Identification of indole inhibitors of human hematopoietic prostaglandin D2 synthase (hH-PGDS)

Article abstract of DOI:10.1016/j.bmcl.2015.04.065

Abstract Human H-PGDS has shown promise as a potential target for anti-allergic and anti-inflammatory drugs. Here we describe the discovery of a novel class of indole inhibitors, identified through focused screening of 42,000 compounds and evaluated using a series of hit validation assays that included fluorescence polarization binding, 1D NMR, ITC and chromogenic enzymatic assays. Compounds with low nanomolar potency, favorable physico-chemical properties and inhibitory activity in human mast cells have been identified. In addition, our studies suggest that the active site of hH-PGDS can accommodate larger structural diversity than previously thought, such as the introduction of polar groups in the inner part of the binding pocket.

Full text of DOI:10.1016/j.bmcl.2015.04.065

Bioorganic & Medicinal Chemistry Letters 25 (2015) 2496–2500
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Identification of indole inhibitors of human hematopoietic
prostaglandin D synthase (hH-PGDS)
2
Fredrik Edfeldt a, , Johan Evenäs , Matti Lepistö , Alison Ward , Jens Petersen , Lisa Wissler ,
⇑
b
b
b
a
a
a
b
b
b
a
b
Mattias Rohman , Ulf Sivars , Karin Svensson , Matthew Perry , Isabella Feierberg , Xiao-Hong Zhou ,
b
b
Thomas Hansson , Frank Narjes
a
Discovery Sciences, Innovative Medicines, AstraZeneca R&D, 431 83 Molndal, Sweden
Respiratory, Inflammation and Autoimmunity, Innovative Medicines, AstraZeneca R&D, 431 83 Molndal, Sweden
b
a r t i c l e i n f o
a b s t r a c t
Article history:
Human H-PGDS has shown promise as a potential target for anti-allergic and anti-inflammatory drugs.
Here we describe the discovery of a novel class of indole inhibitors, identified through focused screening
of 42,000 compounds and evaluated using a series of hit validation assays that included fluorescence
polarization binding, 1D NMR, ITC and chromogenic enzymatic assays. Compounds with low nanomolar
potency, favorable physico-chemical properties and inhibitory activity in human mast cells have been
identified. In addition, our studies suggest that the active site of hH-PGDS can accommodate larger struc-
tural diversity than previously thought, such as the introduction of polar groups in the inner part of the
binding pocket.
Received 10 March 2015
Revised 19 April 2015
Accepted 20 April 2015
Available online 28 April 2015
Keywords:
Prostaglandin D
PGDS inhibitors
Indole
2
synthase
Ó 2015 Elsevier Ltd. All rights reserved.
Focused screening
Hit validation
Prostaglandin D
2
(PGD
2
) is an allergic and inflammatory medi-
and more ligand efficient hits. A fluorescence polarization assay10
was used to screen two library subsets of 25,000 and 17,000 com-
ator produced by mast cells and Th2 cells after cross-linking of
an allergen with the specific IgE antibody on the cell membrane.
1
pounds at 125
determined for compounds showing a >30% displacement effect,
resulting in 1040 compounds with IC50 values <100 M. To further
validate the actives a 1D NMR binding assay was deployed. K val-
ues were determined using the competitive displacement of repor-
lM and 250 lM, respectively. IC50 values were
PGD
that have been associated with inflammatory conditions.
Production of PGD in the peripheral tissues and in immune and
inflammatory cells from the precursor prostaglandin H (PGH ) is
2 1
acts on two G-protein-coupled receptors, DP and CRTH2
2
l
2
d
2
2
1
1
mainly catalyzed by human hematopoietic prostaglandin D syn-
thase (hH-PGDS). This has led hH-PGDS to be envisioned as a target
for the treatment of asthma and inflammatory diseases. hH-PGDS
is a 26-kDa cytosolic homodimer of the sigma class glutathione-S-
transferase (GST) family. It depends on glutathione (GSH) for cat-
alytic activity, which is further increased by divalent metal ions.
ter ligands with known affinity.
The hits were clustered using structural similarity and 216 clus-
ter representatives were selected for testing in the NMR assay. 187
of these hits were confirmed in the NMR assay, corresponding to a
validation rate of 87%. Validated hits were then tested in a GST
enzymatic assay using CDNB (1-chloro-2,4-dinitrobenzene) as
3
4
1
2
Several inhibitors of hH-PGDS, most of them containing a signature
bis-aryl amide motif, have been disclosed in recent years by a
chromogenic substrate and IC50 values were generated. We chose
to monitor the GST enzymatic activity of hH-PGDS rather than its
5
number of organizations, such as (1 and 2) Pfizer, (3) Sanofi-
PGD
2
synthase activity due to technical feasibility. Both enzymatic
Aventis,6 (4) Evotec, (5) AstraZeneca8 and (6) Taiho9 (Fig. 1). For
7
activities reside in the same binding pocket.
4b,13
For a set of four
a recent review of hH-PGDS inhibitors, see Ref. 3.
reference compounds (1–3, and 5) activity in the NMR and GST
Our previous efforts led to the identification of compound (5) as
an hH-PGDS inhibitor. The current study reports our continued
work on identifying novel inhibitors of this enzyme applying a
fragment based screening approach, aiming to identify smaller
assays, as well as isothermal titration calorimetry (ITC),14 corre-
lated overall well with the inhibition of PGD
2
production in
1
5
megakaryoblast cells (Table 1). We therefore felt confident that
our screening set-up was fit for purpose in driving hit expansion
chemistry in a high throughput manner. We observed excellent
correlation between the NMR binding assay and inhibition of GST
activity for our screening hits. 168 compounds had an IC50
⇑
Corresponding author. Tel.: +46 31 776 1604; fax: +46 31 776 3792.
E-mail address: Fredrik.Edfeldt@astrazeneca.com (F. Edfeldt).
http://dx.doi.org/10.1016/j.bmcl.2015.04.065
960-894X/Ó 2015 Elsevier Ltd. All rights reserved.
0

F. Edfeldt et al. / Bioorg. Med. Chem. Lett. 25 (2015) 2496–2500
2497
F
O
H
N
O
N
N
O
H
N
H
N
N
N
HN
HN
N
N
N
N
O
N
O
O
CF3
N
CF3
N
2
4
7
1
8
OH
O
HN
N
N
HN
O
N O
O
N
N
O
S
9
O
SO2NH2
N
N
N
Figure 2. Initial screening hits containing the indole motif.
3
N
N
N
O
O
S
N
N
O
N
5
N
6
Figure 1. Examples of published inhibitors of hH-PDGS.
<10
l
d
M, in good agreement with the NMR K values. We subse-
quently used binding (NMR) and enzymatic activity (GST) in com-
bination to generate compound SAR.
Known chemical motifs, such as aryl amides, were well repre-
sented among the hits, but a range of more novel structures were
also identified. We found compounds containing an indole motif to
be of particular interest (7, 8 and 9, Fig. 2). This class of compounds
had reasonable physico-chemical properties in terms of calculated
logP values (clogP), ligand efficiency (LE), ligand lipophilic effi-
ciency (LLE) and solubility, and was therefore considered a promis-
ing starting point (Table 1). Compound 8 was tested for inhibition
Figure 3. Binding mode of indole hits 7 (brown) and 9 (green) overlaid with
compound 2 (cyan) in complex with hH-PGDS, showing the conserved water, and
key residues Trp104 and Tyr152 (pink). X-ray structure coordinates have been
deposited in the RSCB Protein Data Bank (PDB) with accession codes 5AIX (2), 5AIS
(
7) and 5AIV (9). Crystallization protocol as previously reported.8
2
cellular PGD synthase activity and gave an EC50 value of 0.35 lM.
We then proceeded to determine the crystal structures with
compounds 7, 9, and the reference 2 bound to hH-PGDS, which
(
in 7) and oxadiazole (in 9) rings make the critical stacking interac-
tion with Trp104 and the important hydrogen-bonding interaction
with the conserved water molecule previously described.5
,8,19
In
were obtained by co-crystallization at a resolution of 1.8, 2.0 and
8
2
.1 Å, respectively.1 We were unable to determine the structure
both 7 and 9 the indole binds in the inner cavity, making a hydro-
gen bond interaction with the hydroxyl group of Tyr152, which to
our knowledge has not been observed before. The amine and ether
moieties both protrude out of the binding pocket making no criti-
cal interactions. These structural features convinced us to further
explore the structure activity relationship (SAR) of this series by
synthesizing an amide library of related analogs.
A set of 20 analogs were synthesized and tested in the NMR
binding and GST enzymatic assays; the most relevant examples
are shown in Table 2. Several compounds gave significantly
improved IC50 values while also improving clogP, LE and LLE. The
with 8, but the binding mode is expected to be similar to 7.
Several crystal structures of inhibitors in complex with hH-
PGDS have been described.5
,8,19
The substrate pocket can be
subdivided into three different regions: the inner cavity, typically
occupied by a phenyl residue as evidenced in structures 1–6; the
central cavity, typically occupied by a heterocycle residue, that
stacks on Trp104, whereas the amide residue can be involved in
interactions with Arg14 and/or the thiolate anion of bound GSH;
and the peripheral solvent exposed part of the pocket, where the
9
largest structural variations of inhibitors have been observed.
The overall binding mode of indoles 7 and 9 resembles that of
compound 2, with the indole motif overlapping with the phenyl
ring of 2 (Fig. 3). The amide of 7 is reversed relative to 2 and 9.
This moiety makes no apparent interactions, but the reverse amide
provides a different way to extend compounds. Both the pyridine
most potent compound (10) had an IC50 of 0.026
l
M, a 7-fold
improvement over 7. A large variety of functional R-groups were
tolerated, consistent with SAR observed in the literature.5
–7,9
This
part of the molecule can be utilized to modulate physico-chemical
properties, although modest affinity gains were also made. For
Table 1
FP IC50a
(
lM)
NMR K
a
(l
M)
ITC K
(lM)
GST IC50a
(l
M)
Cell EC50a
(l
M)
cLogPb
Solubilityc
(lM)
LEd
LLEe
Compd
d
d
1
2
3
5
7
8
9
—
0.21
—
0.0095
0.044
0.023
0.051
0.57
0.020
0.012
0.035
0.30
—
0.20
0.035
0.13
0.074
0.11
—
3.0
2.9
1.3
3.2
3.0
1.7
2.1
2.6
4.2
—
1.4
84
95
77
0.34
0.35
0.31
0.65
0.38
0.37
0.42
3.7
4.3
4.9
4.4
3.7
4.8
4.7
0.060
0.070
0.023
0.18
0.30
0.17
—
0.82
0.46
2.9
0.34
0.65
0.21
—
0.35
—
a
b
c
Values are the mean of at least two independent experiments.
Calculated logP.
Measured solubility.
Ligand efficiency in kcal per mol per heavy atom.
d
e
16
17
Ligand lipophilic efficiency defined as pIC50-clogP, calculated using GST IC50
.

2
498
F. Edfeldt et al. / Bioorg. Med. Chem. Lett. 25 (2015) 2496–2500
Table 2
Table 3
N
O
H
N
R
CF3
R
HN
N
HN
O
N
a
GST IC50a
M)
cLogP LEb
LLEb
Compd
R
NMR K
d
a
GST IC50a
M)
cLogP LEb
LLEb
Compd
R
NMR K
M)
d
(l
M)
(l
(
l
(l
O
∗
∗
N
O
2
0.076
0.06
0.060
0.031
2.9
3.1
0.35 4.3
7
8
0.57
0.34
0.18
0.30
3.0
1.7
0.38 3.7
0.37 4.8
∗
N
1
6
∗
0.38 4.5
O
1
1
0
1
0.05
0.06
0.026
0.032
3.0
2.9
0.40 4.6
0.51 4.6
N
∗
N
17
<0.1
0.2
14
0.048
0.23
—
2.4
1.8
1.5
1.0
0.95
0.37 5.0
0.36 4.8
∗
∗
1
1
8
9
N
∗
H
N
1
1
1
2
3
4
∗
O
0.3
0.2
1
0.035
0.043
0.13
1.4
1.9
3.9
0.42 6.1
0.42 5.5
0.36 3.0
N
∗
—
—
N
∗
N
N
20
O
N
∗
0.4
13
0.32
—
0.34 5.5
N
2
2
1
2
HN
N
∗
—
—
N
∗
∗
∗
HN
O
N
1.0
0.2
0.59
0.55
0.5
0.32 5.7
0.38 6.6
N
1
5
<20
1.2
1.8
0.32 4.2
HN
∗
N
2
2
3
4
0.070
a
O
N
Values are the mean of at least two independent experiments.
Calculated using GST IC50
b
.
∗
2
—
2.3
—
—
O
instance, the lactam 12 and tetrazole 13 showed significant
improvement in LLE. Compound 11 represents the highly ligand
efficient minimal scaffold for the series with an IC50 of 0.032 lM
and LE of 0.51. This is roughly equipotent to our previously
reported compound 5, but it represents a more attractive starting
point for further optimization.
In parallel to these screening and chemistry efforts we pursued
a structure-based design approach to further probe the binding
pocket of hH-PGDS. The indole motif suggested that a larger vari-
ety of functional groups than previously thought could be tolerated
in the inner binding pocket. To further explore this hypothesis a
series of compounds related to 1 and 2 were synthesized. A variety
of functional groups pointing into the inner cavity were introduced
whilst keeping the rest of the molecule constant. We also included
polar residues, which we envisioned would be able to engage with
Tyr152, and result in overall good lipophilic ligand efficiency.
Compounds were pre-screened in the NMR binding assay, and
∗
2
2
5
6
NH
0.4
10
0.071
—
0.7
0.5
0.39 6.4
O
O
∗
—
—
HN
a
Values are the mean of at least two independent experiments.
Calculated using GST IC50
b
.
investigation around this motif revealed that the related pyridones
3 and 25 showed greatly improved activity, similar to compound
, and due to the lower cLogP a significant increase in LLE.
2
2
Regioisomer 26 on the other hand was not active, whilst methoxy-
pyridine 24 resulted in a 10-fold loss in potency compared to 23,
pointing to a critical role of the pyridone oxygen for activity. Our
attempts to co-crystallize 23 with hH-PGDS were unsuccessful.
To better understand the SAR around the novel motifs in the
inner pocket, we docked the compounds into the available X-ray
structures of hH-PGDS. hH-PGDS functions as a homodimer and
the apo-enzyme is activated by Mg and Ca ions, which make
multiple cross-linking interactions at the homodimer interface
coordinating clusters of acidic residues from both monomers.
Structures co-crystallized with inhibitors tend to be asymmetric
d
those with a K <1 lM were then further profiled in the GST assay
(Table 3). All analogs shown in Table 3 proved to be competitive
with respect to our reporter ligand. We used 2 for benchmarking,
since it gave more consistent data in our hands and we had
obtained a co-crystal structure.
2+
2+
The methoxyphenyl group in 2 could not only be replaced with
a phenoxy group as in 16 (IC50 = 0.031
pound 6, but also with saturated rings such as piperidine (17,
IC50 = 0.048 M). This resulted in a slight increase in potency and
an increase in LLE, due to reduction of c LogP. Further SAR around
7 revealed that a pyrrolidine (18, IC50 = 0.23 M) was still toler-
l
M), found also in com-
2+
2+
compared to the Mg /Ca -complexed apo-structures, often with
partial or no occupancy of ligand in the second pocket as well as
a more disordered GSH. This asymmetry results in two slightly dif-
ferent kinds of inner cavities, where the Tyr152 OH can act either
l
1
l
20
as an acceptor or as a donor. The overlay of pyridones 23 and 25
ated, whereas further reduction to a dimethylamino group (19)
led to a substantial loss in binding. Introduction of a morpholine
residue (20) further increased LLE, despite a 10-fold loss in potency
with respect to 17, whereas a piperazine (21) was not tolerated,
presumably due to the charged amine. Interestingly, the more
polar, but neutral piperazinone 22 showed improved binding in
the NMR assay and was also active in the GST assay. Further
with indole 7 is shown in Figure 4. Both 23 and 25 appear to form a
hydrogen bonding interaction with Tyr152 via the carbonyl oxygen
rather than the NH, as seen for 7. The pyridinone 23 may also act as
a donor, perhaps explaining its slightly higher potency versus 25.
The 25 to 50-fold loss in binding potency of pyridone 26 could
be explained by the less favorable positioning of the carbonyl oxy-
gen with respect to 23. Similarly, the loss in potency for the

F. Edfeldt et al. / Bioorg. Med. Chem. Lett. 25 (2015) 2496–2500
2499
Compounds 16–22 were obtained by nucleophilic substitution
of the 2-chloro-pyridyl carboxamide 29 with the appropriate phe-
nol or amine. A Suzuki coupling between 29 and the appropriate
methoxypyridyl boronate followed by deprotection with boron
tribromide gave pyridines 23–26 (Scheme 2).
Acknowledgments
The authors would like to acknowledge BioDuro (Beijing, China)
for assistance in chemical synthesis and Ewa Nilsson for the pro-
duction of recombinant protein.
References and notes
1.
(a) Lewis, R. A.; Soter, N. A.; Diamond, P. T.; Austen, F.; Oates, J. A.; Roberts, L. J.
J. Immunol. 1982, 129, 1627; (b) Tanaka, K.; Ogawa, K.; Sugamura, K.;
Nakamura, M.; Takano, S.; Nagata, K. J. Immunol. 2000, 164, 2277.
Figure 4. Alternative hydrogen bond network induced by indole versus pyridone
substituents. The neighboring dimer subunit is indicated by the light-blue surface.
According to modeling the electrostatics of the binding pocket and dimer
2
.
(a) Nagai, H. Allergology 2008, 57, 187; (b) Matsuoka, T.; Hirata, M.; Tanaka, H.;
Takahashi, Y.; Murata, T.; Kabashima, K.; Sugimoto, Y.; Kobayashi, T.; Ushikubi,
F.; Aze, Y.; Eguchi, N.; Urade, Y.; Yoshida, N.; Kimura, K.; Mizoguchi, A.; Honda,
Y.; Nagai, H.; Narumiya, S. Science 2000, 287, 2013.
20
2+
interface, the Mg is coordinated by Asp96 and Asp97 from one subunit with
À
2+
2
OH /H O bridging the Mg to Asp97 of the other subunit. This may then induce
alternative side chain conformations for Arg14 and Asp96 and hydrogen bond
arrangements involving Tyr152, which makes direct interactions with the
inhibitors. This allows the phenolic group to act as an acceptor for indoles like
compound 7 (light gray; Tyr152 in light blue) or a donor for pyridones 23 (blue;
Tyr152 in brown) and 25 (green).
3
4
.
.
Thurairatnam, S. Prog. Med. Chem. 2012, 51, 97.
(a) Meyer, T. J.; Thomas, M. Biochem. J. 1995, 311, 739; (b) Kanaoka, Y.; Ago, H.;
Inagaki, E.; Nanayama, T.; Miyano, M.; Kikuno, R.; Fujii, Y.; Eguchi, N.; Toh, H.;
Urade, Y.; Hayaishi, O. Cell 1997, 90, 1085; (c) Inoue, T.; Irikura, D.; Okazaki, N.;
Kinugasa, S.; Matsumura, H.; Uodome, N.; Yamamoto, M.; Kumasaka, T.;
Miyano, M.; Kai, Y.; Urade, Y. Nat. Struct. Biol. 2003, 10, 291. Corrigendum ibid,
1
0, 409.
5.
(a) Carron, C. P.; Trujillo, J. I.; Olson, K. L.; Huang, W.; Hamper, B. C.; Dice, T.;
Neal, B. E.; Pelc, M. J.; Day, J. E.; Rohrer, D. C.; Kiefer, J. R.; Moon, J. B.;
Schweitzer, B. A.; Blake, T. D.; Turner, S. R.; Woerndle, R.; Case, B. L.; Bono, C. P.;
Dilworth, V. M.; Funckes-Shippy, C. L.; Hood, B. L.; Jerome, G. M.; Kornmeier, C.
M.; Radabaugh, M. R.; Williams, M. L.; Davies, M. S.; Wegner, C. D.; Welsch, D.
J.; Abraham, W. M.; Warren, C. J.; Dowty, M. E.; Hua, F.; Zutshi, A.; Yang, J. Z.;
Thorarensen, A. Med. Chem. Lett. 2010, 1, 59; (b) Blake, T.; Hamper, B. C.; Huang,
W.; Kiefer, J. R.; Moon, J. B.; Neal, B. E.; Olson, K. L.; Pelc, M. J.; Schweitzer, B. A.;
Thorarensen, A.; Trujillo, J. I.; Turner, S. R. US 2008/0146569.
H
N
a, b
R1
NH
2
c
Br
HN
HN
HN
O
N
N
2
7
28
7, 10-15
Scheme 1. Synthesis of indole-amide library. Reagents: (a) B
2
(pin)
2
, (Ph
3
P)
4
Pd,
2
KOAc, dioxane, water (10:1), 90 °C, 86%; (b) 3-amino-5-bromopyridine, Pd(dppf)Cl ,
6.
(a) Li, Z.; Xiong, J.; Sabol, J. S. WO 2004/016223; (b) Cao, Y.; Sabol, J. S.; Ayers, S.
WO 2006/015195; (c) Aldous, S. C.; Jiang, J. Z.; Lu, J.; Ma, L.; Mu, L.; Munson, H.
R.; Sabol, J. S.; Thurairatnam, S.; Vandeusen, C. L. WO 2007/041634; (d) Aldous,
S. C.; Fennie, M. W.; Jiang, J. Z.; John, S.; Mu, L.; Pedgrift, B.; Pribish, J. R.;
Rauckman, B.; Sabol, J. S.; Stoklosa, G. T.; Thurairatnam, S.; Vandeusen, C. L. WO
1
Na
2
CO
3
, dioxane, water (10:1), 90 °C, 81%; (c) R CO
2
H, POCl
3
, À20 °C, add 27, then
to RT; 5-40%.
2
008/121670; (e) Aldous, S. C.; Fennie, M. W.; Jiang, J. Z.; John, S.; Pedgrift, B.;
N
O
N
O
Pribish, J. R.; Rauckman, B.; Sabol, J. S.; Stoklosa, G. T.; Thurairatnam, S.;
Vandeusen, C. L. WO 2008/121670; (f) VanDeusen, C. L.; Weiberth, F. J.; Gill, H.
S.; Lee, G.; Hillegass, A. WO 2011/044307; (g) Weiberth, F. J.; Yu, Y.;
Subotkowski, W.; Pemberton, C. Org. Proc. Res. Dev. 2012, 16, 1967.
Cl
CF
3
a or b
X
CF
3
HN
N
R
HN
N
2
9
16
X = O
1
7-22 X = N
7. (a) Hesterkamp, T.; Barker, J.; Davenport, A.; Whittaker, M. Curr. Top. Med.
Chem. 2007, 7, 1582; (b) Aicher, B.; Kelter, A. R.; Coulter, T. S.; Taylor, S.;
Davenport, A. J.; Hesterkamp, T.; Kirchhoff, C. WO 2008/122787.
Br
B
8
.
Hohwy, M.; Spadola, L.; Lundquist, B.; Hawtin, P.; Dahmén, J.; Groth-Clausen, I.;
Nilsson, E.; Persdotter, S.; von Wachenfeldt, K.; Folmer, R. H. A.; Edman, K. J.
Med. Chem. 2008, 51, 2178.
(a) Urade, Y.; Tanaka, Y.; Yamane, K.; Togawa, M. WO 2007/007778; (b) Urade,
Y.; Shigeno, K.; Tanaka, Y.; Kuze, J.; Tsuchikawa, M.; Hosoya, T. JP 2007051121;
c
d
2
3-26
N
OMe
N
OMe
9.
Scheme 2. Synthetic scheme for inner cavity library. Reagents: (a) ROH, NaH, DMF,
microwave, 150 °C; (b) RNH, DMF, K CO , 80 °C; (c) B (pin) Pd(dppf)Cl , KOAc,
DMF, 80 °C; (d) 28, Pd(dppf)Cl , Na CO , dioxane, water (10:1), 80 °C; BBr , DCM,
°C.
(
c) Yamane, K.; Tanaka, Y.; Shigeno, K.; Hosoya, T.; Inoue, S.; Kitade, M.; Harada,
2
3
2
2
2
T.; Aoyagi, H.; Miyoshi, N.; Mutoh, T.; Togawa, M.; Kiniwa, M.; Yamasaki, Y.
Abstracts of Papers, 235th National Meeting of the American Chemical Society,
New Orleans, LA, 2008; Abstract MEDI-26.
2
2
3
3
0
1
0. Recombinant wild-type hH-PGDS used for in vitro assays and crystallization
was expressed in Escherichia coli and purified using cation exchange
chromatography and gel filtration as described in Ref. 8. The fluorescence
polarization (FP) assay was adapted from that commercially supplied by
Cayman Chemical Inc (Cat. No. 600007). Compound displacement of a potent
hH-PGDS inhibitor conjugated to fluorescein (Cayman Chemical Inc Cat. No.
piperazinone 22 could be explained by less than ideal geometry for
the putative hydrogen bond interaction involving the carbonyl.
These modeling results could help to rationalize how the structural
diversity beyond simple phenyls can be accommodated in the
inner part of the substrate binding pocket of hH-PGDS. Both affin-
ity and physico-chemical properties could potentially be improved
by manipulating this moiety.
In conclusion, we have applied fragment based screening to
identify novel inhibitors hH-PGDS containing an indole motif.
This in turn led us to the discovery of previously unknown polar
head groups for hH-PGDS, which could be used to tune compound
properties.
The synthesis of the indolyl-pyridine library is laid out in
Scheme 1. Commercial 4-bromo-indole 27 was converted to the
pinacolato boronate and then coupled with 3-amino-5-bromo pyr-
idine to yield intermediate 28, which was acylated with acid chlo-
rides obtained in situ from the corresponding acid (Scheme 1).
6
00025) was monitored in dose–response. Fluorescence polarization was
measured with excitation and emission wavelengths of 470 nm and 530 nm,
respectively, using an Envision reader (Perkin Elmer). Data analysis was
performed with the HBase software package developed by AstraZeneca.
1
1. (a) A ligand-observed 1D NMR T1
q
binding assay was set-up monitoring the
reporter molecule, either compound 1-
acid or 6-(3-fluorophenyl)pyridine-3-
binding and displacement of
phenylpyrazole-4-carboxylic
carboxamide. Active site binding for the reporter molecules was confirmed
by competition with compound 4. K values were calculated based on the level
of displacement of reporter at two different compound concentrations, taking
into account the experimentally determined concentrations of reporter and
compound using the exact mathematical expression described in Wang, Z. X.
FEBS Lett. 1995, 360, 111. K
6
and 0.80 lM using isothermal titration calorimetry (ITC); (b) Dalvit, C.; Flocco,
M.; Knapp, S.; Mostardini, M.; Perego, R.; Stockman, B. J.; Veronesi, M.; Varasi,
a
d
d
values for 1-phenylpyrazole-4-carboxylic acid and
-(3-fluorophenyl)pyridine-3-carboxamide were determined to be 140 lM
M. J. Am. Chem. Soc. 2002, 124, 7702; (c) Jahnke, W.; Floersheim, P.; Ostermeier,

2
500
F. Edfeldt et al. / Bioorg. Med. Chem. Lett. 25 (2015) 2496–2500
C.; Zhang, X.; Hemmig, R.; Hurth, K. Angew. Chem., Int. Ed. 2002, 41, 3420; (d)
Holdgate, G. A.; Anderson, M.; Edfeldt, F.; Geschwindner, S. J. Struct. Biol. 2010,
72, 142.
2. Rice, G.; Bump, E.; Shrieve, D.; Lee, W.; Kovacs, M. Quant. Cancer Res. 1986, 46,
105. 60 nL of compound was added to a 1536 well plate (greiner) using an
Echo dispenser, followed by the addition of 2 L enzyme, 1.5 g/ml in reaction
buffer (50 mM tris, 2 mM MgCl , 0.1% Chaps) using a Flexdrop dispenser. After
incubation for 30 min, 2 L of 0.4 mM MCBL and 2 L of glutathione were
GSH four-fold and stabilizes the thiolate nucleophile. The thiolate attacks the
substrate endoperoxide and induces the rearrangement reaction. It also makes
important hydrogen bonds with amide- or heterocycle-NH’s of many inhibitors
including the ones reported here (see Ref. 4c). The inner cavity is separated
from the dimer interface by the GSH thiolate, Arg14, Ser94, Asp96 and two
water molecules forming what appears to be a flexible hydrogen bonding
network. The presence of a hydroxide ion rather than a water molecule at the
interface creates asymmetry around the Mg2 ion coordination for Asp96. In
order to evaluate our novel head groups we needed to construct a model which
would not deviate from the initial coordinates under non-constrained
minimizations. To obtain a stable model with respect to the position of the
1
1
6
l
l
2
+
l
l
added using the Flexdrop dispenser. The background was analysed by
detecting fluorescence (ex 360, em 460) using a Pherastar reader, followed
by incubation for 30 min. Finally, the fluorescence after 30 min was detected
using the same settings as for the background. 0% inhibition was defined as the
fluorescence for wells containing only DMSO, and 100% inhibition was defined
as the fluorescence for a control compound.
2+
Mg ions, the bridging water was deprotonated, which seems reasonable,
2+
since the Mg ion is a fairly strong Lewis acid. Protonating the GSH amino
terminus yielded a model that did not show significant deviations from the
initial state, and which was used to evaluate the binding modes and
approximate binding enthalpies of our compounds. The electrostatics of the
system was assessed at the approximate conditions of crystallization, with the
pH at 8–8.5 range. The PROPKA (Maestro/Schrodinger Inc.) routine was used to
generate initial states. The refined models stabilities were checked by
1
1
3. Urade, Y.; Fujimoto, N.; Ujihara, M.; Hayaishi, O. J. Biol. Chem. 1987, 262, 3820.
4. Isothermal titration calorimetry experiments were performed by titrating 2
aliquots of ligand solution at an experimentally determined concentration
M) into a solution of 30–40 M hH-PGDS. For compounds 1
ll
(
typically ꢀ500
l
l
and 5 reverse titrations, for which protein was titrated into ligand solution,
were performed due to low ligand solubility.
subjecting them to a 1000 step non-constrained minimization in the
1
5. PGD
differentiated with 0.1
6-well plates and incubated in the presence of hH-PGDS inhibitor in
concentration response, 10 M–0.15 nM, for 30 min at 37 °C, followed by
stimulation with 5 M Ionomycin (Sigma, I0634-5MG) for 30 min at 37 °C.
Secreted PGD present in the media was then measured with the PGD enzyme
immunoassay (EIA) kit, Prostaglandin -MOX Express EIA Kit (Cat. No.
00151) according to the manufacturer’s instructions (Cayman Chemical Co.).
2
production assays. MEG-01 (ECACC, cat.no.94012401) cells were
OPLS2005 force field (Macromodel/Maestro/Schrodinger Inc.). None of these
models showed sufficient stability, especially regarding the position of the
lM PMA (Sigma, P1585) for 16 h. Cells were plated in
2+
9
Mg -ion. The most stable models had one of the GSH N-termini deprotonated,
losing interaction with Asp96 and staying fairly symmetrical with a water
molecule bridging the Mg2 -ion and Asp97 from both monomers. However,
deprotonating the bridging water and protonating the GSH amino terminus
l
+
l
2
2
D
2
yielded a model that did not deviate significantly from the initial state
2+
5
(backbone RMSD 0.53, Mg 0.44 Å). This model was used to evaluate the
binding modes and approximate binding enthalpies of our compounds. The
presence of a hydroxide ion rather than a water molecule at the interface
creates asymmetry around the Mg2 -ion coordination for the Asp96–97 pairs.
We obtained an EC50 of 8.1
results.
lM for HQL-79, which is in line with published
+
1
1
1
6. Hopkins, A. L.; Groom, C. R.; Alex, A. Drug Discovery Today 2004, 9, 430.
7. Springthorpe, B.; Leeson, P. D. Nat. Rev. Drug Disc. 2007, 6, 881.
8. Crystallization and structure determination was performed essentially as
described in Ref. 8. Compounds were co-crystallized with hH-PGDS. X-ray
diffraction data were collected at the European Synchrotron Radiation Facility
2+
In one monomer, both Asp96 and Asp97 coordinate the Mg ion and GSH-
amino and Arg14 OH-groups, respectively. In the other monomer, however, the
hydroxide now bridges the Mg2 -ion and Asp97 (in turn bound to GSH-amino),
while Asp96 accepts a hydrogen bond from the Tyr152-OH, rendering the
latter a better acceptor for the indole-NH’s in the inhibitors 7–15 in Table 2. In
effect, this creates two slightly different kinds of bottoms to the binding
pockets of the two monomers, with the Tyr152 OH acting as an acceptor in one
and as a donor in the other. It is unclear whether this is an artifact of our
crystallization conditions or a functional allostery allowing for regulation of
activity or reactivity (isomerase vs GSH-transferase).
+
(
ESRF), Grenoble, France.
1
2
9. Trujillo, J. I.; Kiefer, J. R.; Huang, W.; Day, J. E.; Moon, J.; Jerome, G. M.; Bono, C.
P.; Kornmeier, C. M.; Williams, M. L.; Kuhn, C.; Rennie, G. R.; Wynne, T. Y.;
Carron, C. P.; Thorarensen, A. Bioorg. Med. Chem. Lett. 2012, 22, 3795.
0. The metal ions change the hydration and the hydrogen bonding network at the
dimer interface, releasing Arg14 from hydrogen bonding to one of the metal-
coordinating Asp96 to stabilize glutathione thiolate. This lowers the K
M
for