the canola LPAAT does not accept 12:0-CoA (Bafor et al., 1990). However, when CLP is present, the trilaurin fraction at a given laurate level is drastically higher. Even at 30% of total laurate, more than 5% of the TAGs are trilaurin. The trilaurin fraction increased to up to 15% when total laurate was at 50%. In summary, at all laurate levels measured, the trilaurin fraction was significantly higher with the CLP present than what is found in lines harboring BTE alone; it was even higher than calculated values that assume random distribution for laurate at all three glycerol positions. This result demonstrated that CLP is active in vivo and that during TAG biosynthesis it facilitates the entry of a significant portion of laurate into position 2. Since the seeds look normal and germinate, it also demonstrates that the invasion of laurate into the sn-2 position of glycerolipids is not harmful to the canola seeds.

Figure 3

Relationship between total laurate content and trilaurin. Seed oil was extracted from transformed canola plants and the total laurate content of the oil and proportion of trilaurin compared with other TAGs in the oil were determined. Each symbol reflects a sample derived from a seed pool from one dihaploid plant as described. ×, Canola lines transformed with BTE alone; ○, seeds from F₂ plants resulting from crosses of several different Q04 CLP transformants with the homozygous BTE plant DH22. The line was calculated by assuming that laurate was positioned randomly at all three positions of the triglyceride.

CLP Can Channel Laurate Preferentially to sn-2

Since we had previously found that measuring LPAAT activity in maturing seeds of primary transformants was a poor predictor of induced phenotype (Fig. ​(Fig.2),2), we initiated a second transformation (this time in a conventional canola variety with higher agronomic yield than QO4) and screened the resulting CLP transformants by measuring the in vivo impact on triglyceride assembly. To achieve this goal, all primary CLP transformants were crossed to a homozygous BTE plant. Assuming Mendelian segregation, we could predict that all resulting F₁ seeds had an identical BTE gene dosage, namely one copy of each BTE parent locus. Because of the nature of the plant transformation, all CLP parents were hemizygous for CLP, and therefore the CLP transgene should segregate in the cross according to locus number. In the resulting F₁ seeds, the impact of the different CLP transgenic alleles on laurate deposition at sn-2 was measured. Since the direct determination of laurate at sn-2 in single cotyledons was not feasible, we choose to measure trilaurin levels. F₁ seeds with highest levels of trilaurin were considered to harbor the highest CLP-expression alleles and were selected for the establishment of the BTE/CLP lines.

For this experiment, the canola var Quantum was transformed with pCGN5511, and 20 independent transformants were generated. Each original transformant was crossed with the homozygous BTE line DH63, which featured 59 mol % laurate in the seeds. In this BTE line the laurate phenotype was caused by three or four genetic loci (not shown). After seed maturation, one cotyledon each of several F₁ seeds from each cross was analyzed for trilaurin and the total laurate fraction. Total laurate levels were between 50 mol % and 60 mol %, and the trilaurin fraction ranged from very low to up to 25% (not shown). For further study, we selected F₁ lines that harbored the coconut LPAAT at one genetic locus and had induced the highest trilaurin levels in single F₁ seeds.

Since in such lines BTE and CLP together were expected to segregate in four to five genetic loci, breeding to complete homozygosity would have required a multigeneration breeding program. For a more efficient strategy, microspores were grown from the pollen of F₁ plants, and approximately 100 dihaploid plants were generated (Eickenberry, 1994). Since genetic segregation during the generation of dihaploid plants is at the haploid level, the probability for any given combination of four loci is 1 in 32. We therefore calculated that 100 independent dihaploids should contain all possible combinations of these four loci with a high probability. Because of the nature of dihaploid plants, they are exclusively in the homozygous state.

In the resulting population of 100 dihaploid plants, the CLP transgene was detected via PCR of leaf tissue extracts. In addition, the total fatty acid composition was determined in mature seed pools. The laurate fraction ranged from 0 mol % to 67 mol %. We also analyzed the proportion of laurate at sn-2 of the seed oil TAGs in a subset of dihaploids. In Figure ​Figure44 the total laurate levels for each selected line is plotted against the laurate at sn-2 for both populations. The graph demonstrates that in BTE-only plants with less than 45 mol % total laurate, very little laurate accumulates at sn-2, as was observed for the two lines in Voelker et al. (1996). However, at laurate levels above 45 mol %, sn-2 laurate unexpectedly rose rapidly to significant levels. When CLP was present, however, laurate sn-2 infusion was much higher at any given total laurate level. In the very highest laurate plants, laurate makes up to 75% of the acyl groups in this position, indicating that in this plant sn-2 is the preferred position for this saturated fatty acid.

Figure 4

Laurate proportion at sn-2 is dependent on total laurate levels and coconut LPAAT. The primary CLP transformants (pCGN5511 in var Quantum) were crossed with the homozygous BTE line DH63. A F₁ plant harboring CLP and BTE alleles was grown. Independently segregating F₂ microspores derived from this plant were made diploid and grown into (homozygous) dihaploid plants as described. The presence or absence of the coconut LPAAT gene in the individual dihaploid plants was determined via PCR of leaf tissue. All plants were selfed, and oil was extracted from the resulting seeds. The sn-2 analysis of seed oil was executed using R. arrhizus lipase. Each symbol represents a seed pool derived from one dihaploid plant. ○, CLP-positive plant; ×, CLP-negative plant. For this analysis, we selected the top 12 laurate producers of the generated dihaploids, as well as randomly chosen plants throughout the laurate range.

CLP Can Boost Laurate Levels in Seeds

There are several lines of evidence indicating that in developing seeds of BTE canola that accumulate more than 45% laurate, a fraction of newly synthesized laurate might enter a futile cycle during seed development (Voelker et al., 1996; Eccleston and Ohlrogge, 1998). It was reasoned that in such seeds, since the sn-2 position in the TAG could not be used efficiently and the sn-1 and sn-3 positions were nearly filled with laurate, a fraction of the laurate output of the plastid could not be utilized and was subsequently degraded. The introduction of the CLP was therefore predicted to not only allow access of laurate to position sn-2, but also to increase total laurate levels when expressed in very high laurate lines.

To test this hypothesis, we determined the absence or presence of the coconut LPAAT gene in all 100 dihaploid plants used in this study. As predicted by the underlying genetics, the dihaploid population was split equally into CLP positives and CLP negatives. In Figure ​Figure5,5, the data are sorted by total laurate levels and by the absence or presence of CLP. Plants without CLP have oil with up to 59 mol % laurate, and the BTE parent used for the cross had 59 mol % of laurate; therefore, it was expected that a dihaploid (homozygous) plant with all of the BTE loci reassembled after the cross should also have about the same laurate levels. However, plants with the coconut LPAAT gene had up to 67 mol % laurate. Altogether, we obtained about 12 BTE/LPAAT plants that had more laurate than any of their BTE-only siblings (up to almost 10% more). It is also interesting to note that a disproportionate number of BTE plants had laurate levels between 56 mol % and 58 mol % laurate. Detailed genomic analysis revealed no trend of higher BTE gene dosage in CLP-positive versus CLP-negative plants (not shown).

Figure 5

Coconut LPAAT can boost laurate levels. Each symbol represents a seed-pool analysis of an individual dihaploid plant. All dihaploid plants resulting from crosses described in Figure ​Figure44 were sorted via PCR into a CLP-containing (CLP +) and a CLP-free (CLP −) populations. Plants of both populations were grouped into 1% laurate intervals.

We conclude that the presence of CLP allows a more efficient laurate TAG deposition in very high BTE-expressing plants. We also wanted to determine whether the presence of CLP leads to more total lipid per seed. Therefore, we determined the total oil as a percentage of dry weight in mature seeds of all dihaploid plants. In Figure ​Figure6,6, these data are shown correlated with laurate levels. It is obvious that there was wide plant-to-plant variability, which is common for greenhouse-grown plants, but for both subgroups the distribution is similar. The average oil percentage for plants with laurate levels less than 40 mol % laurate was 34% to 35%, but the average of plants with more than 50 mol % laurate was only 30% to 32%.

Figure 6

Correlation of oil levels with BTE and CLP. The total seed oil mass as a percentage of dry weight of the dihaploid plants described in Figure ​Figure44 were determined by NMR. The data are shown separately for CLP-free (CLP−) and CLP-containing (CLP+) populations, with the oil percentage plotted against total laurate levels.