Index | Search | Home | Table of Contents

Davies, H.M., L. Anderson, J. Bleibaum, D.J. Hawkins, C. Fan, A.C. Worrell, and T.A. Voelker. 1993. Fatty acid synthesis genes: Engineering the production of medium-chain fatty acids. p. 176-181. In: J. Janick and J.E. Simon (eds.), New crops. Wiley, New York.

Fatty Acid Synthesis Genes: Engineering the Production of Medium-chain Fatty Acids

H. Maelor Davies, L. Anderson, J. Bleibaum, D.J. Hawkins, C. Fan, A.C. Worrell, and Toni A. Voelker

  6. Table 1
  7. Fig. 1
  8. Fig. 2
  9. Fig. 3
  10. Fig. 4
  11. Fig. 5
  12. Fig. 6
  13. Fig. 7

Certain plant species uniquely accumulate "medium-chain" acyl groups in their seed triacylglycerols (Hilditch and Williams 1964), and many of these fatty acids are important commercially. For example, coconut oil is a major source of laurate (12:0) for use in detergent formulations. The genus Cuphea, is particularly noteworthy for the considerable diversity of medium-chain compositions among its many species (Graham and Kleiman 1985). Such unusual acyl compositions have attracted the attention of lipid biochemists, not only because of their commercial relevance but also because they pose the challenge of understanding an uncommon variation in fatty acid biosynthesis. A number of theoretical explanations have been suggested (Stumpf 1987), but until recently neither measurements of metabolism in vitro (Deerberg et al. 1990), nor direct assays for hypothetical enzymes (Oo and Stumpf 1979), have provided any conclusive evidence to support them.

At Calgene, we are interested in isolating the genes involved in many areas of lipid biosynthesis, with a view to increasing the range of fats and oils that can be produced commercially in domestic rapeseed crops. Our interests include the medium-chain fatty acids, especially those such as 8:0, 10:0, and 14:0 that are not readily obtained from natural, renewable resources. We therefore began an investigation of medium-chain fatty acid biosynthesis, hoping that only one or two genes would have to be isolated and transferred to rapeseed in order to "re-program" the fatty acid biosynthesis pathway of that oilseed for medium-chain production.


The choice of species for these investigations was determined by practical convenience. The Lauraceae contains several species that accumulate medium-chain fatty acyl groups in their seed triglycerides. One species is native to northern California and abundant, namely the California bay (Umbellularia californica). The seeds of this tree contain large embryos (2 to 2.5 g fresh weight before desiccation) whose oil content (55%) comprises almost entirely 10:0 (36%) and 12:0 (58%). To our knowledge, the mechanism of medium-chain biosynthesis in these embryos had never previously been investigated.

Ammonium sulfate fractions of immature oilseed embryos can synthesize fatty acids in vitro when supplied with the necessary substrates and co-factors. The requirement for acyl-carrier protein (ACP) is usually satisfied by providing E. coli ACP. Such an in vitro fatty acid synthesis (FAS) system from, for example, safflower seeds produces long-chain fatty acids de novo from acetyl-CoA and malonyl-CoA (Pollard and Singh 1987). An analogous system prepared from immature California bay (bay) cotyledons produced chiefly 10:0 and 12:0, in approximately a 1:2 ratio as in the intact seeds (Pollard et al. 1991). This suggested that the mechanism of medium-chain fatty acid formation continued to operate in vitro much as it did in vivo, and that the in vitro system could be used to investigate it.

The in vitro FAS system from bay cotyledons accumulated pools of acyl-thioesters as well as free fatty acids. Whereas the latter were chiefly 10:0 and 12:0, the acyl groups in the thioester pools were primarily 8:0 and 10:0 (Table 1). There was no detectable pool of 12:0 thioester. Taking the thioester pools to be acyl-ACPs (for which some evidence was obtained), this suggested to us that normal fatty acid biosynthesis was being modulated for medium-chain production by specific hydrolysis of 10:0-ACP and 12:0-ACP. Thus, although 8:0-thioester was accumulated it was not hydrolyzed to release free 8:0 in any appreciable amount. But 12:0-ACP was hydrolyzed so effectively that none of the thioester could be detected. The situation with 10:0-ACP was intermediate between these two extremes, resulting in some 10:0 production and 10:0-ACP remaining for extension to 12:0-ACP by the normal FAS pathway.

The simplest model to explain these results invoked a medium-chain specific acyl-ACP thioesterase, acting on 12:0-ACP and (to a lesser extent) on 10:0-ACP. By detaching these acyl groups from ACP the thioesterase would prevent their further extension to long-chain fatty acids and cause them to accumulate. Direct assay of bay cotyledon preparations showed the presence of such an enzyme activity (Pollard et al. 1991; Fig. 1). The usual long-chain thioesterase activity was also present, acting on 16:0-ACP, 18:1-ACP etc. (Fig. 1). Surprisingly, the medium-chain enzyme was primarily active on 12:0-ACP. It showed low activity on 14:0-ACP, but only a trace of activity was ever observed with 10:0-ACP as substrate. This apparent lack of an activity to account for the considerable 10:0 content of the bay seeds remains puzzling. One possible explanation is that the thioesterase interacts in some way with the FAS enzymes for more efficient interception of acyl-ACPs (in a manner perhaps analogous to the medium-chain acyl-ACP thioesterases of animal systems: deRenobales et al. 1980; Smith 1980), and that its specificity is a little different when working in this way. We hope to shed more light on this enigma when we have the bay 12:0-ACP thioesterase expressed in plants which normally produce long-chain fatty acids. Meanwhile, additional evidence was obtained to show that the 12:0-ACP thioesterase is involved in medium-chain production in vivo. For example, its activity is very low in young embryos that are not accumulating medium chains, and then it rises coincidentally with the onset of medium-chain deposition (Fig. 2).


The 12:0-ACP thioesterase was substantially purified from immature bay cotyledons by the scheme shown in Fig. 3 (Davies et al. 1991). The ACP affinity column not only contributed considerable purification, but also separated the activity from long-chain thioesterase (assayed with 18:1-ACP) and a 12:0-CoA hydrolase that was present (Fig. 4). The latter separation indicated that the 12:0-ACP thioesterase was much more active on 12:0-ACP than on 12:0-CoA, as expected for an enzyme involved in plastid-localized fatty acid biosynthesis. In spite of the effectiveness of this affinity chromatography step, the resulting preparation still contained several protein species as seen by SDS-PAGE and silver-staining. This preparation was subjected to additional chromatography and electrophoresis (examples shown in Fig. 3). We identified the proteins whose behavior in these experiments most closely matched the behavior of enzyme activity, and thereby prioritized the proteins as "candidates" for the thioesterase. The best candidates were a group of proteins (a single one in some preparations) of approximately 34 kDa molecular weight.

These proteins are quite "rare" in bay cotyledons, perhaps because the enzyme has a very high specific activity. Several kilograms of cotyledons had to be processed through the purification scheme to yield a few tens of µg of 34 kDa proteins for sequencing. Sequence of one of the candidates was obtained from several tryptic peptides and from the N-terminus. To generate a partial cDNA probe for use in gene isolation, we followed the strategy of mixed oligonucleotide primed amplification of cDNA. Poly(A) RNA was isolated from developing bay cotyledons, and reverse-transcribed to provide a single-strand cDNA to act as template in polymerase chain reactions (PCR). Sense and antisense degenerate oligonucleotides corresponding to the sequenced peptides were used as primers. The PCR procedure amplified a 0.8 kb DNA fragment, which was subsequently used to screen a plasmid cDNA library, resulting in the isolation of several clones.

The longest clone that was isolated contains an 1,163 bp open reading frame, within which is located an ATG surrounded by sequences which match the "rules" for plant initiation of translation (Fig. 5). This translation start predicts a 382 amino acid polypeptide, and the available N-terminal sequence indicates an 83-residue transit sequence and 299-residue mature protein. The transit sequence contains typically conserved features of plastid transit peptides, and its presence is consistent with the protein being involved in fatty acid biosynthesis.


We tested the identity of this cDNA, and the correctness of the original protein identification, by expression in E. coli. A translational fusion was created using the "mature" bay sequence and the modified N-terminal coding sequence of the bacterial lacZ gene on a plasmid. E. coli cells containing this plasmid produced very high 12:0-ACP thioesterase activity, approximately 1,000-fold greater than the cells' own acyl-ACP hydrolysis activities (Fig. 6). The acyl-group specificity of this additional activity corresponded exactly with that of the seed-purified enzyme, confirming that the correct protein had been identified and the correct cDNA isolated.

Liquid cultures of E. coli transformants expressing the bay thioesterase cDNA had approximately twice the 12:0 content of control cultures. The absolute amount of 12:0 in the cultures was still small however (6.5% of the total fatty acids), which is surprising considering the very high thioesterase activity that was present. To ascertain whether a large amount of laurate was being produced and then catabolized, we transformed E. coli strains that were defective in fatty acid breakdown. The resulting 12:0 accumulation was considerable. For example, fadD mutants (lacking medium-chain acyl-CoA synthetase) accumulated sufficient 12:0 over a 20 h growth period to comprise 90% of the culture fatty acid content, a mass of 12:0 approximately equal to the total dry weight of the cells themselves (Fig. 7). The other fatty acids were accumulated to only 50% of their levels in fadD control cultures, and the transformed fadD cells reached stationary phase at a lower cell density (Fig. 7).

The growth rate of transformed E. coli fadD colonies on agar at 25°C was comparable to that of untransformed controls (results not shown). After several days' incubation, crystals formed on the plates of transformed colonies, both in the colonies themselves and on the agar surface. We identified these crystals as the potassium salt of 12:0.


These results show that bay 12:0-ACP thioesterase is capable of interacting with the FAS system of E. coli in vivo to effect medium-chain fatty acid production. The products were what would be expected from the in vitro specificity of the enzyme, only very small amounts of 10:0 being formed. At present, it is unclear whether this particular thioesterase is responsible for both 10:0 and 12:0 formation, or only 12:0 production, in bay seeds. Experiments are underway to introduce the gene into rapeseed, under the control of embryo-specific promoters such as napin. These experiments will test (a) the efficiency of medium-chain production when this enzyme encounters a heterologous plant system; (b) the question of whether this thioesterase can effect 10:0 as well as 12:0 production in vivo; and (c) the efficiency with which medium-chains can be transported from the plastid to the cytoplasm and incorporated into triacylglycerols in a species that does not normally contain them.


Table 1. Products of in vitro fatty acid synthesizing system prepared from developing Umbellularia californica cotyledonsz.

Chain length
Product 8 10 12 18
Free fatty acids (pmol) 0 4.1 15.5 0
0 3.6 11.9 0.8
Thioester acyl groups (pmol) 6.9 3.1 0 0
5.7 1.4 0.3 0.5
zAn ammonium sulfate fraction from developing cotyledons was supplied with E. coli ACP (0.2 mg/ml), acetyl-CoA, radiolabeled malonyl-CoA, reduced pyridine nucleotide and other typical requirements. Duplicate incubations are shown. Free fatty acids were extracted from the incubation mixture with hexane saturated with isopropanol and water. Acyl groups in the remaining aqueous fraction were considered to derive from thioesters. In a separate experiment it was confirmed that acyl-ACPs were present in this aqueous fraction, having the proportions of acyl groups shown above, i.e. C8 > C10 and very little C12 or C9.

Fig. 1. Hydrolysis of acyl-ACP substrates by crude extract of developing Umbellularia californica cotyledons. Substrates were prepared from E. coli ACP and [1-14C]-fatty acids. Incubation was with 4.5 µM substrate for 30 min at 30°C in 50 mM Na phosphate buffer pH 7.5. The radioactive products were extracted in ether, and in the case of laurate verified to be fatty acid by TLC.

Fig. 2. Extractable 12:0-ACP thioesterase activity and medium-chain fatty acyl contents of Umbellularia californica cotyledons at different stages in development. Seeds were harvested periodically from the same tree at the times indicated from an arbitrary starting date (plotted as day 8). At least 5 cotyledon pairs were combined and powdered at each sampling time for enzyme extraction and fatty acyl analysis. The activity results are duplicate assays of simple, crude extracts. Total fatty acyl compositions were determined singly, by acidic methanolysis of the powdered cotyledonary tissue.

Fig. 3. Scheme for purification of 12:0-ACP thioesterase from immature cotyledons of Umbellularia californica.

Fig. 4. Chromatography of partially purified 12:0-ACP thioesterase and (a) the accompanying 18:1-ACP hydrolysis activity, or (b) the accompanying 12:0-CoA hydrolysis activity, on immobilized E. coli ACP. Experiments (a) and (b) were performed with different thioesterase preparations.

Fig. 5. Relative sizes of Umbellularia californica 12:0-ACP thioesterase cDNA, preprotein, and mature protein.

Fig. 6. Substrate specificities of thioesterases purified from Umbellularia californica seeds, and expressed in E. coli. The activity axes have been set so that the 12:0-ACP activity columns are of equal height, in order to show clearly the plant/bacterial comparisons for the other substrates. Background E. coli acyl-ACP hydrolysis activities were negligible at the dilutions required to maintain the introduced 12:0-ACP thioesterase within the linear range of the assay. The small amount of activity shown by the plant preparation towards 18:1-ACP resulted from incomplete separation of the medium- and long-chain thioesters during purification.

Fig. 7. Laurate accumulation (A), and growth curves (B), of control and transformed fadD E. coli cultures. Bacterial colonies were resuspended in growth medium and grown under continuous shaking at 30°C for 2 h in order to enter logarithmic growth (time zero). At the indicated times 10 ml samples were removed for fatty acyl analysis (A). The sum of 16:0 and 16:1 served as a measure of the endogenous bacterial fatty acyl content, comprising 66% of the total fatty acyl groups of control E. coli cultures. Laurate (12:0) levels are shown only for the E. coli transformed with Umbellularia californica 12:0-ACP thioesterase, the control 12:0 level being below the detection limit in this experiment.

Last update September 9, 1997 aw