CA2206984A1 - Engineering plant thioesterases and disclosure of plant thioesterases having novel substrate specificity - Google Patents

Abstract

Methods of altering substrate specificity of plant acyl-ACP thioesterases, and engineered plant acyl-ACP thioesterases so produced are provided. The C-terminal two-thirds portion of plant thioesterases is identified as desirable for such modifications. DNA sequences and constructs for expression of engineered thioesterases, as well as the novel thioesterases produced therefrom are also provided. Such DNA sequences may be used for expression of the engineered thioesterases in host cells, particularly seed cells of oilseed crop plants, for the modification of fatty acid composition. A C12 preferring plant acyl-ACP thioesterase described herein may be altered to obtain a plant thioesterase having approximately equal activity on C14 and C12 substrates.
Further modification of the C12 enzyme yields a thioesterase having greater activity on C14 as compared to C12 substrates.

Description

W O96/36719 PCTrUS96/07064 ENGTM~RTNG PLANT THIOEST~R~.C~S AND DISCLOSURE OF
PLANT THIOESTERASES HaVING NOV~L SUBSTRATE
SPECIFICITY
Technical Field The present invention is directed to proteins, nucleic acid sequences and constructs, and methods related thereto.

INTRODUCTION
Backaround Fatty acids are organic acids having a hydrocarbon chain of from about 4 to 24 carbons. Many different kinds of fatty acids are known which differ from each other in chain length, and in the presence, number and position of double bonds. In cells, fatty acids typically exist in covalently bound forms, the carboxyl portion being referred to as a fatty acyl group. The chain length and degree of saturation of these molecules is often depicted by the formula CX:Y, where "X" indicates nllmh~r of carbons and "Y"
indicates number of double bonds.
The production of fatty acids in plants begins in the plastid with the reaction between acetyl-CoA and malonyl-ACP
to produce butyryl-ACP catalyzed by the enzyme, $-ketoacyl-ACP synthase III. Elongation of acetyl-ACP to 16- and 18-carbon fatty acids involves the cyclical action of the following sequence of reactions: con~n~ation with a two-carbon unit from malonyl-~CP to form a $-ketoacyl-ACP ($-ketoacyl-ACP synthase), reduction of the keto-function to an alcohol (g-ketoacyl-ACP reductase), dehydration to form an enoyl-ACP ($-hydroxyacyl-ACP dehydrase), and finally reduction of the enoyl-ACP to form the elongated saturated acyl-ACP (enoyl-ACP reductase). $-ketoacyl-ACP synthase I, catalyzes elongation up to palmitoyl-ACP (C16:0), whereas ~-ketoacyl-ACP synthase II catalyzes the final elongation to stearoyl-ACP (C18:0). ~he longest chain fatty acids produced by the FAS are typically 18 carbons long. A further fatty acid biochemical step occurring in the plastid is the Sll~S 111 IJTE S~EET (l~ LE 26) W O96/36719 PCT~US96/07064 desaturation of stearoyl-ACP (C18:0) to form oleoyl-ACP
(C18:1) in a reaction catalyzed by a A-g desaturase, also often referred to as a ~stearoyl-ACP desaturase~ because of its high activity toward stearate the 18 carbon acyl-ACP.
Carbon-chain elongation in the plastids can be t~rmin~ted by transfer of the acyl group to glycerol 3-phosphate, with the resulting glycerolipid retained in the plastidial, "prokaryotic", lipid biosynthesis pathway.
Alternatively, specific thioesterases can intercept the prokaryotic pathway by hydrolyzing the newly produced acyl-ACPs into free fatty acids and ACP.
Subsequently, the free fatty acids are converted to fatty acyl-CoA's in the plastid envelope and exported to the cytoplasm. There, they are incorporated into the "eukaryotic" lipid biosynthesis pathway in the endoplasmic reticulum which is responsible for the formation of phospholipids, triglycerides and other neutral lipids.
Following transport of fatty acyl CoA's to the endoplasmic reticulum, subsequent sequential steps for triglyceride production can occur. For example, polyunsaturated fatty acyl groups such as linoleoyl and a-linolenoyl, are produced as the result of sequential desaturation of oleoyl acyl groups by the action of membrane-bound enzymes.
Triglycerides are formed by action of the 1-, 2-, and 3-acyl-ACP transferase enzymes glycerol-3-phosphate acyltransferase, lysophosphatidic acid acyltransferase and diacylglycerol acyltransferase. The fatty acid composition of a plant cell is a reflection of the free fatty acid pool and the fatty acids (fatty acyl groups) incorporated into triglycerides as a result of the acyltransferase activities. The properties of a given triglyceride will depend upon the various combinations of fatty acyl groups in the different positions in the triglyceride molecule.
For example, if the fatty acyl groups are mostly saturated fatty acids, then the triglyceride will be solid at room temperature. In general, however, vegetable oils tend to be mixtures of different triglycerides. The triglyceride oil properties are therefore a result of the combination of SlJI~I 11 ~JTE SHEET (RULE 26) W O96/36719 PCTrUS96/07064 triglycerides which make up the oil, which are in turn influenced by their respective fatty acyl compositions.
Plant acyl-acyl carrier protein thioesterases are of biochemical interest because of their roles in fatty acid synthesis and their utilities in bioengineering of plant oil seeds. A medium-chain acyl-ACP thioesterase from California bay tree, Umbellularia cali~ornica, has been isolated (Davies et al. (1991) Arch. Biochem. Biophys~
290:37-45), and its cDNA cloned and expressed in E.coli (Voelker et al. (1994) J. Bacterial. 176:7320-7327) and seeds of Arabidopsis thaliana and Brassica napus (Voelker et al. (1992) Science 257:72-74). In all cases, large amounts of laurate (12:0) and small amounts of myristate (14:0) were accumulated. These results ~m~n~trated the role of the ~E in detPrm; n; ng chain-length during de novo fatty acid biosynthesis in plants and the utility of these enzymes for modifying seed oil compositions in higher plants.
Recently, a number of cDNA encoding different plant acyl-ACP thioesterases have been cloned (Knutzon et al.
(1992) Plant Physiol. 100:1751-1758; Voelker, et al. (1992) supra; Dormann et al. (1993) Planta 189:425-432; Dormann et al. ( 1994) Biochim. Biophys. Acta 1212: 134-136; Jones et al. (1995) The Plant Cell 7:359-371). Sequence analyses of these thioesterases show high homology, implying similarity in structure and function. Some of these thioesterase cDNAs have been expressed in E. coli, and their substrate specificities determined by in vitro assays. The fact that these enzymes share significant se~uence homology, yet show different substrate specificities, indicates that subtle changes of amino acids may be sufficient to change substrate selectivity.
Little information is available on structural and functional divergence amongst these plant thioesterases, and the tertiary structure of any plant thioesterase has yet to be det~rm;ne~. Protein engineering may prove to be a powerful tool for underst~n~;ng the merh~n;sm of thioesterase substrate recognition and catalysis, and thus SU~3~TITUT~ SHE:T ~ LE 26) lead to the rational design of new enzymes with desirable substrate specificities. Such new enzymes would find use in plant bioengineering to provide various modifications of fatty acyl compositions, particularly with respect to production of vegetable oils having significant proportions of desired fatty acyl groups, including medium-chain fatty acyl groups (C8 to C14) and longer chain fatty acyl groups (C16 or C18). In addition, it is desirable to control the relative proportions of various fatty acyl groups that are present in the seed storage oil to provide a variety of oils for a wide range of applications.

Literature The strategy of using ch;m~ric gene products has been applied to study the structure and function of phosphotransferases in yeast (Hjelmstad et al. (1994) ~.
Biol. Chem. 269: 20995-21002) and restriction endonucleases of Flavobacterium Kim et al. (1994) Proc. Natl. Acad. Sci.
USA. 91: 883-887).
Domain swapping to rearrange functional ~mAinq of proteins has been used in protein engineering (Hedstrom (1994) Current Opinion in Structural Biology 4:608-611).
Recently the structure of a myristoyl-ACP thioesterase from Vi~rio harveyi has been determined (Lawson et al.
(1994) Biochemist~y 33:9382-9388). This thioesterase, like other bacterial or mAmm~lian thioesterases, shares no sequence homology with plant thioesterases (Voelker et al.
(1992) supra).
DESCRIPTION OF THE FIGURES
Figure 1. An am.ino acid sequence alignment of representative Class I (FatA) and Class II (FatB) thioesterases is provided. UcFatB1 (SEQ ID NO:1) is a California bay C12 thioesterase. CcFatB1 (SEQ ID NO:2) is a cAmphor C14 thioesterase. CpFatB1 (SEQ ID NO:3) is a Cuphea palustris C8 and C10 thioesterase. CpFatB2 (SEQ ID
NO:4) is a Cuphea palustris C14 thioesterase. GarmFatA1 (SEQ ID NO:5) is a mangosteen 18:1 thioesterase which also has considerable activity on C18:0 acyl-ACP substrates.

SU~TITUTE SHEE~ (RU~ E ~B) CA 02206984 l997-06-04 W O96/36719 PCTrUS96/07064 BrFatAl (SEQ ID NO:6) is an 18:1 thioesterase from Brassica rapa (aka Brassica campestris) . Amino acid sequences which are identical in all of the represented thioesterases are indicated by bold shading.
Figure 2. Results o~ thioesterase activity assays of wild-type bay (Figure 2A) and wild-type c~m~hor (Figure 2B) thioesterases upon expression in E. col i is presented.
Figure 3. Nucleic acid and translated amino acid sequence of a PCR fragment (SEQ ID NO:7) cont~;n;ng the encoding region for the mature protein portion o~ a ~m~h~r Class II acyl-ACP thioesterase is pro~ided.
Figure 4. Nucleic acid and translated amino acid sequence (SEQ ID NO:8) o~ a mangosteen Class I acyl-ACP
thioesterase clone (GarmFatAl) is provided. GarmFatAl demonstrates primary thioesterase activity on 1~:1 acyl-ACP
substrate, but also demonstrates considerable activity on 18:0 substrate (approximately 10-20% of 18:1 activity).
Figure 5. Nucleic acid and translated amino acid sequence (SEQ ID N0:9) of a mangosteen Class I acyl-ACP
thioesterase clone, GarmFatA2, is pro~ided. GarmFatA2 has thioesterase activity primarily on 18:1 acyl-ACP substrate, and equally low activity on 16:0 and 18:0 substrates.
Figure 6. Nucleic acid and translated amino acid sequence (SEQ ID NO:10) of a ~uphea palustris Class II
acyl-ACP thioesterase clone (CpFatBl) having preferential activity on C8 and C10 acyl-ACP substrates is provided.
Figure 7. Nucleic acid and translated amino acid sequence (SEQ ID NO:ll) of a Cuphea palustris Class II
acyl-ACP thioesterase clone (CpFatB2) having preferential activity on C14 acyl-ACP substrates is provided.
Figure 8. An amino acid sequence comparison of bay (C12) (SEQ ID NO:l) and c~mphor (C14) (SEQ ID NO:2) acyl-ACP thioesterases is provided. Amino acid residues which differ between the thioesterases are indicated by bold ~h~; ng, Figure 9. Bay/camphor ch;m~ric constructs, Ch-l and Ch-2, are shown as in-frame fusions of N- and C-t~rm; n~i portions of the thioesterases (from left to right). The SUi3~TITUTE S~Ç~tr (P~UI E ~) W O96/36719 PCTAUS9G/07064 KpnI site used in constructing the ch;mPric constructs is shown.
Figure 10. An amino acid sequence comparison of C.
palustris CpFatBl (C8/C10) (SEQ ID NO:3) and C. palustris CpFatB2 (C14) (SEQ ID NO:4) acyl-ACP thioesterases is provided. Amino acid residues which differ between the thioesterases are indicated by bold .~h~A; ng .
Figure 11. Substrate specificities of the bay/~m~hor ~hlm~ric enzymes and two bay mutant thioesterases are provided (dark shaded columns). Control (E. coli transformed with vector alone) background activities are indicated by the light hatched columns. (A) Ch-l (B) Ch-2 (C) bay mutant M197R/R199H, and (D) bay mutant M197R/R199H/T231K.
Figure 12. Relative thioesterase activity of wild-type (5247) and mutant Garcinia mangifera thioesterases (GarmFatAl) on 18:1, 18:0 and 16:0 acyl-ACP substrates are provided.
Figure 13. An amino acid sequence comparison of B.
rapa BrFatAl (C18:1) (SEQ ID NO:6) and Garcinia mangifera GarmFatAl (C18:1/C18:0) (SEQ ID NO:5) acyl-ACP
thioesterases is provided. Amino acid residues which differ between the thioesterases are indicated by bold .~h~A;ng Figure 14. Short ~nm~;n-swapping by PCR. The full-length gene is shown by two long, parallel lines. The hatched area represents the Aom~;n of interest. For each PCR primer (a, b, c, and d), an arrow-head is pointing to the 3' end. Primers a and b are forward and reverse primers for the full-length DNA. The thin lines in primers c and d represent sequences that exactly match 3' down-stream of the Aom~;n. The thick tails of primers c and d are the 5' overhangs correspo~A;n~ to the new Aom~;n sequence.
Figure 15. Long Anm~;n-swapping by PCR. Two PCR (PCR
1 and ~) are carried out with gene I as template. A third PCR is performed simultaneously with gene II as template.
Primers a and b are forward and reverse primers for the full-length gene I. Primer c matches the sequence ;mm~A;ate SUBSTITUTE SHEET (RULE 26) CA 02206984 l997-06-04 W O96/36719 PCTrUS96/07064 3' down-stream o~ the original clom~;n in gene I. The thin line in primer d represents sequence that matches 3~ down-stream of the original domain in gene I, whereas the thick tail matches the 3' end sequence of the replacement ~om~; n in gene II. Primer e primes the 5' end o~ the ~m~; n in gene II, while f primes the other end. The thin tail in primer f represents sequence that matches 3' down-stream o~
the original domain in gene I.

SUk~L~RY OF THE l~V ~:L~ lON
By this invention, methods of producing engineered plant acyl-ACP thioesterases are provided, wherein said engineered plant acyl-ACP thioesterases ~mon~trate altered substrate specificity with respect to the acyl-ACP
substrates hydrolyzed by the plant thioesterases as compared to the native acyl-ACP thioesterase., Such methods comprise the steps of (1) modifying a gene sequence encoding a plant thioesterase protein targeted for modification to produce one or more modified thioesterase gene sequences, wherein the modified sequences encode engineered acyl-ACP thioesterases having substitutions, insertions or deletions of one or more amino acid residues in the mature portion of the target plant thioesterase, (2) expressing the modified encoding sequences in a host cell, whereby engineered plant thioesterases are produced and, (3) assaying the engineered plant thioesterases to detect those ha~ing desirable alterations in substrate specificity.
Of particular interest for amino acid alterations is the C-t~rm; n~ 1 two thirds portion of plant thioesterase, and more particularly, the region correspon~;ng to amino acids 229 to 285 (consensus numbering above sequences) of plant thioesterase sequences as represented in the sequence alignment of Figure 1. Additionally, the region of from amino acid 285-312 is of interest for modification of -thioesterase substrate specificity towards shorter chain fatty acids such as C8 and C10.

SUBSTITUTE SHEET (RULE 26) W O96/36719 PCTrUS96/07064 Useful information regarding potential modification sites in a targeted thioesterase may be obtained by comparison of related plant acyl-ACP thioesterase amino acid sequences, wherein the compared thioesterases demonstrate different hydrolysis activities. Comparisons of plant thioesterase amino acid sequences having at least 75% sequence identity in the mature protein region are particularly useful in this regard. In this manner, amino acid residues or peptide ~om~;n.S which are different in the related thioesterases may be selected for mutagenesis.
Other methods for selecting amino acids or peptide ~om~'n~ for modification include analysis of thioesterase protein sequences for predicted effects of substitutions, insertions or deletions on flexibility and/or secondary structure of the target thioesterase.
In addition, useful thioesterase gene mutations may be discovered by random mutation of plant acyl-ACP
thioesterase encoding sequences, followed by analysis of thioesterase activity or fatty acid composition to detect alterations in substrate specificity.
To produce an engineered thioesterase, a DNA sequence encoding the thioesterase may be altered by ~om~; n swapping or mutagenesis, either random or site-directed, to introduce amino acid substitutions, insertions or deletions. The DNA sequences may then be expressed in host cells for pro~uction of engineered thioesterases and for analysis of resulting fatty acid compositions. Engineered thioesterases produced in this m~nn~r are also assayed to determine effects of the amino acid sequence modifications on the substrate specificity of the thioesterase. In this manner, novel thioesterases may be discovered which ~m~n ~trate a variety of profiles with respect to the carbon chain lengths of the acyl-ACP substrates which may be hydrolyzed or with respect to the relative activity of the thioesterase on different carbon chain length acyl-ACP
substrates.
Thus, DNA sequences and constructs for expression of engineered thioesterases, as well as the novel SU~SrlTUTE SHEET (RULE 26) CA 02206984 l997-06-04 W O96/36719 PCT~US96/07064 thioesterases produced therefrom are also considered within the scope of the invention described herein. Such DNA
seguences may be used for expression of the engineered thioesterases in host cells for the modification of fatty acid composition. Of particular interest in the instant invention are DNA constructs for expression of engineered thioesterases in plant cells, especially in plant seed cells of oilseed crop plants. As the result of expression of such constructs, plant triglyceride oil may be produced, wherein the composition of the oil reflects the altered substrate specificity of the engineered thioesterases.
Thus, plant cells, seeds and plants comprising the constructs provided herein are all encompassed by the instant invention, as well as novel plant oils that may be harvested from the plant seeds.
For exa-m-ple/ a C12 preferring plant acyl-ACP
thioesterase described herein may be altered to obtain a plant thioesterase having approximately equal activity on C14 and C12 substrates. Further modification of the C12 enzyme yields a thioesterase having greater activity on C14 as compared to C12 substrates.
Also provided in the instant invention are novel plant acyl-ACP thioesterase se~uences from Cuphea pal ustris and mangosteen ( Garcinia mangifera). The C. palustris sequence, CpFatBl, ~m~n~trateS substrate specificity towards C8 and C10 fatty acyl-ACPs with higher activity on C8. A mangosteen thioesterase gene, GarmFatAl, ~mnn.~trates primary activity on 18:1-ACP substrates, but also ~mo~trates substantial activity on 18:0-~CP.
Importantly, this clone does not ~mon~trate specificity for 16:0 substrates. Methods of modifying the specificity of these novel C8/C10 and C18:1/C18:0 plant thioesterases are also provided in the instant invention. In particular, mutations which increase the 18:0/18:1 activity ratio of the mangosteen clone are provided.

DET~TT-~n DESCRIPTION OF THE lNv~L.llON

Sl)B~TIME SH~ET (RULE 2~) W O96/36719 PCTrUS96/07064 By this invention methods to produce engineered plant thioesterases having altered substrate specificity are provided. An engineered plant thioesterase of this invention may include any sequence of amino acids, such as a protein, polypeptide or peptide fragment obtA;nAhle from a plant source which ~mon~trates the ability to catalyze the production of free fatty acid(s) from fatty acyl-ACP
substrates under plant enzyme reactive conditions. By "enzyme reactive conditions" is meant that any necessary conditions are available in an environment (i.e., such factor.s as temperature, pH, lack of inhibiting substances) which will permit the enzyme to function.
Engineered plant thioesterases may be prepared by random or specific mutagenesis of a thioesterase encoding sequence to provide for one or more amino acid substitutions in the translated amino acid sequence.
Alternatively, an engineered plant thioesterase may be prepared by ~omA;n swapping between related plant thioesterases, wherein extensive regions of the native thioesterase encoding sequence are replaced with the correspon~;ng region from a different plant thioesterase.
Targets for ~omA1n swapping can include peptides ranging from five or six to tens of amino acids in length.
In an ideal case, this type of interchange can be accomplished by the presence of unique, conserved restriction sites at the exact points of exchAnge in the genes encoding both proteins. Oligo-based mutagenesis (looping) may be applied when convenient restriction sites are not available, although this process may be time-consuming when large ~omA; n sequences are to be swapped.
Alternatively, as described in the following Examples, a rapid method for ~nmA; n swapping may be employed which is a modification of an overlap extension technique using polymerase chain reaction (PCR) described by Horton et al .
(BioTechnigues (1990) 8:528-535). The entire procedure can be done within six hours (time for two PCR runs) without in vivo manipulation. The basis for the overlap extension method is that in a PCR the primers must match their Sl~TlTlJTE SHEEr (RVLE 2~

W O96136719 PCTrUS96107064 template sequence well enough to prime, but they need not match exactly, especially toward the 5' end. In fact, PCR
primers with 5' overhangs (non-match sequences) are routinely used. The PCR-based ~m~; n swapping is designed for applications where the ~om~;n contains about six amino acids or less (short ~m~ ; n swapping), or where ~nm~ ; n.~
cont~; n; ng much larger numbers of amino acids are to be swapped (long ~om~; n swapping).
Altered substrate specificities of an engineered thioesterase may be reflected by the presence of hydrolysis activity on an acyl-ACP substrate of a particular chain length which is not hydrolysed by the native thioesterase enzyme. The newly recognized acyl-ACP substrate may differ from native substrates of the enzyme in various ways, such as by having a shorter or longer carbon chain length (usually reflected by the addition or deletion of one or more 2-carbon units), by having a greater or lesser degree of saturation, or by the presence of a methyl group, such as in certain fatty acids which are not c~mmonly present in plant cells, i.e. iso- and anti-iso- fatty acids.
Alternatively, altered substrate specificity may be reflected by a modification of the relative hydrolysis activities on two or more acyl-ACP substrates of differing chain length and/or degree of saturation.
DNA and amino acid sequence information for ~lore than thirty plant acyl-ACP thioesterases is now available, and these sequences may be used in the methods of the instant invention to identify desirable regions for modification to produce sequences for expression of engineered thioesterases.
Plant thioesterases can be classified into two classes by sequence homology. All of these plant thioesterases contain a transit peptide, of 60 to 80 amino acids in length, ~or plastid targeting. The transit peptides bear little homology between species while the mature protein regions (minus transit peptide) show significant amino acid sequence identity.

SUB~ITL~E $HEET (RU~E 28J

W O96/36719 PCTrUS96/07064 The ~irst class, Class I (or FatA) includes long chain acyl-ACP thioesterases having activity primarily on 18:1-ACP. 18:1-ACP is the ;mme~;~te precursor of most fatty acids found in phospholipids and tyiglycerides synthesized by the eukaryotic pathway. This class of thioesterase has been found in essentially all plant sources ex~m; ne~ to date, and is suggested to be an essential "housekeeping~
enzyme (Jones et al. (supra) required for membrane biosynthesis. Examples of Class I thioesterases from safflower, Cuphea hookeriana and Brassica rapa (campestris), which have activity primarily on 18:1-ACP
substrate, have been described (WO 92/20236 and WO
94/10288). Other 18:1 thioesterases have been reported in Arabidopsis thaliana (Dormann et al. (1995) Arch. Biochem.
Biophys. 316:612-618), Brassica napus (Loader et al. (1993) Plant Mol. Biol. 23:769-778) and cor; ~n~e~ (Dormann et al.
(1994) Biochem. Biophys. Acta 1212:134-136). A similar 18:1-ACP specific Class I thioesterase (GarmFatA2) has been discovered in developing embryos from mangosteen ( Garcinia mangifera), and is described herein. A Class I
thioesterase from soybean (WO 92/11373) was reported to provide 10- and 96-fold increases in 16:0-ACP and 18:1-ACP
activity upon expression in E. coli, and a smaller (3-4 fold) increase in 18:0-ACP activity. The mature protein regions of Class I plant thioesterases are highly homologous, ~mo~trating greater than 80% sequence identity.
In addition, another mangosteen Class I thioesterase (GarmFatA1), also described herein, has been discovered which ~m~trateS thioesterase activity primarily on 18:1-ACP substrates (100-fold increase upon expression in E.
coli), but also demonstrates selective activity on 18:0-ACP
versus 16:0-ACP. The 18:0 activity of GarmFatA1 is approximately 25% of the 18:1 activity, whereas in most Class I thioesterases analyzed to date, the 18:1 activity is highly pre~om;n~nt~ with activity on 16:0 and 18:0 substrates detectable at less than 5~ of the 18:1 activity levels.

SUB~ITUTE SHEET (RULE 26) CA 02206984 l997-06-04 W O96/36719 PCTrUS96/07064 A second class of plant thioesterases, Class II (or FatB) thioesterases, includes enzymes that utilize fatty acids with shorter chain-lengths, from C8:0 to C14:0 (medium chain fatty acids) as well as C16:0. Class II
thioesterases preferably catalyze the hydrolysis of substrates cont~;n;ng saturated fatty acids. Class II (or FatB) thioesterases have been isolated from California Bay, elm, Cuphea hookeriana, Cup~ea palustris, Cuphea lanceolata, nutmeg, Arabidopsis thaliana, mango, leek and camphor. The mature protein regions of Class II plant thioesterases are also highly homologous, demonstrating 70-80% sequence identity.
One of the characteristics of Class II thioesterases is the presence of a relatively hydrophobic region of approximately 40 amino acids in the N-t~rmi n~l region of the mature proteins. This hydrophobic region is not found in 18:1-ACP thioesterases, and has no apparent e~fect on the enzyme activity. Recombinant expression of a bay Class II thioesterase with or without this region showed identical activity profiles in vitro (Jones et al.
(supra)).
~ s ~m~n~trated more fully in the following examples, the acyl-ACP substrate specificity of plant thioesterases may be modified by various amino acid changes to the protein sequence, such as amino acid substitutions, insertions or deletions in the mature protein portion of the plant thioesterases. Modified substrate specificity can be detected by expression of the engineered plant thioesterases in E. coli and assaying to detect enzyme activity.
Modified substrate specificity may be indicted by a shift in acyl-ACP substrate preference such that the engineered thioesterase is newly capable of hydrolysing a substrate not recognized by the native thioesterase. The newly recognized substrate may vary from substrates of the native enzyme by carbon chain length and/or degree of saturation of the fatty acyl portion of the substrate.
Alternatively, modified substrate specificity may be Sl)B~Tl~UTE SHEET (RlJLE 261 W O96/36719 PCTrUS96/07064 reflected by a shift in the relative thioesterase activity on two or more substrates o~ the native thioesterase such that an engineered thioesterase exhibits a different order of preference for the acyl-ACP substrates.
For example, a plant thioesterase having primary hydrolysis activity on C12 substrate and some minor activity on C14 substrate may be modified to produce an engineered thioesterase which exhibits increased activity on C14, for example so that the engineered thioesterase has approximately equal activity on C12 and C14 substrates.
Similarly, such plant C12 thioesterases may be further modified to produce an engineered thioesterase having primary activity on C14 substrates and little or no activity on C12 substrates. Alternatively, a plant thioesterase may be modified so as to alter the relative activity towards a substrate having higher or lesser degree of saturation. For example, a Class I (18:1) thioesterase may be modified to increase the relative activity on C18:0 substrates as compared to activity on other substrates of the enzyme, such as C18:1 and C16:0. Examples of these types of thioesterase modifications are provided in the following exam~ples. Further modification of plant thioesterases are also desirable and may be obtained using the methods and sequences provided herein. For example, plant thioesterases may be modified to shift the enzymatic activity towards hydrolysis of shorter chain fatty acids, such as C8 and C10. Comparison of closely related thioesterase sequences, such as the C. palustris C8/10, the C. palustris C14 and the C. hookeriana C8/10 thioesterase sequences provided herein may be used to identify potential target amino acid residues for alteration of thioesterase specificity.
In initial experiments aimed at altering substrate specificity of plant thioesterase enzymes, two highly related Class II thioesterases were studied, a C12 preferring acyl-ACP thioesterase from California bay (Umbellularia californica) and a C14 preferring acyl-ACP
thioesterase from c~mphor (C;nnAmom77m cAm~7ora). These SUB~ltiJTE SHEET (~E 2~) W O96/36719 PCTrUS96/07064 enzymes ~m~trate 90% amino acid sequence identity in the mature protein region yet have different substrate speci~icities. Constructs for expression o~ chimeric mature thioesterases were prepared which encoded ~h;m~riC
thioesterase enzymes contA;n;ng the N-t~rm;nAl mature protein region of either the camphor or bay thioesterase and the C-t~rm;n~l portion of the other thioesterase. The N-t~rm;nAl thioesterase portion as encoded in these constructs contains approximately one third of the mature thioesterase protein, and the C-t~rm; n~ 1 portion contains the r~mA;n;n~ two thirds of the mature thioesterase re~ion.
As described in more detail in the ~ollowing examples, we have discovered that the C-t~rm--; n~ 1 two thirds portion of these plant thioesterases is critical in determ;nin~ the substrate speci~icity. The ch;m~ic enzyme cont~;n;ng the C-term;nAl portion of the c-Amp~or thioesterase (Ch-l) ~ trates the same activity profile as native cAmphor thioesterase (specific for 14:0), and the ch;m~ric protein with the bay thioesterase C-t~rm;nll~ tCh-2) ~mon~trates the same activity profile as native bay thioesterase (12:0 specific).
Additional studies of the C-t~rm;nA1 end of the protein were conducted to further locate regions of thioesterase proteins critical for substrate specificity.
In one such study, the 13 consecutive C-term; n~l ~mino acids of the bay thioesterase were deleted by procluction of a mutant gene lacking the coding DNA for this region. The activity of the expressed mutant thioesterase was compared to an expressed wild-type bay thioesterase protein. The activity profiles of the 17 C-t~rm;nAl meutant and the wild type bay thioesterase proteins were the same, ~mon~trating that the very C-t~rm;nAl end of thioesterase proteins is ~ not a critical region for substrate specificity.
Further analysis of the C-t~rm;nAl two thirds portion~ 35 of the bay C12 preferring acyl-ACP thioesterase was conducted to identify particular amino acids involved in substrate specificity. By e~Am; n; ng a sequence alignment of the bay and cAmphor thioesterases, the least SVBSrJTUTE SI~ET (RULE 2~;) CA 02206984 l997-06-04 W O96/36719 PCTrUS96/07064 conservative amino acid substitutions between the two thioesterases in the C-t~rm;n~l two thirds portion of the proteins were identified. Non-conservative amino acid substitutions include those in which the substituted amino acid has a different charge than the native amino acid residue. Amino acids considered as having positively charged side c~;n~ at pH 7 are lysine and arginine.
Histidine can also have a positively charged side chain under conditions of acidic pH. Amino acids considered as having negatively charged side ch~; n~ at pH 7 are aspartate and glutamate. Non-conservative amino acid substitutions may also be indicated where the size of the substituted amino acid differs considerably from the size of the amino acid normally located at that position. Examples of non-conserved amino acid di~ferences between the bay and camphor thioesterases are M197 -> R (Bay TE -> C~mr~or TE), Rl99 -> H, T231 -> K, A293 -> D, R327 -> Q, P380-> S,and R381 -> S (amino acid sequence numbering for bay and c~m~or thioesterases is shown in Figure 8).
Secondary structure predictions may be used to identify amino acid substitutions likely to have affects on the secondary structure of the thioesterase protein. For example, according to secondary structure predictions using methods of Chou and Fasman, the tripeptide M-R-R amino acids 197-199 of bay and the corresponding tripeptide R-R-H
of c~m~h~r are located behind a g-sheet and a turn anchored by two highly conserved glycines (G193 and G196). This region o~ plant thioesterases is highly conserved, and the ~-sheet and a turn structure is also predicted in other plant thioesterases.
As described in the following examples, when the bay M-R-R tripeptide is changed to R-R-H, mimicking the sequence in c~m~or thioesterase, the activity of the mutant towards 12:0, but not 14:0, is reduced about 7 fold compared to the wild type. This results in an engineered thioesterase which has approximately equal specific activity with respect to the 12:0 and 14:0 substrates.

St~TI~ St~EET (RULE 26~

W O96/36719 PCTrUS96/07064 An additional modification of the engineered bay M197R/R199H thioesterase which converts the threo~ine residue at amino acid 231 to a lysine (T231K) alters the substrate specificity such that the engineered thioesterase M197R/R199H/T231K is highly 14:0-ACP speci~ic.
Interestingly, the mutation T231K alone does not affect the bay thioesterase activity. The non-additive, combinatorial ef~ect o~ the T231K substitution on M197R/R199H engineered thioesterase suggests that the altered amino acid sites are folded close to each other (Sandberg, et al . (1993) Proc.
Natl. Acad. Sci. 90: 8367-8371).
As described in the following Examples, amino acid substitutions near the active site (YRREC, amino acids 357-361 in Figure 1 consensus numbering) of the plant acyl-ACP
thioesterases may result in large reductions in thioesterase activity. Modification of bay thioesterase to produce R327Q results in a 100-fold decrease in the bay thioesterase activity. The decreased activity of R327Q is likely due to the fact that this amino acid position is located very close to the active site cysteine, C320 of the bay thioesterase sequence in Figure 8.
Expression of engineered thioesterases having altered substrate specificities in host cells and analysis of resulting fatty acid compositions ~m~n~trates that the altered substrate specificities of the engineered thioesterases are reflected in the fatty acid composition profiles of the host cells. This is significant because enzyme activity in vivo might have involved sequential interactions or parameters such as lifetime and folding/unfolding rates which would not be reflected in in vitro activity assays. The major lipid components of E. coli membranes are phosphatidyl-ethanolamine and ~ phosphatidylglycerol, which contain pre~m;n~ntly long chain fatty acyl moieties. Recombinant expression of~ 35 native bay thioesterase cDNA in fadD cells redirects the bacterial type II fatty acid synthase system from long-chain to medium-chain production, and similar results are obtained upon expression o~ native bay thioesterase in JTE Sl IEET (RULE 2 CA 02206984 l997-06-04 W O96/36719 PCTrUS96/07064 seeds of transgenic plants (Voelker et al . ( 1994) supra;
Voelker et al . ( 1992) supra) . Thus, E. coli in vivo data may be used to predict the effects of expression of engineered thioesterases in transgenic plants.
With native bay thioesterase, E. coli fadD cells produce large amounts of 12:0 free fatty acid and small amounts of 14:0 (about 5 to 10~ of 12:0 levels) (Voelker et al. (1994) and Table I). However, as demonstrated in the following examples, following two amino acid substitutions 10 (M197R/R199H), expression of an engineered bay thioesterase enzyme results in accumulation of similar amounts of 12:0 and 14:0 fatty acids. Similarly, expression of the engineered bay thioesterase with three amino acid substitutions (M197R/R199H/T231K) completely reverses the 15 12:0/14:0 ratio of fatty acids produced as compared to results with native bay thioesterase.
Thus, as the result of modifications to the substrate specificity of plant thioesterases, it can be seen that the relative amounts of the fatty acids produced in a cell where various substrates are available for hydrolysis may be altered. Furthermore, molecules which are formed from available free fatty acids, such as plant seed triglycerides, may also be altered as a result of expression of engineered thioesterases having altered substrate specificities.
In addition to known acyl-ACP thioesterases and encoding sequences, such as provided herein, other acyl-ACP
thioesterase sequences may be obtained from a variety of plant species, and such thioesterases and encoding sequences will find use in the methods of this invention.
As noted above, plant thioesterase encoding sequences are highly conserved, particularly for those thioesterases which are members of the same class of thioesterase, i.e.
Class I or Class II. Thus, for isolation of additional thioesterases, a genomic or other appropriate library prepared from a candidate plant source of interest is probed with conserved sequences from one or more Class I or Class II plant thioesterase sequences to identify Sl)~llUTE S~EET ~R~E 26) W O96136719 PCT~US96/07064 homologously related clones. Positive clones are analyzed by restriction enzyme digestion and/or sequencing. Probes can also be considerably shorter than the entire sequence.
Oligonucleotides may be used, ~or example, but should be at least about 10, preferably at least about 15, more preferably at least 20 nucleotides in length. When shorter length regions are used for comparison, a higher degree of sequence identity is required than for longer sequences.
Shorter probes are often particularly useful for polymerase chain reactions (PCR) (Gould, et al., PNAS USA (1989) 86:1934-1938), especially for isolation of plant thioesterases which contain highly conserved sequences.
~ PCR using oligonucleotides to conserved regions of plant thioesterases may also be used to generate homologous probes for library screening.
When longer nucleic acid fragments are employed (>100 bp) as probes, especially when using complete or large cDNA
sequences, one can still screen with moderately high stringencies (for example using 50% formamide at 37~C with m;n;m~l w~ch;ng) in order to obtain signal from the target sample with 20-50% deviation, i.e., homologous sequences.
(For additional information regarding screening techniques see Beltz, et al. Methods in Enz~ ology (1983) 100:266-285.).
The nucleic acid or amino acid sequences encoding an engineered plant acyl-ACP thioesterase of this in~ention may be combined with other non-native, or "heterologous", sequences in a variety of ways. By "heterologous"
sequences is meant any sequence which is not naturally found joined to the plant acyl-ACP thioesterase, including, for example, combinations of nucleic acid sequences from the same plant which are not naturally found joined together.
For expression in host cells, sequence encoding an engineered plant thioesterase is combined in a DNA
construct having, in the 5' to 3' direction of transcription, a transcription initiation control region capable of promoting transcription and translation in a SU~I I I ~JTE S~FeT (RU~E 2&) W O96/36719 PCTrUS96/07064 host cell, the DNA sequence encoding the engineered plant acyl-ACP thioesterase and a transcription and translation term;n~tion region.
DNA constructs may or may not contain pre-processing sequences, such as transit peptide sequences. Transit peptide sequences facilitate the delivery of the protein to a given organelle and are cleaved from the amino acid moiety upon entry into the organelle, releasing the "mature" sequence. The use of the precursor plant acyl-ACP
thioesterase DNA sequence is preferred in plant cell expression cassettes. Other plastid transit peptide sequences, such as a transit peptide of seed ACP, may also be employed to translocate plant acyl-ACP thioesterases to various organelles of interest.
Thus, engineered plant thioesterase sequences may be used in various constructs, such as for expression of the thioesterase of interest in a host cell for recovery or study of the enzyme in vitro or in vivo. Potential host cells include both prokaryotic and eukaryotic cells. A
host cell may be unicellular or found in a multicellular differentiated or undifferentiated organism dep~n~;ng upon the intended use. Cells of this invention may be distinguished by having an engineered plant acyl-ACP
thioesterase present therein.
DepPn~;ng upon the host, the regulatory regions will vary, including regions from viral, plasmid or chromosomal genes, or the like. ~or expression in prokaryotic or eukaryotic microorg~n; ~sms, particularly unicellular hosts, a wide variety of constitutive or regulatable promoters may be employed. Expression in a microorganism can provide a ready source of the engineered plant enzyme and is useful for identifying the particular characteristics of such enzymes. Among transcriptional initiation regions which have been described are regions from bacterial and yeast hosts, such as E. col i, B . subtil is, Saccharomyces cerevisi.ae, including genes such as beta-galactosidase, T7 polymerase, tryptophan E and the like.

StlBSrlllJ~E SHEET (RULE 28 W O96/36719 PCTAUS96/07064 For the most part, the constructs will involve regulatory regions functional in plants which provide for expression of the plant acyl-ACP thioesterase, and thus result in the modification of the fatty acid composition in plant cells. The open reading frame, coding for the plant acyl-ACP thioesterase will be joined at its 5' end to a transcription initiation regulatory region such as the wild-type sequence naturally found 5~ upstream to the thioesterase structural gene. Numerous other transcription initiation regions are available which provide for a wide variety of constitutive or regulatable, e.g., inducible, transcription of the structural gene functions. Among transcriptional initiation regions used for plants are such regions associated with the structural genes such as for nopaline and mannopine synthases, or with napin, ACP
promoters and the like. The transcription/translation initiation regions corresponding to such structural genes are found ;mmeAiAtely 5' upstream to the respective start codons. In embodiments wherein the expression of the engineered thioesterase protein is desired in a plant host, the use o~ part of the native plant acyl-ACP thioesterase gene is considered. Namely, all or a portion of the 5' upstream non-coding regions (promoter) together with 3' downstream non-coding regions may be employed. If a different promoter is desired, such as a promoter native to the plant host of interest or a modified promoter, i.e., having transcription initiation regions derived from one gene source and translation initiation regions derived from a different gene source (enh~nced promoters), such as double 35S CaMV promoters, the sequences may be joined together using st~nA~d techniques.
For such applications when 5' upstream non-coding ~ regions are obtained from other genes regulated during seed maturation, those preferentially expressed in plant embryo tissue, such as ACP and napin-derived transcription initiation control regions, are desired. Such "seed-specific promoters" may be obt~;neA and used in accordance with the teachings of U.S. Serial No. 07/147,781, filed SUB~llUlE Stl~ET (RULE 2~) W O96/36719 PCTrUS96/07064 1/25/88 (now U.S. Serial No. 07/550,804, filed 7/9/90), and U.S. Serial No. 07/494,722 filed on or about March 16, 1990 having a title "Novel Sequences Pre~erentially Expressed In Early Seed Development and Methods Related Thereto," which references are hereby incorporated by reference.
Transcription initiation regions which are preferentially expressed in seed tissue, i.e., which are undetectable in other plant parts, are considered desirable for fatty acid modifications in order to m;n;m;ze any disruptive or adverse effects of the gene product.
Regulatory transcript t~rm;n~tion regions may be provided in DNA constructs of this invention as well.
Transcript t~rm;n~tion regions may be provided by the DNA
sequence encoding the plant acyl-ACP thioesterase or a convenient transcription term;n~tion region derived from a different gene source, for example, the transcript t~rm;n~tion region which is naturally associated with the transcript initiation region. Where the transcript t~rm;n~tion region is from a different gene source, it will contain at least about 0.5 kb, preferably about 1-3 kb of sequence 3' to the structural gene from which the t~rm;n~tion region is derived.
Plant expression or transcription constructs having a plant acyl-ACP thioesterase as the DNA sequence of interest may be employed with a wide variety of plant life, particularly, plant life involved in the production of vegetable oils for edible and industrial uses. Most especially preferred are temperate oilseed crops. Plants of interest include, but are not limited to, rapeseed tCanola and High Erucic Acid varieties), sunflower, safflower, cotton, Cuphea, soybean, peanut, coconut and oil palms, and corn. Dep~n~ng on the method for introducing the recombinant constructs into the host cell, other DNA
sequences may be required. Importantly, this invention is applicable to dicotyledon and monocotyledon species alike and will be readily applicable to new and/or improved transformation and regulation techniques.

SUB~T~lU~E SI~LET (RULE 2~) W O96/36719 PCTrUS96107064 The method of transformation is not critical to the instant invention; various methods of plant trans~ormation are currently available. As newer methods are available to transform crops, they may be directly applied hereunder.
For example, many plant species naturally susceptible to Agrobacterium infection may be successfully transfor-m-ed via tripartite or binary vector methods of Agrobacterium mediated transformation. In addition, techniques of microinjection, DNA particle bombardment, electroporation have been developed which allow for the transformation of various monocot and dicot plant species.
In developing the DNA construct, the various components of the construct or fragments thereof will normally be inserted into a convenient cloning vector which is capable of replication in a bacterial host, e.g., E.
coli. Numerous vectors exist that have been described in the literature. After each cloning, the plasmid may be isolated and subjected to further manipulation, such as restriction, insertion of new fragments, ligation, deletion, insertion, resection, etc., so as to tailor the components of the desired sequence. Once the construct has been completed, it may then be transferred to an appropriate vector for further manipulation in accordance with the manner of transformation of the host cell.
Normally, included with the DNA construct will be a structural gene having the necessary regulatory regions for expression in a host and providing for selection of transformant cells. The gene may provide for resistance to a cytotoxic agent, e.g. antibiotic, heavy metal, toxin, etc., complementation providing prototrophy to an auxotrophic host, viral ;mmlln;ty or the like. Dep~n~;ng upon the number of different host species the expression ~ construct or components thereof are introduced, one or more markers may be employed, where different conditions for ~ 35 selection are used for the different hosts.
It is noted that the degeneracy of the DNA code provides that some codon substitutions are permissible of S(JB~TI~E S~EET (P~ULE 28~

W O96/36719 PCTrUS96/07064 DNA sequences without any correspnn~;n~ modification of the amino acid sequence.
The manner in which the DNA construct is introduced into the plant host is not critical to this invention. Any method which provides for efficient trans~ormation may be employed. Various methods for plant cell transformation include the use of Ti- or Ri-plasmids, microinjection, electroporation, DNA particle bombardment, liposome fusion, DNA bom~bardment or the like. In many instances, it will be desirable to have the construct bordered on one or both sides by T-DNA, particularly having the left and right borders, more particularly the right border. This is particularly useful when the construct uses A. tumefaciens or A. rhizogenes as a mode for trans~ormation, although the T-DNA borders may find use with other modes of transformation.
Where Agrobacterium is used for plant cell transformation, a vector may be used which may be introduced into the Agrobacterium host for homologous recombination with T-DNA or the Ti- or Ri-plasmid present in the Agrobacterium host. The Ti- or Ri-plasmid cont~;n;ng the T-DNA for recombination may be armed (capable of causing gall formation) or disarmed (incapable of causing gall formation), the latter being permissible, so long as the vir genes are present in the transformed Agrobacterium host. The armed plasmid can give a mixture of norm~al plant cells and gall.
In some instances where Agrobacterium is used as the vehicle for transforming plant cells, the expression construct bordered by the T-DNA border(s) will be inserted into a broad host spectrum vector, there being broad host spectrum vectors described in the literature. Commonly used is pRK2 or derivatives thereof. See, for example, Ditta et al., PNAS USA, (1980) 77:7347-7351 and EPA 0 120 515, which are incorporated herein by reference. Included with the expression construct and the T-DNA will be one or more mar]cers, which allow for selection of transformed Agrobacterium and transformed plant cells. A number of SU~S1111~ S~ET (Rt.'LE 2~i) W O96/36719 PCTrUS96/07064 markers have been developed for use with plant cells, such as resistance to chlor~mrh~n;col, the aminoglycoside G418, hygromycin, or the like. The particular marker employed is not essential to this invention, one or another marker being preferred depen~;n~ on the particular host and the manner of construction.
Once a transgenic plant is obtained which is capable of producing seed having a modi~ied fatty acid composition, traditional plant breeding techniques, including methods of mutagenesis, may be employed to further manip~llate the fatty acid composition. Alternatively, additional forei~n fatty acid modifying DNA seguence may be introduced via genetic engineering to further manipulate the fatty acid composition. It is noted that the method of transformation is not critical to this invention. However, the use of genetic engineering plant transformation methods, i.e., the power to insert a single desired DNA sequence, is critical.
Heretofore, the ability to modify the fatty acid composition of plant oils was limited to the introduction of traits that could be sexually transferred during plant crosses or viable traits generated through mutagenesis.
Through the use of genetic engineering techniques which permits the introduction of inter-species genetic information and the means to regulate the tissue-specific expression of endogenous genes, a new method is available for the production of plant seed oils with modified fatty acid compositions. In addition, there is the potential for the development of novel plant seed oils upon application of the tools described herein.
One may choose to provide for the transcription or transcription and translation of one or more other sequences of interest in concert with the expression of an engineered plant acyl-ACP thioesterase in a plant host cell. In particular, the expression of a plant LPAAT
protein having activity on medium-chain or ver~ long-chain fatty acids in combination with expression of an engineered plant acyl-ACP thioesterase may be preferred in some Sl~BSI ITl~TE ~HEET (RULE 26 CA 02206984 l997-06-04 W O96/36719 PCTrUS96/07064 applications. See WO 95/27791 for plant LPAAT encoding sequences.
When one wishes to provide a plant transformed for the combined effect of more than one nucleic acid sequence of interest, typically a separate nucleic acid construct will be provided for each. The constructs, as described above contain transcriptional or transcriptional or transcriptional and translational regulatory control regions. One skilled in the art will be able to det~rm; n~
regulatory sequences to provide for a desired t; m; ng and tissue specificity appropriate to the final product in accord with the above principles set forth as to the respective expression or anti-sense constructs. When two or more constructs are to be employed, whether they are both related to the same fatty acid modifying sequence or a different fatty acid modifying se~uence, it may be desired that different regulatory sequences be employed in each cassette to reduce spontaneous homologous recombination between se~uences. The constructs may be introduced into the host cells by the same or different methods, including the introduction of such a trait by crossing transgenic plants via traditional plant breeding methods, so long as the resulting product is a plant having both characteristics integrated into its genome.
The invention now being generally described, it will be more readily understood by reference to the following examples which are included for purposes of illustration only and are not intended to limit the present invention.

EXAMPLES
ExamPle 1 Sequences of Plant Acyl-ACP Thioesterases A. California Bay ( Umbellularia californica) DNA sequence and translated amino acid sequence of California bay Class II thioesterase clone pCGN3822 is provided in Figure 1 of WO 92/20236. Expression of the mature portion of the bay thioesterase protein in E. coli and analysis of thioesterase activity reveals a strong specificity of the bay thioesterase for 12: O-ACP substrate, ~3lJBSTlTUTE SI~EET (RULE 26~

CA 02206984 l997-06-04 W O96/36719 PCTrUS96/07064 although some activity towards 14:0-ACP is also observed (Voelker et al. (1994) supra, and Figure 2A herein).
Furthermore, when bay thioesterase is expressed in E. coli fadD cells, large amounts of laurate (more than 500-fold above control background) and small amounts of myristate (about 10% of that of laurate) are produced. Prod~ction of similar ratios of laurate and myristate are also observed upon expression o~ the bay thioesterase in seeds of Brassica napus or Arabidopsis thaliana (Voelker et al.
10 ( 1992 ) supra) .
B. ~mrhor (C;nn~m~m77m c~rhora) DNA sequence and translated amino acid sequence of a Class II c~m~hor thioesterase encoding region generated by PCR is provided in Figure 5B of WO 92/20236. Sequence (SEQ
ID NO:7) of a DNA fragment obt~;ne~ by PCR from reverse transcribed cDNA and cont~; n; n~ the mature protein region of the c~mphor clone is provided in Figure 3. The sequence begins at the XbaI site located at the beg; nn; n~
of the presumed mature protein encoding region of the c~m~hor thioesterase.
The c~mp~or PCR fragment described above is cloned into a pAMP vector resulting in pCGN5219. pCGN5219 is digested with XbaI and SalI and the resulting cAmrh~r thioesterase fragment is cloned into XbaI and SalI digested 25 pBCSK+ (Stratagene), resulting in pCGN5220. pCGN5220 is used to transform E. coli fadD for analysis of acyl-ACP
thioesterase activity as described in Pollard et al. (Arch.
Biochem & Biophys. (1991) 281:306-312). Results of thioesterase activity assays on c~mrhor thioesterase clones 30 using 8:0, 10:0, 12:0, 14:0, 16:0, 18:0 and 18:1 acyl-ACP
substrates ~mo~trate substrate specificity mainly on 14:0 substrates, although a lesser increase in 12:0 hy~rolysis activity is also observed (Fig. 2B).
C. Mangosteen (Garcinia mangifera) ~ 35 A cDNA bank is prepared from seeds extracted from mature mangosteen fruit using the methods as described in Stratagene Zap cDNA synthesis kit (Stratagene; La Jolla, CA). Oil analysis of the mangosteen tissues used for RNA

SU~Sl I l IJTE SHEET ~R~E 26~

W O96/36719 PCTrUS96/07064 isolation reveals 18:0 levels of approximately 50%. Oil analysis of seeds from less mature mangosteen fruit reveals 18:0 levels of 20-40%. Total RNA is isolated from the mangosteen seeds by modifying the CTAB DNA isolation method of Webb and Knapp (Plant Mol. Biol. Reporter (1990) 8:180-195). Buffers include:

REC: 50 mM TrisCl pH 9, O.7 M NaCl, 10 mM EDTA pH8, 0.5% CTAB.
REC+: Add B-mercaptoethanol to 1% ;mme~; Ately prior to use.

RECP: 50 mM TrisCl pH9, 10 mM EDTA pH8, and 0.5%
CTAB.

RECP+: Add B-mercaptoethanol to 1~ ;mm~; Ately prior to use.

For extraction of 1 g of tissue, 10ml of REC+ and 0.5 g of PVPP is added to tissue that has been ground in liquid nitrogen and homogenized. The homogenized material is centrifuged for 10 min at 12000 rpm. The supernatant is poured through miracloth onto 3ml cold chloroform and homogenized again. After centrifugation, 12,000 RPM for 10 min, the upper phase is taken and its volume dete~m;ne~. An equal volume of RECP+ is added and the mixture is allowed to stand for 20 min. at room temperature. The material is centrifuged for 20 min. at 10,000 rpm twice and the supernatant is discarded after each spin. The pellet is dissolved in 0.4 ml of 1 M NaCl (DEPC) and extracted with an equal volume of phenol/chloroform. Following ethanol precipitation, the pellet is dissolved in 1 ml of DEPC
water.
Briefly, the cloning method for cDNA synthesis is as follows. First strand cDNA synthesis is according to Stratagene Instruction MAnllAl with some modifications accordi~g to Robinson, et al.(Methods in Molecular and SU~; 111 ~ITE S~EE~ (RU' Ç: 2~

W 096/3671~ P~ ,"07064 Cellular Biology (1992) 3:118-127). In particular, approximately 57~g of LiC1 precipitated total RNA was used instead of 5~g of poly(A)+ RNA and the reaction was incubated at 45~C rather than 37~C for 1 hour.
Probes for library screening are prepared by PCR from mangosteen cDNA using oligonucleotides to conserved plant acyl-ACP thioesterase regions. Probe Garm 2 and Garm 106 are prepared using the following oligonucleotides. The nucleotide base codes for the below oligonucleotides are as follows:

A = ~n;ne C = cytosine T = thymine U = uracil G = guanine S = guanine or cytosine K = ~lAn;n~ or thymine W = adenine or thymine M = ~en;ne or cytosine R = ~n;ne or gll~n;n~
Y = cytosine or thymine B = guanine, cytosine or thymine H = ~Pn;ne, cytosine or thymine N = A~n;ne, cytosine, guanine or thymine Garm 2 4874: 5' CUACUACUACUASYNTVNGYNATGATGAA 3' tSEQ ID NO:12) 4875: 5' CAUCAUCAUCAURCAYTCNCKNCKRTANTC 3' (SEQ ID NO:13) Primer 4874 is a sense primer designed to correspond to possible encoding sequences for conserved peptide V/L/A W/S/Y V/A M M N, where the one letter amino acid code is used and a slash between amino acids indicates more than one amino acid is possible for that position. Primer 4875 is an antisense primer designed to correspond to possible encoding seguences for peptide D/E Y R R E C.

Garm 106 5424: 5' AUGGAGAu~u~u~AWCRBTAYCCTAMHTGGGGWGA 3' (SEQ ID
NO:14) 5577: 5' ACGCGUACUA~UllNKK~K~CAYTCNGT 3' (SEQ ID NO:15) SU~SI 11 ~TE SHEET (Ri)~E 26) Primer 5424 is a sense primer designed to correspond to possible encoding sequences ~or peptide E/D H/R Y P K/T W G
D.
Primer 5577 is an antisense primer designed to correspond to possible encoding sequences for peptide T E W R K/P K.
The DNA fragments resulting from the above reactions are amplified for use as probes by cloning or by further PCR and radiolabeled by random or specific priming.
Approximately 800,000 plaques are plated according to manufacturer's directions. For screening, plaque filters are prehybridized at room temperature in 50~ formamide, 5X
SSC, lOX Denhardt's, O.1% (w/v) SDS, 5mM Na2EDTA, O.lmg/ml denatured salmon sperm DNA. Hybridization with a mixture of the Garm 2 and Garm 106 probes is conducted at room temperature in the same buffer as above with added 10%(w/v) dextran sulfate and probe. Plaque purification and phagemid excision were conducted as described in Stratagene Zap cDNA Synthesis Kit instructions.
Approximately 90 acyl-ACP thioesterase clones were identified and sorted as to thioesterase type by DNA
sequencing and/or PCR analysis. Of the analyzed clones, at least 28 were Class I (FatA) types, and 59 were Class II
(FatB) types. Two subclasses of FatA type clones were observed, the most pr~m;n~nt type is termed GarmFatAl and the single clone of the second subclass is termed GarmFatA2. DNA and translated amino acid sequence of GarmFatAl clone C14-4 (pCGN5252) (SEQ ID NO:8) is presented in Figure 4. DNA sequence and translated amino acid sequence of the FatA2 clone C14-3 (SEQ ID NO:9) is presented in Figure 5.
Constructs for expression of the Figure 4 Garm FatAl clone in E. coli are prepared as follows. Restriction sites are inserted by PCR mutagenesis at amino acid 49 (SacI), which is near the presumed mature protein amino t~rm;n-l~, and following the stop codon for the protein encoding region (BamHI). The mature protein encoding region is inserted as a SacI/BamHI fragment into pBC SK
(Stratagenei La Jolla, CA) resulting in pCGN5247, which may S13BSTITUTE SHEET (RULE 2~;) CA 02206984 l997-06-04 W O96/36719 PCTnUS96/07064 be used to provide for expression of the mangosteen thioesterase as a lacZ fusion protein.
Results of thioesterase activity assays on mangosteen Class I thioesterase clone GarmFatAl using 16:0, 18:0 and 18:1 acyl-ACP substrates are shown below.

Acyl-ACP Thioesterasè activity (cpm/min) 16:0 18:0 18:1 Control 1400 3100 1733 GarmFatAl 4366 23916 87366 The GarmFatAl cone ~monqtrates preferential activity on C18:1 acyl-ACP substrate, and also ~m~n~trates substantial activity (a~uximately 25% of the 18:1 activity) on C18:0 acyl-ACP substrates. Only a small increase in C16:0 activity over activity in control cells is observed, and the 16:0 activity represents only approximately 3~ of the 18:1 activity.
Expression of GarmFatA2 thioesterase in E. coli and assay of the resultant thioesterase activity ~mo~trates that C18:1 is highly preferred as the acyl-ACP substrate.
The thioesterase activity on 16:0 and 18:0 acyl-ACP
substrates are approximately e~ual and represent less than 5% of the observed 18:1 activity.
D. Brassica campestris (rapa) DNA sequence and translated amino acid sequence of a Brassica campestris Class I acyl-ACP thioesterase are provided in WO 92/20236 (Figure 6).
E. Cuphea palustris C8/C10 Total RNA is isolated from developing seeds of C.
palustris using the modified CTAB procedure described above. A lambda ~;pTox (BRL; Gaithersburg, MD) cDNA
library cont~;n;ng approximately 6 X 106 pfu is constructed from total RNA. Approximately 500,000 plaques from the unamplified library are screened using a mixed probe cont~;ning the thioesterase coding regions from Cuphea hookeriana Class II thioesterase clones CUPH-l (CMT-9), SU~ST~TUTE S~E~T (RU~E 26) W O96/36719 PCTrUS96/07064 CUPH-2 ~CMT-7) and CUPH-5 (CMT-10). (DNA se~uences o~
these clones are provided in WO 94/10288). Low stringency hybridization conditions are used as follows: hybridization is conducted at room temperature in a solution of 30%
formamide and 2X SSC (lX SSC = 0.15 M NaCl; 0.015 M Na citrate). Eighty two putative positive clones were identified, thirty of which were plaque purified. The nucleic acid sequence and translated amino acid sequence of a clone designated as MCT29 (CpFatB1) (SEQ ID NO:10) is provided in Figure 6. The translated amino acid sequence of this clone is a~oximately 83% identical to the sequence of a Cuphea hookeriana CUPH-2 clone (CMT-7 in Figure 7 of WO 94/10288) having primary thioesterase activity on C8:0 and C10:0 fatty acyl-ACP substrates.
Constructs for expression of MCT29 in E. coli are prepared. SphI and StuI sites are inserted 5' to the presumed mature protein N-t~rm;n~l~ located at amino acid 114 by PCR. Mature N-t~rm;nll~ predicted by correspon~nce to Leu 84 originally identified as bay thioesterase mature protein N-t~rm;nll~. The mature protein encoding region is cloned as a StuI/XbaI fragment into pUC118, resulting in clone MCT29LZ, to provide for expression of the C.
palustris thioesterase in E. coli as a lacZ fusion protein.
Lysates of transformed E. coli cells expressing the MCT29 thioesterase protein are assayed for acyl-ACP thioesterase activity. The results ~m~trate that CpFatB1 encodes a thioesterase enzyme having activity primarily on C8- and C10-ACP substrates, with 50% higher activity on C8-ACP than on C10-ACP. Low activity on C14-ACP substrate is also observed at levels of approximately 10% of the C8-ACP
activity.
MCT29LZ is also transformed into E. coli fadD, an E.
coli mutant which lacks medium-chain specific acyl-CoA
synthetase (Overath et al., Eur. J. Biochem (1969) 7:559-574) for analysis of lipid composition. Results of theseanalyses ~mo~trate a substantial increase in the production of 8:0 and 10:0 fatty acids in cells transformed with the C. palustris MCT29LZ clone.

SUBSTllUT~ SHEET (RU~E 26~

W O96/36719 PCTrUS96/07064 The closely related C. hookeriana ChFats2 clone also ~mnn~trates pre~erential activity on C8:0 and C10:0 acyl-ACP substrates, with 50% higher activity on C10:0 as opposed to C8:0 substrates. Expression of the ChFatB2 clone in seeds of transgenic Brassica plants results in increased production of C8 and C10 fatty acids in the ~ seeds, with C10 levels higher than C8 levels. (See co-p~n~;n~ application SN 08/261,695 ~iled June 1~, 1994.) F . Cuphea palustris C14 The nucleic acid seguence and translated amino acid seguence of an additional C. palustris Class II
thioesterase clone, MCT34 (CpFatB2) (SEQ ID NO:ll), is provided in Figure 7. The translated amino acid seguence of this clone is approximately 80% identical to the sequence of a Cuphea hookeriana CUPH-4 clone (~MT-13 in Figure 8 of WO 94/10288).
Constructs for expression of MCT34 in E. coli are prepared. SphI and StuI sites are inserted 5' to the presumed mature protein N-t~m;nn~ located at amino acid 108 by PCR. The mature protein encoding region is cloned as a StuI/X~aI fragment into pUC118, resulting in clone MCT34LZ, to provide for expression of the C. palustris thioesterase in E. col i as a l acZ ~usion protein. Lysates of transformed E. coli cells expressing the MCT34 thioesterase protein are assayed for acyl-ACP thioesterase acti~ity. ~he results ~monctrate that CpFatB2 encodes a thioesterase enzyme having activity primarily on C14-ACP
substrate. Activity on C16-ACP substrate is also observed at levels oE approximately 3096 of the C14-ACP activity.
MCT34LZ is also transformed into E. coli ~adD, an E.
coli mutant which lacks medium-chain specific acyl-CoA
synthetase (Overath et al., Eur. ~. Biochem (1969) 7:559-574) for analysis of lipid composition. Results of these analyses ~mnn~trate a substantial increase in the 35 production of 14: 0 and 14: 1 fatty acids in cells transformed with the C. palustris MCT34LZ clone.

Exam~ple 2 ~h;m~ric Thioesterase Constructs Sll~TlT~E S~JEET-(RULE ~8~

W O96/36719 PCTrUS96/07064 Both cDNA's of the bay and cAmphor thioesterases contain open re~;ng frames encoding 382 amino acids. Only 31 ami.no acids are different, among them more than half are conservative substitutions (Fig. 8). The codon usage is highly conserved between the two genes, suggesting their the common origin.
Plasmid pCGN3823 (WO 92/20236 and Voelker et al.
( 1994 ) supra) contains a 1.2-kb XbaI fragment of a bay C12 preferring thioesterase cDNA in a pBS- (Stratagene; La Jolla, CA) plasmid backbone and encodes the mature bay thioesterase protein beg;nn;ng at amino acid 84 (as num.bered in Voelker et al. (1992) supra). Amino acid 84 of the bay thioesterase was initially identified as the amino term;nll~ for the mature protein based on amino acid sequence Analysis of the purified protein. Comparison to translated amino acid sequences of other cloned plant medium-chain acyl-ACP thioesterases, however, indicates that the amino terminus may be located further upstream of the leu 84 residue (Jones et al. (1995) supra). Plasmid pCGN5220, described above, contains an XbaI/XhoI fragment of a cAm~hor C14 preferring thioesterase cDNA inserted into pBC+ plasmid (Stratagene). The XbaI site in the cAm~or cDNA is present at amino acid residue 84, a leucine, as in the bay thioesterase encoding region.
There is a conserved, unique Kpn I site in both the bay and cAmphor cDNA clones at amino acid residue 177 of the encoding sequence for the precursor bay and cAmphor thioesterases (Fig. 9). A second Kpn I site is located within the polylinkers of the plasmids 3' to the stop codons of the thioesterase sequences. The interchange of the two KpnI fragments between pCGN3823 and pCGN5220 allows the fusion of the N-t~rm;nAl region of one thioesterase to the C-t~rm;nAl region of the other, forming two rh;m~ric enzymes.
To prepare the c~;m~ric constructs, pCGN3823 and pCGN5220 were digested with KpnI and the resulting fragments gel-purified and ligated into the backbone plasmid from the opposite origin. DNA mini-preparations and SUB~;T~TUTE SI~EET (RULE 26) restriction digestions were used to identify the correct fusion constructs. The t-h;m~ic constructs used ~or expression and enzyme assays were also confirmed by DNA
sequencing.
The resulting ~;m~ric enzymes contain 92 amino acids ~rom the N-t~rm;n~l of one thioesterase and 207 amino acids from the C-t~rm~ n~ 1 portion of the other. The ~usion protein cont~;n;n~ the C-t~rm;n~l portion of the ~m~or thioesterase is referred to as Ch;m~ric 1 (Ch-1), and the other fusion protein is called ~;m~ric 2 (Ch-2) (Fig. 9).

Exam~le 3 Flexibility and Secondary Structure Analyses Predicted secondary structures of plant acyl-ACP
thioesterases are dete~m;ned be computer analysis.
Secondary structure predictions are based on methods of Chou and Fasman (Chou et al. (1974) Biochem. 13:222-245;
Prevelige et al. (1989) in Prediction of Protein Structure and the Principles of Protein Conformation (Fasman, G.D.
ed.), pp 391-416, Plenum, New York); and Garnier et al.
(1978) ~. Mol. Biol. 120:97-120~.
Flexibility of various regions of plant acyl-ACP
thioesterase regions are predicted by computer analysis using MacVector (International Biotechnologies, Inc.), based on flexibility prediction methods of Karplus and Schulz (Naturwiss. (1985) 72:212-213).

Exam~le 4 Engineering FatB Thioesterases A. Bay C12 Thioesterase PCR site-directed mutagenesis (Higuchi et al. (1988) Nucl. Acids ~es. 16:7351-7367) is used for amino acid replacements. The sense mutant primers used for the mutagenesis are as follows:

M197R/R199H 5'-GGA~ATAATGGCCCACGACATGATTTC~ll~l~-3' 35 (SEQ ID NO:16) T23lK 5'-G~ll~lCCAAAATCCC-3' (SEQ ID NO:17) R327Q s~-GCGTGCTGC_GTCCCTGACC-3' (SEQ ID NO:18) SUi3~TlTUTE SI~EET (RULE 26) W O96/36719 PCTrUS96/07064 R322M/R327Q s~-GAGAGAGTGCACGATGGATAGCGTGCTGCAGTCCCTGACC-3' (SEQ ID NO:l9) where bold letters M, R, H, T, K and Q are one-letter abbreviations for amino acids methionine, arginine, histidine, threonine, lysine and glutamine respectively, and the mutated nucleotides are underlined.
PCR conditions were as follows: five cycles of the PCR
were programmed with denaturation for 1 min at 94~C, renaturation for 30 seconds at 48~C, and elongation for 2 min at 72~C. These first five cycles were followed by 30 cycles with renaturation for 30 seconds at 60~C. The amplified DNA was recovered by ethanol precipitation, and examined by gel electrophoresis. The DNA was then digested with XbaI and BamHI, ethanol precipitated and ligated into XbaI/BamHI cut pBC plasmid. The ligation mixture was used to transform Sure cells (Stratagene) by electroporation, and the transformed cells were plated on LB medium cont~; n; ng 50 mg/l of chloramphenicol. Constructs cont~;n;ng the correct inserts were identified by mini-DNA
preparation and restriction digestion. The inserted DNA was sequenced to confirm the mutations.
The same designations noted above for the PCR primers were used for the mutant clones. As an example, M197R/R199H refers to a clone in which the methionine at residue 197 (of precursor bay thioesterase) was changed to an arginine, and where the arginine at residue 199 was changed to a histidine. Similarly, T231K indicates a mutant in which the threonine at residue 231 was changed to a lysine.
B. Cuphea palustris C14 Thioesterase To det~rm;n~ possible amino acid modifications for alteration of thioesterase substrate specificity towards shorter chain length fatty acyl-ACPs, sequences for C14:0 preferring thioesterases may be compared to se~uences for C8:0 and C10:0 preferring thioesterases. A comparison of amino acid sequences of thioesterase CpFatB2 (C14) to CpFatBl (C8/C10) is shown in Figure 10. The most striking S~BST~TUTE Sl~ (R!ILE 26) W O 96136719 PCTrUS96107064 differences in these thioesterase sequences is found in amino acids 230 to 312. Substitutions, such as H229I, H241N, W253Y, E275A, R290G, F292L, L295F, and C304R, can be made in single- and combinatory-form. Alternatively, ~r~m~;n swapping clones may be prepared which pro~ided for switching o~ portions o~ the C8/10 and C14 sequences. Of particular interest in this regard are sequences IEPQFV
starting at amino acid 274, and DRKFHKL starting at amino acid 289.
Exam~le 5 Specificity of t~l~; m~-~ic Enzymes and Bay Mutants Transformed E. coli cells in lacZ expression constructs are grown to 0.6 O.D.60o at 30~C, followed by addition of lmM IPTG and continuous growth at 30~C for 2 hours. The s~;m~nted cells were resuspended and sonicated in the assay buffer, and acyl-ACP hydrolysis is measured as previously described (Davies, H.M. (1993) Phyto~h~m;~t~y 33, 1353-1356). Sure cells transformed with pCGN3823 and pBC served as positive and negative controls, respectively.
Figure 11 shows the thioesterase specific activities o~ the ch; meric bay/c~mrhor enzymes when E. coli cells transformed with Ch-1 and Ch-2 were induced and assayed.
For Ch-1 (Fig. llA) the preferred substrate is 14:0-ACP, whereas for Ch-2 (Fig. llB) it is 12:0-ACP. These results indicate that the C-t~m; n~ 1 portion of the thioesterase protein determines the substrate specificity.
The enzyme specificities of two of the bay mutants are shown in Fig. llC and llD. A mutant in which Metl97 becomes an arginine and Argl99 becomes a histidi~e (M197R/R199H) results in altered specificity of the bay thioesterase such that the enzyme is equally specific towards both 12:0-ACP and 14:0-ACP substrates (Fig. llC).
Another mutant, T231K, gives an identical activity profile as the wild type (data not shown). Howe~er, the triple mutant M197R/R199H/T231K, which combines the three mutations, ~mo~trates 14:0-ACP specific thioesterase activity (Fig. llD). When this triple mutant enzyme is SlJ13STlTU~~ SHErT ~U: E 26) _ _ assayed at high concentration, very low levels of 12:0-ACP
activity are detectable.
Two more mutants (R327Q and R322M/R327Q) were also tested for thioesterase activity. Both mutants show identical activity profiles, and their specific activities toward 12:0-ACP and 14:0-ACP decrease about 100- and 30-fold, respectively, compared to the wild type bay thioesterase. These data indicate that the mutation R327Q
is responsible for the decreased activity. Decreased activity of R327Q is likely due to the fact that this amino acid position is located very close to the active site cysteine, C320. Studies which ~mo~.~trated the catalytic activity of C320 were conducted as follows. C320 was changed by site-directed mutagenesis to either serine or alanine. The mutant C320A completely lost thioesterase activity, while C320S retained approximately 60% of the wild-type activity. Interchange of cysteine and serine in the active site has also been ~mo~trated for An;mAl thioesterases (Witkowski et al. (1992) ~. Biol . Chem.
267:18488-18492). In An;mAls, the active site is a serine, and the change thus was from serine to cysteine.

Exam~le 6 Expression of Bay Mutants in E. coli fadD Cells The E. coli fatty acid-degradation mutant strain K27 (fadD88), a strain lacking acyl-coenzyme A synthetase, is unable to utilize free fatty acids when they are supplied in the medium (Klein et al. (1971) Eur. ~. Biochem. 19:442-450). Thus, it is an ideal host for observing the impact of recombinant thioesterases on the bacterial fatty acid synthase without interference from fatty acid degradation.
E. coli fadD was obtained from the E. coli Genetic Stock Center, Yale University (CGSC 5478). The fadD cells were transformed with either the pBC, a wild-type bay thioesterase gene or the mutant constructs, and grown overnight at 30~C in LB medium con~A; n; ng 50 mg/1 chlorAmph~n;col and 1 mM IPTG. Total lipids were analyzed as described previously (Voelker et al. (1994) supra).
Results of these analyses are presented in Table I below.

SUBSrllUTE SHEET (RU~E 28 W O96/36719 PCTrUS96/07064 Table I
Free Fatty Acid Accumulation (nmole/ml culture) Strain 12:0 14:0 Control* 0.3 1.6 Bay Thioesterase505.5 39.0 M197R/R199H 123.5 181.1 M197R/R199H/T231K35.4 352.9 *fadD cells transformed with the pBC vector only.

When bay thioesterase is expressed in fadD cells, large amounts of laurate (more than 500-fold above control background) and small amounts of myristate (about 10% of that of laurate) are produced (Table I). This result is consistent with the previous report (Voelker et al . ( 1994) supra). When mutant M197R/R199H is expressed in fadD cells, the ratio o f 12:0 to 14:0 accumula~ion changes to 1:1.5 (Table I), reflecting the thioesterase specificity of this mutant (Fig.llC). When mutant M197R/R199H/T231K is expressed in fadD cells, the ratio of 12:0 to 14:0 is completely reversed from that seen with the wild-type bay thioesterase. This result is also consistent with enzyme specificity of the mutant (Fig. llD).

Exam~le 7 Kinetic Analysis In order to gain insight into the impact of the mutations to the bay thioesterase, basic kinetics and inhibition studies were performed. Progress curves of thioesterase activity were obtained by scaling up the assay volume and sampling 100~1 at 5 minute intervals into 0.5 ml stop solution. Kinetic assays were performed at 30~C in buffer cont~;n;ng 100 mM Tris-HCl, pH 8.0, 0.01% Triton X-100, 1 mM DTT, 10% glycerol. After extraction of eachreaction mixture with 2.0 ml dimethyl ether, the radioactivity in 900 ~l of the organic fraction was determined by liquid scintillation counting. This SiJB~ll~lT~ Stl~ET (RULE 28) W O 96/36719 PC~rAUS96/07064 procedure allows accurate measurement of the total extractable free fatty acid (14C-labeled) without the interference of interphase between the organic and a~ueous fractions. Production of laurate and myristate in this assay was linear with respect to time for at least 30 min, and with respect to enzyme concentrations up to 1 mU. All assays were done in duplicate. Initial rate data were fitted to the following e~uations using kinetics software from Bio-Metallics, Inc. (KCat) for competitive inhibition V = VmaxS /[Km~app (1 + I /Kis) + S]; for noncompetitive inhibition v = VmaXs / [Km~app(l + I / KiS) + S(l+ I / Kii)];
and for uncompetitive inhibition v = VmaXs / [Km~app + S (1 + I / Kii)]; where v is velocity; V~aX is maximum velocity;
S is substrate concentration; Km,app is apparent Michaelis constant; KiS and Kii are slope and intercept inhibition constants, respectively; I is inhibitor concentration.
Results of these analyses are presented in Table II below.

Table II
Kinetic Constants of Wild-type Bay TE and Triple Mutant M197R/R199H/T231K

Enzyme R~m,app (~J~I)Ki (~I)*
14:0-ACP12:0-ACP12:0-ACP
Bay TE 6.4 + 1.91.9 + 0.5 10.2 + 1.2 (competitive)**
Mutant 2.3 + O.4 ND 11.6 + O.2 (competitive) *slope inhibition constants of 12:0-ACP with 14:0-ACP as varied substrates **competitive inhibition with respect to 14:0-ACP.
ND - not det~rm;ne~.
Under the same experimental conditions, both bay thioesterase and the triple mutant M197R/R199H/T231K have similar values of Km,app with respect to 14:0-ACP. The SUB~TITUTE SHEET (P~LI~ 26) W O96/36719 PCT~US96107064 specific activity of the mutant towards 12: O-ACP is too low to obtain any ~n;n~ful kinetic parameters under our assaying system. Nevertheless, these results indicate that the mutations do not significantly increase the substrate (14:0-ACP) b;n~ing affinity to the mutant enzyme.
Inhibition assays were conducted under the conditions described above using cold 12:0-ACP to compete with the substrate (14C labeled 14:0-ACP). Results of these assays are presented in Table III below.
Table III
Inhibition of 14:0-ACP Thioesterase Activity by 12:0-ACP

Enzyme Substrate (14:0-ACP) Inhibitor (12:0-ACP) Inhibition Concentration (~M) Concentration (~M) (%) Bay TE 5 5 53 Mutant 5 5 48 In these inhibition assays, a very similar result is seen with the wild-type and the mutant enzymes. When equal amounts of inhibitor (12: 0-ACP) and substrate (14:0-ACP) are present in the assay, the 14:0-ACP TE activity is reduced approximately 50%. If the amount of 12: 0-ACP is 5 times that of 14:0-ACP, the 14:0-ACP TE activity is reduced more than 75%. Consistent with what has been observed before (Pollard et al ., supra), a similar kinetic mer~n;sm is used by the wild-type bay TE, i.e. both 12:0- and 14:0-ACP have similar R~'s, but V= is highly favorable for 12:0-ACP. These data suggest that the specificity of the mutant enzyme is det~rm;ne~ in the acyl hydrolysis step, that is both 12:0- and 14:0-ACP can bind to the mutant enzyme with s;m;l~r affinity, however 14:0-ACP is cleaved at a much higher rate. This conclusion is further SUBSrll~lFE SHEET (RULE ~6 W O96/3671g PCTrUS96/07064 supported by inhibition kinetics, which show that 12:0-ACP
is a competitive inhibitor with respect to 14:0-ACP (Ki values are 10.2 + 1.2 ~M and 11.6 + O.2 ~M for the wild-type and mutant enzymes, respectively (Table II).
Thus, the amino acid substitutions described for the bay thioesterase apparently do not directly impact the substrate b;n~;ng site, as 12:0-ACP is a good competitive inhibitor to 14:0-ACP in both the wild type and the mutant enzymes. In fact, the Michaelis constants are similar and independent of substrate length for bay thioesterase and the engineered bay enzyme, suggesting that specificity must be largely det~rm;ne~ in the acyl hydrolytic step. Because the substrates (acyl-ACP) are relatively large molecules (Mr of ACP is about 9 Kd), it is likely that plant thioesterases have very relaxed b;n~;ng pockets. However, the enzymes have high selectivities with respect to fatty acid chain length or structure (i.e. the presence or absence of double bonds).
Furthermore, the tripeptide Met-Arg-Arg of native bay thioesterase is not the sole the det~rm; n; n~ factor for selectivity towards 12:0-ACP, as this tripeptide is commonly found at the same location in other medium chain specific thioesterases. Therefore, the changes in the engineered bay thioesterases may only slightly alter certain secondary structures, similar to what was observed when surface loops of Bacillus stearofh~rmophilus lactate dehydrogenease were modified (El Hawrani et al. (1994) Trends in Biotech. 12:207-211). Changing the tripeptide from M-R-R to R-R-H apparently reduced the flexibility of the $-structure ;mm~ tely following this tri-peptide, according to the predictions of chain flexibility in proteins (Karplus et al. (1985) Naturwiss. 77, 212-213).
This may lead to reduction of the flexibility of the substrate b; n~; ng pocket and active site.
Exam~le 8 Engineering FatA Thioesterases Alteration of thioesterase enzyme specificity of a mangosteen Garm FatAl clone is provided as an example of &l.~ 111 ~TiE SHEET (R~

W O96/36719 PCTrUS96107064 modification of FatA or Class I type thioesterases.
Desir2ble modifications with respect to FatA thioesterases include alteration in the substrate specificity such that activity on C18:0 fatty acyl-ACP is increased relative to activity on C18:1 or C16:0 fatty acyl-ACP substrates.
For example, in order to increase the relative activity on saturated fatty acids, such as C18:0, mutations in regions of Class I thioesterases which differ from the corresponding regions in Class II thioesterases, which act primarily on saturated fatty acids, may be useful. The data from bay thioesterase engineering experiments indicate that the region from amino acids 229 to 285 (as numbered in the top line consensus sequence on Figure 1) is important in thioesterase substrate b; n~; n~. Amino acid sequence comparison of this region indicates that in the highly conserved region from amino acids 250-265, several charged amino acids are different in FatA as compared to FatB
thioesterases. In FatA thioesterases, amino acid 261 is negatively charged with a few exceptions, whereas in FatB
clones analyzed to date, amino acid 261 is in most cases positively charged. Also, in FatA thioesterases~ amino acid 254 is positively charged in all FatA thioesterases studied to date, whereas in FatB clones analyzed to date, amino acid 254 is in all cases an amino acid having no charge. Thus, alteration of the amino acid charge at these positions may lead to alteration o~ substrate preference.
A FatA TE mutant in amino acid 261 (Figure 1 concensus numbering), D261K of mangosteen FatA1, is generated using PCR site-directed mutagenesis similar to the methods described for modification of bay thioesterase se~uences.
Mutant D261K is measured for thioesterase activity as described above (Davies, H.M. (1993) supra). Results of these analyses (Figure 12) ~mon~trate that the preference for 18:0 versus 18:1 was 35~ :0/18:1) in mutant D261K, as compared to 25~ in the wild-type Garm FatA1. Both the wild-type and mutant Garm FatA1 clone ~m~n~trate very low activity on 16:0 and no activity on medium-chain length substrates such as C10:0 through C14:0. An additional Garm SUa~3TITUTF St~EE~ (RU~E 26~

W O96/36719 PCTrUS96107064 FatA1 mutant was prepared having the D26lK mutation indicated above, as well as a mutation to change amino acid 254 from lysine to valine. This mutant, K254V/D261K, ~mn~trated an increased 18:0/18:1 ratio of 40%. These results once again supports the bay evidence which indicates that modification of this region can change the enzyme activity and specificity. A triple mutant, G249T/K254V/D261K, is under construction to further modify the Garm~ FatA1 clone towards the FatB thioesterase structure for evaluation of further specificity modification.
Other desirable amino acid modifications of mangosteen Garm FatA1 clones may be selected by comparison of the 18:0 enriched Garm FatA1 thioesterase amino acid sequence to the amino acid sequence for a FatA clone having activity primarily on 18:1 substrates, with little or no activity on 18:0 substrates. A comparison of the amino acid sequences of Garm FatA1 and an 18:1 preferring thioesterase clone from Brassica campestris (rapa), Br FatA1, is provided in Figure 13. In view of the bi n~; ng substrate alterations ~m~n~trated for the bay thioesterase in the region following the predicted ~-sheet and turn (anchored by amino acids G169 and G172 of the Figure 13 mangosteen and Brassica thioesterase comparison), this region is also a target for substrate specificity alteration of mangosteen thioesterase clone GarmFatA1. Secondary structure analysis and amino acid sequence com~rison of the mangosteen and Brassica rapa Class I thioesterases result in identification of several target mutations for further altering the substrate specificity of the mangosteen thioesterase, GarmFatA1. Target amino acids include Y182V, Q186E, D209S, V210D and H219F.
Furth~rmore, the unique restriction sites, BglII and SpeI, at amino acids 241 and 293 of Garm FatA1 (numbering as in Figure 4), provide for convenient ~m~;n swapping of the mangosteen thioesterase region between amino acids 242 and 293 (Figure 4 numbering). This region contains both the histidine 248 and cysteine 283 active site amino acids SU~3STITUTE St~FEr ~ LE 2~) W O96/36719 PCTrUS96/07064 which have been identi~ied by mutagenesis and biochemical assay. Thus, the major portion of the mangosteen thioesterase active site may be removed and replaced by the corresponding region (obt~;nAhle by PCR amplification) from other acyl-ACP thioesterases. Such methods allow for further modification of acyl-ACP thioesterase activity, such as increasing specific activity of the mangosteen thioesterase by substituting the active site of the high specific activity bay thioesterase clone, Uc FatB1.
Exam~le 9 Domain Swapping Techniques Methods for preparing thioesterase ~om~; n swapping constructs where convenient restriction sites are not available are provided.
A method for short ~m~; n swapping is illustrated in Figure 14. Two separate PCR result in two fragments (products of primers a + d, and primers b + c), which contain overlapping se~uence identical to the new ~mA i n, Primers c and d are synthesized to match the exact sequence at the 3' end down-stream of the original ~nm~;n, plus a 5' overhang correspon~;n~ to new ~om~; n sequence. The length of the matching sequence should be long enough to give a Tm of 50~C or above (calculated by assuming a C or G = 4~C and a T or A = 2~C). Ideally, the length of the 5' overhangs should not be greater than 18 bases (6 amino acids), although longer overhangs may also work at lower efficiencies. The first two PCR are carried out with approximately 0.2 ~M of primers and 0.1 ~g of template DNA
under PCR conditions described below. The second PCR run (PCR 3) is performed by m;x;ng 10 ~l of each product of PC~
1 and 2, and adding primers a and b to final concentration of 0.2 ~M. The resulting product is the targeted gene with the original ~mA; n replaced by a new ~nm~; n sequence. The PCR product may be Px~m;ned on an agarose gel before precipitation and restriction-digestion for subcloning.
The modified DNA fragment should be sequenced to ~erify the desired mutation.

SU~3STlTUTE SHE~ (RIJL- 2~) W O96/36719 PCTrUS96/07064 For swapping of longer ~m~; n~ / as illustrated in Figure 15, the switch of a ~om~1n from gene II to gene I
can be achieved by first amplifying three fragments from PCR 1, 2, and 3. These partly overlapped fragments are then mixed together for the next PCR with primers a and b. PCR
conditions are described below. The resulting full-length product is gene I with a new ~mA; n from gene II. By the same principle, two ~om~;n~ can be swapped into gene I
simultaneously by an additional PCR in the first run, followed by the second PCR in the presence of the four fragments (not shown).
PCR conditions which have been successfully used are as follows: five cycles were programmed with denaturation for 1 min at 94~C, renaturation for 30 seconds at 48~C, and elongation for 2 min at 72~C. The first five cycles were followed by 30 cycles using the same program except with renaturation for 30 seconds at 60~C. The rationale for the first five cycles at lower temperature is to ensure ~nnP~ling of the PCR primers with 5' overhangs. The increased temperature for the later cycles limit the further amplification to sequences amplified during the first five cycles. The Tm's for all primers should be designed at around 60~C. For the convenience of subsequent cloning, the full-length anchor primers (a and b, ~ig. 14 and 15) usually include additional restriction sites and/or overhangs for various PCR subcloning vectors. It is important to use as little amount of template DNA as possible (usually less than 0.1 ~g) to reduce the non-mutagenized background.

The above results ~mon~trate the ability to modify plant acyl-ACP thioesterase sequences such that engineered thioesterases having altered substrate specificity may be produced. Such thioesterases may be expressed in host cells to provide a supply of the engineered thioesterase and to modify the existing pathway of fatty acid synthesis such that novel compositions of fatty acids are obt~; n~ .

SUBSTITUTE S~EE~ (RULE 2~

W O96/36719 PCTnUS96/07064 In particular, the engineered thioesterases may be expressed in the seeds of oilseed plants to provide a natural source of desirable TAG molecules.

All publications and patent applications mentioned in this specification are indicative of the level of skill o~
those skilled in the art to which this invention pertains.
All publications and patent applications are herein incorporated by reference to the same extent as if each individual publication or patent application was specifically and individually indicated to be incorporated by reference.

Although the foregoing invention has been described in some detail by way of illustration and example for purposes of clarity of understAn~; n~, it will be obvious that certain changes and modifications may be practiced within the scope of the appended claims.

SU~3ST~TUTE SHEET (R~JLE 21

Claims (16)

What is claimed is:

1. A method for obtaining an engineered plant acyl-ACP thioesterase having an altered substrate specificity with respect to the acyl-ACP substrates hydrolyzed by said thioesterase, wherein said method comprises:
(a) modifying a gene sequence encoding a first plant thioesterase protein to produce one or more modified thioesterase gene sequences, wherein said modified sequences encode engineered acyl-ACP thioesterases having substitutions, insertions or deletions of one or more amino acid residues in the mature portion of said first plant thioesterase, (b) expressing said modified gene sequences in a host cell, whereby said engineered plant thioesterases are produced and, (c) assaying said engineered plant thioesterases to detect altered substrate specificity.

2. A method according to Claim 1 wherein said amino acid substitutions, insertions or deletions are in the C-terminal two/thirds portion of said first plant thioesterase.

3. A method according to Claim 1 wherein said amino acid substitutions, insertions or deletions are in the region corresponding to amino acids 230 to 285 of the consensus numbering of thioesterase amino acid sequences shown in Figure 1.

4. A method according to Claim 1 wherein said amino acid substitutions, insertions or deletions are in the region corresponding to amino acids 315 to 375 of the consensus numbering of thioesterase amino acid sequences shown in Figure 1.

5. A method according to Claim 1 wherein one or more amino acid residues in the mature portion of said first plant thioesterase are substituted with the corresponding amino acids of a second plant thioesterase, wherein the preferential acyl-ACP substrates for said first and second plant thioesterases are different with respect to carbon chain length and/or degree of saturation.

6. A method according to Claim 5 wherein said first thioesterase is modified by substitution of a peptide domain from said second thioesterase.

7. A method according to Claim 6 wherein said peptide domain comprises the active histidine and active cysteine residues of said second thioesterase protein.

8. An engineered plant acyl-ACP thioesterase, wherein said engineered thioesterase demonstrates an altered substrate specificity with respect to the acyl-ACP substrates hydrolyzed by said thioesterase as compared to wild-type acyl-ACP
thioesterase in said plant.

9. An engineered thioesterase of Claim 8, wherein said wild-type thioesterase is a Class II thioesterase.

10. An engineered thioesterase of Claim 8, wherein said wild-type thioesterase is a Class I thioesterase.

11. A DNA sequence encoding an engineered plant acyl-ACP
thioesterase, wherein said engineered thioesterase demonstrates an altered substrate specificity with respect to the acyl-ACP
substrates hydrolyzed by said thioesterase as compared to the wild-type plant acyl-ACP thioesterase.

12. A DNA sequence of Claim 11, wherein said wild-type thioesterase is a Class II thioesterase.

13. A DNA sequence thioesterase of Claim 11, wherein said wild-type thioesterase is a Class I thioesterase.

14. A DNA sequence encoding a mangosteen Class I acyl-ACP
thioesterase, wherein said acyl-ACP thioesterase has preferential activity on C18:1 acyl-ACP substrates, substantial activity on C18:0 acyl-ACP substrates, and only a small increase in activity on C16:0 acyl-ACP substrates as compared to control cells.

15. A DNA sequence according to Claim 14 wherein said DNA
sequence encodes the mangosteen GarmFatA1 thioesterase represented by the translated amino acid sequence shown in Figure 2.

16. A method of producing oilseed crop plant seeds containing triglyceride oil having a modified fatty acid composition, wherein said method comprises:
growing an oilseed crop plant having integrated into its genome a DNA construct, said construct comprising in the 5' to 3' direction of transcription, a promoter functional in a plant seed cell, a DNA sequence according to any one of Claims 11-15, and a transcription termination region functional in a plant cell.