Manipulation of kinetic profiles in 2-aryl propionic acid cyclooxygenase inhibitors

Article abstract of DOI:10.1016/j.bmcl.2003.11.034

The nonsteroidal anti-inflammatory drugs flurbiprofen and ibuprofen were modified in an attempt to alter the kinetics of inhibitor binding by COX-1. Contrary to prior predictions, a halogen substituent is not sufficient to confer slow tight-binding behavior. Conversion of the carboxylate moiety of flurbiprofen to an ester or amide abolishes slow tight-binding behavior, regardless of halogenation state.

Full text of DOI:10.1016/j.bmcl.2003.11.034

Bioorganic & Medicinal Chemistry Letters 14 (2004) 667–671
Manipulation of kinetic profiles in 2-aryl propionic acid
cyclooxygenase inhibitors
Kushol Gupta,a,b Carl J. Kaub,c Kristen N. Carey,c Eduard G. Casillas,c
Barry S. Selinskyc and Patrick J. Lolla,*
aDepartment of Biochemistry, Drexel University College of Medicine, Philadelphia, PA 19102, USA
bGraduate Group in the Pharmacological Sciences, University of Pennsylvania, Philadelphia, PA 19104, USA
cDepartment of Chemistry, Villanova University, Villanova, PA 19085, USA
Received 2 October 2003; revised 14 November 2003; accepted 17 November 2003
Abstract—The nonsteroidal anti-inflammatory drugs flurbiprofen and ibuprofen were modified in an attempt to alter the kinetics of
inhibitor binding by COX-1. Contrary to prior predictions, a halogen substituent is not sufficient to confer slow tight-binding
behavior. Conversion of the carboxylate moiety of flurbiprofen to an ester or amide abolishes slow tight-binding behavior,
regardless of halogenation state.
# 2003 Elsevier Ltd. All rights reserved.
1. Introduction
significant, as such behavior can convert a micromolar
inhibitor into a psuedo-irreversible nanomolar inhi-
The cyclooxygenase enzymes (COX-1/2, EC 1.14.99.1)
catalyze important early steps in the biosynthesis of the
prostanoid hormones. These hormones mediate a variety
of physiological processes, including inflammation and
platelet aggregation, and have been implicated in pathol-
ogies such as cardiovascular disease, Alzheimer’s disease,
and colon cancer.1ꢀ3 Both isoforms of this enzyme possess
two distinct activities, catalyzing both cyclooxygenase and
peroxidase reactions. The cyclooxygenase active site is the
target of the nonsteroidal antiinflammatory drugs
(NSAIDs).4 Considerable effort has been invested in the
identification of isoform-selective cyclooxygenase inhibi-
tors; such molecules offer the promise of modulating
clinically significant physiological processes such as
inflammation (COX-2) or platelet aggregation (COX-1),
while minimizing unwanted side effects.
bitor.8 The slow tight-binding phenomenon appears to
underlie the isoform specificity of many COX-2 selective
inhibitors; selective compounds are slow tight-binding
inhibitors of the target isoform and reversible competi-
tive inhibitors of the other isoform.9ꢀ11 While a wealth
of structure–activity data has identified structural
modifications of NSAIDS that confer isoform selectiv-
ity, there is little structural understanding of the phe-
nomenon of time-dependent inhibition.
Early work by Rome and Lands drew attention to the
carboxylic acid moiety possessed by many NSAIDs as a
potential determinant of time-dependent inhibition.7
The free carboxylic acid group found in NSAIDs such
as flurbiprofen and ibuprofen forms critical interactions
with residues Arg-120, Glu-524, and Tyr-355 within the
cyclooxygenase active site,12ꢀ14 and esterification of this
group converts such compounds from time-dependent
to time-independent inhibitors. Alteration of the car-
boxylic acid moiety has recently been exploited to con-
vert nonselective inhibitors into COX-2 selective
inhibitors.15ꢀ17 Rome and Lands also hypothesized,
based on a comparison of the chemical structures of
related inhibitors, that a halogen substituent should
confer time-dependent inhibition upon an NSAID.
However, little evidence has been available to test this
hypothesis.
Most NSAIDs can be classified as either slow tight-
binding inhibitors (time-dependent binders) or rever-
sible
competitive
inhibitors
(time-independent
binders).5ꢀ7 Slow tight-binding behavior is clinically
Keywords: NSAIDS; Slow tight-binding inhibitors; Flurbiprofen;
Ibuprofen.
* Corresponding author. Tel.: +1-215-762-7706; fax: +1-215-762-
4452; e-mail: pat.loll@drexel.edu
0960-894X/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2003.11.034

668
K. Gupta et al. / Bioorg. Med. Chem. Lett. 14 (2004) 667–671
Table 1. Inhibition of cyclooxygenase-1 by flurbiprofen derivatives
Two members of the 2-aryl propionic acid class of
NSAIDs were used to probe structure–activity relation-
ships relevant to inhibitor binding kinetics. Flurbiprofen,
a halogenated, slow tight-binding inhibitor, potently inhi-
bits cyclooxygenase activity in both isoforms. Ibuprofen,
while similar in structure to flurbiprofen, lacks any halo-
gen substituent and is a reversible competitive inhibitor of
both isoforms.10,18 Crystal structures are available for
flurbiprofen in complex with both COX-114 and COX-2,13
and for ibuprofen in complex with COX-1.14 The struc-
ture–activity findings are reported here and discussed in
the context of available structural information.
Compd
R1
R2
IC50 (mM)a
Time
Dependent?
1 (flurbiprofen)
2
3
4
5
6
7
8
9
10
F
H
F
F
F
OH
OH
OCH3
NH2
NH(CH3)
N(CH3)2
OCH3
NH2
NH(CH3)
N(CH3)2
0.012ꢃ0.004
0.032ꢃ0.002
210ꢃ28
8.3ꢃ2
Yes
Yes
No
No
No
No
No
No
No
No
24ꢃ6
COX-1 enzyme was purified from ovine seminal vesicles
as described previously;19 purity, as estimated from
Coomassie-stained SDS PAGE, was greater than 95%.
Cyclooxygenase activity was measured at 22 ꢁC using a
coupled cyclooxygenase-peroxidase assay, in which the
oxidation of the reducing peroxidase cosubstrate
N,N,N0,N0-tetramethyl-1,4-phenylenediamine (TMPD)
was monitored after the addition of arachidonic acid.20
Assay conditions were as follows: 0.1 M Tris–HCl pH
8.0, 120 mM TMPD, and 80 nM hematin-reconstituted
enzyme; reactions were initiated with 60 mM arachi-
donic acid. Because COX-1 auto-inactivates, initial rate
measurements were used in all experiments. IC50 mea-
surements were carried out by pre-warming aliquots of
enzyme to 37 ꢁC, adding inhibitor, and incubating at
37 ꢁC for an additional 5 min. Samples were then cooled
on ice and activity measurements taken. In order to
classify inhibitors as time-dependent or time-indepen-
dent, time-course experiments were carried out in which
enzyme was pre-warmed to 37 ꢁC, inhibitor was added at
time zero, and aliquots were withdrawn for activity mea-
surements at different time points. Ki and kinact values for
1 and 2 were estimated by incubating enzyme (100 nM)
with 120 mM TMPD and different concentrations of
inhibitor at 22 ꢁC for varying times, after which the reac-
tion was initiated by the addition of 60 mM arachidonic
acid. Control experiments using hydrogen peroxide as an
initiator and 250 mM inhibitor showed no peroxidase
inhibition by any of the compounds studied.
F
215ꢃ7
H
H
H
H
28.5ꢃ1
64.5ꢃ5
19.5ꢃ2
4.8ꢃ1
a Standard errors were estimated from two independent trials, each
performed in triplicate.
identical positions and orientations in the cyclooxygen-
ase active site, and no changes in the active site geometry
are apparent which could account for the difference in
potency (Fig. 1). Two alternate positions have been
identified for the fluorine atom of flurbiprofen, corre-
sponding to 180ꢁ rotations of the phenyl ring; the two
positions have approximately equal occupancies. It is
unclear if one or both of these alternate binding modes
contributes to the greater potency of 1 versus 2.
Since 2, like 1, proved to be a time-dependent inhibitor
of COX-1, experiments were undertaken to determine if
removal of the fluorine atom significantly alters the
kinetics of binding. The most general model commonly
used for time-dependent inhibition of COX-1 involves
two reversible steps: first a rapid reversible binding of
the inhibitor, followed by a second, much slower inac-
tivation step (Scheme 1A).24 However, since reversal of
the inactivation step is generally quite slow (k < <k2),
ꢀ2
a considerably simpler model in which kꢀ2ꢂ0 has also
been used (Scheme 1B).7
To examine the contribution of the fluorine atom to the
kinetic properties of the inhibitor flurbiprofen [1, Sigma
Chemical (St. Louis, MO)], the defluorinated analogue,
2-(1,10-biphenyl-4-yl) propanoic acid (2), was synthe-
sized as described previously.21 Compound 2 has pre-
viously been shown to be a potent antispasmotic
agent;22 however, no COX inhibition data were avail-
able for this compound. Racemic preparations of 1 and
2 were used for kinetic analysis, but only the S-enan-
tiomer is expected to bind.12 Contrary to the prediction
of Rome and Lands, removal of the halogen does not
convert flurbiprofen to a reversible competitive inhi-
bitor. However, the fluorine atom does contribute to
potency, 1 being almost three-fold more potent than its
defluorinated analogue 2 (Table 1). A similar increase
in potency is associated with the iodination of the time-
dependent 2-aryl propionic acid NSAID suprofen.23
Recently, high-resolution crystal structures of COX-1
in complex with 1 and 2 have become available21
(Gupta et al., in preparation). The inhibitors adopt
To estimate the parameters associated with the kinetic
model, enzyme can be preincubated with inhibitor, and
then assayed after a time t by addition of a large excess
of substrate ([S]> >Km). The excess of substrate ensures
that any enzyme–inhibitor complexes in rapid equili-
brium with free enzyme will bind substrate and catalyze
its transformation to product. For such a preincubation
experiment, the simple model shown in Scheme 1B pre-
dicts that the fractional remaining enzyme activity will
vary with time as follows:7
ꢀ
ꢁ
Eactive Etotal ꢁ EIꢂ
ꢁkinact½IŠt
Ki þ ½IŠ
ꢀ
¼ exp
ð1Þ
Etotal
Etotal
Ki and kinact values for compounds 1 and 2 were deter-
mined by fitting to eq 1 above and are summarized in
Table 2. The kinetic constants obtained for flurbiprofen
(1) are consistent with previous determinations.6,7,24 The
Ki value for the defluorinated analogue 2 is seven-fold

K. Gupta et al. / Bioorg. Med. Chem. Lett. 14 (2004) 667–671
669
Figure 1.
Scheme 1. Kinetic models for the inactivation of COX by NSAIDs:
(A) General two-step mechanism. Free inhibitor is in rapid equili-
brium with the enzyme-bound form (EI); EI slowly undergoes a tran-
sition to the inactive EI* form. k Is typically < <k2, making the EI
ꢀ2
to EI* transition pseudo-irreversible; (B) Simplified two-step mechan-
ism assuming kꢀ2=0, making k2=kinact
.
Table 2. Kinetic constants for the inhibition of COX-1 by flurbipro-
fen (1) and its defluorinated analogue (2)
Scheme 2. Synthesis of amide derivatives of flurbiprofen.
Compd
Ki (mM)a
kinact (minꢀ1
)
1
2
1.0ꢃ0.5
1.0ꢃ0.4
for kinetic analysis were racemic mixtures.
0.13ꢃ0.03
0.19ꢃ0.02
The methyl ester of flurbiprofen (3) is a weak inhibitor
that displays no time-dependence (Table 1). Interest-
ingly, the methyl ester 7 is over seven-fold more potent
than its halogenated analogue 3. The structural basis for
this is not readily apparent from the crystal structures of
COX-1 bound to 1, 2, and 3. It has recently been shown
that certain carboxylic acid-containing NSAIDs can
bind to COX-2 in an ‘upside-down’ orientation, with the
acidic group interacting with Ser-530, rather than Arg-
120;25 perhaps, in a similar manner, the loss of the fluor-
ine atom allows 7 to flip from the expected orientation
and find a more favorable binding site with its ester
moiety in the upper part of the cyclooxygenase channel.
a Time–velocity curves were obtained at 5 different inhibitor con-
centrations ranging from 5 to 5000 nM. Data from three independent
determinations were combined and simultaneously fit to eq 1 using
GraphPad Prism 4 nonlinear regression software. Asymptotic stan-
dard errors were estimated by the program.
lower than that determined for 1; however, the rate of
inactivation seen with 2 is only about 20% of the rate
seen with 1. Hence, while the fluorine atom might
slightly hinder initial binding within the cyclooxygenase
active site, it appears to enhance the rate of formation
of the EI* complex.
To further probe the structural determinants of slow
tight-binding inhibition in flurbiprofen, derivatives of 1
and 2 in which the carboxylic acid moiety was altered
were synthesized and characterized for their ability to
inhibit cycloxygenase activity in a time-dependent fash-
ion. Methyl esters of 1 and 2 were synthesized as
described previously.14 Amide and amide-substituted
derivatives of 1 and 2 were produced by using thionyl
chloride to generate the acid chlorides, which were then
treated with the appropriate amines to generate the
amides 4–6 and 8–10 (Scheme 2). All preparations used
All of the amide and substituted amide derivatives
examined proved to be reversible competitive inhibitors
of COX-1 (Table 1), with no slow tight-binding character.
Increasing substitution on the amide moiety decreased
potency in the series derived from compound 1. However,
the same pattern of substitution had the opposite effect in
the dehalogenated series derived from 2: potency
increased with increasing steric bulk at the amide moiety.
This result was unexpected, and the structural basis for
this trend is not understood, as no crystal structure is
currently available for a 2-aryl propionamide NSAID

670
K. Gupta et al. / Bioorg. Med. Chem. Lett. 14 (2004) 667–671
complexed with COX. Similar conflicting trends have
been seen with COX-2 inhibitors; for example, increasing
the bulk of the substituents on indomethacin amides
increases COX-2 inhibitory potency,26 whereas increas-
ing substituent size on meclofenac amides reduces affi-
nity for the enzyme.17
Removal of the halogen from the slow tight-binding
inhibitor 1 does not alter its time-dependent character. A
converse approach was also attempted, to determine if the
addition of a halogen to a time-independent inhibitor
could convert it into a slow tight-binding inhibitor. The
reversible competitive inhibitor ibuprofen (11) was nitra-
ted or brominated at the 3-position on the aromatic ring
(analogous to the position of the fluorine substituent in
1), and the products kinetically characterized. The syn-
thetic Scheme is described in Scheme 2. Again, racemic
mixtures were used.
The characteristics of the ibuprofen series are summar-
ized in Table 3. Like the parent compound 11, both the
3-nitro- and 3-bromo-substituents are reversible com-
petitive inhibitors of cyclooxygenase activity. Nitration
reduces potency substantially, increasing the IC50
Scheme 3. Bromination of ibuprofen. Reaction conditions: (a) HNO3,
H2SO4, 0 ꢁC, 0.5 h; (b) CH3OH, H2SO4, rt, overnight; (c) NH4Cl, Fe
powder, CH3OH, reflux, 14 h; (d) NaNO2, CuBr, 0 ꢁC, 1 h; (e) NaOH,
CH3OH, reflux, 5 h.
approximately
60-fold.
However,
bromination
improved the potency 10-fold. This result is surprising,
given the large atomic radius of bromine and the small
size of the constriction (formed by Arg-120, Glu-524,
and Tyr-355) through which the ligand must pass in
order to reach the cyclooxygenase active site. A small
hydrophobic cleft exists within the COX-1 cyclooxy-
genase active site, and is lined by the side chains of Leu-
352, Phe-518, and Ile-523. Based on the crystal structure
of the COX-1-ibuprofen complex, it is possible to model
13 into the cyclooxygenase site with its bromine sub-
stituent pointing into this hydrophobic cleft. The crystal
structure of 11 bound to COX-1 shows that the phenyl
and isobutyl groups of the inhibitor make relatively few
close contacts within the active site, and that the ligand
is somewhat smaller than the actual volume of the
cyclooxygenase active site.14 Therefore, the higher
potency of 13 may stem from its improved com-
plementarity with the COX-1 active site (Scheme 3).
tion of bromine to ibuprofen was sufficient to alter these
inhibitors’ kinetic profiles. Halogenation does con-
tribute significantly to inhibitor potency for compounds
such as 1 and 13; however, this is not generally true, as
the bromo-, chloro-, and iodo- derivatives of the related
NSAID ketoprofen are ineffective towards COX.27
Halogenation can also exert more subtle effects on
NSAID recognition by COX, as is illustrated by the
interplay between the substituents at the 3-position and
the amide group in compounds 4–6 and 8–10.
In contrast, alteration of the carboxylic acid moiety of
2-aryl propionic acid NSAIDs appears to be a depend-
able route to abolishing slow tight-binding binding
behavior. These results recapitulate findings in the
fenamate and indole acetic acid classes of cyclooxygen-
ase inhibitors, where neutralization of the carboxylic
group also leads to reversible competitive inhibitors of
COX-1.15,16,26 The apparent generality of this phenom-
enon is noteworthy, given that at least one member of
the fenamate class of NSAIDs, diclofenac, does not
form an ion pair with Arg-120.25
The results presented in this paper show that halogen-
ation is not a determinant of time-dependent inhibition
in the 2-aryl propionic acid class of NSAIDs. Neither
the removal of fluorine from flurbiprofen nor the addi-
Table 3. Inhibition of cyclooxygenase-1 by ibuprofen derivatives
Acknowledgements
This work was supported by the Fanny Ripple Foun-
dation and NIH grant R01-GM55171 (P.J.L.).
Compd
R3
IC50 (mM)a
Time
dependent?
11 (ibuprofen)
12
13
H
NO2
Br
1.06ꢃ0.13
59ꢃ2.8
No
No
No
References and notes
0.11ꢃ0.06
1. Breitner, J. C. Annu. Rev. Med. 1996, 47, 401.
2. DuBois, R. N. Progress in Experimental Tumor Research
2003, 37, 124.
a Standard errors were estimated from two independent trials, each
performed in triplicate.

K. Gupta et al. / Bioorg. Med. Chem. Lett. 14 (2004) 667–671
671
3. Smith, W. L.; DeWitt, D. L.; Garavito, R. M. Annu. Rev.
Biochem. 2000, 69, 145.
4. Dannhardt, G.; Kiefer, W. Eur. J. Med. Chem. 2001, 36, 109.
5. Morrison, J. F.; Walsh, C. T. Adv. Enz. Relat. Areas Mol.
Biol. 1988, 61, 201.
17. Kalgutkar, A. S.; Rowlinson, S. W.; Crews, B. C.; Mar-
nett, L. J. Bioorg. Med. Chem. Lett. 2002, 12, 521.
18. Laneuville, O.; Breuer, D. K.; DeWitt, D. L.; Hla, T.;
Funk, C. D.; Smith, W. L. J. Pharm. Exp. Ther. 1995,
271, 927.
6. Kulmacz, R. J.; Lands, W. E. M. J. Biol. Chem. 1985, 260,
12572.
19. Mevkh, A. T.; Sud’ina, G. F.; Golub, N. B.; Varfolomeev,
S. D. Anal. Biochem. 1985, 150, 91.
7. Rome, L. H.; Lands, W. E. M. Proc. Natl. Acad. Sci.
U.S.A. 1975, 72, 4863.
8. Marnett, L. J. Prostaglandins and Other Lipid Mediators
2002, 68–69, 153.
20. Kulmacz, R. J.; Lands, W. E. M. In Prostaglandins and
Related Substances, A Practical Approach; Benedetto, C.;
McDonald-Gibson, R. G.; Nigam, S.; Slater, T. F. Eds.;
IRL Press: Washington, DC, 1987; pp 209.
9. Copeland, R. A.; Williams, J. M.; Biannaras, J.; Nurn-
berg, S.; Covington, M.; Pinto, D.; Pick, S.; Trzaskos,
J. M. Proc. Natl. Acad. Sci. U.S.A. 1994, 91, 11202.
10. Gierse, J. K.; Koboldt, C. M.; Walker, M. C.; Seibert, K.;
Isakson, P. C. Biochem. J. 1999, 339, 607.
21. Gupta, K.; Selinsky, B. S.; Kaub, C. J.; Katz, A. K.; Loll,
P. J. J. Mol. Biol. 2003, in press.
22. Blicke, F. F.; Grier, N. J. Am. Chem. Soc. 1943, 65, 1725.
23. Loll, P. J.; Picot, D.; Ekabo, O.; Garavito, R. M. Bio-
chemistry 1996, 35, 7330.
11. Ouellet, M.; Percival, M. D. Biochem. J. 1995, 306, 247.
12. Picot, D.; Loll, P. J.; Garavito, R. M. Nature 1994, 367, 243.
13. Kurumbail, R. G.; Stevens, A. M.; Gierse, J. K.; McDo-
nald, J. J.; Stegeman, R. A.; Pak, J. Y.; Gildehaus, D.;
Miyashiro, J. M.; Penning, T. D.; Seibert, K.; Isakson,
P. C.; Stallings, W. C. Nature 1996, 384, 644.
14. Selinsky, B. S.; Gupta, K.; Sharkey, C. T.; Loll, P. J.
Biochemistry 2001, 40, 5172.
15. Kozak, K. R.; Prusakiewicz, J. J.; Rowlinson, S. W.;
Marnett, L. J. Bioorg. Med. Chem. Lett. 2002, 12, 1315.
16. Kalgutkar, A. S.; Marnett, A. B.; Crews, B. C.; Remmel,
R. P.; Marnett, L. J. J. Med. Chem. 2000, 43, 2860.
24. Callan, O. H.; So, O. Y.; Swinney, D. C. J. Biol. Chem.
1996, 271, 3548.
25. Rowlinson, S. W.; Kiefer, J. R.; Prusakiewicz, J.; Pawlitz,
J. L.; Kozak, K. R.; Kalgutkar, A. S.; Stallings, W. C.;
Kurumbail, R. G.; Marnett, L. J. J. Biol. Chem. 2003,
278, 45763.
26. Kalgutkar, A. S.; Crews, B. C.; Rowlinson, S. W.;
Marnett, A. B.; Kozak, K. R.; Remmel, R. P.; Mar-
nett, L. J. Proc. Natl. Acad. Sci. U.S.A. 2000, 97,
925.
27. Cignarella, G.; Curuzu, M. M.; Grella, G.; Loriga, M.;
Anania, V.; Desole, M. S. Farmaco [Sci] 1983, 28, 187.