Structure and Function of Animal Fatty Acid Synthase
Subrahmanyam S. Chirala and Salih J. Wakil*
Verna and Marrs McLean Department of Biochemistry and Molecular Biology, Baylor College of Medicine, Houston, Texas 77030
ABSTRACT: Fatty acid synthase (FAS; EC 2.3.1.85) of animal tis-
sues is a complex multifunctional enzyme consisting of two
identical monomers. The FAS monomer (~270 kDa) contains
six catalytic activities and from the N-terminus the order is -
ketoacyl synthase (KS), acetyl/malonyl transacylase (AT/MT), -
hydroxyacyl dehydratase (DH), enoyl reductase (ER), -ketoa-
cyl reductase (KR), acyl carrier protein (ACP), and thioesterase
(TE). Although the FAS monomer contains all the activities
needed for palmitate synthesis, only the dimer form of the syn-
thase is functional. Both the biochemical analyses and the
small-angle neutron-scattering analysis determined that in the
dimer form of the enzyme the monomers are arranged in a
head-to-tail manner generating two centers for palmitate syn-
thesis. Further, these analyses also suggested that the compo-
nent activities of the monomer are organized in three domains.
Domain I contains KS, AT/MT, and DH, domain II contains ER,
KR, and ACP, and domain III contains TE. Approximately one
fourth of the monomer protein located between domains I and
II contains no catalytic activities and is called the interdo-
main/core region. This region plays an important role in the
dimer formation. Electron cryomicrographic analyses of FAS re-
vealed a quaternary structure at approximately 19 resolution,
containing two monomers (180 130 75 ) that are separated
by about 19 , and arranged in an antiparallel fashion, which is
consistent with biochemical and neutron-scattering data. The
monomers are connected at the middle by a hinge generating
two clefts that may be the two active centers of fatty acid syn-
thesis. Normal mode analysis predicted that the intersubunit
hinge region and the intrasubunit hinge located between do-
mains II and III are highly flexible. Analysis of FAS particle im-
ages by using a simultaneous multiple model single particle re-
finement method confirmed that FAS structure exists in various
conformational states. Attempts to get higher resolution of the
structure are under way.
Paper no. L9623 in Lipids 39, 10451053 (November 2004).
Fatty acid synthase (FAS) of animal tissues is a homodimer,
and the monomer is a multifunctional protein containing
seven catalytic domains and a site for the prosthetic group,
4-phosphopantetheine. The FAS dimer complex catalyzes the
synthesis of saturated fatty acids, myristate, palmitate, and
stearate by using the substrates acetyl-CoA, malonyl-CoA,
and NADPH (14). It is noteworthy that nearly every tissue
in the human body has some level of FAS expression, but
*To whom correspondence should be addressed at Department of Biochem-
istry and Molecular Biology, Baylor College of Medicine, One Baylor Plaza,
Houston, TX 77030. E-mail: swakil@bcm.tmc.edu
Abbreviations: ACP, acyl carrier protein; AT/MT, acetyl/malonyl transacy-
lase; DH, dehydratase; DI to DIII, domains I to III; ER, enoyl reductase;
FAS, fatty acid synthase; KR, ketoacyl reductase; KS, ketoacyl synthase; and
TE, thioesterase.
Copyright 2004 by AOCS Press
FAS is highly expressed in tissues like liver, adipose, and lac-
tating mammary glands (5).
FAS plays an important role in energy homeostasis by con-
verting the excess food consumed into lipids for storage and
providing energy when needed via -oxidation. FAS is also
required for the generation of milk lipids during lactation. Be-
sides being the apolar constituent of various membrane lipids
required for membrane functions and its biogenesis, the prod-
ucts of FAS myristate (C14:0) and palmitate (C16:0) are in-
volved in the myristoylation and palmitoylation of cellular
and viral proteins for membrane targeting. Also, the products
palmitate and stearate serve as substrates for chain elongation
to produce very long chain fatty acids, which are important
constituents of sphingolipids, ceramides, and glycolipids
needed for cell division progression, brain structures, and
neurological functions (6). Furthermore, increased FAS lev-
els in cancer tissues indicate a poor prognosis and inhibition
of FAS and leads to apoptosis of cancer cells (7,8). Inhibition
of FAS also suppresses HER2/neu (erbB-2) oncogene over-
expression in cancer cells (9). When mice were treated with
FAS inhibitor c75, their food intake and body weight were re-
duced (10). However, saturated fatty acids synthesized by
FAS are readily available from food sources.
To understand the importance and contribution of the de
novo fatty acid synthesis catalyzed by FAS in animals, trans-
genic knockout mice were generated (11). Studies of these
mice showed that not only did the Fasn/ null mutant em-
bryos die before their implantation in the uterus, but that due
to haploid insufficiency most of the Fasn+/ heterozygotes
also died at various stages of their embryonic development
(11). Whole-mount in situ hybridization performed to detect
FAS mRNA expression in E7.5E10.5-day-old embryos,
using digoxigenin-labeled anti-sense FAS mRNA probes,
showed that the FAS gene was expressed broadly in the
mouse embryos, but the regions of strong FAS mRNA expres-
sion varied in the developmental stages examined. These ob-
servations suggest that the sites of most intense FAS expres-
sion coincide with regions known to be undergoing tissue-tis-
sue interactions, tissue outgrowth, and progressive
modeling/remodeling (11); hence, FAS is essential during
embryonic development.
In animal and human tissues palmitate (16:0 fatty acid),
the major saturated constituent of lipids, is synthesized de
novo by FAS. FAS was the first to make an entry into the most
sophisticated class of multifunctional proteins in which a sin-
gle polypeptide contained all the required constituent activi-
ties and performs all of the reactions involved in the synthe-
sis of a metabolite. In addition to FAS, the polyketide syn-
1045 Lipids, Vol. 39, no. 11 (2004)
1046 S.S. CHIRALA AND S.J. WAKIL

thases, the nonribosomal peptide synthases, and the eukary-

otic carboxylases such as acetyl-CoA, propionyl-CoA, and
pyruvate carboxylases, fall into this category of multifunc-
tional proteins; however, much of our knowledge and appre-
ciation of these multifunctional proteins came from the stud-
ies of fatty acid synthesis in animal tissues. There are several
extensive reviews describing the structure and function of
FAS (14), but this review will focus mainly on the recent
findings related to the structure and function of FAS and will
give a brief introductory description of the pertinent earlier
findings.

FATTY ACID SYNTHESIS BY ANIMAL FAS

FAS synthesizes palmitate from acetyl-CoA and malonyl-
CoA, which is produced according to the following equation:

CH3COS-CoA + 7HOOCCH2COS-CoA + 14NADPH + 14H+

CH3(CH2CH2)7COOH + 14NADP+ + 8CoA-SH + 7CO2 + 6H2O

The synthesis of palmitate is carried out by a very complex

overall reaction that involves sequential condensation of
seven C2 units derived from malonyl-CoA to the primer
acetyl moiety derived from acetyl-CoA. The individual reac-
tions involved in the synthesis became known from studies of
the multienzyme complex of Escherichia coli FAS (type 2) in
which each reaction is catalyzed by a separate specific en-
zyme of the FAS complex (1,2,12). These studies identified
three important characteristics of fatty acid biosynthesis.
First, ATP and CO2 are required for the synthesis of the key
substrate, malonyl-CoA, catalyzed by a biotin enzyme,
acetyl-CoA carboxylase (13). Second, the essentiality of the
acyl carrier protein (ACP) containing 4-phosphopantetheine
prosthetic group in fatty acid synthesis (14), and third, that all
of the acyl intermediates in fatty acid synthesis are bound to
ACP as thioester derivatives. The observation that ACP-
bound acyl-intermediates are involved in fatty acid synthesis
distinguished the pathway of synthesis from that of fatty acid
oxidation, which utilizes the acyl-CoA intermediates (12,13).
The fatty acid synthesis reactions, however, are catalyzed by
several component enzymes of the multifunctional animal
FAS (type 1) as illustrated in Figure 1. The transfer of sub-
strates, acetate to -ketoacyl synthase (KS), and malonate to
acyl carrier protein (ACP), from the respective CoA deriva-
tives is catalyzed by acetyl and malonyl transacylases respec-
tively. KS catalyzes the condensation of these substrates to
form acetoacetyl-ACP with the release of carbon dioxide. The
reduction of acetoacetyl to -hydroxyacyl chain is catalyzed
by -ketoacyl reductase (KR). The product then undergoes
dehydration by -hydroxyacyl dehydratase (DH) and is fol-
lowed by a second reduction by enoyl reductase (ER) leading
to the formation of a saturated four-carbon fatty acid attached
to ACP. KS mediates the transfer of the saturated fatty acyl
chain to its own Cys-SH, thus freeing ACP to accept malonate
from malonyl-CoA for the next round of elongation. Thus se-
quential addition of two carbon moieties derived from mal-
FIG. 1. Catalytic reactions involved in the synthesis of palmitate by ani-
mal fatty acid synthase (FAS). Acetyl and malonyl groups from their re-
spective coenzyme A derivatives are transferred to the enzyme by
acetyl/malonyl transacylase (AT/MT). In the first cycle the condensation
of the acetyl group bound to -ketoacyl synthase (KS) with the malonyl
moiety bound to the acyl carrier protein (ACP) and subsequent -car-
bon processing reactions involving -ketoacyl reductase (KR), -hy-
droxy-acyl dehydratase (DH), and enoyl reductase (ER) leads to the for-
mation of butyryl-ACP. The butyryl group is then transferred to KS, thus
freeing ACP-SH to accept malonyl moiety from its CoA derivative for
the next round of elongation. After an additional six cycles of elonga-
tion reactions, the palmitate bound to ACP and is released by
thioesterase (TE) to initiate next round of palmitate synthesis. Excluding
the reversible steps, there are more than 50 reactions involved in the
synthesis of one palmitate molecule, although most of them are repeti-
onyl-CoA to the growing acyl chain leads to the synthesis of
palmitate attached to ACP, which is then hydrolyzed by
thioesterase (TE) and released as palmitic acid (14).
ORGANIZATION OF COMPONENT ACTIVITIES IN
THE MULTIFUNCTIONAL FAS
Initially, the animal FAS was thought to be a multienzyme en-
zyme complex like E. coli FAS due to the presence of several
protein bands on SDS-PAGE. By controlling proteolysis dur-
ing isolation and denaturing immediately before electrophore-
sis, it was determined that FAS contains a single polypeptide
of approximately 250 kDa (15). Active enzyme centrifuga-
tion analysis revealed that it is a homodimer (15,16). FAS
dimer dissociates in low ionic-strength buffers and at cold
temperatures, and the FAS monomer was found to be inactive
in fatty acid synthesis (1,2). Even before cDNA derived,
amino acid sequences were available, the order of the compo-
nent activities in the monomer was established by partial pro-
teolysis using various proteases, isolation of various compo-

Lipids, Vol. 39, no. 11 (2004)

 FATTY ACID SYNTHASE 1047

FIG. 2. The structural and functional organization of animal and human FAS dimer. In A, the
model is based on the biochemical data in which FAS monomer has three domains. The two
subunits with their domains I, II, and III are drawn in an antiparallel arrangement (subunit divi-
sion) so that two sites of palmitate synthesis are constructed (functional division). Domain I
contains KS, AT/MT, and DH. Domain II contains ER, KR and ACP. Domain III contains TE.
The wavy line represents the 4-phosphopantetheine prosthetic group of ACP. The domains I
and II are separated by the interdomain peptide, consisting of about 640 amino acids, with no
known catalytic function. In the head-to-tail arrangement of the monomers, two sites for the
synthesis of palmitate are generated (functional division). Each of these sites is composed of
domain I of one monomer and the domains II and III of the other monomer. In B, a model pro-
posed based on the small-angle neutron-scattering analyses (26). Abbreviations used are indi-
cated for Figure 1.

nent fractions, and by determining the FAS-related activities
contained in them (1,2,17). These analyses revealed that the
FAS monomer contains all the component activities in the fol-
lowing order: NH2-KS, acetyl/malonyl transacylase
(AT/MT), DH, ER, KR, ACP, and TE (1,2,17).
The multifunctional FAS cDNA sequences of chicken
(1820), rat (21,22), human (5), and mouse (23) have been
determined. Comparison of the amino acid sequences of ani-
mal and human FAS showed them to be highly homologous.
In the deduced amino acid sequences, the already known ac-
tive site peptide sequences of KS, AT/MT, the putative nu-
cleotide binding motifs, Gly-X-Gly-X-X-Gly of ER and KR,
the phosphopantetheine binding site of ACP, and the active
site Ser of TE could be readily located (5,1824). Compar-
isons of the amino acid sequences derived from cDNA se-
quences of animal and human FAS, with those of the subunits
of prokaryotic FAS complex and polyketide synthases helped
in defining the boundaries of the FAS component enzyme se-
quences (3,4,24). Interestingly, the highly conserved se-
quences of component enzymes are separated by short pep-
tide stretches (linker regions) of least conserved sequences.
Furthermore, there is a long, less conserved sequence of ap-
proximately 640 amino acids, termed the interdomain/core
region, located between DH and ER, which does not code for
any known functional activity of FAS (Fig. 2). The determi-
nation of the N-terminal sequences of chicken FAS peptides,
which retained component activities (24) generated by lim-
ited proteolysis using kallikrein (2,17), revealed that most of
the protease cleavage sites are located in the linker regions.
Thus, in FAS monomer, the protein sequences of component
activities are folded independently and are organized like
beads on a string, an arrangement that suggests that fusion of
genes coding for individual type 2 FAS activities resulted in
complex multifunctional type 1 FAS.
Proteolytic analyses of animal FAS, described above, also
suggested that these activities are organized in three domains
(1,2,17). The domain organization was further refined by
using the available amino acid sequences of animal and
human synthases and heterologous expression of various seg-
ments of the cDNA. As shown in Figure 2, Domain I contains
KS, AT/MT, and DH. Domain II contains ER, KR, and ACP,
and domain III contains TE (1,2). Domains I and II are sepa-
rated by the interdomain/core region. In addition, thiol cross-
linking reagent experiments using 1,3-dibromo-2-propanone
showed that the active site Cys-SH of KS and phosphopan-
tetheine thiol of ACP were cross-linked, and the cross-linked
enzyme fractionated as a dimer (~500 kDa) in SDS-PAGE
(25). Hence, in the FAS homodimer, the two monomers are
arranged in a head-to-tail fashion, bringing the amino termi-
nal KS-Cys-SH of one monomer within 2 distance to the
carboxy terminally located ACP-phosphopantetheine-SH of
the other monomer (25). This arrangement generates two ac-
tive centers for fatty acid synthesis as shown in Figure 2A.
The domain organization and the dimeric nature of FAS were
further supported by physical studies of the chicken FAS by
using electron microscopy of the dibromopropanone cross-
linked enzyme and the small-angle neutron-scattering analy-
sis of native enzyme (26). These studies suggested a struc-

Lipids, Vol. 39, no. 11 (2004)

1048 S.S. CHIRALA AND S.J. WAKIL
tural model with dimensions of 160 146 73 and an over-
all appearance of two side-by-side cylinders arranged in an
antiparallel fashion. Each cylinder has a length of 160 and
a diameter of 36.5 , and is divided into three domains with
lengths of 32, 82, and 46 . In the antiparallel arrangement
of these cylinders, two crevices are generated on the major
axis of the model at the opposite ends of the molecular diad.
These two crevices probably represent the two active centers
of the FAS dimer (Fig. 2B). This model was further substan-
tiated by the observation that the two sites of palmitate syn-
thesis function simultaneously (27).
COMPONENT ENZYMATIC ACTIVITIES OF ANIMAL
FAS
Biochemical studies, heterologous expression of various con-
stituent activities and full length enzyme, and mutational analy-
sis of amino acid residues of active sites have helped in refining
the organization of component activities and their mechanism of
action. The substrates acetyl-CoA and malonyl-CoA enter the
reaction cycle through a common component enzyme
acetyl/malonyl transacylase. This component activity of chicken
FAS (amino acids 455 to 726), when expressed with a His6 fu-
sion in E. coli, was found to be in the insoluble inclusion bodies
(24). However, the enzyme protein can be extracted from the in-
soluble pellet with denaturing agents such as guanidinium-HCL
and urea, purified using Talon column, and can be renatured by
removing the denaturing agents to obtain a protein that had ap-
proximately 60% of the native AT/MT activity (24,28). AT/MT
has a serine at its active center in a highly conserved motif Gly-
His-Ser (28) and binds either acetyl or malonyl moieties from
their CoA derivatives. The serine-bound acetyl moiety is trans-
ferred to the active site Cys-SH of KS via ACP, and the malonyl
group is transferred to the pantetheine-SH of ACP (Fig. 1). Mu-
tation of the active site Ser581, of the AT/MT to Ala in rat FAS,
resulted in the loss of both acetyl and malonyl transacylase ac-
tivities (28). Mutation of another conserved His683 to Ala also
resulted in the loss of both transacylase activities (28). Hence,
His683 is considered as a proton acceptor from the hydroxyl of
Ser581 to increase its nucleophilicity. Interestingly, when the
conserved Arg606 mutated to Ala, the enzyme has reduced ac-
tivity with the substrate malonyl-CoA. In addition, the Arg606A
mutant enzyme showed a dramatically increased acetyl transacy-
lase activity (29). Thus, Arg606 is essential for the malonyl
transacylase activity due to its role in positioning the free car-
boxyl anion of malonate at the active center (29). Since the ac-
tive site serine of AT/MT can accept both acetyl-CoA and mal-
onyl-CoA as substrates, and the primer acetyl-CoA is needed
only once in the synthesis of a palmitate molecule, the AT/MT
has to scan for the required substrate at any given time and trans-
fers the substrate that is not required to CoA. Hence, free CoA is
required for fatty acid synthesis (14).
The condensing enzyme, KS, has a highly reactive cysteine
at its active center (14,25). KS performs condensation reaction
between acetyl/acyl chains bound to its Cys-SH and malonyl
moiety bound to the phosphopantetheine thiol of ACP, generat-
ing a -ketoacyl chain attached to ACP. KS also functions as an
Lipids, Vol. 39, no. 11 (2004)
acyl transferase, since it transfers the growing saturated acyl
chains from ACP to its own active site Cys-SH (14). In doing
so, KS frees the ACP-SH so that it can accept the malonyl group
from malonyl-CoA, to continue the next round of elongation
(Fig. 1). The amino acid sequence of KS is highly conserved
across all species. Mutation of the active site Cys161 of rat FAS
to Ala, Ser, or Thr, resulted in a practically inactive enzyme (30);
however, mutation of Cys161 to Gln led to the loss of condensa-
tion activity but this mutant enzyme was able to decarboxylate
malonate (31). The decarboxylation of malonyl-CoA may be fa-
cilitated by His293 and His331 (31,32).
The -ketoacyl fatty acid, generated by KS, is reduced by KR
to the -hydroxy fatty acid, which is dehydrated by the dehy-
dratase to the enoyl derivative, which is then reduced by ER to
the saturated fatty acyl derivative. Both ER and KR contain a
typical nucleotide binding motif, GX(X)GXXG, identified in a
Rossmann fold. In ER, the sequence GSGGVG, and in KR, the
sequence GGLGGFG, are highly conserved, and as was ex-
pected, mutation of some of these glycine residues (underlined)
in ER and KR of rat FAS resulted in loss of the reductase func-
tions (33). Similarly, in DH, mutation of a conserved histidine,
His873, in rat FAS led to the loss of its activity, which suggests
that this histidine may be involved in the DH active site (34).
Thioesterase performs the ultimate step in fatty acid synthesis
by releasing the long chain fatty acid palmitate from ACP. The
cloning and expression of chicken FAS ACP-TE in E. coli
showed that the animal ACP can be phosphopantetheinylated by
E. coli holo-ACP synthase (35). Mutagenesis of TE active site
residues of chicken FAS, Ser2302 or His2475 to Ala, resulted in
an inactive enzyme, suggesting that these two residues are part
of Ser-His-Asp catalytic triad of serine esterases (36). Conver-
sion of Ser2302 to Cys made the TE function like a thiol-active
thioesterase and was inhibited by iodoacetamide (36). In addi-
tion, the double mutant containing Ser2302Cys and His2475Ala
showed diminished reactivity to iodoacetamide, and bound
palmitate as palmitoyl-S-intermediate without hydrolyzing the
thioester. These results suggested that Ser2302 is a strong nucle-
ophile and that His2475 is the general base required to withdraw
the proton from Ser2302/Cys2302 (36). Likewise, in the rat
mammary gland-specific thioesterase (TEII), which binds to
FAS and releases medium chain fatty acids during lactation,
Ser101 and His237 are required for the esterase function (37).
By collaborating with Quiocho and his associates, we have re-
cently determined the 3-D structure of the human FAS TE do-
main. The structure revealed that the active site pocket of TE
contains the conserved residues, Ser2308, His2481, and
Asp2338, that constitute the active site catalytic triad (38). Mu-
tagenesis of these conserved amino acids in human FAS TE to
Ala, individually, confirmed their importance in the catalysis
(38), suggesting that FAS-TE and TEII evolved from serene pro-
teases.
DIMERIC STRUCTURE OF FAS AND THE ROLE OF THE
INTERDOMAIN/CORE
As shown in Figure 2A, each palmitate synthesis center is
constituted by the N-terminal KS, AT/MT, and DH compo-
 FATTY ACID SYNTHASE
nents of one monomer and the C-terminal ER, KR, ACP, and
TE components of the other monomer. Synthesis of palmitate
involves seven cycles of 2-carbon additions to the primer
acetyl group and involves more than 50 reactions (Fig. 1).
Based on the average specific activity of purified human FAS
(~70 nmol of palmitate/min/mg), one molecule of FAS dimer
generates approximately 40 molecules of palmitate per
minute by catalyzing more than 2000 individual
reactions/min/dimer. Hence, the two centers of palmitate syn-
thesis in the dimer are highly dynamic and flexible, and it is
difficult to imagine that the component activities can hold the
dimer together while performing palmitate synthesis. How-
ever, the interdomain/core region, which constitutes approxi-
mately 25% of the FAS monomer (Fig. 2A) and has no cat-
alytic function, is thought to play an essential role in dimer
formation and holding it together during palmitate synthesis.
Based on the head-to-tail arrangement of the monomers in the
FAS dimer, the interdomain dimer regions of the two
monomers also interact with each other in a head-to-tail man-
ner (Fig. 2). Hence, the FAS model (Fig. 2A) predicts that the
two halves of a FAS monomer could associate to form a func-
tional FAS dimer containing one catalytic center. To demon-
strate such an association, recombinant proteins containing
domain I and the N-terminal half of interdomain and domains
IIIII containing the C-terminal half of the interdomain at its
N-terminus (Fig. 3) were expressed in E. coli and purified
(39). Recombinant proteins containing no interdomain re-
gions were also produced. All of these recombinant proteins
contained N-terminal thioredoxin and His6 tag fusions. As
shown in Figure 3, fatty acid synthesis could be reconstituted
only when the reaction mixture contained both the domain I
with the N-terminal half of interdomain and the C-terminal
FIG. 3. Requirement of interdomain region in the FAS dimer formation.
FAS dimer is shown in a schematic representation (not drawn to scale).
Arrows indicate the interaction between the interdomain regions (ID) of
the monomers. The interaction of the two halves of the FAS monomer
to generate a palmitate synthesis site is shown in the boxed region. The
recombinant proteins of domain I (DI), and domains IIIII (DIIIII) either
containing or lacking interdomain regions were expressed in E. coli as
thioredoxin (TRX) fusion proteins, purified, and used in the reconstitu-
tion of FAS activity. Abbreviations used are indicated for Figure 1.
1049
half of the interdomain containing domain IIIII proteins
(39). The addition to the reconstitution mixture containing a
constant amount of either domain I or domain IIIII protein
and increasing amounts of domain IIIII or domain I protein,
respectively, showed that the rate of fatty acid synthesis was
determined by the limiting amount of the recombinant do-
main protein present in the reaction mixture (39). Hence, to
generate an active center for fatty acid synthesis, stoichiomet-
ric amounts of interdomain containing domain I and domain
IIIII proteins were needed. As expected, the bifunctional
reagent dibromopropanone cross-linked these two proteins
and inhibited the fatty acid synthesis (39). That the two halves
of a FAS monomer can generate palmitate synthesis further
confirmed the antiparallel arrangement of FAS monomers in
the dimer. In addition, these studies also showed that the in-
terdomain region has an important structural role in the FAS
dimer formation and function. That the interdomain regions
interact with each other in vivo was further confirmed by
using the yeast two-hybrid system (40).
As described above, the two sites of palmitate synthesis of
the FAS dimer function simultaneously and independently
(20); however, some variations to this theme were reported
(4,30,33,4144). Smith and his associates analyzed fatty acid
synthesis by several combinations of mutant FAS dimers in
which one monomer had a mutation in one component activ-
ity and the other monomer had a mutation in a different com-
ponent activity (4,30,33,4144). These analyses mostly sup-
ported the FAS dimer model with two active centers shown
in Figure 2; however, it was observed that KS and AT/MT can
work partially with the ACPs located in both of the palmitate
synthesis centers (41). Although this observation can be ex-
plained by the flexible nature of the monomers in the FAS
dimer as described below, however, the observation that DH
functions only with the ACP of the same monomer (4,33) ap-
pears to be inconsistent with the functional division of the
FAS dimer (Fig. 2) and our finding that the two halves of FAS
monomer can form a fatty acid synthesis center as shown in
Figure 3 (39). To explain the inter- and intrasubunit catalytic
domain interactions observed, Smith and his colleagues have
proposed an alternative model in which KS, AT/MT, and ACP
domains of the opposite subunits are in structural and func-
tional contact with each other and with the rest of the func-
tional units that spread in the opposite directions (4,33). Al-
though this model might explain the observations of Smith
and his associates, it ignores the importance of interdomain
region in the formation of FAS dimer shown in Figures 2 and
3 (39,40). Furthermore, the mutation complementation analy-
ses performed by Smith and his colleagues showed fatty acid
synthesis levels in the range of 1620% (except for a few that
showed 3040%), a less than expected level of FAS activity
(4,33). Even when one of the active centers was supposed to
be active (for example, KS mutant monomer when combined
with ACP mutant monomer), the FAS activity was only 19%
(33). This variation in complemented FAS activity and the
less than expected activity might be due to alteration in the
structures of the individual monomers because of mutations
Lipids, Vol. 39, no. 11 (2004)
1050 S.S. CHIRALA AND S.J. WAKIL
or to structural constraints and catalytic inefficiencies related
to sharing of activities between the two fatty acid synthesis
centers. Nonetheless, the apparent redundancy of KS and
AT/MT functions might provide small but measurable
(~20%) catalytic advantage to the FAS dimer function (4,33).
In the head-to-tail dimer model depicted in Figure 2, dibro-
mopropanone cross-links Cys-SH of KS of one monomer
with the pantetheine-SH of ACP present in the other
monomer, and hence, there will be two cross-links/dimers
(25). Analysis of the dibromopropanone cross-linking
showed that 45% of FAS dimers contained two cross-linkers,
15% contained one cross-link/dimer, and 35% of the mole-
cules were monomers containing intramolecular KS and ACP
thiol cross-linking (43). The presence of cross-linked
monomers shows another degree of FAS flexibility. This flex-
ibility could also explain how a FAS dimer formed by using a
nonfunctional monomer, in which all of the component activ-
ities and ACP were mutated, and a fully functional monomer
can synthesize fatty acids, albeit at 16% of the normal FAS
dimer activity (44). Taken together, these observations sug-
gest that monomers of the FAS dimer also have the ability to
fold in such a way that the KS and ACP of the monomer may
be close enough to generate some FAS activity. However,
based on the dibromopropanone cross-linking analysis
(25,26,43), which showed that head-to-tail dimer is the pre-
dominant form of FAS, the dimer with two distinct active cen-
ters proposed in Figure 2 is the most favorable and catalyti-
cally preferred structure. When some of the component en-
zymes are nonfunctional, the built-in flexibility of the
monomers in the dimer described below might help in de
novo fatty acid synthesis.
QUATERNARY STRUCTURE OF FAS
Attempts to crystallize FAS have not been successful thus far.
Hence, we resorted to determining the structure of the highly
FIG. 4. Typical micrograph of fatty acid synthase (FAS) embedded in
vitreous ice at 2.7-m defocus. Particles can be observed in numerous
different views ranging from dumbbell shape to double arches. Images
are collected as focal pairs; image shown here is a far-from-focus image.
The close-to-focus image is used for actual image processing.
Lipids, Vol. 39, no. 11 (2004)
pure human FAS using electron cryomicroscopy, by collaborat-
ing with Chiu and his associates (45). Electron cryomicroscopy
showed FAS particles as dimers in all possible orientations (Fig.
4). A variety of shapes are observed in the micrograph, all con-
sistent with the model of two roughly cylindrical particles in an
antiparallel orientation with a linker region, forming an H-shaped
molecule. X-ray solution-scattering analysis (45) confirmed FAS
is a dimer consistent with dimensions described previously by
small angle neutron-scattering analysis (26). Standard single par-
ticle reconstruction methodologies were then followed (46).
Briefly, a dimer of two featureless the ellipsoids with dimensions
160 146 73 (26) was used as an initial model. Projections
of this model in all possible orientations were generated, and in-
dividual FAS particles selected from micrograph were compared
to each projection. This classification process determines the ori-
entation of each particle image. Particles in nearly identical ori-
entations were then aligned relative to each other (in the image
plane) and averaged together to generate class averages (Fig. 5).
Finally, the class averages were combined to produce a new 3-D
model (45). The procedure was repeated by using the new model
as an initial model. This process was iterated until convergence
was achieved. The analysis was performed both with and with-
out imposing 2-fold symmetry (Fig. 6A and B). Both models
showed dimensions of 180 130 75 with monomers sepa-
rated by about 19 and arranged in an antiparallel orientation,
which is consistent with biochemical data (1,2,1720). Each
monomer subunit appears to be subdivided into three structural
domains. The two clefts between the monomers, while asym-
metric, may be the two active centers of fatty acid synthesis (Fig.
6). Recent studies on the structure of TE domain of human FAS
(38), and the preliminary electron cryomicroscopy of the struc-
ture of FAS lacking TE indicated that the C-terminal TE domain
is located at the end of the structural domain I (functional do-
FIG. 5. Several projections of the final refined 3-D model in four orien-
tations are shown with the corresponding class average and some indi-
vidual particle images corresponding to each. In a properly refined
structure, the projections and class averages will be identical except for
noise. In cases of particle heterogeneity often small discrepancies will
appear between the two.
 FATTY ACID SYNTHASE 1051

FIG. 6. Three-dimensional structure of fatty acid synthase (FAS). In AD,

the structure was obtained by applying C2 symmetry and in EH, with-
out symmetry. From left to right, different views of the molecule are
shown, i.e., top view (A and E), side view (B and F), end-on view (C and
G), and the view corresponding to the double-arch view visible in Fig-
ure 5 (D and H). Individual domains of the subunits have been indi-
cated as IIII (A and E). In H, the asymmetric structure, the two clefts
formed by the monomers are labeled as narrow (N) and wide (W). The
proposed locations of the active sites of FAS are indicated by the two
gray oval regions.

main III) of each monomer (Fig. 6), further confirming the an-
tiparallel arrangement of FAS dimer. A strong bridge connects
the two monomers, which may be the combination of the inter-
domain regions of both monomers (Fig. 6).
Although the FAS particle images (Fig. 6) are similar to
what we have proposed based on partial proteolysis and bio-
chemical analysis (Figs. 2), the resolution of the structure
even after analyzing approximately 25,000 particles is only
approximately 20 , suggesting that failure to get a much
higher resolution may be due to inter- and intramolecular
flexibility of the FAS dimer. To understand the inherent con-
formational flexibility of the FAS supermolecular complex,
the domain movements of FAS were analyzed by using a
computational method called quantized elastic deformational
model (normal mode analysis) based on the electron density
maps of the synthase at 19 , shown in Figure 6A, with the
assumption that the monomers of FAS have enormous flexi-
bility (47). This method has the ability to predict large-scale
conformational changes such as domain movements by treat-
ing the protein as an elastic object without the knowledge of
protein primary sequence and atomic coordinates (48). As
shown in Figure 7, this analysis suggested that FAS is a very
flexible molecule. FAS structure has two types of flexible
hinges (Fig. 7). One is the intersubunit connection (interdo- FIG. 7. The motional patterns of the fatty acid synthase (FAS) dimer.
main) and the other is an intrasubunit hinge located between Four lowest-frequency deformational modes are shown. Mode 7 de-
the structural domains I and II (Fig. 7), which are equivalent scribes an in-plane bending motion around the hinge between two
to functional domains III and II (Fig. 2). Despite that the monomers (intersubunit hinge). This is the most flexible hinge in the
system. Mode 8 describes an out-of-plane bending motion around the
dimeric FAS has a symmetric structure, large domain move- same hinge. Modes 9 and 10 describe the motions around two smaller
ments around the hinge region occur in various directions and intrasubunit hinges. They are smaller in magnitude. For each mode, two
allow the molecule to adopt a wide range of conformations opposite conformations (left and right) during harmonic vibration are
(47). These domain movements are likely to be important in shown to illustrate the direction of the motion. The amplitude of the
facilitating and regulating the entire palmitate synthesis by motion was arbitrarily chosen for visual clarity. The arrows are used to
indicate the directions of the motions. The larger circles in modes 7 and
coordinating the communication between components of the 8 indicate the intersubunit hinge, and the smaller circles in mode 9 and
molecule, for instance, adjusting the distance between vari- mode10 indicate the intrasubunit hinges. The dotted lines in mode 10
ous active sites inside the catalytic reaction centers. indicate the longest axes of the subunits.

Lipids, Vol. 39, no. 11 (2004)

1052 S.S. CHIRALA AND S.J. WAKIL
FIG. 8. Experimental verification of conformational variation in fatty
acid synthase (FAS) predicted by normal mode analysis. FAS particle
images (~20,000) were processed by using simultaneous multiple
model single particle analysis (48). This single data set was used to si-
multaneously produce all three structures shown in the top row. The
bottom row shows predicted structures from normal mode analysis
(mode 7) as applied to the symmetric structure from Figure 7.
Based on this theoretically predicted flexibility described
above, FAS dimer might exist in several different conforma-
tional states. To demonstrate the existence of this flexibility
in FAS dimer, approximately 20,000 FAS particle images
were processed by using a simultaneous multiple model sin-
gle particle refinement method to search for their presence
(49). For this analysis only the extreme conformational states
shown in Figure 7 (mode 7) were used. This analysis con-
firmed the presence of these conformational states of FAS
dimers in the electron cryomicrographic images (Fig. 8). This
implies that the resolution limiting factor in our current struc-
tures is structural heterogeneity among the data. That is, our
final structure consists of the average of many different states.
To improve the resolution will require splitting the data up
into many more groups, each of which will then be more
structurally homogeneous. While this type of analysis would
require approximately 106 FAS particles to be analyzed, it
will provide multiple FAS dimer structures at much higher
resolution instead of a single FAS structure at low resolution.
ACKNOWLEDGMENTS
The authors would like to express their thanks to the National Insti-
tutes of Health for their support of this research (GM19091,
GM63115 and GM068826). We thank Drs. Steven J. Ludtke and
Jianpeng Ma for their critical review of the manuscript and provid-
ing some of the figures.
REFERENCES
1. Wakil, S.J., Stoops, J.K., and Joshi, A.C. (1983) Fatty Acid Syn-
thesis and Its Regulation, Annu. Rev. Biochem. 52, 537579.
2. Wakil, S.J. (1989) The Fatty Acid Synthase: A Proficient Multi-
functional Enzyme, Biochemistry 28, 45234530.
3. Smith, S. (1994) The Animal Fatty Acid Synthase: One Gene,
One Polypeptide, Seven Enzymes, FASEB J. 8, 12481259.
4. Smith, S., Witkowski, A., and Joshi, A.K. (2003) Structural and
Functional Organization of Animal Fatty Acid Synthase, Prog.
Lipids, Vol. 39, no. 11 (2004)
Lipid Res. 42, 289317.
5. Jayakumar, A., Tai, M.-H., Huang, W.-Y., Al-Feel, W., Hsu, M.,
Abu-Elheiga, L., Chirala, S.S., and Wakil, S.J. (1995) Human
Fatty Acid Synthase: Properties and Molecular Cloning, Proc.
Natl. Acad. Sci. USA 92, 86958699.
6. Hannun, Y.A., and Obeid, L.M. (2002) The Ceramide-Centric
Universe of Lipid-Mediated Cell Regulation: Stress Encounters
of the Lipid Kind, J. Biol. Chem. 277, 2584725850.
7. Kuhajda, F.P., Jenner, K., Wood, F.D., Hennigar, R.A., Jacobs,
L.B., Dick, J.D., and Pasternack, G.R. (1994) Fatty Acid Syn-
thesis: A Potential Selective Target for Antineoplastic Therapy,
Proc. Natl. Acad. Sci. USA 91, 63796383.
8. Pizer, E.S., Thupari, J., Han, W.F., Pinn, M.L., Cherst, F.J., Fre-
hywot, G.L., Townsend, C.A., and Kuhajda, F.P. (2000) Mal-
onyl-CoA Is a Potential Mediator of Cytotoxicity Induced by
Fatty Acid Synthase Inhibition in Human Breast Cancer Cells
and Xenografts, Cancer Res. 60, 213238.
9. Menendez J.A., Vellon, L., Mehmi, I., Oza, B.P., Ropero, S.,
Colomer, R., and Lupu, R. (2004) Inhibition of Fatty Acid Syn-
thase (FAS) Suppresses HER2/neu (erbB-2) Oncogene Overex-
pression in Cancer Cells, Proc. Natl. Acad. Sci. USA 101,
1071510720.
10. Loftus, T.M., Jaworsky, D.E., Frehywot, G.L., Towsand, C.A.,
Ronnett, G.V., Lane, M.D., and Kuhajda, F.P. (2000) Reduced
Food Intake and Body Weight in Mice Treated with Fatty Acid
Synthase Inhibitors, Science 288, 23792381.
11. Chirala, S.S., Chang, H., Matzuk, M., Abu-Elheiga, L., Mao, J.,
Mahon, K., Finegold, M., and Wakil, S.J. (2003) Fatty Acid
Synthesis Is Essential in Embryonic Development: Fatty Acid
Synthase Null Mutants and Most of the Heterozygotes Die In
Utero, Proc. Natl. Acad. Sci. USA 100, 63586363.
12. Wakil, S.J. (1970) Fatty Acid Metabolism, in Lipid Metabolism
(Wakil, S.J., ed.), pp. 148, Academic Press, New York.
13. Wakil, S.J. (1958) Requirement of Bicarbonate in Fatty Acid
Synthesis, J. Am. Chem. Soc. 80, 2908.
14. Wakil, S.J., Pugh, E.L., and Sauer, F. (1964) The Mechanism of
Fatty Acid Synthesis, Proc. Natl. Acad. Sci. USA 52, 106114.
15. Stoops, J.K., Arslanian, A.W., Oh, Y.H., Aune, K.C., Vanaman,
T.C., and Wakil, S.J. (1975) Presence of Two Polypeptide
Chains Comprising Fatty Acid Synthetase, Proc. Natl. Acad.
Sci. USA 75, 19401944.
16. Stoops, J.K., Arslanian, A.W., Chalmers, J.H., Joshi, V.C., and
Wakil, S.J. (1977) Fatty Acid Synthase Complexes, in Bioor-
ganic Chemistry (Van Tamelien, E.E., ed.), Vol. 1, pp. 339370,
Academic Press, New York.
17. Tsukamoto, Y., and Wakil, S.J. (1988) Isolation and Mapping
of the Beta-Hydroxyacyl Dehydratase Activity of Chicken Liver
Fatty Acid Synthase, J. Biol. Chem. 263, 1622516229.
18. Chirala, S.S., Kasturi, R., Pazirandeh, M., Stolow, D.T., Huang,
W.-Y., and Wakil, S.J. (1989) A Novel cDNA Extension Proce-
dure: Isolation of Chicken Fatty Acid Synthase cDNA Clones,
J. Biol. Chem. 264, 37503757.
19. Holzer, K.R., Liu, W., and Hammes, G.G. (1989) Molecular
Cloning and Sequencing of Chicken Liver Fatty Acid Synthase
cDNA, Proc. Natl. Acad. Sci. USA 86, 43874391.
20. Haung, W.-Y., Chirala, S.S., and Wakil, S.J. (1994) Amino-Ter-
minal Blocking Group and Sequence of the Animal Fatty Acid
Synthase, Arch. Biochem. Biophys. 314, 4549.
21. Amy, C.M., Witkowski, A., Naggert, J., Williams, B., Rand-
hawa, Z., and Smith, S. (1989) Molecular Cloning and Sequenc-
ing of cDNAs Encoding the Entire Rat Fatty Acid Synthase,
Proc. Natl. Acad. Sci. USA 86, 31143118.
22. Beck, K.-I., Schreglmann, R., Stathopulos, I., Klein, H., Hoch,
J., and Schweizer, M. (1992) The Fatty Acid Synthase (FAS)
Gene and Its Promoter in Rattus Norvegicus, DNA Seq. 2,
359386.
23. Paulauskis, J.D., and Sul, H.S. (1989) Structure of Mouse Fatty
 FATTY ACID SYNTHASE
Acid Synthase mRNA: Identification of the Two NADPH Bind-
ing Sites, Biochem. Biophys. Res. Commun. 158, 690695.
24. Chirala, S.S., Huang, W.-Y., Jayakumar, A., Sakai, K., and
Wakil, S.J. (1997) Animal Fatty Acid Synthase: Functional
Mapping and Cloning and Expression of the Domain I Con-
stituent Activities, Proc. Natl. Acad. Sci. USA 94, 55885593.
25. Stoops, J.K., and Wakil, S.J. (1981) Animal Fatty Acid Syn-
thase: A Novel Arrangement of the -Ketoacyl Synthase Site
Comprising Domains of the Subunits, J. Biol. Chem. 256,
51285133.
26. Stoops, J.K., Wakil, S.J., Uberbacher, E.C., and Bunick, G.J.
(1987) Small-Angle Neutron-Scattering and Electron Micro-
scopic Studies of the Chicken Liver Fatty Acid Synthase, J. Biol.
Chem. 262, 1024610251.
27. Singh, N., Wakil, S.J., and Stoops, J.K. (1984) On the Question
of Half or Full-Site Reactivity of Animal Fatty Acid Synthetase,
J. Biol. Chem. 259, 36053611.
28. Rangan, V.S., and Smith, S. (1996) Expression in Escherichia
Coli and Refolding of the Malonyl-Acetyltransferase Domain of
the Multifunctional Animal Fatty Acid Synthase, J. Biol. Chem.
271, 3174931755.
29. Rangan, V.S., and Smith, S. (1997) Alteration of the Substrate
Specificity of the Malonyl-CoA/Acetyl-CoA:Acyl Carrier Pro-
tein S-Acyltransferase Domain of the Multi Functional Fatty
Acid Synthase by Mutation of a Single Arginine Residue, J.
Biol. Chem. 272, 1197511978.
30. Joshi, A.K., Witkowski, A., and Smith, S. (1997) Mapping of
Functional Interactions Between Domains of the Animal Fatty
Acid Synthase by Mutant Complementation In Vitro, Biochem-
istry 36, 23162322.
31. Witkowski, A., Joshi, A.K., Lindqvist, Y., and Smith, S. (1999)
Conversion of -Ketoacyl Synthase to a Malonyl Decarboxy-
lase by Replacement of the Active-Site Cysteine with Gluta-
mine, Biochemistry 38, 1164311650.
32. Witkowski, A., Joshi, A.K., and Smith, S. (1996) Mechanism of
the _-Ketoacyl Synthase Reaction Catalyzed by the Animal
Fatty Acid Synthase, Biochemistry 41, 1088710887.
33. Rangan, V.S., Joshi, A.K., and Smith, S. (2001). Mapping the
Functional Topology of the Animal Fatty Acid Synthase by Mu-
tant Complementation In Vitro, Biochemistry 40, 1079210799.
34. Joshi, A.K., and Smith, S. (1993). Construction, Expression, and
Characterization of a Mutated Animal Fatty Acid Synthase De-
ficient in the Dehydrase Function, J. Biol. Chem. 268,
2250822513.
35. Pazirandeh, M., Chirala, S.S., Huang, W.-Y., and Wakil, S.J
(1989) Characterization of Recombinant Thioesterase and Acyl
Carrier Protein Domains of Chicken Fatty Acid Synthase Ex-
pressed in Escherichia Coli, J. Biol. Chem. 264, 1819518201.
36. Pazirandeh, M., Chirala, S.S., and Wakil, S.J. (1991) Site-Di-
rected Mutagenesis Studies on the Recombinant Thioesterase
Domain of Chicken Fatty Acid Synthase Expressed in Es-
cherichia Coli, J. Biol. Chem. 266, 2094620952.
37. Tai, M.H., Chirala, S.S., and Wakil, S.J. (1993) Roles of Ser101,
Asp236, and His237 in Catalysis of Thioesterase II and of the
1053
C-terminal Region of the Enzyme in Its Interaction with Fatty
Acid Synthase, Proc. Natl. Acad. Sci. USA 90, 18521856.
38. Chakravarty, B., Gu, Z., Chirala, S., S., Wakil, S.J., and Quio-
cho, F.A. (2004) Human Fatty Acid Synthase: Structure and
Substrate Selectivity of the Thioesterase Domain, Proc. Natl.
Acad. Sci. USA 101, 1556715572.
39. Jayakumar, A., Chirala, S.S., and Wakil, S.J. (1997) Human
Fatty Acid Synthase: Assembling Recombinant Halves of the
Fatty Acid Synthase Subunit Protein Reconstitutes Enzyme Ac-
tivity, Proc. Natl. Acad. Sci. USA 94, 1232612330.
40. Chirala, S.S., Jayakumar, A., Gu, Z.-W., and Wakil, S.J. (2001)
Human Fatty Acid Synthase: Role of Interdomain in the Forma-
tion of Catalytically Active Synthase Dimer, Proc. Natl. Acad.
Sci. USA 98, 31043108.
41. Joshi, A.K., Witkowski, A., and Smith, S. (1998) The Mal-
onyl/Acetyltransferase and -Ketoacyl Synthase Domains of the
Animal Fatty Acid Synthase Can Cooperate with the Acyl Car-
rier Protein Domain of Either Subunit, Biochemistry 37,
25152523.
42. Rangan, V.S., Joshi, A.K., and Smith, S. (1998) Differential
Affinity Labeling of the Two Subunits of the Homodimeric An-
imal Fatty Acid Synthase Allows Isolation of Heterodimers
Consisting of Subunits that Have Been Independently Modified,
J. Biol. Chem. 273, 49374943.
43. Witkowski, A., Joshi, A.K., Rangan, V.S., Falick, A.M.,
Witkowska, H.E., and Smith, S. (1999). Dibromopropanone
Cross-Linking of the Phosphopantetheine and Active-Site Cys-
teine Thiols of the Animal Fatty Acid Synthase Can Occur Both
Inter- and Intrasubunit. Reevaluation of the Side-by-Side, An-
tiparallel Subunit Model, J. Biol. Chem. 274, 1155711563.
44. Joshi, A.K., Rangan, V.S., Witkowski, A., and Smith, S. (2003).
Engineering of an Active Animal Fatty Acid Synthase Dimer
with Only One Competent Subunit, Chem. Biol. 10, 169173.
45. Brink, J., Ludtke, S.J., Yang, C.-Y., Gu, Z.-W., Wakil, S.J., and
Chiu, W. (2002) Quaternary Structure of Human Fatty Acid
Synthase by Electron Cryomicroscopy, Proc. Natl. Acad. Sci.
USA 99, 13143.
46. Ludtke, S.J., Baldwin, P.R., and Chiu, W. (1999). EMAN: Semi-
automated Software for High-Resolution Single-Particle Recon-
structions, J. Struct. Biol. 128, 8297.
47. Ming, D., Kong, Y., Wakil, S.J., Brink, J., and Ma, J. (2002) Do-
main Movements in Human Fatty Acid Synthase by Quantized
Elastic Deformational Model, Proc. Natl. Acad. Sci. USA 99,
78957899.
48. Ming, D., Kong, Y., Lambert, M.A, Huang, Z., and Ma, J.
(2002). How To Describe Protein Motion Without Amino Acid
Sequence and Atomic Coordinates, Proc. Natl. Acad. Sci. USA
99, 86208625.
49 Brink, J., Ludtke, S.J., Kong, Y., Wakil, S.J., Ma, J., and Chiu,
W. (2004) Experimental Verification of Conformational Varia-
tion of Human Fatty Acid Synthase as Predicted by Normal
Mode Analysis, Structure (Cambr.) 12, 185191.
[Received October 4, 2004; accepted November 4, 2004]
Lipids, Vol. 39, no. 11 (2004)