Allosteric competitive inhibitors of the glucose-1-phosphate thymidylyltransferase (RmlA) from pseudomonas aeruginosa

Article abstract of DOI:10.1021/cb300426u

Glucose-1-phosphate thymidylyltransferase (RmlA) catalyzes the condensation of glucose-1-phosphate (G1P) with deoxy-thymidine triphosphate (dTTP) to yield dTDP-d-glucose and pyrophosphate. This is the first step in the l-rhamnose biosynthetic pathway. l-R

Full text of DOI:10.1021/cb300426u

Articles
pubs.acs.org/acschemicalbiology
Allosteric Competitive Inhibitors of the Glucose-1-phosphate
Thymidylyltransferase (RmlA) from Pseudomonas aeruginosa
Magnus S. Alphey,†,+ Lisa Pirrie,†,‡,+ Leah S. Torrie,§ Wassila Abdelli Boulkeroua,† Mary Gardiner,§
Aurijit Sarkar,§ Marko Maringer,∥ Wulf Oehlmann,⊥ Ruth Brenk,§ Michael S. Scherman,¶
Michael McNeil,¶ Martin Rejzek,# Robert A. Field,# Mahavir Singh,⊥ David Gray,§
Nicholas J. Westwood,*,†,‡ and James H. Naismith*,†
†Biomedical Sciences Research Complex, University of St. Andrews, St. Andrews KY16 9ST, U.K.
‡School of Chemistry, University of St. Andrews and EaStCHEM, St. Andrews KY16 9ST, U.K.
§Biological Chemistry and Drug Discovery, College of Life Sciences, University of Dundee, Dundee DD1 5EH, U.K.
∥mfd Diagnostics GmbH, Mikroforum Ring 5, 55234 Wendelsheim, Germany
⊥Lionex GmbH, Salzdahlumer Str. 196, 38126 Braunschweig, Germany
¶Department of Microbiology, Immunology and Pathology, Colorado State University, 1682 Campus Delivery, Ft. Collins, Colorado
80523-1682, United States
#Department of Biological Chemistry, John Innes Centre, Norwich NR4 7UH, U.K.
S
* Supporting Information
ABSTRACT: Glucose-1-phosphate thymidylyltransferase
(RmlA) catalyzes the condensation of glucose-1-phosphate
(G1P) with deoxy-thymidine triphosphate (dTTP) to yield
dTDP-D-glucose and pyrophosphate. This is the first step in
the L-rhamnose biosynthetic pathway. L-Rhamnose is an
important component of the cell wall of many microorganisms,
including Mycobacterium tuberculosis and Pseudomonas aerugi-
nosa. Here we describe the first nanomolar inhibitors of P.
aeruginosa RmlA. These thymine analogues were identified by
high-throughput screening and subsequently optimized by a
combination of protein crystallography, in silico screening, and
synthetic chemistry. Some of the inhibitors show inhibitory
activity against M. tuberculosis. The inhibitors do not bind at the active site of RmlA but bind at a second site remote from the
active site. Despite this, the compounds act as competitive inhibitors of G1P but with high cooperativity. This novel behavior was
probed by structural analysis, which suggests that the inhibitors work by preventing RmlA from undergoing the conformational
change key to its ordered bi-bi mechanism.
one-third of the world’s population is estimated to be infected,
and the emergence of drug-resistant tuberculosis has been
noted with alarm.6,7 Development of novel antimicrobial
compounds with new chemical scaffolds against novel enzyme
targets is highly desirable.
The bacterial cell wall is a target for many antibiotics since it
is vital for bacterial survival and is composed of peptidoglycans
and lipopolysaccharides not found in eukaryotic cells. The
biosynthesis of sugars in bacteria is in particular an important
area of research (reviewed in ref 8). L-Rhamnose is part of the
O-antigen of many Gram-negative bacterial cell wall lip-
opolysaccharides including P. aeruginosa but is not thought to
be essential. In M. tuberculosis L-rhamnose links the
Pseudomonas aeruginosa is one of the most prevalent and
opportunistic pathogens in hospital-acquired infections. It is a
particularly difficult microorganism to treat due to its intrinsic
chemo-resistance and its ability to acquire further resistance
mechanisms against antimicrobial agents such as β-lactams,
aminoglycosides, and fluoroquinolones.1 The outer membrane
of P. aeruginosa has low drug permeability properties,2 and the
species expresses a variety of efflux pumps.3 The emergence of
multidrug-resistant or even pan-resistant phenotypes of
Pseudomonas means that even currently recommended
combination therapies are less effective, having a negative
impact on patient outcomes.4 Currently, only a small number
of novel anti-Pseudomonal drugs are in preclinical or clinical
development, and very few have reached the market in recent
years.5 Although the World Health Organisation’s recom-
mended DOTS therapy (directly observed therapy, short
course) is highly effective against Mycobacterium tuberculosis,
Received: August 15, 2012
Accepted: November 8, 2012
Published: November 8, 2012
© 2012 American Chemical Society
387
dx.doi.org/10.1021/cb300426u | ACS Chem. Biol. 2013, 8, 387−396

ACS Chemical Biology
Articles
peptidoglycan to the arabinogalactan components, and genes
associated with L-rhamnose biosynthesis have been shown to be
essential in this microorganism.9 dTDP-L-rhamnose is synthe-
sized from glucose-1-phosphate (G1P) and deoxy-thymidine
triphosphate (dTTP) by four enzymes. These enzymes are
potential therapeutic targets, and inhibitors of the pathway
enzyme RmlC are being investigated.10−13 The first enzyme,
glucose-1-phosphate thymidylyltransferase (RmlA), catalyzes
the condensation of G1P with dTTP to give dTDP-D-glucose
(Figure 1A).14,15 RmlA is homologous to other bacterial sugar
nucleotide transferases (e.g., G1P uridylyltransferase16), though
RmlA’s tetrameric arrangement is distinct (Figure 1B). The
tetramer can be thought of as a dimer of dimers. The dimer
“building block” contains what we term the monomer−
monomer interface, and the dimer−dimer interface links
together the two dimers (Figure 1B). The inhibition of RmlA
by dTDP-L-rhamnose, the final pathway product, has been
reported to occur in both a competitive and noncompetitive
manner.17 The presence of this feedback mechanism suggests
that RmlA is the point of control for the pathway. Based on
sequence conservation and interactions with substrates,
previous studies have identified the active site of RmlA as
adjacent to the dimer−dimer interface.15 Specifically, residues
in loops 10−25, 138−148, and 224−232 are involved not only
in forming the active site but also in subunit interactions. A
second (or allosteric) site in RmlA was observed in two earlier
studies15,18 and found to bind thymidine-containing com-
pounds at the monomer−monomer interface. In summary,
each monomer has one active site completely enclosed by the
monomer and one second site. The second site is some 20 Å
distant from the active site and sits at the monomer−monomer
interface, utilizing residues from both monomers.15,18 RmlA has
attracted considerable attention from a protein engineering
standpoint and its use in so-called glycorandomization.18−20
The allosteric modulation of enzyme activity by small
molecules, resulting in either activation or inhibition, has
received considerable attention in recent years.21−25 Allosteric
inhibitors offer several advantages over traditional active site
inhibitors as they generally exhibit a higher selectivity due to
the allosteric site residues having conservation across a protein
superfamily lower than that of the corresponding active site
residues. Lower off target effects result from this and may lead
to a decrease in inhibitor toxicity and side effects.26 The
mechanisms by which allosteric modulators exert their activity
can shed light on enzyme mechanisms. In terms of kinase
inhibition, the development of allosteric inhibitors that show
selective isoform inhibition has proved possible in cases where
selectively targeting the active site is challenging.27 The
identification of novel allosteric inhibitors, however, remains a
considerable challenge.
Figure 1. Enzymatic reaction and quaternary structure of RmlA. (A)
Enzyme substrates dTTP and G1P are combined to produce dTDP-D-
glucose and pyrophosphate. (B) Tetrameric arrangement of RmlA,
with the four interacting subunits shown in different colors. The active
and allosteric site positions are marked; there is one of each site per
monomer. The active sites are located at the dimer−dimer interfaces
but are formed by residues from only one monomer. The allosteric
sites are located at the monomer−monomer interfaces and are formed
by residues from both monomers.
To date no small molecule inhibitors of RmlA have been
reported. Here, we report the discovery and optimization of a
series of Pseudomonas aeruginosa RmlA inhibitors, several of
which have nanomolar activity against the enzyme. Weak
activity against M. tuberculosis has also been demonstrated. The
inhibitors interact with RmlA in a highly cooperative manner
with the inhibitor core structure mimicking the thymine ring of
RmlA’s substrate dTTP. The inhibitors bind only to the
allosteric site on RmlA not at the active site.
compounds28 at a single concentration of 30 μM using an
assay that monitors phosphate levels.29 Forty-three compounds
displayed >30% inhibition (hit rate of 0.27%) and were
designated potential hits. Retesting this set in duplicate as 10-
point dose−response curves enabled IC50 value determination.
Compounds 1 and 2 gave reliable data with IC50 values of 0.22
RESULTS AND DISCUSSION
Identification of Initial Hits. Pseudomonas aeruginosa
RmlA was screened against a diverse library of 15,667
■
388
dx.doi.org/10.1021/cb300426u | ACS Chem. Biol. 2013, 8, 387−396

ACS Chemical Biology
Articles
a
Table 1. Inhibition Data against RmlA for Analogues of 1
b
c
c
d
compound
R1
nBu
R51
R52
SO2Ph
R6
% inhibition at 10 μM
% inhibition at 60 μM
IC50 (μM)
1
Me
H
NH2
OH
100
100
0.21 0.03
1.14 0.19
n.d.
2
benzyl
nBu
nBu
H
100
85.7 2.1
75.7 3.9
39.4 4.6
100
3
Me
H
C(O)Ph
cyclopentyl
C(O)-2-furan
SO2Ph
NH2
NH2
NH2
NH2
NH2
NH2
NH2
NH2
NH2
NH2
NH2
NH2
NH2
NH2
30.3 4.1
30.0 2.0
71.1 1.2
100
4
n.d.
5
nBu
Me
Me
Me
H
5.9 2.9
0.073 0.01
>60
8a
8f
benzyl
benzyl
benzyl
benzyl
benzyl
benzyl
benzyl
benzyl
benzyl
benzyl
benzyl
100
COPh
0
14.5 0.6
100
15a
8b
8c
12a
12b
8j
SO2Ph
94.2 0.8
100
1.32 3.7
0.102 0.02
>10
Et
SO2Ph
100
nPr
Me
Et
SO2Ph
100
43.4 2.3
3.1 3.5
82.4 1.9
100
H
10.7 3.7
97.8 1.5
100
>60
H
4.7 4.0
0.175 0.067
0.205 0.009
0.387 0.065
0.105 0.012
Me
Me
Me
Me
SO2-n-Bu
SO2-4-F-Ph
SO2-3-Me-Ph
SO2-2-furan
8k
8p
8t
100
100
100
100
100
100
a
b
For assay data relating to additional analogues see Table S7 in Supporting Information. The following PDB codes are assigned to structures of the
complexes of RmlA bound to 1 (4ARW), 2 (4AS6); 3 (4B42); 4 (4B2X); 8a (4ASJ); 8f (4ASY); 15a (4ASP); 12b (4B3U); 8j (4B4G); 8k (4B4M);
c
8p (4B5B); for data collection and model refinement statistics for all RmlA-inhibitor complexes see Supporting Information. SE, standard error (n =
d
2). SD, standard deviation (n = 3).
μM (95% CI 0.18−0.27 μM) and 1.23 μM (95% CI 0.92−1.67
μM), respectively (Table 1 and Supplementary Figure S1).
Both 1 and 2 gave large Hill coefficients (3.55 0.25 for 1 and
in a hydrophobic pocket surrounded by Leu45, Val250, Ala251,
and Ile256 (Figure 2A). The N1-n-butyl chain of 1 packs
against the side chain of Arg259 from a neighboring subunit.
The pyrimidinetrione core of 2, including the N3H and the C4
carbonyl, and the N1-benzyl substituent make essentially the
same contacts with RmlA as described for 1 (Figure 2B and
Supplementary Figure S4). In this structure with 2, an
additional glycerol molecule is bound in the equivalent position
to the phenyl group of the C5-N-methylsulfonamide group in 1
(Supplementary Figure S5).
Commercial Analogues. A 2D-similarity screening
approach was undertaken to identify commercially available
analogues of 1 and 2. Structural analysis suggested that the
interactions of the central ring including the N3H and C4
carbonyl group would be crucial for binding to RmlA
(Supplementary Figure S6). Commercially available analogues
containing a pyrimidinedione or pyrimidinetrione ring system
were identified, and 15 analogues were purchased and assayed
(Table 1, Supplementary Tables S2 and S3). None of these
compounds were more potent than 1; however, these studies
showed that replacement of the sulfonamide in 1 by an amide
in 3 or an alkyl substituent in 4 was detrimental (Table 1).
Replacement of the amide phenyl ring by a furan in 5 partially
compensated for the loss in activity. Complexes with 3, 4, and 6
(Supplementary Table S4) showed that the central ring of these
inhibitors remained in essentially the same position as 1, with
only minor angular distortions with respect to each other as a
result of variations in stacking of N1-substituents with Arg259
(Supplementary Figure S7A). However, two of the compounds
appear to react with the buffer at the central carbon of the 1,3-
diketone functional group under the crystallization conditions
(see analogues 6 and 7 in Supplementary Figures S7B and
S7C).
1.75
0.42 for 2, Supplementary Figure S1). Isothermal
calorimetry (ITC) gave a Kd = 0.074 0.006 μM for 1 and Kd
= 0.395 0.02 μM for 2 (Supplementary Figure S2).
Structural Characterization of Initial Hits. Multiple
crystal structures of substrate/product complexes of RmlA from
P. aeruginosa have been reported.15,18 RmlA crystals were
soaked overnight with 1 or 2, and clear density for four bound
inhibitors was identified in the tetramer (Figure 2 and
Supplementary Table S1). The inhibitor binding sites were
not the active sites, rather the allosteric sites.15 The allosteric
site is predominantly hydrophobic in character and sits at an
intersubunit interface formed by residues from two adjoining
subunits of the tetramer: Tyr38, Ser41, Leu45, Tyr114, Gly115,
Phe118, His119, Leu122, Val250, Ala251, Glu255, Ile256,
Arg259 from one monomer and Gly218′, Arg219′ from the
second monomer.
The pyrimidinedione core in 1 stacks against Arg219′ and
forms hydrophobic interactions with Leu45 and Ile256 (Figure
2A and Supplementary Figure S3). The C4 carbonyl hydrogen
bonds directly to Ala251N, and N3H interacts via a single water
molecule with Lys249O, Cys252N, and Glu255OE1. These
interactions directly match those previously observed in
complexes of RmlA with the thymidine-containing compounds
dTTP, dTDP, dTMP, dTDP-D-glucose, or dTDP-L-rhamnose
bound at this allosteric site.15,30 The C6-amino group interacts
with two water molecules, one of which hydrogen bonds to
Gly115O (Supplementary Figure S3). The C5-N-methylsulfo-
namide makes a hydrogen bond with Gly115N and a water-
mediated hydrogen bond with Phe118N and His119N. In
addition, the sulfonamide phenyl ring makes a face-edge
hydrophobic interaction with the side chain of Phe118 and sits
389
dx.doi.org/10.1021/cb300426u | ACS Chem. Biol. 2013, 8, 387−396

ACS Chemical Biology
Articles
N-benzyl urea (9) with 10 led to the formation of the
pyrimidinedione ring in 11.31 Selective C5-bromination of 11
enabled incorporation of the required methylamine function-
ality to give 12a, which was subsequently converted to 8a in
good yield. Analogous methodology was used to prepare 8b−e
via the corresponding amines 12b−e (Scheme 1) and a further
18 C5-N-sulfonamide analogues were prepared from 12a
(Table 1 and Supplementary Table S5 for structures). Reaction
of 12a with benzoyl chloride gave 8f, the amide analogue of 8a,
whereas conversion of 11 to the corresponding nitroso-
derivative 13 enabled the synthesis of analogues 14 and 15a,b.
Analysis of Novel Analogues of 1. Replacement of the
N1-n-butyl substituent in 1 with an N1-benzyl substituent in
8a, a change inspired by the structure of 2, led to a considerable
increase in potency with 8a having an IC50 of 0.073 0.01 μM
against RmlA (Table 1, entry 6) and a KD of 0.028 0.003 μM
(Supplementary Figure S2). Superposition of the structures of
the RmlA-1 with RmlA-8a (Supplementary Figure S8 and
Table S6) showed no change in the interaction made by the
inhibitor core with the enzyme and almost identical positioning
of the C5-N-methylsulfonamide substituent. The significant
difference in potency between 8a and 1 seems most likely to
result from improved protein−ligand hydrophobic interactions
in the N1-binding pocket (Supplementary Figure S8).
As in the N1-n-butyl series (cf. Table 1, entries 1 and 3),
replacement of the sulfonamide substituent in 8a by an amide
proved detrimental (8f, entry 7). An overlay of the structure of
RmlA-8a and RmlA-8f complexes showed that the contacts
made between the enzyme and the pyrimidinedione rings of 8a
and 8f were preserved (Supplementary Figure S9 and Table
S6). However, binding of the amide in 8f to RmlA resulted in a
decrease in the hydrogen bonding network associated with the
amide group compared to that present with the sulfonamide in
8a (Figure 3A−D). In addition, the N-methylamide substituent
in 8f was forced to rotate by 50° compared to its position in the
RmlA-8a complex with a corresponding 1.4 Å movement in the
position of the methyl group. An analogous decrease in
hydrogen bonding and rotation of the N-methyl substituent
was also seen when the structures of the RmlA-1 and RmlA-3
complexes were compared (Supplementary Figure S10).
The N-methyl group in 8a was shown to be important for
inhibition of RmlA (cf. Table 1, entries 6 and 8) with the C5-
NH analogue 15a being 18-fold less potent than 8a. The
difference in potency between 8a and 15a was rationalized on
the basis of the occupancy of a small hydrophobic pocket by the
N-methyl group in 8a. The top of the N-methyl pocket is lined
by hydrophobic groups including Val250 and the methylene
units of the Arg219′ side chain. Analysis of a series of N-alkyl
analogues of 8a showed that while an ethyl substituent was
tolerated, a rapid drop off in potency was observed for longer
alkyl chain lengths (8b−e, Table 1 and Supplementary Table
S5). On replacement of the sulfonamide substituent by a
hydrogen atom, only the C5-N-ethyl analogue 12b inhibited
RmlA (12a−e, Table 1 and Supplementary Table S5).
Structural studies showed that in the RmlA-12b complex the
C5-N-ethyl substituent no longer occupied the previously
identified hydrophobic pocket but instead was positioned in a
nearby cleft (Supplementary Table S7 and Figure S11).
Figure 2. Representative stick images of RmlA crystal complexes with
screening hits 1 and 2. (A) RmlA-1 complex. (B) RmlA-2 complex.
Both show omit (ligand removed, structure rerefined) 2Fo − Fc
electron density maps (contoured at 1.2σ) and selected hydrogen
bonds (black dotted lines) (PDB codes 4ARW and 4AS6). Important
residues are highlighted with the sulfonamide binding region residues
colored green in panel A. 1 is colored yellow, 2 orange, nitrogen atoms
blue, oxygen atoms red, and water molecules cyan. Figures were
prepared using PyMOL.41 Further details of the binding of RmlA to 1
and 2 are shown in Supplementary Figures S3 and S4. Unbiased
original Fo − Fc maps are shown for all ligands in Supplementary
Figure S19.
To probe further the pocket occupied by the sulfonamide
group in 8a, a set of 18 analogues with varying sulfonamide
groups was synthesized. In brief, replacement of the phenyl ring
in the sulfonamide with an alkyl substituent was tolerated with
8j inhibiting RmlA with an IC50 of 0.175 0.067 μM (Table 1,
Synthesis of Analogues of 1. We synthesized inhibitor
8a, the N1-benzyl analogue of 1 (Scheme 1 and Table 1), to
explore the role of the N1 substituent more fully. Reaction of
390
dx.doi.org/10.1021/cb300426u | ACS Chem. Biol. 2013, 8, 387−396

ACS Chemical Biology
Articles
a
Scheme 1. Synthesis of Analogues of 1
a
Reagents and conditions: (a) Ac2O, 80 °C, 2 h then 10% aq NaOH, 85 °C, 1 h, 51%;31 (b) Br2, NaHCO3, MeOH, 0 °C to rt, 30 min then 30% aq
amine, 70 °C, 4 h, 69% for 12a; 30% (12b); 28% (12c); 11% (12d); 8% (12e); (c) for 8a−e phenylsulfonyl chloride, pyridine, DCM, rt, 16 h, 65%
for 8a; 54% (8b); 54% (8c); 60% (8d); 51% (8e); 44−85% for 8g−w; for 8f benzoyl chloride, pyridine, DCM, rt, 16 h, 73%; (d) NaNO2, AcOH/
H2O, 70 °C to rt, 1 h, 72%; (e) Na2S2O4, NH4OH, 6 h, 75%; (d) phenylsulfonyl chloride or benzoyl chloride, pyridine, DCM, rt, 26 h, 71% for 15a;
75% for 15b.
entry 13). Structural analysis of the RmlA-8j complex showed
that hydrophobic interactions between the alkyl side chain and
Phe118 were present (Supplementary Figure S12). Substitution
in the phenylsulfonamide ring of 8a was in general poorly
tolerated with the exception of small groups at the 3- or 4-
positions (Table 1 and Supplementary Table S5; see
Supplementary Figure S13 for structural insights gained from
analysis of the RmlA-8k and RmlA-8p complexes). Replace-
ment of the phenyl ring with a 2-furan ring gave the potent
inhibitor 8t (IC50 of 0.105 0.012 μM, Table 1, entry 16);
however, exchange for a larger aromatic or heteroaromatic ring
or a benzyl substituent was not tolerated (Supplementary Table
S5).
Biological Data. P. aeruginosa PAO1, which had the RmlA
gene knocked out, grew in culture normally but showed
significantly reduced numbers of colonies in mouse lung after
infection compared to wild type P. aeruginosa PAO1
(Supplementary Figure S14). This confirmed RmlA is not
essential for survival but is important for virulence. In simple
assays of bacterial growth (both plate and suspension), no
compound exhibited inhibition of P. aeruginosa PAO1. Selected
compounds (Supplementary Table S8) were tested against M.
tuberculosis, for which RmlA is essential,9 and compounds 8f
and 8p showed MIC100 values of 25 and 50 μg mL−1,
respectively (the value for isoniazid is 0.078 μg mL−1). We have
as yet been unable to express RmlA from M. tuberculosis in
sufficient quantity to assay the compounds in vitro.
Mode of Inhibition. To fully understand the enzyme
reaction mechanism and mode of inhibition of compound 8a,
biochemical assays and surface plasmon resonance (SPR)
methods were utilized. Initial biochemical experiments reveal
the substrate Michaelis constants to be 7.8 μM (95% CI 4.3−
14.4 μM) and 9.5 μM (95% CI 6.8−13.5 μM) for dTTP and
G1P, respectively (Supplementary Figure S15). For both
substrates the Hill coefficient was found to be ∼1, indicating
no evidence of cooperativity within the tetrameric protein.
SPR experiments showed that dTTP binds to apo-RmlA with
a KD of 39.9 μM (95% CI 35.9−44.3 μM),whereas apo-RmlA
does not bind G1P (Supplementary Table S9 and Figure S16).
However, when each substrate was titrated in the presence of a
fixed, saturating concentration of the second substrate, KD
values comparable to the Michaelis constants were measured
(6.70 μM (95% CI 5.55−8.10 μM) for G1P and 6.02 μM (95%
CI 5.03−7.20 μM) for dTTP) (Supplementary Table S9 and
Figure S16). The binding of dTTP is therefore required to
create a functional G1P binding site, consistent with the
previously described17,30 sequential ordered bi-bi mechanism of
catalysis employed by RmlA (Supplementary Figure S16).
Since both substrates are negatively charged, a simple charge−
charge attraction can be ruled out. The most likely explanation
is that dTTP binding triggers a conformational change in the
protein that forms the G1P binding site.
As previously described, inhibitor 8a was tested in the
biochemical assay with an IC50 of 0.073 μM calculated. In
addition to the biochemical studies, SPR experiments confirm
8a binds to RmlA (Supplementary Figure S17). This SPR
binding data could not be fitted to a 1:1 binding model,
indicating 8a binds to the RmlA tetramer with some level of
cooperativity.
To investigate the mode of inhibition of 8a, kinetic assays
were performed. With G1P in excess and varying dTTP
concentrations in the presence of different concentrations of
8a, only weak (essentially no) inhibition of dTTP binding was
observed until inhibitor concentrations were raised to around
400 nM, where almost complete inhibition occurred (Figure
4A). Using a saturating concentration of dTTP and varying the
concentrations of G1P and 8a showed that 8a displayed
competitive inhibition with respect to G1P (Figure 4B)
(estimated Ki = 0.020
0.020 μM). Analysis of these data
shows a Hill slope of 3.6. These findings were confirmed using
SPR, with no binding of G1P observed in the presence of 1 μM
8a and a saturating concentration of dTTP (Figure 5 and
Supplementary Table S9). Conversely binding of dTTP was
observed when titrated in the presence of 1 μM 8a and a
saturating concentration of G1P (albeit with a reduced binding
affinity compared to the KD calculated in the absence of
compound 8a) (Figure 5 and Supplementary Table S9).
Given the SAR relationship of the compounds and the
structural data that show they all bind in the same place, we
conclude that our inhibitor series all act as competitive
391
dx.doi.org/10.1021/cb300426u | ACS Chem. Biol. 2013, 8, 387−396

ACS Chemical Biology
Articles
Figure 3. Crystal complexes of RmlA-8a and RmlA-8f. (A) 8a bound in the allosteric site of RmlA. (B) Schematic representation of the key
interactions present in the RmlA-8a complex. (C) 8f bound in the allosteric site of RmlA. (D) Schematic representation of the key interactions
present in the RmlA-8f complex. Hydrogen bonds are shown as black dotted lines in the stick figures and green dotted lines in the schematics.
inhibitors of G1P. ITC data for three compounds (1, 2, and 8a)
correlate with the IC50 values, thus the tighter the binding the
more potent the inhibitor. This is in contrast to common
assumption that an inhibitor binding at a site remote from the
active site would behave as a noncompetitive inhibitor. The
potential for compounds binding at a second (allosteric) site to
act as competitive, noncompetitive, or uncompetitive inhibitors
is however known. dTTP, which binds at the allosteric site,
weakly inhibits the enzyme (Supplementary Figure S18).
dTDP-D-glucose, dTDP, and dTDP-L-rhamnose also bind at
this site (all products in the dTDP-L-rhamnose pathway). 15 We
determined IC50 for dTMP, dTTP, dTDP-glucose, and dTDP-
L-rhamnose (Supplementary Table S10). We suggest that
binding at this second site is the point of control of the
rhamnose pathway; dTDP-L-rhamnose (as well as any
intermediates on the pathway) shuts down RmlA thus
regulating pathway flux.
is ≫1, the expected value for a 1:1 binding relationship. Indeed
the Hill slope for 8a was calculated as 3.61 0.04, which is
approaching the theoretical maximum of 4 for the RmlA
tetramer. This highly cooperative behavior is unusual, and we
can find no comparable case of such high values for a tetramer
(hemoglobin has a Hill coefficient of 2.8). Slopes of this
magnitude imply the tetramer is either completely free from
inhibitor or all four sites are occupied.
The RmlA active sites are located at the dimer−dimer
interfaces, and crucial residues in loops 10−24, 138−148, 160−
162, and 224−232 key in binding and orienting G1P and dTTP
are also involved in intersubunit contacts (Figure 6).
Comparisons of all inhibitor-bound structures we have
determined show that the loop structure and tetramer
arrangement are essentially unchanged between them. The
tetramer arrangement we see in our inhibited complexes is
indeed different from that observed in an E. coli RmlA-dTTP
complex with an empty allosteric site (PDB code 1MC314).
(We have been unable to obtain an equivalent P. aeruginosa
RmlA complex, the ideal comparison point. In our attempts
We re-examined our structural and kinetic data to identify a
molecular basis for this inhibition. It is interesting to note that
the Hill slope for inhibitor 8a (and many other hit compounds)
392
dx.doi.org/10.1021/cb300426u | ACS Chem. Biol. 2013, 8, 387−396

ACS Chemical Biology
Articles
Figure 5. SPR sensorgrams showing substrate interactions with RmlA
in the presence of 8a. (A) dTTP titration in the presence of 8a and
G1P reveals dTTP binds to RmlA with a KD of 101 μM (95% CI 79−
130 μM). (B) G1P titration in the presence of 8a and dTTP reveals
G1P does not bind to the RmlA, thus confirming that 8a displays
competitive inhibition with respect to G1P binding.
Figure 4. Lineweaver−Burke analysis showing the effects of 8a on
binding of substrates. (A) Compound 8a has little effect on dTTP
binding unless at a very high concentration. (B) Compound 8a shows
competitive inhibition pattern with respect to G1P binding. Controls
with no inhibitor are shown as red lines.
At the active site we note, in particular, changes in two
regions that are both involved in binding G1P: the side chains
of Glu161−Lys162 and the backbone at Tyr145. Both regions
are shifted by around 2 Å, creating a larger pocket in the
uninhibited structure relative to inhibited. These regions flank
the turn at Gln152, which as mentioned previously is sensitive
to the changes at the allosteric site. The changes in the loops
that bind dTTP between the structures are much less
significant, consistent with the lack of inhibition of dTTP
binding. We suggest that the inhibitors act by locking the
conformational state of the RmlA tetramer and thereby
preventing dTTP from inducing the conformational changes
necessary to create the G1P binding site.
Conclusion. We report a novel small molecule scaffold that
is a potent inhibitor of the first enzyme (RmlA) of the dTDP-L-
rhamnose biosynthetic pathway. The pathway is essential for M.
tuberculosis where the compounds show some biological
activity. In a mouse model we have also demonstrated that
RmlA is essential for virulence of P. aeruginosa PAO1. This
compound series seem to work by exploiting a fundamental
property of RmlA, that it possesses an allosteric site that acts as
the point of control for the biosynthetic pathway. The second
site is present in all RmlA enzymes but absent in other
members of the nucleotidyl transferase superfamily, for example
human glucose-1-phosphate uridylyltransferase.32 Although the
both sites are occupied or both empty, resulting in a tetramer
arrangement identical to that of the inhibited structures.) The
change in the tetramer organization can be magnified by
superimposing the structures using only one subunit of each
structure (Figure 6). Using this method of superposition, the
main chain Arg219 whose side chain inserts into the allosteric
site from the other monomer in the inhibited structure (where
it makes a monomer−monomer salt bridge as well as stacking
with the inhibitor) has shifted over 6 Å in the uninhibited
structure. The four C-terminal residues (290−293) that fold
back over the allosteric site in the inhibited structure extend
into space in the uninhibited structure. The different structure
at the C-terminus results in a significant structural shift of the
turn at Gln152 in the other monomer. These changes in the
monomer−monomer interface are reflected at the dimer−
dimer interface where the inhibited structure has a “tighter”
packing arrangement. This transmission of structural change
from one allosteric binding site across the monomer−monomer
interface to the other allosteric site and then across the dimer−
dimer interface is we believe the origin of the remarkable
cooperativity of inhibitor binding.
393
dx.doi.org/10.1021/cb300426u | ACS Chem. Biol. 2013, 8, 387−396

ACS Chemical Biology
Articles
Figure 6. Ribbon diagrams showing the effects on RmlA of having an occupied or empty allosteric site. (A) Overlay of the RmlA-8a complex
(subunits colored light blue, dark blue, green, and cyan) and E. coli RffH (PDB code 1MC3; subunits colored pink, wheat, yellow, and magenta),
which has an empty allosteric site. Subunit A from each was overlaid, and the different tetrameric organization can be seen. (B) Close up of the active
and allosteric site regions in the overlay. Color scheme as in panel A. The highlighted loops (red for RmlA-8a, orange for RffH-dTTP) show the
conformational changes that occur between the empty allosteric site and the inhibitor-occupied allosteric site. Loops II and IV form the G1P binding
site. The loops correspond to the following residues: I = 10−24; II = 138−148; III = 150−157; IV = 160−162; V = 224−232; VI = 290−293. The
substrate dTTP from 1MC3 is shown in black sticks (phosphates in green, oxygens in red), and an overlay of the product dTDP-glucose from 1H5T
is shown as white sticks (phosphates in green, oxygens in red) to indicate the binding region in the active site. Inhibitor 8a is shown as yellow sticks
(nitrogens in blue, oxygens in red, and sulfur in orange).
obtained were tested experimentally with results being reported in
Table 1 and Supplementary Tables S2 and S3.
Inhibitor Synthesis. All analogues of 1 were prepared according
compounds are clearly related to the thymidine ring, they are
not identical. Their failure to bind at the active site of RmlA
would suggest that they are imperfect mimics of thymidine and
thus may not be generally useful against other thymidine sugar-
utilizing enzymes.8 Despite binding remotely from the active
site, the inhibitors are competitive with G1P and bind in a
highly cooperative manner. Our data suggest this is because
they bind at a protein−protein interface, locking the
conformation of the protein tetramer and thereby the active
site loops. This prevents RmlA from adopting the required
conformational change necessary to bind G1P.
to the protocols supplied in the Supporting Information.
Protein Crystallization. Initial crystallization conditions were
found by sparse matrix screening. Precipitant conditions were
optimized to 4% PEG 6000, 100 mM MES pH 6.0, 50 mM MgCl2,
100 mM NaBr, and 1% β-mercaptoethanol. Crystals were grown using
the sitting drop method with drops comprising 1 μL protein (10 mg
mL−1) mixed with 1 μL precipitant. Crystals were grown overnight to
dimensions of 0.2 mm × 0.2 mm × 0.1 mm, and cryoprotected in
precipitant substituted with 25% glycerol for 5 s, prior to be frozen in a
stream of nitrogen gas at 100 K for data collection. Crystallization
conditions different from previously published.33
Soaking/Co-crystallization. Complexes of RmlA and inhibitor
were prepared by soaking or co-crystallization. For soaking, apo-
crystals were grown as described in the Supporting Information. Once
formed, solid compound was added and left to incubate with the
crystals overnight. A selected crystal was then cryo-protected and
frozen in a stream of nitrogen gas at 100 K for data collection. For co-
crystallization, drops were set up as before, but solid compound was
added to the drop prior to sealing and incubation. Crystals typically
grew overnight.
METHODS
■
Cloning, Expression, and Purification. P. aeruginosa RmlA was
cloned, expressed, and purified based on protocols previously reported
by us33 (see Supporting Information for details). A similar procedure
has not yet yielded sufficient amounts of pure RmlA from M.
tuberculosis for robust in vitro assay.
RmlA Hit Identification and Validation. The RmlA high-
throughput screen was performed using a Dundee Drug Discovery
Unit in-house diverse compound collection of 15,667 molecules. Each
assay was performed in a 50 μL reaction volume containing 50 mM
Tris, pH 7.4, 5 mM MgCl2, 1 mM dithiothreitol, 0.05% NP-40, 0.1
mM EDTA, 0.1 mM EGTA, 1.5 nM recombinant RmlA, 0.8 μg mL−1
pyrophosphatase, 5 μM dTTP, 5 μM G1P, and 30 μM test compound.
Test compound was transferred to all assay plates before 25 μL of
enzyme mix was added. The reaction was initiated and stopped with
the additions of 25 μL of substrate mix and 50 μL of BIOMOL Green,
respectively. The assay was run for 30 min at RT, and the BIOMOL
Green signal was allowed to develop for 30 min before the absorbance
of each well was read at 650 nm. Compounds were designated as hits if
the percentage inhibition (PI) was >30% (this value represents 3
standard deviations from the mean data set PI). Preliminary hits were
either cherry picked from the original library plates or repurchased.
Ten-point inhibitor IC50 curves were generated and assays carried out
as described in the Supporting Information. All IC50 curve fitting was
undertaken using ActivityBase XE from IDBS with curves fitted to a
four-parameter logistic dose−response curve. All test compound
curves had floating top and bottom, and prefit was used for all four
parameters.
Data Collection. Data were processed with iMOSFLM34 or
XIA235 incorporating XDS.36 Each structure was solved using
MOLREP,37 initially using a monomer from 1FZW15 (apo P.
aeruginosa RmlA structure), and subsequent complexes using the
RmlA-1 subunit of the complex with the inhibitor removed.
Refinement used REFMAC538 and statistics shown in Supplementary
Tables S1 and S4−S6. Model building was performed using COOT,39
and ligands built using the PRODRG server.40 Structural figures were
prepared with PyMOL.41 Ligands were built into difference electron
density which was also stronger for the “thymidine-like” ring and
weaker for substituents. Occupancy was set to 1 for all atoms in ligand
except 4, where the five-membered ring was modeled in two (0.5
occupancy) positions. The unit cells are different from those in the
PDB for other previously reported P. aeruginosa complexes. Unbiased
maps for each complex are shown in Supplementary Figure S19.
Surface Plasmon Resonance. Binding of analytes (substrates and
8a) to RmlA was investigated using a Biacore 3000 instrument as
described in the Supporting Information.
Kinetic Parameter Determinations. The enzymatic activity of
RmlA was determined using a protocol analogous to that described
above for the HTS. Michaelis constants for each substrate (dTTP and
2D Similarity Screens. 2D similarity searches were run on our in-
house database using the Dotmatics browser. Purchased compounds
394
dx.doi.org/10.1021/cb300426u | ACS Chem. Biol. 2013, 8, 387−396

ACS Chemical Biology
Articles
G1P) were determined with data fitted to the Michaelis−Menten
equation using GraFit. Data sets for each substrate (dTTP and G1P)
were also fitted to the Hill equation. Further details are given in the
Supporting Information.
Pseudomonas aeruginosa RmlA Gene Knockout Virulence
Studies. Wild-type and RmlA knockout mutant strains of P.
aeruginosa were prepared and used to infect NMRI outbred mice.
Bacterial quantification using mouse lung homogenates is detailed in
the Supporting Information.
Mycobacterium tuberculosis MIC Determination. MIC values
were determined against M. tuberculosis (H37Rv) by the microbroth
dilution method as described in the Supporting Information.42
Accession Codes. Protein Data Bank codes 4ARW (RmlA-1),
4AS6 (RmlA-2), 4B42 (RmlA-3), 4B2X (RmlA-4), 4B4B (RmlA-6),
4ASJ (RmlA-8a), 4ASY (RmlA-8f), 4ASP (RmlA-15a), 4B3U (RmlA-
12b), 4B4G (RmlA-8j), 4B4M (RmlA-8k), 4B5B (RmlA-8p).
(8) Singh, S., Phillips, G. N., Jr., and Thorson, J. S. (2012) The
structural biology of enzymes involved in natural product
glycosylation. Nat. Prod. Rep. 29, 1201−1237.
(9) Qu, H., Xin, Y., Dong, X., and Ma, Y. (2007) An rmlA gene
encoding d-glucose-1-phosphate thymidylyltransferase is essential for
mycobacterial growth. FEMS Microbiol. Lett. 275, 237−243.
(10) Li, W., Xin, Y., McNeil, M. R., and Ma, Y. (2006) rmlB and
rmlC genes are essential for growth of mycobacteria. Biochem. Biophys.
Res. Commun. 342, 170−178.
(11) Babaoglu, K., Page, M. A., Jones, V. C., McNeil, M. R., Dong, C.,
Naismith, J. H., and Lee, R. E. (2003) Novel inhibitors of an emerging
target in Mycobacterium tuberculosis; substituted thiazolidinones as
inhibitors of dTDP-rhamnose synthesis. Bioorg. Med. Chem. Lett. 13,
3227−3230.
(12) Mills, J. A., Motichka, K., Jucker, M., Wu, H. P., Uhlik, B. C.,
Stern, R. J., Scherman, M. S., Vissa, V. D., Pan, F., Kundu, M., Ma, Y.
F., and McNeil, M. (2004) Inactivation of the mycobacterial
rhamnosyltransferase, which is needed for the formation of the
arabinogalactan-peptidoglycan linker, leads to irreversible loss of
viability. J. Biol. Chem. 279, 43540−43546.
(13) Sivendran, S., Jones, V., Sun, D., Wang, Y., Grzegorzewicz, A. E.,
Scherman, M. S., Napper, A. D., McCammon, J. A., Lee, R. E.,
Diamond, S. L., and McNeil, M. (2010) Identification of triazinoindol-
benzimidazolones as nanomolar inhibitors of the Mycobacterium
tuberculosis enzyme TDP-6-deoxy-D-xylo-4-hexopyranosid-4-ulose
3,5-epimerase (RmlC). Biorg. Med. Chem. 18, 896−908.
(14) Sivaraman, J., Sauve, V., Matte, A., and Cygler, M. (2002)
Crystal structure of Escherichia coli glucose-1-phosphate thymidylyl-
transferase (RffH) complexed with dTTP and Mg2+. J. Biol. Chem.
277, 44214−44219.
(15) Blankenfeldt, W., Asuncion, M., Lam, J. S., and Naismith, J. H.
(2000) The structural basis of the catalytic mechanism and regulation
of glucose-1-phosphate thymidylyltransferase (RmlA). EMBO J. 19,
6652−6663.
(16) Kim, H., Choi, J., Kim, T., Lokanath, N. K., Ha, S. C., Suh, S. W.,
Hwang, H. Y., and Kim, K. K. (2010) Structural basis for the reaction
mechanism of UDP-glucose pyrophosphorylase. Molecules Cells 29,
397−405.
(17) Melo, A., and Glaser, L. (1965) The nucleotide specificity and
feedback control of thymidine diphosphate D-glucose pyrophosphor-
ylase. J. Biol. Chem. 240, 398−405.
ASSOCIATED CONTENT
* Supporting Information
■
S
IC50 and Hill coefficient determination of 1 and 2; data
collection and refinement statistics; enzyme−inhibitor inter-
action schematics; structural overlays; RmlA-knockout in mice;
MIC values in M. tuberculosis; Km determination; SPR binding
data; unbiased Fo − Fc maps for each ligand; Methods details;
synthesis of inhibitors; NMR spectra of inhibitors from Table 1.
This material is available free of charge via the Internet at
http://pubs.acs.org.
AUTHOR INFORMATION
Corresponding Author
*E-mail: naismith@st-andrews.ac.uk; njw3@st-andrews.ac.uk.
■
Author Contributions
+These authors contributed equally to this work.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
The authors wish to thank the Diamond Light Source staff for
assistance with data collection and K. Barrack for assistance
with ITC. This work was supported by grants from the
European Community’s Seventh Framework Programme
(Aeropath; grant agreement no. 223461), The Scottish
Universities Life Science Alliance (L.P., Ph.D. studentship)
and the Royal Society (N.W., URF Fellowship).
(18) Barton, W. A., Lesniak, J., Biggins, J. B., Jeffrey, P. D., Jiang, J.,
Rajashankar, K. R., Thorson, J. S., and Nikolov, D. B. (2001) Structure,
mechanism and engineering of a nucleotidylyltransferase as a first step
toward glycorandomization. Nat. Struct. Biol. 8, 545−551.
(19) Barton, W. A., Biggins, J. B., Jiang, J., Thorson, J. S., and
Nikolov, D. B. (2002) Expanding pyrimidine diphosphosugar libraries
via structure-based nucleotidylyltransferase engineering. Proc. Natl.
Acad. Sci. U.S.A. 99, 13397−13402.
The coordinates of the RmlA complexes have been deposited
in the Protein Data Bank.
(20) Williams, G. J., Yang, J., Zhang, C., and Thorson, J. S. (2011)
Recombinant E. coli prototype strains for in vivo glycorandomization.
ACS Chem. Biol. 6, 95−100.
REFERENCES
■
(1) Strateva, T., and Yordanov, D. (2009) Pseudomonas aeruginosa−a
phenomenon of bacterial resistance. J. Med. Microbiol. 58, 1133−1148.
(2) Livermore, D. M. (1984) Penicillin-binding proteins, porins and
outer-membrane permeability of carbenicillin-resistant and -susceptible
strains of Pseudomonas aeruginosa. J. Med. Microbiol. 18, 261−270.
(3) Livermore, D. M. (2001) Of Pseudomonas, porins, pumps and
carbapenems. J. Antimicrob Chemother. 47, 247−250.
(21) Cid, J. M., Duvey, G., Tresadern, G., Nhem, V., Furnari, R.,
Cluzeau, P., Vega, J. A., de Lucas, A. I., Matesanz, E., Alonso, J. M.,
Linares, M. L., Andres, J. I., Poli, S. M., Lutjens, R., Himogai, H.,
Rocher, J. P., Macdonald, G. J., Oehlrich, D., Lavreysen, H., Ahnaou,
A., Drinkenburg, W., Mackie, C., and Trabanco, A. A. (2012)
Discovery of 1,4-disubstituted 3-cyano-2-pyridones: a new class of
positive allosteric modulators of the metabotropic glutamate 2
receptor. J. Med. Chem. 55, 2388−2405.
(4) Hirsch, E. B., and Tam, V. H. (2010) Impact of multidrug-
resistant Pseudomonas aeruginosa infection on patient outcomes. Expert
Rev. Pharmacoecon. Outcomes Res. 10, 441−451.
(22) Trabanco, A. A., Cid, J. M., Lavreysen, H., Macdonald, G. J., and
Tresadern, G. (2011) Progress in the developement of positive
allosteric modulators of the metabotropic glutamate receptor 2. Curr.
Med. Chem. 18, 47−68.
(23) Barda, D. A., Wang, Z. Q., Britton, T. C., Henry, S. S.,
Jagdmann, G. E., Coleman, D. S., Johnson, M. P., Andis, S. L., and
Schoepp, D. D. (2004) SAR study of a subtype selective allosteric
potentiator of metabotropic glutamate 2 receptor, N-(4-phenox-
(5) Page, M. G., and Heim, J. (2009) Prospects for the next anti-
Pseudomonas drug. Curr. Opin. Pharm. 9, 558−565.
(6) LoBue, P. (2009) Extensively drug-resistant tuberculosis. Curr.
Opin. Infect. Dis. 22, 167−173.
(7) World Health Organisation (2011) Towards universal access to
diagnosis and treatment of multidrug-resistant and extensively drug-
resistant tuberculosis by 2015, pp 1−119, WHO, Geneva.
395
dx.doi.org/10.1021/cb300426u | ACS Chem. Biol. 2013, 8, 387−396

ACS Chemical Biology
Articles
yphenyl)-N-(3-pyridinylmethyl)ethanesulfonamide. Bioorg. Med.
Chem. Lett. 14, 3099−3102.
(24) Gill, J. K., Dhankher, P., Sheppard, T. D., Sher, E., and Millar, N.
S. (2012) A series of alpha7 nicotinic acetylcholine receptor allosteric
modulators with close chemical similarity but diverse pharmacological
properties. Mol. Pharmacol. 81, 710−718.
(25) Kessl, J. J., Jena, N., Koh, Y., Taskent-Sezgin, H., Slaughter, A.,
Feng, L., de Silva, S., Wu, L., Le Grice, S. F., Engelman, A., Fuchs, J. R.,
and Kvaratskhelia, M. (2012) Multimode, cooperative mechanism of
action of allosteric HIV-1 integrase inhibitors. J. Biol. Chem. 287,
16801−16811.
(26) Christopoulos, A. (2002) Allosteric binding sites on cell-surface
receptors: novel targets for drug discovery. Nat. Rev. Drug Discovery 1,
198−210.
(27) Lewis, J. A., Lebois, E. P., and Lindsley, C. W. (2008) Allosteric
modulation of kinases and GPCRs: design principles and structural
diversity. Curr. Opin. Chem. Biol. 12, 269−280.
(28) Brenk, R., Schipani, A., James, D., Krasowski, A., Gilbert, I. H.,
Frearson, J., and Wyatt, P. G. (2008) Lessons learnt from assembling
screening libraries for drug discovery for neglected diseases.
ChemMedChem 3, 435−444.
(29) Martin, B., Pallen, C. J., Wang, J. H., and Graves, D. J. (1985)
Use of fluorinated tyrosine phosphates to probe the substrate
specificity of the low molecular weight phosphatase activity of
calcineurin. J. Biol. Chem. 260, 14932−14937.
(30) Zuccotti, S., Zanardi, D., Rosano, C., Sturla, L., Tonetti, M., and
Bolognesi, M. (2001) Kinetic and crystallographic analyses support a
sequential-ordered bi bi catalytic mechanism for Escherichia coli
glucose-1-phosphate thymidylyltransferase. J. Mol. Biol. 313, 831−843.
(31) Lafleur, K., Huang, D., Zhou, T., Caflisch, A., and Nevado, C.
(2009) Structure-based optimization of potent and selective inhibitors
of the tyrosine kinase erythropoietin producing human hepatocellular
carcinoma receptor B4 (EphB4). J. Med. Chem. 52, 6433−6446.
(32) Yu, Q., and Zheng, X. (2012) The crystal structure of human
UDP-glucose pyrophosphorylase reveals a latch effect that influences
enzymatic activity. Biochem. J. 442, 283−291.
(33) Blankenfeldt, W., Giraud, M. F., Leonard, G., Rahim, R.,
Creuzenet, C., Lam, J. S., and Naismith, J. H. (2000) The purification,
crystallization and preliminary structural characterization of glucose-1-
phosphate thymidylyltransferase (RmlA), the first enzyme of the
dTDP-L-rhamnose synthesis pathway from Pseudomonas aeruginosa.
Acta Crystallogr. Sect. D: Biol. Crystallogr. 56, 1501−1504.
(34) Leslie, A. G. (2006) The integration of macromolecular
diffraction data. Acta Crystallogr. Sect. D: Biol. Crystallogr. 62, 48−57.
(35) Winter, G. (2010) xia2: an expert system for macromolecular
crystallography data reduction. J. Appl. Crystallogr. 43, 186−190.
(36) Kabsch, W. (2010) Xds. Acta Crystallogr. Sect. D: Biol.
Crystallogr. 66, 125−132.
(37) Vagin, A., and Teplyakov, A. (2010) Molecular replacement
with MOLREP. Acta Crystallogr. D: Biol. Crystallogr. 66, 22−25.
(38) Murshudov, G. N., Vagin, A. A., and Dodson, E. J. (1997)
Refinement of macromolecular structures by the maximum-likelihood
method. Acta Crystallogr. D: Biol. Crystallogr. 53, 240−255.
(39) Emsley, P., and Cowtan, K. (2004) Coot: model-building tools
for molecular graphics. Acta Crystallogr. D: Biol. Crystallogr. 60, 2126−
2132.
(40) Schuttelkopf, A. W., and van Aalten, D. M. (2004) PRODRG: a
tool for high-throughput crystallography of protein-ligand complexes.
Acta Crystallogr. D: Biol. Crystallogr. 60, 1355−1363.
(41) DeLano, W. L. (2002) The PyMOL Molecular Viewer.
(42) Hurdle, J. G., Lee, R. B., Budha, N. R., Carson, E. I., Qi, J.,
Scherman, M. S., Cho, S. H., McNeil, M. R., Lenaerts, A. J., Franzblau,
S. G., Meibohm, B., and Lee, R. E. (2008) A microbiological
assessment of novel nitrofuranylamides as anti-tuberculosis agents. J.
Antimicrob. Chemother. 62, 1037−1045.
396
dx.doi.org/10.1021/cb300426u | ACS Chem. Biol. 2013, 8, 387−396