2538
M. P. DeNinno et al. / Bioorg. Med. Chem. Lett. 19 (2009) 2537–2541
(4) The target segment was poorly populated but contained sev-
eral 2-substituted benzyl substituents.
100
80
60
40
20
0
The second array combined eight templates with a selection of
monomers informed with data from the first library, plus addi-
tional 2-substituted phenylacetic acids from our monomer collec-
tion, resulting in 164 novel analogues. Figure 2 shows that the
substituents on the pyrazole template also play a part in determin-
ing selectivity. Interestingly, introduction of a 1-methylsubstituent
or increasing the steric bulk at the 3-position (isopropyl to t-butyl)
reduced selectivity. Selectivity was enhanced by replacement of
the isopropyl substituent by a 3-pyridyl group, albeit at the ex-
pense of inferior physicochemical properties. Constraint of the iso-
propyl in a cyclopentyl ring provided little advantage in either
potency or selectivity. Two of the most interesting hits (4 and 5
in Table 1) from the libraries were profiled against the PDE1 iso-
forms, which confirmed the importance of ortho substitution on
selectivity. Compound 5 became the focus of further SAR efforts.
Areas for improvement included increasing potency and selectivity
as well as polarity and solubility.
2
3
4
1
-20
0
20
40
60
80
100
Despite lacking a protein crystal structure early in the program,
it was inferred that the pyrimidone portion of the core was forming
key hydrogen bonds to the conserved glutamine in the catalytic
site, similar to PDE5, and was therefore considered essential.6
Small aliphatic groups were preferred at C3, although this SAR will
not be presented within the scope of this publication. One obvious
area for modification was the benzyloxy group, which was a met-
abolic and solubility liability. Replacement of the benzyl group
with a variety of amines and other polar substituents was well tol-
erated suggesting that this region was exposed to solvent. This
hypothesis was supported by homology modeling and later con-
firmed by X-ray crystallography.7
Figure 1. Profile of diverse 1st array in PDE 1 and 9 assays (1
l
M concentration).
The first array combined the same isopropyl-substituted template
as in compound 1 but explored a wide diversity of carboxylic acids,
resulting in 272 5-substituted pyrazolopyrimidinone derivatives.
These analogs were profiled in a single point inhibition assay
against both PDE1 and PDE9 (results displayed in Fig. 1) and
through endpoint assay determination for selected examples. A
number of SAR conclusions were drawn from this array based on
a number of common features of the C5 substituent in the com-
pounds from the four marked segments:
Compound 6 emerged as an analog of interest due to its im-
proved selectivity and solubility. As such, it was deemed ade-
quately potent and selective to begin in vivo testing. After
substantial investigation, it was found that no acute glucose lower-
ing could be detected. However, sub-chronic dosing in ob/ob mice
showed robust glucose lowering and reduced weight gain (Table
2). Poor pharmacokinetics in mice required the use of very high
doses admixed with powdered chow to achieve sustained plasma
drug exposure. The fact that insulin was also reduced suggested
improved glucose disposal. This profile was not unlike the KO phe-
notype and gave us the confidence to continue with the program.
Given that the core structure in 6 was the same as found in sil-
denafil, we attempted to leverage the SAR learned in that program.
Methylation at N1 for example, was found to increase PDE5 po-
tency and selectivity over PDE1.8 Unfortunately, this modification,
analog 7, resulted in loss of both potency and selectivity for PDE9
confirming the results from the libraries. It was noted that 7
showed substantial improvement in permeability. The inability
to mask the pyrazole NH on the core would have implications later
in the program (vide infra).
The improved selectivity of the saturated analog 8 over PDE1a
and PDE1b prompted further exploration in the cyclohexyl series.
An added benefit of this modification was that it attenuated the
metabolic lability of the doubly benzylic methylene group found
in 6.
To expand the SAR, the amino cyclohexyl intermediate 20 was
targeted. Its synthesis required an efficient preparation of the acid
19 which initially proved elusive. A scalable route was eventually
developed as shown in Scheme 2. The key steps were using
dissolving metal reduction of oxime 16 to achieve selectivity for
the desired trans isomer, and the use of the trifluoroacetamide
protecting group which provided the highly crystalline intermedi-
ate 18 which facilitated purification. The absolute configuration of
the active enantiomer of the final products was unambiguously
(1) Compounds showing weak activity against both PDE1 and 9
were mostly derived from arylcarboxylic acids.
(2) Compounds showing strong activity against PDE1 but
weakly active against PDE9 featured alpha-branched benzyl
groups.
(3) Potent inhibitors of both PDE1 and 9 included a range of 3-
and 4-substituted benzyl groups.
100
80
60
40
20
0
-20
-20
0
20
40
60
80
100
Figure 2. Profile of directed 2nd array in PDE 1 and 9 assays (1 lM concentration).
Colored by R1,R3: H, cPent; H, iBu; H, iPr; H, nBu; H, tBu; H, pyridin-2-
yl; H, pyridin-3-yl; Me, nPr.