1622
K. A. Atkinson et al. / Bioorg. Med. Chem. Lett. 21 (2011) 1621–1625
Table 1
products were obtained. These could be separated by standard col-
Pharmacological, ADME and safety properties of 5a
umn chromatography. The esters were saponified to provide the
required N-benzyl pyrazole and triazole acids 13 as generally de-
picted in Scheme 5. The acids were then coupled to appropriate
heterocyclic amines.
O
H2N
N
N
N
H2N
Cl
General SAR trends could be gleaned from the tail and amide re-
gion libraries. These trends are represented by the selected com-
pounds shown in Table 2. Benzylic tail moieties (6b–f, 6i) were
highly preferred over phenyl (data not shown) or phenethyl (6h)
substitution. A single meta substituent, such as Cl or CF3, on the
pendant benzyl group generally led to more ligand efficient inhib-
itors. When the core or amide group was less optimal, potency
could often be boosted by addition of a para halogen (e.g., 6c, 6i,
15). However, this offered no advantage toward our goal for
in vivo activity since the increase in log P associated with this
change often resulted in lower microsomal stability and lower sol-
ubility. In some cases, highly specialized aliphatic groups were also
tolerated tail substituents (6g). Similar trends were noted for the
amide region. Small five-membered aminoheterocycles were
highly preferred (e.g., 6b, 6d, 6f). Aliphatic cyclic (data not shown)
or acyclic amines (e.g., 6i) or selected six-membered aminohetero-
cycles were tolerated but usually resulted in a one to two order of
magnitude loss in potency.
5a
polypharmacology
pharmacophore
element
risk
5a
MW 291.7, e Log D 1.2, TPSA 99
HTMS rSCD-1 IC50
68 nM
94 nM
217 nM
2.6
74.1
15.5
>15 assays with IC50 <10 lM
<1%
64%
[3H]-stearoylCoA rSCD-1 IC50
hSCD-1 IC50
PAMPAa (Papp, 10À6 cm/s)
RLM CLint (
HLM CLint (
l
l
L/min/mg)
L/min/mg)
CEREP/Bioprint™ panel
CYP1A2, 2C9, 2D6, 3A4 (pct inh at 3
l
M)
DOFb (pct inh at 10
lM)
a
PAMPA = Parallel artificial membrane permeability assay.
DOF = [3H]-dofetilide binding assay.12a,b
b
of the imidazole served as the key pharmacophoric element similar
to the o- or m- (pseudo)halo substituted aryl groups of known SCD-
1 matter. We also hypothesized that the acyl guanidine moiety in
5a contributed to the polypharmacology and hERG liabilities since
this functional group is present in drugs known to exert effects on
ion channels (e.g., amiloride,10 zoniporide11).
Multiple five- and six-membered ring cores, with both N-linked
and C-linked tails, were assessed. For ease of synthesis and best
balance of physical properties and ADME characteristics, the
imidazole core was preferred (e.g., 6b). The 1H-pyrazole-3-carbox-
amide 14 and triazole 15 cores were well tolerated, while the
1H-pyrazole-4-carboxamide 17 was less potent.
Early on, a small cohort of compounds was tested in both the rat
and human enzyme assays. The IC50 data from these assays were
generally within two- to three-fold of each other. We found that
replacement of the guanidine moiety of 5a with an N-methylpyra-
zole 6a gave a compound that was not only more potent (Table 2)
but also more selective as judged by polypharmacology risk with
Compounds 5b, 6b, 14, and 15 were selected for further evalu-
ation based upon their promising in vitro profiles. The in vitro hu-
man and rat microsomal clearance as well as the in vivo rat
pharmacokinetic data are summarized in Table 3. In vitro, all com-
pounds had low to moderate clearance in human liver microsomes
(HLM) but were highly variable in rat liver microsomes (RLM). In
vivo, compounds 5b and 17 exhibited moderate clearance, moder-
ate volume of distribution, moderate half life and good bioavail-
ability, while 6b and 15 had clearances that were significantly
higher than rat liver blood flow. The stark contrast in the metabolic
fate of 6b versus 5b in rats was puzzling. Further assessment of 6b
revealed very low renal and bile excretion as well as good stability
in rat plasma. While amide hydrolysis can be offered as a possible
explanation, a definitive explanation for the very high rat in vivo
clearance for 6b (and 15) was not determined. It is notable that
both 6b and 15 were susceptible to degradation in RLM in the ab-
sence of co-factors for CYP enzymes (nonmetabolic decline).
Compound 5b was further assessed in an advanced battery of
in vitro assays (Table 4). Of note was the high selectivity over
P450 enzymes and selectivity over a wide panel of enzymes, recep-
tors and ion channels. Based on the rSCD-1 potency, low human
and rat in vitro CL, moderate in vivo rat CL and a good in vitro
safety profile, compound 5b was selected for further in vivo
characterization.
In rodents, it has been demonstrated that short-term feeding of
a low-fat, high-carbohydrate diet increases hepatic SCD-1 expres-
sion.14 Thus, an acute rat study, using the desaturation index (DI)
as an in vivo biomarker, was used to assess efficacy of 5b. DI, which
correlates with hyperlipidemia,15 was calculated as ratio of total
oleic acid: stearic acid concentrations in plasma. Concentrations
of stearic and oleic acid were determined by direct quantitation
of extracted stearic and oleic acid from plasma by LC/MS.
Sprague–Dawley rats (n = 8 per dose group) were fasted for
20 h, then fed a low essential fatty acid (LEFA) diet16 for 24 h.
The next day, the rats were dosed (po) with 5b (vehicle 0.5% meth-
ylcellulose). Plasma samples were taken three hours post dose.
Compound 5b showed a dose-dependent decrease in DI with a
only two assays in the CEREP/Bioprint™ panel with IC50 <10 lM.
Based on the encouraging data from 6a, the chemical matter in
this series was developed using a combination of parallel synthesis
and singleton synthesis. We elected to use the [3H]-stearoylCoA as-
say system with rat liver microsomes as the primary assay. Com-
pounds made by library synthesis were first screened in triplicate
for percent inhibition at a single 5
mination if warranted.
lM dose followed by IC50 deter-
Iterative library design was used to rapidly probe the tail and
amide regions of the molecule. A three-step library procedure, out-
lined in Scheme 1, was used to probe the tail region. Treatment of 7
with DMF–DMA provided enamine 8, which was reacted with a
wide range of substituted aryl, benzyl and aliphatic amines and
saponified to provide imidazole acids 9. The acids 9 were treated
with a narrow range of aminoheterocycles under standard amide
coupling conditions to provide imidazole amides 5. Gram quanti-
ties of selected N-benzyl imidazole acids 9 and 1013 were synthe-
sized as outlined in Schemes 1 and 2. A one-step library protocol
was designed to probe variation of the amide substituent. A range
of amino heterocycles similar to N-methylpyrazole and a diverse
set of aliphatic amines were coupled under standard amide bond
formation conditions (HATU or HBTU) with 9 or 10 to provide
amides 6 and 5, respectively.
The heterocyclic core determines the position of each substitu-
ent and can have strong impact on the physicochemical properties
of the final compounds. Such changes were less amenable to vari-
ation through parallel chemistry methodology, therefore these
changes were probed with singleton synthesis to afford analogs
14–17. Heterocyclic esters, either from commercial sources or syn-
thesized according to Schemes 3 and 4, were alkylated with the
appropriate benzyl halide. In some cases, regioisomeric alkylation
Atkinson, Karen A.
