Elucidation of a structural basis for the inhibitor-driven, p62 (SQSTM1)-dependent intracellular redistribution of cAMP phosphodiesterase-4A4 (PDE4A4)

Article abstract of DOI:10.1021/jm200070e

Figure Presented. A survey of PDE4 inhibitors reveals that some compounds trigger intracellular aggregation of PDE4A4 into accretion foci through association with the ubiquitin-binding scaffold protein p62 (SQSTM1). We show that this effect is driven by i

Full text of DOI:10.1021/jm200070e

ARTICLE
pubs.acs.org/jmc
Elucidation of a Structural Basis for the Inhibitor-Driven, p62
(SQSTM1)-Dependent Intracellular Redistribution of cAMP
Phosphodiesterase-4A4 (PDE4A4)
Jonathan P. Day,† Barbara Lindsay,‡ Tracy Riddell,† Zhong Jiang,‡ Robert W. Allcock,‡ Achamma Abraham,‡
Sebastian Sookup,† Frank Christian,† Jana Bogum,§ Elisabeth K. Martin,‡ Robert L. Rae,‡ Diana Anthony,†
Georgina M. Rosair,‡ Daniel M. Houslay,† Elaine Huston,† George S. Baillie,† Enno Klussmann,§
Miles D. Houslay,† and David R. Adams*,‡
†Molecular Pharmacology Group, Division of Biochemistry and Molecular Biology, Institute of Biomedical and Life Sciences,
University of Glasgow, Glasgow G12 8QQ, U.K.
‡Department of Chemistry, School of Engineering and Physical Sciences, Heriot-Watt University, Riccarton Campus,
Edinburgh EH14 4AS, U.K.
§Max-Delbr€uck-Centrum f€ur Molekulare Medizin (MDC), Berlin-Buch, Robert-R€ossle-Strasse 10, 13125 Berlin, Germany
S Supporting Information
b
ABSTRACT: A survey of PDE4 inhibitors reveals that some
compounds trigger intracellular aggregation of PDE4A4 into
accretion foci through association with the ubiquitin-binding
scaffold protein p62 (SQSTM1). We show that this effect is
driven by inhibitor occupancy of the catalytic pocket and
stabilization of a “capped state” in which a sequence within the
enzyme’s upstream conserved region 2 (UCR2) module folds
across the catalytic pocket. Only certain inhibitors cause PDE4A4
foci formation, and the structural features responsible for driving
the process are defined. Switching to the UCR2-capped state
induces conformational transition in the enzyme’s regulatory N-terminal portion, facilitating protein association events responsible for
reversible aggregate assembly. PDE4-selective inhibitors able to trigger relocalization of PDE4A4 into foci can therefore be expected to
exert actions on cells that extend beyondsimpleinhibition of PDE4 catalytic activity and that may arise fromreconfiguringthe enzyme’s
protein association partnerships.
’ INTRODUCTION
tissue injury.13,14 The resulting local down-regulation of cAMP
suppresses fibrinolysis, and the action of PDE4A4 has consequently
been implicated in the pathophysiological processes underlying
fibrosis.13 Inhibition of this particular isoform, or its targeting to
p75NTR, may therefore have useful antifibrotic potential in diseases
such as chronic obstructive pulmonary disease (COPD), where
PDE4A4 is specifically up-regulated in the lungs of patients15 and
also in spinal cord repair.16 COPD patients also frequently suffer
from sleep deprivation,17 a condition that, in rodents, causes co-
gnitive deficits through the specific up-regulation of the rodent
PDE4A4 orthologue, PDE4A5, in the hippocampus.18
We have previously shown that chronic exposure to certain
PDE4-selective and competitive inhibitors, such as rolipram,
Ro20-1724, and RS-25344, in addition to inhibition of the cAMP
hydrolase activity of PDE4A4, also triggers a profound intracel-
lular redistribution of the enzyme into compact accretion foci
(Figure 1).19 This fully reversible process is only associated with
Enzymes of the phosphodiesterase type 4 (PDE4) family
mediate hydrolytic degradation of cyclic adenosine 30,50-mono-
phosphate (cAMP), thereby regulating intracellular concentra-
tion gradients of this important second messenger. These
enzymes are encoded by four genes (PDE4AꢀD)1,2 that each
give rise to multiple isoforms.3ꢀ7 Cyclic AMP hydrolysis activity is
provided by a metallohydrolase core catalytic domain (Figure 1A),
which is identical for isoforms arising from any one gene and
highly homologous across different gene products. Individual
isoforms are distinguished by a unique N-terminal signature
sequence that plays a role in targeting the protein to specific
signaling complexes and intracellular sites.2,8ꢀ11 The distinct
targeting characteristics of PDE4 isoforms are important for
controlling the overall spatial integrity of cAMP signaling and
orchestrating cross-talk between signaling pathways.2,12 In this
context, it has recently been shown that a fraction of PDE4A4 can
be targeted to the plasma membrane in cells expressing the p75
neurotrophin receptor (p75NTR), a member of the tumor necrosis
factor (TNF) receptor superfamily that is up-regulated following
Received: January 21, 2011
Published: April 04, 2011
r
2011 American Chemical Society
3331
dx.doi.org/10.1021/jm200070e J. Med. Chem. 2011, 54, 3331–3347
|

Journal of Medicinal Chemistry
ARTICLE
Figure 1. Structural organization of PDE4A4 and model for foci formation. (A) Long PDE isoforms such as PDE4A4 comprise a unique N-terminal
sequence connected to a core catalytic domain by two regulatory modules, upstream conserved regions 1 and 2 (UCR1 and UCR2), in series with linker
regions (LR1 and LR2). The UCR modules transduce the functional consequences of regulatory phosphorylation by cAMP-dependent protein kinase
(PKA). PDE4A4 recruitment to p75NTR targets cAMP hydrolase activity to the plasma membrane and has been linked to the pathophysiological
suppression of fibrinolysis contributing to fibrosis. Inhibition of PDE4A4 has an antifibrotic action, but in some cases (rolipram, Ro20-1724 and
RS-25344) inhibitor occupancy of the catalytic pocket also causes the enzyme’s accretion into intracellular foci by inducing or stabilizing protein binding
to the N-domain. Other established PDE4 inhibitors (cilomilast and piclamilast) fail to induce foci and antagonize rolipram-induced foci formation.
(B) Redistribution of PDE4A4 into accretion foci is most conveniently monitored by confocal microscopy in live CHO cells stably transfected to express
a PDE4A4 construct tagged at the C-terminus with green fluorescent protein (GFP). Left, cytoplasmic distribution of PDE4A4-GFP in untreated cells;
right, redistribution of PDE4A4-GFP into foci (bright spots) after incubation with (rac)-rolipram (3 μM, 10 h).
human PDE4A4 (and the rodent orthologue, PDE4A5) and is
not seen with other PDE4 species (including all other known
PDE4A isoforms6,7), indicating a specific, additional require-
ment for the unique N-terminal region that characterizes
PDE4A4/5. Recently we have further shown that the inhibitor-
induced PDE4A4 foci constitute a novel intracellular aggregate
state for this enzyme with links to both autophagy and the pro-
teasome system through the association of PDE4A4 with the
signaling scaffold and ubiquitin-binding protein p62 (sequesto-
some-1, SQSTM1).20 This protein, which is required for the
formation of autophagy vesicles,21 is also associated with ubiquiti-
nated protein aggregates seen in neurodegenerative diseases, such
as Lewy bodies in Parkinson's disease, neurofibrillary tangles
in Alzheimer's disease, and huntingtin aggregates in Hunting-
ton's disease as well as Mallory bodies of alcoholic and non-
alcoholic steatohepatitis and hyaline bodies in hepatocellular
carcinoma.22,23
partnerships with key proteins linked to fibrosis and autophagy,
indicates that such compounds are likely to exert effects that
extend beyond those elicited simply by increasing intracellular
cAMP levels. An understanding of the SAR surrounding com-
pounds that trigger this action may therefore have an important
bearing on the future development of pharmacologically useful
PDE4 inhibitors. Our initial studies19 showed that the effect is
driven by inhibitor occupancy of the catalytic pocket but not by
elevation of cAMP levels consequent upon inhibition of the
enzyme’s phosphodiesterase activity. Remarkably, not all PDE4
inhibitors were found to elicit foci formation despite targeting the
same catalytic pocket binding site. The structural determinants in
an inhibitor responsible for the effect therefore required defini-
tion. Here we address this issue, providing insight into the
mechanistic basis underpinning the effect.
’ RESULTS AND DISCUSSION
It is now well-appreciated that the intracellular location of
PDE4 species and their recruitment to specific signaling com-
plexes are critical to their functional role in cells not only by
underpinning compartmentalized cAMP signaling2 but, through
sequestration of specific proteins, in regulating other signaling
pathways, like that of mTor.24 That certain PDE4 inhibitors can
cause the redistribution of PDE4A4, and thereby reconfigure its
Survey of Established PDE4 Inhibitors for Foci-Forming
Activity. We previously showed19 that both enantiomers of
rolipram, Ro20-1724 and RS-25344 induce foci formation with
PDE4A4, whereas cilomilast and piclamilast competitively antag-
onize rolipram-induced foci formation. In order to probe the ligand
features responsible for driving the redistribution, we have now
3332
dx.doi.org/10.1021/jm200070e |J. Med. Chem. 2011, 54, 3331–3347

Journal of Medicinal Chemistry
ARTICLE
Figure 2. Survey of established PDE4 inhibitors for foci-inducing activity. Compounds were tested according to the procedure described in Terry et al.19
(A) Structures of PDE4 inhibitors found to induce PDE4A4 foci formation. (B) List of compounds that do not elicit foci formation and that block rolipram-
induced foci formation; structures of representative compounds are shown and a full listing by structure is available in the Supporting Information.
undertaken an extensive survey of established PDE4 inhibitors
and synthesized novel compounds to gain insight into the mechan-
istic basis for this action. Most of the existing PDE4 inhibitors
evaluated were non-foci-inducing (Figure 2B) and, as might be
expected for active-site directed inhibitors, served to block roli-
pram-induced foci formation. The foci-inducing compounds fell
into three distinct structural classes. The catechol ethers, rolipram
and Ro20-1724, identified in our earlier study19 constitute one
group (Figure 2A, class I). None of the other 15 catechol ether
inhibitors tested (Figure 2B) possessed foci-inducing activity.
A second group of foci-inducing compounds (Figure 2A, class II)
were all structurally related to RS-25344. Five new compounds
were found in this class, namely, NVP, RO-458,2640, RS-14203,
RS-33793, and RO-106,7146. These compounds are all based on
a 6-substituted 8-arylquinoline template with various combinations
of aza and oxo modification in the core quinoline benzo ring. One
xanthine-based PDE4 inhibitor, cipamfylline (Figure 2A, class III)
was also found to induce foci formation. In contrast, two other
established xanthine inhibitors, denbufylline and IBMX (Figure 2B),
which are not selective for PDE4 and inhibit various other PDE
families, lacked foci-inducing capacity.
Rolipram Lactam Engagement of Metal Centers Is Not
Responsible for Foci Formation. The archetypal PDE4-selec-
tive inhibitor, rolipram, shares an identical 3-cyclopentyloxy-4-
methoxyphenyl core subunit with a number of foci-blocking
PDE4-selective inhibitors (piclamilast, cilomilast, and CDP-
840). This consideration, reinforced by the observation of a
common catechol ether binding mode in the available PDE4
cocrystal structures of rolipram,25ꢀ28 piclamilast,25 and cilo-
milast25 (Figure 3), initially suggested that the determinants for
foci induction might reside with the inhibitor auxiliary subunits.
These structurally diverse groups, corresponding to the lactam
ring in rolipram, are bound proximal to the enzyme’s catalytic
center, where a more deeply set and tightly bound Zn2þ ion
(Figure 3 Me1) is paired with a second bivalent metal ion
(Me2), which is typically Mg2þ
.
At the outset of our work, the coordinates of six rolipram-bound
PDE4 core catalytic domain cocrystal structures, acquired either
with racemic material or with the (R)-antipode, were available in
the RCSB Protein Data Bank. In five structures (PDB codes 1OYN,
1OYM, 1TBB, 1XMY, 1XN0)25,26,28 the rolipram lactam ring (of
either enantiomer) makes no direct contact with the metal centers
3333
dx.doi.org/10.1021/jm200070e |J. Med. Chem. 2011, 54, 3331–3347

Journal of Medicinal Chemistry
ARTICLE
Figure 3. Comparison of rolipram, cilomilast, and piclamilast PDE4-binding modes. (A) PDE4 core catalytic domain with (R)-rolipram (yellow stick,
PDB code 1XN0)25 bound in a solvent-orientated lactam conformation within the catalytic pocket (residues in green stick); mutation of residues
highlighted in orange stick has been shown19 to ablate rolipram-induced foci formation. (B) Overlay of cilomilast (magenta stick) and piclamilast (cyan
stick) from their respective PDE4 cocrystal structures (PDB codes 1XLX, 1XM4)25 onto (R)-rolipram (yellow stick, PDB code 1XN0)25 showing
common binding mode for the core catechol ether subunit. (C) Detail of alternative rolipram binding modes from the 1RO6 PDE4 cocrystal structure:27
(R)-enantiomer (yellow stick) in metal-orientated mode, (S)-enantiomer (orange stick) in solvent-orientated mode.
but adopts a solvent-orientated conformation with the ring directed
toward the opening of the catalytic pocket. In one structure (PDB
code 1RO6), acquired with (rac)-rolipram, a mixed population was
observed with the inhibitor bound in both the solvent-orientated
state and a metal-orientated conformation, the latter with the
lactam carbonyl functionality engaging the metal centers and their
bridging hydroxide ligand (Figure 3C).27 Although both enantio-
merscouldbefitted to the electron density for either binding mode,
with a slight difference in the axis of the carbonyl group, the
electron density for the solvent-orientated binding mode was
provisionally attributed to the (S)-enantiomer and that for the
metal-orientated mode to the (R)-isomer. In this crystal structure
the distance between the Me2 center and the purine-scanning
glutamine, the residue that anchors the rolipram catechol ether
subunit with hydrogen bonds, is some 3ꢀ4% shorter than in the
other cocrystal structures with rolipram, piclamilast, and cilomilast.
This slightly contracted state allows rolipram to fit between the
opposite ends of the pocket, simultaneously engaging the purine-
scanning glutamine and the bridging hydroxide at the catalytic
center.29 Thus, the special feature of rolipram as a foci-forming
inhibitor might, in principle, be its ability to either induce or
stabilize such a contracted state in the cellular environment,
perhaps with induced movement in the metal centers providing a
basis for the structural transition underpinning foci formation.
We therefore set out to determine whether a metal-orientated
enzyme-bound conformation of rolipram is responsible for the
compound’s foci-inducing capacity. To this end we prepared and
evaluated a spirocyclic, torsionally constrained analogue (6,
Scheme 1) of rolipram. Molecular models suggested that (R)-6
ought to closely mimic the metal-orientated enzyme-bound
conformation of (R)-rolipram in the 1RO6 cocrystal structure.
Similarly, (S)-6 might mimic (S)-rolipram bound in a conforma-
tion with the lactam carbonyl orientated toward the enzyme’s
metal centers. Spirocycle 6 was initially prepared in racemic form
according to Scheme 1. Thus, the 4-methoxy group of commercial
dimethoxyindanone 1 was selectively demethylated as described
by Kemperman et al.30 Phenol 2 was alkylated with cyclopentyl
bromide and the resulting indanone (3) converted into nitrile 4
by treatment with p-toluenesulfonylmethyl isocyanide. Alkylation
of 4 by treatment with sodium hexamethyldisilazide followed by
methyl bromoacetate afforded cyanoester 5. Synthesis of (rac)-6
was completed by hydrogenation of the latter at elevated tem-
perature and pressure.
The single crystal X-ray structure of 6 (Figure 4) confirmed
that the compound’s (R)-enantiomer should be a close structural
match for (R)-rolipram bound in the metal-orientated state of
the 1RO6 cocrystal structure. However, when tested at concen-
trations up to 30 μM, spirocycle 6 failed to elicit foci formation in
3334
dx.doi.org/10.1021/jm200070e |J. Med. Chem. 2011, 54, 3331–3347

Journal of Medicinal Chemistry
ARTICLE
Scheme 1a
a Reagents and conditions: (a) Me3N HCl, AlCl3, CH2Cl2, reflux (93%); (b) bromocyclopentane K2CO3, DMF, 100 °C (85%); (c) TOSMIC, tBuOK,
3
MeOH, MeOCH2CH2OMe, 0 °C to room temp (50%); (d) NaHMDS, THF, ꢀ78 °C, then BrCH2CO2Me (68%); (e) H2 (16 bar), 10% PdꢀC, 60 °C,
EtOH, 43 h (66%); (f) LiOH aq, THF, 0 °C (quant); (g) EDC, HOBt, DIPEA, (S)-1-phenylethylamine (1:1 mixture 98%; separated samples of 8a
and 8b were obtained by HPLC); (h) H2, Raney Ni, EtOH, 90 bar, 90 °C (H-cube); (i) p-xylene, 138 °C, 18 h [two steps, 6S (75%), 6R (38%)];
(j) p-bromobenzyl bromide, NaH, DMF (88%).
derivative (9) and acquisition of a single crystal X-ray diffraction
structure (Figure S1, Supporting Information). The separate spiro-
cycle enantiomers (6R/6S) both exhibited low micromolar PDE4
inhibitory potency (Table 1) but failed to induce PDE4A4 foci.
N-Alkylated Derivatives of (R)-Rolipram Retain Foci-Indu-
cing Activity. The failure of spirocycle 6 to induce foci suggested
that the enzyme-bound state of rolipram responsible for the foci
formation may adopt a solvent-orientated lactam conformation.
Figure 4. Single crystal X-ray structure of spirocycle 6 (green stick)
In order to test this proposal, we next evaluated a set of N-alkyl
rolipram derivatives, as restricted space in the catalytic pocket pro-
ximal to the metal centers (Figure 3C) should sterically preclude
the metal-orientated lactam arrangement for such compounds
and, effectively, constrain them to a solvent-orientated binding
mode. (Bulky structural extensions from the rolipram lactam
nitrogen, presumably directed to subsites around the rim of the
catalytic pocket, are already known to support or enhance PDE4-
inhibitory activity.)31 Accordingly, we prepared both antipodes
of N-methyl and N-benzyl rolipram for assessment (Scheme 2).
Significantly, the foci-inducing capacity of (R)-rolipram was re-
tained in both its N-methyl and N-benzyl derivatives (respec-
tively, 10R and 11R), supporting the notion that the inhibitor
binding with its lactam ring in a solvent-orientated conformation
drives foci formation. In contrast, the foci-forming activity of (S)-
rolipram was ablated by both N-methylation and benzylation,
although derivatives 10S and 11S retained activity for inhibition
of cAMP hydrolysis by PDE4A4 (Table 1).
shows that the semirigid core of the structure closely superimposes onto
the metal-orientated (R)-rolipram conformation (yellow stick) in the
1RO6 crystal structure.
CHO cells expressing PDE4A4-GFP. Moreover, at 30 μM, (rac)-
6 severely inhibited (79 ( 7%; mean ( SD, n = 3) foci induction
by (rac)-rolipram (1 μM) and dispersed foci formed by pretreat-
ment of cells with (rac)-rolipram, thus confirming that the
compound does compete effectively for the enzyme’s catalytic
pocket. In order to evaluate the PDE4 inhibitory activity of
spirocycle 6 more fully, we prepared the enantiomers for separate
assessment by an enantiodivergent extension to the Scheme 1
route. Thus, hydrolysis of ester 5 followed by coupling of the
resulting acid (7) with (S)-1-phenylethylamine afforded a mix-
ture of diastereomeric amides (8a/8b). Chromatographic se-
paration of these compounds followed by hydrogenation of the
nitrile and cyclization in hot xylene gave the spirocycle enantio-
mers (6R/6S) for independent evaluation. Unfortunately the
intermediate amides (8a/8b) did not afford the diffraction quality
crystals required for stereochemical elucidation by X-ray crystal-
lography. The absolute configuration of the indanyl stereocenter
was therefore determined by subsequent conversion of one of the
spirocyclic lactams (6S) into its crystalline N-(p-bromobenzyl)
The Cyclopentyloxy Group Drives Rolipram-Induced Foci
Formation. The failure of 10S and 11S to induce accretion foci
was surprising because the available cocrystal structures show the
lactam NꢀH bond axis of (S)-rolipram to be orientated directly
toward the opening of the catalytic pocket. This suggested that
3335
dx.doi.org/10.1021/jm200070e |J. Med. Chem. 2011, 54, 3331–3347

Journal of Medicinal Chemistry
ARTICLE
Table 1. PDE4A4 Inhibitory Activity and Capacity To Induce
Foci Formation for Test Compoundsa
Scheme 2a
PDE4A4 inhibition
foci induction
compd
calcd Ki (μM)b
IC50 (μM)
EC50 (μM)
(rac)-rolipram
(R)-(ꢀ)-rolipram
(S)-(þ)-rolipram
Ro20-1724
NVP
0.58
0.38
3.0
0.83 ( 0.06
0.55 ( 0.4
4.3 ( 0.3
4.3 ( 0.5
yes 0.50 ( 0.08
yes 0.48 ( 0.09
yes 3.4 ( 0.9c
yes 3.3 ( 0.2
3.0
0.020
0.24
0.015
0.0077
0.20
0.24
1.1
0.029 ( 0.004 yes 0.020 ( 0.008
0.34 ( 0.02 yes 0.22 ( 0.03
RS-14203
RS-25344
RS-33793
RO-458,2640
RO-106,7146
cipamfylline
6S
0.022 ( 0.007 yes 0.017 ( 0.003
0.011 ( 0.006 yes 0.015 ( 0.007
a Reagents and conditions: (a) MeI, NaH, DMF (10R 86%, 10S 63%);
(b) BnBr, NaH, DMF (11R 76%, 11S 94%).
0.29 ( 0.05
0.34 ( 0.04
1.6 ( 0.25
12 ( 2
yes 0.25 ( 0.04
yes 0.27 ( 0.03
yes 0.91 ( 0.3
no n/a
strategy similar to that used for synthesis of the spirocycle (6R/
6S, Scheme 1). A related route was used by Demnitz and
colleagues to prepare the parent rolipram enantiomers.34 Thus,
conjugate addition of the nitromethane anion to enone 13
afforded adduct 14. The latter was converted in three steps into
a pair of diastereomeric amides (17a/17b) derived from (S)-
1-phenylethylamine. Chromatographic separation of these com-
pounds followed by hydrogenation and cyclization of the inter-
mediate aminoamides at elevated temperature afforded the
enantiomeric lactams 18R and 18S. We had commenced with
a benzyloxy group at the 3-position of the precursor arene (13)
so that the hydrogenation step in the route would also unmask
the phenolic function in 18R and 18S, allowing convenient
parallel alkylation as a final step to deliver the target compound
set. In this way we prepared the 3-methoxy, ethoxy, and
propyloxy analogues of both rolipram antipodes. The absolute
stereochemistry of the two enantiomer series was determined by
alkylation of the phenol derived from 17b with iodocyclopen-
tane; polarimetry identified the product as (S)-(þ)-rolipram.
Replacement of the rolipram cyclopentyloxy group by meth-
oxy (compounds 19R/S) reduced the compound’s PDE4-in-
hibitory activity by an order of magnitude (Table 1). Such activity
loss is reasonable given the expected reduction in surface contact
between the inhibitor and protein. Significantly, neither enantio-
mer (19R/S) possessed foci-inducing activity at concentrations
up to 100 μM. However, both suppressed induction by (rac)-
rolipram (1 μM) at these concentrations. Increasing the size of
the Q2-targeted substituent from methoxy to ethoxy (20R/S) or
propyloxy (21R/S) unsurprisingly increased the PDE4-inhibi-
tory potency (Table 1), although the compounds were some-
what less effective than the respective parent rolipram enan-
tiomers. Significantly, these compounds were all found to induce
accretion foci at concentrations comparable to their PDE4-
inhibitory potencies. Thus, substituent occupancy of the Q2 sub-
pocket appears to be a key parameter that controls an inhibitor’s
capacity to induce foci, with the critical size threshold for foci
induction reached with an ethoxy group in the case of the
rolipram analogue series.
8.4
6R
17
24 ( 5
no n/a
10R
0.7
1.0 ( 0.3
2.4 ( 0.2
4.3 ( 0.8
0.73 ( 0.09
38 ( 8
yes 0.5 ( 0.2
no n/a
10S
1.7
11R
2.9
yes 2.1 ( 0.9
no n/a
11S
0.51
27
19R
no n/a
19S
31
44 ( 14
no n/a
20R
4.5
6.5 ( 0.6
6.4 ( 0.8
6.2 ( 0.8
8.2 ( 1.5
5.4 ( 1.0
1.3 ( 0.22
yes 5.6 ( 0.8
20S
4.5
yes
yes 4.6 ( 1.9
yes
c
21R
4.3
21S
5.7
c
28
3.8
no n/a
ibudilast
0.91
yes 0.90 ( 0.1
a Data are given as the mean ( SD for n = 3 independent experiment
sets. IC50 values with 1 mM cAMP as substrate for PDE4A4 assays. n/a:
not applicable. b Ki values were calculated using the ChengꢀPrussof
relationship (Ki = IC50/(1 þ ([cAMP]/Km))), where the cAMP
substrate concentration was 1 mM and the Km for cAMP hydrolysis
by PDE4A4 was 2.3 mM. c Maximal accretion of PDE4A4 into foci with
(S)-enantiomer was maximally <20% of that achieved by the corre-
sponding (R)-enantiomer.
N-alkylation of (S)-rolipram obstructs an induced structural
transition that is pivotal for foci formation, perhaps the folding
or docking of a protein sequence (e.g., from the enzyme’s
regulatory N-terminal/UCR regions or a partner protein) across
the face of the catalytic pocket. Our previous identification32 of a
protein docking interface for RACK1 on the surface of the PDE4
catalytic unit close to the distal region of the catalytic pocket
suggested this idea and caused us to refocus our attention on the
role of the rolipram cyclopentyloxy group. This bulky substituent
occupies a hydrophobic subpocket (Q2, Figure 5) formed at the
junction of orthogonal helices 14 and 15 and their connecting
loop (Figure 3A). We considered that the cyclopentyloxy group
might stabilize binding of a protein sequence across the catalytic
pocket because, like the solvent-orientated lactam, it contributes
to the exposed edge of the inhibitor when bound within the
catalytic pocket.33 The cyclopentyloxy group might therefore be
essential for foci induction by rolipram. In order to evaluate this,
we tested a set of rolipram analogues with smaller alkoxy groups
targeted to the Q2 subpocket in place of the cyclopentyloxy. The
compounds were prepared (Scheme 3) by an enantiodivergent
The Nitrophenyl Group in NVP Drives Foci Formation.
During the course of our study, Ke and co-workers disclosed35
the first PDE4 cocrystal structures of a compound belonging to
the class II foci-inducing inhibitors that we have identified.
Structures were obtained for NVP in complex with the core
catalytic domains from the PDE4A, PDE4B and PDE4D sub-
families (PDB codes 2QYK, 2QYL, and 2QYN). A common
3336
dx.doi.org/10.1021/jm200070e |J. Med. Chem. 2011, 54, 3331–3347

Journal of Medicinal Chemistry
ARTICLE
NVP binding mode is seen in these three structures, with the
NVP naphthyridine N-1 center anchoring the inhibitor to the
purine-scanning Gln and the nitrophenyl group taking the
position of rolipram’s cyclopentyloxy substituent in the Q2 site
(Figure S2, Supporting Information). We thus predicted the
nitrophenyl group to be a key foci-driving subunit in NVP and
reasoned that an ethyl replacement substituent, which would
closely match the 3-methoxy group in 19R/S for size and bound
position, might abolish foci-inducing activity in this represen-
tative of the class II group. On this basis, the 8-ethylnapthyr-
idine analogue (28) of NVP was prepared for evaluation
(Scheme 4).
8-Bromo-1,7-naphthyridin-6-amine (22), an intermediate in
the synthesis of NVP, was prepared in three steps from pyri-
dine-3-acetonitrile according to the published procedure.36 Intro-
duction of the ethyl group was achieved by successive Sonogashira
coupling of (trimethylsilyl)acetylene, desilylation, and hydrogena-
tion to afford naphthyridinamine 25. The carboxyphenyl group
was then installed following the published route to NVP.36
Accordingly, diazotization of 25 in the presence trifluorometha-
nesulfonic acid furnished the triflate (26) which was subjected to
Suzuki coupling with 4-(methoxycarbonyl)benzeneboronic acid.
The resulting ester (27) was hydrolyzed under standard condi-
tions to afford target naphthyridine 28.
Evaluation of 28 in cells expressing PDE4A4-GFP demon-
strated, as predicted, that replacement of the NVP nitrophenyl
group by the smaller ethyl substituent abolishes the inhibitor’s
foci-inducing property. However, loss of PDE4-inhibitory
potency was also observed with 28 relative to NVP (Table 1).
Despite this, 28 clearly retains the capacity to bind in the enzyme’s
Figure 5. Detail of the PDE4 catalytic pocket taken from the PDE4
cocrystal structure, 1OYN,26 showing bound (R)-rolipram (yellow
stick) and its (S)-enantiomer (cyan stick). The cyclopentyloxy sub-
stituent occupies a hydrophobic subpocket formed at the junction of
helices 14 and 15 and their connecting loop. Card et al. define this site,
adjacent to the purine-scanning Gln, as the Q2 subpocket.25 The Q1
subpocket accommodates the rolipram methoxy group.
Scheme 3a
a Reagents and conditions: (a) Ph3PdCHCO2Me, CH3CN, reflux; (b) 1,1,3,3-tetramethylguanidine (cat.), MeNO2, 75 °C (70% two steps); (c) NaOH
aq, MeOH, 0 °C (94%); (d) SOCl2, CH2Cl2 (94%); (e) (S)-1-phenylethylamine (1:1 mixture 96%; 17a and 17b were then partially separated
chromatographically); (f) H2, 10% PdꢀC, EtOH, H2, EtOAc, EtOH, 90 bar, 90 °C (H-cube); (g) p-xylene, Δ (two steps, 18R 36%, 18S 81%);
(h) iodomethane, K2CO3, DMF (19R 87%, 19S 97%); (i) iodoethane, K2CO3, DMF (20R 78%, 20S 96%); (j) 1-iodopropane, K2CO3, DMF
(21R 95%, 21S 95%); (k) iodocyclopentane, K2CO3, DMF, 60 °C (38%).
3337
dx.doi.org/10.1021/jm200070e |J. Med. Chem. 2011, 54, 3331–3347

Journal of Medicinal Chemistry
ARTICLE
Scheme 4a
a Reagents and conditions: (trimethylsilyl)acetylene, Pd(PPh3)2Cl2, CuI, Et3N, THF, sealed vessel 50ꢀ70 °C (30%); (b) TBAF, 95% THF (aq) (85%);
(c) H2, Pd/C, EtOH (73%); (d) CF3SO3H, NaNO2, DMF, 0ꢀ20 °C (53%); (e) 4-(methoxycarbonyl)benzeneboronic acid, Pd(dba)2, Ph3P, Na2CO3
(aq), THF, reflux (77%); (f) LiOH aq, THF, MeOH (59%).
Figure 6. (A) Detail from the IBMX PDE4 cocrystal structure (PDB code 1ZKN37) highlighting key residues in the binding site (PDE4A4 numbering
3
used). (B) Key residueinteraction mapfor bound IBMX (H-bonded wateras red circle). (C) Adoption of an IBMX-like binding mode for cipamfylline directs
the N(1)-(cyclopropylmethyl) group into the Q2 subpocket to drive foci induction. Another binding mode (not shown) in which the cipamfylline 8-NH2 and
N-9 center engage Q642 would place the N(3)-(cyclopropylmethyl) group into the Q2 subpocket and is a conceivable alternative to that illustrated.
catalytic pocket and, consistent with this, at 10 μM concentration
28 severely inhibited (98 ( 5%; mean ( SD, n = 3) foci induction
by (rac)-rolipram (1 μM). The other five class II inhibitors of
Figure 2A are likely to share a related PDE4 binding mode with
NVP, and thus the feature that drives foci formation by these
compounds is predicted to be the chlorophenyl group in RO-
106,7146, the nitrophenyl unit in the RS series inhibitors, and the
benzo[c][1,2,5]oxadiazole appendage in RO-458,2640.
The N(1)-Cyclopropylmethyl Group of Cipamfylline Is
Predicted To Drive Foci Formation. Analysis of the structure
of cipamfylline, a foci-forming xanthine inhibitor that is structu-
rally distinct from the class I and class II compounds (Figure 2A),
suggests that it too is likely to drive the redistribution of PDE4A4
by insertion of a hydrophobic substituent into the Q2 subpocket.
No PDE4 cocrystal structure is available for cipamfylline, but a
cocrystal structure has been obtained with the structurally related
compound, IBMX,37 a nonselective PDE inhibitor that does not
trigger foci formation.19 In this structure the IBMX xanthine core
docks into the catalytic pocket with the imidazole ring orientated
toward the rear and the 1-methyl group oriented to the front,
directed toward the Q2 site (Figure 6A,B). The compound’s O6
center engages the purine-scanning Gln (PDE4A4:Gln642) and
a water molecule forms a hydrogen-bonded bridge from N(7) to
an asparagine residue in the rear wall of the pocket (PDE4A4:
Asn594). Adoption of a similar orientation for cipamfylline
(Figure 6C) would be expected to direct a cyclopropylmethyl
group into the same Q2 subpocket in which the cyclopentyloxy
and nitrophenyl groups of rolipram and NVP act to drive foci
induction. The failure of IBMX to show the effect could then be
explained by the lack of a sufficiently large substituent targeted to
the Q2 subpocket. Denbufylline (Figure 2B), another PDE4-
inhibitory xanthine, possesses substituents at N-1 and N-3 (butyl
groups) that are comparable in size to the cyclopropylmethyl
groups of cipamfylline. In the case of denbufylline, however, the
3338
dx.doi.org/10.1021/jm200070e |J. Med. Chem. 2011, 54, 3331–3347

Journal of Medicinal Chemistry
ARTICLE
Figure 7. (A) PDE4-bound (R)-rolipram (yellow stick, PDB code 1TBB28) superimposed into the enzyme’s catalytic pocket capped by the UCR2
gating sequence (orange ribbon, PDB code 3G4538); lactam N-alkyl groups may be accommodated by a notch in the mouth of the catalytic pocket. (B)
Similar superimposition to (A) but with (S)-rolipram (cyan stick, PDB code 1TBB); lactam N-alkylation is expected to abolish hydrogen bonding to
Tyr286 and obstruct capping.
2-oxopropyl substituent at N-7 is likely too bulky for accom-
modation in the confined Q1 subpocket within the interior of the
cAMP substrate binding cleft. The forced adoption of a binding
mode distinct from that observed for IBMX, and postulated here
for cipamfylline, may account for the lack of foci-inducing
capacity for denbufylline.
recent crystal structures of Burgin et al.38 support our hypothesis
(vide supra) that residues abutting and associated with the
catalytic pocket might present a protein docking site. On this
basis, we propose a mechanism in which inhibitor-stabilized
capping of the catalytic pocket by UCR2 might relay structural
transition into the isoform-specific PDE4A4 N-terminal region,
thereby inducing or stabilizing the proteinꢀprotein associations
involved in assembly of accretion foci. To confirm that stabiliza-
tion of the UCR2-capped state is a key requirement for this
process, we mutated Tyr286 and Phe291 independently to
Ala and found that both mutations in the UCR2 gating sequence
do indeed ablate rolipram-induced foci formation (>97% loss of
foci formation compared to WT PDE4A4; n = 3 separate
experiments).
Inhibitors Drive Foci Induction through Occupancy of the
Q2 Subpocket and Stabilization of a UCR2-Capped State. The
above analysis suggests a unified mechanistic basis for inhibitor-
induced PDE4A4 aggregation across the three structurally distinct
classes of compound defined in Figure 2A. In each case the process
is driven by the targeting of a moderately bulky hydrophobic or
aromatic substituent to the enzyme’s Q2 locus in the mouth of
the catalytic pocket. Very recently Burgin and co-workers have
managed to resolve PDE4 crystal structures that not only contain
the PDE4 core catalytic unit but also encompass part of the UCR2
regulatory domain.38 From this they have postulated that Q2
subpocket occupancy influences reversible folding of an R-helical
motif from thePDE4 regulatory UCR2 module acrossthe opening
of the substrate-binding cleft, thereby gating access to the catalytic
site.38 The binding of certain inhibitors within the catalytic pocket
may thus serve to stabilize this folding. Significantly, they were able
to acquire cocrystal structures for PDE4B and PDE4D in the
UCR2-capped state that were stabilized as complexes with RS-
14203 and RS-25344, both compounds belonging to the class II
foci-inducer group (Figure 2A). UCR2 capping was shown to be
critically dependent on the insertion of two aromatic residues from
the “UCR2 gating sequence”39 into the mouth of the catalytic
pocket. The cognate residues in PDE4A4 are Tyr286 and Phe291,
and the residue corresponding to Tyr286 in particular makes close
contact with the Q2-targeted nitrophenyl group of these inhibitors.
Inother respects the binding mode of the RS-14203 and RS-25344
core structures closely resembles that observed for NVP (Figure
S2, Supporting Information), and the key hydrogen bond from the
purine-scanning Gln (to N-1 in NVP) is conserved.
As recently discussed,39 the UCR2 capping process in PDE4
may involve a multimeric assembly with the N-termini of paired
protein molecules cross-capping the catalytic domains of one
another. The importance of dimerization to the function of PDE4
enzymes has excited debate, with controversy still surrounding
any possible physiological relevance of the crystallographic di-
merization interface (Figure 1A) that is observed almost uni-
versally in PDE4 core catalytic domain structures. A key Arg:Asp
salt bridge contributes to the crystallographic dimerization inter-
face, and Lee et al. have reported that mutations in this salt bridge
compromise dimerization of a PDE4D construct in solution.40 To
probe the relevance of the crystallographically observed core
catalytic domain dimer structure to formation of foci, we installed
a charge reversal mutation (Arg534Asp) to disrupt the corre-
sponding salt bridge in PDE4A4. Significantly, this mutation
also fully ablated induction of foci by rolipram (>96% loss of
foci formation compared to WT PDE4A4; n = 3 separate
experiments), indicating that assembly of PDE4A4 into foci
may involve the crossed-UCR2-capped state invoked from recent
structural studies.38,39
Foci Induction Is Sensitive to the Structure of Inhibitor
Auxiliary Subunits. While the presence of a Q2-targeted hydro-
phobic substituent of appropriate size is a key parameter that de-
termines the capacity of an inhibitor to drive formation of PDE4A4
accretion foci, the effect is clearly sensitive to the structure of the
Unfortunately, no crystal structures are currently available for
full-length PDE4A4 to provide structural detail for the regulatory
UCR1/2 domains and the isoform-specific N-terminal region
that is essential for the enzyme’s accretion into foci. However, the
3339
dx.doi.org/10.1021/jm200070e |J. Med. Chem. 2011, 54, 3331–3347

Journal of Medicinal Chemistry
ARTICLE
inhibitor auxiliary subunit. This sensitivity is strikingly demon-
strated in the N-alkyl derivatives of rolipram, where alkylation
destroys the foci-inducing activity of (S)-rolipram but not (R)-
rolipram (vide supra). A UCR2-capped PDE4 cocrystal structure
for rolipram was not obtained by Burgin et al.38 However, super-
imposition of bound rolipram from the available crystal structures
into the UCR2-capped PDE4B RS-14203 complex (PDB code
3
3G45, Figure 7) reveals that the solvent-orientated lactam of both
enantiomers is well disposed for hydrogen bonded engagement of
the Tyr286 phenolic hydroxyl. For (R)-rolipram a close contact
from the carbonyl oxygen is expected of the order of 2.5 Å
(Figure 7A). In the case of (S)-rolipram it is the lactam NH that
provides the engagement (Figure 7B), but here the contact is
Figure 8. The structure of ibudilast and its predicted PDE4 binding
mode are consistent with foci-inducing activity.
expected to be less tight (O N distance, ∼3.2 Å). This analysis
3 3 3
suggests that both the lactam and cyclopentyloxy groups of
rolipram contribute to promotion of the UCR2-capped state.
The prediction of a weaker hydrogen bond from (S)-rolipram to
Tyr286 may result in a less pronounced equilibrium shift to the
capped state upon binding by this enantiomer compared to (R)-
rolipram, however, and this may account for our earlier
observation19 that maximal levels of PDE4A4 accretion induced
by (S)-rolipram are only ∼20% of those induced by (R)-rolipram,
an effect analogous to partial agonism that is uniquely seen for (S)-
rolipram among the foci-inducing inhibitors that we have
identified here.
Inspection of the docked models in Figure 7 also provides a
clear rationalization for the impact of N-alkylation on the foci-
inducing capacity of the rolipram enantiomers. For (S)-rolipram
the predicted NꢀC bond axis of its N-alkyl derivatives will be
orientated toward Tyr286; N-alkylation will therefore not only
abolish the predicted hydrogen bond from the lactam but
sterically obstruct capping by the UCR2 gating sequence. In
contrast, the corresponding NꢀC bond axis for N-alkyl deriva-
tives of (R)-rolipram would be orientated toward a pronounced
notch in the rim of the catalytic pocket (Figure 7A) proximal to
the metal centers. N-Alkylation of (R)-rolipram may therefore be
tolerated without disruption of the predicted hydrogen bonded
interaction with Tyr286.
require the assistance of a hydrogen bond to Tyr286 in order
promote the UCR2-capped state. The targeting of an aromatic
inhibitor subunit to the Q2 site may therefore intrinsically
stabilize UCR2-capping more effectively, perhaps by establish-
ing efficient π-stacking interactions with Tyr286, than the
targeting of an aliphatic group to this location.
These considerations led us to test the impact of a Tyr286Phe
mutation on foci induction by rolipram and RS-25344, since our
results suggested that loss of a hydrogen bond to Tyr286 should
critically compromise the capacity to stabilize UCR2 capping in
PDE4A4 (and therefore induction of foci) by the former
inhibitor but not the latter. Pleasingly, this analysis was borne
out by experiment, and CHO cells expressing Tyr286Phe
PDE4A4-GFP exhibited foci when challenged with RS-25344
(94 ( 8% foci formation compared to WT PDE4A4 challenged
with RS-25344; mean ( SD, n = 3 separate experiments) but not
with rolipram (>97% loss of foci formation compared to WT
PDE4A4 challenged with rolipram; n = 3 separate experiments).
Our studies imply that effective stabilization of UCR2 capping in
PDE4A4 by inhibitors targeting the Q2 site with aliphatic groups
such as cyclopentyloxy is very finely balanced and dependent on
the establishment of a strong hydrogen bond between the
inhibitor auxiliary subunit and Tyr286.
Superimposition (not shown) of enzyme-bound structures
for cilomilast and piclamilast into the PDE4B RS-14203 co-
The UCR2 gating sequence is highly conserved across the
PDE4 enzyme family,38 but interestingly PDE4D isoforms are
uniquely distinguished by the presence of a phenylalanine at the
position corresponding to Tyr286 in PDE4A4. Our results
therefore suggest that among the PDE4A, PDE4B, and PDE4D
subfamilies, which exhibit widespread tissue distribution, roli-
pram and related inhibitors should preferentially stabilize UCR2
capping in PDE4A and PDE4B isoforms, whereas inhibitors
belonging to the class II foci-inducing group (Figure 2A) may be
less discriminating. It is probable, however, that the adoption of a
UCR2-capped PDE4 state will be sensitive to the regulatory
status of the enzyme, that is, to its specific phosphorylation and
protein association status.39 The impact of inhibitors in mod-
ulating UCR2 capping will therefore likely be dependent on
context, and extension of our analysis to isoforms beyond
PDE4A4 will require investigation in separate studies.
Ibudilast Is Correctly Predicted to Be a Foci-Forming
Inhibitor. Having developed an understanding of the inhibitor
features responsible for foci induction, we next sought to test this
understanding predictively. Ibudilast (Figure 8), a nonselective
PDE inhibitor used clinically as an anti-inflammatory agent,42
was not included in our initial compound screen (Figure 2),
which was focused primarily on PDE4-selective inhibitors.
3
crystal structure reveals that these compounds, although sharing
a Q2-targeted cyclopentyloxy group, lack a hydrogen bonding
feature sufficiently proximal to the side chain of Tyr286 to
establish a strong hydrogen bond.41 This suggests that inhibitor
occupancy of the Q2 site with bulky aliphatic groups such as
cyclopentyl may alone be insufficient to stabilize the UCR2-
capped state and drive formation of foci. In such cases, the
docking of Tyr286 appears to require assistance by interaction
with additional inhibitor features such as a hydrogen bonding
site in the auxiliary subunit, the lactam carbonyl in the case of
(R)-rolipram and, less optimally, the lactam NH for (S)-roli-
pram. In the case of cipamfylline (Figure 6C) insertion of the
cyclopropylmethyl group into the Q2 site would be augmented
by a predicted hydrogen bond between the Tyr286 phenol and
the inhibitor’s 2-O center (O O distance of ∼2.5 Å ex-
3 3 3
pected). In contrast the foci-inducing inhibitor, RS-14203
(Figure 2A), clearly lacks additional hydrogen bonding features
for engagement of Tyr286. Moreover, this compound’s pyr-
idylmethyl extension does not form extensive close contacts
directly with the UCR2 gating sequence (PDB code 3G45).
This suggests that inhibitors that target the Q2 site with an
aryl substituent (nitrophenyl in the case of RS-14203) may not
3340
dx.doi.org/10.1021/jm200070e |J. Med. Chem. 2011, 54, 3331–3347

Journal of Medicinal Chemistry
ARTICLE
We knew from our experience of developing inhibitors related to
ibudilast that pyrazolo[1,5-a]pyridines can bind to the PDE4
catalytic pocket with the N(1) center engaging the purine-scan-
ning-Gln.43 In the case of ibudilast this binding mode would
position the 2-isopropyl substituent in the enzyme’s Q2 subpocket
with a similar steric demand to the ethoxy and propyloxy groups of
the foci-inducing rolipram analogues 20R/S and 21R/S. Moreover,
in this pose ibudilast’s isopropyl ketone would be well-positioned
as a hydrogen bond acceptor site to engage Tyr286 in the UCR2
gating sequence (Figure 8), suggesting that the compound should
meet the criteria for induction of foci. We therefore tested ibudilast
and found that the compound does indeed drive accretion of
PDE4A4 into foci (Table 1; 96 ( 7% foci formation compared to
PDE4A4 challenged with rolipram; mean ( SD, n = 3 separate
experiments).
that such inhibitors may elicit functional effects not only through
PDE4 inhibition but also by disrupting PDE4A4 partnerships
and altering subcellular localization. Whether foci formation has
positive or confounding actions on therapeutic development has
yet to be appreciated and may well depend upon the particular
disease target. However, our SAR analysis now provides inves-
tigators with a firm structural basis and a library of compounds
from different structural classes to assess whether the foci
forming attribute of PDE4 inhibitors has positive, deleterious,
or null actions on their disease target. This library may addition-
ally provide a toolkit for investigating whether there are endo-
genous, physiological effectors able to elicit the redistribution of
PDE4A4 and thus to dynamically reprogram cAMP compart-
mentalization by regulating the subcellular targeting of this
isoform in the cell. Given the link provided by p62, a scaffold
that associates with PDE4A4 and is critical in both foci
formation20 and formation of autophagic vesicles,21 the com-
pounds described here will also be useful for probing the
connection between PDE4A4 foci and autophagy, which may
also have bearing on a variety of diseases.
’ CONCLUSION
In summary, evidence for cross-talk between the PDE4 core
catalytic and N-terminal domains is well established.1,2,5ꢀ8,11,38,39
Events occurring at regions N-terminal to the catalytic unit,
namely, phosphorylation and binding of partner proteins, not
only can affect the enzyme’s substrate binding and kinetic proper-
ties but also can alter sensitivity to inhibition.39,44ꢀ46 Conversely,
as in the case of foci formation with PDE4A4, the binding of
inhibitors in the catalytic pocket can clearly modulate events that
depend on the isoform-specific N-terminus.19 We show here that
the key inhibitor feature driving the intracellular redistribution
of PDE4A4 into accretion foci is a moderately bulky hydrophobic
or aromatic group targeted to the Q2 subpocket formed by
the junction of catalytic domain helices 14 and 15 and their
connecting loop. Reducing the size of this group abolishes foci
formation. Our mutagenesis experiments suggest that the Q2-
targeted substituent stabilizes the UCR2-capped state of the
enzyme through interaction with Tyr286 in the UCR2 gating
sequence, with the resulting structural transition relayed into the
N-terminal region. Interactions between Tyr286 and the inhibi-
tor’s auxiliary subunit, bound proximal to the enzyme’s catalytic
metal centers, also play an important role in promoting the
UCR2-capped state that leads to foci induction. The presence
of an appropriately positioned hydrogen bonding feature that
engages the phenolic function of Tyr286 is a characteristic of foci-
inducing inhibitors and may be essential for those that possess an
aliphatic Q2-targeted substituent.
’ EXPERIMENTAL SECTION
Biology. Analyses of inhibitor-driven PDE4A4 foci formation were
done as described in detail by us previously.19,20 PDE4 inhibition was
performed as described in detail elsewhere for full-length PDE4A4,45,48
using a two-step radiochemical assay.49 Site-directed mutagenesis of
plasmid DNA was carried out using the Stratagene (La Jolla, CA, U.S.)
QuikChange site-directed mutagenesis kit, according to the manufac-
turer’s instructions. The concentration of DNA samples was determined
by spectrophotometrically measuring the absorption at 260 nm (A260).
Purified plasmid DNA was produced using the QIAGEN (Crawley,
U.K.) QIAprep kits and stored either at 4 °C when eluted in 10 mM
Tris-Cl, pH 8.5, or at ꢀ20 °C when eluted in sterile water.
Chemistry. Commercially available reagents from Aldrich, Alfa
Aesar, and Acros chemical companies were generally used as supplied
without further purification. (R)-(ꢀ)-Rolipram and (S)-(þ)-rolipram
were purchased from Ascent Scientific Ltd. Tetrahydrofuran, diethyl
ether, and toluene were dried by distillation from sodium benzophenone
ketyl under argon. “Light petroleum” refers to the fraction boiling
between 40 and 60 °C. Anhydrous N,N-dimethylformamide (DMF)
was purchased from Aldrich and used as supplied from Sure/Seal bottles.
Reactions were routinely carried out under an inert atmosphere of argon
or nitrogen. Analytical thin layer chromatography was carried out using
aluminum backed plates coated with Merck Kieselgel 60 GF254 (No.
05554). Developed plates were visualized under ultraviolet light
(254 nm) and/or alkaline potassium permanganate dip. Flash chroma-
tography was performed using DAVISIL silica (60 Å, 35ꢀ70 μm) from
Fisher (catalog no. S/0693/60). Fully characterized compounds were
chromatographically homogeneous. Melting points were determined
using a Stuart Scientific SMP10 apparatus and are uncorrected. IR
spectra were recorded on a Perkin-Elmer 1600 FT-IR spectrometer
(as KBr discs, as solutions in CHCl3, or as films between NaCl plates) or
on a Perkin-Elmer Spectrum 100 FT-IR spectrometer (as neat solids or
liquids). Mass spectra were obtained on Kratos Concept IS EI (electron
impact) and Fisons VG Quattro (electrospray) spectrometers. 1H NMR
spectra were recorded at 200 and 400 MHz on Bruker AC200
and DPX400 spectrometers; 13C NMR spectra were recorded at 50
and 101 MHz on the same instruments. Chemical shifts are recorded in
parts per million (δ in ppm) and are referenced against solvent signals
(δC 77.16 for chloroform and δC 39.52 for methyl sulfoxide) for
13C spectra and solvent residual resonances (δH 7.26 for chloroform
and δH 2.50 methyl sulfoxide) for 1H spectra.50 Chemical shift values are
There are likely to be over 25 different PDE4 isoforms, each
expressed in a cell-type specific fashion and providing specific
functional roles in the cells where they are expressed.8 PDE4A4
is a critical isoform in regulating fibrin deposition through its
association with p75NTR 13
This has implications for scarring
.
seen in various fibrotic diseases, such as COPD,14 but also for
neuronal growth repair following spinal cord injury,16 where in
each instance the PDE4 inhibitor rolipram has efficacy. The value
of PDE4A4 as a potential therapeutic target in COPD is further
strengthened by the observation that this isoform is specifically
up-regulated in macrophages of COPD patients.15 Up-regulation
of the enzyme in the hippocampus has also recently been shown
to underpin cognitive deficits associated with sleep deprivation,18
a condition prevalent in COPD. Our finding that chronically
challenging cells with certain PDE4-selective inhibitors redis-
tributes PDE4A4 into accretion foci through association with
p62 but with concomitant loss of partnerships with certain other
proteins (such as p75NTR, DISC1, and β-arrestin)20,47 indicates
3341
dx.doi.org/10.1021/jm200070e |J. Med. Chem. 2011, 54, 3331–3347

Journal of Medicinal Chemistry
ARTICLE
accurate to (0.01 ppm and (0.1 ppm, respectively. J values are given in
Hz. Multiplicity designations used are as follows: s, d, t, q, sept, and m for
singlet, doublet, triplet, quartet, septet, and multiplet, respectively. In
13C NMR spectra, signals corresponding to CH, CH2, or CH3 groups
are assigned from DEPT. NMR signal assignments, where reported,
over Celite, washing with EtOAc. The filtrate was concentrated in vacuo and
the crude product subjected to flash column chromatography (EtOAc)
followed by crystallization (CH2Cl2ꢀEt2O) to afford 6 (58.6 mg, 66%)
as a colorless solid: NMR data, as presented for 6S; MS (EI) m/z 301
(Mþ, 49%. HRMS (EI) calcd for Mþ C18H23NO3 301.1678, found
301.1673. The structure was confirmed by single crystal X-ray diffraction
(Supporting Information).
1
were deduced from 1Hꢀ H COSY, NOESY, HSQC, and HMBC
experiments, as appropriate. All compounds tested were >95% pure
by elemental and/or HPLC analysis. Elemental analyses were carried out
by the analytical service of the Department of Chemistry at Heriot-Watt
University using an Exeter CE-440 elemental analyzer.
2-(1-Cyano-4-cyclopentyloxy-5-methoxy-2,3-dihydro-
1H-inden-1-yl)ethanoic Acid (7). Ester 5 (2.99 g, 9.07 mmol) was
hydrolyzed with LiOH H2O (577 mg, 13.8 mmol) in a vigorously
3
stirred mixture of THF (35 mL) and water (17 mL) at 0 °C for 1 h and
then at ambient temperature for 3 h. The mixture was concentrated in
vacuo to afford a white slurry that was taken up in water (500 mL).
NaOH solution (2 M, 10 mL) was added and the basic solution washed
with Et2O (100 mL). The organic extract was discarded; the aqueous
phase was then acidified to pH 1 with concentrated hydrochloric acid
and extracted with CHCl3 (3 ꢁ 100 mL). The combined extract was
dried (Na2SO4) and evaporated to afford acid 7 (2.90 g, quant) as a
white amorphous solid: δH (200 MHz, CDCl3) 1.52ꢀ1.91 (8H, m),
2.37 (1H, ddd, J 13.3, 7.4, and 6.0), 2.69ꢀ3.16 (3H, m), 2.76 (1H, d,
J 17.6), 3.06 (1H, d, J 17.6), 3.84 (3H, s), 4.82ꢀ4.90 (1H, m), 6.83 (1H,
d, J 8.4), 7.07 (1H, d, J 8.3).
4-Cyclopentyloxy-5-methoxy-1-indanone (3). A stirred mix-
ture of 230 (12.6 g, 70.7 mmol; prepared as described in the Supporting
Information), finely powdered K2CO3 (17.5 g, 126 mmol), and cyclo-
pentyl bromide (10.0 mL, 99.0 mmol) in anhydrous DMF (100 mL) was
heated at 100 °C. Further cyclopentyl bromide (4.00 mL, 40.0 mmol)
was added after 13 h and a final increment (2.00 mL, 19.8 mmol)
together with K2CO3 (8.23 g, 59.5 mmol) after 20 h. Following a total
reaction time of 48 h, the mixture was cooled, decanted into brine (1 L),
and extracted with Et2O (4 ꢁ 100 mL). The combined extracts were
dried (Na2SO4) and concentrated in vacuo to a dark brown oil. The title
compound (12.4 g, 71%) was isolated as a pale brown solid by crystal-
lization from light petroleumꢀEtOAc: δH (200 MHz, CDCl3)
1.52ꢀ1.95 (8H, m), 2.62ꢀ2.68 (2H, m), 3.03ꢀ3.09 (2H, m), 3.92
(3H, s), 4.89ꢀ4.98 (1H, m), 6.95 (1H, d, J 8.4), 7.50 (1H, d, J 8.4); MS
(EI) m/z 246 (Mþ, 9%), 178 (100). Anal. (C15H18O3) C, H.
4-Cyclopentyloxy-5-methoxy-2,3-dihydro-1H-indene-1-
carbonitrile (4). A solutionof3 (2.11g, 8.59 mmol) in anhydrousDME
(20 mL) was added dropwise to a stirred, ice-cooled mixture of t-BuOK
(3.29 g, 29.3 mmol), TOSMIC (3.10 g, 15.8 mmol), and MeOH
(0.988 mL, 24.4 mmol) in anhydrous DME (60 mL) under argon. The
mixture was stirredfor 30 min at0 °C and thenat ambient temperature for
24 h. The mixture was then partitioned between brine and EtOAc and the
organic layer dried (Na2SO4) and concentrated in vacuo. The resulting
crude product was subjected to flash column chromatography (9:1 light
petroleumꢀEtOAc) to afford 4 (1.11 g, 50%) as a yellow solid: δH
(400 MHz; CDCl3) 1.54ꢀ1.64 (2H, m), 1.64ꢀ1.76 (2H, m), 1.78ꢀ1.88
(4H, m), 2.34 (1H, dq, J 12.8 and 8.1), 2.52 (1H, dtd, J 12.6, 8.0 and 4.4),
2.88(1H, dt, J16.1 and 8.0), 3.08 (1H, ddd, J16.2, 8.6 and 4.4), 3.83 (3H, s),
4.05 (1H, t, J 8.0), 4.83ꢀ4.87 (1H, m), 6.81 (1H, d, J 8.2), 7.06 (1H, d,
J 8.3); MS (EI) m/z 257 (Mþ, 51%). Anal. (C16H19NO2) C, H, N.
Methyl 2-(1-Cyano-4-cyclopentyloxy-5-methoxy-2,3-di-
hydro-1H-inden-1-yl)ethanoate (5). NaN(SiMe3)2 (1.0 M THF
solution, 22.0 mL, 22.0 mmol) was added dropwise to a solution of
4 (4.50 g, 17.5 mmol) in anhydrous THF (200 mL) at ꢀ78 °C. The
mixture was maintained at ꢀ78 °C for 5 h, brought to 10 °C for 1 h, and
then returned to ꢀ78 °C before adding methyl bromoacetate (3.60 mL,
38.0 mmol). The mixture was then allowed to attain ambient tempera-
ture and, after a further 16 h, partitioned between EtOAc (500 mL) and
brine (200 mL). The organic phase was dried (Na2SO4) and concen-
trated in vacuo to yield the crude product, which was subjected to flash
column chromatography (1:4 to 7:13 Et2Oꢀhexane gradient elution)
and crystallization (Et2OꢀhexaneꢀCH2Cl2) to afford 5 (3.92 g, 68%)
as a colorless solid: δH (400 MHz, CDCl3) 1.53ꢀ1.64 (2H, m),
1.64ꢀ1.76 (2H, m), 1.77ꢀ1.88 (4H, m), 2.37 (1H, ddd, J 13.3, 7.8
and 5.6), 2.69 (1H, d, J 16.0), 2.72 (1H, ddd, J 13.3, 8.4, and 6.5), 2.98
(1H, d, J 16.0), 2.91ꢀ3.09 (2H, m) 3.74 (3H, s), 3.83 (3H, s), 4.81ꢀ4.90
(1H, m), 6.81 (1H, d, J 8.7), 7.04 (1H, d, J 8.3); MS (EI) m/z 329 (Mþ,
42%). HRMS (EI) calcd for Mþ C19H23NO4 329.1627, found 329.1615.
Anal. (C19H23NO4) C, H, N.
2-[(S)-1-Cyano-4-cyclopentyloxy-5-methoxy-2,3-dihydro-
1H-inden-1-yl]-N-[(S)-1-phenylethyl]ethanamide (8a) and
2-[(R)-1-Cyano-4-cyclopentyloxy-5-methoxy-2,3-dihydro-
1H-inden-1-yl]-N-[(S)-1-phenylethyl]ethanamide (8b). To a
stirred solution of 7 (2.86 g, 9.07 mmol) and 1-hydroxybenzotrazole
(1.53 g, 9.96 mmol) in anhydrous DMF (30 mL) at 0 °C was added
1-ethyl-3-(30-dimethylaminopropyl)carbodiimide hydrochloride (2.11 g,
11.0 mmol) followed by N,N-diisopropylethylamine 1.60 mL, 10.3
mmol). After 30 min (S)-1-phenylethylamine (1.65 mL, 12.0 mmol)
was added. The mixture was stirred at 0 °C for 3 h and then at ambient
temperature for 24 h. Et2O (300 mL) and EtOAc (75 mL) were added,
and the mixture was washed successively with brine (100 mL), hydro-
chloric acid (1 M, 3 ꢁ 100 mL), saturated NaHCO3 solution (2 ꢁ
100 mL), and brine (100 mL). The organic phase was dried (Na2SO4)
and evaporated. The resulting residue was subjected to flash column
chromatography to afford the product (3.73 g, 98%), a 1:1 mixture of 8a
and 8b, as a colorless powder. Separated samples of 8a (500 mg, 99.5%
purity) and 8b (402 mg, 98% purity) were obtained by HPLC. Isomer 8a:
HPLC tR = 9.21 min (160 mm ꢁ 4.5 mm Thermo Scientific Hypersil
5 μm silica column; 2:12:86 MeOHꢀCH2Cl2ꢀhexane mobile phase at
1 mL/min); δH (400 MHz, CDCl3) 1.50 (3H, d, J 6.9), 1.54ꢀ1.64
(2H, m), 1.64ꢀ1.75 (2H, m), 1.77ꢀ1.88 (4H, m), 2.53 (1H, d, J 14.1),
2.54 (1H, ddd, J 13.3, 7.9, and 5.7), 2.63 (1H, ddd, J 13.3, 8.3, and 6.6),
2.75 (1H, d, J 14.1), 2.93 (1H, app dt, J 16.4 and 7.3), 3.02 (1H, ddd,
J 16.4, 8.3, and 5.8), 3.82 (3H, s), 4.82ꢀ4.86 (1H, m), 5.12 (1H, quintet,
J 7.2), 5.90 (1H, br d, J 7.8), 6.77 (1H, d, J 8.3), 6.96 (1H, d, J 8.3),
7.24ꢀ7.35 (5H, m). Isomer 8b: HPLC tR = 9.74 min (160 mm ꢁ 4.5 mm
Thermo Scientific Hypersil 5 μm silica column; 2:12:86 MeOHꢀ
CH2Cl2ꢀhexane mobile phase at 1 mL/min); δH (400 MHz, CDCl3)
1.43 (3H, d, J 6.9), 1.54ꢀ1.64 (2H, m), 1.64ꢀ1.75 (2H, m), 1.77ꢀ1.88
(4H, m), 2.47 (1H, ddd, J 13.3, 7.8, and 5.6), 2.57 (1H, d, J 14.2), 2.59
(1H, ddd, J 13.3, 8.4, and 6.7), 2.75 (1H, d, J 14.2), 2.94 (1H, app dt, J 16.4
and 7.9), 3.02 (1H, ddd, J 16.4, 8.6, and 5.8), 3.82 (3H, s), 4.83ꢀ4.89
(1H, m), 5.09 (1H, quintet, J 7.2), 5.83 (1H, br d, J 7.7), 6.76 (1H, d,
J 8.3), 6.99 (1H, d, J 8.3), 7.23ꢀ7.35 (5H, m). Anal. (1:1 8a/8b mixture,
C26H30N2O3) C, H, N.
(S)-4-Cyclopentyloxy-5-methoxy-2,3-dihydrospiro[indene-
1,30-pyrrolidin]-50-one (6S). A solution of 8a (396 mg, 0.946 mmol)
in absolute EtOH (52 mL) was hydrogenated at 90 °C and 90 bar of H2
pressure in three passes over a Raney Ni cartridge using a ThalesNano
H-cube hydrogenation reactor (flow rate 0.7 mL/min). The mixture was
then evaporated; the residue was reconstituted in p-xylene (40 mL) and
(rac)-4-Cyclopentyloxy-5-methoxy-2,3-dihydrospiro[indene-
1,30-pyrrolidin]-50-one (6). A solution of 5 (97.0 mg, 0.295 mmol) in a
mixture of absolute EtOH (5 mL) and EtOAc (5 mL) containing AcOH
(2 drops) was hydrogenated over 10% PdꢀC (10.0 mg) at 60 °C and 16 bar
of H2 pressure for 43 h. The mixture was then purged with N2 and filtered
3342
dx.doi.org/10.1021/jm200070e |J. Med. Chem. 2011, 54, 3331–3347

Journal of Medicinal Chemistry
ARTICLE
heated in an oil bath thermostated at 155 °C. After 18 h the solvent was
removed in vacuo and the residue subjected to flash column chromatog-
raphy (EtOAc) followed by crystallization (EtOAcꢀhexane) to afford 6S
(215 mg, 75%) as a white powder: δH (400 MHz, CDCl3) 1.53ꢀ1.63
(2H, m), 1.63ꢀ1.77 (2H, m), 1.79ꢀ1.88 (4H, m), 2.13 (1H, dt, J 12.7 and
7.5), 2.23 (1H, ddd, J 12.8, 7.0, and 6.2), 2.44 (1H, d, J 16.8), 2.61 (1H, d,
J 16.8), 2.84ꢀ2.96 (2H, m), 3.42 (1H, d, J 9.5), 3.47 (1H, d, J 9.5), 3.81
(3H, s), 4.81ꢀ4.87 (1H, m), 6.55 (1H, br s), 6.77 (1H, d, J 8.2), 6.89 (1H,
solid: δH (400 MHz, CDCl3) 2.66 (1H, dd, J 13.9 and 5.3), 2.71 (1H, dd,
J 13.9 and 5.0), 3.60 (3H, s), 3.83ꢀ3.90 (1H, m), 3.85 (3H, s), 4.52 (1H, dd,
J 12.5 and 7.8), 4.64 (1H, dd, J 12.5 and 7.1), 5.12 (2H, s), 6.74 (1H, d, J 2.1),
6.77 (1H, dd, J 8.3 and 2.1), 6.84 (1H, d, J 8.2), 7.29ꢀ7.33 (1H, m),
7.35ꢀ7.39 (2H, m), 7.41ꢀ7.44 (2H, m); m/z (EI) 359 (Mþ, 11%),
91 (100). Anal. (C19H21NO6) C, H, N.
3-(3-Benzyloxy-4-methoxyphenyl)-4-nitrobutanoic Acid (15).
To a stirred solution of 14 (6.50 g, 18.1 mmol) in MeOH (120 mL) at 0 °C
was added dropwise NaOH solution (2 M, 60 mL). The mixture was allowed
to attain ambient temperature over 16 h and then concentrated in vacuo. The
residual oil was partitioned between hydrochloric acid (2 M, 60 mL) and
EtOAc (150 mL), separating and further extracting the aqueous layer with
EtOAc (2 ꢁ 40 mL). The combined organic phase was washed with brine
(50 mL), dried (MgSO4), and evaporated to afford 15 (5.89 g, 94%) as a
beige powder: δH (400 MHz, CDCl3) 2.69 (1H, dd, J 15.2 and 6.2), 2.75
(1H, dd, J 15.2 and 5.7), 3.80ꢀ3.88 (1H, m), 3.85 (3H, s), 4.50 (1H, dd,
J 12.5 and 7.7), 4.61 (1H, dd, J 12.5 and 7.1), 5.11 (2H, s), 6.73 (1H, d, J 2.1),
6.77 (1H, dd, J 8.3 and 2.1), 6.84 (1H, d, J 8.3), 7.28ꢀ7.33 (1H, m),
7.34ꢀ7.39 (2H, m), 7.40ꢀ7.43 (2H, m), 9.93 (1H, br s); m/z (EI) 345
(Mþ, 74%), 91 (100). Anal. (C18H19NO6) C, H, N.
3-(3-(Benzyloxy)-4-methoxyphenyl)-4-nitrobutanoyl Chlor-
ide (16). A solution of 15 (5.85 g, 16.9 mmol) in anhydrous CH2Cl2
(55 mL) was refluxed with SOCl2 (2.04 mL, 28.0 mmol). After 16 h the
volatiles were removed in vacuo to afford 16 (6.13 g, 99%) as an amorphous
yellow solid: δH (400 MHz, CDCl3) 3.24 (1H, dd, J 17.5 and 8.0), 3.31
(1H, dd, J 17.5 and 6.4), 3.84ꢀ3.93 (1H, m), 3.86 (3H, s), 4.51 (1H, dd,
J 12.8 and 7.3), 4.56 (1H, dd, J 12.8 and 7.6), 5.13 (2H, s), 6.71 (1H, d,
J 2.2), 6.77 (1H, dd, J 8.3 and 2.2), 6.86 (1H, d, J 8.3), 7.30ꢀ7.34 (1H, m),
7.36ꢀ7.40 (2H, m), 7.41ꢀ7.44 (2H, m); MS (EI) m/z 363 (35Cl Mþ,
72%) 91 (100). Anal. (C18H18NO5Cl) C, H, N.
(R)-3-(3-Benzyloxy-4-methoxyphenyl)-4-nitro-N-[(S)-1-
phenylethyl]butanamide (17a) and (S)-3-[(3-Benzyloxy-4-
methoxyphenyl)-4-nitro-N-[(S)-1-phenylethyl]butanamide
(17b). To a solution of 3-(3-benzyloxy-4-methoxyphenyl)-4-nitrobu-
tanoyl chloride (6.11 g, 16.8 mmol) in CH2Cl2 (65 mL) under reflux was
added dropwise a solution of (S)-(ꢀ)-1-phenylethylamine (4.2 mL, 33.4
mmol) in CH2Cl2 (6 mL). After 2 h the mixture was cooled and washed
successively with 2 M hydrochloric acid (50 mL), saturated NaHCO3
solution (50 mL), and brine (50 mL). The organic phase was dried
(MgSO4) and evaporated to afford a 1:1 mixture of amides 17a and 17b
(7.19 g, 96%) as a buff powder. The amides were partially separated by
flash column chromatography (1:4:5 CH2Cl2ꢀEtOAcꢀhexane) to
afford 17a (1.72 g) and 17b (2.84 g) as colorless powders. Amide
17a: TLC Rf = 0.37 (1:4:5 CH2Cl2ꢀEtOAcꢀhexane); δH (400 MHz,
CDCl3) 1.30 (3H, d, J 6.9), 2.41 (1H, dd, J 14.5 and 7.5), 2.52 (1H, dd,
J 14.5 and 6.9), 3.81ꢀ3.88 (1H, m), 3.86 (3H, s), 4.56 (1H, dd, J 12.5
and 8.0), 4.70 (1H, dd, J 12.5 and 6.4), 4.99 (1H, quintet, J 7.2), 5.08
(1H, d, J 12.2), 5.11 (1H, d, J 12.2), 5.39 (1H, br d, J 7.9), 6.71 (1H, d,
J 2.1), 6.74 (1H, dd, J 8.2 and 2.1), 6.80 (1H, d, J 8.1), 7.13ꢀ7.16 (2H,
m), 7.22ꢀ7.35 (6H, m), 7.39ꢀ7.42 (2H, m); MS (EI) m/z 448 (Mþ,
8%), 105 (100). HRMS (EI) calcd for Mþ C26H28N2O5 448.1998,
found 448.1993. Anal. (C26H28N2O5) C, H, N. Amide 17b: TLC Rf =
0.24 (1:4:5 CH2Cl2ꢀEtOAcꢀhexane); δH (400 MHz, CDCl3) 1.38
(3H, d, J 6.9), 2.34 (1H, dd, J 14.2 and 8.3), 2.55 (1H, dd, J 14.2 and 6.4),
3.83ꢀ3.90 (1H, m), 3.87 (3H, s), 4.58 (1H, dd, J 12.4 and 8.1), 4.67
(1H, dd, J 12.4 and 6.6), 4.99 (1H, quintet, J 7.3), 5.06 (1H, d, J 12.3),
5.09 (1H, d, J 12.3), 5.33 (1H, br d, J 7.9), 6.69 (1H, d, J 2.1), 6.73 (1H,
dd, J 8.3 and 2.1), 6.79 (1H, d, J 8.3), 6.93ꢀ6.96 (2H, m), 7.18ꢀ7.34
(6H, m), 7.36ꢀ7.39 (2H, m); MS (EI) m/z 448 (Mþ, 13%), 105 (100).
HRMS (EI) calcd for Mþ C26H28N2O5 448.1998, found 448.1997. Anal.
(C26H28N2O5) C, H, N.
22
d, J 8.2); [R]D þ11.1 (c 1.08, CHCl3). Anal. (C18H23NO3) C, H, N.
(R)-4-Cyclopentyloxy-5-methoxy-2,3-dihydrospiro[indene-
1,30-pyrrolidin]-50-one (6R). Following the procedure for 6S, enan-
tiomer 6R (106 mg) was prepared in 38% yield from 8b (383 mg):
22
[R]D ꢀ12.1 (c 1.09, CHCl3). Anal. (C18H23NO3) C, H, N.
(R)-4-(3-Cyclopentyloxy-4-methoxyphenyl)-1-methylpyr-
rolidin-2-one (10R). To a stirred solution of (R)-(ꢀ)-rolipram (30.0
mg, 0.109 mmol) in anhydrous DMF (3 mL) at ambient temperature
was added NaH (60% w/w dispersion in mineral oil; 6.0 mg, 0.15 mmol)
followed after 10 min by MeI (0.014 mL, 0.22 mmol). After 16 h
the mixture was diluted with EtOAc (50 mL), washed with brine (3 ꢁ
10 mL), dried (Na2SO4), and evaporated to afford a residue from which
10R (27 mg, 86%) was isolated as a colorless oil by flash column
chromatography (1:99 to 1:49 MeOHꢀCH2Cl2 gradient): δH (400 MHz,
CDCl3) 1.56ꢀ1.66 (2H, m), 1.78ꢀ1.97 (6H, m), 2.51 (1H, dd, J 16.8and
8.4), 2.78 (1H, dd, J 16.8 and 9.1), 2.90 (3H, s), 3.36 (1H, dd, J 9.6 and
7.1), 3.48 (1H, app quintet, J 8.2), 3.71 (1H, dd, J 9.6 and 8.3), 3.83 (3H, s),
4.74ꢀ4.78 (1H, m), 6.73 (1H, d, J 2.2), 6.74 (1H, dd, J 8.0 and 2.2), 6.82
(1H, d, J 8.0); m/z (EI) 289 (Mþ, 45%), 221 (89), 150 (100). HRMS
(EI) calcd for Mþ C17H23NO3 289.1678, found 289.1665. Anal.
(C17H23NO3) C, H, N.
(S)-4-(3-Cyclopentyloxy-4-methoxyphenyl)-1-methylpyr-
rolidin-2-one (10S). Following the above procedure for 10R, enantio-
mer 10S (20 mg, 63%) was prepared from (S)-(þ)-rolipram (30 mg,
0.11 mmol). Anal. (C17H23NO3) C, H, N.
(R)-1-Benzyl-4-(3-cyclopentyloxy-4-methoxyphenyl)pyrr-
olidin-2-one (11R). To a stirred solution of (R)-(ꢀ)-rolipram (23.2
mg, 0.0845 mmol) in anhydrous DMF (5 mL) at ambient temperature
was added NaH (60% w/w dispersion in mineral oil; 3.7 mg, 0.093
mmol) followed after 45 min by BnBr (0.011 mL, 0.093 mmol). After
20 h the mixture was diluted with Et2O (75 mL), washed with brine (3 ꢁ
15 mL), dried (Na2SO4), and evaporated to afford a residue from which
11R (23.4 mg, 76%) was isolated as a colorless oil by flash column
chromatography (1:9 Et2OꢀCH2Cl2): δH (400 MHz, CDCl3) 1.55ꢀ
1.65 (2H, m), 1.76ꢀ1.92 (6H, m), 2.58 (1H, dd, J 16.9 and 8.2), 2.86
(1H, dd, J 16.8 and 9.2), 3.24 (1H, dd, J 9.6 and 7.0), 3.47 (1H, app
quintet, J 8.4), 3.61 (1H, dd, J 9.6 and 8.3), 3.81 (3H, s), 4.46 (1H, d,
J 14.6), 4.55 (1H, d, J 14.6), 4.67ꢀ4.71 (1H, m), 6.66 (1H, d, J 2.1), 6.68
(1H, dd, J 8.2 and 2.1), 6.77 (1H, d, J 8.2), 7.25ꢀ7.25 (5H, m). Anal.
(C23H27NO3) C, H, N.
(S)-1-Benzyl-4-(3-cyclopentyloxy-4-methoxyphenyl)pyrr-
olidin-2-one (11S). Following the above procedure for 11R, en-
antiomer 11S (25 mg, 94%) was prepared in an identical manner from
(S)-rolipram (20 mg, 0.073 mmol). Anal. (C23H27NO3) C, H, N.
Methyl 3-(3-Benzyloxy-4-methoxyphenyl)-4-nitrobutanoate
(14). A stirred suspension of (methoxycarbonylmethylene)triphenylphos-
phorane (152 g, 456 mmol) and 3-benzyloxy-4-methoxybenzaldehyde
(102 g, 419 mmol) was refluxed in MeCN (600 mL). After 7 h the dark
brown homogeneous solution was cooled and evaporated to dryness.
The solid residue comprising alkene 13 and Ph3PO was resuspended in
MeNO2 (500 mL). 1,1,3,3-Tetramethylguanidine (12 mL) was added and
the stirred mixture heated at 75 °C for 24 h. The mixture was then cooled
and concentrated in vacuo. Ph3PO was partially removed by crystallization
of the residual oil from Et2O (600 mL) and filtration. The filtrate was
evaporated and subjected to flash column chromatography (1:9 to 1:4
EtOAcꢀlight petroleum gradient to afford 14 (105 g, 70%) as a colorless
(S)-4-(3-Hydroxy-4-methoxyphenyl)pyrrolidin-2-one (18S).
Amide 17b (2.65 g, 5.91 mmol) was hydrogenated over 10% Pd/C as a
solution in EtOAc and EtOH (1:1 v/v, 250 mL) using a ThalesNano
3343
dx.doi.org/10.1021/jm200070e |J. Med. Chem. 2011, 54, 3331–3347

Journal of Medicinal Chemistry
ARTICLE
H-Cube(90 °C, 90 bar H2, flow rate 0.8 mL/min). The crude aminoamide
(1.76 g) obtained following evaporation of the eluate was reconstituted
with p-xylene (150 mL) and heated in a sealed pressure vessel at 160 °C for
36 h to facilitate closure of the lactam ring. The solvent was then removed
in vacuo and the resulting residue subjected to flash column chromatog-
raphy (5:95 MeOHꢀEtOAc) to afford 18S (990 mg, 81%) as a light
brown powder: δH (400 MHz, CD3ODꢀCDCl3) 2.30 (1H, dd, J 17.0 and
8.7), 2.53 (1H, dd, J 17.0 and 8.9), 3.22 (1H, dd, J 9.7 and 7.3), 3.45 (1H,
app. quintet, J 8.1), 3.58 (1H, dd, J 9.7 and 8.2), 3.71 (3H, s), 6.54 (1H,
ddd, J8.2, 2.2, and 0.5), 6.62 (1H, d, J2.2), 6.67 (1H, d, J8.3); MS m/z(EI)
207 (Mþ, 90%), 150 (100), 135 (64). Anal. (C11H13NO3) C, H, N.
(R)-4-(3-Hydroxy-4-methoxyphenyl)pyrrolidin-2-one (18R).
18R was prepared as described for 18S in 43% yield from 17a (1.92 g).
Anal. (C11H13NO3) C, H, N.
(R)-4-(3,4-Dimethoxyphenyl)pyrrolidin-2-one (19R). To a
stirred mixture of 18R (115 mg, 0.555 mmol) and finely powdered
K2CO3 (460 mg, 3.33 mmol) in anhydrous DMF (58 mL) was added
iodomethane (70 μL, 1.1 mmol). After 42 h the mixture was filtered over
Celite and the filtrate concentrated in vacuo. The resulting residue was
subjected to flash column chromatography (2:98 MeOHꢀEtOAc) to
afford 19R (107 mg, 87%) as a pale yellow powder: δH (400 MHz,
CDCl3) 2.48 (1H, dd, J 16.9 and 8.8), 2.72 (1H, dd, J 16.9 and 8.9), 3.40
(1H, dd, J 9.2 and 7.3), 3.65 (1H, app quintet, J 8.2), 3.74ꢀ3.79 (1H, m),
3.87 (3H, s), 3.89 (3H, s), 5.86 (1H, br s), 6.75 (1H, d, J 2.0), 6.80 (1H,
dd, J 8.2 and 2.0), 6.84 (1H, J 8.2); m/z (EI) 221 (Mþ, 64%), 164 (100).
Anal. (C12H15NO3) C, H, N.
8-Ethynyl-1,7-naphthyridin-6-amine (24). A solution of 23
(593 mg, 2.46 mmol) in THF (25 mL) was treated with tetrabutylam-
monium fluoride (1 M in 5% H2OꢀTHF solution; 2.71 mL,
2.71 mmol). After 10 min the mixture was concentrated in vacuo. The
resulting residue was taken up in EtOAc (150 mL) and washed with
water (3 ꢁ 25 mL). The organic phase was dried (Na2SO4) and
concentrated in vacuo to afford a residue from which 24 (356 mg,
86%) was isolated as a yellow solid by flash column chromatography
(1:19 MeOHꢀCH2Cl2): δH (200 MHz, CDCl3) 3.62 (1H, s), 4.69
(2H, br s), 6.67 (1H, s), 7.36 (1H, dd, J 4.0 and 8.5), 7.83 (1H, dd, J 1.6
and 8.5), 8.75 (1H, dd, J 1.6 and 4.0).
8-Ethyl-1,7-naphthyridin-6-amine (25). A solution of 24 (350
mg, 2.07 mmol) in EtOH (30 mL) was hydrogenated at 1 atm over 10%
PdꢀC (121 mg) at ambient temperature for 10 h. The reaction mixture
was then purged with N2 and filtered over Celite, washing with CHCl3.
The filtrate was evaporated to afford a residue from which 25 (263 mg,
73%) was isolated by flash column chromatography (2.5:97.5
MeOHꢀCH2Cl2): δH (200 MHz, CDCl3) 1.36 (3H, t, J 7.5), 3.36
(2H, q, J 7.5), 4.57 (2H, s), 6.46 (1H, s), 7.29 (1H, dd, J 4.1 and 8.4), 7.75
(1H, dd, J 1.7 and 8.5), 8.61 (1H, dd, J 1.7 and 4.1); m/z (EI) 173 (Mþ,
38%), 172 (Mþ ꢀ H, 100). HRMS (ESIþ) calcd for MHþ C10H12N3
174.1026, found 174.1027.
8-Ethyl-1,7-naphthyridin-6-yl Trifluoromethanesulfonate
(26). To an ice-cooled solution of 25 (256 mg, 1.48 mmol) in DMF
(4 mL) was added trifluoromethanesulfonic acid (1.0 mL, 11.3 mmol)
followed after 5 min by NaNO2 (204 mg, 2.96 mmol) in small portions.
After the addition the mixture was stirred at 0 °C for 2 h and then at
ambient temperature for a further 2 h. The mixture was then diluted with
EtOAc (80 mL) and washed successively with water (20 mL), 2 M
NaOH solution (10 mL), and water (2 ꢁ 10 mL). The organic phase
was dried (Na2SO4) and concentrated in vacuo to afford 26 (239 mg,
53%) as an orange oil: δH (200 MHz, CDCl3) 1.38 (3H, t, J 7.5), 3.49
(2H, q, J 7.5), 7.36 (1H, s), 7.60 (1H, dd, J 4.1 and 8.5), 8.14 (1H, dd,
J 1.7 and 8.5), 8.99 (1H, dd, J 1.7 and 4.1).
Methyl 4-(8-Ethyl-1,7-naphthyridin-6-yl)benzoate (27). A
stirred solution of 26 (239 mg, 0.780 mmol), 4-(methoxycarbonyl)
phenylboronic acid (183 mg, 1.02 mmol), Ph3P (16 mg, 0.061
mmol), and bis(dibenzylideneacetone)palladium(0) (18 mg, 0.031
mmol) in THF (6 mL) was heated with 2 M Na2CO3 solution
(1.5 mL) in a bath thermostated at 80 °C. After 17 h the brown
mixture was cooled, diluted with EtOAc (50 mL), filtered through
Celite, and washed with brine (3 ꢁ 10 mL). The organic phase was
dried (Na2SO4) and concentrated in vacuo to give a residue from
which 27 (175 mg, 77%) was isolated by flash column chromatog-
raphy (1:4 EtOAcꢀlight petroleum) as a colorless solid: δH (200
MHz, CDCl3) 1.50 (3H, t, J 7.5), 3.58 (2H, q, J 7.5), 3.92 (3H, s), 7.54
(1H, dd, J 4.2 and 8.3), 7.91 (1H, s), 8.11ꢀ8.18 (3H, m), 8.25 (2H,
m), 8.95 (1H, dd, J 1.7 and 4.2).
(S)-4-(3,4-Dimethoxyphenyl)pyrrolidin-2-one (19S). 19S was
prepared as described for 19R in 97% yield from 18S (120 mg). Anal.
(C12H15NO3) C, H, N.
(R)-4-(3-Ethoxy-4-methoxyphenyl)pyrrolidin-2-one (20R).
20R was prepared as described for 19R in 78% yield from 18R (157 mg)
and iodoethane: δH (400 MHz, CDCl3) 1.45 (3H, t, J 7.0), 2.47 (1H, dd,
J 16.9 and 8.9), 2.70 (1H, dd, J 16.9 and 8.9), 3.38 (1H, dd, J 9.4 and 7.4),
3.62 (1H, app quintet, J 8.2), 3.72ꢀ3.77 (1H, m), 3.85 (3H, s), 4.09 (2H,
q, J 7.0), 6.54 (1H, br s), 6.75 (1H, d, J 2.0), 6.78 (1H, dd, J 8.2 and 2.0),
6.82 (1H, J 8.2); m/z (EI) 235 (Mþ, 100%). Anal. (C13H17NO3) C, H, N.
(S)-4-(3-Ethoxy-4-methoxyphenyl)pyrrolidin-2-one (20S).
20S was prepared as described for 19R in 96% yield from 18S (302 mg)
and iodoethane. Anal. (C13H17NO3) C, H, N.
(R)-4-(4-Methoxy-3-propyloxyphenyl)pyrrolidin-2-one (21R).
21R was prepared as described for 19R in 95% yield from 18R (400 mg) and
1-iodopropane: δH (400 MHz, CDCl3) 1.04 (3H, t, J 7.4), 1.85 (2H, sextet,
J 7.4), 2.47 (1H, dd, J 16.9 and 8.9), 2.70 (1H, dd, J 16.9 and 8.9), 3.38 (1H,
dd, J 9.4 and 7.4), 3.62 (1H, app quintet, J 8.4), 3.72ꢀ3.77 (1H, m), 3.84
(3H, s), 3.96 (2H, t, J 6.8), 6.59 (1H, br s), 6.75ꢀ6.79 (2H, m), 6.82 (1H,
J 8.1); m/z (EI) 249 (Mþ, 59%), 150 (100). Anal. (C14H19NO3) C, H, N.
(S)-4-(4-Methoxy-3-propyloxyphenyl)pyrrolidin-2-one (21S).
21S was prepared as described for 19R in 95% yield from 18S (120 mg) and
1-iodopropane. Anal. (C14H19NO3) C, H, N.
4-(8-Ethyl-1,7-naphthyridin-6-yl)benzoic Acid (28). A solu-
8-[(Trimethylsilyl)ethynyl]-1,7-naphthyridin-6-amine (23). A
sealable 500 mL pressure vessel was charged with 2236 (2.24 g, 8.50 mmol,
prepared as described in the Supporting Information) and a mixture of
THF and Et3N (4:1 v/v, 250 mL) under Ar. To the resulting turbid
mixture was added (Ph3P)2PdCl2 (580 mg, 0.826 mmol), CuI (290 mg,
1.52 mmol), and (trimethylsilyl)acetylene (7.00 mL, 49.9 mmol). The
vessel was sealed and the mixture stirred for 1 h at 50 °C followed by 14 h
at 70 °C. The vessel was then cooled and the mixture filtered through
Celite, washing with EtOAc (150 mL). The filtrate was washed with water
(250 mL) followed by brine (3 ꢁ 50 mL), dried (Na2SO4), and
concentrated in vacuo. The title compound (617 mg, 30%) was isolated
from the resulting residue as a light brown powder by flash column
chromatography (1:3 to 1:1 EtOAcꢀlight petroleum gradient): δH (200
MHz, CDCl3) 0.32 (9H, s), 4.73 (2H, br s), 6.64 (1H, s), 7.33 (1H, dd,
J 3.9 and 8.5), 7.79 (1H, dd, J 1.5 and 8.5), 8.71 (1H, dd, J 1.5 and 3.9).
tion of LiOH H2O (50.2 mg, 1.20 mmol) in water (4 mL) was added to
3
a stirred solution of 27 (175 mg, 0.599 mmol) in a mixture of THF and
MeOH (2:1, 6 mL) at ambient temperature. After 20 h the mixture was
concentrated in vacuo to remove the organic solvent and the aqueous
residue adjusted to pH 6 by addition of 2 M HCl, depositing a fine white
precipitate of product (42 mg) that was collected by filtration. The
filtrate was extracted with EtOAc (50 mL) and the organic layer
subsequently washed with brine (3 ꢁ 10 mL), dried (Na2SO4), and
evaporated. A second batch of product (57 mg) was isolated from the
residual pale yellow solid following trituration with Et2O (15 mL) to
afford 28 (99.0 mg in total, 59%) as a colorless powder: δH (200 MHz,
DMSO-d6) 1.40 (3H, t, J 7.5), 3.50 (2H, q, J 7.5), 7.78 (1H, dd, J 4.1 and
8.4), 8.08 (2H, m), 8.35 (2H, m), 8.42 (1H, s), 8.44 (1H, dd, J 1.7 and
8.4), 9.02 (1H, dd, J 1.7 and 4.1). HRMS (EI) calcd for Mþ C17H14N2O2
279.1128, found 279.1128. Anal. (C17H14N2O2) C, H, N.
3344
dx.doi.org/10.1021/jm200070e |J. Med. Chem. 2011, 54, 3331–3347

Journal of Medicinal Chemistry
ARTICLE
X-ray CIF Files. The X-ray CIF files for (rac)-6 and 9 have been
deposited at the Cambridge Crystallographic Data Center with the
deposition numbers CCDC 808534 and CCDC 808535. Copies of the
data can be obtained, free of charge, from CCDC, 12 Union Road,
Cambridge, CB2 1EZ, U.K. (e-mail, deposit@ccdc.cam.ac.uk; Internet,
http://www.ccdc.cam.ac.uk/).
(5) Vicini, E.; Conti, M. Characterization of an intronic promoter of
a cyclic adenosine 30,50-monophosphate (cAMP): specific phosphodies-
terase gene that confers hormone and cAMP inducibility. Mol. Endocri-
nol. 1997, 11, 839–850.
(6) Wallace, D. A.; Johnston, L. A.; Huston, E.; MacMaster, D.;
Houslay, T. M.; Cheung, Y. F.; Campbell, L.; Millen, J. E.; Smith, R. A.;
Gall, I.; Knowles, R. G.; Sullivan, M.; Houslay, M. D. Identification and
characterization of PDE4A11, a novel, widely expressed long isoform
encoded by the human PDE4A cAMP phosphodiesterase gene. Mol.
Pharmacol. 2005, 67, 1920–1934.
(7) Rena, G.; Begg, F.; Ross, A.; Mackenzie, C.; McPhee, I.; Camp-
bell, L.; Huston, E.; Sullivan, M.; Houslay, M. D. Molecular cloning,
genomic positioning, promoter identification, and characterization of
the novel cyclic AMP-specific phosphodiesterase PDE4A10. Mol. Phar-
macol. 2001, 59, 996–1011.
(8) Houslay, M. D.; Baillie, G. S.; Maurice, D. H. cAMP-specific
phosphodiesterase-4 enzymes in the cardiovascular system: a molecular
toolbox for generating compartmentalized cAMP signaling. Circ. Res.
2007, 100, 950–966.
’ ASSOCIATED CONTENT
S
Supporting Information. Experimental procedures for
b
preparation of 2 and 22; procedures for the derivatization of 6S
and 18S as 9 and (S)-rolipram, respectively; crystallographic data
for (rac)-6 and 9 (Figure S1); 1H and 13C NMR data presented
with signal assignments; compound elemental analysis data;
melting point and IR data; comparison of PDE4 binding modes
for rolipram and NVP (Figure S2); CAS registry numbers,
primary references, and structures for PDE4 inhibitors listed in
Figure 2 (Figure S3). This material is available free of charge via
the Internet at http://pubs.acs.org.
(9) Baillie, G. S.; Scott, J. D.; Houslay, M. D. Compartmentalisation
of phosphodiesterases and protein kinase A: opposites attract. FEBS Lett.
2005, 579, 3264–3270.
’ AUTHOR INFORMATION
(10) Baillie, G. S.; Houslay, M. D. Arrestin times for compartmenta-
lised cAMP signalling and phosphodiesterase-4 enzymes. Curr. Opin.
Cell Biol. 2005, 17, 129–134.
(11) Houslay, M. D. Underpinning compartmentalised cAMP sig-
nalling through targeted cAMP breakdown. Trends Biochem. Sci. 2010,
35, 91–100.
Corresponding Author
*Phone: þ44 (0)131 221 9790. Fax: þ44 (0)131 221 3180.
E-mail: D.R.Adams@hw.ac.uk.
(12) McCahill, A. C.; Huston, E.; Li, X.; Houslay, M. D. PDE4
associates with different scaffolding proteins: modulating interactions as
treatment for certain diseases. Handb. Exp. Pharmacol. 2008, 125–166.
(13) Sachs, B. D.; Akassoglou, K. Regulation of cAMP by the p75
neurotrophin receptor: insight into drug design of selective phospho-
diesterase inhibitors. Biochem. Soc. Trans. 2007, 35, 1273–1277.
(14) Sachs, B. D.; Baillie, G. S.; McCall, J. R.; Passino, M. A.;
Schachtrup, C.; Wallace, D. A.; Dunlop, A. J.; MacKenzie, K. F.;
Klussmann, E.; Lynch, M. J.; Sikorski, S. L.; Nuriel, T.; Tsigelny, I.;
Zhang, J.; Houslay, M. D.; Chao, M. V.; Akassoglou, K. p75 neurotro-
phin receptor regulates tissue fibrosis through inhibition of plasminogen
activation via a PDE4/cAMP/PKA pathway. J. Cell Biol. 2007, 177,
1119–1132.
(15) Barber, R.; Baillie, G. S.; Bergmann, R.; Shepherd, M. C.;
Sepper, R.; Houslay, M. D.; Heeke, G. V. Differential expression of
PDE4 cAMP phosphodiesterase isoforms in inflammatory cells of
smokers with COPD, smokers without COPD, and nonsmokers. Am.
J. Physiol.: Lung Cell. Mol. Physiol. 2004, 287, L332–L343.
(16) Hannila, S. S.; Filbin, M. T. The role of cyclic AMP signaling in
promoting axonal regeneration after spinal cord injury. Exp. Neurol.
2008, 209, 321–332.
(17) Sharafkhaneh, A.; Jayaraman, G.; Kaleekal, T.; Sharafkhaneh,
H.; Hirshkowitz, M. Sleep disorders and their management in patients
with COPD. Ther. Adv. Respir. Dis. 2009, 3, 309–318.
(18) Vecsey, C. G.; Baillie, G. S.; Jaganath, D.; Havekes, R.; Daniels,
A.; Wimmer, M.; Huang, T.; Brown, K. M.; Li, X. Y.; Descalzi, G.; Kim,
S. S.; Chen, T.; Shang, Y. Z.; Zhuo, M.; Houslay, M. D.; Abel, T. Sleep
deprivation impairs cAMP signalling in the hippocampus. Nature
2009, 461, 1122–1125.
(19) Terry, R.; Cheung, Y. F.; Praestegaard, M.; Baillie, G. S.;
Huston, E.; Gall, I.; Adams, D. R.; Houslay, M. D. Occupancy of the
catalytic site of the PDE4A4 cyclic AMP phosphodiesterase by rolipram
triggers the dynamic redistribution of this specific isoform in living cells
through a cyclic AMP independent process. Cell. Signalling 2003, 15,
955–971.
(20) Christian, F.; Anthony, D. F.; Vadrevu, S.; Riddell, T.; Day, J. P.;
McLeod, R.; Adams, D. R.; Baillie, G. S.; Houslay, M. D. p62 (SQSTM1)
and cyclic AMP phosphodiesterase-4A4 (PDE4A4) locate to a novel,
reversible protein aggregate with links to autophagy and proteasome
degradation pathways. Cell. Signalling 2010, 22, 1576–1596.
’ ACKNOWLEDGMENT
PyMOL from W. L. DeLano, DeLano Scientific, was used to
create the protein structure images in this work. We thank the
EPSRC Mass Spectrometry Centre at Swansea for mass spectra.
M.D.H. and G.S.B. acknowledge grants from the Medical Re-
search Council (U.K., Grant G0600765) and Fondation Leducq
(Grant 06CVD02) for their involvement in this work.
’ ABBREVIATIONS USED
cAMP, cyclic adenosine 30,50-monophosphate; CHO, Chinese
hamster ovary; COPD, chronic obstructive pulmonary disease;
DISC1, disrupted-in-schizophrenia 1; GFP, green fluorescent pro-
tein; LR1, linker region 1; LR2, linker region 2; p62/SQSTM-1,
sequestosome-1; mTor, mammalian target of rapamycin; p75NTR
,
p75 neurotrophin receptor; PDE4, phosphodiesterase type 4; PKA,
protein kinase A; TOSMIC, toluenesulfonylmethyl isocyanide;
TNF, tumor necrosis factor; UCR1, upstream conserved region 1;
UCR2, upstream conserved region 2; WT, wild type
’ REFERENCES
(1) Conti, M.; Richter, W.; Mehats, C.; Livera, G.; Park, J. Y.; Jin, C.
Cyclic AMP-specific PDE4 phosphodiesterases as critical components
of cyclic AMP signaling. J. Biol. Chem. 2003, 278, 5493–5496.
(2) Houslay, M. D.; Adams, D. R. PDE4 cAMP phosphodiesterases:
modular enzymes that orchestrate signalling cross-talk, desensitization
and compartmentalization. Biochem. J. 2003, 370, 1–18.
(3) Le Jeune, I. R.; Shepherd, M.; Van Heeke, G.; Houslay, M. D.;
Hall, I. P. Cyclic AMP-dependent transcriptional up-regulation of
phosphodiesterase 4D5 in human airway smooth muscle cells: identi-
fication and characterization of a novel PDE4D5 promoter. J. Biol. Chem.
2002, 277, 35980–35989.
(4) Monaco, L.; Vicini, E.; Conti, M. Structure of 2 rat genes coding
for closely-related rolipram-sensitive cAMP phosphodiesterases: multi-
ple messenger-RNA variants originate from alternative splicing and
multiple start sites. J. Biol. Chem. 1994, 269, 347–357.
3345
dx.doi.org/10.1021/jm200070e |J. Med. Chem. 2011, 54, 3331–3347

Journal of Medicinal Chemistry
ARTICLE
(21) Pankiv, S.; Lamark, T.; Bruun, J. A.; Overvatn, A.; Bjorkoy, G.;
Johansen, T. Nucleocytoplasmic shuttling of p62/SQSTM1 and its role
in recruitment of nuclear polyubiquitinated proteins to promyelocytic
leukemia bodies. J. Biol. Chem. 2010, 285, 5941–5953.
(22) Moscat, J.; Diaz-Meco, M. T. p62 at the crossroads of auto-
phagy, apoptosis, and cancer. Cell 2009, 137, 1001–1004.
(23) Ichimura, Y.; Komatsu, M. Selective degradation of p62 by
autophagy. Semin. Immunopathol. 2010, 32, 431–436.
(24) Kim, H. W.; Ha, S. H.; Lee, M. N.; Huston, E.; Kim, D. H.; Jang,
S. K.; Suh, P. G.; Houslay, M. D.; Ryu, S. H. Cyclic AMP controls mTOR
through regulation of the dynamic interaction between Rheb and
phosphodiesterase 4D. Mol. Cell. Biol. 2010, 30, 5406–5420.
(25) Card, G. L.; England, B. P.; Suzuki, Y.; Fong, D.; Powell, B.; Lee,
B.; Luu, C.; Tabrizizad, M.; Gillette, S.; Ibrahim, P. N.; Artis, D. R.;
Bollag, G.; Milburn, M. V.; Kim, S. H.; Schlessinger, J.; Zhang, K. Y. J.
Structural basis for the activity of drugs that inhibit phosphodiesterases.
Structure 2004, 12, 2233–2247.
(26) Huai, Q.; Wang, H. C.; Sun, Y. J.; Kim, H. Y.; Liu, Y. D.; Ke,
H. M. Three-dimensional structures of PDE4D in complex with
roliprams and implication on inhibitor selectivity. Structure 2003, 11,
865–873.
(27) Xu, R. X.; Rocque, W. J.; Lambert, M. H.; Vanderwall, D. E.;
Luther, M. A.; Nolte, R. T. Crystal structures of the catalytic domain of
phosphodiesterase 4B complexed with AMP, 8-Br-AMP, and rolipram.
J. Mol. Biol. 2004, 337, 355–365.
(28) Zhang, K. Y. J.; 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. A glutamine switch mechanism for
nucleotide selectivity by phosphodiesterases. Mol. Cell 2004, 15,
279–286.
(foci-inducer) and piclamilast and cilomilast (foci-blockers that share
rolipram’s 3-cyclopentyloxy-4-methoxyphenyl core) show no significant
differences in the conformation of the M-loop, the induced fit model for
PDE4A4 foci formation also required initial consideration because
structural differences between the complexes of foci-blocker and foci-
former might only be evident with the full-length enzyme; all of the
currently available crystal structures are N-terminally truncated.
(34) Demnitz, J.; LaVecchia, L.; Bacher, E.; Keller, T. H.; Muller, T.;
Schurch, F.; Weber, H. P.; Pombo-Villar, E. Enantiodivergent synthesis
of (R)- and (S)-rolipram. Molecules 1998, 3, 107–119.
(35) Wang, H.; Peng, M. S.; Chen, Y.; Geng, J.; Robinson, H.;
Houslay, M. D.; Cai, J.; Ke, H. Structures of the four subfamilies of
phosphodiesterase-4 provide insight into the selectivity of their inhibi-
tors. Biochem. J. 2007, 408, 193–201.
(36) Hersperger, R.; Bray-French, K.; Mazzoni, L.; Muller, T.
Palladium-catalyzed cross-coupling reactions for the synthesis of 6,8-
disubstituted 1,7-naphthyridines: a novel class of potent and selective
phosphodiesterase type 4D inhibitors. J. Med. Chem. 2000, 43, 675–682.
(37) Huai, Q.; Liu, Y.; Francis, S. H.; Corbin, J. D.; Ke, H. Crystal
structures of phosphodiesterases 4 and 5 in complex with inhibitor
3-isobutyl-1-methylxanthine suggest a conformation determinant of
inhibitor selectivity. J. Biol. Chem. 2004, 279, 13095–13101.
(38) Burgin, A. B.; Magnusson, O. T.; Singh, J.; Witte, P.; Staker,
B. L.; Bjornsson, J. M.; Thorsteinsdottir, M.; Hrafnsdottir, S.; Hagen, T.;
Kiselyov, A. S.; Stewart, L. J.; Gurney, M. E. Design of phosphodiesterase
4D (PDE4D) allosteric modulators for enhancing cognition with
improved safety. Nat. Biotechnol. 2010, 28, 63–70.
(39) Houslay, M. D.; Adams, D. R. Putting the lid on phosphodies-
terase 4. Nat. Biotechnol. 2010, 28, 38–40.
(40) Lee, M. E.; Markowitz, J.; Lee, J. O.; Lee, H. Crystal structure of
phosphodiesterase 4D and inhibitor complex(1). FEBS Lett. 2002, 530,
53–58.
(29) A variety of metal ions may occupy the Me2 site and support
catalytic activity, but the physiologically relevant ion, which also
occupies the site in the majority of PDE4 crystal structures, is Mg2þ
.
(41) The amide carbonyl group of PDE4-bound piclamilast is
orientated toward the position that would be occupied by Tyr286 in
In the 1RO6 structure the Me2 site is occupied by Mn2þ (or a mixed
Mn2þ/Ca2þ population). We considered that the presence of the Mn2þ
ion, with a slightly larger ionic radius than Mg2þ, might simulate the
enzyme with its catalytic pocket in a mildly contracted state.
the UCR2-capped state but would not form a close contact (O
O
3 3 3
distance of ∼3.7 Å expected).
(42) Gibson, L. C.; Hastings, S. F.; McPhee, I.; Clayton, R. A.;
Darroch, C. E.; Mackenzie, A.; Mackenzie, F. L.; Nagasawa, M.; Stevens,
P. A.; Mackenzie, S. J. The inhibitory profile of ibudilast against the
human phosphodiesterase enzyme family. Eur. J. Pharmacol. 2006, 538,
39–42.
(30) Kemperman, G. J.; Roeters, T. A.; Hilberink, P. W. Cleavage of
aromatic methyl ethers by chloroaluminate ionic liquid reagents. Eur.
J. Org. Chem. 2003, 1681–1686.
(31) Keller, T. H.; Bray-French, K.; Demnitz, F. W.; Muller, T.;
Pombo-Villar, E.; Walker, C. Synthesis and structureꢀactivity relation-
ship of N-arylrolipram derivatives as inhibitors of PDE4 isozymes. Chem.
Pharm. Bull. (Tokyo) 2001, 49, 1009–1017.
(32) Bolger, G. B.; Baillie, G. S.; Li, X.; Lynch, M. J.; Herzyk, P.;
Mohamed, A.; Mitchell, L. H.; McCahill, A.; Hundsrucker, C.; Klussmann,
E.; Adams, D. R.; Houslay, M. D. Scanning peptide array analyses identify
overlapping binding sites for the signalling scaffold proteins, β-arrestin and
RACK1, in cAMP-specific phosphodiesterase PDE4D5. Biochem. J. 2006,
398, 23–36.
(33) We also initially considered an alternative model in which the
N-alkyl groups in 10S and 11S might ablate the foci-forming property of
(S)-rolipram by obstructing an induced fit of the inhibitor within the Q2
subpocket rather than obstructing protein docking or folding. Such a fit
could conceivably drive structural transition leading to foci formation by
exerting leverage at the junction between helices 14 and 15. Occupancy
of the Q2 pocket in PDE5, typically with more penetrating hydrophobic
substituents, is important for achieving high affinity binding of
inhibitors.51,52 Such occupancy undoubtedly influences the conforma-
tion of the enzyme’s M-loop, encompassing the connecting sequence
between helices 14 and 15, albeit that the situation is likely complicated
by coordinated movement in both H- and M-loops.52,53 There is no
evidence that the cognate regions in PDE4 share the extensive con-
formational and positional mobility of the PDE5 H- and M-loops.
Nevertheless, the influential role of Q2 occupancy in PDE5 suggested
that this should be considered as a possible driver for structural transition
in PDE4 and reinforced the decision to probe the role of the cyclopen-
tyloxy group. Although the available cocrystal structures of rolipram
(43) Johnston, K. A.; Allcock, R. W.; Jiang, Z.; Collier, I. D.; Blakli,
H.; Rosair, G. M.; Bailey, P. D.; Morgan, K. M.; Kohno, Y.; Adams, D. R.
Concise routes to pyrazolo[1,5-a]pyridin-3-yl pyridazin-3-ones. Org.
Biomol. Chem. 2008, 6, 175–186.
(44) Alvarez, R.; Sette, C.; Yang, D.; Eglen, R. M.; Wilhelm, R.;
Shelton, E. R.; Conti, M. Activation and selective inhibition of a cyclic
AMP-specific phosphodiesterase, PDE-4D3. Mol. Pharmacol. 1995, 48,
616–622.
(45) McPhee, I.; Yarwood, S. J.; Scotland, G.; Huston, E.; Beard,
M. B.; Ross, A. H.; Houslay, E. S.; Houslay, M. D. Association with the
SRC family tyrosyl kinase LYN triggers a conformational change in the
catalytic region of human cAMP-specific phosphodiesterase HSPDE4A4B.
Consequences for rolipram inhibition. J. Biol. Chem. 1999, 274, 11796–
11810.
(46) Saldou, N.; Obernolte, R.; Huber, A.; Baecker, P. A.; Wilhelm,
R.; Alvarez, R.; Li, B.; Xia, L.; Callan, O.; Su, C.; Jarnagin, K.; Shelton,
E. R. Comparison of recombinant human PDE4 isoforms: interaction
with substrate and inhibitors. Cell. Signalling 1998, 10, 427–440.
(47) Houslay, M. D.; Christian, F. p62 (SQSTM1) forms part of a
novel, reversible aggregate containing a specific conformer of the cAMP
degrading phosphodiesterase, PDE4A4. Autophagy 2010, 6, 1198–1200.
(48) Huston, E.; Pooley, L.; Julien, P.; Scotland, G.; McPhee, I.;
Sullivan, M.; Bolger, G.; Houslay, M. D. The human cyclic AMP-specific
phosphodiesterase PDE-46 (HSPDE4A4B) expressed in transfected
COS7 cells occurs as both particulate and cytosolic species that exhibit
distinct kinetics of inhibition by the antidepressant rolipram. J. Biol.
Chem. 1996, 271, 31334–31344.
3346
dx.doi.org/10.1021/jm200070e |J. Med. Chem. 2011, 54, 3331–3347

Journal of Medicinal Chemistry
ARTICLE
(49) Lobban, M.; Shakur, Y.; Beattie, J.; Houslay, M. D. Identifica-
tion of two splice variant forms of type-IVB cyclic AMP phosphodies-
terase, DPD (rPDE-IVB1) and PDE-4 (rPDE-IVB2) in brain: selective
localization in membrane and cytosolic compartments and differential
expression in various brain regions. Biochem. J. 1994, 304, 399–406.
(50) Gottlieb, H. E.; Kotlyar, V.; Nudelman, A. NMR chemical shifts
of common laboratory solvents as trace impurities. J. Org. Chem.
1997, 62, 7512–7515.
(51) Allerton, C. M.; Barber, C. G.; Beaumont, K. C.; Brown, D. G.;
Cole, S. M.; Ellis, D.; Lane, C. A.; Maw, G. N.; Mount, N. M.; Rawson,
D. J.; Robinson, C. M.; Street, S. D.; Summerhill, N. W. A novel series of
potent and selective PDE5 inhibitors with potential for high and dose-
independent oral bioavailability. J. Med. Chem. 2006, 49, 3581–3594.
(52) Chen, G.; Wang, H.; Robinson, H.; Cai, J.; Wan, Y.; Ke, H. An
insight into the pharmacophores of phosphodiesterase-5 inhibitors from
synthetic and crystal structural studies. Biochem. Pharmacol. 2008, 75,
1717–1728.
(53) Wang, H.; Ye, M.; Robinson, H.; Francis, S. H.; Ke, H.
Conformational variations of both phosphodiesterase-5 and inhibitors
provide the structural basis for the physiological effects of vardenafil and
sildenafil. Mol. Pharmacol. 2008, 73, 104–110.
3347
dx.doi.org/10.1021/jm200070e |J. Med. Chem. 2011, 54, 3331–3347