ACS Medicinal Chemistry Letters
Letter
and screening of approved and clinical compounds.12 We
followed a virtual screening approach to identify BLT2 agonists
using the MOE software suite (Chemical Computing Group,
Montreal, Canada). After downloading the database, all
compounds with a molecular weight below 250 Da and
above 650 Da were removed. Furthermore, only compounds
exhibiting an acidic group (indicated by the MOE descriptor
a_acid > 0) were considered for virtual screening. For the
remaining compounds, adjustment of the protonation state
(MOE wash routine) and energy minimization using default
settings were performed. Multiple low energy conformations
were generated using default settings of the LowModeMD
routine, setting the energy window to 5 kJ/mol and the RMSD
cutoff to 0.75 A.
For the generation of the pharmacophore model, three
known active BLT2 agonists CAY10583, 12-(S)-HHT, and
LTB4 were pasted into the MOE database. Subsequently,
adjustment of the protonation state (MOE wash routine) and
energy minimization using default settings was performed. The
initial alignment of the active compounds was performed using
the Pharmacophore Elucidation routine with default settings,
while the conformers were generated during the search by
bond rotation. The top rated superposition was examined
manually and subsequently refined using the Flexible Align-
ment routine. A previous publication on the conformation of
12-(S)-HHT in the receptor bound state by Giusti et al.
demonstrated the importance of the interactions of the
carboxylic group and the −OH group which is supposed to
act as a H-bond acceptor.13 These observations were in perfect
agreement with the SAR of CAY10583, where the removal of
the acidic or the H-bond acceptor group resulted in fully
inactive derivatives. Therefore, we used default settings for
refinement of the existing alignment, while weights for H-Bond
acceptor and carboxylate (O2)-type Centroid were set to 10.
The resulting molecular alignment was used to generate a
pharmacophore model manually, considering the knowledge of
SAR from our previous publication (Figure 2A).10 The
pharmacophore model consisted of six features; the
importance of the anionic group and the H-bond acceptor
group (F4, cyan) was mentioned above, while three hydro-
phobic/aromatic features (F2, F3, F6) as well as a hydrophobic
feature F5 were added.
The pharmacophore search was performed on the prepared
multiconformer database. For this search, a partial match of at
least 5 features was required. Features F1, F4, and F5 were
considered to be essential due to our prior knowledge of the
SAR (vide supra). The pharmacophore search delivered 54
hits, which were inspected manually. The full list of the
The manual inspection revealed that most of the hits do not
superpose well with the shape of the active ligands, and these
were discarded. The 10 most promising hits were purchased
and tested for their potency in a functional cell-based assay
with detection of accumulation of inositol monophosphate
(IP-1) as a measure for activation of BLT2 (Table 1).
Figure 2. (A) Pharmacophore model based on CAY10583 (purple
sticks), 12-(S)-HHT (pink sticks), and LTB4 (cyan sticks). The
anionic group (F1, red), the H-bond acceptor group (F4, cyan), and
four hydrophobic features (F2, F3, F5, F6) are represented by
spheres. (B) Overlay of irbesartan (orange sticks) and CAY10583
(purple lines) with the pharmacophore model.
the irbesartan scaffold was selected to perform an initial SAR
study.
Most irbesartan derivatives were prepared according to the
synthetic route depicted in Scheme 1.
In the first step of the synthesis, the imidazolone ring was
built in a one-pot procedure by condensation reaction between
the corresponding substituted 2-aminoacetamide (2) and acid
chloride (3) in the presence of triethylamine, followed by
treatment with potassium hydroxide in methanol to give
intermediates 6a−g. In the next step, the amide was
deprotonated with sodium hydride and reacted with benzylic
bromides (5, 6) to yield compounds 7a−g. In the case that R4
is a methyl ester group, hydrolysis of the ester under alkaline
conditions provided the final products 8a−8e. In the case that
R4 is a nitrile group, the final product containing a tetrazole
ring was obtained by reaction of the nitrile precursors (7f,g)
with azidotributyltin (8f,g).
Alternatively, the synthetic routes depicted in Scheme 2
were used to obtain products 13−15. In this case, an
imidazolinidone ring was formed in the first step by reaction
of substituted 2-aminoacetamide (2) and valeraldehyde (9) in
the presence of PTSA. Intermediate 10 can react chemo-
selectively with 5 either through the amino or amide functional
groups. Thus, compound 10a reacted with 5 in the presence of
potassium carbonate in DMF to give tertiary amine 11. On the
contrary, deprotonation of 10b with sodium hydride in DMF
and subsequent nucleophilic attack to 5 gave intermediate 12.
Of the 10 compounds selected, 3 compounds from the
“sartan” family showed BLT2 activation while the other 7 were
inactive. Candesartan and valsartan showed weak activation
with EC50’s around 15−16 μM, while irbesartan (1), which
overlays very nicely with the pharmacophore model and with
the reference agonist CAY10583 (Figure 2B), was confirmed
as a moderate BLT2 activator with EC50 = 410 nM. Therefore,
B
ACS Med. Chem. Lett. XXXX, XXX, XXX−XXX