8-Mercaptoguanine Derivatives as Inhibitors of Dihydropteroate Synthase

Article abstract of DOI:10.1002/chem.201704730

Dihydropteroate synthase (DHPS) is an enzyme of the folate biosynthesis pathway, which catalyzes the formation of 7,8-dihydropteroate (DHPt) from 6-hydroxymethyl-7,8-dihydropterin pyrophosphate (DHPPP) and para-aminobenzoic acid (pABA). DHPS is the long-standing target of the sulfonamide class of antibiotics that compete with pABA. In the wake of sulfa drug resistance, targeting the structurally rigid (and more conserved) pterin site has been proposed as an alternate strategy to inhibit DHPS in wild-type and sulfa drug resistant strains. Following the work on developing pterin-site inhibitors of the adjacent enzyme 6-hydroxymethyl-7,8-dihydropterin pyrophosphokinase (HPPK), we now present derivatives of 8-mercaptoguanine, a fragment that binds weakly within both enzymes, and quantify sub-μm binding using surface plasmon resonance (SPR) to Escherichia coli DHPS (EcDHPS). Eleven ligand-bound EcDHPS crystal structures delineate the structure–activity relationship observed providing a structural framework for the rational development of novel, substrate-envelope-compliant DHPS inhibitors.

Full text of DOI:10.1002/chem.201704730

A Journal of
Accepted Article
Title: 8-Mercaptoguanine derivatives as inhibitors of dihydropteroate
synthase
Authors: James Swarbrick, Matthew Dennis, Michael Lee, Jitendra
Harjani, Mohamed Ahmed, Aaron Debono, Noel Pitcher,
Zhong-Chang Wang, Sandeep Chhabra, Nicholas Barlow,
Raphael Rahmani, Ben Cleary, olan Dolezal, Meghan
Hattarki, Luigi Aurelio, Jeremy Shonberg, Bim Graham, Tom
Peat, and Jonathan Baell
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Chem. Eur. J. 10.1002/chem.201704730
Link to VoR: http://dx.doi.org/10.1002/chem.201704730
Supported by

Chemistry - A European Journal
10.1002/chem.201704730
FULL PAPER
8
-Mercaptoguanine derivatives as inhibitors of dihydropteroate
synthase
Matthew L. Dennis,[a],[b] Michael D. Lee,[a][b] Jitendra R. Harjani, Mohamed Ahmed,
[a]
[a],[d]
Aaron J.
[
a]
[a]
[a],[c]
[a]
[a]
DeBono, Noel P. Pitcher, Zhong-Chang Wang,
Sandeep Chhabra, Nicholas Barlow, Raphaël
[
a]
[a]
[b]
[b]
[a]
[a]
Rahmani, Ben Cleary, Olan Dolezal, Meghan Hattarki, Luigi Aurelio, Jeremy Shonberg, Bim
[
a]
[b]
[a]
[a]
Graham, Thomas S. Peat, Jonathan B. Baell, and James D. Swarbrick*
Abstract: Dihydropteroate synthase (DHPS) is an enzyme of the
folate biosynthesis pathway which catalyzes the formation of 7,8-
dihydropteroate (DHPt) from 6-hydroxymethyl-7,8-dihydropterin
pyrophosphate (DHPPP) and para-aminobenzoic acid (pABA).
DHPS is the long-standing target of the sulfonamide class of
antibiotics that compete with pABA. In the wake of sulfa drug
resistance, targeting the structurally rigid (and more conserved)
pterin site has been proposed as an alternate strategy to inhibit
DHPS in wild-type and sulfa drug resistant strains. Following work
developing pterin-site inhibitors of the adjacent enzyme 6-
hydroxymethyl-7,8-dihydropterin pyrophosphokinase (HPPK), we
now present derivatives of 8-mercaptoguanine, a fragment that binds
weakly within both enzymes, and quantify sub-μM binding using
Surface Plasmon Resonance (SPR) to Escherichia coli DHPS
pyrophosphate (DHPPP) into dihydropteroate (DHPt) (Figure 1).
Sulfonamides exert their effect through competing with pABA by
reacting to form “dead-end” drug-pterin adducts, depleting
cellular DHPPP as well as inhibiting downstream enzymes.[
2]
(EcDHPS). Eleven ligand-bound EcDHPS crystal structures
delineate the structure-activity relationship observed providing a
structural framework for the rational development of novel, substrate
envelope-compliant DHPS inhibitors.
Introduction
The sulfonamide class of antibiotics was discovered in the
Figure 1. (A) Reaction pathway of DHPS showing the formation of DHPt from
DHPPP and pABA. (B) Chemical structures of DHPS inhibitors. (C) Structure
of the EcDHPS/pteroic acid complex displayed. The pteroic acid molecules are
displayed in gray.
1930s and, along with the related sulfone class, is still widely
used today to treat infectious diseases including malaria,
tuberculosis, toxoplasmosis, Pneumocystis jirovecii pneumonia
(
the most common fungal infection in HIV-infected patients), and
methicillin-resistant Staphylococcus aureus.[1] Sulfonamides
Despite decades of success in the clinic, mechanistic data
is still emerging to aid the rational design of DHPS inhibitors.
Crystallization studies have unraveled the catalytic mechanism
and the structural basis behind sulfa drug resistance.[3]
target the folate-pathway enzyme dihydropteroate synthase
(
DHPS), which catalyzes the condensation of p-aminobenzoic
acid (pABA) and 6-hydroxymethyl-7,8-dihydropterin-
Structural data has shed light on how the off-target inhibition of
sepiapterin reductase has led to the neurological side effects of
sulfa drugs following treatment.[4] These structural details and
other studies promise to help the development of newer
generation sulfa drugs with improved therapeutic effect and
longevity. Sulfa drug resistance rapidly emerged after usage
and is predominantly a consequence of point mutations in the
folP gene altering residues near to the pABA site.[6]
Given the dearth of new antibiotic classes, the pterin-
binding site of DHPS has been recognized as an alternative
druggable pocket to reinvigorate the design of DHPS inhibitors.
Crystal structures of DHPS from various organisms[3, 7] reveal
the pterin pocket to be rigid and lined with highly conserved
residues located deep in the enzyme, in stark contrast to the
pABA site which is nestled in the flexible loop regions. Taking
advantage of these properties of the pterin pocket (which is
[
[
a]
b]
ML Dennis, MD Lee, JR Harjani, M Ahmed, AJ DeBono, NP Pitcher,
Z-C Wang, S Chhabra, N Barlow, R Rahmani, B Cleary, L Aurelio, J
Shonberg, B Graham, JB Baell, JD Swarbrick.
Monash Institute of Pharmaceutical Sciences, Monash University,
[5]
381 Royal Parade, Parkville 3052, Victoria, Australia
ML Dennis, MD Lee, O Dolezal, M Hattarki, TS Peat.
CSIRO Biomedical Program, Manufacturing, Parkville, 3052,
Victoria, Australia
Z-C Wang, State Key Laboratory of Pharmaceutical Biotechnology,
Nanjing University, Nanjing 210093, People’s Republic of China
M Ahmed, School of Pharmacy, University College London,
Bloomsbury, London, WC1N 1AX, England
[
[
c]
d]
E-mail: james.swarbrick@monash.edu
Supporting information for this article is given via a link at the end of
the document.
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201704730
FULL PAPER
required for substrate specificity) is potentially beneficial; under
selective pressure, viable mutations conferring resistance to
pterin-site inhibitors would be difficult to evolve without
compromising activity. The conserved nature of the pterin pocket
indicates that a suitably designed pterin-site agent could
potentially elicit broad-spectrum activity and be effective against
organisms that have become sulfa drug resistant.[8] There are
however potential difficulties targeting the substrate site which
derive from drugging a pocket that is highly selective for a rather
polar substrate that binds with high affinity.
Results and Discussion
Crystallization of EcDHPS. Initially, protein crystallization
trials were performed using Tris-HCl (pH 8.5) as the protein
buffer, however poor reproducibility of crystal growth, suspected
to arise from stability issues, led us to perform a differential
scanning fluorimetry (DSF) assay[17] to screen for more suitable
buffer conditions. A buffer solution containing 50 mM trisodium
2 4 2
citrate, 150 mM NaH PO , and 350 mM K HPO4, pH 6.9, was
thus determined to provide EcDHPS with markedly improved
The molecular details of the transition state have been
trapped in crystallo revealing interesting details regarding the
shape, loop conformational changes, and intermolecular
interactions within the substrate binding envelope that can be
used to guide affinity, selectivity, and monitor substrate-
envelope ‘compliance’ during inhibitor design. Such compliance,
according to the ‘substrate-envelope hypothesis’,[9] would
prevent intermolecular interactions beyond those available to the
substrate and therefore limit mutations that are tolerated by the
substrate but not by any inhibitors. A fragment approach to
target allosteric pockets away from the active site may offer a
novel way to perturb DHPS function[10] and potentially mitigate
difficulties associated with polar substrate-like compounds.
Interest in pterin-site inhibitors started in the 1970s, when
stability (Figure S4), at both low and high temperatures, over the
previously reported conditions,[
7a, 18]
which enabled consistent
crystal growth and good diffraction for a range of complexes.
8MG is scaffold hop of the pterin substrate.
a
Previously, 8MG was crystallized in BaDHPS, despite not
showing any enzyme inhibition (at 250 µM) in a radiolabelled
pABA assay.[13] 8MG is
respectable ligand efficiency (LE) to EcDHPS (K
12 heavy atoms, corresponding to a LE of 0.46 kcal·mol·heavy
a
small fragment, binding with
= 76 µM over
D
-
1 [15c]
atom ).
It represents a close scaffold hop from the pterin
core[15a] sharing the same isocytosine ring and a synthetically
useful mercapto-group facilitating extension from the 8 position.
8
8
Elaboration at this position is justified by the fact that the C -S
vector points out from the deep pterin pocket in DHPS (Figure 2),
but is also oriented ~30° away from the methylene linker of the
native substrate. While direct derivatization of the pterin scaffold
has been carried out in earlier work (vide supra), elaboration of
8MG has not been investigated as a means to generate pterin-
site inhibitors of DHPS.
4,5-dioxo-1,4,5,6-tetrahydropyrimido[4,5-c]pyridazines
were
developed to inhibit Escherichia coli DHPS (EcDHPS) by
Burroughs Wellcome Co. (London, UK)[11] and a decade later
various
monocyclic
6-(alkylamino)-5-nitrosoisocytosines,
including 6-methylamino-5-nitrosoisocytosine (1, Figure 1), were
published by Lever et al.[12] Both compound series were pterin-
site inhibitors as verified through recent crystallographic studies
[
7c,
with Bacillus anthracis DHPS (BaDHPS) by the White group.
1
3]
The same group also identified other pterin-like compounds,
such as 8-mercaptoguanine (8MG) (Figure 1B) and several of
these were crystallized in the pterin site.[13-14]
We have been generating focused libraries based around
the 8MG core, having shown that the parent compound inhibits
6-hydroxymethyl-7,8-dihydropterin pyrophosphokinase (HPPK,
the enzyme preceding DHPS in the folate pathway), through
binding at the equivalent pterin site.[15] Our recent study
demonstrated that some of our 8MG derivatives also bound to
EcDHPS,[15c] which supports the rationale that suitably designed
pterin-like analogues may elicit beneficial polypharmacological
activity.[ Here, we have extended our 8MG chemical probe
series and measured equilibrium binding constants using SPR.
We report an array of crystal structures comprising novel
compounds in complex with EcDHPS that delineate key
intermolecular interactions and rationalize SAR of the compound
Figure 2. X-ray crystal structures looking into the pterin pocket of EcDHPS
bound to (A) 7 or (B) 1. Key residues are displayed in green. A water molecule
is displayed as a red sphere. Hydrogen bonds from the ligand are displayed as
dashed black lines
16]
7
9
8
Functionalization of 8MG at N , N , and C . Prior to
8
derivatization from the S handle, we briefly probed the chemical
space from the N , N , and C positions. Both the N7 and N9
7
9
8
series. Sub-micromolar affinities were observed (62, K
D
= 390
nM, Table 5), including one compound that was shown to bind
positions looked to be rather poor choices for elaboration of the
closely within the substrate envelope (59, K
constituting a 200-fold improvement in affinity and a rational
evolution of the parent fragment, 8MG.
D
= 650 nM, Table 4), 8MG core from simple superposition of 8MG on the pterin
9
moiety in the EcDHPS structure. For N , the equivalent nitrogen
group of the pterin substrate participates in a hydrogen bond
with residue D96.[7a] Methylation at this position (and therefore
removal of this interaction), as per 1 (Figure 2) and other
inhibitors,[
7c, 13-14]
is tolerated, however there appears to be
limited space to extend this substituent further. To test this, four
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201704730
FULL PAPER
simple 8MG derivatives with either methyl (3), ethyl (4), benzyl
apart from the direct interactions from the introduced substituent,
N -methylation potentially improves binding by more favorable
9
9
(
(
5), or phenethyl (6) N -functionalization were tested by SPR
Table 1). Indeed, only methyl-substitution maintained affinity for
desolvation as well as promoting the bound tautomer, i.e. an
unsaturated N -C bond in which the resulting HBA at N7
interacts with the sidechain of K221. As extension from N9
showed limited potential, strategically S8 represented the
obvious group for elaboration, offering synthetic ease for a range
for chemical extensions. We therefore focused on
7
8
EcDHPS compared to 8MG (3, K
compounds returned K values of 420 μM, 350 μM, and 160 μM,
respectively. A crystal structure of EcDHPS/9-methylguanine (7,
= 58 μM) (Figure 2) shows that the methyl group is
D
= 70 µM), while the other
D
K
D
accommodated by the slight displacement of the catalytic
residue D96 — a similar displacement is observed in the solved
complex of EcDHPS/1 — while the remaining hydrogen bonding
network of the pterin substrate core is retained (residues N115,
D185, and K221, and a water molecule, Figure 2). These
structures suggest that any further extension from the N9
position would be to a first approximation disfavored by potential
close contacts to the D96 residue. However, the affinity trend for
8
functionalization at the S position.
Table 2. Binding of 8MG-benzyl derivatives as determined by SPR
9
these compounds appears to indicate that, as the N substituent
further increases in size, additional interactions to the protein
may be formed that could compensate for these unfavorable
contacts. Inspection of the EcDHPS structure reveals that
compd
13
R
K
D
EcDHPS (µM)
8.5 ± 2.1a
7.0 ± 0.8
14.0 ± 0.2
9.5 ± 0.6
12.7 ± 0.3
33 ± 2
7
extension from the N position is predicted to be energetically
2-fluoro
disfavored, since the N7 atom of 8MG is part of a bifurcated
hydrogen bonding network involving the sidechain of residue
K221 and the carbonyl of 8MG.
14
2-chloro
15
2-bromo
16
17
18
19
20
21
22
23
24
25
26
27
28
29
2-methyl
Table 1. Binding of N7 and N9 8MG derivatives as determined by SPR
2-cyano
2-nitro
3-methyl
11 ± 2
3-methoxy
3-cyano
7.5 ± 0.5
11 ± 2
compd
X
SH
SCH
SH
SH
SH
SH
H
Y
H
K
D
EcDHPS (µM)
76.2 ± 12b
330a
8MG
4-fluoro
11 ± 2
2
3
H
2,2-difluoro
2,4-difluoro
3,4-difluoro
2-fluoro,3-methyl
2-methyl,4-fluoro
3-methyl,4-fluoro
2,3-dimethyl
2,5-dimethyl
3,5-dimethyl
14 ± 3
3
Methyl
Ethyl
Benzyl
70 ± 9
420a
12 ± 3
4
11 ± 1
5
350a
31 ± 8
6
CH
2
Benzyl
160a
13 ± 1
7
Methyl
58 ± 8
42a
8.2 ± 0.9
14 ± 2
8
Br
H
H
9
Methyl
OH
71a
200a
30
11.5 ± 0.1
14 ± 3
1
0
1
2
H
31
1
1
OH
Methyl
H
101 ± 2
160a
[a] Dennis et al.15c
N-morpholino
[
a]n=1. [b] Dennis et al.15c
A variety of substituents on the S8 benzyl group are
broadly tolerated. Previously we demonstrated that
derivatization from the S position can show improvement over
Guanine analogues with various functionalities at the C8
position — bromo (8), methyl (9), hydroxyl (10), and morpholino
12) — were tested (Table 1). Of these, only 8 showed improved
8
D
the affinity of 8MG to EcDHPS (K = 76 µM) with benzyl
(
functionalization yielding binding constants as low as 8.5 µM
(compound 13).[
15c]
Here, further substitution from the benzyl
binding relative to 8MG (~2-fold), returning an affinity of 42 μM.
9
N -methylation of 10 improved binding 2-fold (11, K
D
= 100 μM);
ring was investigated (Table 2).
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201704730
FULL PAPER
Figure 3. X-ray crystal structures of EcDHPS bound to (A) 13, (B) 36, or (C)
the three previously published EcDHPS crystal structures[7a]
where the dimer interface was present through crystallographic
symmetry. The new complex is similar, with overall backbone
41. Key residues are shown (green) and hydrogen bonds to residue R255 in
dashed black lines. Both monomers in the EcDHPS/13 complex is shown.
rmsd values of ~0.5 Å, when compared to these three previous
Testing with small substituents at all positions (ortho, meta,
para), in addition to naphthyl (32, 33), phenylethynyl (34), and
pyridyl (35) analogues, revealed affinities between 7 and 16 µM
for 21 of the 22 compounds tested (Tables 2 and 3). The 8MG-
benzyl derivative possessing o-chloro substitution (14) bound
with the highest affinity of these compounds. While the SAR is
rather broad, these analogues maintain increased binding
EcDHPS structures (1AJ2, 1AJ0, and 1AJZ).[
7a]
The biggest
differences between the two monomers of the 13/EcDHPS
complex are for residues 57–70 (loop 2) and 143–151 (loop 5,
absent in monomer B). Loop 1 (residues 25–34) is also
disordered for both monomers. A key difference between the
pterin-binding sites of the monomers is found in the orientation
of catalytic residue D96. This interacts with the ligand in
monomer A, but is instead oriented towards residue D56 in
monomer B — a trend consistent for most of the crystal
structures in this study.
D
compared with the simple unsubstituted benzyl (K = 44 µM)
previously measured by SPR.[15c]
Table 3. Binding of 8MG-benzyl analogues as determined by SPR.
The structure shows the 8MG core mimicking the
hydrogen bonding network of the substrate. While electron
density was strong for the 8MG moiety it was markedly weaker
for the benzyl moiety (Figure S2) and its orientation is somewhat
ambiguous, although the benzyl moiety appears to interact with
residues F190 and K221. The weak density for this group is
indicative of mobility and an absence of specific interactions,
which may help to explain the broad SAR observed for the 8MG-
benzyl series. The site beyond the 8MG-binding pocket is rather
compd
R
K
D
EcDHPS (µM)
15 ± 3
K
D
EcHPPK (µM)a
7.3 ± 0.6
8
open (depending on loop conformations) such that S -
3
2b
functionalized derivatives with appropriate flexibility can
generally avoid clashes with the enzyme.
8
S -methylenecarbonyl derivatives exhibit improved
D
binding. 8MG-acetophenone derivatives, for example 36 (K =
.9 µM[ ), were previously shown to display higher affinities for
15c]
3
33
16 ± 2
1.46 ± 0.04
EcDHPS than 8MG (Table 4).[
showed improved affinity (38, K
15c]
The acetyl derivative also
= 7.8 µM[15c]) relative to 8MG,
D
suggesting that the carbonyl group was primarily responsible for
the increased potency. Several amidic compounds (41–45) with
alkyl N-substitution were therefore tested (Table 4, Table S3), a
simple N-methylene acetamide group (41, K = 4.5 μM) proving
D
the most beneficial. Extending off the methylene linker between
the S and amide group was found to significantly reduce affinity
3
4
9.8 ± 0.9
15 ± 1
nd
35
15.9 ± 0.1
(
45, K
D
> 200 μM).
[
a] In presence of 10 mM Mg2+ and 1mM ATP. [b] 84% pure.
X-ray structures of EcDHPS bound to 36 and 41 reveal a
hydrogen bond between the carbonyl group of these compounds
and the guanidinium group of R255 (Figure 3). The amide of 41
and ketone of 36 overlay well and are oriented approximately
perpendicular to the plane of the guanidinium group. The phenyl
ring of 36 makes a favorable van der Waals interaction with the
alkyl sidechain of K221, though it should be noted that electron
density is noticeably poorer for this portion of the molecule,
A crystal structure of EcDHPS in complex with compound
3 featuring an o-fluorobenzyl group, was solved to 1.84 Å
resolution (Figure 3A). Two protomers are present in the
asymmetric unit, contrasting with the one protomer observed in
1
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201704730
FULL PAPER
possibly indicating mobility. The p-brominated variant of 36,
compound 37, bound with a K of 2.0 μM. These two crystal
D
4
7
8
22 ± 2
nd
nd
nd
nd
structures indicated that further extension from the amide of 41
may pick up additional interactions to the nearby residues,
particularly K221 and F190. 11 compounds (46–56) with
hydrophobic substituents, and either amide or ester linkers, were
thus synthesized. An N-phenyl acetamide bound the tightest of
4
7.1 ± 0.6
these (54, K
linker was observed to result in a loss of affinity (cf. 46 and 47,
= 4.5 and 22 μM, respectively).
D
= 2.1 μM). Again, extension from the methylene
49
2.9 ± 0.1
42 ± 2
62 ± 7
2.6 ± 0.1
2.8 ± 0.7
K
D
50
5.90 ± 0.04
Table 4. Binding of 8MG derivatives as determined by SPR
41.8 ±
5
1
2
4.08 ± 0.01
4.9 ± 0.2
2.20 ± 0.01
4.7 ± 0.3
4.7 ± 0.4
1.4 ± 0.3
0
.1
compd
R
K
D
EcDHPS
µM)
K
D
EcHPPK K
D
SaHPPK
(µM)e
5
66 ± 1
64 ± 8
12 ± 1
(
(µM)e
.9 ± 0.2ac (4.3 ± 0.82 ±
.2d) 0.04b
53
9.6 ± 0.1
3
36
0.60 ± 0.04b
0
5
4
5
2.05 ± 0.02
0
.70 ±
3
7
8
2.0 ± 0.1
0.57 ± 0.01b
1.85 ± 0.07a
0
.01b
5
12.1 ± 0.1
67 ± 1
57 ± 1
4.7 ± 0.8
4.0 ± 0.5
3
7.8 ± 1.0a
nd
c
1
8 ± 4 (11.8 ± 89.2 ±
56
9.9 ± 0.6
7.9 ± 0.3
3
9
0
7.6 ± 0.2b
4.6 ± 0.3b
0
.4d)
0.5b
4
20.0 ± 0.5
68 ± 1b
57
nd
nd
nd
nd
c
4
.5 ± 0.1 (7.6 ± 50.0 ±
8.8 ± 1.0b
c
41
0.83 ± 0.01
.3d)
0.1b
58
0
(
0.86 ± 0.01d)
4
2
3
8.2 ± 0.1
7.5 ± 0.2
53 ± 10b 1.5 ± 0.1b
5
9
0
0.65 ± 0.03
> 200
2.7 ± 0.1
5
9.4 ±
4
13 ± 1b
11 ± 1b
0
.4b
6
4.6 ± 0.1
8.8 ± 0.5
12 ± 3
74.2 ±
44
6.9 ± 0.1
0
.1b
[
a]Dennis et al.15c [b]Dennis et al.15d [c]Determined using Biacore T200 or [d]
ProteOn XPR36. [e] Tested in presence of 10mM Mg2+ and 1 mM ATP.
4
5
6
> 200
120 ± 10b
26 ± 3
46 ± 4b
4
4.53 ± 0.03
2.8 ± 0.2
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201704730
FULL PAPER
Figure 4. (A) Overlay of both monomers from the X-ray crystal structure of
9
5
), the N methyl group displaces these interactions, with D96
58/EcDHPS. X-ray crystal structure of the 59/DHPS complex, monomer (B) A
interacting instead with D56. In the 62-bound EcDHPS crystal
structure (Figure 5), D96 is likewise displaced, but is oriented
away from D56, instead interacting with residue T62. In the
and (C) B. Key residues are displayed in orange and green, and the ligand is
displayed in purple and yellow, for monomer A and B, respectively. Hydrogen
bonds to residue R255 are displayed as dashed black lines.
6
2/EcDHPS complex, the amide carbonyl of 62 interacts with
R255 as per the previous structures, while the N-phenyl moiety
projects up to interact with residues R220, K221, and H257.
Substituting a secondary amide for a carboxylic acid gave
a marked improvement in affinity (58, K = 830 nM). Addition of
D
9
an extra methylene unit between the sulfur atom and carboxylic
Table 5. N -methylated 8MG derivatives
acid group (57) reduced binding by 10-fold. Aryl substitution at
the methylene linker of 58 maintained affinity (59, K = 650 nM),
D
contrasting with the effect of adding a methyl substituent to the
methylene linker in the N-methyl acetamide-functionalized
compounds, 45 and 47. Crystal structures of EcDHPS in
complex with 58 and 59 were solved, revealing that the carbonyl
groups in the carboxylic acid moieties are positioned similarly to
the carbonyl groups within the 36- and 41-bound structures, and
that the benzyl group of 59 projects towards loop 2 (Figure 4). In
the 59-bound complex, the interacting loop 2 residues differ
between the monomers, with the key residues being T62 for
monomer A and R63/P64 for monomer B. The racemic form of
compd
R
K
D
EcDHPS (µM)
0.42 ± 0.04
6
1
2
6
0.39 ± 0.08
59 was synthesized and tested, however only the (R)-
enantiomer was present in the X-ray structure, suggesting that
the affinity reported in Table 4 is underestimated up to 2-fold.
The (S)-enantiomer would likely clash with residues F190 and
K221. The interactions involving the carboxylic acid groups in
these two complexes are not dissimilar to those observed for the
carboxylic acid-functionalized pterin analogues reported
previously by Zhao et al.[14]
6
3
4
19 ± 4
6
30 ± 10
9
As discussed above, methylation at the N position of the
8
MG core was hypothesized to improve binding for this
compound series. Two of the top compounds, 58 and 54, were
therefore methylated to give 61 and 62, respectively.
respective 2-fold and 5-fold improvement to binding was
observed, over their desmethyl equivalents, with K values of
20 nM and 390 nM (Table 5). Two 8MG-benzyl derivatives
A
D
4
9
were also N -methylated, however binding was marginally worse
than the majority of the 8MG-benzyl series (63, K
D
= 19 μM, and
9
64, K
D
= 30 μM). N -methylation of these inhibitors may,
therefore, not be generally applicable, or may be specific to the
8
S -functionalization present. A crystal structure of EcDHPS
solved in complex with 61 largely resembles the equivalent 58-
bound structure. In the 58-bound structure (Figure 4), the N9
nitrogen interacts with either D96 or a water molecule, in
monomer A or B, respectively. In the 61-bound EcDHPS (Figure
Figure 5. X-ray crystal structure of EcDHPS bound to (A) 61 or (B) 62.
Hydrogen bonds are displayed as dashed black lines.
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201704730
FULL PAPER
A crystal structure of EcDHPS solved in complex with 61
largely resembles the equivalent 58-bound structure. In the 58-
bound structure (Figure 4), the N nitrogen interacts with either
Relationship to the substrate envelope. Superposition of
the 11 X-ray structures onto the volume occupied by the
transition state bound structure[3] (green) is shown in Figure 7.
As anticipated, the pterin-like 8MG, the pteroic acid, and the
8MG-conjugate sit closely within the envelope as described by
the substrate ligands in the transition state structure. The
9
D96 or a water molecule, in monomer A or B, respectively. In the
6
1-bound structure (Figure 5), the N9 methyl group displaces
these interactions, with D96 interacting instead with D56. In the
2-bound EcDHPS crystal structure (Figure 5), D96 is likewise
6
acetophenone sidechain of compound 36 (K
out from the volume occupied by the envelope as does the
related but higher affinity 62 (K = 0.39 µM). In contrast, the
= 0.83 µM) resides
D
= 3.6 µM) projects
displaced, but is oriented away from D56, instead interacting
with residue T62. In the 62/EcDHPS complex, the amide
carbonyl of 62 interacts with R255 as per the previous structures, smaller and simpler acid derivative (58, K
D
D
while the N-phenyl moiety projects up to interact with residues
within the substrate envelope as expected based on its size.
R220, K221, and H257.
Notably, the larger phenyl analogue, compound 59, not only
D
maintained potency (K = 0.65 µM) but has good shape
The pteroic acid and sulfa conjugate complexes. To
further our understanding of the binding site envelope and key
interactions within EcDHPS, we crystallized and solved (2.04 Å)
the structure of EcDHPS complexed with pteroic acid, the
oxidized analogue of the catalysis product (Figure 6A). A single
pteroic acid molecule is bound in the active site for each
monomer, although the quality of electron density differs (Figure
S2). The binding of pteroic acid is similar to that observed in the
equivalent Yersinia pestis,[3] Bacillus anthracis,[7c] and Coxiella
burnetii[7i] complexes. The pterin moiety hydrogen bonds to
residues D96, N115, D185, G217, and K221 while the pABA
moiety interacts with the sidechains of F190 (tyrosine in C.
burnetii) and K221, and the carboxylate hydrogen bonds with the
backbone and sidechain of S222 (Figure 6).
complementarity to the substrate envelope and suitably
positioned for further analoguing to engage the phosphate
subsite.
Figure 7. Two views showing the positions of the 8MG analogues as observed in the
EcDHPS X-ray structure relative to the substrate (green) and the envelope in the
transition state structure of DHPS from Y. pestis (PDB: 3TYZ). A single magnesium
atom (lime green) and a sulfate (orange) are shown from the pteroic acid product.
Two compounds (62, 36, pink) project outside the substrate envelope whereas 59
(
cyan) is confined largely within the volume occupied by the substrate.
Dual DHPS/HPPK binding. Consultation of Tables 3 and
4
indicates that the acetophenone series of 8MG analogues,
such as 36 and 37, offers robust (sub micromolar) binding to
both HPPK from E. coli and S. aureus.[
15c]
Previously we showed
Figure 6. X-ray crystal structure of EcDHPS bound to (A) pteroic acid or (B) 60.
Both conformations of R63 in the 60/EcDHPS complex are displayed.
Hydrogen bonds to S222 are displayed as dashed black lines.
that the mode of binding of this series to the S. aureus HPPK
enzyme was very similar to that observed in the E. coli enzyme
accounting for the remarkably similar affinity.[15c] In this, the p-
methoxy or p-bromo group projected out of the active site pocket
into the solvent and therefore further extension from this position
had little effect on affinity. In comparison, the benzyl series of
8MG analogs bound far tighter to SaHPPK by inducing a cryptic
binding pocket, specific to the S. aureus enzyme. Combined with
the structural data presented here, future structure based design
of the acetophenone series may therefore have the potential for
dual E. coli HPPK/DHPS inhibition. Alternatively, 59 shows tight
binding to EcDHPS and good binding to SaHPPK, with a 100-
fold selectivity to SaHPPK over EcHPPK.[15d] As mentioned
earlier, sequence and structural conservation over the pterin site
in DHPS is high and a superposition of the EcDHPS/59 complex
with the deposited SaDHPS structure (Midwest Centre for
Structural Genomics, to be published) shows essentially
Recently, a series of pterin-sulfa conjugates were reported,
with one of these crystallized in the binding site of Yersinia
pestis DHPS.[19] To probe the potential for 8MG to act as a core
from which to develop equivalent sulfa conjugates, we used the
above pteroic acid/EcDHPS complex as a guide for our design.
Compound 60 (K
between 8MG and
D
= 4.6 μM), incorporating an ethylene linker
p-phenylsulfonamide moiety, was
a
synthesized and then crystallized with EcDHPS. This compound
was found to bind as predicted — the phenyl ring forms
hydrophobic interactions with F190 and K221, in addition to loop
2
residues, while the sulfonamide group hydrogen bonds to the
backbone and sidechain S222 (Figure 6B). Strong electron
density is observed for the entire ligand.
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201704730
FULL PAPER
equivalent intermolecular interactions (Figure 8). For example
the position of R255 is exactly superimposable and would make
the key interaction between the guanidinium group and to the
carboxylic acid of 59. Notably, as 59 is also confined within the
substrate envelope of DHPS, further structure based inhibitor
design of SaHPPK in parallel may produce dual enzyme,
envelope-compliant, pterin-site inhibitors in S. aureus.
interactions into the pyrophosphate subsite, the 8MG core may
have to be synthetically revisited and either modified (or scaffold
hopped) in order to reduce its inherently large polar surface area
2
(tPSA ~100 Å ) to improve permeability. Given that a pterin-site,
substrate envelope-compliant DHPS inhibitor will reduce the
growth of wild-type and sulfonamide-resistant pathogenic strains,
be refractory to point mutation-based resistance mechanisms,
and potentially target a network of other folate synthesis
enzymes, our 8MG analogues present interesting scaffolds
towards targeting the pterin site of this well-validated target.
Experimental Section
Protein preparation, differential scanning fluorimetry, crystallization and
deposited PDB ID codes, SPR, chemical synthesis and characterization
are reported in the supplementary information.
Acknowledgements
All crystals were grown at the C3 Crystallisation Centre at
CSIRO, Parkville, Australia and X-ray data were obtained at the
MX1 and MX2 beamlines at the Australian Synchrotron, Victoria,
Australia. We thank OpenEye Scientific Software for a license to
use their software. We thank Janet Newman for help with
crystallization, and Shane Seabrook for help with DSF. JBB was
supported by an NHMRC Research Fellowship 1020411.
Figure 8. Superposition of 59/EcDHPS (cyan) and SaDHPS (grey)
(PDB:4HB7) showing the close sequence and structural similarity in the active
site region. Selected hydrogen bonds are shown in pink and equivalent
binding site residues are labelled for EcDHPS.
Keywords: DHPS • inhibitor • pterin • antibiotic • drug discovery.
Conclusions
[
1] a) C. Manyando, E. M. Njunju, U. D'Alessandro and J.-P. Van Geertruyden,
8
MG is a pterin-like fragment that binds DHPS projecting a thiol
PLoS One 2013, 8, e56916; b) P. Forgacs, N. L. Wengenack, L. Hall, S. K.
Zimmerman, M. L. Silverman and G. D. Roberts, Antimicrob. Agents
Chemother. 2009, 53, 4789‒ 4793; c) S. I. Martin and J. A. Fishman, Am. J.
Transplant. 2013, 13, 272–279; d) M. L. Campbell, D. Marchaim, J. M. Pogue,
B. Sunkara, S. Bheemreddy, P. Bathina, H. Pulluru, N. Chugh, M. N. Wilson, J.
Moshos, K. Ku, K. Hayakawa, E. T. Martin, P. R. Lephart, M. J. Rybak and K.
S. Kaye, Ann. Pharmacother. 2012, 46, 1587‒ 1597.
group as a convenient reaction handle for structural elaboration,
allowing for the efficient generation of derivatives. We have
presented and structurally characterized a number of novel, S -
functionalized 8MG derivatives that possess sub-micromolar
affinities for EcDHPS constituting a gain of over two orders of
magnitude in potency while maintaining an admirable ligand
efficiency (0.5 kcal·mol·heavy atom ) of the parent.
Characterizing the structure of tight binding novel compounds to
the pterin site is an important aspect of our approach and our
efforts provide a detailed structural framework to guide further
studies aimed at increasing affinity whilst guiding substrate
envelope compliance. While structural data revealed that some
of our better compounds projected out from the envelope,
gratifyingly some of our best compounds reside within the
volume occupied by the substrate. These offer considerable
promise for further rational elaboration into DHPS inhibitors in a
range of species with the potential for dual HPPK/DHPS activity.
While we have observed encouraging inhibition in an E. coli K-
8
-1
[
2] a) D. D. Woods, Br. J. Exp. Pathol. 1940, 21, 74‒ 90; b) G. M. Brown, J.
Biol. Chem. 1962, 237, 536‒ 540; c) S. Roland, R. Ferone, R. J. Harvey, V. L.
Styles and R. W. Morrison, J. Biol. Chem. 1979, 254, 10337‒ 10345; d) G.
Swedberg, S. Castensson and O. Sköld, J. Bacteriol. 1979, 137, 129‒ 136; e)
O. Patel, J. Satchell, J. Baell, R. Fernley, P. Coloe and I. Macreadie, Microb.
Drug Resist. 2003, 9, 139‒ 146.
[
3] M. K. Yun, Y. Wu, Z. Li, Y. Zhao, M. B. Waddell, A. M. Ferreira, R. E. Lee,
D. Bashford and S. W. White, Science 2012, 335, 1110‒ 1114.
[4] H. Haruki, M. G. Pedersen, K. I. Gorska, F. Pojer and K. Johnsson,
Science 2013, 340, 987–991.
[5] a) S. Mondal, S. M. Mandal, T. K. Mondal and C. Sinha, Spectrochim. Acta
A Mol. Biomol. Spectrosc. 2015, 150, 268–279; b) W. Boufas, N. Dupont, M.
Berredjem, K. Berrezag, I. Becheker, H. Berredjem and N.-E. Aouf, Journal of
Molecular Structure 2014, 1074, 180–185.
12
minimal media based assay[20] (data not shown),
antimicrobial activity was achieved in the presence of sub-MIC
levels of colistin acting as a membrane-permeating/depolarizing
agent. This is not unexpected given the polar nature of the
guanine core and the tight binding of the substrate. As our
compounds reach comparable potency to the substrate through
[
6] O. Sköld, Drug Resist. Updat. 2000, 3, 155–160.
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201704730
FULL PAPER
[
7] a) A. Achari, D. O. Somers, J. N. Champness, P. K. Bryant, J. Rosemond
[12] a) O. J. Lever, L. Bell, H. McGuire and R. Ferone, J. Med. Chem. 1985,
28, 1870–1874; b) O. W. J. Lever, L. N. Bell, C. Hyman, H. M. McGuire and R.
Ferone, J. Med. Chem. 1986, 29, 665–670.
and D. K. Stammers, Nat. Struct. Biol. 1997, 4, 490–497; b) I. C. Hampele, A.
D'Arcy, G. E. Dale, D. Kostrewa, J. Nielsen, C. Oefner, M. G. Page, H. J.
Schönfeld, D. Stüber and R. L. Then, J. Mol. Biol. 1997, 268, 21–30; c) K.
Babaoglu, J. Qi, R. E. Lee and S. W. White, Structure 2004, 12, 1705–1717;
d) C. W. t. Pemble, P. K. Mehta, S. Mehra, Z. Li, A. Nourse, R. E. Lee and S.
W. White, PLoS One 2010, 5, e14165; e) A. M. Baca, R. Sirawaraporn, S.
Turley, W. Sirawaraporn and W. G. J. Hol, J. Mol. Biol. 2000, 302, 1193–1212;
f) B. Bagautdinov and N. Kunishima, RIKEN Structural Genomics/Proteomics
Initiative doi:10.2210/pdb2212dqw/pdb; g) C. Levy, D. Minnis and J. P. Derrick,
Biochem. J. 2008, 412, 379–388; h) R. E. Morgan, G. O. Batot, J. M. Dement,
V. A. Rao, T. C. Eadsforth and W. N. Hunter, BMC Struct. Biol. 2011, 11, 1–9;
i) M. C. Franklin, J. Cheung, M. J. Rudolph, F. Burshteyn, M. Cassidy, E. Gary,
B. Hillerich, Z. K. Yao, P. R. Carlier, M. Totrov and J. D. Love, Proteins: Struct.,
Funct., Bioinf. 2015, 83, 2124–2136; j) M. C. Lawrence, P. Iliades, R. T.
Fernley, J. Berglez, P. A. Pilling and I. G. Macreadie, J. Mol. Biol. 2005, 348,
[13] K. E. Hevener, M. K. Yun, J. Qi, I. D. Kerr, K. Babaoglu, J. G. Hurdle, K.
Balakrishna, S. W. White and R. E. Lee, J. Med. Chem. 2010, 53, 166–177.
[14] Y. Zhao, D. Hammoudeh, M. K. Yun, J. Qi, S. W. White and R. E. Lee,
ChemMedChem 2012, 7, 861–870.
[15] a) S. Chhabra, O. Dolezal, B. M. Collins, J. Newman, J. S. Simpson, I. G.
Macreadie, R. Fernley, T. S. Peat and J. D. Swarbrick, PLoS One 2012, 7,
e29444; b) S. Chhabra, N. Barlow, O. Dolezal, M. K. Hattarki, J. Newman, T. S.
Peat, B. Graham and J. D. Swarbrick, PLoS One 2013, 8, e59535; c) M. L.
Dennis, S. Chhabra, Z.-C. Wang, A. Debono, O. Dolezal, J. Newman, N. P.
Pitcher, R. Rahmani, B. Cleary, N. Barlow, M. Hattarki, B. Graham, T. S. Peat,
J. B. Baell and J. D. Swarbrick, J. Med. Chem. 2014, 57, 9612–9626; d) M. L.
Dennis, N. P. Pitcher, M. D. Lee, A. J. DeBono, Z.-C. Wang, J. R. Harjani, R.
Rahmani, B. Cleary, T. S. Peat, J. B. Baell and J. D. Swarbrick, J. Med. Chem.
2016, 59, 5248–5263.
6
55–670.
8] D. I. Hammoudeh, Y. Zhao, S. W. White and R. Lee, Future Med. Chem.
013, 5, 1331–1340.
9] Y. Shen, M. D. Altman, A. Ali, M. N. L. Nalam, H. Cao, T. M. Rana, C. A.
Schiffer and B. Tidor, ACS Chem. Biol. 2013, 8, 2433–2441.
10] D. Hammoudeh, M. Date, M. Yun, W. Zhang, V. Boyd, A. Viacava Follis,
E. Griffith, R. Lee, D. Bashford and S. White, ACS Chem. Biol. 2014,
294−1302.
11] R. W. Morrison, W. R. Mallory and V. L. Styles, J. Org. Chem. 1978, 43,
844–4849.
[
[16] C. R. Bourne, Antibiotics 2014, 3, 1–28.
2
[17] S. A. Seabrook and J. Newman, ACS Comb. Sci. 2013, 15, 387–392.
[18] W. S. Dallas, J. E. Gowen, P. H. Ray, M. J. Cox and I. K. Dev, J. Bacteriol.
1992, 174, 5961–5970.
[
[
[19] Y. Zhao, W. R. Shadrick, M. J. Wallace, Y. Wu, E. C. Griffith, J. Qi, M.-K.
Yun, S. W. White and R. E. Lee, Bioorg. Med. Chem. Lett. 2016, 26, 3950–
3954.
1
[
[20] S. Zlitni, L. F. Ferruccio, E. D. Brown. Nat. Chem. Biol. 2013, 9, 796–804.
4
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201704730
FULL PAPER
FULL PAPER
M. L. Dennis, M. D. Lee, J. R. Harjani, M.
Ahmed, A. J. DeBono, N. P. Pitcher, Z-C
Wang, S. Chhabra, N. Barlow, R.
Rahmani, B. Cleary, O. Dolezal, M.
Hattarki, L. Aurelio, J. Shonberg, B.
Graham, T. S. Peat, J. B. Baell and J. D.
Swarbrick*
Within the envelope: Structure base evolution of 8-mercaptoguanine derivatives leads
to tight binding to the antibiotic target DHPS and substrate envelope compliance.
8
-Mercaptoguanine derivatives as
inhibitors of dihydropteroate synthase
This article is protected by copyright. All rights reserved.