doi:10.1006/jmbi.2001.4643 available online at http://www.idealibrary.com on J. Mol. Biol.

(2001) 309, 171±180

A Study of the Structure-Activity Relationship for

Diazaborine Inhibition of Escherichia coli Enoyl-ACP
Reductase
Colin W. Levy1, Clair Baldock1, Alistair J. Wallace1, Sveta Sedelnikova1
Russell C. Viner2, John M. Clough2, Antoine R. Stuitje3
Antoni R. Slabas4, David W. Rice1 and John B. Rafferty1*
1
Krebs Institute for Enoyl acyl carrier protein (ACP) reductase catalyses the last reductive
Biomolecular Research step of fatty acid biosynthesis, reducing the enoyl group of a growing
Department of Molecular fatty acid chain attached to ACP to its acyl product using NAD(P)H as
Biology and Biotechnology, The the cofactor. This enzyme is the target for the diazaborine class of anti-
University of Shef®eld bacterial agents, the biocide triclosan, and one of the targets for the front-
Shef®eld S10 2TN, UK line anti-tuberculosis drug isoniazid. The structures of complexes of
2 Escherichia coli enoyl-ACP reductase (ENR) from crystals grown in the
Zeneca Agrochemicals, Jealott's
presence of NAD‡ and a family of diazaborine compounds have been
Hill Research Station
determined. Analysis of the structures has revealed that a mobile loop in
Bracknell, Berkshire RG42 6ET
the structure of the binary complex with NAD‡ becomes ordered on
UK
binding diazaborine/NAD‡ but displays a different conformation in the
3
Department of Genetics, Vrije two subunits of the asymmetric unit. The work presented here reveals
Universiteit, De Boelelaan how, for one of the ordered conformations adopted by the mobile loop,
1087, 1081 HV, Amsterdam the mode of diazaborine binding correlates well with the activity pro®les
The Netherlands of the diazaborine family. Additionally, diazaborine binding provides
4 insights into the pocket on the enzyme surface occupied by the growing
Department of Biological fatty acid chain.
Sciences, The University of # 2001 Academic Press
Durham, Durham DH1 3LE
UK
Keywords: diazaborine; inhibitor binding; enoyl reductase; crystal
*Corresponding author structure; antibacterial activity

Introduction be the target for a range of diazaborine

Enzymes that form the biosynthetic apparatus
for fatty acid production, the fatty acid synthase
(FAS), make ideal targets for the design of new
antibacterials because the molecular organisation
of FAS in most bacteria is different from that
found in mammals.1 Enoyl reductase (ENR) from
Mycobacterium tuberculosis is a target for isoniazid,
the frontline chemotherapeutic treatment for
tuberculosis.2 Resistance to tuberculosis (TB) is
becoming an increasingly grave threat with some
resistant strains being insensitive to all TB
treatments.3 ENR from Escherichia coli is known to
compounds4,5 and the biocide triclosan.6,7 Diaza-
borines attack Gram-negative bacteria by inhibiting
the synthesis of lipopolysaccharide,8 a vital com-
ponent in bacterial cell walls. Previously, we have
described the mode of action of diazaborines
against ENR, which involves the formation of a
covalent bond from the boron atom in the inhibitor
to the enzyme-bound cofactor NAD‡ (ref.9) and
how a mutation in E. coli ENR leads to diazaborine
resistance.4,5 Here, we report the analysis of a
family of diazaborines bound to E. coli ENR to elu-
cidate features of the interactions with the enzyme
that modulate binding af®nity.

Abbreviations used: TB, tuberculosis; ENR, enoyl Results and Discussion

reductase; MIC, minimum inhibitory concentration;
ED50, median effective dose; FAS, fatty acid synthase. We have determined the structures of ENR com-
E-mail address of the corresponding author: plexed with NAD‡ and ®ve different diazaborine
j.rafferty@shef®eld.ac.uk compounds (compounds C-G in Figure 1), repre-

0022-2836/01/010171±10 $35.00/0 # 2001 Academic Press

172
Figure 1. A representation of the structures of the
seven diazaborine compounds discussed in this study.
senting three of ®ve different classes of diazabor-
ines (diazaborine classes 1, 2 and 4, Figure 2), to
better than 2.5 AÊ resolution in order to analyse the
structure-activity relationships. The structures of
these ®ve complexes have been combined with
those formed with two other diazaborine com-
pounds (compounds A and B in Figure 1) pre-
viously determined9 and their structures analysed.
The disordered loop (residues 190-207) in the struc-
ture of the binary complex with NAD‡ that
becomes ordered on binding diazaborine/NAD‡
can now be seen to adopt signi®cantly different
conformations in the two subunits of the asym-
metric unit (A and B) in all the diazaborine com-
plexes. In the initial report of the structure of a
diazaborine complex,9 both subunits were built
with identical conformations of the loop. In the B
subunit the loop adopts the same conformation as
that described by Baldock et al.,9 forming limited
interactions with the drug compound bound in the
active site (Figure 3(a)). In the A subunit, however,
which was noted by Baldock et al.9 to possess poor
density for this region, analysis of the diazaborine
complex with compound C provided clear
SAR of E. coli ENR-Diazaborine Complexes
Figure 2. A representation of the generic class of com-
pounds known as diazaborines and the individual nam-
ing convention for each of the different classes
numbered 1 to 5.
evidence for an alternative conformation in which
the loop forms a largely helical structure. In this
helical conformation, the loop is involved in form-
ing an additional edge to the diazaborine binding
site, providing new interactions with the drug
(Figures 3(b) and 4) and decreasing the solvent-
accessibility of the diazaborine from approximately
10 % to 6 %. The reordering of the loop draws the
conserved residues Ala196 and Ala197 closer to the
diazaborine such that their side-chain and main-
chain atoms make extensive van der Waals con-
tacts with the edge of the fused rings of the inhibi-
tor. In addition, the hydroxyl group of Thr194 now
approaches within hydrogen bonding distance of
one of the oxygen atoms of the pyrophosphate
moiety of the cofactor.
Diazaborine compounds are thought to mimic
the enzyme's natural enoyl substrate,10 therefore,
these ®ndings provide a possible explanation for
the strong conservation of the alanine residue at
position 196 of E. coli ENR in the aligned
sequences of ten enoyl reductases (data not
shown). Furthermore, at position 194 the ENR
sequences show a preference for either threonine
or serine, both of which possess an equivalent
hydroxyl group that could form a hydrogen bond
to the pyrophosphate moiety of the nucleotide
cofactor. The high levels of conservation observed
for these two residues, and their location at the
active site implicates them as residues important
for ENR activity. The apparent inability of the loop
region (residues 190-207) in subunit B to adopt the
helical conformation seen in subunit A can be
attributed to steric hindrance from an adjacent
molecule in the crystal lattice. There are no closely
packed molecules in the vicinity of the loop region
SAR of E. coli ENR-Diazaborine Complexes
from subunit A that might effect the adopted con-
formation. However, when the helical loop confor-
mation observed in the A subunit is modelled onto
the B subunit via least-squares ®tting of the Ca
atoms of the rest of the subunit, there is a clear
clash between the side-chain of Leu195 of the loop
and Pro83 and Lys 84 in an adjacent molecule.
This is in contrast to the situation observed for the
binding of the compound triclosan,10 wherein the
loops in both subunits are ordered helices. In the
triclosan complex, the Ca of Leu195 is approxi-
mately 0.5 A Ê closer to the drug molecule and the
nicotinamide ring of the NAD‡ cofactor and hence
Figure 4. A close-up view of the loop organisation in
the A subunit of the E. coli ENR diazaborine compound
A complex structure. The loop is present in the closed,
mainly helical, conformation and was re®ned in this
position for several cycles of TNT re®nement.19 The
resulting j2Fo ÿ Fcj electron density map contoured at
0.5 s (grey/blue) extends continuously over the model
(all-atom representation, coloured by atom type) of the
loop, diazaborine compound and associated NAD‡
cofactor (blue).
173
Figure 3. (a) and (b) Ribbon dia-
gram of the B subunit of E. coli
ENR, highlighting the location of
the mobile loop (yellow) in the
open conformation and of the A
subunit of E. coli ENR, highlighting
the location of the mobile loop in
the closed conformation. Shown in
stick representation are diazaborine
compound B (cyan) along with its
associated NAD‡ cofactor (blue) in
the active site. The protein is dis-
played with a-helices as cylinders
and b-strands as arrows. The
Figure was produced using Turbo
Frodo.21
the steric clash is avoided. Thus, on the basis of the
use of conserved residues, absence of crystal pack-
ing interference, and similarity to the triclosan
complex, we believe that the helical conformation
adopted by the loop in the A subunit of the E. coli
ENR-NAD‡-diazaborine complexes more closely
represents a conformation important in the
enzyme's catalytic cycle than that observed in the
B subunit.
In both the A and B subunits the diazaborines
bind in a closely related manner, adjacent to the
nicotinamide ring of the cofactor, in a pocket
formed by the side-chains of Tyr146, Tyr156,
Met159, Ile200, Phe203, Leu100, Lys163 and the
main-chain peptide between Gly93 and Ala95 (sub-
unit A has the additional interactions with the
side-chain of Leu195 and the main chain and side-
chain of Ala196). A covalent bond is formed
between the 20 OH of the nicotinamide ribose and
the boron atom of the diazaborine compound, as
described by Baldock et al.9 This observation is
repeated for all the diazaborine compounds stu-
died here. The formation of this covalent bond,
resulting in a tetrahedral anionic boron atom, is
characteristic of boron chemistry, and is a conse-
quence of the fact that in accepting a dative bond,
the boron atom achieves an octet of electrons in its
valence shell. A further interaction appears to be a
hydrogen bond between the phenolic hydrogen
atom of Tyr156 and the boron hydroxyl group. We
infer that Tyr156 is acting as the hydrogen bond
donor in this interaction, whilst the boron hydroxyl
group is, in turn, donating an intramolecular
hydrogen bond to one of the sulphonyl oxygen
atoms. This arrangement satis®es the hydrogen
bonding requirements of the buried face of the dia-
zaborine molecule. The sulphonyl group occurs in
all the diazaborines, and we suggest that, in
addition to providing a hydrogen bonding partner
to stabilise the boron adduct, it ful®ls two other
functions. Firstly, its electron-withdrawing nature
stabilises the negative charge localised on the
boron atom, and secondly, its shape introduces a
bend into the molecule at this position, allowing
174 SAR of E. coli ENR-Diazaborine Complexes

the substituent on the sulphonyl group to project
into the cavity bounded by Phe94, Met159, Tyr156
and Ile200.
The bicyclic ring system of the diazaborines
forms a face-to-face interaction with the nicotina-
mide ring, allowing extensive p-p stacking inter-
actions, which is presumably enhanced by
electrostatic attraction between the positive charge
on the nicotinamide ring in the oxidised cofactor,
and the negative charge of the diazaborine mol-
ecule. The effect of the nearby amino group of the
putative catalytic residue Lys163 is to produce a
charge ``sandwich'' in which the ®lling is the nega-
tively charged diazaborine molecule.
The residues of the active site undergo subtle
shifts in order to accommodate the different diaza-
borine compounds (Figure 5). The individual
active-site residues are themselves shifted only
very slightly, with the large differences in diaza-
borine structure being accommodated as the result
of several small movements distributed throughout
the active site.
Structure-activity relationships
Analysis of the diazaborine complexes may pro-
vide some explanations for the structure-activity
relationships of the family of diazaborines whose
activities have been tested in the work of Grassber-
ger et al.11 Such an analysis might be helpful in the
design of better inhibitors. Grassberger screened a
number of diazaborine compounds to ®nd the
minimum inhibitory concentration (MIC) and
median effective dose (ED50) values for E. coli,
Neisseria gonorrhoea, Proteus mirabilis, Enterobacter
aerogenes and Klebsiella pneumoniae. In this discus-
sion a compound has been considered active if its
MIC value in E. coli is less than 50 mg mlÿ1 as
determined by Grassberger et al.11
The vast majority of diazaborines synthesised
and characterised to date are derived from struc-
tures in which a six-membered diazaborine ring is
attached through a sulphonyl group to a side-
chain that is generally hydrophobic in character.
The six-membered diazaborine ring is also fused to
a further ®ve or six-membered ring, with, in the
case of the former, an additional variation arising
from the introduction of ®ve-membered heterocyc-
lic rings containing O, N or S (Figure 2).
Of the diazaborines whose inhibitory properties
have been studied in detail, the thieno-diazaborine
class of compounds are the most potent inhibitors
(types A2 and A3 in Figure 2). Benzo-diazaborines
and furo-diazaborines (types A1 and A4 in
Figure 2) appear to be less active and the limited
numbers of pyrrolo-diazaborines (type A5 in
Figure 2) tested were totally inactive. This simple
analysis is sometimes somewhat undermined by
the lack of directly comparable compounds where
all other moieties in the diazaborine are the same.
It is clear, however, that substitutions in the nature
of the side-chain attached to the sulphonyl group
and around the fused rings are responsible for dra-
matic differences in potency of these molecules on
ENR inhibition. These large differences in potency
may be due, in part, to differences in uptake and
bioavailability of the compounds.
The analysis of the compounds studied has
enabled us to examine the effects of substitution at
three sites around the diazaborine molecule, and to
infer the effects of substitution at three further
sites. This analysis has been restricted to the con-
formation adopted by the A monomer of ENR in
the asymmetric unit in which the loop (residues
190-207) forms an intrinsic part of the active site.
Across all diazaborines, two positions of substi-
tution (at the sulphonyl tail and para to the boron
atom) are common. There are in addition, two or

Figure 5. Stereoview of a structural superposition of ENR complexed with each of the seven diazaborine com-
pounds A-G. The superposition was carried out using LSQKAB22 and was based on the Ca atoms of the core b-
strands and all atoms of the NAD‡ cofactor. For clarity only a single diazaborine compound (B) and its associated
NAD‡ cofactor are shown. Key residues that form the boundary to the active site are shown (Ile92, Gly93, Phe94,
Ala95, Leu100, Tyr146, Tyr156, Met159, Lys163, Pro191, Phe203 & Met206) in all-atom stick representation and
coloured consistently for each of the different diazaborine compounds: A, light green; B, dark green; C, red; D,
orange; E, blue; F, yellow; and G, pink). The Figure was produced using Turbo Frodo.21
SAR of E. coli ENR-Diazaborine Complexes
three further sites of substitution associated with
the fusion of the boron-containing ring with a ®ve-
membered ring (Figure 6, two sites for X ˆ O,S;
three sites for X ˆ N) or four further sites associ-
ated with fusion to a six-membered ring. These
positions around the diazaborine molecule are
designated R1-R7 or R1-R8, where the boron atom
occupies position R1 (Figure 6).
Substitution in diazaborines
Substitution at R2
In the series of compounds examined, three
different side-chain groups have been observed at
this position: an n-propanesulphonyl side-chain, a
benzenesulphonyl side-chain and a para-toluenesul-
phonyl side-chain.
Comparison of the structures and biochemical
data has revealed that diazaborines with n-propa-
nesulphonyl side-chains at position R2 show the
highest antibacterial activity and decreasing the
chain length of the alkyl group to methyl greatly
reduces activity (Table 1A). The n-propanesulpho-
nyl side-chain of diazaborine compound A makes
interactions with the side-chains of Met159 and
Ile200 and the main-chain peptide of both Gly93
and Phe94 (Figure 7(a)). The n-propyl group folds
over the bicyclic ring system and packs against the
175
side-chains of Met159 and Ile200. Modelling stu-
dies suggest that shortening this group may result
in the loss of packing interactions.
Derivatives with benzenesulphonyl side-chains
show considerable activity. The phenyl group at
R2 of the benzo-diazaborine studied here packs
against the main-chain between Gly93 and Ala95
and also against the side-chain of Leu100. Grass-
berger et al.11 observed that substitutions on the
phenyl rings generally lead to lower activity. How-
ever substitutions of H at position R24 for CH3,
NO2 or NH2 are well tolerated (Table 1A) as the
substituted grouping is partially solvent-exposed
and able to occupy a pocket in the active site
formed by residues Ala95, Gly97, Gly199 and the
side-chain of Leu100 (Figure 7(b)).
Substitution at R4
Earlier studies9 of the interactions made between
ENR and diazaborine compounds had failed to
rationalise the deleterious nature of substitutions at
the R4 position. However, comparison of the struc-
tures of diazaborine compounds C and E, and cor-
relation with the biochemical data provides an
explanation for the reduction in antibacterial
activity following substitution of H at position R4
by CH3 or C6H5 (Table 1B). Position R4 is located
Figure 6. A representation of the
numbering convention adopted in
this discussion to describe the
different sites of substitution
around the diazaborine com-
pounds. Sites common to all diaza-
borine classes are labelled R2 and
R4, additional sites associated with
the fused ®ve or six-membered
ring are labelled R5, R6, R7 and
R8. Five-membered heterocyclic
rings may be accommodated in the
structure in one of two confor-
mations. The hetero atom can
occupy a position remote to the
boron (3,2-d conformation, shown)
and an opposite conformation
where the hetero atom lies on the
same face of the diazaborine as the
boron atom (2,3-d conformation,
not shown).
Table 1. Comparison of substituted diazaborines

176
SAR of E. coli ENR-Diazaborine Complexes
(a) and (b), Summary of directly comparable diazaborine compounds substituted with a variety of groupings at the R2 and R4 positions, respectively, and their associated MIC (mg mlÿ1)
(E. coli). MIC values were determined by serial broth dilutions in trypticase soy broth after incubation for 16 hours at 37  C. The MIC is de®ned as the lowest concentration that inhibited visible
growth. (c)-(e), Summary of directly comparable benzo-diazaborine compounds substituted at the R5, R6 and R7 positions and their associated MIC (mg mlÿ1) (E. coli). (f), Summary of the pyr-
rolo-diazaborine compounds studied by Grassberger et al. with a variety of groupings at the R5 position and their associated MIC (mg mlÿ1) (E. coli). (g), Summary of directly comparable thieno-
and furo-diazaborine compounds substituted with a variety of groupings at the R6 position and their associated MIC (mg mlÿ1) (E. coli). (h), Example of a thieno-
diazaborine compound substituted at position R7 and its associated MIC (mg mlÿ1) (E. coli). (A11 data reviewed by Grassberger et al.).11
SAR of E. coli ENR-Diazaborine Complexes 177

Figure 7. (a) and (b) Stereo view highlighting the important active-site residues (all-atom representation, colour by
atom) involved in making interactions with diazaborine compounds containing R2 alkyl groupings (compound A
shown in cyan), and in (b), with R2 benzenesulphonyl and toluenesulphonyl groupings (compound C shown in
cyan). Also shown are the associated NAD‡ cofactor (stick representation, blue) and the Ca trace of the E. coli ENR
protein (line representation, red). The Figure was produced using Turbo Frodo.21

approximately 3.8 A Ê from Ala196. Substitutions at
this position are likely to form adverse steric inter-
actions with Ala196 and Ala197, and in so doing
disrupt the ordering of the loop region (Figure 8).
This is supported by our analysis of diazaborine
compound E (MIC value 50 mg mlÿ1 unpublished
data) which has a methyl group substituted at this
position and where the electron density for the
loop in the A subunit of this enzyme-diazaborine
complex is poorly de®ned.

Figure 8. Stereoview highlighting the important active-site residues (all-atom representation, colour by atom)
involved in making interactions with the R4 grouping of diazaborine compounds and with the R6, R7 and R8 group-
ings of benzo-diazaborine compounds (compound B shown in cyan). Also shown are the associated NAD‡ cofactor
(stick representation, blue) and the Ca trace of the E. coli ENR protein (line representation, red). The Figure was pro-
duced using Turbo Frodo.21
178
Substitution in benzo-diazaborines
Substitution at R5
Modelling studies suggest that position R5 may
be highly signi®cant in determining the biological
activity of a diazaborine compound. Substitutions
of CH3, Cl or Br at position R5 all have the effect
of reducing antibacterial activity (Table 1C). Substi-
tutions at this position are likely to produce steric
clashes with the side-chain of Phe203, which lies
approximately 3.6 A Ê away.
Substitution at R6
Substitutions at the R6 position of benzo-diaza-
borine compounds must be small and hydrophobic
in nature to bene®t from increased favourable
packing interactions with the side-chain of Phe203
(Figure 8; Table 1D).
Substitutions at R7 and R8
Substitutions at position R7 are not well toler-
ated in the benzo-diazaborine compounds. This
position lies tightly packed against Tyr146 and any
substituted grouping will be likely to produce a
large steric clash between itself and Tyr146, which
is approximately 3.7 A Ê away at its nearest point
(Table 1E).
Benzo-diazaborine compounds have an
additional point of substitution at position R8.
Analysis of the structure suggests that substi-
tutions at this point would have a negative effect
on the antibacterial activity of a compound as a
result of steric clashes with Tyr156.
Substitution in pyrrolo-diazaborines
Substitution at R5
Modelling studies suggest that substitutions at
position R5 are likely to produce steric clashes
with the side-chain of Phe203, which lies approxi-
Figure 9. Stereoview highlighting the important active-site residues (all-atom representation, colour by atom)
involved in making interactions with the R6 and R7 groupings of thieno and furo-diazaborine compounds (com-
pound D shown in cyan). Also shown are the associated NAD‡ cofactor (stick representation, blue) and the Ca trace
of the E. coli ENR protein (line representation, red). The Figure was produced using Turbo Frodo.21
SAR of E. coli ENR-Diazaborine Complexes
mately 3.6 AÊ away, and therefore even a small
group substituted at this point would be likely to
cause a steric clash. The pyrrolo class of diazabor-
ines studied by Grassberger et al.11 all had at least
a CH3 grouping at this position, and this may
explain their lack of activity against E. coli ENR
(Table 1F).
Substitution in furo and thieno-diazaborines
Substitution at R6
Substitutions at R6 can occupy a clear pocket
within the active site and gain van der Waals pack-
ing interactions with the rings of Phe203 and
Pro191, the OH group of Tyr146 and the side-chain
of Met206 (Figure 9). These residues are highly
conserved in the sequence of ENR either in type or
nature across all species.
There is a clear decrease in MIC values for com-
pounds shown in Table 1(G) as the substituted
hydrophobic group at position R6 increases in
length.
Diazaborine D in this study, which has a
methylthio group attached to the R6 position of its
furan ring, has highlighted the advantage of substi-
tution at R6. It has also provided a possible insight
into the substrate binding of enoyl reductase. The
S atom of the methylthio group occupies the pock-
et described above, whilst the CH3 group lies in an
orientation pointing up away from the NAD‡
cofactor and approximately perpendicular to the
plane of the furan ring. The CH3 group gains
favourable packing interactions with the side-chain
of Met206 and the rings of Tyr146 and Tyr156
(Figure 9). The binding of diazaborine is believed
to mimic that of the natural substrate, with the
planar heterocyclic boron-containing rings of the
diazaborine positioned above the nicotinamide
ring of the NAD‡ in a fashion analogous to the
planar enoyl moiety of the substrate. ENR has to
bind a hydrophobic chain of increasing length as
the fatty acid is extended with successive
SAR of E. coli ENR-Diazaborine Complexes
elongation cycles, and modelling of a growing
chain bound to ENR indicates the region occupied
by the R6 position of the diazaborine molecule to
be a likely direction for growth. Thus, the tail
region of this diazaborine may be binding in the
same way as the growing fatty acid chain. Indeed,
a comparison of the recently determined structure
of a complex of InhA, the M. tuberculosis enoyl
ACP reductase, with NAD‡ and an acyl-cystea-
mine substrate analogue12 and the complex of the
E. coli enzyme and diazaborine D shows that the
general direction and location of the acyl tail on
the cysteamine derivative is mimicked by the
methylthio tail of the diazaborine. In the InhA
complex, the growing chain is observed to fold
back upon itself but this may re¯ect the unique
substrate preference of the enzyme for long-chain
acyl compounds and an alternative route can be
seen for the extension of the growing chain in the
E. coli enzyme that would not require it to be
folded back. However, the location of the acyl-
cysteamine substrate relative to the nicotinamide
cofactor in the crystal complex described by Roz-
warski et al.12 requires an explanation for the mech-
anism of hydride transfer in M. tuberculosis enoyl
ACP reductase that is markedly different from that
proposed previously for Brassica napus ENR13 and
InhA itself.14
Substitution at R7
Substitution at position R7 appears to have an
adverse effect on diazaborine activity (Table 1H).
Modelling studies show that any substitution at
this position is likely to form unfavourable steric
interactions with the side-chain of Tyr156 and the
b-carbon atom of Tyr146.
Materials and Methods
Data collection and processing
E. coli ENR is a homo-tetramer with a subunit mass of
approximately 28,000 Da and was prepared from an
over-expressing E. coli strain as described.15 Crystals of
E. coli ENR complexed with NAD‡ and a range of diaza-
borines were grown by the hanging-drop vapour diffu-
sion method from buffered PEG400 solutions as
described.15 Crystals of the ENR-NAD‡-diazaborine
complex belong to spacegroup P6122 with a dimer in the
asymmetric unit.
Data sets were collected at the CLRC Daresbury
Synchrotron and were processed using the MOSFLM
suite of programs16 for compounds A and B, and scaled
and merged using the programs ROTAVATA and
AGROVATA.17 The data for the complexes of com-
pounds C to G were processed using the HKL suite of
programs18 and the resulting data were merged and
scaled using SCALEPACK.
Structure determination
A preliminary report of the structure determination of
complexes of E. coli ENR with NAD‡ and diazaborines
179
A and B has been published.9 The two diazaborine data
sets A and B were non-isomorphous with a mean frac-
tional isomorphous difference of 46 % and therefore the
structures were solved independently by molecular
replacement using an appropriate dimer from the E. coli
ENR-NAD‡ structure. The data from complexes with
diazaborines C to G were suf®ciently isomorphous to
one of the known structures to allow these structures to
be solved by difference Fourier methods. The initial elec-
tron density maps were readily interpretable and unam-
biguous density could be observed for the location of the
diazaborine compounds, which were subsequently incor-
porated into the re®nement. The structures were sub-
mitted to least-squares re®nement using the program
TNT,19 and further cycles of model building and re®ne-
ment were performed. Models comprising 257 of the 261
amino acid residues in each monomer of the E. coli ENR,
could be built with con®dence using the graphics pro-
gram FRODO20 running on an Evans and Sutherland
ESV system.
During the process of re®nement and rebuilding, a
new interpretation for the conformation of the loop (resi-
dues 190-207) in subunit A relative to that reported pre-
viously, was seen for the structure of the complex with
diazaborine C. Subsequent analysis revealed that for
each of the complexes C to G this was the appropriate
interpretation of the electron density and the structures
were re®ned accordingly. Compound E, which possesses
a methyl substitution at R4, produces an electron density
map that is broken and uninterpretable for the loop resi-
dues 196A and 197A. The remaining residues of the
loop, however, appear to be in the same conformation as
those seen for compounds C, D, F and G, and therefore
in this structure the loop residues were inserted and
re®ned but in the absence of residues 196A and 197A.
This observation of a new loop conformation led to a
re-examination of the density for complexes with diaza-
borines A and B. For both of these structures, it had
been noted that the density for the loop in subunit B was
good, whilst that for the loop in subunit A was poor. In
the light of the re®ned structures for compounds C to G,
the density for the loop in the A subunit was better inter-
preted and exhibited the same conformation seen for the
other compounds. Upon re®nement with the new model
for the loop, the density in this region improved signi®-
cantly for diazaborines A and B.
Protein Data Bank accession codes
The co-ordinates have been deposited with the Protein
Data Bank (accession codes: 1DFG, 1DFH and 1DFI).
Acknowledgements
We thank the support staff at the Synchrotron Radi-
ation Source at Daresbury Laboratory for assistance with
station alignment. This work is supported by grants
from the BBSRC to D.W.R. and A.R.S. J.B.R. is a Royal
Society Olga Kennard Fellow, C.W.L. is supported by
BSAC. The Krebs Institute is a designated BBSRC Biomo-
lecular Science Centre and a member of the North of
England Structural Biology Centre (NESBIC).
180
References
1. McCarthy, A. D. & Hardie, D. G. (1984). Fatty acid
synthase - an example of protein evolution by gene
fusion. Trends Biochem. Sci. 9, 60-63.
2. Banerjee, A., Dubnau, E., Quemard, A.,
Balasubramanian, V., Sun Um, K., Wilson, T.,
Collins, D., de Lisle, G. & Jacobs, W. R., Jr (1994).
inhA, a Gene encoding a target for isoniazid and
ethionamide in Mycobacterium tuberculosis. Science,
263, 227-230.
3. Pfyffer, G. E. (2000). Drug-resistant tuberculosis:
resistance mechanisms and rapid susceptibility test-
ing. Schweizerische Medizinische Wochenschrift, 130,
1909-1913.
4. Bergler, H., Wallner, P., Ebeling, A., Leitinger, B.,
Fuchsbichler, S., Aschauer, H., Kollenz, G.,
Hogenauer, G. & Turnowsky, F. (1994). Protein
EnvM is the NADH-dependent enoyl-ACP reductase
(FabI) of Escherichia coli. J. Biol. Chem. 269, 5493-
5496.
5. Kater, M. M., Koningstein, G. M., Nijkamp, H. J. J.
& Stuitje, A. R. (1994). The use of a hybrid genetic
system to study the functional relationship between
prokaryotic and plant multi-enzyme fatty acid
synthetase complexes. Plant Mol. Biol. 25, 771-790.
6. McMurry, L. M., McDermott, P. F. & Levy, S. B.
(1999). Genetic evidence that InhA of Mycobacter-
ium smegmatis is a target for triclosan. Antimicrob.
Agents Chemother. 43, 711-713.
7. Levy, C. W., Roujeinikova, A., Sedelnikova, S.,
Baker, J. B., Stuitje, A. R., Slabas, A. R., Rice, D. W.
& Rafferty, J. B. (1999). Molecular basis of triclosan
activity. Nature, 398, 383-384.
8. Hogenauer, G. & Woisetschlager, M. (1981). A
diazaborine derivative inhibits lipopolysaccharide
biosynthesis. Nature, 293, 662-664.
9. Baldock, C., Rafferty, J. B., Sedelnikova, S. E., Baker,
P. J., Stuitje, A. R., Slabas, A. R., Hawkes, T. R. &
Rice, D. W. (1996). A mechanism of drug action is
revealed by structural studies of enoyl reductase.
Science, 274, 2107-2110.
10. Roujeinikova, A., Levy, C. W., Rowsell, S.,
Sedelnikova, S., Baker, P. J., Minshull, C. A., Mistry,
A., Colls, J. G., Camble, R., Stuitje, A. R., Slabas,
A. R., Rafferty, J. B., Pauptit, R. A., Viner, R. & Rice,
D. W. (1999). Crystallographic analysis of triclosan
bound to enoyl reductase. J. Mol. Biol. 294, 527-535.
(Received 22 June 2000; received in revised form 14 March 2001; accepted 16 March 2001)
SAR of E. coli ENR-Diazaborine Complexes
11. Grassberger, M. A., Turnowsky, F. & Hildebrandt, J.
(1984). Preparation and antibacterial activities of
new 1,2,3-diazaborine derivatives and analogues.
J. Med. Chem. 27, 947-953.
12. Rozwarski, D. A., VilcheÁze, C., Sugantino, M.,
Bittman, R. & Sacchettini, J. C. (1999). Crystal struc-
ture of the Mycobacterium tuberculosis enoyl-ACP
reductase, InhA, in complex with NAD‡ and a C16
fatty acyl substrate. J. Biol. Chem. 274, 15582-15589.
13. Rafferty, J. B., Simon, J. W., Baldock, C., Artymiuk,
P. J., Baker, P. J., Stuitje, A. R., Slabas, A. R. & Rice,
D. W. (1995). Common themes in redox chemistry
emerge from the X-ray structure of oilseed rape
(Brassica napus) enoyl acyl carrier protein reductase.
Structure, 3, 927-938.
14. Dessen, A., QueÂmard, A., Blanchard, J. S., Jacobs,
W. R. & Sacchettini, J. C. (1995). Crystal structure
and function of the isoniazid target of Mycobacterium
tuberculosis. Science, 267, 1638-1641.
15. Baldock, C., Rafferty, J. B., Sedelnikova, S. E.,
Bithell, S., Stuitje, A. R., Slabas, A. R. & Rice, D. W.
(1996). Crystallisation of Escherichia coli enoyl
reductase and its complex with diazaborine. Acta
Crystallog. sect. D, 52, 1181-1184.
16. Leslie, A. G. W. (1992). Recent changes to the
MOSFLM package for processing ®lm and image
plate data. In Joint CCP4 and ESF-EACBM Newsletter
on Protein Crystallography, no. 26, SERC Laboratory,
Daresbury, UK, Warrington WA4 4AD.
17. Collaborative Computational Project No. 4 (1994).
The CCP4 suite: programs for protein crystallogra-
phy. Acta Crystallog. sect. D, 50, 760-763.
18. Otwinowski, Z. & Minor, W. (1997). Processing of
X-ray diffraction data collected in oscillation mode.
Macromol. Crystallog. sect. A, 276, 307-326.
19. Tronrud, D. E., Ten Eyck, L. F. & Matthews, B. W.
(1987). An ef®cient general-purpose least-squares
re®nement program for macromolecular structures.
Acta Crystallog. sect. A, 43, 489-501.
20. Jones, T. A. (1978). A graphics model building and
re®nement system for macromolecules. J. Appl. Crys-
tallog. 11, 268-270.
21. Roussel, A. & Cambillau, C. (1991). Silicon Graphics
Geometry Partners, vol. 86, Silicon Graphics, Moun-
tain View, CA.
22. Kabsch, W. (1976). A solution for the best rotation
to relate two sets of vectors. Acta Crystallog. sect. A,
32, 922-923.
Edited by I. A. Wilson