M. Azimioara et al. / Bioorg. Med. Chem. Lett. 24 (2014) 5478–5483
5479
Table 1
Table 2
SAR of the core pyrazolo[1,5-a]pyrimidine substitution The numbers in italics are
Some selected ADME/PK properties for analogs 1, 4 and 9
positional labels
Entry
clogP
Precipitationa
ÀlogPAMPA
ERb (h/r/m)
AUCc h*
lM
R2
1
4
8
5.74
6.19
5.96
Low
Med
High
5.5
5.4
5.5
0.3–0.7
0.3–0.5
0.6–1.0
2.1
0.6
0.3
2
1
N
N
3
EtO2C
7
R1
4
a
b
c
N
Precipitation in PEG/D5W vehicles.
Extraction ratio.
Dose-normalized AUC (po) in Balb/C mice.
R6
6
5
R5
R6
Entry
R1
R2
R5
hGPR119a
mGPR119a
EC50 Eff (%)
EC50
Eff (%)
displayed high potency with increased agonist efficacy on both
human and rodent receptors (Fig. 2).
This increase in receptor activity of compound 8 came at the
price of a decrease in exposures compared to the lead compound
1 (Table 2). Whereas clogP values and permeabilities did not
change dramatically, we noticed a further drop in the already poor
4
5
6
7
CF3
Cl
CF3
CF3
H
H
Me
H
H
Me
H
Me
H
H
1
80
15
83
75
11
62
>10
23
3
24
8
72
73
H
Cl
a
EC50 values in nM; 100% efficacy = max. response of compound 2.
solubility (generally <5 lM), the extent of which was difficult to
pick up in our solubility assay. However, in several organic and
aqueous media, there was significantly more precipitation
observed for compounds 4 and 8 compared to 1. The in vitro hepa-
tic clearance (depicted as a range of microsomal extraction ratio in
human, mouse and rat liver microsomes) had a clear trend to
increase dramatically in the tricyclic system 8. Concomitantly, oral
exposure in mice significantly decreased from 1 to 8 (in vivo
hepatic clearance not measured).
We hypothesized that the gains in receptor potency and efficacy
for the tricyclic analogs would allow us to implement changes to
the scaffold previously not tolerated in the bicyclic series, thereby
leading to increased exposures via more favorable physicochemical
properties, while keeping activities in the acceptable range
required for in vivo efficacy. An SAR summary (human and mouse
GPR119 receptor activity, physicochemical properties, metabolic
stability) for analogs 8–38 is depicted in Table 3.
A general synthetic pathway to these tricyclic structures is
described in Scheme 1 (detailed procedures can be accessed in
the patent application WO 2011/014520 A2).15 Benzylation of the
appropriate ketoester using KHMDS, followed by treatment with
DMF-dimethyl acetal yielded the racemic (dimethylamino)-meth-
ylene ketone. The optimal synthetic pathway for enantiomerically
pure analogs proved to be analytical resolution (preparative chiral
chromatography)16 at this stage. The final products were afforded
by acid-catalyzed condensation with the appropriate 4-aryl amino-
pyrazole, which was derived from the corresponding aryl acetoni-
trile via condensation with ethyl formate and subsequent
cyclization with hydrazine.
In line with previous observations, the (R) enantiomer 8 was
significantly more active than the (S) enantiomer 9. The ethyl
and methyl esters had similar activities and properties for a variety
of analogs (e.g. 8 vs 10). The 3-fluoro substitution on the benzyl
group (R7 = F in Table 3) did not bring about any major changes
(entries 8 vs 11).
Expansion of the carbocyclic ring beyond the 5-membered car-
bacycle in 8 led to reduced GPR119 activity (entries 12–14). As
described in the preceding report, SAR around the position 3 and
7 of the core heterocycle was rather steep, the more polar substit-
uents being particularly not tolerated. Indeed, small changes in the
para-substituent of the 3-aryl group led to a marked drop in recep-
tor agonist potency and efficacy (entries 15–18). In an effort to
decrease hydrophobicity and increase solubility in the series, we
thus focused our attention to the newly formed carbacycle in the
tricyclic core. To our surprise, either inserting a heteroatom posi-
tion 7 of the core heterocycle (entries 19 and 20) or next to the
quarternary carbon (entries 21 and 22) did not result in a complete
loss of GPR119 agonist activity. Even the introduction of a physio-
logical-conditions ionizable secondary amine 22 was well
in vivo inC57Bl/6 mice,13 it was only marginally active in an acute
rat OGTT study.14 We hypothesized that the low efficacy may be
due to three main factors: (1) partial agonism (680%)11 on the
receptor, (2) high plasma protein binding and (3) poor exposure
due to low solubility and low metabolic stability. Hence we sought
to improve the physicochemical properties of the scaffold while
also enhancing its agonistic efficacy on the receptor.
First we expanded the existing SAR of 1 by substituting posi-
tions of the pyrrolopyrimidine core not previously investigated.
The results depicted in Table 1 suggest that substitution is toler-
ated in positions 2 and 6 (entries 6, 4 and 7, respectively), but
not in position 5 (entry 5). It is noteworthy that in both positions
only small substituents such as methyl and halogen groups were
tolerated, whereas larger substituents led to loss of GPR119
activity.
However, both potency and efficacy were not improved. We
observed a significant drop of activity, especially on the rodent
receptor (e.g. compound 4). Not surprisingly, these compounds
did not significantly elevate GLP-1 levels in rodents (data not
shown).
In order to gain agonist efficacy on the receptor, we intended to
find if there was a preferred ‘active’ conformation of the scaffold
ameliorating the binding and activation of both the human and
mouse receptor. One way to sample some of the potential conforma-
tions is to reduce the degree of rotational freedom and rigidify the
scaffold into a productive, pharmacologically relevant conforma-
tion. Hence a number of tricyclic compounds were synthesized,
using the 6-position of the scaffold as an anchoring point for cycliza-
tions. The cyclizations involved each of the three substituents of the
quaternary carbon next to position 7 of the core (e.g. the ester,
methyl or benzyl group), thereby allowing to ‘freeze’ three distinct
conformations of this particular substituent. Cyclizing the ester or
the benzylic position with the 6-Me group led to inactive com-
pounds (results not shown), but tying both methyl groups into a ring
led to the more restricted analog 8. Analogs such as 8 consistently
N
N
N
N
EtO2C
EtO2C
CF3
CF3
N
N
4
8
hGPR119: 0.001 µM (80%)
mGPR119: 0.011 µM (62%)
hGPR119: 0.0005 µM (85%)
mGPR119: 0.008 µM (80%)
Figure 2. Rigidification of 4 leads to increased GPR119 agonist efficacy.