Solubility-driven optimization of phosphodiesterase-4 inhibitors leading to a clinical candidate

Article abstract of DOI:10.1021/jm300459a

The solubility-driven optimization of a series of 1,7-napthyridine phosphodiesterase-4 inhibitors is described. Directed structural changes resulted in increased aqueous solubility, enabling superior pharmacokinetic properties with retention of PDE4 inhib

Full text of DOI:10.1021/jm300459a

Article
pubs.acs.org/jmc
Solubility-Driven Optimization of Phosphodiesterase‑4 Inhibitors
Leading to a Clinical Candidate
Neil J. Press,*,† Roger J. Taylor,† Joseph D. Fullerton, Pamela Tranter,† Clive McCarthy,
Thomas H. Keller, Nicola Arnold, David Beer,∥ Lyndon Brown, Robert Cheung, Julie Christie,
Alastair Denholm, Sandra Haberthuer, Julia D. I. Hatto,† Mark Keenan, Mark K. Mercer, Helen Oakman,†
Helene Sahri, Andrew R. Tuffnell, Morris Tweed, John W. Tyler,† Trixie Wagner,‡ John R. Fozard,§
and Alexandre Trifilieff‡
†Novartis Institutes for Biomedical Research, Wimblehurst Road, Horsham, West Sussex RH12 5AB, U.K.
‡Novartis Institutes for Biomedical Research, Lichtstrasse 35, CH-4056 Basel, Switzerland
§Novartis Distinguished Scientist, Novartis Institutes for Biomedical Research, Lichtstrasse 35, CH-4056 Basel, Switzerland
∥Novartis Institute for Tropical Diseases, 10 Biopolis Road, 05-01 Chromos, 138670 Singapore
S
* Supporting Information
ABSTRACT: The solubility-driven optimization of a series of
1,7-napthyridine phosphodiesterase-4 inhibitors is described.
Directed structural changes resulted in increased aqueous
solubility, enabling superior pharmacokinetic properties with
retention of PDE4 inhibition. A range of potent and orally
bioavailable compounds with good in vivo efficacy in animal
models of inflammation and reduced emetic potential
compared to previously described drugs were synthesized.
Compound 2d was taken forward as a clinical candidate for the
treatment of COPD.
We have previously described the synthesis of a range of
PDE4 inhibitors around the 1,7-naphthyridine scaffold. Primary
among these was compound NVP-ABE171 (4, Figure 1), which
INTRODUCTION
■
Phosphodiesterase-4 (PDE4) has been an important therapeu-
tic target for some time, with a strong rationale for the potential
of an inhibitor to be of therapeutic benefit in a wide range of
disease states.1,2 PDE4 is mainly expressed in inflammatory
cells, where it catalyzes the degradation of cAMP to AMP.
Inhibiting this process is expected to result in an accumulation
of intracellular cAMP, which is known to lead to a broad range
of anti-inflammatory effects, making this an attractive target for
inter alia respiratory diseases such as asthma and chronic
obstructive pulmonary disease (COPD).3−5 Indeed, drugs such
as roflumilast and cilomilast have proven to be clinically
beneficial in these disorders.6 One reason why so many
published PDE4 inhibitors have failed so far to be progressed to
market is their propensity to show a high side effect profile
(initially nausea and emesis) and low therapeutic index. While
several theories have been published suggesting ways to design
molecules in order to avoid side effects (ranging from
selectively targeting a particular isoform such as PDE4B to
avoiding the high affinity rolipram binding site (HARBS)), it is
true to say that there is no current consensus as to the validity
of any of these.7a Alternative approaches could utilize delivery
methods focused on obtaining peripheral selectivity, thereby
avoiding putative CNS-generated nausea effects, for example,
by the introduction of hydrophilic moieties, or local
administration by inhalation.7b
Figure 1. Structure and properties of 4.
was shown to be a potent anti-inflammatory agent in in vivo
models of inflammation.8−10 Although 4 was efficacious when
dosed orally in the inflammation model, the poor pharmaco-
kinetics and low solubility of the compound precluded further
development. The project therefore continued with the specific
aim of producing a PDE4 inhibitor with improved solubility
Received: April 2, 2012
Published: August 13, 2012
© 2012 American Chemical Society
7472
dx.doi.org/10.1021/jm300459a | J. Med. Chem. 2012, 55, 7472−7479

Journal of Medicinal Chemistry
Article
Scheme 1. Synthetic Scheme for the Preparation of Key Intermediate 1
Scheme 2. Elaboration of Key Intermediate 1
and pharmacokinetics. Furthermore, it was felt that a
compound with a particularly reproducible and predictable
pharmacokinetic profile (enabled by high solubility) would be
of great benefit in this project, where we anticipated that the
therapeutic index would be low.
and 2. Thus 6-amino-8-bromo-1,7-naphthyridine is synthesized
in three steps from commercially available pyridine-3-
ylacetonitrile. From this intermediate, Suzuki coupling proceeds
smoothly at the 8-position to give a range of substituted
aromatics, which were used as replacements for the
benzoxadiazole moiety of 4 (Scheme 1). Diazotization followed
by substitution with triflate gave key intermediate 1 (with
various ‘R’), which we were able to elaborate at the 6-position
either through Negishi coupling to form the carbocyclic series
2, or through direct displacement to give the amino series 3
(Scheme 2). The former reaction is an elegant example of the
coupling of a heteroaromatic triflate to an alkyzincate, allowing
rapid access to a series of carbon-linked analogues. The
cyclohexyl iodide was synthesized from the alcohol as a mixture
of geometric isomers. However, the ratio of isomers was not
important, as after the ester hydrolysis step the trans-
stereochemistry was observed exclusively on the cyclohexyl
ring; this geometry was proven by nuclear magnetic resonance
analysis and by the X-ray structure determination of several
analogues (see for example Figure 2).11
RESULTS AND DISCUSSION
■
We decided to keep the 1,7-naphthyridine core scaffold and
attempt to improve solubility by breaking up what we saw as
the very flat, highly aromatic structure of 4. We hypothesized
that removal of two aromatic rings was feasible: the oxadiazole
ring (replacing it with m-electron withdrawing groups) and the
benzoic acid moiety (replacing, for example, with a cyclohexyl
group). These changes were proposed in order to introduce a
more three-dimensional morphology to the compound, the aim
being to weaken the crystal structure of 4 and improve
dissolution rate. We were keen to maintain the presence of the
carboxylic acid in the molecule both as a handle for salt
formation, and also with an aim to limiting the penetration of
the compound into the central nervous system, it being
considered that this would help to reduce the nausea/emesis
side effect profile.
Table 1 contains the results of several representative
compounds from each series in our PDE4 assays. Pleasingly,
replacement of the benzoic acid moiety of 4 with either the
saturated analogue or the piperidine analogue resulted in
The synthesis of the compounds in two series (6-amino- or
6-carbo-substituted 1,7-naphthyridines) is shown in Schemes 1
7473
dx.doi.org/10.1021/jm300459a | J. Med. Chem. 2012, 55, 7472−7479

Journal of Medicinal Chemistry
Article
Also shown in Table 1 are results from an assay measuring
inhibition of TNF-α release from stimulated human peripheral
blood mononuclear cells (PBMCs); this is significant because it
demonstrates that these acids are able to permeate the cell wall
in order to act intracellularly. It is seen that the trend for the 6-
cyclohexyl series to show highest inhibition of PDE4 activity is
paralleled in this functional system. In this case, the difference
between the two series is more obvious and led to the
deprioritization of the piperidyl series due to the critical nature
of achieving cellular activity (compare 2b with 3b, 2d with 3d,
and 2f with 3f). This may be due to inherently lower affinity to
the enzyme of the piperidinyl (3) series combined with a
reduced cell permeability of that series (cLogP of series 3
compared with 2 is generally 1 order of magnitude lower, with a
higher polar surface area). The cell penetration hypothesis was
supported with passive diffusion permeability data from the
parallel artificial membrane permeability assay (PAMPA, Table
1), which clearly shows poorer permeability for the piperidyl
class (3) compared to cyclohexyl (2).
Figure 2. Structure of compound 2b (R = CN) in the crystal, showing
the trans-stereochemistry of the substituents at the cyclohexyl ring.11
compounds which were still inhibitors of the various PDE4
isoforms. The cyclohexyl series generally showed higher affinity
across all isoforms than the piperidinyl analogues. As expected
from our previous SAR knowledge, substitution on the 8-
phenyl group with a m-electron withdrawing group generally
led to good affinity at PDE4, with minor differences in rank
order according to substituent in each series.
Crucially, these structural modifications led to a considerable
and general increase in solubility compared to the highly
unsaturated 4 (Table 1). We believe that the elimination of two
aromatic rings in new compound series 2 and 3, and the
introduction of sp3 structures by way of a cyclohexyl or
piperidyl group, has resulted in the breakup of the flat,
ab
,
Table 1. In Vitro Activity of PDE4 Inhibitors
a
b
The data represent the mean of at least two separate experiments, each performed in duplicate. Error limits represent standard errors of the mean
c
d
(SEM) and are shown for those experiments performed at least n = 3 times. TNF-α release measured after stimulation with LPS. n.d. = not
determined. Measured at pH6.8. Parallel artificial membrane permeability assay. For method, see ref 17.
e
f
7474
dx.doi.org/10.1021/jm300459a | J. Med. Chem. 2012, 55, 7472−7479

Journal of Medicinal Chemistry
Article
a
Table 2. Pharmacokinetic Parameters of PDE4 Inhibitors in the Rat
iv Vss (L/kg)
iv Cl (mL/min/kg)
po Cmax (uM)
po Tmax (min)
po AUCtot (μM·min)
po T1/2 (h)
F (%)
b
c
2b
2d
0.25 0.06
0.09 (n = 2)
0.36 (n = 2)
0.72 (n = 2)
0.24 (n = 2)
0.25 0.02
0.37 0.03
0.22 0.1
0.64 (n = 2)
0.65 (n = 2)
1.16 (n = 2)
2.07 (n = 2)
0.98 0.09
2.34 0.39
8.3 1.4
4.6 1.0
3.6 0.5
2.8 0.6
4.82 0.97
4.49 0.51
2.4 1.4
1500 990
200.0 51.8
445.0 200.9
129.0 37.2
172.5 104.2
172.50 48.34
210 87
112900 42700
4408 799.0
5039 849
2768 566
3919 1655
3267 361
523 140
193
9
5
1
92 18
52
9
d
2e
29 18
107 14
158 32
70 12
e
3b
3d
23
17
7
5
6
1
7
f
g
3e
4
149 17
12
8
3
a
b
Error limits are shown as SEM where n ≥ 3. Compound dosed as the potassium salt. Vehicle (iv and po): PBS to 4.2 μM/mL dosed at 4.2
c
d
μmol/kg. Compound dosed as the potassium salt. Vehicle (iv and po): PBS to 2.1 μM/mL dosed at 2.1 μmol/kg. Compound dosed as the
potassium salt. Vehicle: iv, PBS to 1.05 μmol/mL dosed at 1.05 μ mol/kg; po, PBS to 2.1 μmol/mL dosed at 2.1 μmol/kg. Vehicle: iv, 25/75
PEG200/PBS to 1.05 μmol/mL dosed at 1.05 μmol/kg; po, 25/75 PEG200/PBS to 2.1 μmol/mL dosed at 2.1 μmol/kg. Compound dosed as the
e
f
sodium salt. Vehicle: iv, 25/75 PEG200/PBS to 2.1 μmol/mL dosed at 2.1 μmol/kg; po, 25/75 PEG200/PBS to 4.2 μmol/mL dosed at 4.2 μmol/
g
kg. Compound dosed as the sodium salt. Vehicle: iv, PBS to 1.05 μmol/mL dosed at 1.05 μmol/kg; po, PBS to 2.1 μmol/mL dosed at 2.1 μmol/kg.
Figure 3. Effects of compounds 2d, 2e, and roflumilast on the pulmonary inflammatory response in the airways measured in samples of BAL fluid
recovered 24 h after antigen challenge in actively sensitized brown Norway rats. Compounds or vehicle (PEG200:saline 1:1) were given orally 2 h
prior to antigen challenge. V = challenge with saline; OA = challenge with ovalbumin. Results are expressed as mean values SEM of 5−10 animals
per group. *p < 0.05 indicates significant difference by comparison with vehicle-treated and OA challenged animals.
Figure 4. Effect of 2d and roflumilast on inhibition of LPS-induced lung inflammation in mice. One hour before the LPS challenge, mice were
treated via the oral route with vehicle, PEG200:saline 1:1 (LPS) or increasing doses of 2d or roflumilast (1−30 mg/kg). As a control, a group of
animals was treated with vehicle and challenged with PBS (PBS). Animals were killed at 3 h postchallenge. Results are from two experiments and are
expressed as individual data and means (horizontal bars) of 6−15 mice per group. Significance indicated by * (p < 0.05) was determined using a
Mann−Whitney test against the vehicle-treated and LPS-challenged animals (LPS).
lipophilic ABE171 and disrupted the regular, ordered crystal
packing structure. In this way, solubility of the new compounds
is vastly improved compared to our starting point. Piperidyl
series 3 has CLogP at least 1 log unit less than the
corresponding cyclohexyl derivative (e.g., CLogP 3d = 3.198,
CLogP 2d = 4.28), helping the solubility and providing a basic
center. The compound which stands out as having surprisingly
low solubility is example 2e. The reason behind this is not
entirely clear, however this compound does have a high ClogP
at 5.26, and perhaps the electronic effects of the trifluor-
omethoxy group in this example are such that solubility at this
critical pH happens to be low. This generally excellent
improvement in solubilities must have been a strong factor in
the concomitant improvement in pharmacokinetic properties in
the rat. As shown in Table 2, these new compounds typically
showed high bioavailability and excellent exposure, in contrast
to 4 (Figure 1).
There were no obvious trends in pharmacokinetics when
looking across the cyclohexyl or piperidinyl series (compare 2b
with 3b and 2d with 3d). For example, although the half-life of
7475
dx.doi.org/10.1021/jm300459a | J. Med. Chem. 2012, 55, 7472−7479

Journal of Medicinal Chemistry
Article
2b was exceptionally long, and this was reduced by going to the
piperidinyl derivative 3b, there was no reduction in half-life of
3d compared to 2d.
Given the excellent in vitro affinity of the PDE4 inhibitors
described, it was decided to evaluate selected compounds
further in in vivo models of disease.
Effect of PDE4 Inhibitors on the Pulmonary Inflam-
matory Response in Sensitized Brown Norway (BN) Rats
Following Challenge with Inhaled Ovalbumin (OA).
Given the data so far on the PDE4 inhibitors, several
compounds were selected for in vivo efficacy studies. The
OA challenge of actively sensitized BN rats results in a marked
increase in eosinophil numbers in bronchoalveolar lavage
(BAL) fluid, obtained 24 h following antigen exposure. This
model is a well characterized and recognized model for
asthma.12 As shown in Figure 3, oral administration of
compounds 2d and 2e (0.3, 1, and 3 mg/kg) given 2 h prior
to antigen challenge induced dose-dependent inhibition of the
numbers of eosinophils present in BAL fluid obtained from
actively sensitized BN rats, 24 h after OA challenge. The ED50
values against eosinophilia were estimated at approximately 1.2
and 2 mg/kg for compounds 2d and 2e, respectively. The
known PDE4 inhibitor roflumilast was used as a benchmark in
this experiment and seen to have ED50 = 0.74 mg/kg.
Inhibition of Lipopolysaccharide (LPS)-Induced Lung
Inflammation in Mice. COPD is characterized by an increase
in the activation of and/or numbers of macrophages and
neutrophils Therefore, it was important to test compounds in a
macrophage-dependent model. To do so, we used the LPS-
induced lung inflammation model in BALB/c mice, which has
been shown to be dependent upon macrophage activation.13
When given orally, one hour before the LPS challenge, 2d
inhibited neutrophil influx into the BAL fluid with an ED50
Figure 5. Effect of PDE4 inhibitors in a ferret model of emesis.
Compounds were given orally, in a cumulative fashion, every 4 h in
PEG200/saline (1/1) and animals observed for signs of emesis. Data
are expressed as mean SEM of 4−8 ferrets.
As can be seen from Table 3, the maximum concentration
and area under the curve indicate good absorption and
exposure of compounds 2d and 2e. The emetic events occurred
at the time when there was maximum concentration of 2d,
however, there was no distinction in the Tmax between
individual animals that did or did not exhibit emesis. The
half-life did not appear to be affected by the emetic episodes
noted, although the area under the curve (extrapolated) was
lower for individual animals that exhibited emesis. No plasma
levels were detected for roflumilast or for its possible N-oxide
metabolite. Possible explanations could be that in the ferret
roflumilast is metabolized quickly to a metabolite other than
the N-oxide or that roflumilast is not absorbed to a great extent.
However, animals did show emetic events with a very low
systemic exposure of this compound.
value of 11.2
1.9 mg/kg. Roflumilast was as potent as
compound 2d, with an ED50 value of 12.4 2.9 mg/kg (Figure
4).
The potent activity of the PDE4 inhibitors in attenuating
both ovalbumin- and LPS-driven lung inflammation in these
experiments is supportive of their having therapeutic potential
in asthma and COPD.
It is surprising and of great interest that compound 2e did
not show any emesis in the ferret model despite high systemic
exposure. There was no obvious rationale as to why this was the
case; certainly, the PDE4 receptor subtype profile is very similar
between 2d and 2e (although only the human isorforms were
tested), as is the ferret pharmacokinetic profile. Other
possibilities could be differences in plasma protein binding
between the two compounds and differences in brain
penetration; however, these parameters were not measured in
the ferret.
The aim of this project was to take an established PDE4
inhibitor with good activity on target and to improve the
pharmacokinetic and solubility profile in order to produce a
developable drug candidate. We have shown that through
targeted chemical modifications it is possible to improve the
physicochemical properties of molecules while retaining
potency on target, resulting in more “drug-like” and
pharmaceutically developable compounds. A series of 1,7-
naphthyridines has been presented, all of which had excellent
pharmacological and physicochemical profiles. Focusing on
compounds with the best pharmacokinetics and in vivo activity,
we progressed two compounds to the ferret model of emesis,
benchmarked against known PDE4 inhibitor roflumilast. In this
model, a significant disparity was seen between the emetic
potential of two very closely related compounds (2d and 2e),
although both could be reasoned to have a better therapeutic
window than roflumilast (lower emesis seen at pharmacolog-
Effect of PDE4 Inhibitors in a Ferret Model of Emesis.
One of the observed side effects of PDE4 inhibitors in the clinic
is their potential to cause nausea and emesis at therapeutic
doses. The aim of this study was to assess the emetic potential
of our PDE4 inhibitors in a ferret model, the details of which
have been described.14 Roflumilast, currently the most
advanced PDE4 inhibitor, was used as the internal standard
and compared against compounds 2d and 2e (Figure 5).
Among the compounds used, 2e showed the most promising
profile because no emesis was observed at all the doses tested,
while roflumilast was shown to be the most emetic agent.
Compound 2e has a similar level of efficacy in a rodent model
of inflammation as 2d. An impression of the relative therapeutic
index of the compounds can be gained by examining the extent
of emesis at doses concomitant with ED50 values in our rat
models (albeit there is a species difference). In the rat OA
model, the respective ED50s are: 2d = 1.2 mg/kg, 2e = 2 mg/kg,
roflumilast = 0.74 mg/kg; it may be seen from the graph in
Figure 5 that the degree of emesis observed at these doses is
less for 2d compared to roflumilast. In addition, the
pharmacokinetic properties of the compounds were examined
after oral dosing to investigate whether the degree of emesis
was related to exposure (Table 3).
7476
dx.doi.org/10.1021/jm300459a | J. Med. Chem. 2012, 55, 7472−7479

Journal of Medicinal Chemistry
Article
a
Table 3. Pharmacokinetic Parameters of PDE4 Inhibitors in the Ferret
compd
Cmax (μM)
Tmax (min)
AUCtot (μM·min)
T1/2 (min)
b
2d
12.3 6.0
90.0 34.6
105.0 30.0
not detected
3560 1690
2980 3460
not detected
192 37
112 83
c
2e
roflumilast
10.5 12.8
d
e
not detected
not detected
a
b
Error limits are shown as
SEM. Formulated as solution in PEG200/saline (50/50, v/v) for intragastric administration at 3.42 μmol/kg.
c
d
Formulated as solution in PEG200/saline (50/50, v/v) for intragastric administration at 3.42 μmol/kg. Formulated as solution in PEG200/saline
e
(50/50, v/v) for intragastric administration at 2.98 μmol/kg. No plasma levels were detected for roflumilast or for its possible N-oxide metabolite
(limits of detection were 0.5 and <0.02 μM, respectively).
s, pulse width of 33°, and data size of 64K points. Again, chemical
shifts are reported in ppm relative to the residual solvent resonances.
All final compounds were purified to >95% purity as determined by
high performance liquid chromatography (HPLC) with UV detection
at 254 and 220 nm and ELSD, using Acquity/Waters SQD LCMS
instruments. Low resolution mass spectrometry (MS) data were
obtained using a Waters SQD single quadrupole mass spectrometer
operating in electrospray ionization (ESI) mode (positive or negative).
The following methods were used:
ically equivalent doses and much higher levels of systemic
exposure).
Compounds 2d and 2e were selected as potential clinical
candidates and were profiled through further preclinical assays.
Although compound 2e appeared by far the better candidate
based on the ferret model data, it was felt that emesis as a side
effect was less important than other toxicological readouts
associated with PDE4 inhibitors such as vasculitis.2 During a
two-week toxicology study in the rat, again using roflumilast as
a benchmark, it was found that based on the histopathological
assessment of selected organs at the dose group of 2 mg/kg,
compound 2d showed a better toxicological profile compared
to 2e and roflumilast, despite similar doses of the compounds
resulting in much higher exposure to 2d as compared to
roflumilast (or its N-oxide metabolite). Thus it was interesting
to note that there appeared to be no correlation between
emetic potential of the compounds in the ferret and
toxicological profile in the rat.
(1) Column: Acquity CSH C18, 100 mm × 2.1 mm, 1.7 μ. A =
water + 0.1% formic acid A, B = MeCN + 0.1% TFA. Gradient,
0−8 min 2−98%B, 8−9 min 98%B, 9−9.1 min 98−2%B. Flow
rate, 0.7 mL/min
(2) Column: Acquity CSH C18, 100 mm × 2.1 mm, 1.7 μ. A =
water + 0.1% ammonia, B = MeCN + 0.1% ammonia. Gradient,
0−8 min 2−98%B, 8−9 min 98%B, 9−9.1 min 98−2%B. Flow
rate, 0.7 mL/min
Chemistry. The general procedure for the preparation of 8-aryl-
1,7-naphthyridin-6-yltrifluoromethane sulfonate derivatives 1, includ-
ing 1b and 1c, has been previously described16 and is followed for the
variants in this disclosure. The characterization of representative new
compounds is given below and in the Supporting Information:
8-(3-Fluorophenyl)-1,7-naphthyridin-6-yl trifluoromethanesulfo-
nate (1d). 1H NMR (500 MHz, DMSO-d6) δ 9.22 (dd, J = 4.1, 1.8 Hz,
1H), 8.65 (dd, J = 8.5, 1.8 Hz, 1H), 8.27 (s, 1H), 8.03 (ddd, J = 8.4,
1.7, 1.0 Hz, 1H), 8.01 (ddd, J = 10.6, 2.7, 1.7 Hz, 1H), 7.95 (dd, J =
8.5, 4.1 Hz, 1H), 7.64 (ddd, J = 8.4, 8.4, 6.2 Hz, 1H), 7.43 (dddd, J =
8.4, 8.4, 2.7, 1.0 Hz, 1H). 13C NMR (126 MHz, DMSO-d6) δ 161.5 (d,
J = 242.5 Hz), 156.8 (d, J = 2.6 Hz), 153.5, 149.9, 140.5, 138.2 (d, J =
8.3 Hz), 136.3, 135.5, 130.0 (d, J = 8.2 Hz), 127.2 (d, J = 2.7 Hz),
126.5, 118.3 (q, J = 321.1 Hz), 118.0 (d, J = 23.7 Hz), 116.8 (d, J =
20.8 Hz), 111.6.
CONCLUSION
■
We have demonstrated that it is possible to take a lead
compound with poor solubility and pharmacokinetic profile
and to reintroduce solubility based on rational design, in turn
achieving an improved pharmacokinetic response. Reproducible
and predictable ADME properties are likely to be important in
the case of a trial drug acting on a target known to give a low
therapeutic window such as PDE4.
In this program, we developed a number of PDE4 inhibitors
with in vivo activity, however, the side effect profile proved to
be difficult to predict or explain based on any measured
physicochemical or biological parameter. Thus, a pragmatic
decision was taken to progress the overall most promising
compound, particularly looking at the in vivo toxicology in
selected organs rather than just at acute emesis. On the basis of
these experiments, compound 2d was chosen for further
advancement, and a large scale synthesis developed.15 The
compound was selected for profiling in Ph1 clinical trials in
humans, the results of which will be reported in future
publications.
4-Iodocyclohexanecarboxylic Acid Ethyl Ester. To a cold (0 °C)
stirred solution of 4-hydroxycyclohexanecarboxylic acid ethyl ester (1.0
g, 5.80 mmol) in 1:2 dichloromethane:carbontetrachloride (52 mL)
was added triphenylphosphine (1.82 g, 6.96 mmol), imidazole (473
mg, 6.96 mmol), and iodine (1.79 g, 7.08 mmol). The mixture was
allowed to warm to room temperature and stirred overnight. The
reaction was quenched by the addition of saturated sodium thiosulfate
(ca. 50 mL) and stirred until the solution became clear. The layers
were separated, and the aqueous layer was extracted with dichloro-
methane (3 × 30 mL). The combined organic phases were washed
with sodium thiosulfate (30 mL) and brine (30 mL), dried with
anhydrous MgSO4, filtered, and evaporated at reduced pressure to an
oily solid. Purification was by dry flash chromatography, using
Keiselgel 15−40 grade silica, eluting with 3% ethyl acetate/isohexane
to yield 4-iodo-cyclohexanecarboxylic acid ethyl ester as a clear
colorless oil (801 mg, 49%).
EXPERIMENTAL SECTION
All H NMR spectra were obtained on either a Bruker AV400 or a
■
1
Bruker DRX500 spectrometer operating at 400 and 500 MHz for
proton, respectively, both at 298K. The AV400 is equipped with a 5
mm inverse 1H/13C dual probe, the DRX500 with a 5 mm BBO
probe. Both instruments are operating under Bruker Topspin software.
Typical acquisition parameters: spectral width 16 ppm, pulse width of
33°, and data size 64k points. Chemical shifts are reported in values
relative to the residual solvent signal (dimethylsulfoxide), and coupling
constants are reported in hertz. 13C NMR experiments were carried
out on the Bruker AV400 and DRX500 instruments operating at 100
and 125 MHz, respectively, also at 298K. Typical acquisition
parameters were a sweep width of 30 kHz, acquisition time of 1.08
Isomer 1: 1H NMR (400 MHz, DMSO-d6) δ 4.81 (m, 1H), 4.07 (q,
J = 7.1 Hz, 2H), 2.47 (m, 1H), 1.96 (m, 1H), 1.76 (m, 1H), 1.75 (m,
1H), 1.73 (m, 1H), 1.18 (q, J = 7.1 Hz, 3H). 13C NMR (100 MHz,
DMSO-d6) δ 174.2, 59.8, 40.4, 35.8, 35.2, 25.8, 14.1.
1
Isomer 2: H NMR (400 MHz, DMSO-d6) δ 4.30 (dddd, J = 11.5,
11.5, 3.9, 3.9 Hz, 1H), 4.03 (q, J = 7.1 Hz, 2H), 2.41 (dddd, J = 11.5,
11.5, 3.8, 3.8 Hz, 1H), 2.29 (m, 1H), 1.94 (m, 1H), 1.74 (m, 1H), 1.47
(m, 1H), 1.15 (q, J = 7.1 Hz, 3H). 13C NMR (100 MHz, DMSO-d6) δ
174.5, 59.8, 40.4, 38.3, 30.3, 30.2, 14.0.
7477
dx.doi.org/10.1021/jm300459a | J. Med. Chem. 2012, 55, 7472−7479

Journal of Medicinal Chemistry
Article
General Procedure for the Preparation of 4-((8-Aryl)-1,7-
naphthyridin-6-yl)cyclohexanecarboxylic Acid Derivatives 2a−2f.
A flask was charged with activated zinc dust (742 mg, 11.13 mmol),
THF (1.80 mL), and 1,2-dibromoethane (25 μL, 0.284 mmol). The
suspension was heated to reflux for 3 min and then allowed to cool
before trimethylsilyl chloride (29 μL, 0.227 mmol) was added. The
mixture was stirred for 15 min, then 4-iodocyclohexanecarboxylic acid
ethyl ester (1.60 g, 5.67 mmol) was added and the mixture stirred at
35 °C for 1.5 h. A second flask was charged with Pd-
(dibenzylideneacetone)2 (101 mg, 0.176 mmol), 1,1′-bis-
(diphenylphosphino)ferrocene (98 mg, 0.176 mmol), N-methylpyrro-
lidinone (NMP) (3 mL):THF (1 mL), tetrabutylammonium iodide
(2.79 g, 7.56 mmol), and the appropriate 8-(3-‘R’-phenyl)-1,7-
naphthyridin-6-yl trifluoromethanesulfonate (956 mg, 2.52 mmol),
and the contents were added to the first flask at 35 °C. The reaction
mixture was stirred for 2 h, then quenched by the addition of water
(15 mL) and stirred for 10 min. Ethyl acetate (40 mL) was then added
and stirred for 5 min. The layers were separated, and the organic layer
was washed with 5% citric acid (25 mL), water (2 × 25 mL), and brine
(40 mL), dried with anhydrous MgSO4, filtered, and evaporated to a
brown viscous oil. Purification was by dry flash chromatography using
15−40 grade Keiselgel silica, eluting with 30% ethyl acetate/iso-hexane
to yield the [1,7]-naphthyridin-6-yl-cyclohexanecarboxylic acid ethyl
ester as an orange gum.
column chromatography, typically eluting with 25% ethyl acetate in
hexane, yielded the desired product as a powder.
1-(8-(3-Fluorophenyl)-1,7-naphthyridin-6-yl)piperidine-4-carbox-
ylic Acid (3d). 1H NMR (500 MHz, DMSO-d6) δ 8.63 (dd, J = 4.0, 1.8
Hz, 1H), 8.11 (dd, J = 8.5, 1.8 Hz, 1H), 7.97 (ddd, J = 8.0, 1.3, 1.3 Hz,
1H), 7.94 (ddd, J = 10.8, 2.6, 1.3 Hz, 1H), 7.53 (ddd, J = 8.0, 8.0. 6.1
Hz, 1H), 7.49 (dd, J = 8.5, 4.0 Hz, 1H), 7.29 (dddd, J = 8.9, 8.0, 2.6,
1.3 Hz, 1H), 7.01 (s, 1H), 4.25 (m, 2H), 2.99 (m, 2H), 2.07 (dddd, J =
10.5, 10.5, 3.3, 3.3 Hz, 1H), 1.84 (m, 2H), 1.57 (m, 2H). 13C NMR
(126 MHz, DMSO-d6) δ 176.8, 161.5 (d, J = 241.8 Hz), 154.9, 154.6
(d, J = 2.6 Hz), 147.2, 140.9 (d, J = 8.2 Hz), 135.1, 135.1, 133.9, 129.4
(d, J = 8.2 Hz), 126.9 (d, J = 2.4 Hz), 124.7, 117.5 (d, J = 22.8 Hz),
115.2 (d, J = 21.1 Hz), 98.0, 45.5, 43.9, 28.9.
PDE4 in Vitro Assays. The PDE4 isoform assays were based on
Amersham Pharmacia Biotech Scintillation Proximity Assay (SPA)
technology. The enzyme was diluted with enzyme dilution buffer (10
mM Tris-HCI, pH7.5 containing 1 mM EDTA) in order to obtain
between 10 and 30% total substrate hydrolysis during the assay. The
enzymatic reaction was started by adding 10 μL of diluted enzyme to
80 μL of substrate (0.1 μCi [3H]-cAMP, 1 μM cAMP) and 10 μL of
inhibitor solution in a 96-well microtiter plate. After 30−60 min
incubation at room temperature, the reaction was stopped by adding
50 μL of PDE SPA beads (20 mg/mL). After 30 min, the plate was
centrifuged (3000g, 10 min) and counted (Packard TopCount).
Inhibitor stock solutions were prepared in 100% dimethylsulfoxide
(DMSO) and diluted with DMSO/water to achieve 10 concentrations
to cover the range of 0−100% inhibition. The concentration of DMSO
is kept constant at 1% (v/v) throughout the assay. The concentration
at which 50% inhibition occurs (IC50) was determined from inhibition
concentration curves.
hPBMC Activation Assay. Peripheral blood mononuclear cells
(PBMC) were isolated from the blood of normal individuals by
Ficoll−Hypaque gradient centrifugation (20 min at 800g). The
interphase was collected, washed twice in PBS, and resuspended in
RPMI1640 supplemented with 10% FCS. Cell density was then
adjusted to 1 × 106 cells/mL. The PBMC suspension (100 μL) was
placed in 96-well culture plates, and 50 μL of either medium or
compound were added (compound solutions were prepared as above).
After a 10 min preincubation, cells were stimulated with LPS (50 μL,
10 ng/mL) for the production of TNF-α. Supernatants were harvested
after incubating the plates for 20 h at 37 °C in a humidified incubator
with 5% CO2.
TNF-α was measured by sandwich ELISA using a primary anti-
TNF-α monoclonal antibody purchased from Pharmingen (UK).
Binding of the secondary antibody was analyzed by stepwise
incubation with streptavidin−alkaline phosphatase conjugate (Sigma,
UK) and 4-nitrophenylphosphate disodium salts (Sigma, UK).
Optical density was measured at 405 nm, and cytokine
concentration was calculated based on the results from serial dilutions
of standard recombinant TNF-α. The sensitivity of the cytokine ELISA
was approximately 10 pg/mL. The concentration at which 50%
inhibition occurs (IC50) was determined from inhibition concentration
curves.
The esters were hydrolyzed to give the final compounds by the
following general procedure:
Aqueous lithium hydroxide (1M, 24.3 mL, 24.3 mmol) was added
to a solution of the appropriate [1,7]-naphthyridin-6-yl-cyclo-
hexanecarboxylic acid ethyl ester (12.14 mmol) in THF/methanol
(40 mL:20 mL) and stirred at room temperature overnight. The
organic solvents were removed by evaporation, then the aqueous
residue diluted with water and basified to pH 9 with 1 M KOH. The
aqueous layer was then washed with ethyl acetate (3×). The aqueous
layer was acidified to pH 4 with 1 M HCI to give a white precipitate,
which was extracted into ethyl acetate. The ethyl acetate layer was then
dried over sodium sulfate, filtered, and concentrated to yield the
desired product as a yellow foam. If necessary, further trituration with
1 M HCl yields the product as a pale-yellow powder.
4 - ( 8 - ( 3 - F l u o r o p h e n y l ) - 1 , 7 - n a p h t h y r i d i n - 6 - y l ) -
cyclohexanecarboxylic Acid (2d). 1H NMR (500 MHz, DMSO-d6) δ
12.08 (s, 1H), 9.01 (dd, J = 4.1, 1.9 Hz, 1H), 8.43 (dd, J = 8.3, 1.9 Hz,
1H), 7.99 (ddd, J = 8.0, 1.3, 1.3 Hz, 1H), 7.95 (ddd, J = 10.8, 2.6, 1.3
Hz, 1H), 7.76 (dd, J = 8.3, 4.1 Hz, 1H), 7.76 (s, 1H), 7.56 (ddd, J =
8.0, 8.0, 6.1 Hz, 1H), 7.33 (dddd, J = 8.9, 8.0, 2.6, 1.3 Hz, 1H), 2.88
(dddd, J = 11.9, 11.9, 3.3, 3.3 Hz, 1H), 2.31 (dddd, J = 12.1, 12.1, 3.3,
3.3 Hz, 1H), 2.10 (m, 2H), 2.07 (m, 2H) 1.69 (m, 2H), 1.54 (m, 2H).
13C NMR (126 MHz, DMSO-d6) δ 176.7, 161.5 (d, J = 241.8 Hz),
158.6, 156.0 (d, J = 2.4 Hz), 151.2, 140.7 (d, J = 8.2 Hz), 139.4, 135.5,
132.5, 129.5 (d, J = 8.3 Hz), 127.1 (d, J = 2.4 Hz), 125.2, 117.7 (d, J =
23.1 Hz), 116.4, 115.4 (d, J = 21.1 Hz), 44.3, 42.1, 31.4, 28.7.
4-(8-(3-(Trifluoromethoxy)phenyl)-1,7-naphthyridin-6-yl)-
cyclohexanecarboxylic Acid (2e). 1H NMR (500 MHz, DMSO-d6) δ
12.09 (s, 1H), 9.00 (dd, J = 4.1, 1.8 Hz, 1H), 8.43 (m, 1H), 8.18 (ddd,
J = 8.0, 1.4, 1.4 Hz, 1H), 8.11 (m, 1H), 7.77 (s, 1H), 7.76 (dd, J = 8.3,
4.1 Hz, 1H), 7.66 (dd, J = 8.0, 8.0 Hz, 1H), 7.49 (m, 1H), 2.88 (dddd,
J = 11.9, 11.9, 3.3, 3.3 Hz, 1H), 2.30 (dddd, J = 12.1, 12.1, 3.3, 3.3 Hz,
1H), 2.10 (m, 2H), 2.07 (m, 2H), 1.68 (m, 2H), 1.54 (m, 2H). 13C
NMR (126 MHz, DMSO-d6) δ 176.7, 158.6, 155.6, 151.3, 147.8 (q, J
= 1.8 Hz), 140.6, 139.4, 135.5, 132.5, 130.1, 129.5, 125.3, 123.3, 121.0,
120.2 (q, J = 255.9 Hz), 116.6, 44.3, 42.1, 31.4, 28.7.
ASSOCIATED CONTENT
■
S
* Supporting Information
Experimental results for examples 1a, 1e, 1f, 2a−b, 2f, 3a−c,
and 3e−f. This material is available free of charge via the
Internet at http://pubs.acs.org.
General Procedure for the Preparation of 1-((8-Aryl)-1,7-
naphthyridin-6-yl)piperidine-4-carboxylic Acid Derivatives 3a−3f.
The appropriate 8-(3-‘R’-phenyl)-1,7-naphthyridin-6-yltrifluorometha-
nesulfonate (0.53 mmol) was dissolved in DMSO (1.2 mL) with the
appropriate amine (l.24 mmol) and the mixture heated at 60 °C for 2
days or until complete by high performance liquid chromatography
(HPLC) analysis. After cooling, ethyl acetate was added to the
mixture, which was washed four times with brine. The organic layer
was dried over sodium sulfate and concentrated. Purification by
AUTHOR INFORMATION
■
Corresponding Author
*Phone: +44 1403 272827. Fax: +44 1403 323 837. E-mail:
neil.press@novartis.com.
Notes
The authors declare no competing financial interest.
7478
dx.doi.org/10.1021/jm300459a | J. Med. Chem. 2012, 55, 7472−7479

Journal of Medicinal Chemistry
Article
6,8-Disubstituted 1,7-Naphthyridines: A Novel Class of Potent and
Selective Phosphodiesterase Type 4D Inhibitors. J. Med. Chem. 2000,
43 (4), 675−682.
(17) Wohnsland, F.; Faller, B. High-Throughput Permeability pH
Profile and High-Throughput Alkane/Water log P with Artificial
Membranes. J. Med. Chem. 2001, 44 (6), 923−930.
ACKNOWLEDGMENTS
■
We thank Professor R. J. Naylor, University of Bradford, for
performing the ferret emesis experiments.
ABBREVIATIONS USED
■
BAL, bronchoalveolar lavage; BN, brown Norway; PDE4,
phosphodiesterase-4; HARBS, high affinity rolipram binding
site; OA, ovalbumin; PBMC, peripheral blood mononuclear cell
REFERENCES
■
(1) Dastidar, S. G.; Rajagopal, D.; Ray, A. Therapeutic benefit of
PDE4 inhibitors in inflammatory diseases. Curr. Opin. Invest. Drugs
(Thomson Sci.) 2007, 8, 364−372.
(2) Press, N. J.; Banner, K. H. PDE4 inhibitorsa review of the
current field. Prog. Med. Chem. 2009, 47, 37−74.
(3) Boswell-Smith, V.; Spina, D. PDE4 inhibitors as potential
therapeutic agents in the treatment of COPD-focus on roflumilast. Int.
J. Chronic Obstruct. Pulm. Dis. 2007, 2, 121−129.
(4) Martin, N.; Reid, P. T. The potential role of phosphodiesterase
inhibitors in the management of asthma. Treat. Respir. Med. 2006, 5,
207−217.
(5) Spina, D. The potential of PDE4 inhibitors in respiratory disease.
Curr. Drug Targets: Inflammation Allergy 2004, 3, 231−236.
(6) Cazzola, M.; Picciolo, S.; Matera, M. G. Roflumilast in chronic
obstructive pulmonary disease: evidence from large trials. Expert Opin.
Pharmacother. 2010, 11, 441−449.
(7) (a) Giembycz, M. A. Can the anti-inflammatory potential of
PDE4 inhibitors be realized: guarded optimism or wishful thinking? Br.
J. Pharmacol. 2008, 155, 288−290. (b) Chapman, R. W.; House, A;
Richard, J; Prelusky, D; Lamca, J; Wang, P; Lundell, D; Wu, P; Ting,
P. C.; Lee, J. F.; Aslanian, R; Phillips, J. E. Pharmacology of a potent
and selective inhibitor of PDE4 for inhaled administration. Eur. J.
Pharmacol. 2010, 643, 274−281.
(8) Hersperger, R.; Dawson, J.; Mueller, T. Synthesis of 4-(8-
benzo[1,2,5]oxadiazol-5-yl-[1,7]naphthyridine-6-yl)-benzoic acid: a
potent and selective phosphodiesterase type 4D inhibitor. Bioorg.
Med. Chem. Lett. 2002, 12, 233−235.
(9) Trifilieff, A.; Wyss, D.; Walker, C.; Mazzoni, L.; Hersperger, R.
Pharmacological profile of a novel phosphodiesterase 4 inhibitor, 4-(8-
benzo[1,2,5]oxadiazol-5-yl-[1,7]naphthyridin-6-yl)benzoic acid (4), a
1,7-naphthyridine derivative, with anti-inflammatory activities. J.
Pharmacol. Exp. Ther. 2002, 301, 241−248.
(10) Tigani, B.; Cannet, C.; Zurbruegg, S.; Schaeublin, E.; Mazzoni,
L.; Fozard, J. R.; Beckmann, N. Resolution of the oedema associated
with allergic pulmonary inflammation in rats assessed noninvasively by
magnetic resonance imaging. Br. J. Pharmacol. 2003, 140, 239−246.
(11) Crystallographic data (excluding structure factors) for the
structure in this paper have been deposited with the Cambridge
Crystallographic Data Centre as supplementary publication numbers
CCDC 873054. Copies of the data can be obtained, free of charge, on
application to CCDC, 12 Union Road, Cambridge CB2 1EZ, UK (fax:
+44 (0)1223−336033 or e-mail: deposit@ccdc.cam.ac.uk).
(12) Renzi, P. M.; Olivenstein, R.; Martin, J. G. Inflammatory cell
populations in the airways and parenchyma after antigen challenge in
the rat. Am. Rev. Respir. Dis. 1993, 147, 967−974.
(13) Goncalves, d. M.; Singer, M.; Vargaftig, B. B.; Chignard, M.
Effects of rolipram on cyclic AMP levels in alveolar macrophages and
lipopolysaccharide-induced inflammation in mouse lung. Br. J.
Pharmacol. 1998, 123, 631−636.
(14) Costall, B.; Domeney, A. M.; Naylor, R. J.; Tattersall, F. D.
Emesis induced by cisplatin in the ferret as a model for the detection
of antiemetic drugs. Neuropharmacology 1987, 26, 1321−1326.
(15) Jiang, X.; Lee, G. T.; Villhauer, E. B.; Prasad, K.; Prashad, M. A
Scalable Synthesis of a 1,7-Naphthyridine Derivative, a PDE-4
Inhibitor. Org. Process Res. Dev. 2010, 14, 883−889.
(16) Hersperger, R; Bray-French, K; Mazzoni, L; Muller, T.
̈
Palladium-Catalyzed Cross-Coupling Reactions for the Synthesis of
7479
dx.doi.org/10.1021/jm300459a | J. Med. Chem. 2012, 55, 7472−7479