Identification of PDE4B Over 4D subtype-selective inhibitors revealing an unprecedented binding mode

Article abstract of DOI:10.1016/j.bmc.2009.03.061

A PDE4B over 4D-selective inhibitor programme was initiated to capitalise on the recently discovered predominance of the PDE4B subtype in inflammatory cell regulation. The SAR of a tetrahydrobenzothiophene (THBT) series did not agree with either of two proposed docking modes in the 4B binding site. A subsequent X-ray co-crystal structure determination revealed that the THBT ligand displaces the Gln-443 residue, invariably ligand-anchoring in previous PDE4 co-crystal structures, and even shifts helix-15 by 1-2 A. For the first time, several residues of the C-terminus previously proposed to be involved in subtype selectivity are resolved and three of them extend into the ligand binding site potentially allowing for selective drug design.

Full text of DOI:10.1016/j.bmc.2009.03.061

Bioorganic & Medicinal Chemistry 17 (2009) 5336–5341
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry
journal homepage: www.elsevier.com/locate/bmc
Identification of PDE4B Over 4D subtype-selective inhibitors revealing
an unprecedented binding mode
*
Michael Kranz , Michael Wall, Brian Evans, Afjal Miah, Stuart Ballantine, Chris Delves, Brian Dombroski,
Jeffrey Gross, Jessica Schneck, James P. Villa, Margarete Neu, Don O. Somers
Medicines Research Centre, GlaxoSmithKline, Gunnels Wood Road, Stevenage, Hertfordshire, SG1 2NY, England, UK
a r t i c l e i n f o
a b s t r a c t
Article history:
A PDE4B over 4D-selective inhibitor programme was initiated to capitalise on the recently discovered
predominance of the PDE4B subtype in inflammatory cell regulation. The SAR of a tetrahydrobenzothio-
phene (THBT) series did not agree with either of two proposed docking modes in the 4B binding site. A
subsequent X-ray co-crystal structure determination revealed that the THBT ligand displaces the
Gln-443 residue, invariably ligand-anchoring in previous PDE4 co-crystal structures, and even shifts
helix-15 by 1–2 Å. For the first time, several residues of the C-terminus previously proposed to be
involved in subtype selectivity are resolved and three of them extend into the ligand binding site
potentially allowing for selective drug design.
Received 27 November 2008
Revised 27 March 2009
Accepted 31 March 2009
Available online 5 April 2009
Keywords:
Lead optimisation
Ligand docking
Induced fit
SAR
Ó 2009 Elsevier Ltd. All rights reserved.
1. Introduction
For exactly 4 decades, inhibitors of 30,50-nucleotide phosphodi-
esterase enzymes (PDE) have been known.1 The initially discovered
natural product inhibitors, like papaverine, were found to interfere
with the hydrolysis of the endogenous substrates of cAMP and
cGMP non-selectively for the various PDE subtypes. The dominant
therapeutic role of the PDE4 protein family was established
through the action of the first-generation PDE4 inhibitor rolipram
1 as an antidepressant agent.2 The dose-limiting side effects of
nausea and vomiting were reduced by second-generation inhibi-
tors like cilomilast (Ariflo, 2) and roflumilast 3, which have suc-
ceeded as anti-inflammatories in clinical trials for COPD and
asthma as oral agents, but their therapeutic index has prevented
market launch so far.3 Most recent studies have linked the PDE4B
subtype to inflammatory cell regulation4 while the PDE4D subtype
is implied in the emetic response.5 Consequently, this paper de-
scribes how we tried to optimise the potency of a PDE4B-selective
hit from an HTS based on docking models. The X-ray structure of a
lead molecule surprised with an unexpected binding mode and an
unprecedented active site deformation.
O
O
O
O
NC
O
N
HO2C
1
2
O
O
F
Cl
F
N
N
O
Cl
3
2. Results and discussion
2.1. Initial docking mode
During an HTS campaign, a tetrahydrobenzothiophene (THBT)
bisamide (4) was discovered to be a potent and modestly PDE4B-
over 4D-selective inhibitor (PDE3 (bovine enzyme SPA assay) and
PDE5 (human enzyme SPA assay) had a pIC50 < 5). With a large
number of in-house and published ligand/PDE4B catalytic domain
co-crystal structures available,6 it was hoped that elucidation of
the THBT binding mode would drive the analogue selection to de-
rive PDE4B- over 4D-selective SAR.7
* Corresponding author. Tel.: +44 1438 763385; fax: +44 1438 763352.
E-mail address: michael.j.kranz@gsk.com (M. Kranz).
0968-0896/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmc.2009.03.061

M. Kranz et al. / Bioorg. Med. Chem. 17 (2009) 5336–5341
5337
H2N
S
O
H
N
O
5
Figure 1. Preferred conformation of the thiophene bisamide and the 2-amide-anchored docking mode (Flo+) of a THBT (magenta) overlaid with the X-ray of piclamilast
(1xm4; green).
H2N
An initial array was undertaken to probe the proposed binding
O
mode. Key compounds were synthesised according to the general
pathway described in Scheme 1. The intermediate ethyl 2-amino-
4,5,6,7-tetrahydro-1-benzothiophene-3-carboxylate 8 and 2-ami-
no-4,5,6,7-tetrahydro-1-benzothiophene-3-carboxamide 11 were
prepared using known reaction conditions13 using a suitably
substituted cycloketone in a one pot procedure and in good yields.
It was demonstrated that functionality could be built into the THBT
scaffold at this stage whereby ketones with a variety of substitu-
tion patterns R1 = (alkyl, aryl, protected amine, protected ketone,
CH2) R2 = (alkyl, CH2) R3 = (Alkyl, CH2) were used to synthesise
the corresponding thiophenes. Regioisomers were separable in
some cases at this stage, however where necessary only, it was
decided to separate these at the final stage in the synthesis.
Diversity at the 3-carboxy position was introduced through
amide coupling. Saponification of the ethyl ester 8 using sodium
hydroxide gave the carboxylic acid. Subsequent treatment of the
acid using HATU as the coupling agent and the appropriate amine
gave compounds of type 9 in good yields. Alternatively, the pri-
mary carboxamide 11 could be accessed in a one pot procedure
using the conditions of cyclisation described above.
H
N
S
O
N
4
7
PDE4B/4D pIC50 6.7/6.5
A search in the CDC database8 confirmed the assumption that
the bisamide system is fixed through an intramolecular H bond
(5 in Fig 1). It is known that the prerequisite for PDE4 activity is
a flat aromatic system fitting into the ‘hydrophobic clamp’ as well
as an H bond acceptor for the Gln-443 amino group.9 With the
amide conformations fixed as described above, both carbonyl oxy-
gens as well as the sulfur (although a very weak H bond acceptor)10
were individually coordinated to the ligand-anchoring Gln-443
while keeping the template ring system in the ‘hydrophobic clamp’
of the PDE4 catalytic site.
After optimisation of the ligand position (Flo+, Thistlesoft),11 the
sulfur or the 3-amide carbonyl coordination to Gln-443 would lead
to serious clashes in the deep end of the enzyme pocket whereas
the 2-amide carbonyl gives a reasonable docking fit for this tem-
plate (Fig. 1). In this pose, the aryl ring occupies the same small
pocket as the methoxy group of piclamilast (1xm4),12 the primary
amide points into the water-filled part of the binding site while the
tetrahydrobenzo group orientates to the pocket exit suggesting
ample scope for substitution at both positions.
Optimal conditions for acylation at the 2-amino thiophene were
found when the reaction was carried out in pyridine in the pres-
ence of the corresponding acyl chlorides or fluorides, either pre-
formed or generated in situ and furnished the final compounds
of types 11 and 12.
H
R
N
O
F
NH
2.2. Chemistry
S
O
A substructure search of in-house thiophene bisamides resulted
in 32 compounds fitting the initial docking model. The most active
hit 6 demonstrated significant (10-fold) selectivity against the 4D
subtype and represented a suitable starting point from which to
launch our hit-to-lead effort.
F
PDE4B pIC50 R = H (13a): 7.1;
R = Me (13b): < 5
2.3. Alternative binding mode
H2N
O
H
N
Initial SAR at the 3-carboxy position showed significantly
reduced potencies when the primary amide was methylated as
exemplified by 13a/b and indicated a clear requirement for the pri-
mary carboxamide at this position. This suggested that the current
docking hypothesis was untenable as the primary carboxamide
S
O
N
O
O
6
PDE4B/4D pIC50 7.3/6.3

5338
M. Kranz et al. / Bioorg. Med. Chem. 17 (2009) 5336–5341
R4
R4
N
N
O
R3
O
R3
R3
O
R3
O
R2
d,e
R2
R1
R2
R1
O
R2
R1
c
b
H
NH2
N
NH2
S
S
S
R1
7
O
9
10
8
a
H2N
R3
H2N
O
R3
O
R2
R1
R2
R1
H
b
N
NH2
S
S
O
12
11
Scheme 1. Reagents and conditions: (a) 2-cyanoacetamide, sulfur, morpholine, ethanol, 80 °C 80–95%. (b) Acid chloride, pyridine, 25 °C 65–95%. (c) Ethyl cyanoacetate,
sulfur, morpholine, ethanol, 80 °C, 80–95%. (d) (i) NaOH, methanol/water (1:1), 100 °C, (ii) H+. (e) HATU, Amine, i-Pr2NEt, DMF, 25 °C. R1, R2, R3 can be 2H, alkyl or aryl,
protected amine or ketone and R4 can be H, alkyl or aryl.
Table 2
SAR around the tetrahydrobenzo group
H2N
O
4
5
H
R
N
S
6
O
O
N
O
R
pIC50 PDE4B
pIC50 PDE4D
H (16)
5.9
5.3
5.3
5.7
5.9
7.3
8.0
7.6
7.3
7.0
6.4
5.2
4.5
5.6
5.2
5.2
6.3
6.9
6.9
6.4
5.6
5.8
4-Me (17)
4-(2,6-Toluyl) (18)
5-Me (19)
5-tBu (20)
6-Me (6)
Figure 2. Overlay of the X-ray of piclamilast in PDE4B (1xm4; green) and the
second lowest energy docking pose for the alternative amide anchoring of a THBT
(magenta).
6-Et (21)
6-tBu (22)
6,6-(Me)2 (23)
6-n-Pn (24)
6-Ph (25)
Table 1
SAR around the phenylamide
H2N
O
site exit (Fig. 2) where the 4B/4D selectivity is thought to be deter-
mined.14 This model also suggests that there is enough space
around the ortho- and meta-positions of the phenyl ring of 13a to
account for the observed SAR, with the para-position blocked by
the protein, also in agreement with the SAR (Table 1).
H
N
S
O
R
The emerging SAR around the tetrahydrobenzo group confirms
the 4B-subtype selectivity of this series and indicates a preference
for small C6 substituents (Table 2). High potency was achieved
where R = 6-ethyl 21, whilst maintaining the 10-fold selectivity in-
dex. However, this feature cannot be reconciled with the alterna-
tive binding mode. Throughout the progress of the programme,
crystallisation efforts were undertaken to facilitate a structure-
based drug design approach.
R
pIC50 PDE4B
pIC50 PDE4D
o-NO2 (6)
m-NO2 (14)
p-NO2 (15)
7.3
6.2
<5.6
6.3
<5.6
<5.6
herein points into a water-filled pocket that should disfavour a
methylated amide but should not render the molecule completely
inactive.
2.4. Binding mode in co-crystal structure
To give credit to the significance of the primary amide, this
group was coordinated to Gln-443 and a conformational search
was performed in the binding site (Flo+). The lowest energy struc-
ture also had the pyridyl ring in a similar vicinity to the piclamilast
methoxy group as shown in the overlay in Figure 1. The second
lowest binding mode has the ligand less deeply buried in the pock-
et thus bringing the thiophene ring system closer to the binding
Constructs previously used in crystallography15 have focused
on the catalytic domain from residue 152 to residues 503, 506
and 528 but resolving a structure beyond about residue 487 has
been successful only once.16 The C-terminal residues 496–508
have been resolved in a PDE4B apo-structure but their provenance

M. Kranz et al. / Bioorg. Med. Chem. 17 (2009) 5336–5341
5339
selectivity-determining residues reside within the C-terminus of
PDE4 located just outside the catalytic pocket.14 This effort led to a
new crystal form produced from construct PDE4B2 152–528 (triple
mutant) co-crystallised with a THBT inhibitor (13) that allowed
the structure of additional C-terminal residues Asp-498 to Glu-509
to be built and shows the THBT bound in a completely unexpected
way.
The THBT ligand binds deep in the binding site, perpendicular to
both computationally predicted poses, forcing an unprecedented
distortion of several catalytic domain helices (Fig. 3). Most remark-
ably, the ligand enlarges the catalytic pocket by forcing the top of
helix-15 a distance of 1–2 Å outwards together with the termini of
helix-14 and -16.
While the ligand still slots into the ‘hydrophobic clamp’, the
obligatory H bond to Gln-443 is not only absent but this residue
is pushed from its position in all known PDE4 X-ray structures
(Fig. 4). The Gln amide plane is almost perpendicular to the usual
orientation and the amino group forms an H bond with Thr-407.
Figure 3. Overlay of PDE4B/THBT ligand complex (cyan) and unliganded 1f0j (1.8 Å
resol.; orange) X-rays; helix-14 above ligand, helix-15 to the right, helix-16 far
right.
from a neighbouring protein led the authors to describe this inter-
action as a crystallographic artefact. In addition, some of the side
chains extend into the binding pocket preventing any potential li-
gand binding.
Further crystallisation attempts were made on the longest pro-
tein construct (and others generated from extensive cloning and
expression studies)17 based on the hypothesis that the PDE4B/4D
Figure 6. C-terminal residues resolved in THBT PDE4B X-ray structure.
Figure 4. Overlay of the X-rays of piclamilast (green) and THBT 13 (magenta).
Figure 5. PDE4B/THBT X-ray and overlay of PDE4B/THBT (magenta) and PDE4D/zardaverine (1xor, 1.5 Å resol.; orange) X-rays and shortest distance of ligand to the 2 closest
4B/4D side chain variations.

5340
M. Kranz et al. / Bioorg. Med. Chem. 17 (2009) 5336–5341
CO2H
N
N
NVP
NO2
Figure 7. Overlay of PDE4B/THBT (magenta) and PDE4B/NVP (2qyl, 1.95 Å resol.; green) X-ray structures; three C-terminal side chains in ball and stick.
An alternative ligand-anchoring H bond is found between the
domain which has been rigid in dozens of previous apo and co-
crystal structures.
primary ligand amide and Asn-395 as well as a water molecule
at the back of the binding pocket (Fig. 4). Whereas the part of
the ligand close to the metals stays within the space commonly
occupied by PDE4 inhibitors, the THBT extends several Ångstroms
beyond this space at the opposite end. The established SAR in the
THBT series can now be rationalised with this docking mode. The
4- and 5-positions of the tetrahydrobenzo group are deeply buried
in the protein pocket and inaccessible for substitution. It is position
6, which is responsible for the subtype selectivity (Table 2), that
extends towards Hlx-15. The 6-ethyl group gives maximal PDE4B
activity of the groups in Table 2 whereas larger and smaller sub-
stituents decrease activity. While the degree of protein deforma-
tion seems to correlate with the level of 4B activity and 4B/4D
selectivity, the mechanistic connection is not clear.
Notably, the THBT is 6 Å removed from the C-terminal helix
fragment 498–509 (Fig. 5). Three C-terminal residues point into
the binding pocket, Leu502Gln (4B amino acid/4B2 numbering/
4D amino acid) Lys505Lys and Phe506Phe (Fig. 6), but only one
of them can potentially be exploited for subtype-selective ligand
design. So a ligand like NVP that extends towards this residue pair
Leu/Gln in the PDE4B/4D co-crystal structures, respectively
(Fig. 7),18 could exhibit subtype selectivity if a differential interac-
tion between ligand and Leu/Glu can be achieved.
4. Experimental
4.1. General
HPLC chromatography was performed using column:
3.3 cm ꢀ 4.6 mm ID, 3
l
m ABZ + PLUS with a flow rate: 3 ml/min
and injection volume of 5
ll at room temperature which was cou-
pled to a UV detector with a range of 215–330 nm. Mass directed
auto preparative chromatography was carried out using a Supelco-
sil ABZ+Plus column; 100 mm ꢀ 21.2 mm ID with a stationary
phase particle size of 5 l
m. 1H NMR data were collected using
the Bruker 400 MHz spectrometer using deuterated solvents and
(CH3)4Si as the internal standard.
4.1.1. General procedure for the synthesis of n-substituted
tetrahydrobenzothiophenes scaffolds
The cyclohexanone (1 equiv) 2-cyanoacetamide (1.05 equiv)
and sulfur (1.05 equiv) were added to ethanol (6 ml/g) and heated
to reflux. Morpholine (1.05 equiv) was added dropwise to the heat-
ing solution and the mixture was allowed to heat to reflux for 2–
4 h. Upon cooling, a precipitate formed which was filtered and
washed with cold ethanol. The precipitate was recrystallised from
hot ethanol, filtered and dried to give the desired compound.
Other catalytic domain residues differing between 4B and 4D,
like Gln416Arg in helix-14 and Thr436Asn in the helix-15/16 loop
are more than 14 Å distant from the ligand. The orientation of both
is affected by the helix shift induced by the THBT but both face to-
wards the solvent (Fig. 5). It is not clear how this PDE4B/4D se-
quence difference around the ligand binding site, the movement
of helix-15 and the nature of the C-terminal residues extending
into the binding pocket contribute to the subtype selectivity ob-
served for some of the THBT ligands.
4.1.2. 2-Amino-6-methyl-4,5,6,7-tetrahydro-1-benzothio-
phene-3-carboxamide (11, where R1 = CH3, R2 = R3 = H2)
Orange solid (2.9 g, 51%); 1H NMR (400 MHz, CDCl3) d 01.07 (d,
J = 8 Hz, 3H), d 1.42–1.48 (m, 1H), d 1.89–1.94 (m, 2H), d 2.16–2.23
(m, 1H), d 2.57–2.63 (m, 2H), d 2.74–2.78 (m, 1H), d 5.40 (brS,
2H), d 6.15 (brs, 2H); MS (ESI) m/z 211.13 (M+H)+; LC (eluent: ace-
tonitrile/water, 0–100% gradient over 5 min): t 2.77 min.
3. Conclusions
4.1.3. 6-Methyl-2-{[(2-nitrophenyl)carbonyl]amino}-4,5,6,
7-tetrahydro-1-benzothiophene-3-carboxamide (6)
In summary, the SAR of a 10-fold PDE4B-selective tetrahydro-
thiophene inhibitor series did not agree with either of two pro-
posed docking modes derived from existing co-crystal structures
for other PDE4 inhibitor templates. Eventually, the PDE4 co-crystal
structure of a lead molecule was solved and surprised us twofold,
with a considerable shift of the helix-15 backbone as well as the
displacement of a crucial residue for ligand coordination (Gln-
443). The unexpected binding mode emphasises the possibility of
an induced fit effect even in a protein like the PDE4 catalytic
To a solution of the amine 7 (100 mg, 0.47 mmol) in pyridine
(1 ml) was added 2-nitrobenzoyl chloride (1.2 equiv 0.57 mmol,
106 mg) and the reaction mixture allowed to stir at room temper-
ature for 16 h. Upon completion, the reaction mixture was concen-
trated under vacuum and the crude mixture crystallised from hot
ethanol to afford the desired compound (110 mg, 65%). 1H NMR
(400 MHz, CDCl3) d 01.12 (d, J = 8 Hz, 3H), d 1.49–1.52 (m, 1H), d

M. Kranz et al. / Bioorg. Med. Chem. 17 (2009) 5336–5341
5341
1.96–2.01 (m, 2H), d 2.35–2.39 (m, 1H), d 2.71–2.75 (m, 1H), d 2.81–
References and notes
2.88 (m, 2H), d 5.72 (br s, 2H), d 7.63–7.74 (m, 3H), d 8.06–8.08 (d,
J = 8 Hz, 1H) d 12.62 (s, 1H); MS (ESI) m/z 360.23 (M+H)+, 358.28
(MꢁH)+; LC (eluent: acetonitrile/water, 0–100% gradient over
5 min); t 3.31 min.
Compounds in Tables 1 and 2. From 11, the general procedure
described for the synthesis of 6 was followed.
1. Cheung, W. Y. Biochemistry 1967, 6, 1079.
2. Schwabe, U.; Miyake, M.; Ohga, Y.; Daly, J. Mol. Pharmacol. 1976, 12, 900.
3. Dastidar, S. G.; Rajagopal, D.; Ray, A. Curr. Opin. Invest. Drugs 2007, 85, 364.
4. Jin, S.-L. C.; Lan, L.; Zoudilova, M.; Conti, M. J. Immun. 2005, 175, 1523.
5. Robichaud, A.; Stamatiou, P. B.; Jin, S.-L. C.; Lachance, N.; MacDonald, D.;
Laliberte, F.; Liu, S.; Huang, Z.; Conti, M.; Chan, C.-C. J. Clin. Invest. 2002, 110, 1045.
6. 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. Mol. Cell 2004, 15, 279.
7. Inhibition of PDE4B and PDE4D were measured using a luminescence-coupled
assay system developed by Cambrex. This assay system couples the formation
of AMP, derived from PDE4-catalyzyed hydrolysis of cAMP, to the formation of
ATP. The ATP is then used as a substrate for Luciferase and results in light as a
signal output. When PDE is inhibited or inactive, no AMP is produced, the
luciferase is inactive, and no light signal is produced. This assay is used in a
5. Crystallography
The PDE4B2B (residues 152–528, triple mutant [S482A, S487A,
S489A]) was expressed in Baculovirus infected Sf9 cells and purified
as described previously.19 To minimise C-terminal cleavage (to 503)
during cellbreakage andextraction, a proteaseinhibitorcocktailwas
included (Roche complete tablets). After clarification by centrifuga-
tion and filtration, PDE4B2 protein was purified on Q-Sepharose FF
and Cibacron Blue Sepharose FF. The eluate was applied to a 10 ml
Q-Sepharose HP column and eluted with a NaCL gradient to separate
out any cleaved protein. Intact protein was further purified by size
exclusion on a Superdex 200 column (26/60) and finally dialysed
into 10 mMHepes, pH7.0, 20 mM NaCL, 0.1 mM EDTAfor crystallog-
raphy. Purified protein was concentrated to 16–20 mg/ml in 10 mM
Hepes, pH 7.0, 20 mM NaCl, 0.1 mM EDTA, 5 mM DTT.
X-ray quality co-crystallisation was set up as a hanging drop
experiment at 4 °C using 1 ll + 1 ll drops with protein (with sixfold
molar excess of inhibitor) and a well solution containing 18% PEG
4000, 18% glycerol, 0.08 M MES pH 6.8, 0.45 M NaCl, 0.15 M MgCl2,
0.05 M Na acetate, 0.05 M Mg acetate and 0.025 M ammonium sul-
fate. The crystals reached full size within four we0eks.
X-ray diffraction data were collected to 2.23 ÅA resolution at the
ESRF (station ID23.1) using a ADSC Q315 CCD detector and pro-
cessed as space group P41 with MOSFLM20 and SCALA within the
CCP4 programme suite.21 The structure was solved by Molecular
Replacement using PHASER22 and a search model derived from
1F0 J. Refinement was performed using REFMAC23 to a final R-fac-
tor = 17.9% (R-free = 22.9%). The coordinates consisted of two mol-
ecules in the asymmetric unit, each with a bound inhibitor. All
residues were built (in both molecules) except 486–494 and
512–528 that were too disordered. The coordinates have been
deposited in the PDB as deposition code 3HMV.
quenched assay format, where PDE4 enzyme (2.5
40 mM Tris–HCl, 10 mM MgCl2, 1 mM CHAPS, 0.01% BSA, pH 7.5.) and cAMP
substrate (2.5 L; 2 M cAMP in 40 mM Tris–HCl, 10 mM MgCl2, 1 mM CHAPS,
lL; ꢂ120 pM enzyme in
l
l
0.01% BSA, pH 7.5.) are added sequentially to a 384 well assay plate (Greiner
784075) pre-stamped with 12.5-50 nL compound at the desired concentration.
The reaction is incubated at room temperature for 1 h, then is quenched by the
addition of enzyme stop solution (1.5
catalog # LT27-253) and then the light signal is generated by the addition of
detection reagent (2.5 L, prepared as described by vendor, catalog# LT27-
lL; prepared as described by vendor;
l
250). The luminescence is then measured on a Viewlux imager (Perkin Elmer)
using emission filters of 613/55 nm or 618/40 nm and a 5 s. Compounds are
prepared in neat DMSO at a concentration of 10 mM. For inhibition curves,
compounds were diluted using a three fold serial dilution and tested at 11
concentrations (e.g., 50 lM-0.8 nM or 25 lM–0.42 nM or 2.5 lM–42 pM).
Curves were analysed as describe above using ActivityBase and XLfit , and
results were expressed as pIC50 values. Average standard deviations are for
PDE4B = 0.10 and PDE4D = 0.11.
8. Allen, F. H. Acta Crystallogr., Sect. B 2002, 58, 380.
9. Lee, M. E.; Markowitz, J.; Lee, J.-O.; Lee, H. FEBS Lett. 2002, 530, 53.
10. Allen, F. H.; Bird, C. M.; Rowland, R. S.; Raithby, P. R. Acta Crystallogr. Sect. B
1997, 53, 696.
11. Mcmartin, C.; Bohacek, R. S. J. Comp. Aided Mol. Des. 1997, 11, 333.
12. 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. Structure 2004, 12, 2233.
13. Gewald, K.; Schael, J. J. Prakt. Chem. 1973, 315, 39.
14. Houslay, M. D.; Adams, D. R. Biochem. J. 2003, 370, 1.
15. McLaughlin, M. M.; Cieslinski, L. B.; Burman, M.; Torphy, T. J.; Livi, G. P. J. Biol.
Chem. 1993, 268, 6470.
16. Xu, R. X.; Hassell, A. M.; Vanderwall, D.; Lambert, M. H.; Holmes, W. D.; Luther,
M. A.; Rocque, W. J.; Milburn, M. V.; Zhao, Y.; Ke, H.; Nolte, R. T. Science 2000,
288, 1822.
17. Manuscript in preparation.
18. Wang, H.; Peng, M.-S.; Chen, Y.; Robinson, H.; Houslay, M. D.; Cai, J.; Ke, H.
Biochem. J. 2007, 408, 193.
19. Rocque, W. J.; Holmes, W. D.; Patel, I. R.; Dougherty, R. W.; Ittoop, O.; Overton,
L.; Hoffman, C. R.; Wisely, G. B.; Willard, D. H.; Luther, M. A. Protein Exp. Purif.
1997, 9, 191.
20. Leslie, A. G. W., Joint CCP4 + ESF-EAMCB Newsletter on Protein
Crystallography, 1992, No. 26.
Acknowledgements
We would like to thank Surjit Bains, Girish Shah and Phil Hard-
wick for protein expression and Martin Fillmore for early chemis-
try development. GlaxoSmithKline funded this work.
21. Bailey, S. Acta Crystallogr. Sect. D 1994, 50, 760.
22. Storoni, L. C.; McCoy, A. J.; Read, R. J. Acta Crystallogr. Sect. D 2004, 60, 432.
23. Murshudov, G. N.; Vagin, A. A.; Dodson, E. J. Acta Crystallogr. Sect. D 1997, 53, 240.