Tài liệu Báo cáo khoa học: Fatty acid synthesis Role of active site histidines and lysine in Cys-His-His-type b-ketoacyl-acyl carrier protein synthases ppt

Fatty acid synthesis
Role of active site histidines and lysine in Cys-His-His-type
b-ketoacyl-acyl carrier protein synthases
Penny von Wettstein-Knowles
1
, Johan G. Olsen
2
, Kirsten A. McGuire
1
and Anette Henriksen
2
1 Genetics Department, Molecular Biology and Physiology Institute, Copenhagen University, Denmark
2 Biostructure Group, Carlsberg Laboratory, Copenhagen, Denmark
The formation of carbon–carbon bonds is a funda-
mental biochemical reaction. A number of enzymes
involved in various biosynthetic pathways accomplish
this by different means. Among these is a large family
of enzymes involved in synthesis of fatty acids, waxes,
flavins, natural drugs, and antibiotics making carbon–
carbon bonds by use of the Claisen condensation prin-
ciple. Initially, an active site nucleophile induces a
transesterification by nucleophilic attack on an acyl-
thioester substrate. In the second step, a b-carbanion
thioester is generated by either proton abstraction or
decarboxylation. This strong nucleophile then attacks
the carbonyl carbon of the first ester, resulting in a
b-keto product (Scheme I). b-Ketoacyl-acyl carrier
protein (ACP) synthase {3-oxoacyl-[acyl-carrier-pro-
tein] synthase (E.C. 2.3.1.41)} I (KAS I) and KAS II
from Escherichia coli represent a set of decarboxylating
condensing enzymes, which we refer to as the CHH

Keywords
active site mutations; condensation reaction;
fatty acid synthase; reaction mechanism;
b-ketoacyl-ACP synthase
Correspondence
P. von Wettstein-Knowles, Genetics
Department, Molecular Biology and
Physiology Institute, Copenhagen University,
Øster Farimagsgade 2A, DK-1353
Copenhagen, Denmark
Fax: +45 35322113
Tel: +45 35322180
E-mail:
A. Henriksen, Carlsberg Laboratory,
Biostructure Group, Gamle Carlsberg Vej 10,
DK-2500 Valby, Denmark
Fax: +45 33274708
Tel: +45 33275222
E-mail:
(Received 10 August 2005, revised 2
December 2005, accepted 12 December
2005)
doi:10.1111/j.1742-4658.2005.05101.x
b-Ketoacyl-acyl carrier protein (ACP) synthase enzymes join short carbon
units to construct fatty acyl chains by a three-step Claisen condensation
reaction. The reaction starts with a trans thioesterification of the acyl pri-
mer substrate from ACP to the enzyme. Subsequently, the donor substrate
malonyl-ACP is decarboxylated to form a carbanion intermediate, which in
the third step attacks C1 of the primer substrate giving rise to an elongated
acyl chain. A subgroup of b-ketoacyl-ACP synthases, including mitochond-

rial b-ketoacyl-ACP synthase, bacterial plus plastid b-ketoacyl-ACP
synthases I and II, and a domain of human fatty acid synthase, have a
Cys-His-His triad and also a completely conserved Lys in the active site.
To examine the role of these residues in catalysis, H298Q, H298E and six
K328 mutants of Escherichia coli b-ketoacyl-ACP synthase I were construc-
ted and their ability to carry out the trans thioesterification, decarboxyla-
tion and ⁄ or condensation steps of the reaction was ascertained. The crystal
structures of wild-type and eight mutant enzymes with and ⁄ or without
bound substrate were determined. The H298E enzyme shows residual
decarboxylase activity in the pH range 6–8, whereas the H298Q enzyme
appears to be completely decarboxylation deficient, showing that H298
serves as a catalytic base in the decarboxylation step. Lys328 has a dual
role in catalysis: its charge influences acyl transfer to the active site Cys,
and the steric restraint imposed on H333 is of critical importance for
decarboxylation activity. This restraint makes H333 an obligate hydrogen
bond donor at N
e
, directed only towards the active site and malonyl-ACP
binding area in the fatty acid complex.
Abbreviations
ACP, acyl carrier protein; KAS, b-ketoacyl-ACP synthase; WT–C8, KAS I–octanoyl complex.
FEBS Journal 273 (2006) 695–710 ª 2006 The Authors Journal compilation ª 2006 FEBS 695
group because of the cysteine and two histidine active-
site residues [1–3]. Another group of decarboxylating
condensing enzymes called CHN (N for asparagine),
represented by KAS III and certain polyketide synth-
ases [4–6], catalyze a similar three-step reaction with
an active site composed of a cysteine nucleophile, a
histidine and an asparagine. Although CHN enzymes
have a substantially altered active site structure and

substrate-binding funnel, they are characterized by the
same ababa-fold as the CHH enzymes.
Several conserved active site residues are important
for the course of the b-ketoacyl-ACP synthase reaction
in the CHH group of condensing enzymes. These
include: (a) the cysteine nucleophile (C163 in E. coli
KAS I), with a lowered pK
a
because of a position at the
N-terminus of helix Na3 [3]; (b) two histidines (H298
and H333), promoting decarboxylation of malonyl and
H333 also playing a role in the condensation reaction
and for the pK
a
value of the nucleophile [7,8]; (c) a lysine
(K328), required for decarboxylation and efficient trans-
fer of the substrate to be elongated to C163 [7–9]; (d)
an aspartate (D306) and a glutamate (E309), essential
for decarboxylation [9]; (e) two threonines (T300 and
T302), speculated to contribute to ACP binding during
malonyl-ACP decarboxylation [2,3], and (f) a phenyl-
alanine (F392), forming an oxyanion hole together with
the backbone nitrogen of the nucleophile that promotes
the transfer reaction [10]. Although several studies,
including crystal structures of both CHH and CHN
enzymes [1,4,5,11,12] have contributed to the under-
standing of the role of conserved residues in the active
site of CHH condensing enzymes, and analogies have
been made with the reaction mechanism proposed for
CHN condensing enzymes, a clear consensus about the

exact role of the conserved residues in CHH enzymes
has not emerged [1–5,8,9,12].
In recent years, condensing enzymes have enjoyed
substantial commercial interest. The efficiency and pre-
cision with which these various enzymes carry out syn-
thesis of rather complicated molecules such as ring
systems [13] and wax components [14] are attractive
properties for drug synthesis research. The fatty acid
condensing enzymes have also come into focus as
targets for new antibiotics [6,15–18] and in cancer
treatment [19,20]. A description of the exact role, elec-
trostatic properties, and hydrogen bonding potentials
of active site residues provides an optimized model of
the ligand-binding potential of the active site, enabling
differentiation between the active site properties of
target enzymes to be made. This study probes the roles
of the active site histidines and lysine in the CHH con-
densing enzyme KAS I from E. coli by use of crystal
structures of active site mutants and biochemical
characterization of the acyl transfer, decarboxylation
and ⁄ or condensation steps of the reaction performed
by these mutants.
The results establish that the CHH reaction mechan-
ism is different from that of the CHN enzymes. They
reveal that: (a) K328 imposes steric restraints on H333
that are necessary for maintenance of the hydrogen
bond network required for decarboxylation, and that
its positive charge influences acyl transfer to the active-
site cysteine; (b) H298 functions as a catalytic base in
the decarboxylation reaction, and (c) H333 stabilizes

the negative charge on C163 in the native enzyme,
whereas in the intermediate fatty acyl complex it parti-
cipates in the active site hydrogen bonding network by
donation of a hydrogen bond.
Results
Structure of KAS I and its C8 complex
These structures were determined to ascertain whe-
ther the previously published structures of KAS I
and KAS I–fatty acid complexes based on room tem-
perature data and ester rather than thioester linkages
(C163S mutant protein [3]) would cause erroneous
interpretation of the stereochemistry in the active
site. The only significant differences, apart from the
length of the fatty acids, in the active site between
C163S–C12 and WT–C8 (KAS I–octanoyl complex)
are (a) a slight rotation of H333 with maximal effect
(0.4 A
˚
) at the N
e
position (Fig. 1) and (b) a cation
detected octahedrally co-ordinated in the vicinity of
the active site (Fig. 2). Three of the six cation lig-
ands are main-chain oxygen atoms, and three are the
side-chain oxygen atoms of N296, E342 and S387.
The glutamic acid and asparagine residues are con-
served among known KAS I and KAS II sequences.
The serine residue is generally conserved, but can be
a cysteine (in E. coli KAS II [21,22]) or an asparagine
(in Mycobacterium tuberculosis and Mycobacterium

leprae KAS I and II [23,24]).
Scheme 1.
Histidines and lysine in KAS I ⁄ KAS II catalysis P. von Wettstein-Knowles et al.
696 FEBS Journal 273 (2006) 695–710 ª 2006 The Authors Journal compilation ª 2006 FEBS
The two subunits of the KAS I homodimer have
slightly larger than average discrepancies in the atomic
positions in the segments 318–323 (r.m.s.d. ¼ 0.7 A
˚
)
and 367–373 (r.m.s.d. ¼ 0.8 A
˚
) in an overall super-
imposition (overall average r.m.s.d. ¼ 0.3 A
˚
). In this
respect the subunit pairs AC and BD have smaller
r.m.s.d. values between backbone atoms than other
combinations of subunit pairs. The two segments of
the AC and BD dimer are involved in crystal packing
at the AC interface, and the observed structural diver-
sity is unlikely to be of biochemical significance. The
same is true for the mutated KAS I structures.
The distance between H333 N
e
and C163 S
c
is 3.2 A
˚
and 3.1 A
˚

in the native and fatty acid complex,
respectively, and infers that H333 N
e
donates a hydro-
gen bond to the nucleophile, although the Cys163
C
b
–Cys163 S
c
–His333 N
e
angle is not optimal (87° and
96°, respectively) [25]. The formation of the fatty acid
complex is accompanied by the emergence of a well-
defined water molecule within hydrogen bond distance
of His333 N
e
(Fig. 3A,B).
Structures of KAS I H298E and the H298E–C12
complex
The overall structure of the H298E mutant is the same
as that of the wild-type, but the active site substructure
presents a few changes in amino acid side-chain orienta-
tions, as revealed by the superimposition of the two
structures in Fig. 3C. As opposed to H298 (Fig. 3A,B),
E298 is involved in hydrogen bonds through both O
e
atoms (Fig. 3D,E). One side-chain oxygen is hydrogen
bonded to F390 N (3.0 A
˚

), and the other to the O
c
atom
of T300 (2.9 A
˚
). T300 is reoriented (Fig. 3C,D,E) and
cannot contribute to malonyl-ACP binding as proposed
on the basis of the C163S structure [3]. T302 does not
change orientation (Fig. 3D). The orientation of the
conserved active site residues H333 and K328 are not
affected by the H298E mutation (Fig. 3C). H333 is
hydrogen bonded to the backbone N of L335, making it
a potential hydrogen bond donor to the active site, and
probably lowers its pK
a
considerably. K328 shares a
bidentate hydrogen bond with E342 and is within
hydrogen bond distance of the E298 backbone O
(Fig. 3D).
The H298E–C12 structure (Fig. 3E) is the same as
that of H298E except that an extra water molecule
appears well defined in the active site. A water mole-
cule in this position is also present in some of the sub-
units in the H298E structure, which is of considerably
poorer quality (R ⁄ R
free
¼ 21.7 ⁄ 27.2; Table 1 [26]). The
formation of the acyl–thioester bond in the H298E–
C12 structure has no impact on the orientation of
T300 (Fig. 3D versus Fig. 3E).

Structures of KAS I H298Q and the H298Q–C12
complex
The H298Q overall structure, including that of its act-
ive site, is similar to those of H298E and H298E–C12.
Fig. 2. The cation site as observed in both KAS I and mutant struc-
tures. The color coding is as in Fig. 1.
Fig. 1. Superimposition of the KAS I C163S-C12 (white, light colors)
[3] and KAS I–C8 (gray, dark colors) active sites. Red spheres are
water molecules. Blue atoms represent nitrogen, red represent
oxygen, and green represents sulfur. Figures 1, 2 and 4 are made
in
MOLSCRIPT [41].
P. von Wettstein-Knowles et al. Histidines and lysine in KAS I ⁄ KAS II catalysis
FEBS Journal 273 (2006) 695–710 ª 2006 The Authors Journal compilation ª 2006 FEBS 697
The active site residue 298Q is involved in hydrogen
bonds through both the O
e
and the N
e
atoms
(Fig. 3F). In this case the side-chain oxygen is hydro-
gen bonded to F390 N (2.9 A
˚
), and the N
e
to the O
c
atom of T300 (3.0 A
˚
). H333 and K328 have insignifi-

cant variations in orientations, but the position of the
390–394 backbone is shifted. The largest effect is seen
for residue F390, which is shifted by  0.9 A
˚
(Fig. 3H).
The formation of the H298Q–C12 complex
(Fig. 3G) induces side-chain reorientation of residue
298Q (Fig. 3H), a shift in the position of the 390–394
backbone (Fig. 3H) to that found in the H298E ⁄
H298E–C12 structures, and a side-chain reorientation
Fig. 3. The active sites of the wild-type KAS I, its H298 mutants and their acyl complexes. (A) Wild-type. (B) WT–C8. (C) Superimposition of
the wild-type (white, light colors) and H298E (orange, dark colors). (D) H298E. (E) H298E–C12. (F) H298Q. (G) H298Q–C12. (H) Superimposi-
tion of H298Q and H298Q–C12. In (A, B) and (D–G), water molecules (red spheres) within hydrogen bonding distance are indicated with
dashed lines. (H) Superimposition of H298Q (orange, dark colors) and H298Q–C12 (white, light colors) not including water molecules. Figure
prepared using
PYMOL [42].
Histidines and lysine in KAS I ⁄ KAS II catalysis P. von Wettstein-Knowles et al.
698 FEBS Journal 273 (2006) 695–710 ª 2006 The Authors Journal compilation ª 2006 FEBS
of T300 to that resembling the orientation found in
the structure of the native enzyme and the WT–C8
complex (Fig. 3A,B,G). A water molecule is found
between H333 N
e
and F390 N in three of the four sub-
units of H298Q–C12 (Fig. 3G). It is not possible to
unambiguously determine the hydrogen bonding
pattern in the active site of the H298Q complex, but
dotted lines have been included to atoms within hydro-
gen bonding distance in Fig. 3G.
Structure of KAS I K328A

The overall structure of the K328A mutant is the same
as that of the wild-type (Fig. 3A), but the active site
substructure has changed. (a) In the absence of the
K328 side chain, a solvent molecule occupies the posi-
tion of the K328 N
f
atom (Fig. 4A). The hydrogen
bonds and distances imply that the properties of the
solvent molecule are similar to the properties of the N
f
atom (compare Fig. 3A and Fig. 4A). (b) Relief of the
steric constraints normally imposed on H333 by the
side chain of K328 results in an altered rotamer con-
formation in the mutant (Fig. 4A). The changed H333
position facilitates formation of a 3.1 A
˚
hydrogen
bond to the solvent molecule, and contrary to the
situation in the wild-type, H333 N
d
does not accept a
hydrogen bond from the L335 backbone nitrogen
(Fig. 4A). Unambiguous determination of which H333
N will carry the proton is therefore not possible unless
the solvent molecule hydrogen bonded to H333 N
d
is
an ammonium ion. Moreover, the hydrogen bond dis-
tance between H333 N
e

and C163 S
c
(3.2 A
˚
) in the
mutant is very similar to the 3.3 A
˚
found in the wild-
type and infers that H333 N
e
donates a hydrogen bond
to the nucleophile, although the Cys163 C
b
–Cys163
S
c
–His333 N
e
angle (79°) is less favorable than in the
wild-type (87°) [25]. Thus, we have introduced an
ammonium ion at this solvent site in our model, an
assignment that is further justified by the fact that the
crystals were obtained in the presence of 1.9 m
(NH
4
)
2
SO
4
.

Structure of the KAS I K328A–C12 complex
Electron density calculated using phases derived from
the K328A polypeptide structure revealed electron
density corresponding to a fatty acid bound to all
four C163 residues in the asymmetric unit. The fatty
acid residues were included in the model and refined.
The structure is similar to the WT–C8 complex
structure, but with the removal of the lysine side
chain, H333 relaxes to a new rotamer conformation
(Fig. 4B) and interacts via N
d
with a solvent mole-
Table 1. Data collection and refinement statistics.
WT–C8 WT H298Q H298Q–C12 H298E H298E–C12 K328A K328A–C12 K328R
Resolution (A
˚
) 15.0–2.40 29.7–1.9 26.1–1.86 116.4–1.95 32.5–2.20 31.6–2.00 29.8–1.70 37.9–2.00 29.9–2.00
R
merge
a
(%) 7.6 (17.9)
b
12.4 (53.4)
b
6.2 (23.0)
b
8.6 (25.2)
b
10.7 (15.5)
b

8.6 (25.0)
b
6.5 (19.8)
b
9.2 (19.3)
b
6.1 (14.3)
b
Average I ⁄ rI 8.4 4.4 9.5 6.8 5.0 6.6 10.3 6.1 11.2
Average redundancy 6.6 2.2 4.3 4.5 2.9 5.8 2.7 4.3 3.5
Completeness 91.2 (80.3)
b
89.9 (60.8)
b
96.1 (75.2)
b
94.8 (80.7)
b
88.2 (72.7)
b
98.1 (96.9)
b
96.8 (86.8)
b
98.1 (92.8)
b
90.8 (90.1)
b
R
factor

c
⁄ R
free
d
(%) 19.3 ⁄ 23.1 21.7 ⁄ 25.6 20.4 ⁄ 23.4 18.9 ⁄ 21.8 21.7 ⁄ 27.2 19.6 ⁄ 22.9 18.1 ⁄ 20.8 17.1 ⁄ 20.6 17.8 ⁄ 22.7
Number of reflections
used in the refinement
62307 124549 144860 121622 79506 117097 180614 216172 106579
Number of nonhydrogen atoms 12722 12825 13012 12941 9492 12929 13003 13071 12804
Mean B-factor (A
˚
2
) 18.6 25.7 20.3 17.3 18.0 17.7 17.5 17.2 19.6
Cross validated estimated
maximum
coordinate error (A
˚
)
e
0.3 0.3 0.2 0.2 0.3 0.2 0.2 0.2 0.3
R.m.s.d. bond lengths (A
˚
) 0.006 0.006 0.005 0.006 0.012 0.005 0.005 0.005 0.015
R.m.s.d. bond angles (°) 1.2 1.2 1.1 1.2 1.5 1.2 1.2 1.2 1.7
a
R
merge
¼ S|I –<I >|⁄SI, with I being the intensity of the measured reflection and < I > the mean intensity of that reflection.
b
Outer resolution shell.

c
R
factor
¼ S(|Fo|–|Fc|) ⁄S|Fo|.
d
R
free
is the same as R
factor
, but calculated with the 5% of the total number of observations not used in the refinement.
e
As determined from a Luzzati plot [26]. WT, wild-type.
P. von Wettstein-Knowles et al. Histidines and lysine in KAS I ⁄ KAS II catalysis
FEBS Journal 273 (2006) 695–710 ª 2006 The Authors Journal compilation ª 2006 FEBS 699
cule as in the unbound K328A structure. Interest-
ingly, the water ⁄ ion structure around H298 changes
when the fatty acid is bound to K328A (Fig. 4B), in
contrast with the wild-type case (Fig. 3B). Contrary
to the situation in K328A (Fig. 4A), any suggestions
as to the nature of the solvent molecule found
between E342 and H333 cannot be justified, because
position and potential hydrogen bonds are shifted
(Fig. 4B). A water molecule has been included in the
model at this position.
Structure of KAS I K328R
Four subunits arranged in two dimers, AB and CD,
form the KAS I asymmetric unit in the P2
1
2
1

2
1
space group obtained in all KAS I crystallization
experiments so far [3,11]. The relatively good quality
of the K328R diffraction data makes it possible to
determine the structure of each K328R subunit inde-
pendently. In K328R, subunits A and C have identi-
cal orientation of active site residues with a rotated
H333 (Fig. 5A), corresponding to the situation in
K328A (Fig. 4A). N
g1
and N
g2
of R328 interact
with E342 by forming a salt bridge. K328 makes a
similar interaction with E342 in the wild-type enzyme
(Fig. 3A). R328 also interacts with H333 by dona-
ting a hydrogen bond from N
e
to H333 N
d
(2.7–
2.9 A
˚
) (Fig. 5A). The situation is a bit different for
subunits B and D. This pair of subunits also shares
orientation of active site residues although with a
higher positional r.m.s.d. than found between sub-
units A and C. Surprisingly, R328 interacts with
H333 via N

g1
rather than via N
e
(Fig. 5B). The
H333 rotamer falls between the wild-type orientation
and the orientation observed in K328A, with N
d
being oriented more towards the backbone N of resi-
due 335 (on average the H333 N
d
–L335 N distance
is 3.9 A
˚
in subunit A and C versus 3.5 A
˚
in subunit
B and D). Nevertheless, the shortest interatomic dis-
tance from H333 N
d
to R328 N
g1
is on average
2.9 A
˚
with the average H333 N
e
–C163 S
c
distance
being 3.2 A

˚
. As only the K328R mutant shows vari-
ation in the orientation of residue 328, whereas all
structures have variations in the 318–323 and
367–373 fragments, this difference in residue 328 ori-
entation is significant and cannot be ascribed to pro-
pagation of the crystal-packing effects observed in
fragments 318–323 and 367–373.
A
B
Fig. 4. Hydrogen bonding patterns in the
active sites of KAS I K328A. (A) The hydro-
gen bonding pattern in K328A resembles
the pattern observed in the wild-type,
except for rotation of the H333 side chain.
A solvent molecule is located at a position
corresponding to the wild-type K328 N
f
position. The solvent molecule is proposed
to be a NH
4
+
ion and is represented by a
blue sphere. (B) In the K328A–C12 complex,
the hydrogen bond pattern changes, and the
solvent molecule found hydrogen bonded to
H333 N
d
has been modeled as a water
molecule. Water molecules are represented

by red spheres.
Histidines and lysine in KAS I ⁄ KAS II catalysis P. von Wettstein-Knowles et al.
700 FEBS Journal 273 (2006) 695–710 ª 2006 The Authors Journal compilation ª 2006 FEBS
Decarboxylation of malonyl-ACP by wild-type
and mutant KAS I proteins
The ability of the wild-type and mutant KAS I proteins
to form acetyl-ACP from malonyl-ACP (Scheme 1) was
measured by visualizing the decarboxylation of
[2-
14
C]malonyl-ACP to [2-
14
C]acetyl-ACP. In the assay,
the substrate is synthesized from radiolabeled malonyl-
CoA and ACP by malonyl-CoA–ACP transacylase
before the addition of the KAS protein to be tested.
The amount of labeled substrate generated was inde-
pendent of pH over the tested range (3–8), generating
adequate substrate for the decarboxylation reaction, as
illustrated in Fig. 6A for pH 6.8 and 4, lanes 1 and 10,
respectively. Figure 6A (lanes 2–5) illustrates for the
wild-type enzyme the rapid decrease in the malonyl-
ACP substrate and the much slower appearance of the
product acetyl-ACP at pH 6.8 in assays from 1 to
30 min in length. Analogous results were obtained at
pH 6 and 8. Reducing the pH to 5 results in slower loss
of the malonyl-ACP substrate (compare Fig. 6A, lanes
2 and 6), and only at 30 min can the acetyl-ACP prod-
uct be detected (Fig. 6A, lanes 6–9). At pH 4 loss of the
malonyl-ACP substrate was first visible in the 30 min

assay (lane 14).
A previous study comparing decarboxylation activit-
ies of the E. coli enzymes revealed that, in contrast
with KAS I, no acetyl-ACP was recovered in KAS II
assays even after 30 min at pH 6.8 [9,27]. That the
KAS II enzyme was active in decarboxylation was con-
firmed in elongation assays that resulted in the synthe-
sis of long-chain acyl-ACPs. We suggested [9] that
perhaps the acetyl carbanion resulting from this
decarboxylation was transferred directly to the active-
site cysteine, as happens in the synthesis of triacetic
acid lactone, which is a derailment product in E. coli
[28] and humans [10], but the natural product of the
chalcone synthase related enzyme in Gerbera hybridia
[29]. Triacetic acid lactone would not be precipitated
in our assay designed to visualize acyl-ACPs. The lag
A
B
Fig. 5. The active sites of the K328R mutant
protein. (A) Active site of the
C monomer;
(B) active site of the
D monomer. Dotted
lines represent possible hydrogen bonds or
salt bridges. A W indicates the water mole-
cule found to interact with R328 N
g
in the
C and D monomer.
P. von Wettstein-Knowles et al. Histidines and lysine in KAS I ⁄ KAS II catalysis

FEBS Journal 273 (2006) 695–710 ª 2006 The Authors Journal compilation ª 2006 FEBS 701
periods observed between malonyl-ACP disappearance
and acetyl-ACP appearance in the present experiments
rule out the triacetic acid lactone hypothesis, as triace-
tic acid lactone cannot breakdown to give acetyl-ACP.
Either the acetyl carbanion has a significant lifetime in
the absence of an acyl acceptor or it is transferred to
an unknown, nonprecipitable intermediate before for-
mation of acetyl-ACP. Additional studies will be
required to unravel this unexpected phenomenon. In
the present experiments, loss of substrate gives an
unambiguous picture of the enzyme’s ability to decarb-
oxylate the extender substrate.
At pH 6.8 in 30 min assays, the H298E mutant
evinced only a slight decrease in the malonyl-ACP sub-
strate unaccompanied by formation of acetyl-ACP
(Fig. 6B, lanes 4 and 5), which can be compared with
the wild-type activity (lanes 2 and 3). The H298E
activity is about the same as that exhibited by the
wild-type enzyme at pH 5 in assays approaching
30 min in length (Fig. 6A, lanes 8 and 9), or better
than the wild-type activity at pH 4 in 30 min assays
(Fig. 6A, lane 14). At and below pH 5, H298E was
inactive even in 60 min assays, as was H298Q in the
pH range tested (pH 6–8; Fig. 6B, lanes 6 and 7).
Thus, H298E is able to decarboxylate at pH 6.8, albeit
with a much reduced efficiency compared with the
wild-type, whereas H298Q appears to be totally inhib-
ited.
The decarboxylase assays with the K328 mutants

were carried out at pH 6.8. The amount of labeled
substrate at the start of the 10 min or 30 min reaction
is shown in Fig. 7A, lane 3, and Fig. 7B, lane 1,
respectively. The wild-type protein accomplished essen-
tially 100% formation of acetyl-ACP in 30 min assays
even at the lowest protein concentration of 2.2 lm
(Fig. 7A lanes 4–6) [9], although no conversion could
be detected in 10 min assays using 0.3 lm wild-type
protein. In analogous assays with the C163A mutant,
by comparison, acetyl-ACP was readily formed
pH 6.8
pH 4
pH 5
1
101110 10 3030 30 55500
min
Mal-ACP
Ac-ACP
Mal-ACP
Ac-ACP
7.2
3.6
7.2 3.65.0 2.5
WT
H298QH298E
0
µM
C
1543111213141028769
15432876

A
B
Fig. 6. Decarboxylation activities of the wild-type KAS I and H298 mutants. Reactions were carried out as detailed in Experimental proce-
dures. (A) Time courses (0–30 min) of [2-
14
C]malonyl-ACP (Mal-ACP) decarboxylation to acetyl-ACP (Ac-ACP) by 5 lM wild-type at pH 6.8, 5
and 4. The amount of labeled malonyl substrate generated at pH 6.8 and 4 and present at the start of the reactions are shown at time 0 in
lanes 1 and 10, respectively. (B) Decarboxylation at pH 6.8 by two different amounts of wild-type, H298E and H298Q proteins in 30 min
assays. The amount of labeled substrate present at the start of the reactions is shown in the first lane. C is a standard consisting of malo-
nyl-ACP and acetyl-ACP (lane 8). WT, wild-type.
0.1
2.2 4.38.6 4.94.3 2.2 2.20 9.7 8.70.8
Mal-ACP
Ac-ACP
154311121028769
A
µM
WT
30
K328H
30
K328R
30
C163A
10
min
10
B
8.7
3.6 4.32.2 4.37.1 2.2 2.24.3 8.7 8.70

Mal-ACP
Ac-ACP
13154311121028769
µM
K328I K328E
K328A
K328F
1.8
Fig. 7. Decarboxylation activity of mutant K328 KAS I proteins. The
concentration (l
M) of KAS added per 100 lL assay is shown as
well as the assay time (min). The reaction was carried out as
detailed in Experimental procedures. After resolution of the ACP
species by electrophoresis on conformationally sensitive 13.3%
polyacrylamide ⁄ 1
M urea gels, the proteins were blotted to a
poly(vinylidene difluoride) membrane followed by autoradiography.
(A) Wild-type and basic mutant KAS I proteins compared with the
very active C163A mutant protein. Lane 3 represents the amount
of labeled substrate at the start of the 10 min reaction. (B) The
bulky and acidic mutant proteins compared with the inactive K328A
protein. Lane 1 represents the amount of labeled substrate at the
start of the 30 min reaction. WT, wild-type.
Histidines and lysine in KAS I ⁄ KAS II catalysis P. von Wettstein-Knowles et al.
702 FEBS Journal 273 (2006) 695–710 ª 2006 The Authors Journal compilation ª 2006 FEBS
(Fig. 7A, lanes 1 and 2), whereas hardly any activity
could be detected, either as loss of the malonyl sub-
strate or accumulation of the acetyl product, for the
six lysine mutants (data not shown). In the 30 min
assays, however, the two basic mutants, K328H and

K328R, and the bulky mutant, K328F, evinced some
activity (Fig. 7A, lanes 7–12, and Fig. 7B, lanes 2–4).
By comparison, the acidic mutant K328E and to a les-
ser extent the bulky mutant K328I appear totally
decarboxylation deficient (Fig. 7B, lanes 5–10), which
is characteristic for K328A (lanes 11–13) as shown pre-
viously in 10 min assays [9]. Only trace amounts of
label are seen in the position characteristic of acetyl-
ACP in the K328I and K328E lanes. That more sub-
strate appeared to be present in some of the assay
lanes than at the start of the assay (Fig. 7B, lane 1)
results from the continued activity of the malonyl-
CoA–ACP transacylase.
To summarize, the K328 isoleucine, glutamic acid
and alanine mutants appear to totally lack decarboxy-
lation activity, whereas the histidine, arginine and phe-
nylalanine mutants are active, although less efficient
than the wild-type.
Transfer of fatty acid from ACP to the wild-type
and K328 mutant KAS I protein
The initial step in the Claisen condensation carried out
by a CHH group enzyme is transfer of the acyl sub-
strate from the phosphopantetheine arm of ACP to its
active site cysteine (Scheme 1). With the use of ACP
carrying
3
H-labeled myristate (C
14
fatty acid), this
transfer can be readily monitored with the aid of a

size-exclusion column and scintillation counting of the
fractions. We have previously shown [9] that, under
saturation conditions, wild-type KAS I transfers 42%
of the labeled myristate, whereas 100% transfer is
accomplished by KAS II under the same conditions
[9]. That additional transfer to wild-type KAS I does
not occur has been attributed to inhibition by free
ACP released during the reaction [9]. The K328A
mutant, in comparison with the wild-type, exhibits
very unusual, apparent sigmoidal kinetics for the trans-
fer reaction [9]. Thus, although in 10 min assays the
transfer efficiency is only 68% of that characterizing
the wild-type, transfer continues until all the myristate
is bound to K328A, revealing that, in this mutant,
ACP does not inhibit transfer. To probe the basis for
the noted difference in transacylation between the
wild-type and K328A [9], transfer to five additional
K328 mutants was investigated.
The results presented in Table 2 were obtained using
about the same amount of protein per assay that
resulted in maximum transfer by the wild-type. The
transfer efficiencies of the mutant proteins relative to
that of the wild-type can be deduced by dividing the
percentage transfer per lg KAS protein found for the
mutants by that characterizing the wild-type. On this
basis the mutants exhibit marked differences from the
wild-type in transfer efficiency in the 10 min assays.
The basic mutants K328H and K328R are more effi-
cient (247% and 227%), the acidic mutant K328E is
less efficient (31%), and the bulky mutants K328F and

K328I are similar to the wild-type (91% and 117%).
Increasing the assay time resulted in an increase in
transfer to the acidic and bulky mutants, so that by
120 min all three were somewhat more efficient than
the wild-type (130–145%). That the acidic and basic
mutant enzymes accept myristate more readily than
the wild-type is in accord with the observation that
they, like K328A, are insensitive to ACP inhibition.
This infers that the bulky mutants are unlikely to be
inhibited by ACP as they also accept more myristate
than the wild-type. Detailed analyses of the bulky
mutants revealed apparent sigmoidal kinetics for trans-
fer (data not shown), but with much lower slopes than
that characterizing the K328A mutant [9]. In these
assays, when maximum transfer was reached, the
transfer efficiencies were similar to that of the wild-
type (99% and 118%).
Whereas carrying out the assays on ice had no effect
on the wild-type, the transfer efficiencies of the mutant
proteins were impeded at the lower temperature
(Table 2). Although the K328H and K328R mutants
had considerable activity at 4 °C (142% and 82% of
Table 2. Transfer of [
3
H]myristate from ACP to wild-type and
Lys328 mutant KAS I proteins. The percentage myristic acid (C
14
)
transferred from C
14

–ACP to the specified KAS proteins was
determined at 10, 60 and ⁄ or 120 min at 22 °Cand⁄ or 4 °C, using
 6 lg protein, at which concentration maximum transfer to wild-
type occurs. WT, Wild-type; –, not measured; tr, < 0.1% of wild-
type activity.
KAS I
% transfer
22 °C, 10 min
% transfer per lg KAS protein
22 °C4°C
10 min 60 min 120 min 10 min
WT 42.5
a
6.4 6.3 – 6.2
K328A 30.1
b
4.1 13.4 – tr
K328H 94.7
b
15.8 16.0 – 8.8
K328R 86.9
b
14.5 14.7 – 5.1
K328E 12.1
b
2.0 4.2 8.5 tr
K328F 34.9
c
5.8 7.5 8.3 tr
K328I 45.1

c
7.5 8.5 9.3 tr
a
Sensitive to ACP;
b
not sensitive to ACP;
c
sensitivity to ACP not
determined.
P. von Wettstein-Knowles et al. Histidines and lysine in KAS I ⁄ KAS II catalysis
FEBS Journal 273 (2006) 695–710 ª 2006 The Authors Journal compilation ª 2006 FEBS 703
that of the wild-type), this was only 58% and 35% of
their activity at 22 °C. At best, a trace of activity
(< 0.1% of that of the wild-type) was detected for the
other four mutants. For all six mutants, adding of
additional KAS protein to the assay resulted in
increased transfer, albeit with lower efficiencies. For
example, 12 lg K328F and K328I gave 7.3% and
17.9% transfer, respectively. The K328A mutant
exhibited only 4% transfer with 15.7 lg protein in a
30 min assay, and K328E 10% with 68 lgina2h
assay. Combined, these results indicate that a posi-
tively charged residue at position 328 is an important
factor for efficient transacylation activity, especially at
4 °C.
Elongation activity of the wild-type and K328
mutant KAS I proteins
The transfer and decarboxylation partial reactions
characterizing the six K328 mutants, as described
above, differ from their respective wild-type partial

reactions. To determine if the mutants were neverthe-
less able to carry out the Claisen condensation
(Scheme 1), their ability to restore activity to elonga-
tion defective protein extracts of the CY244 KAS I ⁄
KAS II double mutant E. coli strain was determined.
As shown in Fig. 8, extracts of this strain are unable
to extend radiolabeled acetate (lane 5) in 30 min assays
at 42 °C. Addition of wild-type KAS I resulted in syn-
thesis of long-chain saturated and unsaturated fatty
acyl-ACPs in a protein concentration-dependent man-
ner (lanes 2–4). Although the basic mutant proteins
K328H and K328R were able to form saturated acyl-
ACPs, more than 70 times as much protein was
required (lanes 6–11). These results demonstrate that,
despite the failure to recover the acetyl-ACP extender
unit in the decarboxylase assays, this substrate is suc-
cessfully formed and available for the condensation
step. By comparison, none of the other four mutants
were capable of elongating the acetyl-CoA primer sub-
strate, i.e. the same picture was obtained in gels as
shown for the CY244 extract in the absence of func-
tional KAS I protein (lane 5) even when 1.11 lg
mutant protein was used. This implies that, although
K328F decarboxylation activity was equivalent to that
characterizing K328H and K328R, its modified trans-
fer activity prohibited condensation at a level that
could be detected in our assay.
Discussion
The conserved cation-binding site
A possible cation site was discovered close to the act-

ive site. This binding site is unaffected by the H298
mutations and is conserved in all published KAS I and
KAS II sequences. The nature of the bound cation has
been determined based on the refined B-factors, the
electron density level, and the hydrogen bond distan-
ces. Models of NH
4
+
,Na
+
and K
+
were constructed,
and the NH
4
+
model best fitted the observed electron
density. The cation site can be detected in all published
CHH class structures [2,3], but has only been described
as a cation site in the crystal structure of Streptococcus
pneumoniae KAS II [30] and in the structure of mito-
chondrial KAS from Arabidopsis thaliana [31]. The
crystals of S. pneumoniae KAS II were grown in the
presence of 250 mm magnesium acetate and revealed a
magnesium ion in this site, whereas the cation in the
mitochondrial KAS structure grown in 1.6 m
(NH
4
)
2

SO
4
was interpreted as potassium. Our crystal-
lization conditions included an ammonium ion concen-
tration of 1.9 m, which probably influences the nature
of the cation under crystallization conditions. The
function of the ion is presumably to fix E342 in the
optimal position and induce the correct pK
a
of this
residue.
Altered transfer kinetics
At least four factors play a role in transfer kinetics
under the assay conditions tested: firstly, the ability of
the enzyme to attract acyl-ACP, secondly the nucleophi-
licity of C163, thirdly the degree to which the enzyme
stabilizes the acyl–enzyme complex, and finally the
WT
K328R
K328H
0.140 1.110.14 0.550.015 0.14 0.550.55 1.11
µg KAS
1543111028769
C#
12
18:1
14
16
18
M

Fig. 8. Ability of the wild-type and mutant K328 KAS I proteins to
enable synthesis of fatty acyl-ACPs by soluble protein extracts of the
E. coli mutant strain CY244 under restrictive conditions. The reaction
was carried out as described in Experimental procedures for 30 min
at 42 °C with 0.015–1.11 lg KAS protein per assay. After resolution
of the ACP species by electrophoresis on conformationally sensitive
13.3% polyacrylamide ⁄ 4
M urea gels, the proteins were blotted to a
poly(vinylidene difluoride) membrane followed by autoradiography.
Addition of up to 1.11 lg K328F, K328I, K328E and K328A mutant
proteins gave the same result as when no KAS protein (O) was
added. M ¼ marker,
14
C
16
-ACP. WT, wild-type.
Histidines and lysine in KAS I ⁄ KAS II catalysis P. von Wettstein-Knowles et al.
704 FEBS Journal 273 (2006) 695–710 ª 2006 The Authors Journal compilation ª 2006 FEBS
sensitivity of KAS I to ACP. Thus far, the last of these
is the only factor identified that can explain why not all
of the wild-type enzyme in an assay actually reacts with
acyl-ACP to form the fatty acid–enzyme complex [9].
The reaction is irreversible under the present conditions
[9]. Interestingly, all the K328 mutants exhibited a
higher efficiency of transfer than the wild-type. The sig-
nificance of this observation is not immediately appar-
ent, as transfer by the four tested lysine mutants (Ala,
His, Arg, Glu) is not influenced by free ACP whereas
transfer by the wild-type is. A future introduction of
analogous mutants in KAS II, a CHH group enzyme

not inhibited by ACP [9], would not only make it poss-
ible to distinguish between reaction rate and ACP inhi-
bition effects, but conceivably also correlate ACP
inhibition with structural properties. The observed con-
sistent substructural differences between subunits within
the asymmetric unit of the crystals studied may imply
the presence of co-operativity between active sites [3].
The possibility that the K328 mutants have altered
co-operativity cannot be excluded. If true, the resulting
effect may be reflected in the acyl-transfer kinetics of
the various mutants.
The proposed model for transfer of the acyl chain
from ACP to KAS III, a CHN group enzyme, consists
of two steps: specific, albeit weak, binding of ACP to
the docking site followed by a conformational change in
ACP proposed to inject the acyl chain into the sub-
strate-binding pocket [32]. If a similar mechanism is
involved in the KAS I reaction then an intermediate
ACP-acyl–KAS I crystal complex remains to be
defined. Such intermediate structures should assist in
interpreting the sigmoidal kinetics of the transfer char-
acterizing the K328A, K328I and K328F mutants as
well as the observed transfer efficiencies.
Decarboxylation
Several theories have been proposed for the decarboxy-
lation mechanism of the CHH family of condensing
enzymes. Huang et al. [1] initially inferred that the
E. coli KAS II residue equivalent of KAS I His298
was positively charged, the charge guiding the localiza-
tion of malonyl-ACP and stabilizing the formation of

the carbanion intermediate. The conserved K328 was
thought to play a structural rather than functional
role. Moche and coworkers [2] later found that the
proximity between the conserved lysine and the H298
equivalent in the 1.5 A
˚
resolution structure of KAS II
from Synechocystis sp., in which no water molecule is
found between H298 and K328, indicated that H298
was unsuitable as proton acceptor. Olsen et al. [3] sug-
gested that K328 was functional in catalysis through
its interaction with H298. This model hypothesized
that hydroxide ion like properties of the solvent mole-
cule, found between K328 and H298 N
d
in KAS I and
most KAS II structures, enabled His298 to function as
a general base in decarboxylation. Price et al. [12] pro-
posed that decarboxylation followed similar mecha-
nisms in CHH and CHN enzymes, and that both
active site histidines donated a hydrogen bond to the
malonyl-ACP thioester oxo group, hence no acid-base
catalysis. Subsequently, Witkowski and coworkers [8]
proposed a decarboxylation mechanism whereby the
rat fatty acid synthase equivalent of KAS I H333 is
protonated and acts as a catalytic acid in decarboxyla-
tion, whereas the KAS I H298 equivalent donates a
hydrogen bond to the malonate carboxylate group in
the same process. This mechanism includes a nucleo-
philic attack by a water molecule on C3 of the malonyl

residue and the formation of bicarbonate as a reaction
product. While this manuscript was in preparation,
White et al. [33], in a review covering type II fatty acid
synthesis, postulated that His298 is a general base acti-
vating a water molecule for nucleophilic attack on the
carboxylate group of malonyl-ACP, and H333 stabil-
izes charge relocalization by hydrogen bond donation.
As is most obvious from the present structural stud-
ies, the role of H333 changes from a dual one in native
enzymes, namely, stabilizing the negative charge of
C163 and donating a hydrogen bond to an active site
water molecule, to only the latter in acyl–KAS com-
plexes. This is revealed by the better defined active site
water molecule in the structures of fatty acyl com-
plexes than in the native structures. The stereochemical
effect of K328 on His333 is critical. This functionality
of K328 derives from its aliphatic side chain imposing
steric constraints on H333 so that it is an obligate
proton donor at N
e
. The crystal structures of
K328 ⁄ K328A–C12 and K328R and the results of the
decarboxylation assays of the ‘bulky’ mutants all point
to the orientation of H333 and its hydrogen bond
donor properties as being the cornerstone of obtaining
and ⁄ or retaining residual decarboxylase activity. The
emerging picture of well-defined water molecules
hydrogen bonded to H333 N
e
in the fatty acid com-

plexes and the critical influence that the orientation of
H333 has on decarboxylation shows that H333 is an
obligate hydrogen bond donor at N
e
, directed only
towards the active site and malonyl-ACP binding area
in the fatty acid complex. The hydrogen bond between
the L335 backbone nitrogen and H333 N
d
makes it
unlikely that H333 participates in catalysis as an acid
or base.
Compared with the role of H333, that of H298
has been an enigma, especially in decarboxylation.
P. von Wettstein-Knowles et al. Histidines and lysine in KAS I ⁄ KAS II catalysis
FEBS Journal 273 (2006) 695–710 ª 2006 The Authors Journal compilation ª 2006 FEBS 705
Recently, bioinformatics and biochemical studies have
substantiated the notion that the two active site histi-
dines in CHH condensing enzymes have chemically dif-
ferent roles in catalysis [8,34]. H298Q mimics the
nonprotonated state of H298 by having the ability to
both donate and accept a hydrogen bond, but does not
mimic its ability to act as a Brønsted base at physiologi-
cal pH values. H298E mimics the nonprotonated state
of H298 by having the ability to act as a base. That
H298E appears to be totally lacking in decarboxylase
activity below pH 6 as does the H298Q enzyme at all
investigated pHs, whereas the wild-type enzyme retains
a vestige of activity at pH 4 demonstrates that decarb-
oxylase activity depends on the presence of a base at

position 298. This observation conflicts with the reac-
tion mechanism proposed by Davies et al. [5] based on
the structure of the CHN enzyme E. coli KAS III and
the interpretation of the structure of the E. coli KAS I–
thiolactomycin complex [12], assuming a 2 A
˚
movement
of H298 on malonyl-ACP binding. The structures of
H298 mutants suggest that the structural changes adop-
ted on acyl transfer, permitting proper binding of
malonyl-ACP and decarboxylation, are restricted to
blockage of the C163 ⁄ F392 oxyanion hole by the thio-
ester [10] and a change in the hydrogen bond potential
of H333.
We have previously hypothesized that K328 is essen-
tial for the catalytic properties of H298 [3]. That the
K328A mutant appeared unable to decarboxylate [9]
supports this hypothesis, but the present observations
that K328E and K328I are similarly deficient whereas
the two basic mutants, K328H and K328R, and the
bulky mutant, K328F, showed some activity reveals
that the positive charge of the lysine is not absolutely
required for decarboxylation. That the orientation of
the conserved K328 is completely unaffected by the
H298 mutations and the binding of acyl thioesters
implies that the mutual interaction between H298 and
K328 is not of structural character. Residues corres-
ponding to K328 are absent from the CHN group of
enzymes, but in CHH enzymes K328 is conserved, as
are the cation site and E342, which are linked to

H298. Possibly the positive charge of K328 is of signi-
ficance for optimal decarboxylation efficiency through
the interactions with conserved E342 and the cation
site, but the present combined structural and kinetic
study only identifies an effect of a positive charge at
the K328 position on the transfer reaction.
An interesting aspect of the H298E mutation is the
orientation of the T300 ⁄ T302 side chains. Both have
been implicated in the binding of ACP [2], probably
by donating hydrogen bonds to one of the carbonyl
oxygens in malonyl-ACP, thereby stabilizing charge
relocalization during catalysis [3]. One reason for the
low decarboxylation activity of H298E could well
be the reorientation of T300 in both the native and the
acyl-bound enzyme, resulting in a lower affinity for the
substrate and inadequate stabilization of charge relo-
calization for decarboxylation.
Experimental procedures
Cloning
The pQE30-fabB plasmid used for mutagenesis and expres-
sion of E. coli KAS I has been described. The mutagenesis
procedure for making the K328 mutant detailed previously
[9] was used to construct seven additional mutants using
the following primers plus their complements:
H298E, CGATTACCTGAACTCCGAGGGTACTTCGA
CTCCG
H298Q, CGATTACCTGAACTCCCAGGGTACTTCGAG
TCCG
K328H, (5¢-GGCGATTTCTGCAACCCACGCCATGAC
CGGTCAC-3¢)

K328R, (5¢-GGCGATTTCTGCAACCCGTGCCATGAC
CGGTCAC-3¢)
K328E, (5¢-GGCGATTTCTGCAACCGAAGCCATGA
CCGGTCAC-3¢)
K328I, (5¢-GGCGATTTCTGCAACCATTGCCATGAC
CGGTCAC-3¢)
K328F, (5¢-GGCGATTTCTGCAACCTTCGCCATGAC
CGGTCAC-3¢)
Confirmation that the desired mutants had been obtained
was achieved by complete sequencing of the initial trans-
formants in XL1-Blue cells.
Expression, purification and activity assay
Wild-type and mutant His-tagged KAS I proteins were puri-
fied as detailed previously [9] except that only the lysozyme
and sonication step was used to open the cells. The resulting
protein had an added N-terminal sequence consisting of
M-R-G-S, a His tag of six residues followed by G-S. Three
assays were carried out on the purified enzymes. (a) The acyl
transferase assay measures the ability of a KAS protein to
accept a fatty acid substrate from the ACP. To 48 lL con-
taining 50 mm potassium phosphate buffer (pH 6.8),
0.5 mgÆmL
)1
thyroglobulin and 0.2 lm [
3
H]myristoyl-ACP
(6000 cpm) is added 12 lL KAS protein (0.3–24 lg) in
storage buffer [9]. Incubations last 10–120 min at 20–22 °C.
Under these conditions, percentage transfer per min is linear
from 1 to 6 lg per assay. The maximum percentage transfer

is 42, and increasing the concentration beyond 6 lg wild-type
KAS I per assay does not increase the percentage transfer. In
the present experiments, a molecular ratio of 10 : 1 KAS to
acyl-ACP was used. Previous work has shown that transfer is
Histidines and lysine in KAS I ⁄ KAS II catalysis P. von Wettstein-Knowles et al.
706 FEBS Journal 273 (2006) 695–710 ª 2006 The Authors Journal compilation ª 2006 FEBS
not an equilibrium reaction, i.e. the bound acyl chain is not
released from the enzyme. One of the factors contributing to
this phenomenon is inhibition by the free ACP released [9].
The reaction mixture was resolved into [
3
H]myristoyl-ACP
and [
3
H]KAS by size-exclusion chromatography, and the
fractions subjected to liquid scintillation using either a Beck-
man L#1701 or an LKB-Wallac 1219 Rackbeta instrument.
Sensitivity of the transfer reaction to ACP was tested by add-
ing 0.17 lm (10 pmol per assay) ACP before addition of
KAS as described previously [9]. Assays were carried out in
triplicate, and the mean values in the presence and absence of
ACP compared. (b) The decarboxylase assay measures the
ability of KAS protein to decarboxylate the donor substrate
malonyl-ACP, forming the acceptor primer acetyl-ACP.
[
14
C]Malonyl-ACP was generated from 12 lm ACP and
10.1 lm [
14
C]malonyl-CoA (10 000 dpmÆnmol

)1
) using malo-
nyl-CoA–ACP transacylase [9] in 25 mm potassium phos-
phate buffer (pH 6, 6.8 or 8) or in 50 mm sodium citrate
buffer (pH 3, 4 or 5). To this reaction mixture was added
0.1–11 lm monomer of the wild-type or mutant protein in
the buffer used to generate the malonyl-ACP substrate.
After 1–60 min at 25 °C, the radiolabeled product and
substrate were precipitated and resolved using 1 m urea
13.3% polyacrylamide before electroblotting to poly(viny-
lidene difluoride) membrane and autoradiography. Addi-
tional details are given in [9]. (c) The elongation assay
measures the ability of a purified KAS I protein preparation
to restore elongation activity to a crude protein extract from
the E. coli CY244 strain that is defective in both the fabB
and fabF genes [35]. The temperature sensitive fabB15
allele determines an A329V mutation in KAS I (GenBank
AJ012163), and the fabF1 allele encodes two mutations,
S220N and G262M, that inactivate KAS II [21]. Soluble pro-
tein extracts of CY244 were prepared as described [36], and
protein concentrations determined using the BCA reagent as
recommended by Pierce. The 100 lL reaction mixture pre-
pared on ice contained 10 lm ACP, 1 mm NADH, 1 mm
NADPH, 10 lm acetyl-CoA, 60.8 lm [2-
14
C]malonyl-CoA
(1670 dpmÆnmol
)1
), 50 mm potassium phosphate (pH 6.8)
buffer, 40 lg CY244 soluble protein and 0.015–1.11 lg puri-

fied KAS protein. After 30 min at 42 °C, the acyl-ACPs were
precipitated and analyzed as described above except that 4 m
urea ⁄ polyacrylamide gels were used. [1-
14
C]Palmitoyl-ACP
(1250 cpm) was included on the gels as standard.
Expression, purification and crystallization
After induction the cells were opened by five freeze ⁄ thaw
cycles. The insoluble fraction was sedimented by centrifuga-
tion at 4000 g for 90 min, and the supernatant loaded on to
a 1-mL HiTrap chelating column (Amersham-Pharmacia) in
buffer A containing 30 mm sodium phosphate, 500 mm
NaCl, and 50 mm imidazole (pH 7.4). The protein was elut-
ed in a linear gradient of 0–100% buffer B (as buffer A plus
500 mm imidazole) in 10 column volumes. The protein was
eluted as a single peak at 25–30% buffer B. The peak frac-
tion was desalted into buffer C containing 25 mm Trizma,
2mm EDTA and 1 mm dithiothreitol, pH 8.0, using a Hi-
Prep 26 ⁄ 10 Sephadex G-25 column. After this, the protein
was loaded on to an anion-exchange column, Mono Q
10 ⁄ 10, in buffer C and eluted in a linear gradient of 0–100%
buffer D (as buffer C with 1 m NaCl). The protein was elut-
ed as two peaks at 12% and 16% buffer D, of which the
earliest (low salt) peak was two to three times as large as the
other. The low salt peak was desalted into buffer C and
concentrated to  14.0 mg proteinÆmL
)1
, as calculated from
the theoretical extinction coefficient (e
280

¼ 24180 m
)1
Æcm
)1
and a molecular mass of 44 011.8 Da. Although the extinc-
tion coefficient and molecular mass refer to the His-tagged
KAS I construct, the same values have been applied to all
samples. An Amicon Centriprep YM-3 device was used to
concentrate the solution. The protein was crystallized in
hanging drop experiments at 294 K with reservoirs contain-
ing 2% (w ⁄ v) poly(ethylene glycol) 400, 100 mm bis-Tris
propane (pH 6.5) and 1.9 m (NH
4
)
2
SO
4
(resulting pH 7.4).
Crystals of the H298E, H298Q and K328A dodecanoyl com-
plexes were formed by adding 1 lL 65.8 mm dodecanoyl-
CoA (Sigma) to crystals in the mother liquor, and incubating
for 24 h. The WT–C8 complex was formed by the same pro-
cedure using 100 mm octanoyl-CoA (Sigma). The mutants
and their fatty acid complexes crystallized in the same space
group and with similar cell dimensions as the wild-type at
cryo-temperatures. In contrast with the other enzymes,
K328A is unstable and starts precipitating in an amorphous
form after three to four days at 277 K.
Data collection
Diffraction data from the H298E, H298Q and K328R

mutants was collected at the MAXLAB, Lund, Sweden at
beam line I711. A D/ ¼ 0.2 ° was used. The data collection
strategy option in Mosflm [37] was used to determine the
optimal / angle span. K328A crystal diffraction data were
collected at the ESRF in Grenoble, France at the ID14.4
beam line (k ¼ 0.97 A
˚
) on a single crystal. The crystal to
detector distance was 150 mm (D/ ¼ 0.25 °). With these
settings spot overlap was not a problem. The K328A–C12
complex data were recorded from a single crystal at the
ESRF ID29 beam line (k ¼ 0.91 A
˚
). The crystal was trans-
lated vertically three times during data collection to avoid
the radiation damage that was readily observable in the dif-
fraction pattern. The data were collected using D/ ¼ 0.50 °
for the first and last run and 0.3 ° for the second run.
Wild-type and WT–C8 complex data were collected on sin-
gle crystals on an in-house Rigaku RU300 generator with
rotating Cobber anode, osmic mirrors, and an R-AXIS
IV++ image plate system. These data were collected with
a crystal to detector distance of 150 mm and a D/ ¼ 0.2 °.
Before data collection, the crystals were cooled to 100 K in
a stream of gaseous N
2
after a 1 min incubation in a
P. von Wettstein-Knowles et al. Histidines and lysine in KAS I ⁄ KAS II catalysis
FEBS Journal 273 (2006) 695–710 ª 2006 The Authors Journal compilation ª 2006 FEBS 707
cryo-protectant containing 28% (v ⁄ v) glycerol, 2% (w⁄ v)

poly(ethylene glycol) 400, 100 mm bis-Tris propane,
pH 6.5, and 1.9 m (NH
4
)
2
SO
4
. All crystals were annealed,
after which the mosaicity decreased from  0.8 ° to 0.2 °.
The mosflm and scala programs [38] were used for data
integration and scaling. Statistics are provided in Table 1.
Structure solution
The temperature of 100 K used for these data collections
as opposed to room temperature as used for previous
data collections (pdb accession number 1EK4) resulted in
cell parameter changes for the a, b and c axes of 1.0,
4.3, and 3.3 A
˚
, respectively, for the wild-type structure. It
was not possible to obtain the correct molecular orienta-
tions directly by rigid body refinement. A molecular
replacement was carried out for the H298E and K328A
mutant proteins with the program AMoRe [39] by using
the KAS I C163S dimer as search model [11]. Water mol-
ecules, ions and substrate were excluded from the search
model.
Initial phases for electron-density calculations were
derived from the C163S model, again excluding water and
ions, except that phases for electron-density calculations for
the K328A–C12 complex and the K328R mutant were

derived from the KAS I K328A model, again excluding
water molecules and ions. Noncrystallographic symmetry
matrices were determined after rigid body refinement in cns
version 1.1 [26], and the relevant amino acid substitutions
made according to the resulting 2|F
o
|–|F
c
| electron density.
Water molecules were added to the models using the auto-
mated procedure in cns. The peak search was performed in
a2|F
o
|–|F
c
| map using a 1.2 r peak height cut-off and dis-
tance constraints based on hydrogen bonding distance
potential. Water molecules were then inspected individually
by eye in the graphics program O [40]. Additional refine-
ment was performed in cns using maximum likelihood
refinement on amplitudes with simulated annealing and
restrained individual B factor refinement. A noncrystallo-
graphic symmetry restraint value of 50 kcalÆmol
)1
ÆA
˚
)2
gave
the lowest value for R
free

for the initial refinements. For all
but the K328A structures, the noncrystallographic sym-
metry restraints were omitted in the last refinements
because of a slight fall in R
free
by this procedure. One
ammonium ion and a dodecanoyl residue per subunit were
added to the H298E–C12 and H298Q–C12 models. In the
wild-type and H298E and H298Q mutant protein struc-
tures, an ammonium ion was added to each of the four
subunits. One ammonium ion and an octanoyl residue were
added to each subunit in the WT–C8 model. Two ammo-
nium ions per subunit were included in the K328A model
and one ammonium ion and a dodecanoyl residue per sub-
unit were added to the K328A–C12 model. In the K328R
mutant protein structure, an ammonium ion was added to
each of the four subunits. Corrections of the models by
visual inspection interspaced by additional rounds of refine-
ment gave the statistics shown in Table 1.
Acknowledgements
This work was supported by grants from the Novo
Nordisk Foundation and the Danish Natural Science
Research Council through DANSYNC and grant
21-01-0622. Dr Andy Thompson (ID29, ESRF), Dr
Sean McSweeney (ID14.4 also ESRF) and Dr Yngve
Cerenius (I711, MAXLAB) are acknowledged for their
assistance with data collection. We thank Annette
Kure Andreassen, Marianne Mortensen and Augusta
Palsdottir for outstanding technical assistance.
References

1 Huang W, Jia J, Edwards P, Dehesh K, Schneider G &
Lindqvist Y (1998) Crystal structure of b-ketoacyl-acyl
carrier protein synthase II from E. coli reveals the
molecular architecture of condensing enzymes. EMBO J
17, 1183–1191.
2 Moche M, Dehesh K, Edwards P & Lindqvist Y (2001)
The crystal structure of b-ketoacyl-acyl carrier protein
synthase II from Synechocystis sp. at 1.54 A
˚
resolution
and its relationship to other condensing enzymes. J Mol
Biol 305, 491–503.
3 Olsen JG, Kadziola A, von Wettstein-Knowles P, Sigg-
aard-Andersen M & Larsen S (2001) Structures of
b-ketoacyl-acyl carrier protein synthase I complexed
with fatty acids elucidate its catalytic machinery. Struc-
ture 9, 233–243.
4 Qiu X, Janson CA, Konstantinidis AK, Nwagwu S,
Silverman C, Smith WW, Khandekar S, Lonsdale J &
Abdel-Meguid SS (1999) Crystal structure of b-ketoa-
cyl-acyl carrier protein synthase III. A key condensing
enzyme in bacterial fatty acid biosynthesis. J Biol Chem
274, 36465–36471.
5 Davies C, Heath RJ, White SW & Rock CO (2000) The
1.8 A
˚
crystal structure and active site architecture of
b-ketoacyl-acyl carrier protein synthase III (FabH) from
Escherichia coli. Structure 8, 185–195.
6 Scarsdale JN, Kazanina G, He X, Reynolds KA &

Wright HT (2001) Crystal structure of the Mycobacter-
ium tuberculosis b-ketoacyl-acyl carrier protein synthase
III. J Biol Chem 276, 20516–20522.
7 Siggaard-Andersen M, Bangera G, Olsen JG & von
Wettstein-Knowles P (1998) Defining the functions of
highly conserved residues in b-ketoacyl-ACP synthases.
In Advances in Plant Lipid Research (Sa
`
nchez, J, Cerda
`
Olmedo, E & Martı
`
nez-Force, E, eds), pp. 67–70. Univer-
sidad de Sevilla, Secretariado de Publicaciones, Sevilla.
8 Witkowski A, Joshi AK & Smith S (2002) Mechanism
of the b-ketoacyl synthase reaction catalyzed by the
Histidines and lysine in KAS I ⁄ KAS II catalysis P. von Wettstein-Knowles et al.
708 FEBS Journal 273 (2006) 695–710 ª 2006 The Authors Journal compilation ª 2006 FEBS
animal fatty acid synthase. Biochemistry 41, 10877–
10887.
9 McGuire KA, Siggaard-Andersen M, Bangera MG,
Olsen JG & von Wettstein-Knowles P (2001) b-Keto-
acyl-[acyl carrier protein] synthase I of Escherichia coli:
aspects of the condensation mechanism revealed by ana-
lyses of mutations in the active site pocket. Biochemistry
40, 9836–9845.
10 Witkowski A, Joshi AK, Lindqvist Y & Smith S (1999)
Conversion of a b-ketoacyl synthase to a malonyl
decarboxylase by replacement of the active-site cysteine
with a glutamine. Biochemistry 38, 11643–11650.

11 Olsen JG, Kadziola A, von Wettstein-Knowles P, Sigg-
aard-Andersen M, Lindquist Y & Larsen S (1999) The
X-ray crystal structure of b-ketoacyl [acyl carrier pro-
tein] synthase I. FEBS Lett 460, 46–52.
12 Price AC, Choi K-H, Heath RJ, Li Z, White SW &
Rock CO (2001) Inhibition of b-ketoacyl-acyl carrier
protein synthases by thiolactomycin and cerulenin.
Structure and mechanism. J Biol Chem 276 , 6551–6559.
13 Shen B (2003) Polyketide biosynthesis beyond the type
I, II and III polyketide synthase paradigms. Curr Opin
Chem Biol 7, 285–295.
14 Kunst L & Samuels AL (2003) Biosynthesis and secre-
tion of plant cuticular wax. Prog Lipid Res 42, 51–80.
15 Mdluli K, Slayden RA, Zhu Y, Ramaswamy S, Pan X,
Mead D, Crane DD, Musser JM & Barry CE III (1998)
Inhibition of a Mycobacterium tuberculosis b-ketoacyl
ACP synthase by isoniazid. Science 280, 1607–1610.
16 Heath RJ, White SW & Rock CO (2001) Lipid bio-
synthesis as a target for antibacterial agents. Prog Lipid
Res 40, 467–497.
17 Jones AL, Herbert D, Rutter AJ, Dancer JE & Har-
wood JL (2000) Novel inhibitors of the condensing
enzymes of the type II fatty acid synthase of pea (Pisum
sativum). Biochem J 347, 205–209.
18 Schaeffer ML, Agnihotri G, Volker C, Kallender H,
Brennan PJ & Lonsdale JT (2001) Purification and bio-
chemical characterization of the Mycobacterium tubercu-
losis b-ketoacyl-acyl carrier protein synthases KasA and
KasB. J Biol Chem 276, 47029–47037.
19 Rendina AR & Cheng D (2005) Characterization of the

inactivation of rat fatty acid synthase by C75: inhibition
of partial reactions and protection by substrates.
Biochem J 388, 895–903.
20 Alli PM, Pinn ML, Jaffee EM, McFadden JM &
Kuhajda FP (2005) Fatty acid synthase inhibitors are
chemopreventive for mammary cancer in neu-N trans-
genic mice. Oncogene 24, 39–46.
21 Magnuson K, Carey MR & Cronan JE Jr (1995) The
putative FabJ gene of Escherichia coli fatty acid synth-
esis is the FabF gene. J Bacteriol 177, 3593–3595.
22 Siggaard-Andersen M, Wissenbach M, Chuck J-A,
Svendsen I, Olsen JG & von Wettstein-Knowles P
(1994) The fabJ-encoded b-ketoacyl-[acyl carrier
protein] synthase IV from Escherichia coli is sensitive to
cerulenin and specific for short-chain substrates. Proc
Natl Acad Sci USA 91, 11027–11031.
23 Cole ST, Brosch R, Parkhill J, Garnier T, Churcher C,
Harris D, Gordon SV, Eiglmeier K, Gas S, Barry CE
III, et al. (1998) Deciphering the biology of Mycobacte-
rium tuberculosis from the complete genome sequence.
Nature 393, 537–544.
24 Cole ST, Eiglmeier K, Parkhill J, James KD, Thomson
NR, Wheeler PR, Honore
´
N, Garnier T, Churcher C,
Harris D, et al. (2001) Massive gene decay in the lep-
rosy bacillus. Nature 409, 1007–1011.
25 McDonald IK & Thornton JM (1994) Satisfying hydro-
gen bonding potential in proteins. J Mol Biol 238,
777–793.

26 Bru
¨
nger AT, Adams PD, Clore GM, DeLano WL, Gros
P, Grosse-Kunstleve RW, Jiang J-S, Kuszewski J, Nilges
M, Pannu NS, et al. (1998) Crystallography & NMR sys-
tem: a new software suite for macromolecular structure
determination. Acta Crystallogr D 54, 905–921.
27 von Wettstein-Knowles P, Olsen JG, McGuire KA &
Larsen S (2000) Molecular aspects of b-ketoacyl
synthase (KAS) catalysis. Biochem Soc Trans 28,
601–607.
28 Yalpani M, Willecke K & Lynen F (1969) Triacetic acid
lactone, a derailment product of fatty acid biosynthesis.
Eur J Biochem 8, 495–502.
29 Eckermann S, Schro
¨
der G, Schmidt J, Strack D, Edrada
RA, Helariutta Y, Elomaa P, Kotilainen M, Kilpela
¨
inen
I, Proksch P, Teeri TH & Schro
¨
der J (1998) New path-
way to polyketides in plants. Nature 396, 387–390.
30 Price AC, Rock CO & White SW (2003) The 1.3-Ang-
strom-resolution crystal structure of b-ketoacyl-acyl car-
rier protein synthase II from Streptococcus pneumoniae.
J Bacteriol 185, 4136–4143.
31 Olsen JG, Rasmussen AV, von Wettstein-Knowles P &
Henriksen A (2004) Structure of the mitochondrial

b-ketoacyl-[acyl carrier protein] synthase from Arabidop-
sis and its role in fatty acid synthesis. FEBS Lett 577,
170–174.
32 Zhang Y-M, Rao MS, Heath RJ, Price AC, Olson AJ,
Rock CO & White SW (2001) Identification and analy-
sis of the acyl carrier protein (ACP) docking site on
b-ketoacyl-ACP synthase III. J Biol Chem 276, 8231–
8238.
33 White SW, Zheng J, Zhang Y-M & Rock CO (2005)
The structural biology of type II fatty acid biosynthesis.
Annu Rev Biochem 74, 791–831.
34 Dawe JH, Porter CT, Thornton JM & Tabor AB (2003)
A template search reveals mechanistic similarities and
differences in b-ketoacyl synthases (KAS) and related
enzymes. Proteins 52, 427–435.
35 Ulrich AK, de Mendoza D, Garwin JL & Cronan JE Jr
(1983) Genetic and biochemical analyses of Escherichia
coli mutants altered in the temperature-dependent
P. von Wettstein-Knowles et al. Histidines and lysine in KAS I ⁄ KAS II catalysis
FEBS Journal 273 (2006) 695–710 ª 2006 The Authors Journal compilation ª 2006 FEBS 709
regulation of membrane lipid composition. J Bacteriol
154, 221–230.
36 Yasuno R, von Wettstein-Knowles P & Wada H (2004)
Identification and molecular characterization of the
b-ketoacyl-[acyl carrier protein] synthase component of
the Arabidopsis mitochondrial fatty acid synthase. J Biol
Chem 279, 8242–8251.
37 Leslie A (1999) MOSFLM. MRC Laboratory of
Molecular Biology, Cambridge, UK.
38 Collaborative Computing Project, number 4 (1994) The

CCP4 suite: programs for protein crystallography. Acta
Crystallogr D 50, 760–763.
39 Navaza J (2001) Implementation of molecular replace-
ment in AMoRe. Acta Crystallogr D 57, 1367–1372.
40 Jones A, Zou JY, Cowan SW & Kjeldgaard M (1991)
Improved methods for building protein models in elec-
tron density maps and the location of errors in these
maps. Acta Crystallogr A 47, 110–119.
41 Kraulis PJ (1991) MOLSCRIPT: a program to produce
both detailed and schematic plots of protein structures.
J Appl Crystallogr 24, 946–950.
42 DeLano WL (2002) The PyMOL molecular graphics
system. .
Histidines and lysine in KAS I ⁄ KAS II catalysis P. von Wettstein-Knowles et al.
710 FEBS Journal 273 (2006) 695–710 ª 2006 The Authors Journal compilation ª 2006 FEBS