Do Acyl Carrier Proteins and Conenzyme A have similar reactivity?

In terms of the reactions they undergo are they roughly equivalent groups?

The acyl groups ligated to CoA and to acyl carrier proteins are actually attached to the same group (a phosphopantetheine group) as shown below, and so have very similar reactivities (i.e. group transfer potentials).

Ability of Streptomyces spp. aryl carrier proteins and coenzyme A analogs to serve as substrates in vitro for E. coli holo-ACP synthase

Introduction: The polyketide natural products are assembled by a series of decarboxylation/condensation reactions of simple carboxylic acids catalyzed by polyketide synthase (PKS) complexes. The growing chain is assembled on acyl carrier protein (ACP), an essential component of the PKS. ACP requires posttranslational modification on a conserved serine residue by covalent attachment of a 4′-phosphopantetheine (P-pant) cofactor to yield active holo-ACP. When ACPs of Streptomyces type II aromatic PKS are overproduced in E. coli, however, typically little or no active holo-ACP is produced, and the ACP remains in the inactive apo-form.

Results: We demonstrate that E. coli holo-ACP synthase (ACPS), a fatty acid biosynthesis enzyme, can catalyze P-pant transfer in vitro to the Streptomyces PKS ACPs required for the biosynthesis of the polyketide antibiotics granaticin, frenolicin, oxytetracycline and tetracenomycin. The catalytic efficiency of this P-pant transfer reaction correlates with the overall negative charge of the ACP substrate. Several coenzyme A analogs, modified in the P-pant portion of the molecule, are likewise able to serve as substrates in vitro for ACPS.

Conclusions:E. coli ACPS can serve as a useful reagent for the preparation of holo-forms of Streptomyces ACPs as well as holo-ACPs with altered phosphopantetheine moieties. Such modified ACPs should prove useful for studying the role of particular ACPs and the phosphopantetheine cofactor in the subsequent reactions of polyketide and fatty acid biosynthesis.

The systematic name of this enzyme class is malonyl-CoA:[acyl-carrier-protein] S-malonyltransferase. Other names in common use include malonyl coenzyme A-acyl carrier protein transacylase,

  • [acyl carrier protein]malonyltransferase,
  • FabD,
  • malonyl transacylase,
  • malonyl transferase,
  • malonyl-CoA-acyl carrier protein transacylase,
  • malonyl-CoA:ACP transacylase,
  • malonyl-CoA:AcpM transacylase,
  • malonyl-CoA:acyl carrier protein transacylase,
  • MAT, and
  • MCAT.

Crystal Structures of FabD from E.coli [1] and Streptomyces coelicolor [2] are known and provide great insight into the catalytic mechanism of FabD. In E.Coli, FabD primarily involved in FAS pathway. However, in Streptomyces coelicolor, FabD is involved in FAS and polyketide synthase pathways. In both cases, the structures and active sites are very similar.

The protein has an α/β type architecture, but the fold is unique. the active site inferred from the location of the catalytic Ser92 contains a typical nucleophilic elbow as observed in α/ β hydrolases. [1] Serine 92 is hydrogen bonded to His 201 in a fashion similar to various serine hyrdolases. however, instead of the carboxylic acid typically found in catalytic triads, the main chain carbonyl of Gln 250 serves as a hydrogen bond acceptor in an interaction with His 201. [1] Two other residues, Arg-117 and Glu-11 are also located in the active site, but their function is not clear.

The fatty acid synthetic pathway is the principal route for the production of membrane phospholipid acyl chains in bacterial and plants. [3] The reaction sequence is carried out by a series of individual soluble proteins that are each encoded by a discrete gene, and the pathway intermediates are shuttled between the enzymes. [3] Malony-CoA:ACP Transacylase (FabD) is one such individual soluble protein and catalyzes the following reaction:

malonyl-CoA + acyl carrier protein ⇌ CoA + malonyl-[acyl-carrier-protein]

The transfer of malonate to acyl-carrier-protein (ACP) converts the acyl groups into thioester forms which are characteristic of acyl intermediates in fatty acid synthesis and which are strictly required for the condensation reactions catalyzed by β-ketoacyl-ACP synthetase. [4]

Mechanism Edit

Malonyl-CoA:ACP Transacylase uses a ping-pong kinetic mechanism with a bound malony ester as the acyl intermediate attached to a serine residue residing within a GHSLG pentapeptide. [5] FabD first binds malonyl-CoA, the malonyl moiety is then transferred to the active siteSer 92, and CoA is released from the enzyme. ACP then binds and the malonyl moiety is transferred to the terminal sulfhydryl of the ACP prosthetic group. This reaction is readily reversible. [3] [6]

Among all known metabolic pathways in living systems, fatty acid biosynthesis yields the most energy dense products. [7] As a result, microbial fatty acid derivatives are emerging as a promising renewable energy alternative to fossil fuel derived transportation fuels. Recently, Khosla et al. [7] have devised a procedure to reconstitute E.Coli Fatty Acid Synthase using purified protein components (including FabD) and reported a detailed kinetic analysis of this in-vitro reconstituted system. [8] Their finding provide a new basis for assessing the scope and limitations of using E.Coli as a biocatalyst for the production of diesel fuels.

FabD as a target for Antibacterial Drug Discovery: An upcoming field Edit

Fatty acid biosynthesis is carried out by the ubiquitous Fatty Acid Synthase. [9] Fatty acid synthase pathways are divided into two distinct molecular forms: Type I and Type II. [10] In Type I, Fatty Acid Synthase (found in humans and other mammals) is a single large polypeptide composed of several distinct domains. [11] On the other hand, each enzymatic activity (Condensation reaction, Reduction Reaction, Dehydration reaction) is found as a discrete protein in type II systems. [12] The difference in active site organization and predominance of type II FAS systems in bacteria make the enzymes of this pathway attractive targets for antibacterials. [9] [12]

FabD (Acyl-Carrier-Protein S-Malonyltransferase) is a reasonable target given that a high resolution crystal structure is available. [9] However, no FabD inhibitors have been reported in the literature and review articles on this topic. [9] The simple structure and acidity of malonate seem to permit few approaches to synthesizing derivatives (acting as potential inhibitors) that retain the character of the molecule.

A second approach for using FabD as a drug target is frequently identified in the literature: FabD can provide a useful tag for locating fab genes because FabD gene is usually adjacent to at least one other fab gene. [13] However (as of 2015), no potential drugs have attempted to exploit this feature.

Biosynthesis of Fatty Acids

The Author: Dr. Peter Reilly, Department of Chemical and Biological Engineering, Iowa State University, Ames, Iowa 50011, U.S.A.

Reactions of the fatty acid synthesis cycle

The standard way for cells to synthesize fatty acids is through the fatty acid synthesis cycle (Figure 1). This cycle of eight enzymes (acyl-CoA synthase, acyl-CoA carboxylase, acyltransferase, ketoacyl synthase, ketoacyl reductase, hydroxyacyl dehydratase, enoyl reductase, and thioesterase) and acyl carrier protein) is initated with acetic acid, CoA, and ATP to make acetyl-CoA using acyl-CoA synthase as catalyst. A second step, using another ATP and bicarbonate ion catalyzed by acyl-CoA carboxylase, yields malonyl-CoA. With ketoacyl synthase and acyltransferase catalysis, malonyl-CoA is added to an acyl chain, usually activated with acyl carrier protein, to make an acyl chain two methylene groups longer. Further reduction, dehydration, and reduction with ketoacyl synthase, hydroxyacyl dehydratase, and enoyl reductase catalysis, respectively, leads to a saturated and unhydroxylated acyl chain activated with acyl carrier protein. If the chain is of appropriate length, it is attacked by thioesterase to release acyl carrier protein, yielding the finished fatty acid.

Figure 1. The fatty acid synthesis cycle and the enzyme groups that are part of it. ACC: acetyl-CoA carboxylase ACS: acyl-CoA synthase AT: acyltransferase ER: enoyl reductase HD: hydroxyacyl dehydratase KR: ketoacyl reductase KS: ketoacyl synthase TE: thioesterase. SX: Coenzyme A or acyl carrier protein. Reproduced in modified form with permission from ref. 1. Copyright 2011, Oxford University Press.

Definitions of enzyme families and clans

It is perhaps not fully appreciated that these eight enzyme groups and one carrier protein are, in fact, usually assemblages within each group of several (or many) different proteins not necessarily related to each other by amino acid sequence (primary structure) or even by three-dimensional structure (tertiary structure). Instead, unrelated proteins can be shaped by convergent evolution to catalyze the same general reaction. Conversely, proteins of related primary and tertiary structures produced by different organisms but catalyzing the same reaction can be part of families whose members are generally considered to have the same common ancestor. Members of these families can be identified by automated interrogation of databases holding enzyme primary structures, where family members must pass a probability test of similarity followed by further multiple sequence alignment and comparison of tertiary structures (if available). In a further step, different families having unrelated (or only slightly related) primary structures but similar tertiary structures, and catalyzing the same general reaction by the same mechanism, can be gathered into clans within the same group. This is confirmed by computationally overlapping the tertiary structures of members of different families to confirm the low root mean square deviations of adjacent amino acid residues. Members of different families in the same clan are considered to have more distant common ancestors than those in the same family. A third diversification, not to be covered here, is that of family members into subfamilies, governed by statistical formulations applied to primary structures.

The ThYme database

The families of the enzymes catalyzing the reactions leading to fatty acid synthesis have been tabulated in the ThYme (Thioester-active enzYme) database [1]. At present this database consists of five families of acyl-CoA synthases, four families of acyl-CoA carboxylase subgroups (one of biotin carboxylases, one of biotin carboxyl carrier proteins, and two of carboxyl transferases) [2], one of acyltransferases, five of ketoacyl synthases [3], four of ketoacyl reductases [4], eight of hydroxyacyl dehydratases [4], five of enoyl reductases [4], 25 of thioesterases [5], and 16 of acyl carrier proteins [6] (Table 1). Although not all enzyme groups have been analyzed, one clan of ketoacyl synthases encompasses four of its five families [3], one clan of ketoacyl reductases covers three of its four families [7], one clan of hydroxyacyl dehydratases takes in two of its eight families [7], and four clans of thioesterases cover four, three, three, and two of its 25 families [5].

Most ThYme entries have resulted from genomic studies and have not been subjected to experimental projects. Therefore they usually do not have firmly established titles, often having been characterized as &ldquouncharacterized proteins&rdquo, &ldquohypothetical proteins&rdquo, or by other indefinite or even clearly incorrect terms. Databases such as ThYme have been helpful, by their use of primary structures to classify proteins, in directing these less characterized proteins into families containing other more studied and clearly described proteins.

Let us now more carefully consider each fatty acid synthesis cycle enzyme group.

Acyl-CoA synthases

Acyl-CoA synthases catalyze the first step in the fatty acid synthesis pathway, activating acetic acid with CoA and with the expenditure of an ATP molecule, leading to the formation of AMP, pyrophosphate, and water. (Figure 1). They are sorted into five families (Table 1), some with many more primary structures than others. Members of three families have similar tertiary structures (folds), while two have yet no known folds. As is common over the different groups of the cycle, the different families have different principal names and different Enzyme Commission (EC) numbers [8]. In some large families, as in family ACS1 here, there are so many different perceived enzyme functions that no single EC number can be assigned to the family. This divergence of families sorted by primary structure similarity, on one hand, and EC numbers based on enzyme substrates and products and therefore on the reactions they catalyze, on the other hand, further illustrates the phenomenon of convergent evolution of protein structures to serve the same purpose.

Archaea produce members of four of the five families, bacteria all five, and eukaryota two of the five (with a few members in a third family). Most eukaryotic members are produced by some combination of fungi, animals, and plants, as well as sometimes by algae and occasionally by other simple life forms.

Acyl-CoA carboxylases

The reaction catalyzed by acyl-CoA carboxylase is superficially a simple one: acetyl-CoA being carboxylated with CO2 (HCO3 &ndash under physiological conditions) and ATP to form malonyl-CoA, with ADP and inorganic phosphate being liberated (Figure 1). However, the reaction is actually much more complex, with biotin attached to a biotin carboxyl carrier protein being carboxylated by a biotin carboxylase domain. The carboxyl group is then transferred to an acetyl-CoA molecule, catalyzed by a carboxyl transferase domain (Figure 2). The whole complex of biotin carboxylase, biotin carboxyl carrier protein, and carboxyl transferase makes up acyl-CoA carboxylase. A recent and very complete review of this enzymatic system has been published by Lombard and Moreira [9].

Figure 2. Schematic of an acetyl-CoA carboxylase-catalyzed reaction producing malonyl-CoA. Reproduced with permission from ref. 2. Copyright 2012, Springer Science+Business Media B.V.

A comparison of primary structures shows that all biotin carboxylases and biotin carboxyl carrier proteins make up two single families, BC1 and BCCP1, while carboxyl transferases are part of two families, CT1 and CT2 [2] (Table 1).

The placement of the biotin carboxylase, biotin carboxyl carrier protein, and carboxyl transferase domains is very complex. In most eukaryotes, acetyl-CoA carboxylases occur as biotin carboxylase-biotin carboxyl carrier protein-fused carboxyl transferase chains. However, in bacterial and eukaryotic acyl-CoA carboxylases more active on propionyl-CoA, 3-methylcro­tonyl-CoA, and geranyl-CoA, the biotin carboxyl carrier protein and biotin carboxyl domains are attached to each other, but they are separate from the fused carboxyl transferase domains. In archaeal biotin-dependent carboxylases specific for acetyl-CoA and propionyl-CoA, biotin carboxyl carrier protein is separate from the biotin carboxylase and fused carboxyl transferase domains [9].


The acyltransferases are made up of one very large family (Table 1) whose members are produced by bacteria, eukaryota, and a few archaea. The specific task of the acyltransferases is to exchange the CoA moiety on the malonyl-CoA entering the fatty acid synthesis cycle with an acyl carrier protein (Figure 1).

Ketoacyl synthases

The general role of ketoacyl synthases, along with acyltransferases, is to add malonyl-CoA to an acyl-acyl carrier protein chain, with the loss of carbon dioxide and CoA and with the acyl chain increasing in length by two methylene groups [3] (Figure 1). There are five families of ketoacyl synthases, three of whose members have similar tertiary structures (Figure 3). Together with the members of a fourth family somewhat related to them by primary structure and by catalytic mechanism (members of all four families have a catalytic triad of Cys, His, and Asn or His), they form a single clan [3] (Table 1). These four families have very different functions, the members of one of them being involved in chalcone and stilbene synthesis. The fifth family, without a known tertiary structure, is made up only by eukaryotic enzymes involved in the elongation of already very long fatty acid chains.

Figure 3. Superimposed ketoacyl synthase (KS) crystal structures. KS1 Escherichia coli &beta-ketoacyl-ACP synthase III (yellow), KS3 Saccharopolyspora erythraea DEBS 2 (cyan), and KS4 Arachis hypogaea stilbene synthase (pink). Reproduced with permission from ref. 3. Copyright 2011, The Protein Society.

The &beta-oxidation pathway

Three enzymes, ketoacyl reductase, hydroxyacyl dehydratase, and enoyl reductase, are required to remove the 3-keto group added to the growing acyl chain by ketoacyl synthase (Figure 1). All three enzymes are also involved in the &beta-oxidation pathway, where fatty acids are broken down instead of being built up. In this pathway the ketoacyl reductase-, hydroxyacyl dehydratase-, and enoyl reductase-catalyzed steps operate in reverse, so these enzymes are given names characteristic of the reverse reactions. Oxidation of acyl-CoA catalyzed by acyl-CoA dehydrogenase (equivalent to enoyl reductase) with a flavin adenine dinucleotide (FAD) prosthetic group gives 2-enoyl-CoA. Hydration of 2-enoyl-CoA yielding 3-hydroxyacyl-CoA is catalyzed by 2-enoyl hydratase (equivalent to hydroxyacyl dehydratase). The third step, oxidation of 3-hydroxyacyl-CoA to give 3-ketoacyl-CoA by catalysis with L-3-hydroxyacyl-CoA dehydrogenase (equivalent to ketoacyl reductase) using nicotinamide adenine dinucleotide (NAD + ). A final step, with addition of CoA, removes an acetyl-CoA molecule and leaves an acyl-CoA molecule two carbon atoms shorter than before. It is catalyzed by 3-ketoacyl-CoA thiolase, which is not a member of the standard fatty acid synthesis cycle.

Polyketide synthesis

A short paragraph can serve to describe the difference between fatty acid synthesis and polyketide synthesis. In the latter, the ketoacyl reductase-, hydroxyacyl dehydratase-, and enoyl reductase-catalyzed steps do not necessarily occur during each turn around the cycle, so keto and hydroxy groups and double bonds may remain in the lengthening chain. Furthermore, the chain can be elongated by substances other than malonyl-CoA, therefore leading to chains with odd numbers of carbon atoms or different functional groups. Finally, if chain growth is terminated not by water but by a hydroxy group, chain cyclization can occur.

Ketoacyl reductases

The ketoacyl reductases are divided into four families (Table 1), with family KR1 having the largest number of tabulated primary structures in the ThYme database (Table 1). This family includes many reductases and dehydrogenases that do not catalyze reactions associated with the fatty acid synthesis or polyketide synthesis cycles or the &beta-oxidation pathway [4]. It also includes the former family ER1, as all the amino acid sequences of these enoyl reductases were found also in family KR1. Members of families KR1, KR3, and KR4 have NAD(P)-binding Rossmann folds and serine, tyrosine, and lysine catalytic residues, and are part of clan KR-A [4,7]. Those in family KR2 have histidine and glutamate catalytic residues, but also NAD(P)-binding Rossmann folds along with C-terminal 6-phosphogluconate dehydrogenase folds. The Rossmann fold is composed of up to seven mainly parallel &beta-strands, with an &alpha-helix between the first two &beta-strands. Members of families KR1 and KR2 are freestanding, while those of families KR3 and KR4 are parts of multi-enzyme chains carrying out fatty acid and polyketide synthesis.

Hydroxyacyl dehydratases

The hydroxyacyl dehydratases are found in eight families, of which families HD1 and HD3 through to HD6 have members with HotDog tertiary structures, while members of family HD2 have ClpP/crotonase folds [4] (Table 1). The HotDog fold is composed of one or occasionally two &alpha-helices encased in a &beta-sheet made up of several &beta-strands, a good example being shown in Figure 4. No tertiary structures are yet known for members of families HD7 and HD8. Although tertiary structures of members of families HD1, HD5, and HD6 can be closely superimposed on each other, the catalytic aspartate and histidine residues of HD1 are not in the same location as the catalytic aspartate and histidine residues of HD5 or the glutamate and histidine residues of HD6 [7]. The latter pairs of residues, on the other hand, overlap each other very closely and therefore allow HD5 and HD6 to make up clan HD-A (Figure 4). Separately, HD3 and HD4 members are part of multi-enzyme fatty acid and polyketide synthases.

Figure 4. A) Superimposed representative hydroxyacyl dehydratase (HD) tertiary structures of clan HD-A: HD5 &beta-hydroxydecanoyl thiol ester-ACP dehydrase from Escherichia coli (green) and HD6 ((3R)-hydroxyacyl-ACP dehydratase (FabZ) from Helicobacter pylori (cyan). B) Superimposed active-site side chains of the same HD family representatives, with colors as in A. Reproduced with permission from ref. 7. Copyright 2014, Springer Science+Business Media Dordrecht.

Enoyl reductases

As previously noted, family ER1 has been merged into family KR1 because all of its entries are also found there. Furthermore, all members of each family catalyze the same types of reductive reactions. The remaining enoyl reductases fall into five families, of which those in families ER3 and ER4 are part of larger fatty acid and polyketide synthase complexes [4] (Table 1). Families ER2 and ER4 have NAD(P)-binding Rossmann folds, while members of family ER5 have GrosES-like structures, and those of family ER6 have TIM (&beta,&alpha)-barrel folds.

A superimposition of a ER2 tertiary structure upon the structures of KR1, KR3, and KR4, members of clan KR-A, is extremely close, suggesting that ER2 members and members of the three ketoacyl reductase families have a distant common ancestor [7].


The thioesterases are the most varied of the eight enzyme groups making up the fatty acid synthesis cycle. They are arranged into 25 families [5] (Table 1), most of which have members with either &alpha/&beta-hydrolase or HotDog tertiary structures (Figure 5). Twelve of these families can be gathered into clans. Families TE5, TE9, TE10, and TE12 with HotDog folds are part of clan TE-A, while families TE8, TE11, and TE13 with different HotDog tertiary structures make up clan TE-B. Meanwhile families TE16, TE17, and TE18 with &alpha/&beta-hydrolase folds are part of clan TE-C, and families TE20 and TE21 with different &alpha/&beta-hydrolase tertiary structures make up clan TE-D. The mechanism of the thioesterases with &alpha/&beta-hydrolase tertiary structures is generally accepted to be based on the Ser/His/Asp catalytic triads usually found with such folds. The mechanisms of those thioesterases with HotDog tertiary structures are much less known [5]. They may well have different mechanisms from each other, as their experimentally identified catalytic residues vary.

Figure 5. Superimposed representative thioesterase (TE) tertiary structures of A) clan TE-A: acyl-CoA hydrolases from TE5 Escherichia coli (green), TE9 Helicobacter pylori (red), TE10 Pseudomonas sp. (yellow), and TE12 Prochlorococcus marinus (blue) B) clan TE-B: acyl-CoA hydrolases from TE8 Homo sapiens (blue), TE11 Arthrobacter sp. (red), and TE13 Escherichia coli (yellow) C) clan TE-C: acyl-ACP hydrolases from TE16 Homo sapiens (blue), TE17 Saccharopolyspora erythraea (red), and TE18 Amycolatopsis mediterranei (yellow) D) clan TE-D: protein-acyl hydrolases from TE20 Bos taurus (blue) and TE21 Homo sapiens (yellow).

Acyl carrier proteins

Acyl carrier proteins have very variable primary structures, and therefore they can be classified into families ACP1 to ACP17 (Figure 6), family ACP14 having been deleted [6] (Table 1). Most acyl carrier proteins have 70 to 100 amino acid residues, but members of family ACP2 have longer chains. Commonly found are three parallel &alpha-helices, usually but not always accompanied by a crosswise &alpha-helix (Figure 7). Members of some families are freestanding, while others are part of longer fatty acid synthase and polyketide synthase chains (Table 1). Members of individual acyl carrier protein families are more likely than members of the eight enzyme groups that are part of the fatty acid synthesis cycle to be made by members of single life kingdoms, usually bacterial but sometimes eukaryotic.

Figure 6. Tertiary structures of single members of acyl carrier protein (ACP) families. ACP1 Thermus thermophilus, ACP2 Saccharomyces cerevisiae, ACP3 Homo sapiens, ACP5 Streptomyces roseofulvus, ACP8 Aspergillus parasiticus, ACP10 Brevibacillus parabrevis, ACP12 Escherichia coli, ACP13 E. coli, and ACP17 Lactobacillus casei. Reproduced with permission from ref. 6. Copyright 2012, The Protein Society.

Figure 7. Phylogenetic tree of acyl carrier protein families. Reproduced with permission from ref. 6. Copyright 2012, The Protein Society.

Computation has established that ACP4 members have folds similar to those of other acyl carrier protein families whose folds have been experimentally determined [6]. However, members of families ACP15 and ACP16 appear by computation to have different tertiary structures, in agreement with their having more divergent primary structures (Figure 6) and different roles (they are associated with enzymes that decrease rather than increase the sizes of substrates and they do not use the same 4&rsquo-phosphopantetheine prosthetic group present in other acyl carrier proteins).

Fatty acid and polyketide synthesis complexes

It has been noted above and in Table 1 that the members of some fatty acid synthesis enzyme families are covalently linked parts of fatty acid and polyketide synthesis complexes. More specifically, these complexes, found in some bacteria, fungi, animals, and slime molds, allow growing fatty acid chains, activated by acyl carrier protein, to be passed sequentially from one fatty acid synthesis enzyme to the next without randomly diffusing among them.

Of the sixteen acyl carrier protein families, members of ACP1, ACP4, ACP5, ACP15, ACP16, and ACP17 are freestanding (Table 1) and engage in what is known as Type II synthesis. ACP1 proteins are produced by bacteria, fungi, animals, and plants, and they engage with many fatty acid synthesis enzymes. ACP4 and ACP5 members are produced by various bacteria and are involved with polyketide synthesis. Members of ACP15 and ACP16 are also produced by various bacteria, and they are associated with malonate decarboxylase and citrate lyase, respectively. ACP17 is composed of D-alanyl carrier proteins.

The remaining acyl carrier proteins covalently linked to enzymes take part in Type I fatty acid synthesis and other related activities (Table 1). ACP2 and ACP3 proteins help to produce fatty acids and are made by fungi and animals, respectively. Members of ACP6 through ACP11 are engaged in polyketide synthesis and are produced by mycobacteria (ACP6), fungi (ACP7, ACP8, and ACP11), and slime molds (ACP9). The very large number of ACP10 proteins are made by various bacteria as well as by fungi. ACP12 and ACP13 members are made by various bacteria and are linked to isochoromatases, peptide synthetases, phenyloxazoline synthases (ACP12), enterobactin synthases, and chromophore lyases (ACP13)

Of the eight fatty acid synthesis enzyme groups, six (acyltransferases, ketoacyl synthases, ketoacyl reductases, hydroxyacyl dehydratases, enoyl reductases, and thioesterases) have families whose members are part of multi-enzyme fatty acid and polyketide synthases (Table 1). One acyltransferase family (AT1) has domains in various synthases produced by bacteria, fungi, and animals, as well as freestanding members. Members of the very large KS3 family are both freestanding and are parts of polyketide and fatty acid synthases produced by bacteria, animals, fungi, and slime molds. Among the ketoacyl reductases, bacterial and fungal forms of KR3 are found in fatty acid synthases, while bacterial, fungal, and animal forms of KR4 are in polyketide and fatty acid synthases. Hydroxyacyl dehydratases in HD3 are part of fatty acid synthases produced by bacteria, fungi, and animals, and those in HD4 are in polyketide and fatty acid synthases made by bacteria and animals. ER3 has enoyl reductase domains in fatty acid synthases produced by bacteria and fungi, while enoyl reductase domains in ER4 are located in bacterial, fungal, animal, and slime mold polyketide synthases. Finally, families TE16, TE17, and TE18 among the 25 thioesterase families contribute domains to multi-enzyme synthases. TE16 members are part of amino acid adenylation proteins, polyketide, fatty acid, and nonrib­osomal peptide synthases, and enterobactin synthase component F made by bacteria, fungi, and animals. Bacterial polyketide synthases have TE17 domains. TE18 domains are found in bacterial amino acid adenylation proteins and nonrib­osomal peptide synthases and in animal S-acyl fatty acid synthases.

A note in passing

Archaeal membranes contain phospholipids that have hydrophobic chains bound to L-glycerol with ether links rather than the chains found in bacteria and eukaryota bound to D-glycerol by ester links. Furthermore, archaeal chains are branched with methyl groups at regular intervals, based on isoprenoid chemistry, while bacterial and eukaryotic organisms have aryl chains with straight-chain fatty acids. Despite this, many archaea have one or more of the enzymes (or acyl carrier protein) found in the fatty acid synthesis cycle. Chen used the ThYme database in 2011 to find that of the 155 archaeal species tabulated then, 33 had one cycle member, 28 had two, 22 had three, 17 had four, 26 had five, 27 had six, one had eight, and one had all nine [10].


This work is supported in part by the National Institute of General Medical Sciences Protein Structure Initiative grants GM094585 (A.J.) and GM098248 (G.N.P.) and National Institutes of Health grants GM109456 (G.N.P.) and GM114353 (B.S.). The use of Structural Biology Center beamlines at the Advanced Photon Source was supported by US Department of Energy, Office of Biological and Environmental Research grant DE-AC02-06CH11357 (A.J.). N.W. is supported in part by the Institute of Applied Ecology, Chinese Academy of Sciences, and a scholarship from the Chinese Scholarship Council (201504910034). J.D.R. is supported in part by an Arnold O. Beckman Postdoctoral Fellowship. C.-Y.C. is supported in part by the Fellowship of Academia Sinica–The Scripps Research Institute Postdoctoral Talent Development Program. This is manuscript #29600 from The Scripps Research Institute.


Strain and culture conditions

Phaeodactylum tricornutum Pt4 strain (UTEX 646) was grown in artificial sea water (Instant Ocean, Spectrum Brands, Blacksburg, VA, USA) supplemented with F/2 nutrients. Cultures were grown at 20°C under constant illumination (100 µmol m −2 sec −1 , 4000 K White and 660 nm LED lighting) and agitated continuously at 70 rpm. Growth was monitored by OD750nm calibrated to cell density measured by an automated cell counter (Cellometer T4, Nexcelom, Lawrence, MA, USA). Lines were maintained on F/4 agar plates grown at 20°C under 50 µmol m −2 sec −1 (3500 K fluorescent tubes).

Construction of acyl-ACP Δ9-desaturase overexpression cassette and transformation

The PAD (Phatr3_J9316) gene was chemically synthesized (Genscript) and codon optimized to remove conflicting restriction sites. The PAD gene was inserted into position 1 in the two-gene cassette transformation vector pPhOS2 (Hamilton et al., 2014 ) behind the EF2 promoter (Seo et al., 2015 ) generating pPhOS2_PAD construct. Construction of the transformation cassette pPhOS2_PAD is described in detail in Data S1. Transformation of P. tricornutum Pt4 via biolistic transformation and screening was carried out as described previously (Hamilton et al., 2014 ).

Generation of acyl-ACP Δ9-desaturase KO lines

A universal KO CRISPR/Cas9 vector was constructed with a dual sgRNA design (Aslan et al., 2015 ) using dual reporter gene selection system (designed by Mark Youles, personal communication). Type IIS LOOP DNA assembly was used for Level 1 and 2 vector constructions, following the method described in Pollak et al. ( 2019 ). The design of KO cassettes is explained in detail in Data S1. Transformation of P. tricornutum Pt4 via biolistic transformation and screening was carried as described previously (Hamilton et al., 2014 ).

Cloning of acyl-ACP Δ9-desaturase (Phatr3_J9316) into Synechocystis expression vector and functional characterization in Synechocystis

To generate the Syn_PAD strain, the Synechocystis PCC6803 glucose-tolerant WT (Syn_WT, Himadri Pakrasi, Department of Biology, Washington University, St. Louis, MO, USA) was transformed with the self-replicative vector pUR (Kim et al., 2016 ) expressing the acyl-ACP Δ9-desaturase gene lacking the 5′ putative signal peptide sequence (as predicted by SignalP online software). Generation of the vector, transformation and functional characterization of the Phatr3_J9316 gene is described in detail in Data S1.

RNA extraction and quantitative reverse transcription-PCR

For RNA extraction, 1–1.5 × 10 8 exponential phase cells were pelleted, flash frozen in liquid nitrogen and stored at −80°C. RNA extraction was carried out using the method described in Rengel et al. ( 2018 ). cDNA was synthesized and quantitative PCR analysis was carried out as described detail in Data S1.

Lipid analysis

Whole biomass FAME analysis was carried out as previously reported (Hamilton et al., 2014 ), using pentadecanoic acid and tricosanoic acid internal standards. Further details are described in Data S1.

Glycerolipids were extracted following a method adapted from Bryant et al. ( 2016 ), further details are described in Data S1.

For positional analysis, lipids were fractionated by 1D and 2D TLC (described in detail in Data S1), and then purified classes were characterized by preferential loss analysis under low-energy collision-induced dissociation as described previously (Abida et al., 2015 ). After species characterization, quantification of each species was carried out by liquid chromatography-MS/MS as previously described (Jouhet et al., 2017 ).

Transmission electron microscopy

Transmission electron microscopy imaging was carried out by Rothamsted Bioimaging (Harpenden, Herts, UK). Processing of P. tricornutum cells for transmission electron microscopy imaging is described in detail in the Data S1.


One-way analysis of variance was applied to data on specific growth, FA, lipid class and lipid species data. Data were transformed by natural log (quantitative data) or logit (relative %). Post-hoc comparison of the means was carried out using LSD at 5%, 1% and 0.1% level of significance. A two-tailed Student’s t-test was carried out in cases where only two strains are compared. Microsoft Excel was used for these analyses.

Acyl-coenzyme A: cholesterol acyltransferase modulates the generation of the amyloid β-peptide

The pathogenic event common to all forms of Alzheimer's disease is the abnormal accumulation of the amyloid β-peptide (Aβ). Here we provide strong evidence that intracellular cholesterol compartmentation modulates the generation of Aβ. Using genetic, biochemical and metabolic approaches, we found that cholesteryl-ester levels are directly correlated with Aβ production. Acyl-coenzyme A:cholesterol acyltransferase (ACAT), the enzyme that catalyses the formation of cholesteryl esters, modulates the generation of Aβ through the tight control of the equilibrium between free cholesterol and cholesteryl esters. We also show that pharmacological inhibitors of ACAT, developed for the treatment of atherosclerosis, are potent modulators of Aβ generation, indicating their potential for use in the treatment of Alzheimer's disease.

Identification of a chloroplast coenzyme A-binding protein related to the peroxisomal thiolases.

A 30-kD coenzyme A (CoA)-binding protein was isolated from spinach (Spinacea oleracea) chloroplast soluble extracts using affinity chromatography under conditions in which 95% of the total protein was excluded. The 30-kD protein contains an eight-amino-acid sequence, DVRLYYGA, that is identical to a region in a 36-kD protein of unknown function that is encoded by a kiwifruit (Actinidia deliciosa) cDNA. Southern blotting also detected a spinach gene that is related to the kiwifruit cDNA. The kiwifruit 36-kD protein that was synthesized in Escherichia coli was imported into chloroplasts and cleaved to a 30-kD form it was processed to the same size in an organelle-free assay. Furthermore, the kiwifruit protein specifically bound to CoA. The kiwifruit protein contains a single cysteine within a domain that is related to the peroxisomal beta-ketoacyl-CoA thiolases, which catalyze the CoA-dependent degradative step of fatty acid beta-oxidation. Within 50 amino acids surrounding the cysteine, considered to be part of the thiolase active site, the kiwifruit protein shows approximately 26% sequence identity with the mango, cucumber, and rat peroxisomal thiolases. N-terminal alignment with these enzymes, relative to the cysteine, indicates that the 36-kD protein is cleaved after serine-58 during import, agreeing with the estimated size (approximately 6 kD) of a transit peptide. The 30-kD protein is also related to the E. coli and mitochondrial thiolases, as well as to the acetoacetyl-CoA thiolases of prokaryotes. Features distinguish it from members of the thiolase family, suggesting that it carries out a related but novel function. The protein is more distantly related to chloroplast beta-ketoacyl-acyl carrier protein synthase III, the initial condensing enzyme of fatty acid synthetase that utilizes acetyl-CoA.



Genomic DNA from the haploid protease deficient S. cerevisiae BJ2168 plasmid (MATa leu2 trp1 ura3–52 prb1-1122 pep4-3 prc1-407 gal2) was obtained by phenol/chloroform extraction. FAS1 and FAS2 genes were amplified from genomic DNA using primers containing overlaps with the vector backbone for cloning with NEBuilder HiFi DNA Assembly (NEB) (Supplementary Table 1). FAS1 was cloned into NcoI-digested pET-28a(+) with a C-terminal His6 tag (JWL02). FAS2 was cloned into NcoI-digested pET-15b(+) (JWL03). The sequences of the FAS1 and FAS2 genes were verified (Supplemantary Data 1) via whole plasmid sequencing of JWL02 and JWL03 at the Center for Computational & Integrative Biology DNA core facility in Massachusetts General Hospital.

Point mutations were generated by the overlap extension method 24 using a forward primer containing the mutation and a reverse primer 200–2000 bp downstream (Supplementary Table 1). PCR fragments were purified using spin columns and used as megaprimers to amplify the entire plasmid and introduce the point mutation. Mutations were verified by Sanger sequencing using custom sequencing primers (Supplementary Table 1).

Transformation, expression, and protein purification

The JWL02 and JWL03 plasmids were co-transformed into Escherichia coli BL21 cells by electroporation, plated on LB agar plates containing 50 μg/mL kanamycin and 100 μg/mL ampicillin, and grown overnight at 37 °C. One colony was inoculated into 50 mL LB media containing kanamycin and ampicillin and grown overnight at 37 °C, 250 RPM. In all, 10 mL of this starter culture was transferred to 1 L LB media containing kanamycin and ampicillin in a 4-L flask and grown at 37 °C, 180 RPM in an Innova 42 shaker (New Brunswick) until the OD600 nm reached 0.6. The culture was cooled to 15 °C and expression was induced with 0.5 mM IPTG. Expression occurred at 15 °C, 180 RPM overnight.

Cells were harvested by centrifugation at 4000 × g for 15 min and resuspended in lysis buffer (200 mM potassium phosphate pH 7.4, 300 mM KCl, 10 mM imidazole, 5 mM β-mercaptoethanol, 0.5 mM PMSF, 1 mM benzamidine, and 5 mM aminocaproic acid). Resuspended cells were incubated with DNAse I and lysozyme for 10 min. Cells were lysed by sonication (Branson Analog Sonifier S-450A) in an ice-water bath with five cycles of

15 W pulses every 0.5 s for 1 min followed by 2 min rest. The lysate was cleared by centrifugation at 40,000 × g for 1 h. The cleared lysate was filtered using a 0.22-μm filter and loaded onto a pre-equilibrated 5 mL HisTrap column (GE Healthcare). The column was washed with 10 column volumes of wash buffer (200 mM potassium phosphate pH 7.4, 300 mM KCl, 20 mM imidazole, and 5 mM β-mercaptoethanol) before elution with a linear gradient of 0–100% elution buffer (200 mM potassium phosphate pH 7.4, 150 mM KCl, 300 mM imidazole, 5 mM β-mercaptoethanol) over 20 column volumes. Fractions were pooled and concentrated to 1 mL and injected into a Superose6 10/300 GL column (GE Healthcare) pre-equilibrated with TBS (50 mM Tris pH 7.4, 150 mM NaCl). Gel filtration fractions were pooled and concentrated to 2 mg/mL, flash frozen in liquid nitrogen and stored at −80 °C. For cryoEM samples, proteins were used fresh on the day of purification.

Activity assays

Activity assays of purified recombinant FAS were performed as described 14 . Briefly, 20 μg FAS was mixed with 0.2 mM acetyl-CoA, 0.7 mM NADPH, 1 mM DTT, and 100 mM potassium phosphate pH 7.4 in a 100-μL reaction. The level of NADPH was monitored at 340 nm for 3 min and the reaction was started by adding 30 nM malonyl-CoA and monitored for 1 h. The specific activity of wild type recombinant FAS was measured from three independent protein preparations to be 78 ± 32 mU/mg, where one unit is defined as consumption of 1 μmol of malonyl-CoA per minute. Only linear part of the curve (i.e. 2 min post malonyl-CoA addition) was used for calculation of the specific activity 25 .

CryoEM sample preparation and image collection

To prepare samples for CryoEM imaging, 3 μl of freshly purified protein complexes were added at a concentration of 1–2 mg/ml to glow-discharged (25 s in air) holey gold grids 26 mounted in a Vitrobot Mark IV. Blotting was done for 3 s at 4 °C and 100% humidity before plunge freezing in liquid ethane. A total of three grids were prepared per condition reported in Supplementary Table 2 and Supplementary Table 3 from the same purifications. Data reported in Supplementary Table 2 were collected from one grid and data reported in Supplementary Table 3 were collected from three grids.

Electron micrographs were collected with a FEI Tecnai F20 field emission electron microscope equipped with a Gatan K2 summit direct detector device (DDD) camera. Micrographs were acquired as movies in counting mode using DigitalMicrograph® software with 1.45 Å/pixel, 2 frames/s for 15 s, and an exposure rate of 1.2 e − /Å 2 /frame (Supplementary Table 2 and Supplemantary Table 3). Images were converted to mrc format using dm2mrc 27 .

Stock solutions of acetyl-CoA, malonyl-CoA, and NADPH were prepared at a concentration of 10 mM in TBS buffer. For the AT and MPT double mutant FAS construct (S274A - S1808A), 1 μL of either acetyl-CoA or malonyl-CoA stock solutions were diluted with TBS to 5 μL before addition to 10 μg (at 2 mg/ml) of purified FAS, before application onto cryoEM grids.

For KR (Y839A), DH (H1564A), and ER (H740A) mutants, 100 nmol NADPH and 50 nmol malonyl-CoA were added to the pooled and concentrated HisTrap fractions (final volume 0.5 ml) and incubated for 1 h at room temperature. Substrates were removed by gel filtration as described above. In total, 10 μg (at 2 mg/ml) of purified FAS was mixed with a 5-μL solution containing 10 nmol acetyl-CoA, 25 nmol NADPH, and 15 nmol malonyl-CoA. Samples were kept on ice for 30 min before application onto cryo-EM grids.

For the WT FAS, 10 μg of purified FAS at 2 mg/ml was mixed with solutions containing either 5 μL of TBS (for the apo sample), or 5 μL of TBS containing 10 nmol acetyl-CoA, 25 nmol NADPH, and 15 nmol malonyl-CoA (for turn-over conditions) on ice, respectively. Samples were applied onto the cryoEM grids immediately after the addition of TBS plus substrates.

Image processing

Data presented in Supplementary Table 2 were processed with cryoSPARC 2.0 28 . Movies were aligned by alignframe_lmbfgs implemented in cryoSPARC 2.0 29 . CTF estimation was done with CTFFIND4 30 . Particle motion correction was performed with alignpart_lmbfgs implemented in cryoSPARC 2.0 29 . High resolution refinement was done with the homogenous refinement algorithm using particles selected from reference free 2D classification. A 30-Å low-pass filtered cryoEM map was generated from PDB: 2UV8 7 with ACP atoms deleted from the model. This map was used as the initial reference in the homogenous refinement. Model to map conversion was carried out in UCSF Chimera 31 . Low-pass filtering of refined maps was carried out using the ‘relion_image_handler’ command. All one-voxel-thin slices were generated using XIMDISP 32 .

Data presented in Supplementary Table 3 were processed in Relion 3.0 33 . Movies were aligned with the Relion implementation of MotionCor2 and CTF estimation was performed using CTFFIND4. Following 2D classification, all 3D classification was done with C1 symmetry using a 30-Å low-pass filtered cryoEM map generated from PDB 2UV8 with ACP atoms deleted as the initial reference. Particle images were selected from 3D classes and refinement was done with C1 symmetry. The refined maps were aligned to the D3 symmetry axis and a second refinement was carried out with D3 symmetry imposed. The reference was aligned to the symmetry axis using the ‘relion_align_symmetry’ command.

Focused classification

A mask for focused classification was made by rigid body docking of a model of a β-chain protomer and the ACP domain from an α-chain protomer from the FAS crystal structure (PDB: 2UV8) to an asymmetric unit of the D3 refined map of the DH-stalled FAS. Residues from the ER and DH domains of the β-chain were selected based on proximity to the observed ACP density in the DH-stalled state and the domain boundaries identified in a previous crystallographic study of S. cerevisiae FAS 7,8 . The residues were 601–704 (ER domain, encompassing the ACP binding sites observed in C. albicans and T. thermophilum) and residues 1236–1658 (DH pseudo-dimer). ACP residues 140–302 from the α-chain of FAS (PDB: 2UV8) were rigid body docked into the D3 refined map of DH-stalled FAS in the asymmetric unit with the docked β-chain fragments. This model comprising partial segments of ER and DH domains plus the whole ACP domain docked at the observed ACP density in the DH-stalled state was used to generate the initial reference and was low-pass filtered to 15 Å for the focused classifications described below. To create a mask that covered the partial model and the position of the ACP in the ER binding mode, we expanded the aforementioned partial model of ER-DH -ACP DH by including a second model of the ACP domain docked at the ER binding site based on homology to previously observed ACP bound to ER in fungal FAS (PDB: 6U5V) 14 . A 15-Å resolution mask (i.e. focused mask) comprising the residues mentioned above was generated with UCSF chimera and Relion 3.0. The focused mask was extended by 3 pixels (1.45 Å/pixel) with 3 pixel padding. This mask was subtracted from a mask generated from the D3 refined map of DH-stalled FAS to produce a subtracted mask. The subtracted mask was then multiplied by the map and the ‘modified map’ containing all regions of FAS except for the region defined by the focused mask, was used for subsequent steps.

The rotation matrix for each asymmetric unit of FAS (i.e. α-,β-chain heterodimer) was computed using the original particle images with alignment information from the D3 symmetry refinement (done in Relion 3.0) via the ‘relion_particle_symmetry_expand’ command 20,34 . Different projections from the ‘modified map’ were then generated based on orientation information from the D3 symmetry refinement. These projections were subtracted from the symmetry expanded particle images. Subtracted images were used for focused 3D classification without orientation search in Relion 3.0 with an optimized regularization parameter of 500. The focused refinement was done in cryoSPARC 2.0 by importing the subtracted particle data set from Relion 3.0 belonging to the ACP containing 3D class in Fig. 3 (i.e. Class 1 in Supplementary Fig. 5) low-pass filtered to 20 Å as the initial reference. The mask for the cryoSPARC focused refinement was generated using the initial reference map in Relion 3.0 and low-pass filtered to 15 Å and expanded by 3 pixels with 3 pixels padding. A B factor of 344 was estimated using Guinier plot analysis implemented in cryoSPARC and used for map sharpening. A local resolution estimate of the focused refined map was done using cryoSPARC 2.0.

Model fitting and visualization

Model fitting was done by rigid body docking of the selected regions of ER, DH, and ACP domains, as described in the manuscript text, into the cryoEM maps of the stalled complexes. An atomic model of FAS (PDB: 2UV8 7 ) was used for docking. Maps and models were visualized with UCSF chimera 31 and PyMol 35 . Flexible fitting was done for Fig. 4b and Supplementary Fig. 6d using PHENIX real space refinement with the resolution limited to 6 Å. The Ramachandran statistics were: 83.8% favored, 15.85% allowed, and 0.35% outlier. For the distance measurements reported in Figs. 4b and 5, side chain conformations for residues H740, D1559, and H1564 were from PDB: 2UV8.

Sequence alignment

Non-redundant protein sequences corresponding to the subphylum Saccharomycotina (i.e. true yeast, taxonomy id: 147537) were chosen for alignment using BLASTp 36 . The β-chain sequence of S. cerevisiae FAS was chosen as the search template. A maximum number of 100 sequences was chosen for initial display. Sequences belonging to different strains of S. cerevisiae were excluded from the analysis. The remainder encompassed 58 sequences from other species of Saccharomycotina (Supplementary Data 2). Multiple sequence alignment was done with Clustal Omega 37 . LOGO plots were generated via the WebLogo server 38 with the y-axis set to represent frequency of amino acids plotted against amino acid number (based on S. cerevisiae sequence) on the x-axis.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.

22.4.7. Citrate Carries Acetyl Groups from Mitochondria to the Cytosol for Fatty Acid Synthesis

The synthesis of palmitate requires the input of 8 molecules of acetyl CoA, 14 molecules of NADPH, and 7 molecules of ATP. Fatty acids are synthesized in the cytosol, whereas acetyl CoA is formed from pyruvate in mitochondria. Hence, acetyl CoA must be transferred from mitochondria to the cytosol. Mitochondria, however, are not readily permeable to acetyl CoA. Recall that carnitine carries only long-chain fatty acids. The barrier to acetyl CoA is bypassed by citrate, which carries acetyl groups across the inner mitochondrial membrane. Citrate is formed in the mitochondrial matrix by the condensation of acetyl CoA with oxaloacetate (Figure 22.25). When present at high levels, citrate is transported to the cytosol, where it is cleaved by ATP-citrate lyase.


Enzymes catalyzing the cleavage of C-C, C-O, or C-N bonds by elimination. A double bond is formed in these reactions.

Figure 22.25

Transfer of Acetyl CoA to the Cytosol. Acetyl CoA is transferred from mitochondria to the cytosol, and the reducing potential NADH is concomitantly converted into that of NADPH by this series of reactions.

Thus, acetyl CoA and oxaloacetate are transferred from mitochondria to the cytosol at the expense of the hydrolysis of a molecule of ATP.


Type II PKSs manufacture many structurally complex and bioactive organic molecules that serve as important antibiotics and anticancer agents. ACPs play a pivotal role in these biosynthetic processes by interacting with virtually all building blocks, intermediates, and enzymes throughout the manufacturing of polyketide products. Engineering strategies, such as hybrid synthase production, precursor-directed biosynthesis, mutagenesis, metabolic engineering, and combinatorial biosynthesis, all rely in some way on the ability of the ACP to function with noncognate enzyme partners and/or molecular substrates. Therefore, understanding the molecular-level structure and function relationships of type II PKS ACPs will be crucial to harnessing the power of nature’s molecular assembly lines. A fully integrated approach to studying type II PKS ACP mechanisms and structures will be important, and we envision that the needle of the field will be moved by merging MD simulations with bioinformatics, mutagenesis work, probe applications, structural characterization techniques, and biochemical assays. With routes to access KSCLF samples that were previously difficult to obtain and recent advances in protein structure prediction software, it is an exciting time to uncover the molecular underpinnings of these remarkable systems.