Castor LPCAT and PDAT1A Act in Concert to Promote Transacylation of Hydroxy-Fatty Acid onto Triacylglycerol.

Oilseeds produce abundant triacylglycerol (TAG) during seed maturation, to fuel the establishment of photoautotrophism in the subsequent generation. Commonly, TAG contains 18-carbon polyunsaturated fatty acids (FA), but plants also produce oils with unique chemical properties highly desirable for industrial processes. Unfortunately, plants that produce such oils are poorly suited to agronomic exploitation, leading to a desire to reconstitute novel oil biosynthesis in crop plants. Here, we studied the production and incorporation of hydroxy-FA (HFA) onto TAG in Arabidopsis (Arabidopsis thaliana) plants expressing the castor (Ricinus communis) FAH12 hydroxylase. One factor limiting HFA accumulation in these plants is the inefficient removal of HFA from the site of synthesis on phosphatidylcholine (PC). In Arabidopsis, lysophosphatidic acid acyltransferase (LPCAT) cycles FA to and from PC for modification. We reasoned that the castor LPCAT (RcLPCAT) would preferentially remove HFA from PC resulting in greater incorporation onto TAG. However, expressing RcLPCAT in Arabidopsis expressing FAH12 alone (line CL37) or together with castor acyl:CoA:diacylglycerol acyltransferase2 reduced HFA and total oil yield. Detailed analysis indicated that RcLPCAT reduced the removal of HFA from PC, possibly by competing with the endogenous LPCAT isozymes. Significantly, co-expressing RcLPCAT with castor phospholipid:diacylglycerol acyltransferase increased novel FA and total oil content, by transferring HFA from PC to diacylglycerol. Our results demonstrate that a detailed understanding is required to engineer modified FA production in oilseeds and suggest that phospholipase A2 enzymes rather than LPCAT mediate the highly efficient removal of HFA from PC in castor seeds.


58
Seeds are the most common form of plant reproduction and supply the necessary reserves 59 to sustain seedlings until they can establish photoautotrophic growth. A critical carbon reserve in 60 seeds of many plant species is oil in the form of triacylglycerol (TAG). Oils from crops like 61 canola (Brassica napus) and soybean (Glycine max) contain fatty acid (FA) that is a crucial 62 source of human and animal nutrition. Some plant species produce seed oils that contain unique 63 bond structures, or side groups that impart unique properties valued in a wide array of industries 64 fluorescence, we identified 30 independent T1 transformants and grew them to maturity. To 149 determine seed fatty acid composition, we selected red T2 seeds, generated fatty acid methyl 150 esters (FAME), and analyzed these by gas chromatography (GC). Compared to the parental 151 CL37 line, transformation with RcLPCAT reduced the proportion of HFA in nearly every case 152 (Supplemental Fig. S1A). These CL37:RcLPCAT lines were designated LPT. We cultivated 153 three independent lines (LPT#4, LPT#15, and LPT#29) with low HFA levels and whose marker 154 segregation ratio in T2 seed (approximately 3red:1brown) indicated a single genomic insertion 155 site. During the cultivation of homozygous T3 plants, we collected developing seed and 156 confirmed strong RcLPCAT expression using RT-qPCR (Supplemental Fig. S1B (Fig. 2B). In contrast to these 222 decreases, line PD_LPT showed higher HFA-TAG accumulation at each time point, averaging 223 25.7±0.3% at 12 DAF, 25% higher than the 20.6±0.6% HFA in TAG found in the seed of PD 224 than CL37, LPT seeds surprisingly remained shrunken and misshapen. Seeds of DG2 and DG2_LPT appeared misshapen only rarely, with many plump seeds similar in size and 237 appearance to seed of fae1. Seeds from PD were often misshapen, while RcLPCAT expression in 238 PD_LPT relieved this phenotype so that PD_LPT seeds were largely indistinguishable from fae1 239 ( Fig. 3A). To quantify the differences in oil content, we generated FAME from mature seed, We analyzed the molecular species of HFA-TAG from mature seed in each RcLPCAT 251 expression line. We used TLC to separate TAG bands corresponding to 0-, 1-, 2-HFA moieties 252 for FAME analysis (Supplemental Fig. S4), using an internal standard to determine the 253 percentage contribution of each TAG species. We discounted 3-HFA TAG from our calculation 254 as it accounted for less than 1% of the total TAG in each line. The seed of LPT contained 255 66.7±1.7% 0-HFA-TAG, a 30% increase over the 51.5±2.3% found in CL37 seed. The increased 256 0-HFA-TAG accompanied a 39% reduction in 1-HFA-TAG from 40.0±1.5% in CL37 to 257 24.6±2.8% in LPT, but no significant change in 2-HFA-TAG (Fig. 4A). Expressing RcLPCAT in 258 DG2 also significantly increased 0-HFA TAG by 68%, from 35.5±2.7% to 59.5±0.3%. This 259 accompanied a 34% reduction in 1-HFA-TAG from 39.2±3.6% to 25.7±1.6%, as well as a 39% 260 reduction in 2-HFA-TAG from 24.6±1.8% to 14.9±1.1% (Fig. 4B). The TAG species in PD and 261 PD_LPT seed revealed a different pattern with 0-HFA TAG declining 19%, from 40.8±1.6% to 262 32.9±2.3%. Paired with this decline in 0-HFA TAG was a 28% increase in 1-HFA TAG from 263 38.7±0.5% to 53.7±2.1%; there was no significant change in 2-HFA TAG (Fig. 4C). These data 264 imply that RcLPCAT expression decreases HFA availability for TAG synthesis in the LPT and 265 DG2_LPT lines but increases it in PD_LPT.

Altered Stereochemical Distribution of HFA in 1-HFA TAG 267
The altered levels of 1-HFA-TAG in our three RcLPCAT expression lines implied a shift 268 in the balance of HFA incorporation through different TAG biosynthesis pathways. Since 269 acyltransferases incorporate acyl groups at a specific location, the distribution of HFA moieties 13 causes feedback inhibition of de novo fatty acid synthesis (Bates et al., 2011). In the PD_LPT 296 line, RcPDAT1A would foster an alternative route for removal of HFA from PC, potentially 297 allowing for the increased oil content we observed in this line relative to the PD parental line. 298 Using developing seeds from our RcLPCAT expression lines, we determined the 299 proportion of HFA on PC at 8, 10, 12, and 14 DAF (Fig. 6). We extracted lipids from these seeds 300 and separated the PC fraction using TLC, followed by FAME analysis. Levels of HFA-PC in 301 CL37 rose continually through the time course, reaching 10.4±0.5% at 14 DAF. In LPT, the 302 proportion of HFA-PC was higher throughout the time course totaling 12.2±0.4% at 14 DAF 303 (Fig. 6A). Similarly, DG2_LPT contained significantly higher HFA in PC than the parental DG 304 line, reaching 12.1±0.0% at 14 DAF compared to 9.7±0.3% in DG2 (Fig. 6B). However, the 305 results for PD and PD_LPT are similar at 8 and 10 DAF and then show less HFA in the PC of 306 PD_LPT. The HFA in PC at 14 DAF averaged 6.4±0.4% for PD_LPT, a reduction of 24% 307 relative to the 8.4±0.1% for PD seeds (Fig. 6C). Collectively, these results are consistent with 308 RcLPCAT expression reducing rather than increasing the flux of HFA from PC into HFA-CoA. Considering the previous discoveries, we believed that expressing the castor LPCAT 326 homolog in hydroxy-accumulating Arabidopsis seeds might enhance the release HFA from PC 327 into the acyl-CoA pool, making it available for TAG synthesis by enzymes of the Kennedy 328 pathway. Further, the efficient removal of HFA from PC by RcLPCAT would also increase oil 329 production by alleviating feedback inhibition of FA synthesis. Surprisingly, the expression of 330 RcLPCAT in both CL37 and DG2 backgrounds markedly reduced both HFA content (Fig. 1A,  331 B) and HFA accumulation in TAG throughout seed development ( Fig. 2A, B). In contrast to 332 these results, RcLPCAT when co-expressed with RcPDAT1A elevated the rate of HFA 333 accumulation and the final proportion of HFA in the seed oil, relative to PD seeds (Figs. 1C and 334 2C), indicating concerted action between these two enzymes. 335 Interestingly, the reduced HFA levels in LPT (CL37_RcLPCAT) and DG2_LPT 336 (CL7_RcDGAT2_RcLPCAT) lines were paired with lower total oil content, relative to CL37 337 and DG2, respectively (Fig. 3). By contrast, PD_LPT (CL37_RcPDAT1A_RcLPCAT) plants 338 had both a higher proportion of HFA and increased total oil content compared to the PD control 339 PC than in the corresponding parental line throughout oil filling (Fig. 6A and B), and in reduced oil contents ( Fig. 3B and C). In contrast, the co-expression of RcPDAT1A in PD_LPT produced 357 an increase in seed oil content relative to the PD line (Fig. 3D) coupled with some reduction in 358 HFA-PC at 12 and 14 DAF (Fig. 6C). These results indicate that HFA in PC is increased upon 359 RcLPCAT expression, resulting in lower yield and exacerbating the metabolic bottleneck and   We dispersed Arabidopsis (Arabidopsis thaliana) seeds on plates with half-strength 431 Murashige and Skoog nutrients (Sigma), 0.75% (w/v) agar, and 1% (w/v) sucrose at pH 5.7. 432 After growth under 100 µmol m -2 s -1 continuous light at 22°C for 12 days. We then transferred 433 plantlets to the soil and continued cultivation until senescence under 120 µmol m -2 s -1 continuous 434 white light from broad-spectrum fluorescent lamps at 22°C with 70% humidity. 435

Statistical Analyses. 436
We analyzed data with GraphPad Prism (GraphPad Software, La Jolla, CA) using either 437 one-way ANOVA with post hoc Tukey or t-test. 438

Cloning and transformation of RcLPCAT. 439
We amplified the RcLPCAT coding sequence from a castor developing seed cDNA 440 library with KOD hot-start polymerase (Novagen) using primers designed to append EcoR1 and 441 Xho1 restriction sites. Using the manufacturer's instructions, we ligated isolated amplicons into 442 ZeroBlunt D-TOPO (Invitrogen). We compared the sequence of a single clone against the 443 reference genome available at http://blast.jcvi.org/er-blast/index.cgi?project=rca1 for fidelity. 444 We liberated the EcoR1-RcLPCAT-Xho1 fragment and ligated it into the same sites in We determined fatty acid proportions, seed oil, and composition of TLC samples by the 465 preparation of FAME followed by GC analysis. Quantification of FAME by GC used flame 466 ionization detection from a wax column (EC Wax; 30m X 0.53 i.d X 1.20µm; Alltech) with 467 parameters of 210°C for 1 minute followed by a ramp to 250°C at 10°C -1 , with a final 9-minute 468 temperature hold. The comparative analysis of FA composition used 15 to 20 µg of seed from 469 each line. We measured oil content by co-derivatization of a 17:0 TAG standard and seed of 470 known weight. Calculation of total oil used the ratio of oil to 17:0 TAG added before derivation 471 and normalization to the seed sample weight.

Accumulation of HFA-PC and HFA-TAG during seed development. 473
We cultivated plant lines as above until the first six siliques emerged. We then scored 474 each subsequent silique at the first visible white tinge as day 1. Counting continued until each 475 independent replicate accounted for 100 siliques at 8, 10, 12, and 14 DAF. We extracted seeds by 476 immersing siliques in liquid nitrogen followed by rapid thawing to break open the wall (Bates et 477 al., 2013). We extracted lipids from the seeds and separated TAG and PC using thin-layer 478 chromatography with a single development in We separated extracted lipids from 50 mg of dried mature seed and isolated bulk TAG 504 using thin layer chromatography in a single development, using a solvent system of 505 chloroform:acetone:acetic acid (96:3.5:0.5, v/v/v). Bulk TAGs were collected and eluted from 506 the silica with two washes of chloroform:methanol (4:1, v/v). We induced phase separation with 507 methanol (2 mL) and 0.88% (w/v) KCl (4 mL), followed by 2-minute centrifugation at 4383g. 508 The chloroform layer was collected and a single back-extraction with chloroform (5 mL) 509 conducted. We removed chloroform by drying the samples under nitrogen and resuspending in 510 toluene containing 0.05% (w/v) butylated hydroxytoluene (0.5 mL). We dried lipid samples of 511 50 µL under nitrogen and resuspended in diethyl ether (1 mL), 0.8 mL buffer (50 mM sodium 512 bromide, pH 7.6, and 5mM calcium chloride) in the presence of Rhizomucor miehei lipase 513 (Sigma). The reaction proceeded for 40 min under constant vortex mixing and was quenched 514 with 2 mL methanol:chloroform. The chloroform layer was collected and lipids separated by 515 TLC with a single development in a solvent system of chloroform:methanol:acetic acid 516 (97.5:2:0.5, v/v/v). We vacuum dried the plate and stained with 0.005% (w/v) primulin in 80% 517