Identification of novel IP receptor agonists using historical ligand biased chemical arrays

Article abstract of DOI:10.1016/j.bmcl.2014.03.089

By considering published structural information we have designed high throughput biaryl lipophilic acid arrays leveraging facile chemistry to expedite their synthesis. We rapidly identified multiple hits which were of suitable IP agonist potency. These re

Full text of DOI:10.1016/j.bmcl.2014.03.089

Bioorganic & Medicinal Chemistry Letters 24 (2014) 2247–2250
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Identification of novel IP receptor agonists using historical ligand
biased chemical arrays
a
a
a
a
Stephen C. McKeown a, , Steven J. Charlton , Brian Cox , Helen Fitch , Christopher D. Howson ,
⇑
Catherine Leblanc b, Arndt Meyer b, Elizabeth M. Rosethorne a, Emily Stanley a
a Novartis Institutes for BioMedical Research, Wimblehurst Road, Horsham, West Sussex RH12 5AB, UK
b Novartis Pharma AG, Novartis Institutes for BioMedical Research, Postfach, CH-4002 Basel, Switzerland
a r t i c l e i n f o
a b s t r a c t
Article history:
By considering published structural information we have designed high throughput biaryl lipophilic acid
arrays leveraging facile chemistry to expedite their synthesis. We rapidly identified multiple hits which
were of suitable IP agonist potency. These relatively simple and strategically undecorated molecules
present an ideal opportunity for optimization towards our target candidate profile.
Ó 2014 Elsevier Ltd. All rights reserved.
Received 5 March 2014
Revised 25 March 2014
Accepted 26 March 2014
Available online 5 April 2014
Keywords:
IP agonist
PGI2 agonist
Pulmonary arterial hypertension
Prostanoid
IP (prostaglandin I2) receptor agonists are one of the most effi-
cacious therapies for the treatment of pulmonary arterial hyper-
tension (PAH). However, current drugs, such as prostacyclin (1)1,
iloprost (2)2 and treprostinil3 (compound 3, Fig. 1) have major
inconveniences (e.g.: iv, sub-cutaneous or nebulized administra-
tion), are short acting or are poorly tolerated.4
Most, if not all, published IP agonists to date have originated
from three distinct structures: the PGI2 scaffold, biphenyl methy-
lene EP1575 and octimibate,6 with the latter two having unknown
specific prostanoid selectivity at the time of their publication
(Fig. 2). In this Letter we describe the facile high throughput array
chemistry based upon the general structure derived from both
EP157 and octimibate to identify bicyclic IP agonists as novel start-
ing points (hits) for further optimization.
The optimization of ligands described around EP157 (4) or
octimibate (10) have mainly focused on achieving pharmacokinet-
ics suitable for good oral bioavailability. Figure 2 outlines the evo-
lution of some key published chemotypes.7
We were particularly interested in lipophilic biaryl structures
since they could offer properties beneficial to the pharmacokinetic
profile we sought following inhaled administration.8 We wanted to
explore molecules that could have a long duration of action
through affinity to local tissue and consequently result in a low
systemic exposure (Cmax) in order to limit potential side effects
Figure 1. Prostacyclin and analogues.
which are observed with current systemically delivered
molecules.9
Since there have been multiple reported adaptations of the biar-
yl system we embarked upon a rapid targeted array strategy using
this published landscape to generate novel biaryl lipophilic acids
with which to initiate a lead optimization project. The two biaryl
motifs that have been the focus of drug discovery efforts (Fig. 2)
have the aryl motifs separated by either 1 or 2 atoms. The biaryl
motifs have been attached to a carboxylic acid through a variety
of spacing groups. We were keen to balance the design features
with a pragmatic ability to broadly cover a range of properties by
synthesizing 100’s of compounds. Whilst we could design specific
target molecules based around computational modelling of prece-
dented structures, not only did these require more complex assem-
bly chemistry, but we appreciated that the design element made
assumptions of structural shape and interactions that we could cir-
cumvent by making larger numbers of ligands. An important con-
sideration in our strategy was to use relatively undecorated
⇑
Corresponding author. Tel.: +44 1403 324531.
E-mail address: steve.mckeown@novartis.com (S.C. McKeown).
http://dx.doi.org/10.1016/j.bmcl.2014.03.089
0960-894X/Ó 2014 Elsevier Ltd. All rights reserved.

2248
S. C. McKeown et al. / Bioorg. Med. Chem. Lett. 24 (2014) 2247–2250
(unsubstituted) building blocks, since firstly from the literature
examples the lipophilic and acidic binding interactions are able
to give enough potency to be able to identify initial starting hits
and secondly that we wanted to avoid jeopardizing chances of
finding good templates due to any negative interactions from other
substituents.10 Our approach involved decorating any ‘hit’ tem-
plates during the optimization phase.
We therefore designed arrays based around three important
considerations: (i) the general pharmacophore features (i.e. the
biaryl motif and a carboxylate terminus), (ii) simple chemical con-
nection steps linking the two pharmacophore features together
and (iii) complexity, producing relatively undecorated products
(Fig. 3).
We considered multiple synthetic avenues and found that we
could construct numerous target templates using available in-
house building blocks and by employing facile well established ar-
ray chemistry. We chose amide, ether and amino linking strategies,
using the biaryl feature in one synthetic partner and a carboxylic
acid protected as an ester. For the amine array we used reductive
alkylation chemistry and decided to choose building blocks with
the amino functionality on the biaryl portion (e.g. 16 and 20) due
to the poor availability of biaryl aldehydes. The reductive alkyl-
ation chemistry, by way of example, is illustrated in Scheme 1.
The amine array was of particular interest as any hits could possess
high affinity for tissues through a strong interaction with phospho-
lipids. The observation of long duration of action, our desired
Figure 2. Evolution of published structures: from EP1575 (a) and octimibate6 (b).

S. C. McKeown et al. / Bioorg. Med. Chem. Lett. 24 (2014) 2247–2250
2249
The ether array (D) was constructed using aryl alcohols (e.g. 21)
and bromoesters (e.g. 25) and employed solid sodium hydride
(NaH) in THF to effect ether bond formation followed by sodium
hydroxide (NaOH) treatment to give the final carboxylic acids
which were purified using preparative HPLC. All the arrays were
performed on a 0.1 mmol scale using 8 mL glass vials as reaction
vessels (see Supplementary material for further synthesis and puri-
fication information).
After synthesis a total of 499 molecules were constructed from
the 4 arrays. The output from the ether array was lower than ex-
pected due to poor product formation that could have come from
in situ quenching of the NaH from the starting materials (which
may not have been dry). The reductive alkylation arrays produced,
after ester hydrolysis, compounds with zwitterionic character
which also presented solubility issues during purification and
may have led to failed recovery of the final products.
Figure 3. Design features of the chemical arrays. The ‘2C’ and ‘1C’ building blocks
were chosen to represent the octimibate and EP157 lineage (Figure 2) respectively.
Figure 4 shows the range of properties that were captured with
the successfully synthesized compounds. Pleasingly we found that
a broad range of properties were covered by the isolated products
which offered opportunities to choose multiple structures to opti-
mize, each with a unique physicochemical profile. For example, ar-
ray C products contained a basic group embedded in the linker that
could not only contribute to phospholipid binding but also aid sol-
ubility and salt formation if required.
Scheme 1. Construction of amine array using reductive alkylation.
profile, of lipophilic amines after inhalation has been reported with
b2-adrenoceptor agonists.8
We included reactants that gave rise to products with a range of
clogP’s and molecular weights (Mw). Table 1 shows the final array
dimensions, example reactants and number of products obtained
for each chemical array.
Each array was then screened in our IP agonist cAMP assay.11
Hits (<10 lM, >40% efficacy) were identified from the arrays and
served as starting templates for further exploration. These identi-
fied hits were resynthesized on a larger scale (50 mg) in order to
gain further characterization and physicochemical property mea-
surements. Figure 5 shows three example templates identified
from the amide and reductive alkylation arrays. No hits were found
from the ether array products which may be partly due to the small
number prepared. In total we identified 9 molecules that had an
Amide array A was constructed using solid supported diisopro-
pylethylamine
and
2-(7-aza-1H-benzotriazole-1-yl)-1,1,3,3-
tetramethyluronium hexafluorophosphate (HATU) in dimethyl-
formamide (DMF). The reactions were performed in sealed 8 mL
glass vials with overnight shaking and then partially purified using
ISOLUTEÒ PE-AX cartridges (BiotageÒ). Ester saponification was
then effected using lithium hydroxide (LiOH) in H2O/tetrahydrofu-
ran (THF). The final products were then purified using preparative
high-performance liquid chromatography (HPLC).
Array B was performed as array A except that Argoresin MP-iso-
cyanate resin (BiotageÒ) was added to remove excess amine re-
agent. After saponification with LiOH, the products were initially
purified using ISOLUTEÒ SCX-2 cartridges (BiotageÒ). Those prod-
ucts requiring further purification were purified using preparative
HPLC.
EC50 of <10 lM and an efficacy (compared to treprostinil) of
>40%. Table 2 shows the potency and physicochemical profile of
three example hits. As observed from the hits (e.g. 26–28), there
is a variability of the accepted lipophilic motifs between the car-
boxylate and biaryl motifs, which is also seen from published
structures (Fig. 2). For example, the inclusion of a basic centre
(compound 15), polar heterocycles (compound 8) and ethers (com-
pound 10) seem to be tolerated which led us to take this approach
with a variety of linking chemistries.
The reductive alkylation array (C) utilized MP-triacetoxyboro-
hydride resin (BiotageÒ) in DMF and acetic acid and after interme-
diate purification again with ISOLUTEÒ SCX-2 cartridges,
saponified using LiOH. The products were then isolated by phase
extraction using 10% citric acid and dichloromethane followed by
preparative HPLC.
Table 1
Array dimensions and final product count
Biaryls: Example
and number of
reagents
Esters: Example
and number of
reagents
Linking chemistry
Isolated
products
Amide coupling
(Array A)
58
8
319
Figure 4. C logP versus Mw of synthesized molecules.
Amide coupling
(Array B)
Reductive
24
25
7
4
125
42
alkylation (Array C)
Ether formation
(Array D)
7
4
13
Figure 5. Example hits from arrays A (26), B (27) and C (28).

2250
S. C. McKeown et al. / Bioorg. Med. Chem. Lett. 24 (2014) 2247–2250
3. LeVarge, B. L.; Channick, R. N. Expert Rev. Respir. Med. 2012, 6, 255.
Table 2
4. Vachiery, J.-L.; Simonneau, G. Eur. Respir. Rev. 2010, 19, 279.
Potency and property data for 26, 27 and 28
5. Armstrong, R. A.; Jones, R. L.; MacDermot, J.; Wilson, N. H. Br. J. Pharmacol.
1986, 87, 543.
26
27
28
6. Seiler, S.; Brassard, C. L.; Arnold, A. J.; Meanwell, N. A.; Fleming, J. S.; Keely, S. L.,
Jr. J. Pharmacol. Exp. Ther. 1990, 255, 1021.
IP cAMP EC50
(%max)a,b
clogP
l
M
1.75 0.12
(85.49 26.89)
2.52
428
0.39
0.25 0.03
(48.37 12.36)
4.75
393
0.078
0.54 0.16
(78.40 1.38)
2.82
414
0.085
7. (a) Hamanaka, N.; Takahashi, K.; Nagao, Y.; Torisu, K.; Tokumoto, H.; Kondo, K.
Bioorg. Med. Chem. Lett. 1995, 5, 1065; (b) Takamura, F.; Tanaka, A.; Takasugi,
H.; Taniguchi, K.; Nishio, M.; Seki, J.; Hattori, K. Bioorg. Med. Chem. Lett. 2006,
16, 4475; (c) Hamanaka, N.; Takahasi, K.; Nagao, Y.; Torisu, K.; Tokumoto, H.;
Kondo, K. Bioorg. Med. Chem. Lett. 1995, 5, 1083; (d) Tsubaki, K.; Taniguchi, K.;
Tabuchi, S.; Okitsu, O.; Hattori, K.; Seki, J.; Sakane, K.; Tanaka, H. Bioorg. Med.
Chem. Lett. 2000, 10, 2787; (e) Ohno, M.; Tanaka, Y.; Miyamoto, M.; Takeda, T.;
Hoshi, K.; Yamada, N.; Ohtake, A. Bioorg. Med. Chem. 2005, 2006, 14; (f)
Meanwell, N. A.; Rosenfeld, M. J.; Trehan, A. K.; Wright, J. J. K.; Brassard, C. L.;
Buchanan, J.; Federici, M. E.; Fleming, J. S.; Gamberdella, M.; Zavoico, G. B.;
Seilert, S. M. J. Med. Chem. 1992, 35, 3483; (g) Meanwell, N. A.; Romine, J. L.;
Rosenfeld, M. J.; Martin, S. W.; Trehan, A. K.; Wright, J. J. K.; Malley, M. F.;
Gougoutas, J. Z.; Brassard, C. L. J. Med. Chem. 1993, 36, 3884; (h) Asaki, T.;
Hamamoto, T.; Sugiyama, Y.; Kuwano, K.; Kuwabara, K. Bioorg. Med. Chem.
2007, 15, 6692; (i) Lampe, T.; Kast, R.; Jeske, M.; Stoll, F.; Schuhmacher, J.;
Becker, E.-M.; Beck, H. WO2008155016, 2008.; (j) Tanaka, A.; Hattori, K.;
Taniguchi, K.; Okitsu, O.; Tabuchi, S.; Nishio, M.; Nagakura, Y.; Maeda, N.;
Muraic, H.; Sekid, J. Bioorg. Med. Chem. Lett. 2006, 16, 4861.
Molecular weight
Solubility@pH6.8
(mg/mL)12
tPSA13
86.71
66.4
65.9
a
Average of at least two experiments; %max relative to treprostinil, 3 (EC50
0.58 0.95 nM).
:
b
MRE-269, 13: EC50: 1.5 1.6 nM (98%).
In summary, by harnessing simple precedented pharmacophore
features (biaryl and carboxylic acid), facile assembly chemistry and
relatively undecorated building blocks together with the strategic
choice of the pharmacokinetic profile that a lipophilic acid struc-
ture may offer, we have rapidly identified multiple, attractive
and novel IP agonist templates with which to initiate an optimiza-
tion effort. The exploration and progression of these templates will
be the subject of future publications.
8. Beattie, D.; Bradley, M.; Brearley, A.; Charlton, S. J.; Cuenoud, B. M.; Fairhurst, R.
A.; Gedeck, P.; Gosling, M.; Janus, D.; Jones, D.; Lewis, C.; McCarthy, C.; Oakman,
H.; Stringer, R.; Taylor, R. J.; Tuffnell, A. Bioorg. Med. Chem. Lett. 2010, 20, 5302.
9. Raina, A.; Coons, J. C.; Kanwar, M.; Murali, S.; Sokos, G.; Benza, R. L. Pulm.
Circulation 2013, 3, 116.
10. Hann, M. M.; Leach, A. R.; Harper, G. J. Chem. Inf. Comput. Sci. 2001, 41, 856.
11. cAMP levels were assessed using the AlphaScreenÒ (Perkin ElmerÒ
)
Supplementary data
competition immunoassay. For EC50 determinations, stable IP receptor
transfected CHO cells were incubated with a range of agonist concentrations
or vehicle for 1 h at room temperature in HBSS buffer (with Ca2+/Mg2+
)
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.bmcl.2014.03
.089.
supplemented with HEPES (5 mM), BSA (0.1% w/v) and rolipram (5
biotinylated-cAMP/bead mix was added, shaken in the dark for 1 hr and levels
of cAMP assessed on an EnVisionÒ (PerkinElmerÒ
plate reader. cAMP
standard curve in
lM). The
)
concentrations in each well were determined using
a
GraphPadPrismÒ (GraphPad Software Inc). Data were normalized to the
maximal response elicited by treprostinil, and expressed as mean SD of at
least 2 independent experiments.
References and notes
1. Moncada, S.; Gryglewski, R.; Bunting, S.; Vane, J. R. Nature 1976, 263, 663.
2. Kawahara, H.; Sakamoto, T.; Takeda, S.; Onodera, H.; Imaki, J.; Ogawa, R. Anesth.
Analg. 2001, 1012, 93.
12. Zhou, L.; Yang, L.; Tilton, S.; Wang, J. J. Pharm. Sci. 2007, 98, 3052.
13. Ertl, P.; Rohde, B.; Selzer, P. J. Med. Chem. 2000, 43, 3714.