Maximizing lipophilic efficiency: The use of Free-Wilson analysis in the design of inhibitors of acetyl-CoA carboxylase

Article abstract of DOI:10.1021/jm201503u

This paper describes the design and synthesis of a novel series of dual inhibitors of acetyl-CoA carboxylase 1 and 2 (ACC1 and ACC2). Key findings include the discovery of an initial lead that was modestly potent and subsequent medicinal chemistry optimization with a focus on lipophilic efficiency (LipE) to balance overall druglike properties. Free-Wilson methodology provided a clear breakdown of the contributions of specific structural elements to the overall LipE, a rationale for prioritization of virtual compounds for synthesis, and a highly successful prediction of the LipE of the resulting analogues. Further preclinical assays, including in vivo malonyl-CoA reduction in both rat liver (ACC1) and rat muscle (ACC2), identified an advanced analogue that progressed to regulatory toxicity studies.

Full text of DOI:10.1021/jm201503u

Article
pubs.acs.org/jmc
Maximizing Lipophilic Efficiency: The Use of Free-Wilson Analysis in
the Design of Inhibitors of Acetyl-CoA Carboxylase
Kevin D. Freeman-Cook,*,† Paul Amor, Scott Bader, Leanne M. Buzon, Steven B. Coffey,
Jeffrey W. Corbett, Kenneth J. Dirico, Shawn D. Doran, Richard L. Elliott, William Esler,
Angel Guzman-Perez, Kevin E. Henegar, Janet A. Houser, Christopher S. Jones, Chris Limberakis,
Katherine Loomis, Kirk McPherson, Sharad Murdande, Kendra L. Nelson, Dennis Phillion, Betsy S. Pierce,
Wei Song, Eliot Sugarman, Susan Tapley, Meihua Tu, and Zhengrong Zhao
Pfizer Worldwide Research and Development, Eastern Point Road, Groton, Connecticut 06340, United States
S
* Supporting Information
ABSTRACT: This paper describes the design and synthesis of a novel series of dual inhibitors of acetyl-CoA carboxylase 1 and 2
(ACC1 and ACC2). Key findings include the discovery of an initial lead that was modestly potent and subsequent medicinal
chemistry optimization with a focus on lipophilic efficiency (LipE) to balance overall druglike properties. Free-Wilson
methodology provided a clear breakdown of the contributions of specific structural elements to the overall LipE, a rationale for
prioritization of virtual compounds for synthesis, and a highly successful prediction of the LipE of the resulting analogues.
Further preclinical assays, including in vivo malonyl-CoA reduction in both rat liver (ACC1) and rat muscle (ACC2), identified
an advanced analogue that progressed to regulatory toxicity studies.
role in m-CoA production, acetyl-CoA carboxylase (ACC) is a
target of significant interest within the pharmaceutical
industry.6−11 Experiments with antisense oligonucleotides
suggest that pharmacological inhibition of ACC may have the
potential to reduce m-CoA levels, to lower ectopic lipid
accumulation, and to improve glucose homestatis.12
ACC is a biotin-dependent heterodimeric protein that uses
bicarbonate ion to form an enzyme−biotin−carbonate complex
in an adenosine triphosphate (ATP)-dependent reaction. It
then transfers carbon dioxide to acetyl-CoA to form m-CoA.13
Two isoforms of ACC are known [acetyl-CoA carboxylase 1
(ACC1) and acetyl-CoA carboxylase 2 (ACC2)], which differ
somewhat in their tissue distribution. ACC1 is primarily located
in liver and adipose tissue, while ACC2 is the major form found
in skeletal and heart muscle. Our research program focused on
ACC1/ACC2 dual inhibitors, to derive potential therapeutic
benefit from reducing fatty acid (m-CoA) synthesis in the liver
as well as stimulating fatty acid oxidation in both liver and
muscle tissue.14
INTRODUCTION
■
Type 2 diabetes mellitus (T2DM) has become a major global
health problem. The World Health Organization (WHO)
estimates that more than 220 million people worldwide have
diabetes. While often considered a disease of the industrialized
world, more than 80% of deaths due to diabetes occur in low-
and middle-income countries. Worse, the WHO predicts a
100% increase in diabetes mortality over the next two decades.1
Approximately 85% of patients with T2DM are overweight or
obese, a key factor underlying the development and
maintenance of insulin resistance.2 Because comorbidities
from diabetes and obesity include stroke, cardiovascular disease,
and some forms of cancer, the cumulative cost from these
conditions is staggering.1
A clear correlation exists between abnormal fatty acid
metabolism and the development of insulin resistance.3,4
Patients who display insulin resistance also exhibit increased
skeletal muscle malonyl coenzyme A (m-CoA), a key building
block in fatty acid synthesis and a regulator of fatty acid
oxidation.5 The same insulin resistant patients also show
marked suppression of fatty acid oxidation and elevated rates of
de novo lipogenesis.3 This has been hypothesized to contribute
to the ectopic accumulation of hepatic and intramyocellular
triglycerides in this patient population. Because of its central
Screening of the Pfizer corporate compound collection,
optimization of the resulting hit series, and structure-based drug
Received: November 7, 2011
Published: December 8, 2011
© 2011 American Chemical Society
935
dx.doi.org/10.1021/jm201503u | J. Med. Chem. 2012, 55, 935−942

Journal of Medicinal Chemistry
Article
design efforts identified a spirochromanone series (1, Figure 1),
which was recently described in the literature.15 While 1 and
further deteriorated. Alternately, if polar functional groups were
simply appended to the periphery of the core, different
problems followed: notably compromised potency, poor
permeability, and poor oral bioavailability.
Lipophilicity is a key parameter that negatively influences
metabolic stability, human ether a-go-go related gene (hERG)
activity, solubility, and overall safety risk,18 all of which had
been concerns in the earlier spirochromanone series. In an
effort to break away from the original series, several less
lipophilic spiro-template structures were designed. These
structures had the potential to make the same key contacts
with ACC but contained additional polarity embedded into the
core of the template. One such designed structure (2) proved
10-fold less potent than the more optimized spirochromanone
1. On balance, however, when evaluating analogues in terms of
their overall intrinsic potential as leads, 2 is much more efficient
in terms of lipophilic ligand efficiency (LipE).19 This parameter
normalizes potency differences relative to a compound's
lipophilicity and allows for a clear comparison between
molecules. Despite its lower overall potency, 2 represents a
compound with higher LipE and thus a higher intrinsic
potential for further optimization to produce druglike
molecules. Chemistry was therefore developed to enable the
production of analogues in this chemical series (Scheme 1).
Following literature precedent, 4-hydroxy-3-acetyl pyrazoles
were prepared in one or two synthetic steps in a straightforward
manner.20 Variation of the nitrogen substituent on the pyrazole
was easily accomplished by choosing the appropriately
substituted hydrazine to derivatize the 1-position. The adjacent
position could be modified by choice of an appropriate 2-oxo-
aldehyde substrate. These 4-hydroxypyrazoles were subse-
quently reacted with suitably protected 4-piperidinone
analogues in a base-catalyzed spiro-cyclization.15 Acid-mediated
deprotection, followed by coupling with 7-methyl-1H-indazole-
5-carboxylic acid, or alternate carboxylic acid derivatives
provided the final compounds for testing. Structure−activity
relationship (SAR) analysis in this fused pyrazole series focused
from the outset on the optimization of LipE as a means of
maximizing potency at a given Log D and understanding the
effects of making small changes to the system.
Figure 1. ACC inhibitor lead structures 1 and 2. IC50 values are the
arithmetic mean of at least three independent replicates
SEM,
except where noted. Each individual IC50 value is generated from
single point inhibition data run in triplicate. Log D values are
calculated using ACD laboratories Log D at pH = 7.4. Because of
protein availability, rat ACC1 was used extensively as a surrogate for
human ACC1 in screening. Ultimately, comparisons to human ACC1
were made for a subset of analogues. These showed no notable
differences for all compounds tested.
related analogues had useful potency, a major challenge in
optimization of this series was a struggle with poor
physicochemical properties. Lipophilic compound 1 (Log D =
3.8)16 suffered from relatively high turnover in human liver
microsomal incubations (Clint = 68 mL/min/kg)17 and poor
aqueous solubility (19 μM at pH = 7.4). Although the addition
of lipophilicity improved potency, clearance and solubility
a
Scheme 1. General Synthesis of the Fused Pyrazole Core and Elaboration To Form Final Compounds
a
Conditions: (A) 2-Oxopropanal (1 equiv), AcOH, H2O, RT. (B) Glyoxal, H2O, reflux. (C) R2COCHO (6 equiv), AcOH, H2O, reflux. (D) tert-
Butyl 4-oxopiperidine-1-carboxylate, pyrrolidine, methanol. (E) HCl, dioxane. (F) 7-Methyl-1H-indazole-5-carboxylic acid, TEA, HATU/EDCI or 2-
chloro-4,6-dimethoxy-1,3,5-triazine, N-methyl morpholine. To prepare compound 2, methylhydrazine is used in step A and glyoxal is used in step B.
936
dx.doi.org/10.1021/jm201503u | J. Med. Chem. 2012, 55, 935−942

Journal of Medicinal Chemistry
Article
a
Table 1. Spiropyrazole SAR
IC50 (nM)
Compound Number
R1
R2
rACCl
hACC2
Log D
LipE
1
2
NA
Me
Me
Et
NA
H
25 1.0
370 99
280 45
62 19
48 5.7
650 150
470 180
63 14
37 15
45 7.0
23 4.5
ND
3.8
1.1
1.5
1.6
2.1
2.1
1.9
2.7
2.0
2.6
3.1
2.3
3.8
5.4
5.0
5.6
5.5
5.2
5.7
3.9
5.8
5.3
4.2
5.9
b
3
Me
H
4
5
Et
Me
H
28 2.5
48 5.6
22 1.7
260 56
15 0.60
14 0.60
46 4.3
7.0 0.60
6
n-Pr
iPr
Bn
7
H
8
H
9
Cyclobutyl
Cyclopentyl
Cyclohexyl
t-Bu
H
14 2.0
9.4 2.4
63 15
6.9 0.35
10
11
12
H
H
H
a
IC50 values are the arithmetic mean of at least three independent replicates SEM, except where noted. Each individual IC50 value is generated
from single-point inhibition data run in triplicate. Log D values are calculated using ACD labs Log D at pH = 7.4. LipE values are calculated using the
b
rat ACC1 IC50 value, according to the formula LipE = pIC50 − Log D. Arithmetic mean of two independent replicates.
The addition of a single methyl group to 2 to produce 3
(Table 1) had a modest, beneficial effect on potency but a
detrimental effect on LipE. For comparison, ethyl compound 4
provided an example of a better position to install a methyl
group, exhibiting 5-fold greater potency than 3 at roughly the
same Log D. Analogue 5 gave another example of improved
LipE relative to 3, due to its 10-fold increase in potency. This fit
the original hypothesis that careful incorporation of lipophilic
substituents into the proper vectors could ultimately provide
the desired potency enhancement. The comparison of 6 and 7
is also intriguing. Both contain three carbon atoms appended to
the pyrazole core, but a linear (n-propyl) arrangement as in 6
yields a less efficient analogue than the more compact
(isopropyl) orientation in 7. That trend continued in analogues
such as 8, where introduction of large, extended groups
afforded inefficient analogues due to decreased potency
(relative to smaller R1 substituents) and increased Log D.
Such groups were easily deprioritized for further SAR
development. Cyclic substituents were tolerated at R1, as
shown in compounds 9, 10, and 11, but consistent with other
trends, these compounds were progressively less efficient as the
ring size increased. While analogues 9 and 10 have virtually
identical potency, the 0.5 unit difference in log D is reflected in
the lower LipE of 10. The most potent and efficient pyrazole
substituent was the tert-butyl group (compound 12).
Because 5 demonstrated a significant potency breakthrough
while maintaining a high LipE, it was chosen as a representative
scaffold for SAR development in the amide “tailpiece” portion
of the molecule. Synthesis of the required amine intermediate
proceeded in good yields according to the general procedure
shown in Scheme 1. Subsequently, custom indazole acids were
coupled to this amine. A small subset of the analogues prepared
is shown in Table 2.
Translation of the SAR from the earlier spirochromanone
series to the new fused pyrazole series focused our efforts to
optimize LipE on 5,6-fused aryl systems. Specifically, interest
centered on 5-carboxy-indazole, 5-carboxy-indole, and 5-
carboxy-benzimidazole ring systems. Only modest changes in
potency were observed among the many analogues examined.
Methyl groups at either the 3-position of the indazole (13) or
at the 7-position (5, Table 1) seemed interchangeably active
and equally efficient. Installation of a methyl at both the 3- and
the 7-positions gave 14, a 12 nM compound that maintained
high LipE. Extension to an ethyl substituent at either the 3-
position (15) or the 7-position (16) gave analogues of lower
efficiency, indicating that the ethyl groups were not optimally
efficient in those vectors for ACC. The 7-ethoxy analogues (17
and 18) demonstrated that potency could be maintained
relative to 14. However, this came at a significant cost in
lipophilicity and resulted in very poor efficiency: despite their
potency, neither analogue represented an attractive avenue for
further optimization. Notably, unsubstituted indazole 19 was
approximately equal in LipE to both monomethyl indazole
compounds (5 and 13) and to dimethylindazole 14. Because 19
is 10-fold less potent than 14, an analysis focused primarily on
potency would have deprioritized the unsubstituted indazole
moiety. This LipE-based analysis, on the other hand, actually
highlighted the indazole moiety in 19 as an attractive option.
Removal of a single nitrogen atom from the dimethylindazole
system to produce dimethylindole 20 gave a familiar result:
enhanced potency at the expense of efficiency. Alternatively,
benzimidazoles 21 and 22 provided high LipE options, albeit at
modest potency.
To more precisely define the individual contributions of
various headgroups and tailpieces to the overall efficiency of the
final analogues, a Free-Wilson analysis was performed (Figure
2).21 Because attention had been focused on the lipophilic
efficiency of the analogues throughout the early SAR analysis,
the Free-Wilson methodology was utilized to assess relative
contributions to LipE rather than to potency, which is most
commonly used as an end point.22 This allowed for a pure,
direct assessment of fragment substructure LipE. Because
binding of the newer pyrazole series was well aligned with the
earlier spirochromanone series, examples of compounds from
937
dx.doi.org/10.1021/jm201503u | J. Med. Chem. 2012, 55, 935−942

Journal of Medicinal Chemistry
Article
a
Once the headgroup and tailpiece contributions were
determined, a leave-one-out (LOO) validation method was
employed on three different random samples (sets 1−3) of 350
analogues from the data set (Figure 3). The R2 values obtained
for the correlation of the actual and predicted values according
to this methodology were 0.87, 0.92, and 0.93. This indicated a
high degree of consistent behavior and additivity within these
series of ACC inhibitors. It also provided confidence in the
predictive power of this method for estimating the LipE of new
headgroup−tailpiece combinations.
Table 2. Indazole, Indole, and Benzimidazole SAR
Upon completion of this validation, focus shifted to LipE
prediction for virtual analogues (i.e., novel combinations of
headgroups and tailpieces) and the subsequent prioritization of
a subset of those analogues for synthesis (Figure 2). The full
matrix of potential combinations derived from 160 headgroups
and 300 tailpieces consists of 48 000 (160 × 300) compounds.
Mathematical recombination of head and tail groups allowed a
prediction of LipE for each of the resulting 48 000 virtual
analogues. Subsequent calculation of log D then allowed further
prioritization. Compounds of the most interest were those
predicted to have the highest LipE values, especially those with
very high LipE values in the log D range of 1.7−2.5, which
would provide the desired excellent potency in compounds
with attractive physical properties. On the basis of prior SAR
knowledge from both the spirochromanone and the spiropyr-
azole series, analogues in this log D range were the most likely
to be free of complications arising from high clearance, low
solubility, or low permeability.23 From these considerations, a
subset of 197 compounds was prioritized for synthesis, both to
identify the highest quality compounds and also to test the
accuracy of the Free-Wilson prediction.
Results from the 197 analogues prepared are shown in Figure
4. An arbitrary, but useful, threshold of 0.5 LipE units was
used to assess the overall accuracy of the prediction.24 In all,
83% of the analogues (164/197) were found to be within 0.5
LipE units of the Free-Wilson predicted value (blue circles,
Figure 4). For 33 compounds (17%), the actual LipE values
were outside of the 0.5 unit threshold (yellow squares, Figure
4), but the majority of these (20 compounds) were still within
0.75 units. While slightly less accurate, the predictions for
these 20 analogues still resulted in measured potency values
that were within 5-fold of the predicted potency values. Thus,
for 184/197 analogues prepared (93%), the Free-Wilson
methodology provided a LipE prediction that was quite
accurate and extremely useful (within 5-fold) in the context
of assessing potency and efficiency of virtual analogues. For the
most part, prediction accuracy remained quite high even for the
less efficient compounds (LipE < 5). This argued for a broad
utility in prediction within this data set and not simply a less
robust model that predicts good compounds well but struggles
to handle less efficient compounds.
a
IC50 values are the arithmetic mean of at least three independent
replicates SEMs, except where noted. Each individual IC50 value is
generated from single-point inhibition data run in triplicate. Log D
values are calculated using ACD labs log D at pH = 7.4. LipE values
are calculated using the rat ACC1 IC50 value, according to the formula
b
LipE = pIC50 − log D. Arithmetic mean of two independent
replicates. Result of a single IC50 determination; the SEM in this assay
is generally <20%.
c
The data for six advanced compounds in the desired property
space (log D = 1.7−2.5) are shown in Table 3. Not surprisingly,
given the large beneficial effect on LipE observed with the tert-
butyl group in 12 (Table 1), the tert-butyl moiety was present
as a headgroup substituent in most of these compounds. The
isopropyl group (23) and tetrahydrofuryl moiety (24) also
provided very efficient analogues. The more polar headgroups
of these two compounds gave products of moderate log D by
virtue of combination with the very lipophilic dimethylindazole
tailpiece. In terms of tailpieces, the majority of these six
analogues contained the indazole moieties previously described,
but a benzimidazole moiety was present in two compounds (25
both series were included in the Free-Wilson analysis. This
allowed for the mathematical examination of a large number of
tailpiece fragments that had only been synthesized in the
spirochromanone series, enabling an evaluation of their
suitability for incorporation into the fused pyrazole series. In
all, 948 analogues were computationally deconstructed into
their head and tail components. This provided 160 unique
headgroups and 300 unique tailpieces (Figure 2). The LipE
contribution of each fragment was then calculated according to
the Free-Wilson methodology.
938
dx.doi.org/10.1021/jm201503u | J. Med. Chem. 2012, 55, 935−942

Journal of Medicinal Chemistry
Article
Figure 2. Schematic depiction of the Free-Wilson analysis for calculation of contributions to LipE from 160 headgroups and 300 tailpieces.
Figure 3. (A) LOO validation of 350 randomly sampled analogues (set 1). A total of 165 analogues predicted from this set, R-Sq value = 0.87. (B)
LOO validation of 350 randomly sampled analogues (set 2). A total of 146 analogues predicted from this set, R-Sq value = 0.92. (C) LOO validation
of 350 randomly sampled analogues (set 3). A total of 167 analogues predicted from this set, R-Sq value = 0.93.
both 23 and to 1) was also beneficial for aqueous solubility
(300 μM at pH = 6.5), and permeability was maintained in an
RRCK cell line (10.9 × 10−6 cm s−1). Because the overall in
vitro profile of 28 warranted its progression to more advanced
in vivo assays, a larger scale synthesis was optimized to support
these studies.
For further analysis, 28 was examined in vivo for both
pharmacokinetic (PK) characteristics and efficacy in reduction
of hepatic and skeletal muscle malonyl-CoA. This compound
displayed a favorable rat PK profile (Clp = 26 mL/min/kg, Vdss
= 2.3 L/kg, t1/2 = 4.2 h, F = 50%, and rat plasma fraction
unbound = 55%). In rat pharmacodynamic studies, 28 blocked
the production of malonyl-CoA in vivo in both liver
(predominantly ACC1, Figure 5A) and muscle (measured in
quadriceps, predominantly ACC2, Figure 5B) after oral dosing.
Figure 4. Predicted vs experimental LipE for all 197 analogues selected
for synthesis. Compounds represented as blue circles (83%) are
analogues where the experimental value for LipE was found to be
The apparent ED50 in liver was calculated to be 1.6 mg/kg. In
quadriceps, the apparent ED50 was calculated to be 2.6 mg/kg.
Maximal inhibition of malonyl-CoA synthesis occurred at the
50 mg/kg dose and reached 78 and 75% in liver and muscle,
respectively.
within 0.5 units of the predicted value. Compounds represented as
yellow squares (17%) are analogues where the predicted and
experimental LipE values differed by more than 0.5 units.
and 26). This group had provided efficient but less potent
analogues in the earlier tailpiece SAR study (Table 2, 21 and
22). Notably, indazole regioisomer 27 proved slightly more
efficient than predicted by the Free-Wilson analysis (actual
LipE = 6.0 vs predicted LipE = 5.7). More importantly, the
indazole regioisomer used in 27 had been employed only
sporadically prior to incorporation into analogue 27 and had
never provided a compound that tested at better than 60 nM
potency for ACC1. An analysis focused on potency would
therefore have missed this fragment, but the Free-Wilson LipE
analysis identified this indazole regioisomer tailpiece as a
particularly valuable contributor to LipE. Although all
compounds were relatively stable as compared to earlier
lipophilic spirochromanone analogues, there seemed to be a
clear advantage in microsomal stability for dimethylindazole 23
and unsubstituted indazole 28. The polarity of 28 (relative to
CONCLUSION
■
In summary, we have outlined an optimization that began with
2, a weak ACC inhibitor lead, and culminated in 28. This
approach used a combination of subtle structural changes of
mainly aliphatic substituents around the headgroup pyrazole
ring and more substantial structural investigations around the
tailpiece amide portion of the compounds. Central to this effort
was a clear focus on lipophilicity and a determination to
identify the modifications, which led to the most efficient (not
simply the most potent) analogues. Application of Free-Wilson
analysis proved useful in this respect, allowing not only a clear
assessment of each group's respective contribution to LipE but
also a highly accurate prospective prediction of the LipE for
virtual compounds. This effort resulted in the design and
939
dx.doi.org/10.1021/jm201503u | J. Med. Chem. 2012, 55, 935−942

Journal of Medicinal Chemistry
Article
a
Table 3. Structure, Potency, LipE, and Microsomal Stability Data for Six Advanced Analogues
a
IC50 values are the arithmetic mean of at least three independent replicates SEMs, except where noted. Each individual IC50 value is generated
from single-point inhibition data run in triplicate. Log D values are calculated using ACD labs log D at pH = 7.4. Actual LipE values are calculated
b
using the rat ACC1 IC50 value, according to the formula LipE = pIC50 − log D. Arithmetic mean of two independent replicates.
Figure 5. Bar graphs of oral dose vs hepatic malonyl Co-A levels (percent relative to control, left panel) and oral dose vs quadriceps malonyl Co-A
levels (percent relative to control, right panel).
spectra were recorded on a Waters (Waters Corp., Milford, MA)
Micromass Platform Il spectrometer. Unless otherwise specified, mass
spectra were recorded on a Waters Micromass Platform Il
spectrometer. Flash chromatography was performed according to the
method described by Still et al. (J. Org. Chem., 1978, 43, 2923). All
Biotage purifications, discussed herein, were performed using either a
40M or 40S Biotage column containing KP-SIL silica (40−63 μM, 60
Å) (Biotage AB; Uppsala, Sweden). All Combiflash purifications,
discussed herein, were performed using a CombiFlash Companion
system (Teledyne Isco, Lincoln, NE) utilizing packed RediSep silica
columns. All tested compounds were determined to be >95% pure by
high-performance liquid chromatography (HPLC). The compounds
and intermediates described below were generally named according to
the IUPAC (International Union for Pure and Applied Chemistry)
synthesis of 28, a novel ACC1/2 inhibitor that advanced to
regulatory toxicity studies as a potential treatment for diabetes.
EXPERIMENTAL SECTION
■
Materials and Methods. Reagents and solvents were of
commercial quality and were used without further purification unless
otherwise noted. Proton NMR chemical shifts are given in parts per
million downfield from tetramethylsilane and were recorded on a
Varian Mercury Plus 300 MHz, Varian Unity 400 or 500 MHz
(megaHertz) spectrometer (Varian Inc.; Palo Alto, CA). NMR
chemical shifts are given in parts per million downfield from
tetramethylsilane (for proton) or fluorotrichloro-methane (for
fluorine). Multiplicities are given as s (singlet), bs (broad singlet), d
(doublet), t (triplet), dt (double of triplets), and m (multiplet). Mass
940
dx.doi.org/10.1021/jm201503u | J. Med. Chem. 2012, 55, 935−942

Journal of Medicinal Chemistry
Article
recommendations on Nomenclature of Organic Chemistry and the
CAS Index rules.
was concentrated to afford a yellow solid, which was triturated with
ether for 1 h. The solid was filtered and dried in vacuo to afford a
white solid; mp 257 °C. Anal. calcd for C22H25N5O3: C, 64.85; H,
6.18; N, 17.19. Found: C, 64.59; H, 6.40; N, 16.87.
HPLC Analytical Conditions. Method LC-1: column, Waters
ACQUITY Ultra Performance LC BEH C18 column, 2.1 mm × 30
mm, 1.7 μm, 0.05% TFA 95/5 to 5/95 water/acetonitrile; flow rate,
1.3 mL/min; run time, 1.1 min. Method LC-2: column, Waters XTerra
C18 4.6 mm × 50 mm, 3.5 μm column. Solvents A and B were Water
w/0.1% TFA and acetonitrile w/0.1% TFA, respectively. Nine minute
total method time with 5−95% B over 5.83 min. Mass spectral data
were acquired from 180 to 850 amu in electrospray positive mode.
The flow rate was 2.0 mL/min.
ASSOCIATED CONTENT
* Supporting Information
■
S
Experimental details for the synthesis of key intermediates as
well as compounds 1−27. This material is available free of
charge via the Internet at http://pubs.acs.org.
Rat ACC1 Assay Description. Rat ACC1 was purified from rat
liver as previously described.14 For ACC1 inhibition studies, test
compounds were dissolved in dimethyl sulfoxide (DMSO), serially
diluted in DMSO, and were added in replicate to 96-well plates. A
comparable volume of DMSO was added to control wells. The enzyme
was activated for 30 min at 37 °C in buffer containing: 50 mM HEPES
(pH 7.5), 10 mM MgCl2, 10 mM tripotassium citrate, 6 mM DTT,
0.75 mg/mL BSA, and 0.8 μg/mL ACC prior to addition to assay
wells. After a 10 min enzyme−compound preincubation, the reaction
was initiated at room temperature in a fume hood by the addition of
the substrate solution (containing 2.4 mM acetyl-CoA, 38.4 mM
KHCO3, 1.6 mM 14C-NaHCO3, and 8.0 mM ATP). The final assay
volume of 100 μL per well consisted of 46 mM HEPES (pH 7.5), 7.5
mM MgCl2, 7.5 mM tripotassium citrate, 2.8 mM DTT, 0.5 mg/mL
BSA, 2.0 mM ATP, 600 μM acetyl-CoA trilithium salt, 9.6 mM
potassium bicarbonate, 0.6 μg/mL ACC1, and 0.4 mM 14C-NaHCO3
(58 mCi/mmol). After 20 min, the reaction was terminated by the
addition of 3 N hydrochloric acid (HCl) with the concomitant
liberation of nonreacted NaHCO3 as CO2. Plates were dried overnight
at 50 °C to allow complete 14C−CO2 liberation. The following day,
water and OptiPhase Supermix liquid scintillation fluid were added to
the dried wells now containing 14C-malonyl-CoA. Following vigorous
shaking, the amount of 14C in each assay well was quantified using
Microbeta LSC luminescence counter.
Human ACC2 Assay Description. Human ACC2 was purified,
and hACC2 inhibition assays were performed as previously
described.15 Briefly, ACC2 was expressed in CHO cells, and
recombinant ACC2 was isolated using soft-link Avidin (Promega,
Madison, WI). ACC2 inhibition assays were performed using the
Transcreener ADP detection FP assay kit (Bellbrook Laboratories,
Madison, WI) using the manufacturers' conditions and read using an
Envision Fluorescence reader (Perkin-Elmer) using a 620 excitation
Cy5 FP general dual mirror, 620 excitation Cy5 FP filter, 688 emission
(S), and a 688 (P) emission filter.
Synthesis of 2′-(tert-Butyl)-1-(1H-indazole-5-carbonyl)-2′H-
spiro[piperidine-4,5′-pyrano[3,2-c]pyrazol]-7′(6′H)-one (28). A
mixture of 1H-indazole-5-carboxylic acid (2.70 g, 16.7 mmol), 2-
chloro-4,6-dimethoxy-1,3,5-triazine (2.93 g, 16.7 mmol), and N-
methylmorpholine (1.72 g, 16.7 mmol) in DMF (98.1 mL) was
stirred for 1 h at room temperature. 2′-(tert-Butyl)-2′H-spiro-
[piperidine-4,5′-pyrano[3,2-c]pyrazol]-7′(6′H)-one hydrochloride (5.0
g, 17 mmol) and an additional quantity of N-methylmorpholine (3.44
g, 33.3 mmol). The reaction mixture was then stirred at ambient
temperature for 1.5 h. The mixture was filtered to remove insoluble
solids. Then, aqueous ammonium chloride was added to the filtrate,
and the mixture was extracted with EtOAc. The layers were separated,
and the aqueous layer was extracted with EtOAc (3×). The combined
organic layers were washed with water and saturated aqueous sodium
chloride, dried (Na2SO4), filtered, and concentrated under reduced
pressure. The crude product was the purified via medium pressure
chromatography (SiO2) eluting with a gradient of chloroform:metha-
nol (100:0 to 97:3) to deliver 6.8 g (89%) of 28 as a pale yellow foam.
1H NMR (500 MHz, DMSO-d6) δ 1.51 (s, 9 H), 1.72-1.78 (m, 2 H),
AUTHOR INFORMATION
Corresponding Author
*Tel: 858-526-4844. E-mail: kevin.freeman-cook@pfizer.com.
■
Present Address
†Pfizer Worldwide Research and Development, 10770 Science
Center Drive, San Diego, California 92121.
ACKNOWLEDGMENTS
■
We thank Katherine Brighty, Martin Edwards, Robert Hoffman,
and David Griffith for editorial suggestions and help with
proofreading.
ABBREVIATIONS USED
■
WHO, World Health Organization; T2DM, type 2 diabetes
mellitus; ACC1, acetyl-CoA carboxylase 1; ACC2, acetyl-CoA
carboxylase 2; LipE, lipophilic ligand efficiency; m-CoA,
malonyl coenzyme A; ATP, adenosine triphosphate; SEM,
standard error of measurement; hERG, human ether a-go-go
related gene; AcOH, acetic acid; HCl, hydrochloric acid; TEA,
triethyl amine; HATU, 2-(1H-7-azabenzotriazol-1-yl)-1,1,3,3-
tetramethyl uronium hexafluorophosphate methanaminium;
EDCI, 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide; SAR,
structure−activity relationships; PK, pharmacokinetic
REFERENCES
■
(1) World Health Organization fact sheets. http://www.who.int/
mediacentre/factsheets/fs312/en/index.html (accessed July 18, 2011).
(2) Mokdad, A. H.; Ford, E. S.; Bowman, B. A.; Dietz, W. H.; Vinicor,
F.; Bales, V. S.; Marks, J. S. Prevalence of obesity, diabetes, and
obesity-related health risk factors, 2001. J. Am. Med. Assoc. 2003, 289
(1), 76−79.
(3) Savage, D. B.; Petersen, K. F.; Shulman, G. I. Disordered lipid
metabolism and the pathogenesis of insulin resistance. Physiol. Rev.
2007, 87 (2), 507−520.
(4) Shulman, G. I. Cellular mechanisms of insulin resistance. J. Clin.
Invest. 2000, 106 (2), 171−176.
(5) Bandyopadhyay, G. K.; Yu, J. G.; Ofrecio, J.; Olefsky, J. M.
Increased Malonyl-CoA levels in muscle from obese and type 2
diabetic subjects lead to decreased fatty acid oxidation and increased
lipogenesis; thiazolidinedione treatment reverses these defects.
Diabetes 2006, 55 (8), 2277−2285.
(6) Bourbeau, M. P.; Allen, J. G.; Gu, W. Recent advances in acetyl-
CoA carboxylase inhibitors. Annu. Rep. Med. Chem. 2010, 45, 95−108.
(7) Chonan, T.; Tanaka, H.; Yamamoto, D.; Yashiro, M.; Oi, T.;
Wakasugi, D.; Ohoka-Sugita, A.; Io, F.; Koretsune, H.; Hiratate, A.
Design and synthesis of disubstituted (4-piperidinyl)-piperazine
derivatives as potent acetyl-CoA carboxylase inhibitors. Bioorg. Med.
Chem. Lett. 2010, 20 (13), 3965−3968.
(8) Corbett, J. W. Review of recent acetyl-CoA carboxylase inhibitor
patents: mid-2007 - 2008. Expert Opin. Ther. Pat. 2009, 19 (7), 943−
956.
1.95 (br. s., 2 H), 2.75 (s, 2 H), 3.18 (br. s., 2 H), 4.23 (br.s., 2 H),
7.40 (dd, J = 8.5, 1.5 Hz, 1 H), 7.58 (d, J = 8.5 Hz, 1 H), 7.81 (s, 1 H),
7.86 (s, 1 H), 8.14 (s, 1 H), 13.26 (br. s., 1 H). MS (ES+) m/z 408
(MH+). HPLC (method LC-1) retention time = 0.38 min. An
analytical sample could be obtained by dissolving the solid in
dichloromethane and filtering through a plug of silica gel. The filtrate
(9) Corbett, J. W.; Harwood, J. H. Jr. Inhibitors of mammalian acetyl-
CoA carboxylase. Recent Pat. Cardiovasc. Drug Discovery 2007, 2 (3),
162−180.
941
dx.doi.org/10.1021/jm201503u | J. Med. Chem. 2012, 55, 935−942

Journal of Medicinal Chemistry
Article
(10) Shinde, P.; Srivastava, S. K.; Odedara, R.; Tuli, D.; Munshi, S.;
Patel, J.; Zambad, S. P.; Sonawane, R.; Gupta, R. C.; Chauthaiwale, V.;
Dutt, C. Synthesis of spiro[chroman-2,4′-piperidin]-4-one derivatives
as acetyl-CoA carboxylase inhibitors. Bioorg. Med. Chem. Lett. 2009, 19
(3), 949−953.
(11) Chonan, T.; Oi, T.; Yamamoto, D.; Yashiro, M.; Wakasugi, D.;
Tanaka, H.; Ohoka-Sugita, A.; Io, F.; Koretsune, H.; Hiratate, A. 4-
Piperidinyl)-piperazine: A new platform for acetyl-CoA carboxylase
inhibitors. Bioorg. Med. Chem. Lett. 2009, 19 (23), 6645−6648.
(12) Savage, D. B.; Choi, C. S.; Samuel, V. T.; Liu, Z.-X.; Zhang, D.;
Wang, A.; Zhang, X.-M.; Cline, G. W.; Yu, X. X.; Geisler, J. G.; Bhanot,
S.; Monia, B. P.; Shulman, G. I. Reversal of diet-induced hepatic
steatosis and hepatic insulin resistance by antisense oligonucleotide
inhibitors of acetyl-CoA carboxylases 1 and 2. J. Clin. Invest. 2006, 116
(3), 817−824.
(13) Polakis, S. E.; Guchhait, R. B.; Zwergel, E. S.; Lane, M. D.;
Cooper, T. G. Acetyl coenzyme A carboxylase system of Escherichia
coli. Mechanisms of the biotin carboxylase and carboxyltransferase
catalyzed reactions. J. Biol. Chem. 1974, 249 (20), 6657−67.
(14) Harwood, H. J. Jr.; Petras, S. F.; Shelly, L. D.; Zaccaro, L. M.;
Perry, D. A.; Makowski, M. R.; Hargrove, D. M.; Martin, K. A.; Tracey,
W. R.; Chapman, J. G.; Magee, W. P.; Dalvie, D. K.; Soliman, V. F.;
Martin, W. H.; Mularski, C. J.; Eisenbeis, S. A. Isozyme-nonselective
N-Substituted Bipiperidylcarboxamide Acetyl-CoA Carboxylase Inhib-
itors Reduce Tissue Malonyl-CoA Concentrations, Inhibit Fatty Acid
Synthesis, and Increase Fatty Acid Oxidation in Cultured Cells and in
Experimental Animals. J. Biol. Chem. 2003, 278 (39), 37099−37111.
(15) Corbett, J. W.; Freeman-Cook, K. D.; Elliott, R.; Vajdos, F.;
Rajamohan, F.; Kohls, D.; Marr, E.; Zhang, H.; Tong, L.; Tu, M.;
Murdande, S.; Doran, S. D.; Houser, J. A.; Song, W.; Jones, C. J.;
Coffey, S. B.; Buzon, L.; Minich, M. L.; Dirico, K. J.; Tapley, S.;
McPherson, R. K.; Sugarman, E.; James Harwood, H.; Esler, W.
Discovery of small molecule isozyme non-specific inhibitors of
mammalian acetyl-CoA carboxylase 1 and 2. Bioorg. Med. Chem. Lett.
2010, 20 (7), 2383−2388.
(16) This Log D value (and all others in this manuscript) are
calculated Log D values, using ACD laboratories software (at pH =
7.4).
(17) Obach, R. S. The prediction of human clearance from hepatic
microsomal metabolism data. Curr. Opin. Drug Discovery Dev. 2001, 4
(1), 36−44.
(18) Hughes, J. D.; Blagg, J.; Price, D. A.; Bailey, S.; DeCrescenzo, G.
A.; Devraj, R. V.; Ellsworth, E.; Fobian, Y. M.; Gibbs, M. E.; Gilles, R.
W.; Greene, N.; Huang, E.; Krieger-Burke, T.; Loesel, J.; Wager, T.;
Whiteley, L.; Zhang, Y. Physiochemical drug properties associated with
in vivo toxicological outcomes. Bioorg. Med. Chem. Lett. 2008, 18 (17),
4872−4875.
(19) Edwards, M. P.; Price, D. A. Role of physicochemical properties
and ligand lipophilicity efficiency in addressing drug safety risks. Annu.
Rep. Med. Chem. 2010, 45, 381−391.
(20) Begtrup, M.; Nytoft, H. P. Reactions of Glyoxals with
Gydrazones: A New Route to 4-Hydroxypyrazoles. J. Chem. Soc.,
Perkin Trans. 1 1985, 1, 81−86.
(21) Free, S. M. Jr.; Wilson, J. W. A Mathematical Contribution to
Structure-Activity Studies. J. Med. Chem. 1964, 7 (4), 395−399.
(22) Patel, Y.; Gillet, V. J.; Howe, T.; Pastor, J.; Oyarzabal, J.; Willett,
P. Assessment of Additive/Nonadditive Effects in Structure-Activity
Relationships: Implications for Iterative Drug Design. J. Med. Chem.
2008, 51 (23), 7552−7562.
(23) The calculation of Log D also allowed a calculation of predicted
potency (predicted pIC50 = predicted LipE + log D).
(24) This level of accuracy for predicted LipE on a logarithmic scale
corresponds to a 3-fold accuracy in potency prediction.
942
dx.doi.org/10.1021/jm201503u | J. Med. Chem. 2012, 55, 935−942