Communications to the Editor
J ournal of Medicinal Chemistry, 1998, Vol. 41, No. 3 269
though somewhat less potent, exhibits strikingly similar
SAR. Thus, hydroxamic is preferred over carboxylic
acid (2a vs 15), R-stereochemistry is preferred over S-
((R)-2b,c vs (S)-2b,c), and activity declines upon in-
creasing steric bulk at the R-position ((R)-2b vs (R)-
2c). These results are explained by the ability of series
2 to adopt a conformation similar to 1, in which the
trans amide bond overlaps with the C3-C4 bond of the
2-isoxazline ring and the R-hydrogen eclipses the amide
carbonyl to make a pseudo five-membered ring.17 The
most potent analogue, (R)-2b, has an IC50 of 0.224 µM
in the HM-PDE assay, which is 15-fold lower than that
of rolipram.
the early stages of a model of the PDE4 active site can
be built. This model predicts that by incorporating
structural features of cAMP and a site for metal
coordination within the rolipram pharmacophore, more
potent and structurally diverse analogues can be ob-
tained. Ultimately, the true manner in which PDE4
inhibitors bind in the active site may be obtained by
X-ray crystallography, and we view (R)-1a as an excel-
lent candidate for cocrystallization studies in order to
confirm our model and provide insight into the role of
the divalent metal in the mechanism of cAMP hydrolysis
and the binding of other inhibitors.
The ability of series 1 and 2 analogues to inhibit the
release of HM-TNFR correlates well with their relative
ability to inhibit HM-PDE4, paralleling a previous
study,5b with the most potent compounds, (R)-1a , (R)-
1b, 2a , and (R)-2b, having IC50 values between 3 and
49 nM.18 In the HWB-TNFR assay, which best reflects
their therapeutic potential in TNFR-related diseases,
these compounds retain excellent potency, exhibiting
IC50 values < 80 nM. Interestingly, (R)-1a , with an IC50
of 30 nM, is nearly 3-fold more potent than RP73401 in
this assay, suggesting that the hydroxamic acid has the
added beneficial effect of reducing protein binding. To
our knowledge, (R)-1a is the most potent PDE4 inhibi-
tor of TNFR release in HWB so far discovered.
Ack n ow led gm en t. We would like to thank Drs.
Alan J . Duplantier, Douglas A. Fisher, and J ohn Lowe
III for helpful discussions and Dr. Thomas J . Carty and
Mr. Francis J . Sweeney for measurement of TACE
inhibition. We also thank Drs. Kelvin Cooper and J ohn
W. Watson for their encouragement.
Su p p or tin g In for m a tion Ava ila ble: Experimental pro-
cedures and data for new analogues and intermediates, mean
standard errors for in vitro data, detailed results of the Sybil
force field calculations of series 2, and details of the X-ray
determination of compound (5S)-7a (19 pages). Ordering
information is given on any current masthead page.
The hydroxamic effect described above leads us to
compare the structures of series (R)-1 and (R)-2 with
cAMP as is hypothetically bound via the phosphate19
to the divalent metal in the active site (Figure 1).
Indeed, some strikingly common structural features
exist. First, the oxygen anions of the metal binding
ligands and the aromatic base or catechol are connected
by a five-membered ring template and separated by a
distance of six atoms. Second, within the template a
hydrogen-bonding acceptor is located at one of the
adjacent positions to the aromatic base or catechol
substituent, and a hydrogen-bonding donor is located
at the alternate position (cAMP and 2 only). Third, the
metal ligand is oriented on the re face of the template.
Fourth, small R substituents (i.e., hydrogen for cAMP)
are preferred at the position adjacent to the metal
binding ligand in the template. Inherent in this model
is the prediction that the catechol and adenosine groups
bind in the same region of the enzyme. Despite the
structural difference, there is a common hydrogen-
bonding acceptor four atoms removed from the corre-
sponding template, namely the methoxy oxygen of the
catechol and N1 of the adenosine, respectively. As
illustrated, this manner of overlap predicts that the
cyclopentyl group of the catechol binds in an auxiliary
lipophilic binding pocket, which, if absent in other PDEs
and PDE4 mutants,20 could account for the PDE4
specificity of rolipram. Clearly, this aspect of the model
is less developed, and it is conceivable that the catechol
and adenosine groups, respectively, are oriented differ-
ently within the active site.
Refer en ces
(1) Beavo, J . A. Physiol. Rev. Cyclic nucleotide phosphodiesteras-
es: functional implications of multiple isoforms. 1995, 75, 725-
748.
(2) (a) Palfreyman, M. N.; Souness, J . E. Phosphodiesterase type
IV inhibitors. Prog. Med. Chem. 1996, 33, 1-52. (b) Torphy, T.
J .; Livi, G. P.; Christensen, S. B. Novel phosphodiesterase
inhibitors for the therapy of asthma. Drug News Persp. 1993, 6,
203-214. (c) Torphy, T. J .; Undem, B. J . Phosphodiesterase
inhibitors: new opportunities for the treatment of asthma.
Thorax 1991, 46, 512-523. (d) Erhardt, P. W. Second-generation
PDE inhibitors. In Cyclic Nucleotide Phosphodiesterases: Struc-
ture, Regulation and Drug Action; Beavo, J ., Houslay, M. D.,
Eds.; J ohn Wiley & Sons: New York, 1990; pp 320-332.
(3) Hughes, B.; Owens, R.; Perry, M.; Warrellow, G.; Allen, R. PDE
4 inhibitors: the use of molecular cloning in the design and
development of novel drugs. Drug Discovery Today 1997, 2, 89-
101.
(4) Wilson, M.; Sullivan, M.; Brown, N.; Houslay, M. D. Purification,
characterization and analysis of rolipram inhibition of a human
type-IVA cyclic AMP-specific phosphodiesterase expressed in
yeast. Biochem. J . 1994, 304, 407-415.
(5) (a) Davidsen, S. K.; Summers, J . B. Inhibitors of TNFR synthesis.
Exp. Opin. Ther. Patents 1995, 5, 1087-1100. (b) Souness, J .
E.; Griffin, M.; Maslen, C.; Ebsworth, K.; Scott, L. C.; Pollock,
K.; Palfreyman, M. N.; Karlsson, J .-A. Evidence that cyclic AMP
phosphodiesterase inhibitors suppress TNFR generation from
human monocytes by interacting with a ‘low-affinity’ phosphodi-
esterase 4 conformer. Br. J . Pharmacol. 1996, 118, 649-658.
(6) Ashton, M. J .; Cook, D. C.; Fenton, G.; Karlsson, J .-A.; Palfrey-
man, M. N.; Raeburn, D.; Ratcliff, A. J .; Souness, J . E.;
Thurairatnam, S.; Vicker, N. Selective type IV phosphodiesterase
inhibitors as antiasthmatic agents. The synthesis and biological
activities of 3-(cyclopentyloxy)-4-methoxybenzamides and ana-
logues. J . Med. Chem. 1994, 37, 1696-1703.
(7) Stafford, J . A.; Veal, J . M.; Feldman, P. L.; Valvano, N. L.; Baer,
P. G.; Brackeen, M. F.; Brawley, E. S.; Connolly, K. M.;
Domanico, P. L.; Han, B.; Rose, D. A.; Rutkowske, R. D.; Sekut,
L.; Stimpson, S. A.; Strickland, A, B.; Verghese, M. W. Introduc-
tion of a conformational switching element on a pyrrolidine ring.
Synthesis and evaluation of (R*,R*)-(()-methyl 3-acetyl-4-[3-
(cyclopentyloxy)-4-methoxyphenyl]-3-methyl-1-pyrrolidinecar-
boxylate, a potent and selective inhibitor of cAMP-specific
phosphodiesterase. J . Med. Chem. 1995, 38, 4972-4975.
(8) Francis, S. H.; Colbran, J . L.; McAllister-Lucas, L. M.; Corbin,
J . D. Zinc interactions and conserved motifs of the cGMP-binding
cGMP-specific phosphodiesterase suggest that it is a zinc hy-
drolase. J . Biol. Chem. 1994, 269, 22477-22480.
In conclusion, we have shown that several analogues
within series 1 and 2 are potent inhibitors of PDE4 and
release of TNFR in HWB, thus raising the potential of
these series for use as anti-TNFR therapy. By overlap-
ping these series with cAMP and taking into account
their ability to bind to a metal in the active site of PDE4,