Fatty acid synthesis is inhibited by inefficient

Fatty acid synthesis is inhibited by inefficient utilization of unusual fatty acids for glycerolipid assembly Philip D. Batesa,b,1, Sean R. Johnsonb, Xia Caoc, Jia Lid, Jeong-Won Namd, Jan G. Jaworskid, John B. Ohlroggec, and John Browseb a Department of Chemistry and Biochemistry, The University of Southern Mississippi, Hattiesburg, MS 39402; bInstitute of Biological Chemistry, Washington State University, Clark Hall, Pullman, WA 99164; cDepartment of Plant Biology, Michigan State University, MI 48824; and dDonald Danforth Plant Science Center, St. Louis, MO 63132

Edited by Chris R. Somerville, University of California, Berkeley, Berkeley, CA, and approved December 13, 2013 (received for review October 1, 2013)

Degradation of unusual fatty acids through β-oxidation within transgenic plants has long been hypothesized as a major factor limiting the production of industrially useful unusual fatty acids in seed oils. Arabidopsis seeds expressing the castor fatty acid hydroxylase accumulate hydroxylated fatty acids up to 17% of total fatty acids in seed triacylglycerols; however, total seed oil is also reduced up to 50%. Investigations into the cause of the reduced oil phenotype through in vivo [14C]acetate and [3H]2O metabolic labeling of developing seeds surprisingly revealed that the rate of de novo fatty acid synthesis within the transgenic seeds was approximately half that of control seeds. RNAseq analysis indicated no changes in expression of fatty acid synthesis genes in hydroxylase-expressing plants. However, differential [14C]acetate and [14C]malonate metabolic labeling of hydroxylaseexpressing seeds indicated the in vivo acetyl–CoA carboxylase activity was reduced to approximately half that of control seeds. Therefore, the reduction of oil content in the transgenic seeds is consistent with reduced de novo fatty acid synthesis in the plastid rather than fatty acid degradation. Intriguingly, the coexpression of triacylglycerol synthesis isozymes from castor along with the fatty acid hydroxylase alleviated the reduced acetyl–CoA carboxylase activity, restored the rate of fatty acid synthesis, and the accumulation of seed oil was substantially recovered. Together these results suggest a previously unidentified mechanism that detects inefficient utilization of unusual fatty acids within the endoplasmic reticulum and activates an endogenous pathway for posttranslational reduction of fatty acid synthesis within the plastid. β-oxidation

understanding of mechanisms that control seed FA synthesis and accumulation. The net accumulation of a metabolic product is controlled by the combined action of anabolic and catabolic pathways. The FAs that accumulate within TAG are initially synthesized up to 18C and 0–1 double bonds within the plastid. Upon exiting the plastid, newly synthesized FAs may be further modified (desaturated, hydroxylated, etc.) while esterifed to endoplasmic reticulum (ER) membrane lipid phosphatidylcholine (PC) before incorporation into TAG (12, 13). FAs esterifed to glycerolipids have long half-lives (14), with minimal turnover in most tissues (15). A prominent exception takes place in germinating seedlings where TAG is broken down through β-oxidation to produce acetyl–CoA for energy production and gluconeogenesis (16). In preparation for germination, enzymes for TAG degradation accumulate during seed development and lead to a loss of ∼10% of seed oil reserves during late seed maturation (17). Thus, oil levels of mature seeds result from a combination of both FA synthesis and FA catabolism, and an alteration of either process could lead to the reduced oil phenotypes of some transgenic oilseeds. The selective breakdown of unusual FAs within transgenic plants has long been suggested as a major factor limiting production of oilseed crops containing industrial oils (12, 18). Multiple lines of evidence support this hypothesis. The castor (Ricinus communis) Significance Many plants produce valuable fatty acids in seed oils that provide renewable alternatives to petrochemicals for production of lubricants, coatings, or polymers. However, most plants producing these unusual fatty acids are unsuitable as crops. Metabolic engineering of oilseed crops, or model species, to produce the high-value unusual fatty acids has produced only low yields of the desired products, and previous research has indicated fatty acid degradation as a potential major factor hindering oilseed engineering. By contrast, we here present evidence that inefficient utilization of unusual fatty acids within the endoplasmic reticulum can induce posttranslational inhibition of acetyl–CoA carboxylase activity in the plastid, thus inhibiting fatty acid synthesis and total oil accumulation.

| feedback inhibition | metabolic engineering

F

atty acids (FAs) that accumulate as triacylglycerols (TAGs) in seeds of plants represent a major source of renewable reduced carbon that can be used as food, fuel, or industrial feedstocks. Within the plant kingdom there are greater than 300 different types of “unusual FAs” that contain functional groups (e.g., hydroxy, epoxy, and cyclopropane) or have physical properties useful for replacing petroleum in the chemical industry (1, 2). Unfortunately, most plants which naturally produce these unusual FAs have agronomic features which make them unsuitable as major crops. Over the past 2 decades, most attempts to genetically engineer unusual FAs into oilseed crops or model species have produced only low proportions of the desired FA within TAG (2–5). Additionally, in many cases, accumulation of unusual FAs in transgenic plants is accompanied by a reduction of total seed oil (6–11); in some instances reductions of up to 50% of total seed oil have been reported (7, 10). The endogenous mechanisms that recognize and respond to unusual FAs and result in reduced seed oil accumulation in transgenic plants are unknown. These limited successes and adverse outcomes of oilseed engineering highlight our lack of knowledge on how plants accumulate TAG and indicate a need for better

1204–1209 | PNAS | January 21, 2014 | vol. 111 | no. 3

Author contributions: P.D.B. and J.B. designed research; P.D.B., S.R.J., J.L., and J.-W.N. performed research; P.D.B., S.R.J., X.C., J.L., J.-W.N., J.G.J., J.B.O., and J.B. analyzed data; and P.D.B., J.B.O., and J.B. wrote the paper. The authors declare no conflict of interest. This article is a PNAS Direct Submission. Data deposition: The RNAseq data reported in this paper have been deposited in the Gene Expression Omnibus (GEO) database, www.ncbi.nlm.nih.gov/geo. 1

To whom correspondence should be addressed at the present address: Department of Chemistry and Biochemistry, The University of Southern Mississippi, Hattiesburg, MS 39406. E-mail: [email protected].

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. 1073/pnas.1318511111/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1318511111

Bates et al.

Results Reduced Oil Accumulation in Plants Producing HFA. We initially set

out to quantify the effect of HFA β-oxidation on mature seed oil levels and the accumulation of TAG during seed development. Fig. 1 displays the oil content of wild-type and transgenic Arabidopsis seeds producing HFA. There is no difference in oil content between wild-type Col-0 and the fatty acid elongation 1 (fae1) mutant (26) which is the background for the transgenic lines. Two independent transformations of fae1 with the castor FA hydroxylase using a seed-specific promoter (CL37 and CL7) each contain ∼17% HFA within seed oil (20). Both hydroxylase-expressing lines have an ∼50% reduction in total FAs per seed (Fig. 1A) as well as a reduction in oil content as a percent of total seed weight (Fig. 1B). In an attempt to increase the proportion of HFA in TAG HFA-selective TAG-synthesis enzymes from castor [phospholipid: diacylglycerol acyltransferase (PDAT) and acyl-CoA:diacylglycerol acyltransferase (DGAT)] were coexpressed with the FA hydroxylase within the CL37 and CL7 backgrounds, respectively (9, 25). The total micrograms of HFA per seed in the PDAT and DGAT lines was doubled (Fig. 1A), indicating a more efficient incorporation of HFA into TAG. However, HFA content as a percent of total FA only increased up to ∼25% due to an even larger increase in the micrograms of normal FA (9). In the PDAT and DGAT lines the oil content of the seeds is recovered to 75 and 85% of the control, respectively (Fig. 1A). We also measured the net accumulation of TAG during the stage of rapid oil accumulation within developing seeds (SI Appendix, Fig. S2). The net rate of TAG accumulation over seed development of CL37 was approximately half the rate of accumulation in the fae1 background line (SI Appendix, Fig. S2 A and B). In contrast, the PDAT and DGAT lines accumulated TAG at ∼0.7 and 0.9 the net rate of fae1, respectively (SI Appendix, Fig. S2 C and D). The correlation between reduced CL37 seed oil levels (Fig. 1) and net rates of TAG accumulation over seed development (SI Appendix, Fig. S2) suggests that the reduced seed oil content of CL37 versus fae1 is mostly due to less TAG accumulation during seed development, rather than increased TAG breakdown during late seed maturation. FA Synthesis Rate Is Reduced When HFA Are Inefficiently Incorporated into TAG. The oil content and net rate of TAG accumulation

of CL37 was approximately half of that in the fae1 control. However, this does not necessarily imply that half of the FAs synthesized in CL37 are broken down through β-oxidation. Previously, β-oxidation induced by unusual FA production or FA synthesis mutants has led to increased levels of FA synthesis to replenish cellular FA content (21, 27). Therefore, we hypothesized that the rate of FA synthesis within CL37 may also be

A

B

Fig. 1. Oil content of wild-type and transgenic Arabidopsis lines. (A) Micrograms of total FA per seed. (B) Oil content calculated as FA percentage of seed weight (seed weights are given in SI Appendix, Fig. S1). CL37 and CL7 express the castor fatty acid hydroxylase (RcFAH12) in the fae1 background . PDAT contains RcPDAT1a in CL37, and DGAT contains RcDGAT2 in CL7 . Data are mean ± SEM of 15–18 replicates for each plant line.

PNAS | January 21, 2014 | vol. 111 | no. 3 | 1205

PLANT BIOLOGY

FA hydroxylase which produces hydroxylated FAs (HFA) and the California bay medium-chain acyl–acyl carrier protein thioesterase (MCTE) which produces 12:0 (FA nomenclature, # carbons: # double bonds) have been constitutively expressed in tobacco and Brassica napus, respectively (12, 19). In each case, small amounts of the unusual FAs were found in the seeds but not in the leaves. Biochemical analysis of MCTE leaves indicated high MCTE activity and 12:0 production, but no accumulation (19). Thus, the lack of unusual FA accumulation in leaves suggests that unusual FAs are rapidly degraded after synthesis. The accumulation of HFA and 12:0 to significant levels within seed TAG of transgenic plants has been achieved by the use of strong seed-specific promoters (18, 20, 21). Sequestration of unusual FAs in TAG of transgenic plants may limit their adverse effects on membranes and thus allow accumulation in seeds. However, some evidence suggests that unusual FAs are also broken down by β-oxidation in developing seeds. Degradation of HFA through β-oxidation was indicated in the seeds of Arabidopsis plants coexpressing the castor FA hydroxylase along with a bacterial polyhydroxyalkanoate (PHA) synthase, which produces PHA from intermediates of β-oxidation (22). However, it is unclear if the PHA accumulation in the transgenic seeds was due to β-oxidation during the TAG accumulation phase of seed development or during the TAG breakdown phase of late seed maturation. B. napus embryos accumulating 12:0 up to 60% of seed FA had increased levels of a 12:0 specific β-oxidation activity. The substantial rates of β-oxidation during seed development did not reduce total oil accumulation in these seeds, implying that a concomitant increase in FA synthesis compensated for a futile cycle of 12:0 synthesis/degradation (21). Together these results demonstrate that even though seed tissue can accumulate unusual FAs within glycerolipids, the proportion and content of unusual FAs within TAG may still be limited by selective β-oxidation of the unusual FAs. The apparent increased FA synthesis that is proposed to offset unusual FA degradation makes it unclear how production of HFA or epoxy–FA has led to substantially reduced levels of seed oil in multiple transgenic plants (8–11). To better understand the metabolic processes that control TAG accumulation in seeds we investigated the basis of reduced oil accumulation in Arabidopsis plants expressing the castor FA hydroxylase (12). Castor seeds accumulate oil containing ∼90% HFA. However, heterologous expression of the FA hydroxylase in different plants has led to a relatively low proportion of HFA in the oil of transgenic seeds, with the best lines producing ∼17% HFA in TAG (9, 18, 20, 23, 24). The accumulation of HFA within transgenic Arabidopsis seeds is also accompanied by a 30–50% reduction in total seed oil (8–10). Interestingly, endeavors to circumvent the mechanisms limiting the proportion of HFA within TAG through coexpression of the castor FA hydroxylase with HFA-selective TAG-synthesis enzymes increased not only the proportion of HFA in the TAG (to over 25%) but also mostly recovered the seed oil content to that of nontransgenic lines (9, 25). These previous results suggest that inefficient utilization of unusual FAs leads to the reduced seed oil levels. However, it is unknown if the reduced oil content is due to increased catabolism of the unusual FAs through β-oxidation, impairment of FA synthesis, or both. Here we report that production of HFA within the ER of developing Arabidopsis seeds by castor FA hydroxylase expression alone induced a large reduction in FA synthesis and seed oil content, apparently by posttranslational inhibition of plastid localized acetyl–CoA carboxylase (ACCase) activity. Our results indicate that the reduction in FA synthesis is a primary mechanism for the reduced seed oil. Additionally, more efficient incorporation of HFA into TAG by HFA-selective TAG-synthesis enzymes alleviates the inhibition of ACCase activity and increases seed oil content. Thus, we demonstrate in vivo that developing oilseeds can detect inefficient glycerolipid metabolism within the ER and respond by posttranslational down-regulation of de novo FA synthesis within the plastid.

increased, and the total amount of CL37 FAs degraded by β-oxidation would be the difference between total FA synthesis and total FA accumulation. To estimate the amount of FA synthesis in fae1 and CL37 developing seeds we measured the in vivo rate of FA synthesis during the phase of TAG accumulation through metabolic labeling. Newly synthesized FAs can be labeled with [14C]acetate in vivo through incorporation of [14C] acetate into the acetyl–CoA substrate for FA synthesis (28). Fig. 2A demonstrates that the rate of [14C]acetate incorporation into both fae1 and CL37 FAs was linear from 0 to 60 min, indicating very little degradation of the newly synthesized FAs over the labeling time period. However, surprisingly, the rate of CL37 FA labeling was approximately half that of fae1. This result is more consistent with reduced FA synthesis in CL37 rather than increased FA β-oxidation. Acetyl–CoA is not only the substrate for FA synthesis but also the product of FA β-oxidation. Therefore, to control against possible differences in acetate metabolism between fae1 and CL37 (such as dilution of [14C]acetate due to high rates of β-oxidation) we also measured the rate of FA synthesis using [3H]2O. Nascent FAs can be labeled with 3H from [3H]2O during the reduction steps of FA synthesis (29). Fig. 2B demonstrates the accumulation of 3H-labeled FAs over 30 min in fae1 and CL37 at three different developmental stages during the period of TAG accumulation in developing seeds. The rate of CL37 FA synthesis was consistently less than half that of fae1 at all developmental stages. Additionally, the rates of FA synthesis in the PDAT and DGAT lines were significantly increased from CL37, but not statistically different from fae1 (Fig. 2B). Together, the metabolic labelings of FA synthesis are consistent with a reduced rate of de novo FA synthesis in CL37 that is responsible for the reduction in seed oil accumulation. Additionally, the more efficient incorporation of HFA into TAG of the PDAT and DGAT lines and the recovered seed oil content (Fig. 1) are consistent with recovered rates of FA synthesis within these lines (Fig. 2B). It is possible that increased β-oxidation activity due to HFA production channels nascent FAs exported from the plastid directly into β-oxidation [similar to the effect of MCTE expression in leaves (19)] and thus produces an apparent reduction in the FA synthesis labeling. However, this scenario is unlikely for two reasons. First, MCTE-expressing leaves produce an incomplete product of FA synthesis (12:0). The 12:0 exported from the plastid is not used for membrane lipid synthesis in the ER (30) and therefore is rapidly degraded. In contrast, the castor FA hydroxylase can only produce HFA after newly synthesized 18:1 is exported from the plastid and incorporated into PC within the ER. The major product of FA synthesis is 18:1 and thus it is unlikely to be selectively degraded. Second, the major flux of nascent FAs exported from the plastid is into PC through acyl editing in Arabidopsis (31, 32) and in other plants (33, 34). Therefore, the removal of HFA from PC and their subsequent β-oxidation would require an increased flux of newly synthesized FAs into PC to maintain membrane integrity. Analysis of nascent [14C]FA incorporation into glycerolipids of fae1 and CL37 (SI Appendix, Fig. S3 and Fig. S3 Discussion), as well as previous analysis of [14C]glycerol labeling of fae1 and CL37 (10), together does

A

B

not indicate a change in PC catabolism or acyl editing (28) as might be expected if there was a large flux of HFA from PC into β-oxidation. Therefore, the reduced rate of [14C]acetate and [3H]2O incorporation into lipids of CL37 compared with fae1 is consistent with a reduced rate of de novo FA synthesis in CL37. Thus, the reduced TAG accumulation in CL37 seeds is more likely due to less FA synthesis rather than increased FA β-oxidation. Expression of FA Synthesis-Related Genes is Not Changed Between the fae1, CL37, and PDAT Lines. TAG deposition within oil-bearing

plant tissues is correlated to large increases in genes involved in FA synthesis including acetyl–CoA carboxylase (ACCase) which is the first committed step in FA synthesis (35–37). We used whole-transcriptome RNAseq to determine if transcriptional regulation of FA synthesis is involved in the altered rates of FA synthesis between fae1, CL37, and PDAT developing seeds. If a particular change in gene expression between fae1 and CL37 is associated with the reduced FA synthesis rates, we would expect to see the opposite expression change between CL37 and PDAT where the rates of FA synthesis and oil levels are recovered. Because FA synthesis gene expression changes temporally in developing seeds we analyzed gene expression at three developmental stages. The fold change in gene expression between fae1 and CL37 or CL37 and PDAT for the four nuclear encoded genes of ACCase is given in Table 1. Surprisingly, there was very little difference in ACCase gene expression between each of the three lines at each developmental stage. Additionally, there was no consistent opposite regulation of FA synthesis genes between CL37/fae1 and PDAT/CL37 as hypothesized. Very similar results were found for all other known genes involved in FA synthesis (Dataset S1). Additionally, there was little change in expression of β-oxidation genes (Dataset S2) between fae1, CL37, and PDAT. Therefore, the large changes in lipid metabolism between fae1 and CL37 as indicated by TAG accumulation (Fig. 1) and metabolic labeling (Fig. 2) are most likely due to posttranscriptional regulation of FA synthesis rather than transcriptional regulation. Acyl–CoA and acyl–ACP Compositions. Recent fatty acid-feeding studies of B. napus embryo-derived microspore cultures indicate that accumulation of oleate (18:1) within the acyl–CoA pool could lead to reduced export of oleate from plastid, accumulation of oleoyl–ACP within the plastid, and product inhibition of FA synthesis by interaction of oleoyl–ACP with ACCase (38). We hypothesized that inefficient utilization of HFA–CoA for TAG synthesis may also cause a backup of oleate export from the plastid and reduced FA synthesis within CL37. Therefore, we measured the acyl–CoA and acyl–ACP compositions of developing fae1, CL37, PDAT, and DGAT lines (Fig. 3). The proportion of oleoyl–CoA was similar in all lines (Fig. 3A), and HFA–CoA accumulated to less than 10% in CL37, similar to levels of HFA in developing castor seeds (39). Additionally, the acyl–ACP compositions of all four lines were very similar (Fig. 3B), and there was no statistically significant change in the acyl– ACP pool sizes (SI Appendix, Fig. S4). Therefore, it is unlikely

C Fig. 2. Rates of FA synthesis in developing Arabidopsis seeds. (A) Time course of [14C]acetate incorporation into total FAs, n = 4. Seed age 9–10 d after flowering. (B) Thirty-minute labeling of FA synthesis (n = 5) with [3H]2O at three developmental stages. (C) Relative [14C] acetate and [14C]malonate incorporation into FAs of developing seeds (n = 5). Total incorporation of each radioisotope is in SI Appendix, Fig. S5. Data are mean of replicates ± SEM.

1206 | www.pnas.org/cgi/doi/10.1073/pnas.1318511111

Bates et al.

CL37 vs. fae1 ACCase subunit 7–8 Days after flowering BC BCCP1 BCCP2 α-CT 9–10 Days after flowering BC BCCP1 BCCP2 α-CT 11–12 Days after flowering BC BCCP1 BCCP2 α-CT

PDAT vs. CL37

Fold change q value Fold change q value −0.10 0.04 0.01 0.05

0.99 1.00 1.00 1.00

−0.37 −0.19 −0.25 −0.24

0.70 0.86 0.82 0.83

0.09 0.07 0.14 0.04

0.98 0.98 0.95 0.99

0.04 0.04 0.04 0.27

0.99 0.99 0.99 0.71

−0.36 −0.26 −0.75 −0.67

0.63 0.74 0.06 0.24

0.12 0.21 0.23 0.16

0.97 0.86 0.86 0.95

Three biological replicates were analyzed at three different stages of development. Fold change (Log2) in gene expression calculated between indicated plant lines is presented. The q value is calculated considering false discovery rates. A q value > 0.05 indicates the fold change in gene expression is not statistically significant. BC, Biotin carboxylase (AT5G35360); BCCP, Biotin carboxyl carrier protein, BCCP1 (AT5G16390), BCCP2 (AT5G15530); α-CT, alpha-carboxyltransferase, (AT2G38040).

that the reduced rates of FAS in CL37 are due to posttranslational regulation of FA synthesis through oleoyl–ACP feedback inhibition of ACCase. ACCase Activity Is Down-Regulated in CL37. Several studies have indicated ACCase as a major posttranslationally regulated enzyme of FA synthesis (38, 40–43). Therefore, to estimate in vivo ACCase activity under the same conditions that indicated changes in the FA synthesis rate we used differential [14C]acetate and [14C]malonate labeling (38) of fae1, CL37, PDAT, and DGAT developing seeds (Fig. 2C). Acetyl–CoA and malonyl–CoA are the substrates and products of ACCase, respectively. A change in ACCase activity will affect the incorporation of [14C]acetate into newly synthesized FAs, but it will not affect the incorporation of [14C]malonate into FAs. Relative to fae1 labeling, the incorporation of [14C]acetate into CL37 FAs was approximately half that of [14C]malonate incorporation into FAs. However, there were no differences between the radiolabeled substrates for the incorporation into PDAT and DGAT FAs, indicating that ACCase activity is reduced in CL37 but not in the PDAT or DGAT lines.

reduced rate of FA synthesis (Fig. 2B), and substantially alleviates the reduced seed oil accumulation (Fig. 1). Similar oil phenotypes have also been reported in soybeans expressing a FA epoxygenase alone and with coexpression of an epoxy–FA selective DGAT (11). Together, these results suggest that plants can detect inefficient glycerolipid synthesis within the ER and can respond by posttranslational down-regulation of ACCase activity, and thus de novo FA synthesis, within the plastid. FA Synthesis Inhibition and β-Oxidation Represent Differential Responses to Unusual FA Production in Transgenic Plants. Pre-

viously, futile cycles of unusual FA synthesis and β-oxidation have been indicated as factors limiting the accumulation of HFA, epoxy–FA, or medium-chain FAs in the seeds or leaves of transgenic plants (19, 21, 22, 44). However, in the previous studies with transgenic plants producing HFA or epoxy–FA, very low levels of the unusual FAs were produced in the seeds (∼6% HFA and ∼3% epoxy–FA), and no report was made on the seed oil levels (22). CL37 seeds accumulate ∼17% HFA and have reduced seed oil (Fig. 1), suggesting that higher levels of unusual FAs may induce different changes in seed metabolism than do low levels. At low levels, the HFA or epoxy–FA that are inefficiently used for ER glycerolipid synthesis may be shunted into β-oxidation with a corresponding small increase in de novo FA synthesis required to achieve the WT level of TAG accumulation (21). However, high levels of HFA production and inefficient incorporation into TAG interferes with ER glycerolipid synthesis (10) and may induce the ER-to-plastid feedback inhibition of FA synthesis measured in CL37. It appears that production of medium-chain FAs within transgenic plants can produce different metabolic responses. Up to 60% of the seed oil can be accumulated by 12:0 within transgenic B. napus plants and induce futile cycles of increased rates of FA synthesis and β-oxidation, yet without an effect on oil levels (21). The differences in metabolic responses to medium-chain FAs and HFA may be due to their chemical structure or their sites of synthesis. Medium-chain FAs are produced within the plastid as intermediates of FA synthesis. The shuttling of incomplete products of FA synthesis exported from the plastid into β-oxidation (or sequestering them in TAG) may represent an endogenous pathway to limit their

A

Discussion Inefficient Glycerolipid Synthesis Induces Down-Regulation of de Novo FA Synthesis. Previously, β-oxidation of unusual FAs in

transgenic plants has been hypothesized as a major factor limiting efficient oilseed engineering. We cannot completely rule out β-oxidation of HFA to some extent during CL37 seed development. However, CL37 plants expressing the castor FA hydroxylase, compared with the fae1 control, have approximately half as much seed oil accumulation (Fig.1), half the rate of FA synthesis (Fig. 2A), and half the ACCase activity (Fig. 2C), yet transcript levels for ACCase subunits are not changed (Table 1). Together these results suggest the major mechanism for reduced oil in transgenic plants accumulating HFA (8–10) is reduced FA synthesis through posttranslational inhibition of ACCase, rather than primarily through degradation of HFA via β-oxidation. Intriguingly, more efficient utilization of HFA for TAG synthesis by coexpression of castor DGAT or PDAT with the FA hydroxylase alleviates the reduced ACCase activity (Fig. 2C), the Bates et al.

B

Fig. 3. Acyl–CoA and acyl–ACP compositions in developing seeds. (A) acyl– CoA. (B) acyl–ACP. A and B, mean (n = 5) ± SEM.


PLANT BIOLOGY

Table 1. Absence of significant changes in ACCase gene expression in developing seeds

incorporation into ER membrane lipids, and the corresponding increased rates of FA synthesis within these plants may restore the flux of FAs into ER glycerolipid synthesis. However, HFA are produced on PC which is a key intermediate of cellular membrane and storage lipid synthesis (13). Many PC-derived unusual FAs are thought to affect membrane structure and function and are typically maintained at very low levels within membrane lipids, even in plant species that naturally accumulate PC-derived unusual FAs within TAG (30). Thus, the inefficient flux of HFA from PC into TAG (9, 10) may be identified by the plant as a problem with membrane lipid synthesis or lipid oxidation and elicit a differential metabolic response than that of incomplete FA synthesis.

A

B C

Possible Mechanisms of ACCase Inhibition from Inefficient ER Glycerolipid Synthesis. Excess FA accumulation (from oleate

feeding) has been shown to induce posttranscriptional downregulation of ACCase activity and reduce de novo FA synthesis in tobacco and B. napus cell cultures (38, 40). However, the cells used the supplied FAs for growth, and thus overall lipid accumulation was not affected by the reduction in de novo FA synthesis. Additionally, the reduction in FA synthesis was recently attributed to a build-up of oleate in the acyl–ACP pool and may lead to direct ACCase inhibition through interaction with oleoyl–ACP (38). CL37 did not have an increased proportion of oleoyl–ACP relative to fae1, PDAT, or DGAT lines (Fig. 3B), and thus this mechanism is unlikely to be responsible for the CL37 phenotype. Additionally, it is unlikely that HFA–CoA is directly related to the reduction in FA synthesis rate in CL37 because both the PDAT and the DGAT lines have recovered FA synthesis rates, but only DGAT has a reduced proportion of HFA–CoA (Fig. 3A). The reduction in HFA–CoA by the castor DGAT is consistent with its enzymatic activity that selectively uses HFA–CoA to produce TAG (25). Fig. 4 displays a possible model for CL37 lipid metabolism based on the results reported here and previous research (10). The diacylglycerol (DAG) substrate used for TAG production can be synthesized de novo from glycerol-3-phosphate (G3P) and acyl–CoAs or it can be derived from PC (28). In wild-type Arabidopsis, TAG is primarily produced from PC-derived DAG (10), indicating a de novo DAG → PC → PC-derived DAG → TAG pathway (Fig. 4). We recently demonstrated that HFA– CoA [produced from PC acyl editing (Fig. 4A)] is efficiently used for de novo DAG synthesis; however, de novo HFA–DAG is not efficiently used for membrane lipid or TAG synthesis within CL37 seeds and is rapidly turned over, limiting the flux of glycerol into PC and thus TAG (Fig. 4, red X) (10). Because we conclude here that reduced FA synthesis (Fig. 2) is the major mechanism limiting oil accumulation in CL37, the FAs released from HFA–DAG turnover likely reenter glycerolipid metabolism. The synthesis of HFA-containing de novo DAG and its subsequent turnover produces a futile cycle of inefficient glycerolipid synthesis within the ER (Fig. 4B, red lines) which leads to posttranslational regulation of ACCase activity in the plastid (Fig. 4C) via phosphorylation, redox, binding proteins, enzyme turnover, or other regulators of ACCase activity (41–43). Further identification of the metabolic signals which control ACCase regulation, and thus FA synthesis, in response to unusual FA production may generate alternative bioengineering strategies for increasing total oil accumulation in plants (45). Closing Remarks. It has long been proposed that β-oxidation is

a major factor limiting the engineering of unusual FAs into crop plants. Here we demonstrate that posttranslational down-regulation of FA synthesis by unusual FAs can also significantly inhibit the accumulation of industrial oils within transgenic plants. The induced down-regulation of plastidial ACCase, due to in vivo metabolic changes within the ER of CL37 seeds, suggests that this transgenic system may be responding to an endogenous intercompartmental posttranslational regulatory pathway that would otherwise be difficult to study in wild-type plants. The

1208 | www.pnas.org/cgi/doi/10.1073/pnas.1318511111

Fig. 4. Proposed model of CL37 lipid metabolism. (A) HFA enters the acyl– CoA pool by acyl editing after HFA production on PC. (B) Red arrows, futile cycle of de novo HFA–DAG synthesis and turnover; red X, bottleneck in HFA– DAG flux into PC. (C) ER-to-plastid signaling that down-regulates ACCase activity and thus plastid FA synthesis. Abbreviations not in the text: LPC, lysophosphatidylcholine; LPA, lyso-phosphatidic acid; PA, phosphatidic acid.

metabolic analysis of transgenic plants producing HFA presented here enhances our knowledge of the mechanisms limiting the accumulation of TAG containing unusual FAs. In addition, the differential responses to various transgenically produced unusual FAs measured here and by others may be even more useful for understanding the basic biochemistry of how plants deal with changes to cellular FA content. Methods Plants, Oil Content, and Metabolic Labeling. All six plant lines were grown randomized across a growth chamber under constant ∼100–170 μE light, ∼22 °C, and ∼60% humidity. Seed oil was determined by whole-seed transmethylation and analysis by gas chromatography with flame ionization detection (46). For labeling, developing seeds were collected from each plant line grown together as in ref. 10. The [14C]acetate time course labeling was conducted as in ref. 10 using 1 mM [14C]acetate within the labeling media; data represent mean ± SEM (n = 4) from two separate experiments (each n = 2). The [3H]2O and [14C]acetate vs. [14C]malonate labelings each consisted of five individual 30-min labelings for each plant line/growth stage/radioisotope. Each individual labeling in 0.2 mL of media contained either 1 mCi [3H]2O, 10 μCi [14C] acetate, or 10 μCi [14C]malonate. The incorporation of radiolabel into FAs was determined by transmethylation of the total lipid extract as above, and purification of the fatty acid methyl esters by TLC on silica gel 60 20 × 20 cm glass plates (EMD, www.emdmillipore.com/) developed in hexane/diethyl ether/ acetic acid, 70/30/1 (vol/vol/vol). Radiolabel was quantified by liquid scintillation counting on a Tri–CARB liquid scintillation analyzer (Packard Instrument Company). RNA Expression Analysis. Frozen developing seeds of 7–8, 9–10, and 11–12 days after flowering (DAF) were collected from liquid N2 frozen siliques (47). Approximately 100 mg of frozen seed samples (three biological replicates for each plant line/developmental stage) were ground to a fine power, RNA extracted (48), and DNA removed (DNA-Free RNA Kit, Zymo Research). Libraries were prepared from 2 to 4 μg total RNA (Illumina TruSeq RNA kits) and sequenced with Illumina HiSeq2000. Reads (50 nt) were trimmed, filtered, and mapped against TAIR10 genome (www.arabidopsis.org/) with Bowtie (v. 2.0.0-beta7). Only alignments consistent with annotated transcripts were considered valid. Each sample and biological replicate was aligned separately. All alignments were then used as input to Cuffdiff (v. 2.0.1) (49) to calculate per sample, per gene fragments per kilobase of exon per million fragments mapped (FPKM), and statistical significance (q value) of differential expression between transgenic lines. Cufflinks averages the FPKM across biological replicates within each sample.

Bates et al.

Acyl–CoA and Acyl–ACP Analysis. Acyl–CoA and acyl–ACP were extracted from developing seeds of 9–12 DAF that were collected from liquid N2 frozen siliques (47). Acyl–CoA were exacted as in ref. 50. Extracted acyl–CoA dried under N2 were suspended in 100 μL of 50% (vol/vol) MeOH and analyzed as in ref. 51, except HPLC solvents A and B contained an additional 7.5 mM ammonium hydroxide. An acyl–ACP-enriched protein fraction was collected by vortexing finely ground frozen seed tissue in 5% (vol/vol) trichloroacetic acid (TCA) and centrifuged 21,000 g (10 min, 4 °C). The pellet was dissolved in 1% TCA and centrifuged again. The pellet was dissolved in 50 mM Mops, pH 7.6, checked for pH above 6.5, incubated on ice (30 min), and centrifuged again. 100% TCA was added to the supernatant

to a final concentration of 10%, frozen (−80 °C, 10 min), thawed on ice, and centrifuged again. The pellet was washed with 1% TCA, centrifuged again, and dissolved in Mops buffer again. The protein samples were digested with Asp–N endoproteinase and analyzed on a 4000 QTRAP LC/MS/MS system (Applied Biosystems).

1. Badami RC, Patil KB (1980) Structure and occurrence of unusual fatty acids in minor seed oils. Prog Lipid Res 19(3–4):119–153. 2. Vanhercke T, Wood CC, Stymne S, Singh SP, Green AG (2013) Metabolic engineering of plant oils and waxes for use as industrial feedstocks. Plant Biotechnol J 11(2): 197–210. 3. Cahoon EB, et al. (2007) Engineering oilseeds for sustainable production of industrial and nutritional feedstocks: Solving bottlenecks in fatty acid flux. Curr Opin Plant Biol 10(3):236–244. 4. Napier JA (2007) The production of unusual fatty acids in transgenic plants. Annu Rev Plant Biol 58:295–319. 5. Haslam RP, et al. (2013) The modification of plant oil composition via metabolic engineering—better nutrition by design. Plant Biotechnol J 11(2):157–168. 6. Knutzon DS, Hayes TR, Wyrick A, Xiong H, Maelor Davies H, Voelker TA (1999) Lysophosphatidic acid acyltransferase from coconut endosperm mediates the insertion of laurate at the sn-2 position of triacylglycerols in lauric rapeseed oil and can increase total laurate levels. Plant Physiol 120(3):739–746. 7. Larson TR, Edgell T, Byrne J, Dehesh K, Graham IA (2002) Acyl CoA profiles of transgenic plants that accumulate medium-chain fatty acids indicate inefficient storage lipid synthesis in developing oilseeds. Plant J 32(4):519–527. 8. Dauk M, Lam P, Kunst L, Smith MA (2007) A FAD2 homologue from Lesquerella lindheimeri has predominantly fatty acid hydroxylase activity. Plant Sci 173(1):43–49. 9. van Erp H, Bates PD, Burgal J, Shockey J, Browse J (2011) Castor phospholipid:diacylglycerol acyltransferase facilitates efficient metabolism of hydroxy fatty acids in transgenic Arabidopsis. Plant Physiol 155(2):683–693. 10. Bates PD, Browse J (2011) The pathway of triacylglycerol synthesis through phosphatidylcholine in Arabidopsis produces a bottleneck for the accumulation of unusual fatty acids in transgenic seeds. Plant J 68(3):387–399. 11. Li R, et al. (2012) Vernonia DGATs can complement the disrupted oil and protein metabolism in epoxygenase-expressing soybean seeds. Metab Eng 14(1):29–38. 12. van de Loo FJ, Broun P, Turner S, Somerville C (1995) An oleate 12-hydroxylase from Ricinus communis L. is a fatty acyl desaturase homolog. Proc Natl Acad Sci USA 92(15): 6743–6747. 13. Bates PD, Stymne S, Ohlrogge J (2013) Biochemical pathways in seed oil synthesis. Curr Opin Plant Biol 16(3):358–364. 14. Roughan PG (1970) Turnover of the glycerolipids of pumpkin leaves. The importence of phosphatidylcholine. Biochem J 117(1):1–8. 15. Gerhardt B (1992) Fatty acid degradation in plants. Prog Lipid Res 31(4):417–446. 16. Theodoulou FL, Eastmond PJ (2012) Seed storage oil catabolism: A story of give and take. Curr Opin Plant Biol 15(3):322–328. 17. Baud S, Boutin JP, Miquel M, Lepiniec L, Rochat C (2002) An integrated overview of seed development in Arabidopsis thaliana ecotype WS. Plant Physiol Biochem 40(2): 151–160. 18. Broun P, Somerville C (1997) Accumulation of ricinoleic, lesquerolic, and densipolic acids in seeds of transgenic Arabidopsis plants that express a fatty acyl hydroxylase cDNA from castor bean. Plant Physiol 113(3):933–942. 19. Eccleston VS, Cranmer AM, Voelker TA, Ohlrogge JB (1996) Medium-chain fatty acid biosynthesis and utilization in Brassica napus plants expressing lauroyl-acyl carrier protein thioesterase. Planta 198(1):46–53. 20. Lu CF, Fulda M, Wallis JG, Browse J (2006) A high-throughput screen for genes from castor that boost hydroxy fatty acid accumulation in seed oils of transgenic Arabidopsis. Plant J 45(5):847–856. 21. Eccleston VS, Ohlrogge JB (1998) Expression of lauroyl-acyl carrier protein thioesterase in brassica napus seeds induces pathways for both fatty acid oxidation and biosynthesis and implies a set point for triacylglycerol accumulation. Plant Cell 10(4): 613–622. 22. Moire L, Rezzonico E, Goepfert S, Poirier Y (2004) Impact of unusual fatty acid synthesis on futile cycling through beta-oxidation and on gene expression in transgenic plants. Plant Physiol 134(1):432–442. 23. Smith MA, Moon H, Chowrira G, Kunst L (2003) Heterologous expression of a fatty acid hydroxylase gene in developing seeds of Arabidopsis thaliana. Planta 217(3): 507–516. 24. Lu CF, Kang JL (2008) Generation of transgenic plants of a potential oilseed crop Camelina sativa by Agrobacterium-mediated transformation. Plant Cell Rep 27(2): 273–278. 25. Burgal J, et al. (2008) Metabolic engineering of hydroxy fatty acid production in plants: RcDGAT2 drives dramatic increases in ricinoleate levels in seed oil. Plant Biotechnol J 6(8):819–831.

26. Kunst L, Taylor DC, Underhill EW (1992) Fatty-acid elongation in developing seeds of Arabidopsis-thaliana. Plant Physiol Biochem 30(4):425–434. 27. Bonaventure G, Bao X, Ohlrogge J, Pollard M (2004) Metabolic responses to the reduction in palmitate caused by disruption of the FATB gene in Arabidopsis. Plant Physiol 135(3):1269–1279. 28. Bates PD, Browse J (2012) The significance of different diacylgycerol synthesis pathways on plant oil composition and bioengineering. Front Plant Sci 3:147. 29. Jungas RL (1968) Fatty acid synthesis in adipose tissue incubated in tritiated water. Biochemistry 7(10):3708–3717. 30. Millar AA, Smith MA, Kunst L (2000) All fatty acids are not equal: Discrimination in plant membrane lipids. Trends Plant Sci 5(3):95–101. 31. Tjellström H, Yang Z, Allen DK, Ohlrogge JB (2012) Rapid kinetic labeling of Arabidopsis cell suspension cultures: implications for models of lipid export from plastids. Plant Physiol 158(2):601–611. 32. Bates PD, et al. (2012) Acyl editing and headgroup exchange are the major mechanisms that direct polyunsaturated fatty acid flux into triacylglycerols. Plant Physiol 160(3):1530–1539. 33. Bates PD, Durrett TP, Ohlrogge JB, Pollard M (2009) Analysis of acyl fluxes through multiple pathways of triacylglycerol synthesis in developing soybean embryos. Plant Physiol 150(1):55–72. 34. Bates PD, Ohlrogge JB, Pollard M (2007) Incorporation of newly synthesized fatty acids into cytosolic glycerolipids in pea leaves occurs via acyl editing. J Biol Chem 282(43):31206–31216. 35. Bourgis F, et al. (2011) Comparative transcriptome and metabolite analysis of oil palm and date palm mesocarp that differ dramatically in carbon partitioning. Proc Natl Acad Sci USA 108(30):12527–12532. 36. Troncoso-Ponce MA, et al. (2011) Comparative deep transcriptional profiling of four developing oilseeds. Plant J 68(6):1014–1027. 37. Ruuska SA, Girke T, Benning C, Ohlrogge JB (2002) Contrapuntal networks of gene expression during Arabidopsis seed filling. Plant Cell 14(6):1191–1206. 38. Andre C, Haslam RP, Shanklin J (2012) Feedback regulation of plastidic acetyl-CoA carboxylase by 18:1-acyl carrier protein in Brassica napus. Proc Natl Acad Sci USA 109(25):10107–10112. 39. Brown AP, et al. (2012) Tissue-specific whole transcriptome sequencing in castor, directed at understanding triacylglycerol lipid biosynthetic pathways. PLoS ONE 7(2): e30100. 40. Shintani DK, Ohlrogge JB (1995) Feedback inhibition of fatty-acid synthesis in tobacco suspension cells. Plant J 7(4):577–587. 41. Feria Bourrellier AB, et al. (2010) Chloroplast acetyl-CoA carboxylase activity is 2oxoglutarate-regulated by interaction of PII with the biotin carboxyl carrier subunit. Proc Natl Acad Sci USA 107(1):502–507. 42. Sasaki Y, Nagano Y (2004) Plant acetyl-CoA carboxylase: Structure, biosynthesis, regulation, and gene manipulation for plant breeding. Biosci Biotechnol Biochem 68(6):1175–1184. 43. Nikolau BJ, Ohlrogge JB, Wurtele ES (2003) Plant biotin-containing carboxylases. Arch Biochem Biophys 414(2):211–222. 44. Poirier Y, Ventre G, Caldelari D (1999) Increased flow of fatty acids toward betaoxidation in developing seeds of Arabidopsis deficient in diacylglycerol acyltransferase activity or synthesizing medium-chain-length fatty acids. Plant Physiol 121(4): 1359–1366. 45. Weselake RJ, et al. (2009) Increasing the flow of carbon into seed oil. Biotechnol Adv 27(6):866–878. 46. Li-Beisson Y, et al. (2013) Acyl-lipid metabolism. Arabidopsis Book 11:e0161. 47. Bates PD, Jewell JB, Browse J (2013) Rapid separation of developing Arabidopsis seeds from siliques for RNA or metabolite analysis. Plant Methods 9(1):9. 48. Suzuki Y, Kawazu T, Koyama H (2004) RNA isolation from siliques, dry seeds, and other tissues of Arabidopsis thaliana. Biotechniques 37(4):542–544, 544. 49. Trapnell C, et al. (2013) Differential analysis of gene regulation at transcript resolution with RNA-seq. Nat Biotechnol 31(1):46–53. 50. Larson TR, Graham IA (2001) Technical advance: A novel technique for the sensitive quantification of acyl CoA esters from plant tissues. Plant J 25(1):115–125. 51. Han J, et al. (2010) The cytochrome P450 CYP86A22 is a fatty acyl-CoA ω-hydroxylase essential for Estolide synthesis in the stigma of Petunia hybrida. J Biol Chem 285(6): 3986–3996.

Bates et al.


PLANT BIOLOGY

ACKNOWLEDGMENTS. We are grateful to the Research Technology Support Facility at Michigan State University for sequencing and to Kevin Carr for bioinformatics analysis. This work was supported by the National Science Foundation (Grant DBI-0701919), the Agricultural Research Center at Washington State University, and Department of Energy–Great Lakes Bioenergy Research Center Cooperative Agreement DE-FC02-07ER6449.

Supplemental Figures and Discussion

g FA per seed

A

8 6 4 2

Seed dry wt (g)

B

fa e1 C L3 7 C L7 PD A T D G A T

C ol -0

0

25 20 15 10 5

% FA of seed weight

C


C ol

-0

0

40 30 20 10


C

ol -0

0

Fig S1. Seed weight and oil content. (A) μg total FA per seed. (B) Seed weights. (C) Oil content as percentage of seed weight. Descriptions of each line are in Fig. 1. Data are mean ± SEM of 15-18 replicates for each plant line.

B

600

fae1

g neutral lipid/ silique


A

Slope: 70.8  9.8

400

200

0

600

CL37 Slope: 37.2  8.9

400

200

0 6

8

10

12

6

Days after flowering D

600

PDAT



C

Slope: 52.1  12.2

400

200

8

10

12

Days after flowering 600

DGAT Slope: 64.5  10.8

400

200

0

0 6

8

10

Days after flowering

12

6

8

10

12

Days after flowering

Fig S2. Accumulation of neutral lipids during seed development. (A) fae1. (B) CL37. (C) PDAT. (D) DGAT. Data are mean ± SEM of 7-10 replicates per time point. Method. For each plant line 1-3 siliques aged 7-12 days after flowering (DAF) were collected from 7-10 individual plants grown as above and quenched in boiling isopropanol. Neutral lipids were collected from the total lipid extract by fractionation on a DSC-Si SPE tube, 100 mg bed wt. (Sigma). Each SPE tube was prewashed with 1 ml MeOH and 2 ml CHCl3. Samples were loaded in 0.2 ml of CHCl3 and neutral lipids eluted with 1 ml 5% MeOH in CHCl3. Total neutral lipids dried under N2 were transmethylated, FAME collected into 0.2 ml of hexane and quantified by GC against 40 μg 17:0 TAG internal standard.

Fig S3. Accumulation of [14C]acetate labeled FA into lipids of fae1 and CL37. 9-10 DAF seeds from figure 2, n = 2. (A) fae1 time course. (B) CL37 time course. (C) Stereochemistry of labeled FA in PC over the fae1 and CL37 time courses. Lipid separation, quantification and stereochemistry preformed as in (1). Fig. S3 Discussion Newly synthesized [14C]FAs exported from the plastid are first incorporated into PC through acyl editing (2) (Fig. 4A). The similar initial proportion (%) of nascent [14C]FA incorporated into glycerolipids of fae1 and CL37 (Fig. S3A-B), and the similar stereochemical labeling of PC (Fig. S3C) indicates that the relative rates of acyl editing and de novo DAG synthesis in relation to nascent FA synthesis are functioning essentially the same in both fae1 and CL37 (3, 4). The relative changes in PC, DAG and TAG over time between fae1 and CL37 are consistent with reduced PC synthesis from de novo DAG by the HFA-DAG to PC bottleneck (Fig. 4, red X) previously characterized (1). If there was a large flux of HFA from PC into β-oxidation then we would expect to see an increase in PC acyl editing represented by increased sn-2 and total PC labeling, which is not apparent.

6.010 6 4.010 6 2.010 6

T D

G

A

T PD A

C L3

7

0

fa e1

Total peak area/ g total protein

8.010 6

Fig. S4. Relative acyl-ACP pool sizes between developing seeds of fae1, CL37, PDAT and DGAT. Relative acyl-ACP pool sizes are determined based on the total MS peak area for all acyl-ACP species normalized to the total protein extract for each sample, mean ± SEM, n = 5. There is no statistically significant difference in the acyl-ACP pool sizes between any of the plant lines.

A DPM/g chlorophyll

40000

[14C]acetate 30 min

30000 20000 10000

DPM/g chlorophyll

B

800

A T G D

PD A T

C L3 7

fa e1

0

[14C]malonate 30 min

600 400 200

A T D G

T PD A

L3 7 C

fa e1

0

Fig S5. Total incorporation of radiolabeled substrates into lipids of 11-12 DAF developing seeds. (A) [14C]acetate. (B) [14C]malonate. 30 min labelings, mean ± SEM, n = 5.

1.

2. 3. 4.

Bates PD & Browse J (2011) The pathway of triacylglycerol synthesis through phosphatidylcholine in Arabidopsis produces a bottleneck for the accumulation of unusual fatty acids in transgenic seeds. The Plant Journal 68(3):387-399. Bates PD, Stymne S, & Ohlrogge J (2013) Biochemical pathways in seed oil synthesis. Curr. Opin. Plant Biol. 16(3):358-364. Bates PD & Browse J (2012) The significance of different diacylgycerol synthesis pathways on plant oil composition and bioengineering. Frontiers in Plant Science 3:147. Bates PD, et al. (2012) Acyl Editing and Headgroup Exchange Are the Major Mechanisms That Direct Polyunsaturated Fatty Acid Flux into Triacylglycerols. Plant Physiol. 160(3):1530-1539.