Synthesis and structure-activity relationship studies of dihydronaphthyridinediones as a novel structural class of potent and selective PDE7 inhibitors

Article abstract of DOI:10.1016/j.bmcl.2011.09.065

The synthesis and SAR studies of a series of structurally novel inhibitors of PDE7 are discussed. The best compounds from the series display low nanomolar inhibitory activity and are selective versus other PDE isoenzymes.

Full text of DOI:10.1016/j.bmcl.2011.09.065

Bioorganic & Medicinal Chemistry Letters 21 (2011) 6652–6656
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Synthesis and structure–activity relationship studies
of dihydronaphthyridinediones as a novel structural class of potent
and selective PDE7 inhibitors
⇑
Rainer Gewald , Carla Rueger, Christian Grunwald, Ute Egerland, Norbert Hoefgen
biocrea GmbH, Meissner Strasse 191, 01445 Radebeul, Germany
a r t i c l e i n f o
a b s t r a c t
Article history:
The synthesis and SAR studies of a series of structurally novel inhibitors of PDE7 are discussed. The best
compounds from the series display low nanomolar inhibitory activity and are selective versus other PDE
isoenzymes.
Received 19 August 2011
Revised 14 September 2011
Accepted 16 September 2011
Available online 21 September 2011
Ó 2011 Elsevier Ltd. All rights reserved.
Keywords:
PDE7 inhibitor
Phosphodiesterase
Dihydronaphthyridinedione
Structure–activity relationship
Molecular docking
Phosphodiesterase enzymes (PDEs) play an important role in
various biological processes by hydrolyzing the key secondary mes-
collection resulted in the identification of 1 as a weak dual PDE4/
PDE7 inhibitor with IC50 values of 0.38 and 0.82
(Fig. 1).
lM, respectively
´ ´
sengers adenosine and guanosine 3,5-cyclic monophosphates
´
(cAMP and cGMP) into their corresponding 5-monophosphate
This communication describes our preliminary efforts towards
optimizing the enzymatic inhibitory activity and the selectivity
nucleotides, thereby decreasing concentrations of cAMP and cGMP,
respectively.1 At least eleven isoenzymes of mammalian cyclic
nucleotide phosphodiesterase have been identified on the basis of
primary structure, substrate specificity, or sensitivity to cofactors
profile of these compounds versus PDE4, which is also
cAMP-specific PDE.
a
During the initial phase of our work to evaluate the structure–
activity relationship (SAR) of this new type of PDE7 inhibitors we
investigated the role of substituents on the pyrido ring of the core
structure. Surprisingly, we detected that the incorporation of a
hydrogen bond acceptor function yielded very different effects on
the target affinity of these derivatives. While the introduction of
or inhibitory drugs.2 Among them, PDE7 is
a
high affinity
cAMP-specific PDE (Km = 0.2
lM) with kinetic properties different
from those of PDE4. The Km of PDE7 is about 10-fold lower com-
pared with PDE4. Interestingly, the inhibition of the activity or
the expression of PDE7A was found to block the activation of T-cells
as well as the proliferation and function of preactivated T-cells and
cytotoxic T-lymphocytes.3 Recently, topical application of a PDE7A
inhibitor in a mouse chronic inflammation model decreased the
number of Ki67-positive keratinocytes, suggesting PDE7A to be a
valuable therapeutic target of skin diseases, for example, psoriasis.4
As PDE7 inhibitors might also be useful in the treatment of CNS
disorders,5 current medicinal chemistry effort are directed towards
the identification of compounds active at very low concentrations.6
In order to avoid potential secondary effects such as emesis or
cardiotoxicity, PDE isoenzyme selectivity is of great importance.
In this context, high throughput screening of the compound
O
N
H
O
N
F
N
F
N
O
1
⇑
Corresponding author. Tel.: +49 351 4043 1204.
E-mail address: rainer.gewald@biocrea.com (R. Gewald).
Figure 1. Structure of hit 1.
0960-894X/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2011.09.065

R. Gewald et al. / Bioorg. Med. Chem. Lett. 21 (2011) 6652–6656
6653
O
N
H
O
N
F
N
O
O
N
F
2
IC50 > 5 µM
1
O
IC50 = 0.82 µM
N
H
H
N
O
N
F
N
O
O
F
3
IC50 = 0.022 µM
Figure 2. Structures of modified hit compounds 2 and 3.
Figure 3. Schematic representation of GOLD docking result generated by LigPlot+.13
Water mediated hydrogen bonds are indicated by green dotted lines. Residues
making nonbonded contacts with compound 3 are shown as spoked arcs.
a methoxy group in 6-position created an inactive compound 2, a
carbonyl function at the same position as in compound 3 shifted
the IC50 value to about 40-fold higher affinity (Fig. 2).
Since compounds 1, 2, and 3 display a wide range of potency but
differ only regarding the substitution in the 6-position of the naph-
thyridine ring it is concluded that interaction of this part with the
PDE7 catalytic domain is crucial for the target affinity. Molecular
docking into the published co-crystal structure7 of PDE7 and 3-iso-
butyl-1-methylxanthine (IBMX) (PDB ID 1ZKL rev. 3.101) was used
to derive a reasonable hypothesis on the binding mode of these
compounds. Prior to docking, conformation and tautomeric forms
of compounds under study were analyzed applying semiempirical
AM1 calculations.8 Results indicate (i) anti conformation of the two
carboxy oxygens of our compounds and (ii) keto tautomerization
O
N
O
O
N
O
O
N
O
N
O
N
O
N
O
O
O
O
a
b
c
NH2
N
O
O
R1
A
B
C
D
R2
H
O
N
N
O
N
route A
O
R2
f
O
N
OH
H
d
O
N
N
H
N
O
N
e
O
O
R1
O
F
e
f
R1
H
N
O
N
OH
R1
O
E
O
route B
H
R1
G
Scheme 1. Reagents and conditions: (a) EtOCHC(COOEt)2, DMF, 120 °C, 92% yield; (b) (C6H5)2O, reflux, 56% yield; (c) R1-substituted benzyl bromide, K2CO3, DMF, 120 °C, 52–
75% yield; (d) KOH, H2O, EtOH, rt, 84–96% yield; (e) R2-substituted aniline, ClCOOCH2CH(CH3)2, Et3N, CH2Cl2, 0 °C to rt, 56–80% yield; (f) 48% aqueous HBr, AcOH, reflux, 72–
95% yield.

6654
R. Gewald et al. / Bioorg. Med. Chem. Lett. 21 (2011) 6652–6656
Table 1
of naphthyridine-diones. Both results are supported by NMR data
which also indicate an internal hydrogen bond between the amide
nitrogen and neighboring naphthyridine carboxy oxygen. These
findings were used to constrain docking calculations employing
the program GOLD.9
In vitro data for compounds 1–32
R2
H
H
O
N
F
N
F
Figure 3 schematically depicts the interaction of compound 3
and the PDE 7 binding site derived from GOLD docking. A direct
hydrogen bond to conserved Gln413 known from natural PDE
substrates10,11 and most PDE inhibitors does not exist. Such inter-
action is prevented sterically and would be found only if the
morpholine-phenyl-amide group was freely rotatable. Instead,
docking calculations indicate two water-mediated hydrogen bond
interactions of carboxy oxygen at 6-position and Gln413 and
Asn365. The NH at the 5-position is not involved in such interac-
tions according to the docking outcome. In addition to hydrogen
bonds, the naphthyridine ring is held between Phe416, Val 380,
and Phe384 known as the ‘hydrophobic clamp’.12 Compared with
these results, no direct or mediated hydrogen bonding interac-
tions of compounds 1 and 2, respectively, with Gln413 were
found, leading to a drop of affinity. Whereas 1 still fits into the
hydrophobic clamp, steric interactions of the MeO group of 2 lead
to a slight shift out of this region causing loss of affinity. Other
interactions should not discriminate between compounds 1, 2,
and 3: There are a number of hydrophobic contacts with the mor-
pholino-phenyl side chain. The morpholine ring sticks out of the
binding site and thus will interact with surrounding polar solvent
in a favorable way. Following our hypothesis that the carbonyl
group in compound 3 is crucial for high target affinity and thus
we focused our optimization strategy on the SAR of this type of
inhibitor.
O
N
O
Compda R2
PDE7A1b (IC50
,
l
M) PDE4A4b (IC50
, lM)
117
0.823
>5.00
0.022
0.124
0.046
0.066
0.082
0.104
0.024
0.094
0.096
1.07
0.179
>1.00
0.300
0.200
0.176
0.029
0.049
0.029
0.020
0.379
>5.00
0.506
>1.00
0.442
0.471
0.994
0.477
0.362
1.33
>1.00
>1.00
>1.00
0.219
>1.00
1.52
218
3
4
5
4-N-morpholinyl
4-N-piperidinyl
4-NEt2
6
4-NMe2
7
8
4-NHMe
4-NH2
9
4-NHCOMe
H
10
11
12
13
14
15
16
1719
18
19
20
21
4-CN
3-CN
4-F
2,6-Di-F
4-CF3
4-COOH
4-CH2COOH
4-COMe
4-CONH2
4-CONHMe
4-CONHCH2-4-
Pyridyl
>1.00
>1.00
0.201
0.560
0.430
The 1,5-dihydro-[1,5]naphthyridine-2,8-dione derivatives were
prepared as described in Scheme 1.
The synthesis of intermediates E followed established proce-
dures for [1,5]naphthyridine ring synthesis.14 5-Amino-2-methoxy-
pyridine A was condensed with diethyl ethoxymethylenemalonate
to afford the enamine B. Thermal cyclization of this enamine in
refluxing diphenyl ether provided 1,4-dihydro[1,5]naphthyridine-
3-carboxylic acid ethyl ester C, which was alkylated in the presence
of potassium carbonate with substituted benzyl bromides to yield
intermediates D. The corresponding carboxylic acids E were ob-
tained by alkaline saponification. Depending on the stability of the
functional groups R2 either the amide formation was carried out at
first followed by cleavage of the methoxy group (route A) or vice ver-
sa (route B). The coupling of the carboxylic acids E and G with substi-
tuted anilines was mediated by isobutyl chloroformate according to
the mixed anhydride method of peptide synthesis.15 Formation of
the second cyclic carbonyl function in intermediates G and target
compounds H was brought about by using standard conditions for
the cleavage of arylalkyl ethers with hydrogen bromide in acetic
acid.16
The compounds described in this paper were assessed against
both PDE7A and PDE7B subtypes. As no selectivity was observed,
only the PDE7A inhibitory activities will be presented herein.
As a first approach, we decided to keep the substitution pattern
at the benzyl group unchanged and focus on the amide part of the
molecule. Using a parallel synthesis approach, a large number of
different R2 substituents were introduced at the phenyl ring and
evaluated. Table 1 lists the structures and inhibitory activity of se-
lected analogs from this series.
22
23
24
25
26
2720
28
29
30
31
32
4-CO-N-morpholinyl
4-SO2CH3
4-SO2NH2
4-SO2NH-thiazol-2-yl
4-SMe
4-OH
4-OMe
4-OCHF2
4-OCF3
3,5-di-OMe
3,4,5-Tri-OMe
0.188
0.027
0.034
0.025
0.086
0.105
0.151
0.071
0.090
0.212
0.018
1.06
>1.00
0.458
0.481
>1.00
>1.00
>1.00
0.568
>1.00
>1.00
0.516
a
Compounds 1–26 prepared by route A; compounds 28–32 prepared by route B.
b
Measured against the human full length enzymes produced in baculovirus
infected sf21 cells.21
While small electron withdrawing functions such as cyano or
fluorine (11, 13 and 15) showed barely any influence on inhibitory
activity, changing their position on the phenyl ring had a signifi-
cant impact. When the point of attachment was moved from para
to meta, an 11-fold loss in potency was observed (12). The fluorine
disubstitution pattern in 14 even shifted the selectivity balance to-
wards PDE4.
Further targeting the 4-position, the introduction of a carboxy
group, connected either directly or by a methylene bridge, led to
less active compounds (16 and 17). These results suggest an influ-
ence of pKa as other carbonyl containing neutral groups (18–21)
with the exception of tertiary amide 22 caused a flat SAR compared
to 3. However, the good selectivity exhibited by 18 turned into
dual inhibition with the primary amide 19. The sulfonyl derivatives
23–25 were found to maintain the level of activity in the same
range as their carbonyl analogs.
Replacement of the morpholine structure at the 4-position by a
number of other amino groups (4–9) was tolerated, but with an
apparent loss of selectivity in line with the decreasing size of the
substituent. . As inhibitory activity was found to be only slightly
lower for nonsubstituted 10, it became clear, that the modification
of physicochemical properties by introducing a variety of other
functional groups should be feasible.
None of the other mono and disubstituted neutral derivatives
(26–31) provided any enhancement compared to reference com-
pound 3. Interestingly, the addition of a third methoxy group
(32) reversed this trend.

R. Gewald et al. / Bioorg. Med. Chem. Lett. 21 (2011) 6652–6656
6655
Table 2
Table 4
In vitro data for compounds 3 and 33–43
In vitro data for compounds 37 and 52–63
O
R2
N
H
H
N
O
N
N
F
H
O
H
N
O
N
N
O
O
O
Cl
R1
Compda
R2
PDE7A1b IC50
,
l
M
PDE4A4b IC50
, lM
Compda
R1
PDE7A1b (IC50
,
lM)
PDE4A4b (IC50
,
lM)
37
52
53
54
55
56
57
58
59
60
61
62
63
4-N-morpholinyl
4-NEt2
4-NMe2
4-NHCOMe
4-COMe
4-CONH2
3-CONH2
4-CONHMe
4-SO2CH3
3-SO2CH3
4-SO2NH2
0.013
0.060
0.039
0.017
0.018
0.010
0.044
0.018
0.012
0.059
0.020
0.032
0.034
>1.00c
>5.00c
0.402
0.362
0.161
0.090
0.580
0.156
0.073
1.42
3
2,6-di-F
H
2-Me
2-CF3
2-Cl
2-Cl, 6-F
2,6-di-Cl
4-F
4-OCF3
2,4-di-F
3,5-di-F
0.022
0.712
0.266
>1.00
0.767
0.013
0.165
2.38
>1.00
>1.00
>1.00
0.317
0.506
>1.00
>1.00
>1.00
>1.00
>1.00
>1.00
>5.00
>1.00
>1.00
>1.00
3.56
33
34
35
36
37
38
39
40
41
42
43
0.385
0.274
>1.00
4-SO2NH-thiazol-2-yl
3-Cl, 4-OMe
2,3,5,6-tetra-F
a
Compounds 37 and 52–62 prepared by route A; compound 63 prepared by
route B.
a
All compounds prepared by route A.
Measured against the human full length enzyme produced in baculovirus
b
b
Measured against the human full length enzyme produced in baculovirus
infected sf21 cells.21
infected sf21 cells.21
c
Applies also to human recombinant PDE1B, 2A, 3A, 5A and 8A–11A, and to PDE6
from bovine retina.
Table 3
In vitro data for compounds 32 and 44–51
the most potent PDE7 inhibitor 37 with a selectivity ratio greater
than 77 versus PDE4A.
O
O
O
Substitution at the para position was shown to be detrimental
to the enzymatic activity (39–41), while a different arrangement
or an increase in the number of F groups (42 and 43) was also
not beneficial.
As 32 did not much differ in terms of activity and stability from
the reference compound 3 we decided to verify the optimal substi-
tution pattern at the benzyl moiety with the preparation of a fur-
ther small subseries (Table 3).
Except for 46, a similar tendency to that presented in Table 2
was observed. The fivefold improvement in potency prompted us
to investigate in more detail ortho halo substituted derivatives
(47–50) but the preferred substituent arrangements still remained
with those found in 32 and 44.
Finally, the most promising structural features from the previ-
ous subseries were combined to complete the SAR studies
(Table 4). Synergistic effects with regard to selectivity versus
PDE4 could particularly be demonstrated with compound 52
(>83-fold).
H
H
N
O
N
N
O
O
R1
Compda
R1
PDE7A1b (IC50
,
lM)
PDE4A4b (IC50
, lM)
32
44
45
46
47
48
49
50
51
2,6-di-F
2-Cl, 6-F
H
2-Cl
2-F
0.018
0.033
0.686
0.178
0.206
>1.00
0.192
0.088
0.868
0.516
>1.00
>1.00
>1.00
>1.00
>1.00
>1.00
0.979
>1.00
2-Br
2-Cl, 4-F
2,6-di-Cl
4-F
In general, smaller groups than dimethylamino caused a signif-
icant drop in selectivity (53–61) with ratios varying from 6- to
24-fold. Shifting the functionality from para to meta position
(57 and 60) resulted again in a loss of activity, although this time
the impact was less pronounced. The trimethoxy substitution
pattern (44) could be replaced while maintaining affinity by the
introduction of a 3-chloro, 4-methoxy template (63).
With the exception of 52 and 62 all compounds were slightly
more potent compared to their analogs presented in Table 1.
In summary, the dihydronaphthyridinedione derivatives repre-
sent a new class of PDE7 inhibitors. A significant improvement in
activity and selectivity was achieved starting from the weak dual
PDE4/PDE7 inhibitor hit compound 1 by using molecular modeling
a
All compounds prepared by route B.
b
Measured against the human full length enzyme produced in baculovirus
infected sf21 cells.21
We next explored the SAR around the benzyl ring while retain-
ing the morpholinophenyl substituent at the amide group
(Table 2). Removal of the two fluorine atoms caused a large drop
in potency (33). An ortho monosubstitution alone turned out not
to be sufficient to regain high activity levels (34–36).
Consequently, we then evaluated the impact of the replacement
of one or both fluoro by chloro substituents (37 and 38) leading to

6656
R. Gewald et al. / Bioorg. Med. Chem. Lett. 21 (2011) 6652–6656
and modification of R1 and R2 chemical groups. The best compounds
tested 37 and 52 displayed low nanomolar inhibitory activity and
also showed good selectivity versus all other PDE isoenzymes. Fur-
ther optimization of these newly discovered leads will be supported
by the findings from this study.
37; (c) Bergmann, J. E.; Cutshall, N. S.; Demopulos, G. A.; Florio, V. A.;
Gaitanaris, G.; Gray, P.; Hohmann, J.; Onrust, R.; Zeng, H. PCT Int. Appl. WO
129036 A1, 2010.
6. (a) Vergne, F.; Bernardelli, P.; Chevalier, E. Annu. Rep. Med. Chem. 2005, 40, 227;
(b) Gil, C.; Campillo, N. E.; Perez, D. I.; Martinez, A. Expert Opin. Ther. Patents
2008, 18, 1127.
7. Wang, H.; Liu, Y.; Chen, Y.; Robinson, H.; Ke, H. J. Biol. Chem. 2005, 280, 30949.
8. Clark, T., Alex, A., Beck, B., Chandrasekhar, J., Gedeck, P., Horn, A., Hutter, M.,
Acknowledgments
Martin, B., Rauhut, G., Sauer, W., Schindler, T., and Steinke, T.
VAMP (version
7.5a). 1999, The Oxford Molecular Group, The Medawar Centre, Oxford Science
Park, Oxford OX4 4GA, UK.
9. GOLD (version 5.0.1). 2011, CCDC Software Ltd, Cambridge.
10. Conti, M. Nat. Struct. Mol. Biol. 2004, 11, 809.
The project was supported by the European Fund for Regional
Development (EFRE) and the Free State of Saxony (SAB 8093).
11. Ke, H.; Wang, H. Curr. Top. Med. Chem. 2007, 7, 391.
12. Zhang, K. Y.; Card, G. L.; Suzuki, Y.; Artis, D. R.; Fong, D.; Gillette, S.; Hsieh, D.;
Neiman, J.; West, B. L.; Zhang, C.; Milburn, M. V.; Kim, S. H.; Schlessinger, J.;
Bollag, G. Mol. Cell 2004, 15, 279.
13. Wallace, A. C.; Laskowski, R. A.; Thornton, J. M. Prot. Eng. 1995, 8, 127.
14. Heindl, J.; Kelm, H.-W.; Dogs, E.; Seeger, A.; Herrmann, C. Eur. J. Med. Chem.
Chim. Ther. 1977, 12, 549.
References and notes
1. (a) Beavo, J. A. Physiol. Rev. 1995, 75, 725; (b) Conti, M. In Principles of Molecular
Regulation; Conn, P. M., Means, A. R., Eds.; Humona, 2000; pp 261–275; (c)
Soderling, S. H.; Beavo, J. A. Curr. Opin. Cell Biol. 2000, 12, 174.
2. (a) Soderling, S. H.; Bayuga, S. J.; Beavo, J. A. J. Biol. Chem. 1998, 273, 15553; (b)
Conti, M.; Jin, S. L. C. In Progress in Nucleic Acid Research and Molecular Biology;
Moldave, K., Ed.; Academic Press: New York, 1999; Vol. 63, pp 1–38.
3. (a) Liu, Y.; Goff, R. D.; Zhou, D.; Mattner, J.; Sullivan, B. A.; Khurana, A.; Cantu, C.,
III; Ravkov, E. V.; Ibegbu, C. C.; Altman, J. D.; Teyton, L.; Bendelac, A.; Savage, P.
B. J. Immunol. Methods 2006, 312, 34; (b) Kadoshima-Yamaoka, K.; Murakawa,
M.; Goto, M.; Tanaka, Y.; Inoue, H.; Murafuji, H.; Nagahira, A.; Hayashi, J.;
Nagahira, K.; Miura, K.; Nakatsuka, T.; Chamoto, K.; Fukuda, Y.; Nishimura, T.
Immunol. Lett. 2009, 122, 193; (c) Kadoshima-Yamaoka, K.; Murakawa, M.;
Goto, M.; Tanaka, Y.; Inoue, H.; Murafuji, H.; Nagahira, A.; Hayashi, J.; Nagahira,
K.; Miura, K.; Nakatsuka, T.; Chamoto, K.; Fukuda, Y.; Nishimura, T. Int.
Immunopharmacol. 2009, 9, 97.
4. Goto, M.; Kadoshima-Yamaoka, K.; Murakawa, M.; Yoshioka, R.; Tanaka, Y.;
Inoue, H.; Murafuji, H.; Kanki, S.; Hayashi, J.; Nagahira, K.; Ogata, A.; Nakatsuka,
T.; Fukuda, Y. Eur. J. Pharmacol. 2010, 633, 93.
5. (a) Perez-Torres, S.; Cortes, R.; Tolnay, M.; Probst, A.; Palacios, J. M.; Mengod, G.
Exp. Neurol. 2003, 182, 322; (b) de Gortari, P.; Mengod, G. Brain Res. 2010, 1310,
15. Hoffmann, J. A.; Tilak, M. A. Org. Prep. Proced. Int. 1975, 7, 215.
16. Burwell, R. L., Jr. Chem. Rev. 1954, 54, 615.
17. Prepared by using 2-aminopyridine as starting material instead of A.
18. Corresponds to precursor F in the preparation of 3.
19. Prepared from the corresponding ethyl ester intermediate F.
20. Prepared from 28 (BBr3, CH2Cl2, reflux).
21. PDE assays: PDE4A1 and PDE7A1 enzymes were generated from full-length
recombinant human clones, expressed in Sf21 cells. The enzymatic activity was
determined using
a plate based Scintillation Proximity Assay (SPA) at a
substrate concentration of 20 nM and 500 nM [3H]-cAMP for PDE7A1 and
PDE4A4, respectively. The incubation in 80 mM Tris/10 mM MgSO4 (pH 7.4)
was allowed to proceed for 30 min at 37 °C before addition of 25
beads. Plates were counted in a Trilux plate reader. The background was
determined without enzyme for PDE7A1 and in the presence of 10
Rolipram for PDE4A4, respectively. IC50 values are means of at least four
experiments.
ll of the SPA
l
M