Application of fragment-based screening to the design of inhibitors of Escherichia coli DsbA

Article abstract of DOI:10.1002/anie.201410341

The thiol-disulfide oxidoreductase enzyme DsbA catalyzes the formation of disulfide bonds in the periplasm of Gram-negative bacteria. DsbA substrates include proteins involved in bacterial virulence. In the absence of DsbA, many of these proteins do not fold correctly, which renders the bacteria avirulent. Thus DsbA is a critical mediator of virulence and inhibitors may act as antivirulence agents. Biophysical screening has been employed to identify fragments that bind to DsbA from Escherichia coli. Elaboration of one of these fragments produced compounds that inhibit DsbA activity in vitro. In cell-based assays, the compounds inhibit bacterial motility, but have no effect on growth in liquid culture, which is consistent with selective inhibition of DsbA. Crystal structures of inhibitors bound to DsbA indicate that they bind adjacent to the active site. Together, the data suggest that DsbA may be amenable to the development of novel antibacterial compounds that act by inhibiting bacterial bacterial virulence,drug design, EcDsbA,fragment-based drugdiscovery,medicinal chemistry

Full text of DOI:10.1002/anie.201410341

Angewandte
Chemie
DOI: 10.1002/anie.201410341
Drug Discovery
Hot Paper
Application of Fragment-Based Screening to the Design of Inhibitors of
Escherichia coli DsbA**
Luke A. Adams, Pooja Sharma, Biswaranjan Mohanty, Olga V. Ilyichova, Mark D. Mulcair,
Martin L. Williams, Ellen C. Gleeson, Makrina Totsika, Bradley C. Doak, Sofia Caria,
Kieran Rimmer, James Horne, Stephen R. Shouldice, Mansha Vazirani, Stephen J. Headey,
Brent R. Plumb, Jennifer L. Martin, BegoÇa Heras,* Jamie S. Simpson,* and Martin J. Scanlon*
Abstract: The thiol-disulfide oxidoreductase enzyme DsbA
catalyzes the formation of disulfide bonds in the periplasm of
Gram-negative bacteria. DsbA substrates include proteins
involved in bacterial virulence. In the absence of DsbA,
many of these proteins do not fold correctly, which renders the
bacteria avirulent. Thus DsbA is a critical mediator of
virulence and inhibitors may act as antivirulence agents.
Biophysical screening has been employed to identify fragments
that bind to DsbA from Escherichia coli. Elaboration of one of
these fragments produced compounds that inhibit DsbA
activity in vitro. In cell-based assays, the compounds inhibit
bacterial motility, but have no effect on growth in liquid
culture, which is consistent with selective inhibition of DsbA.
Crystal structures of inhibitors bound to DsbA indicate that
they bind adjacent to the active site. Together, the data suggest
that DsbA may be amenable to the development of novel
antibacterial compounds that act by inhibiting bacterial
virulence.
slowing down the development of resistance while sparing the
normal host microbiota.[4] The development of antivirulence
agents presents a number of challenges, and targeting
virulence remains to be proven in a clinical setting. None-
theless, compounds have been reported that inhibit virulence
processes.[3c]
Many virulence factors produced by Gram-negative
bacteria are secreted or surface-exposed proteins that are
stabilized by disulfide bonds. Disulfide bond formation in the
periplasm is catalyzed by enzymes of the Dsb (disulfide bond)
family.[5] DsbA is a soluble enzyme of the thioredoxin
superfamily that directly catalyzes disulfide bond formation
in substrates.[6] DsbB is an all-a-helical integral membrane
protein, which catalyzes reoxidation of DsbA.[7] The DsbA/
DsbB system introduces disulfides bonds in a broad range of
virulence factors and is therefore a central mediator of
bacterial virulence.[5] Compounds that inhibit DsbB in
biochemical assays have been reported, although their
activity as antivirulence agents is unknown.[8] Genetic evi-
dence has shown that Gram-negative bacteria lacking a func-
tional DsbA are avirulent in animal infection models.[9] For
example, mice infected with Burkholderia pseudomallei, the
causative agent of melioidosis, all died when infected with
a wild-type organism, but all survived when infected with
a dsbA-deficient mutant.[10] Similarly, deletion of dsbA in
uropathogenic Escherichia coli severely attenuated coloniza-
A
ntibiotic resistance is a growing problem that has been
recognized as one of the major health challenges of the 21st
century.[1] With few antibacterials in the drug pipeline, new
therapeutic targets and compounds with novel mechanisms of
action are needed to treat resistant bacterial strains.[1b,d,2] One
approach is to target bacterial virulence.[3] Antivirulence
compounds are expected to exert less selective pressure, thus
[*] Dr. L. A. Adams,[+] Dr. P. Sharma,[+] Dr. B. Mohanty, O. V. Ilyichova,
Dr. M. D. Mulcair, Dr. M. L. Williams, E. C. Gleeson, Dr. B. C. Doak,
Dr. S. Caria, Dr. K. Rimmer, Dr. J. Horne, M. Vazirani,
Dr. S. J. Headey, B. R. Plumb, Dr. J. S. Simpson, Dr. M. J. Scanlon
Medicinal Chemistry, Monash Institute of
E-mail: b.heras@latrobe.edu.au
[#] Present address: Institute of Health and Biomedical
Innovation, Queensland University of Technology
Kelvin Grove, QLD 4059 (Australia)
[+] These authors contributed equally to this work.
Pharmaceutical Sciences, Monash University
[**] We acknowledge funding from the National Health and Medical
Research Council (grant number 1009785) and the Australian
Research Council for an Australian Laureate Fellowship, a Future
Fellowship, and a Discovery Early Career Researcher Award to
J.L.M., B.H., and M.T., respectively (grant numbers FL0992138,
FT130100580, and DE130101169). We thank the Bio21 Institute
NMR Facility and the CSIRO Collaborative Crystallisation Centre
(www.csiro.au/C3). This research was undertaken on the MX1 and
MX2 beamlines at the Australian Synchrotron and at the UQ ROCX
Diffraction Facility.
381 Royal Parade, Parkville, VIC 3052 (Australia)
E-mail: jamie.simpson@monash.edu
martin.scanlon@monash.edu
Homepage: http://www.pharm.monash.edu.au
Dr. S. R. Shouldice, Prof. J. L. Martin, Dr. B. Heras[$]
Institute for Molecular Bioscience, University of Queensland
St. Lucia, QLD 4072 (Australia)
Dr. M. Totsika[#]
School of Chemistry and Molecular Biosciences
University of Queensland
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201410341.
St. Lucia, QLD 4072 (Australia)
[$] Present address: La Trobe Institute for
Molecular Science, La Trobe University
Melbourne, VIC 3086 (Australia)
Angew. Chem. Int. Ed. 2014, 53, 1 – 7
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1
These are not the final page numbers!

.
Angewandte
Communications
tion of the mouse bladder.[11] The loss of virulence in dsbA-
deficient mutants is attributed to misfolding and loss of
function of DsbA substrates.
compounds, as shown for phenylthiazole 1 (Figure 1a and
b). In this way, 171 fragments were identified as hits from the
STD-NMR screen.
The central role of DsbA in bacterial virulence and our
understanding of its catalytic mechanism make this enzyme
an attractive target for the development of inhibitors. Crystal
structures of DsbA have shown its mode of binding to
substrates[12] and DsbB,[7] suggesting that it is amenable to
structure-based approaches to the design of small-molecule
inhibitors. However, inhibiting protein–protein interactions is
often challenging.[13] In this context, the substrate-binding site
of DsbA comprises a large hydrophobic groove rather than
a well-defined binding pocket. Herein we report our approach
to overcome these challenges by developing functional small-
molecule inhibitors of E. coli DsbA (EcDsbA), which to the
best of our knowledge are the first reported nonpeptide
inhibitors of DsbA.
We screened a library of 1132 fragments sourced from the
Maybridge Ro3 collection to identify compounds that bind to
oxidized EcDsbA. The primary screen used saturation trans-
fer difference (STD) NMR spectra with mixtures of three to
five fragments.[14] Fragments with observable signals in STD-
NMR spectra of the mixtures were retested as single
Hits were validated by measuring chemical shift pertur-
bations (CSP) for backbone amide resonances in 15N hetero-
nuclear single quantum coherence (HSQC) spectra of 15N-
labelled EcDsbA upon addition of single fragments (1 mm).
Under these conditions, 37 fragments produced CSP ꢀ
0.01 ppm and were retained for further characterization.
Each of the fragments caused CSP for residues in a hydro-
phobic groove of EcDsbA (Figure 1c and d) that is adjacent
to the active site and is implicated in the binding of EcDsbA
to both substrates and EcDsbB. CSP analysis showed that
none of the fragments had achieved saturation at the highest
concentration tested (1 mm) indicating weak binding. There-
fore, a preliminary fragment scoring system was devised
based on the magnitude of CSP observed for residues in the
hydrophobic groove of EcDsbA that were unaffected by
changes in DMSO concentration (Figure 2, see also Figur-
es S1 and S2 in the Supporting Information).
Figure 2. Scoring of ligand binding to EcDsbA. Left: Histogram of CSP
per residue for unambiguously assigned residues of EcDsbA induced
by the compound shown. Hydrophobic groove residues are colored
red. Right: Expansion of HSQC spectra highlighting perturbation of
Gln164 and Ser169. The magnitude of CSP was used to score ligand
binding as a) ligand 2 score=1, b) ligand 3 score=2, and c) ligand 4
score=3.
Figure 1. Identification of fragments that bind EcDsbA. a) STD-NMR
showed that phenylthiazole 1 binds to EcDsbA. b) (Top) 1H NMR
spectrum of EcDsbA; (middle) Off-resonance 1H NMR spectrum of
EcDsbA plus 1; (bottom) STD-NMR spectrum of EcDsbA plus 1. STD
signals from 1 are indicated (*); buffer peaks (B) and impurities (I)
are labeled. c) Overlay of 15N HSQC spectra of EcDsbA in the presence
(red) and absence (blue) of 1 (1 mm). d) Residues with
CSPꢀ0.01 ppm in the HSQC spectrum are colored red on the
structure of EcDsbA (PDB ID: 1FVK). Also indicated are the active site
(yellow sphere) and hydrophobic groove (blue).
The 37 fragments could be clustered into 8 distinct
chemical classes and 11 singletons based on linear fingerprints
(Figure 3 and S3). To expedite the initial evaluation of
structure–activity relationships (SAR), singletons were set
aside, and classes 1–5 were examined based on the obvious
scaffold similarities within the clusters. Class 5 was depriori-
tized as an examination of other anilines in the library
generated no clear SAR. Classes 1–4 were evaluated through
2
www.angewandte.org
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2014, 53, 1 – 7
These are not the final page numbers!

Angewandte
Chemie
Figure 4. Chemical structures and HSQC score for phenylthiazole
analogues.
binding. Substitution of the thiazole 2 for either imidazole 31
or oxazole 32 had a detrimental effect on binding, as did
inverting the methyl and acid substituents on the thiazole ring
(33). Replacing the acid with the corresponding hydroxy-
methyl group (35 vs. 34) reduced solubility in some cases and
thus gave conflicting SAR.
The phenylthiazole analogues (4, 36, 37, 38) with the
highest HSQC score were soaked into crystals of EcDsbA,
which resulted in one structure of the complex with com-
pound 4 at 1.45 ꢀ resolution. However, it was observed that
two molecules of 4 were stacked in the hydrophobic groove of
EcDsbA (Figure S5; PDB ID: 4WF5). Therefore co-crystal-
lization of 4 with EcDsbA was undertaken, resulting in
a structure with the expected 1:1 stoichiometry (Figure 5;
PDB ID: 4WF4). In this structure, the thiazole formed a p–p
stacking interaction with His32 and the aromatic ring was
located in a hydrophobic pocket. The carboxy group at
position 5 was orientated toward the more hydrophilic, water-
filled portion of the hydrophobic groove.
Given the strong CSP in the HSQC spectrum and the
available structural information, compound 4 was selected for
further elaboration. A growth vector attached to the carboxy
group of 4 was identified to enable extension along the
hydrophobic groove of EcDsbA. Preliminary studies into
simple esters or amides identified these as poor analogues
because of their lack of solubility in the aqueous buffer used
for screening. Thus we chose to synthesize a series of amino
acid derivatives to extend from the carboxy group at
Figure 3. Structures and HSQC score for the 26 fragment hits clus-
tered using linear fingerprints into the 8 chemical classes shown.
purchase and synthesis of analogues. Analysis of the binding
of thiophenes in class 1 suggested multiple modes of binding
to EcDsbA, while analogues of the benzothiophenes of class 3
showed little improvement in binding as measured from their
HSQC score, suggesting that progression of these classes was
challenging. In contrast, the diphenyl ethers (class 4) and
phenylthiazoles (class 2) both showed interpretable SAR, and
are under development. The phenylthiazoles were prioritized
because crystal structures of complexes were obtained for
analogues in this class (see below) and the progression from
phenylthiazole 1 is reported here.
Initially, 22 analogues of this scaffold were purchased and
the HSQC score of each was determined (Figure 4 and S4).
This produced initial SAR for the phenylthiazoles with the
following trends. 2-Phenylthiazoles, in particular with halogen
substituents on the aryl ring, were strongly favored (for
example, fragments 4 and 36). In contrast, replacing the
phenyl ring with pyridine (2 vs. 30), led to a loss of detectable
Angew. Chem. Int. Ed. 2014, 53, 1 – 7
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
3
These are not the final page numbers!

.
Angewandte
Communications
Figure 6. HSQC score, binding affinity, and inhibitory concentration of
selected amino acid derived ligands. KD and IC50 values were deter-
mined by SPR and POA, respectively.
short polar substituents (41 and 42) are more favorable,
whereas lysine 43 proved detrimental to binding.
Inhibition of EcDsbA activity was measured using
a previously described peptide oxidation assay (POA).[15]
Figures 6 and 7, Figures S7–S17, and Table S1 show that the
inhibitory concentrations correlated well with the KD values
determined by SPR.
The structure of 40 in complex with EcDsbA at 1.63 ꢀ
resolution was determined from crystals prepared by co-
crystallization (Figure 7d; PDB ID: 4WET). Unexpectedly,
the tyrosine of 40 orientated perpendicular to the phenyl-
thiazole core as opposed to extending along the hydrophobic
groove. This structure showed two hydrogen-bond interac-
tions from 40 to His32 and Gln164 of EcDsbA (Figure 7e).
Compound 40 was tested for its ability to inhibit EcDsbA
in a bacterial motility assay.[12] As DsbA is not required for
growth in rich media, but is essential for effective folding of
the FlgI component of the bacterial flagella, the expected
phenotype for an active EcDsbA inhibitor is inhibition of
E. coli motility, but no effect on growth. Addition of 40
(600 mm) in soft agar plates caused a 59% reduction in the
zone of E. coli swarming motility compared to DMSO-
containing control plates (Figure 7g), yet it had no effect on
bacterial growth in liquid culture at 1 mm concentration
(Figure 7h). Thus 40 was capable of inhibiting EcDsbA in an
isolated enzyme assay in vitro. In cell-based assays, 40
inhibited E. coli motility in soft agar, but had no effect on
growth of E. coli in liquid media, which is consistent with
selective inhibition of DsbA.
Figure 5. Co-crystal structure of 4 bound to EcDsbA. a) CSP observed
in HSQC spectra upon addition of 4 are mapped onto the structure of
EcDsbA using a color gradient from white (<0.001 ppm) to dark blue
(ꢀ0.15 ppm). Active-site residues Cys30–Cys33 are yellow. b) Phenyl-
thiazole 4 (depicted as green sticks) binds within the hydrophobic
groove of EcDsbA. A p–p stacking interaction with His32 is indicated.
Residues of EcDsbA within 4 ꢀ of 4 are shown as purple sticks and
labeled. Water molecules are shown as blue spheres and the sulfur
atoms of Cys30 and Cys33 are shown as yellow spheres. c) Simulated
annealing omit sA-weighted mFo-DFc electron density maps contoured
at 2.5s are shown as grey mesh.
position 5, in order to explore chemical space while main-
taining the solubilizing carboxylic acid group.
The binding affinity of these derivatives for EcDsbA was
evaluated using SPR. As shown in Figure 6 and Table S1, the
KD values measured by SPR correlated well with the prelimi-
nary HSQC scores. The majority of compounds identified as
strong hits by HSQC were observed by SPR to bind with
KD values in the 200–400 mm range, while weaker HSQC hits
gave KD values above 1 mm. The strongest binder was the
phenylalanine derivative 39 with a KD value of 196 mm.
Preliminary SAR suggested that aromatic (39 and 40) or
In conclusion, through a combination of fragment-based
screening, biophysical assays, and X-ray structure determi-
nation, we have identified fragments that bind noncovalently
to EcDsbA and inhibit its activity. The fragments show
concentration-dependent effects on EcDsbA in biochemical
(activity) and biophysical (binding) assays. Structural data
indicate that the fragments occupy a hydrophobic groove
4
www.angewandte.org
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2014, 53, 1 – 7
These are not the final page numbers!

Angewandte
Chemie
Received: October 22, 2014
Revised: November 25, 2014
Published online: && &&,
&&&&
Keywords: bacterialvirulence·
.
drug design · EcDsbA ·
fragment-baseddrugdiscovery·
medicinal chemistry
[1] a) M. A. Cooper, D. Shlaes,
Nature 2011, 472, 32; b) B.
Obama, Executive Order—
Combating antibiotic-resist-
ant bacteria, http://www.
whitehouse.gov/the-press-
office/2014/09/18/executive-
order-combating-antibiotic-
resistant-bacteria; c) R. R.
Uchil, G. S. Kohli, V. M.
Katekhaye, O. C. Swami, J.
Clin. Diagn. Res. 2014, 8,
ME01 – 04;
Health Organization in
Antimicrobial resistance:
global report on surveillance
2014; e) G. D. Wright,
d) World
Chem. Biol. 2012, 19, 3 –
10.
[2] J. Fernebro, Drug Resist.
Updates 2011, 14, 125 – 139.
[3] a) R. C. Allen, R. Popat,
S. P. Diggle, S. P. Brown,
Nat. Rev. Microbiol. 2014,
12, 300 – 308; b) B. Heras,
M. J. Scanlon, J. L. Martin,
Br. J. Clin. Pharmacol. 2014,
DOI:
10.1111/bcp.12356;
c) D. A. Rasko, V. Speran-
dio, Nat. Rev. Drug Discov-
ery 2010, 9, 117 – 128.
Figure 7. Biochemical, structural, and phenotypic characterization of 40. a) Raw sensorgrams and b) dose-
response plot of 40 binding to EcDsbA by surface plasmon resonance. c) Dose-dependent inhibition of
EcDsbA by 40 determined using a peptide oxidation assay. d) Crystal structure of 40 in complex with
EcDsbA. The surface is colored according to CSP observed in HSQC spectra of EcDsbA upon addition of 40.
Active-site residues Cys30–Cys33 are colored yellow. e) Detailed view of the binding site for 40 on EcDsbA.
f) Simulated annealing omit sA-weighted mFo-DFc electron-density maps for 40 contoured at 2.5s are
shown as grey mesh g) Motility of E. coli JCB816 on soft LB agar containing DMSO (left) and 40 (right).
h) Growth of E. coli JCB816 in LB broth containing DMSO or 40.
[4] a) A. K. Barczak, D. T.
Hung, Curr. Opin. Micro-
biol. 2009, 12, 490 – 496;
b) A. E. Clatworthy, E.
Pierson, D. T. Hung, Nat.
Chem. Biol. 2007, 3, 541 –
548; c) S. A. Stanley, D. T.
Hung, Biochemistry 2009,
48,
8776 – 8786;
d) M.
Zucca, S. Scutera, D. Savoia, Mini-Rev. Med. Chem. 2011, 11,
888 – 900.
[5] B. Heras, S. R. Shouldice, M. Totsika, M. J. Scanlon, M. A.
Schembri, J. L. Martin, Nat. Rev. Microbiol. 2009, 7, 215 – 225.
[6] J. L. Martin, J. C. Bardwell, J. Kuriyan, Nature 1993, 365, 464 –
468.
[7] K. Inaba, S. Murakami, M. Suzuki, A. Nakagawa, E. Yamashita,
K. Okada, K. Ito, Cell 2006, 127, 789 – 801.
[8] V. Frꢁh, Y. Zhou, D. Chen, C. Loch, E. Ab, Y. N. Grinkova, H.
Verheij, S. G. Sligar, J. H. Bushweller, G. Siegal, Chem. Biol.
2010, 17, 881 – 891.
adjacent to the active site of EcDsbA. Taken together, our
data show that these compounds may represent a starting
point for the development of antivirulence agents that inhibit
EcDsbA. Compounds with this spectrum of activity may exert
less selective pressure than traditional antibiotics that kill
bacteria or inhibit their growth, thus slowing down the
development of resistance, and could therefore represent
a future treatment of multi-drug resistant bacterial infec-
tions.[3c]
[9] T. Miki, N. Okada, H. Danbara, J. Biol. Chem. 2004, 279, 34631 –
34642.
Angew. Chem. Int. Ed. 2014, 53, 1 – 7
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
5
These are not the final page numbers!

.
Angewandte
Communications
[10] P. M. Ireland, R. M. McMahon, L. E. Marshall, M. Halili, E.
Furlong, S. Tay, J. L. Martin, M. Sarkar-Tyson, Antioxid. Redox
Signaling 2014, 20, 606 – 617.
[11] M. Totsika, B. Heras, D. J. Wurpel, M. A. Schembri, J. Bacteriol.
2009, 191, 3901 – 3908.
[12] J. J. Paxman, N. A. Borg, J. Horne, P. E. Thompson, Y. Chin, P.
Sharma, J. S. Simpson, J. Wielens, S. Piek, C. M. Kahler, H.
Sakellaris, M. Pearce, S. P. Bottomley, J. Rossjohn, M. J. Scanlon,
J. Biol. Chem. 2009, 284, 17835 – 17845.
[13] T. L. Nero, C. J. Morton, J. K. Holien, J. Wielens, M. W. Parker,
Nat. Rev. Cancer 2014, 14, 248 – 262.
[14] M. Mayer, B. Meyer, Angew. Chem. Int. Ed. 1999, 38, 1784 –
1788; Angew. Chem. 1999, 111, 1902 – 1906.
[15] F. Kurth, K. Rimmer, L. Premkumar, B. Mohanty, W. Duprez,
M. A. Halili, S. R. Shouldice, B. Heras, D. P. Fairlie, M. J.
Scanlon, J. L. Martin, PLoS One 2013, 8, e80210.
6
www.angewandte.org
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2014, 53, 1 – 7
These are not the final page numbers!

Angewandte
Chemie
Communications
Drug Discovery
L. A. Adams, P. Sharma, B. Mohanty,
O. V. Ilyichova, M. D. Mulcair,
M. L. Williams, E. C. Gleeson, M. Totsika,
B. C. Doak, S. Caria, K. Rimmer, J. Horne,
S. R. Shouldice, M. Vazirani, S. J. Headey,
B. R. Plumb, J. L. Martin, B. Heras,*
J. S. Simpson,*
Combating bacterial virulence: DsbA is
an oxidoreductase enzyme and a key
mediator of virulence in Escherichia coli.
Using fragment-based screening, com-
pounds were developed that inhibit DsbA
activity in vitro and E. coli motility in
a cell-based assay. Crystal structures of
the compounds in complex with DsbA
provide a rationale for their activity.
M. J. Scanlon*
&&&& — &&&&
Application of Fragment-Based Screening
to the Design of Inhibitors of Escherichia
coli DsbA
Angew. Chem. Int. Ed. 2014, 53, 1 – 7
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
7
These are not the final page numbers!