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

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 reacti

Role of active site histidines and lysine in Cys-His-His-type

b-ketoacyl-acyl carrier protein synthases

Penny von Wettstein-Knowles1, Johan G Olsen2, Kirsten A McGuire1and Anette Henriksen2

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: knowles@biobase.dk

A Henriksen, Carlsberg Laboratory,

Biostructure Group, Gamle Carlsberg Vej 10,

DK-2500 Valby, Denmark

Fax: +45 33274708

Tel: +45 33275222

E-mail: anette@crc.dk

(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, bactemitochond-rial 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 Ne, 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.

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 pKabecause 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 pKavalue 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 Ne 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.

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 Neand C163 Scis 3.2 A˚ and 3.1 A˚ in the native and fatty acid complex, respectively, and infers that H333 Nedonates a hydro-gen bond to the nucleophile, although the Cys163

Cb–Cys163 Sc–His333 Neangle 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 Ne(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 Oe atoms (Fig 3D,E) One side-chain oxygen is hydrogen bonded to F390 N (3.0 A˚), and the other to the Ocatom

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 pKa 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⁄ Rfree¼ 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].

The active site residue 298Q is involved in hydrogen

(Fig 3F) In this case the side-chain oxygen is

hydro-gen bonded to F390 N (2.9 A˚), and the Ne to the Oc

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)

(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].

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 Neand F390 N in three of the four sub-units of H298Q–C12 (Fig 3G) It is not possible to

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 Nf atom (Fig 4A) The hydrogen bonds and distances imply that the properties of the solvent molecule are similar to the properties of the Nf 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 Nd 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 Nd is

an ammonium ion Moreover, the hydrogen bond dis-tance between H333 Ne and C163 Sc (3.2 A˚) in the mutant is very similar to the 3.3 A˚ found in the wild-type and infers that H333 Nedonates a hydrogen bond

to the nucleophile, although the Cys163 Cb–Cys163

Sc–His333 Neangle (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 (NH4)2SO4

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 Nd with a solvent

Rmerge

a (%)

Rfactor

c ⁄R

2 )

maximum coordinate

Rmerge

Rfactor

Rfree

Rfactor

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 P212121

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) Ng1 and Ng2 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 Ne to H333 Nd (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 Ng1 rather than via Ne (Fig 5B) The H333 rotamer falls between the wild-type orientation and the orientation observed in K328A, with Nd being oriented more towards the backbone N of resi-due 335 (on average the H333 Nd–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 Nd to R328 Ng1 is on average 2.9 A˚ with the average H333 Ne–C163 Sc distance being 3.2 A˚ As only the K328R mutant shows vari-ation in the orientvari-ation of residue 328, whereas all

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 Nf 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 Ndhas been modeled as a water molecule Water molecules are represented

by red spheres.

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

[2-14C]malonyl-ACP to [2-14C]acetyl-ACP In the assay,

the substrate is synthesized from radiolabeled

malonyl-CoA and ACP by malonyl-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.

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,

pH 6.8 pH 5 pH 4

1 5 10 30 1 5 10 30 0 1 5 10 30 0

min Mal-ACP Ac-ACP

Mal-ACP Ac-ACP

7.2 3.6 7.2 3.6 5.0 2.5

WT H298E H298Q 0

µM

C

1 2 3 4 5 6 7 8 9 10 11 12 13 14

1 2 3 4 5 6 7 8

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 l M 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 0 8.6 4.3 2.2 9.7 4.9 2.2 8.7 4.3 2.2 0.8

Mal-ACP Ac-ACP

1 2 3 4 5 6 7 8 9 10 11 12

A

µM

WT 30

K328H 30

K328R 30 C163A

10 min 10

B

8.7 4.3 2.2 7.1 3.6 8.7 4.3 2.2 8.7 4.3 2.2 0

Mal-ACP Ac-ACP

13

1 2 3 4 5 6 7 8 9 10 11 12

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.

(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 3H-labeled myristate (C14 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 (C14) transferred from C 14 –ACP to the specified KAS proteins was determined at 10, 60 and ⁄ or 120 min at 22
C and ⁄ or 4
C, using

6 lg protein, at which concentration maximum transfer to 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

10 min 60 min 120 min 10 min

a Sensitive to ACP; b not sensitive to ACP; c sensitivity to ACP not determined.

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 lg in a 2 h

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 NH4+, Na+and K+were constructed, and the NH4+ 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

(NH4)2SO4 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 pKa 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

0.14

0.14 0.015 0.55 0.14 0.55

0.55 1.11

µg KAS

1 2 3 4 5 6 7 8 9 10 11

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 C16-ACP WT, wild-type.