Identification and optimisation of a novel series of pyrimidine based cyclooxygenase-2 (COX-2) inhibitors. Utilisation of a biotransformation approach

Article abstract of DOI:10.1016/j.bmcl.2009.02.089

Many years of work have been invested in the identification of potent and selective COX-2 inhibitors for the treatment of chronic inflammatory pain. One issue faced by workers is the balance between the lipophilicity required for potent enzyme inhibition

Full text of DOI:10.1016/j.bmcl.2009.02.089

Bioorganic & Medicinal Chemistry Letters 19 (2009) 4509–4514
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Identification and optimisation of a novel series of pyrimidine based
cyclooxygenase-2 (COX-2) inhibitors. Utilisation of a biotransformation
approach
Paul J. Beswick a,e, Andrew P. Blackaby c, Chas Bountra a, Terry Brown d, Kerry Browning d, Ian B. Campbell d,
d
b
d
d
John Corfield d, Robert J. Gleave a, , Steve B. Guntrip , Richard M. Hall , Sean Hindley , Paul F. Lambeth ,
Fiona Lucas d, Neil Mathews d, Alan Naylor a, Hazel Player d, Helen S. Price a, Phillip J. Sidebottom d,
Nicholas L. Taylor d, Graham Webb d, Joanne Wiseman d
*
a Neurosciences Centre of Excellence for Drug Discovery, GlaxoSmithKline, New Frontiers Science Park, Third Avenue, Harlow, Essex CM19 5AW, United Kingdom
b Biological Reagents and Assay Development, GlaxoSmithKline, New Frontiers Science Park, Third Avenue, Harlow, Essex CM19 5AW, United Kingdom
c Analytical Sciences, GlaxoSmithKline, New Frontiers Science Park, Third Avenue, Harlow, Essex CM19 5AW, United Kingdom
d GlaxoSmithKline, Medicines Research Centre, Gunnels Wood Road, Stevenage, Hertfordshire SG1 2NY, United Kingdom
e Structural Genomics Consortium (SGC), University of Oxford, Old Road Campus Research Building, Old Road Campus, Roosevelt Drive, Headington, Oxford, United Kingdom
a r t i c l e i n f o
a b s t r a c t
Article history:
Many years of work have been invested in the identification of potent and selective COX-2 inhibitors for
the treatment of chronic inflammatory pain. One issue faced by workers is the balance between the lipo-
philicity required for potent enzyme inhibition and the physical properties necessary for drug absorption
and distribution in vivo. Frequently approaches to reduce lipophilicity through introduction of polar
functionality is hampered by highly challenging chemistry to prepare key molecules. We have comple-
mented traditional synthetic chemistry with a biotransformations approach which efficiently provided
access to an array of key target molecules.
Received 28 November 2008
Revised 2 February 2009
Accepted 2 February 2009
Available online 27 February 2009
Keywords:
Cyclooygenase
COX-2
Ó 2009 Elsevier Ltd. All rights reserved.
Pyrimidine
Biotransformation
At the outset of our work the majority of selective COX-2
inhibitors reported were 1,2 diaryl heterocycles, for example,
DUP6971 (1)1 and SC58152 (2)2 (Fig. 1). The core heterocycle
was most commonly a five-membered ring with only a few exam-
ples of six-membered cores being reported. We were interested in
investigating a series of pyrimidines 6. Pyrimidine being the core of
choice as the two nitrogen atoms help to control the lipophilicity of
the final molecule.
Encouraged by this result we decided to investigate further
structure activity relationships (SAR) around this novel template,
initially focusing on modification of, or replacement of the benzyl
group R. The SAR observed were similar in each isomeric series
and in this paper we will focus on the series represented by 6a.
Data from an initial set of compounds is shown in Table 1. All of
these molecules were prepared by the general route depicted in
Scheme 1.
Compounds were prepared by the general route shown in
Scheme 1. Perkin condensation of an appropriately substituted phe-
nylaceticacidwitha benzaldehydegavetheintermediate acid3. The
acids were efficiently converted to the corresponding ketones 4 by a
Curtius reaction, and concomitant chlorination and Vilsmeir formy-
lationgave theversatile chloroaldehyde intermediates 5 whichwere
condensed with amidines to give the final compounds 6.
The first two compounds prepared were the two isomeric
benzyl derivatives (6a and 6b, R = benzyl). Encouragingly both iso-
mers had measurable activity at COX-2, but did not inhibit COX-1.
Substitution of the benzyl unit with hydrophobic groups was
generally well tolerated as illustrated by compounds 6b and 6c.
Heterocyclic replacements were accepted, the optimal activity
residing in the less polar rings (cf. compounds 6h and 6i). Removal
of the benzyl group and replacement with hydrogen or smaller
alkyl entities resulted in a loss of activity (compounds 6d–g), as
did increasing the length of the linker (compound 6j). Encourag-
ingly saturated carbocyclic rings resulted in compounds of
increased potency (compound 6k).
In addition to SAR around the benzyl group we also examined
the effect of replacing the linking methylene group with a hetero-
atom. The molecules were prepared by a modification of the route
shown in Scheme 1 (Scheme 2).
* Corresponding author.
E-mail address: robert.j.gleave@gsk.com (R.J. Gleave).
0960-894X/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2009.02.089

4510
P. J. Beswick et al. / Bioorg. Med. Chem. Lett. 19 (2009) 4509–4514
Table 1
O
O
O
O
COX-2 inhibition and selectivity data for initial set of compounds
S
S
O
S
N
Me
O
CF3
Me
Br
N
S
N
N
R
F
F
F
(1)
(2)
Compound
R
COX-1b IC50
(nM)
COX-2a IC50
(nM)
Selectivityc
>326
Figure 1. Examples of early selective COX2 inhibitors.
6a
>34,000
104
114
277
R1
CO2H
a
CHO
6b
6c
>52,124
>457
>361
CO2H
F
F
R1
R2
R2
>100,000
3
F
R1
R1
6d
6e
6f
H
Me
iPr
NSE
NSE
10,000
NSE
NSE
NSE
1135
NSE
—
—
8.8
—
O
Cl
b
c
6g
CH2CONH2
CHO
R2
R2
6h
6i
>100,000
6541
6323
23
>16
284
5.6
4
5
N
R1
N
N
d
N
R
6j
29,371
2969
5263
12
R2
6
Scheme 1. Reagents: (a) Acetic anhydride, sodium acetate, reflux; (b) (i) SOCl2; (ii)
6k
247
NaN3, toluene; (iii)
D
; (iv) HCl, H2O; (c) POCl3, DMF; (d) amidine, MeCN.
O
O
S
1
2
—
—
870
32,000
2
66
435
485
F
Me
a
IC50 values for inhibition of PGE2 produced by arachidonic acid stimulated COS
cells stably expressing human COX-2 as described in Ref. 7.
N
N
b
IC50 values for inhibition of PGE2 produced by arachidonic acid stimulated COS
N
R
cells stably expressing human COX-1 as described in Ref. 7.
N
R
O
c
IC50 value for COX-1 divided by IC50 value against COX-2.
O
S
F
We were encouraged by the activity seen amongst key mem-
bers of this novel series of inhibitors, however when the more po-
tent examples were studied in vivo poor efficacy was observed
when compared with standard molecules. We had previously ob-
served in other series of COX-2 inhibitors that high lipophilicity of-
ten impaired in vivo efficacy3, and thus focussed the chemistry on
synthesising analogues with reduced lipophilicity. Introducing po-
lar groups around the template proved chemically challenging. We
were particularly interested in hydroxylating the methylene linker
of the original benzyl compound 6a, to give compound 10, as
observations in related series suggested that polar groups were
generally poorly tolerated, but we hypothesized that the potential
for internal hydrogen bonding may ‘mask’ the polarity of the hy-
droxyl group in 10 thus retaining potency. We were initially unable
to prepare this compound by conventional synthetic methodology.
Me
6a
6b
COX-2 IC 50 = 266nM
COX-2 IC50 = 104nM
COX-1 IC50 = >34,000nM COX-1 IC 50 = >100,000nM
The chloroaldehyde 5 was condensed with S-methylisothiourea
to give the 2-methylthiopyrimidine 7, this was oxidised to the cor-
responding sulfone 8, which when treated with a variety of nucle-
ophiles and was readily displaced to give the final compounds 9
activities are shown in Table 2.
Replacing the methylene with a nitrogen linker was generally
well tolerated (6l and 6o), however oxygen and sulfur linkers
resulted in a loss of activity (6m and 6n).

P. J. Beswick et al. / Bioorg. Med. Chem. Lett. 19 (2009) 4509–4514
4511
O
S
O
O
S
O
Me
Me
CHO
a
N
b
Cl
N
SMe
F
F
(5)
(7)
O
S
O
S
O
O
Me
Me
c
N
N
N
SO2Me
N
XR
F
F
(9)
(8)
Scheme 2. Reagents and conditions: (a) S-Methylisothiourea, K2CO3, MeCN, reflux; (b) OxoneÒ, MeOH, H2O, rt; (c) RX, MeCN, rt.
functional group modifications and chiral synthesis.4 Micro-organ-
isms can also be employed to mimic mammalian metabolism (so-
called ‘microbial models of mammalian metabolism5) generating
valuable metabolites to aid drug metabolism studies. In this study,
we employed a whole-cell biotransformation approach looking to
identify micro-organisms capable of hydroxylating the methylene
linker of a key benzyl compound 6a. Conventional synthetic chem-
istry had failed in this regard.
Table 2
COX2 inhibition and selectivity data for compounds with a modified linker
O
O
S
N
N
R
A broad range of bacteria and fungi (>220 taxonomically diverse
organisms selected from the GSK culture collection) were initially
screened for their ability to biotransform compound 6a. Cultures
were grown at 2 ml scale in 24-well plates (7 ml volumes) employ-
ing conditions previously reported.6 Cultures were held on porus
F
Compound
R
COX-1b IC50 (nM)
COX-2a IC50 (nM)
Selectivityc
6l
NHPh
OPh
SPh
N(Me)Ph
NHBenzyl
>100,000
>100,000
>100,000
4340
56
>1785
—
—
51
>145
6m
6n
6o
6p
>4640
>7200
85
TM
beads (Microbank , Prolab Inc., Ontario, Canada) and a single bead
was used to inoculate each 2 ml aliquot of culture medium.
Cultures were incubated on a rotary shaker (250 rpm; 50 mm
throw) at 25 °C (fungi) or 28 °C (bacteria) for three days prior to
>100,000
686
a
IC50 values for inhibition of PGE2 produced by arachidonic acid stimulated COS
cells stably expressing human COX-2 as described in Ref. 7.
IC50 values for inhibition of PGE2 produced by arachidonic acid stimulated COS
cells stably expressing human COX-1 as described in Ref. 7.
IC50 value for COX-1 divided by IC50 value against COX-2.
addition of compound 6a (0.125 mg/ml final concn in 200 ll 50%
b
methanol). Following three days additional incubation, 1 ml sam-
ples were removed and added to 1 ml 0.6% TFA in MeOH, stood
for 1 h, centrifuged and supernatant was subsequently evaluated
by LC–MS. Comparison of these data with those obtained for un-
fed controls allowed an assessment of the extent and nature of
the metabolism of 6a by each organism.
Based on this initial analysis, nine micro-organisms (seven
bacteria and two fungi) were selected for scale up because they
showed significant substrate conversion to products of potential
interest. Scale-up was successfully carried out at the 2 Â 50 ml
scale under conditions that mimicked those shown above. The
biotransformation products were isolated in mg amounts by
c
Direct oxidation using
a variety of reagents failed to give
significant oxidation. We prepared the compound utilising a bio-
transformation approach from the parent 6a (Scheme 3). This is
described in the following section.
Biotransformation approaches have been used to good effect in
the pharmaceutical industry over many years. Enzyme-catalysed
reactions and whole-cell-mediated transformations can often pro-
vide mild and high yielding alternatives to chemical syntheses,
O
O
S
O
O
S
Me
Me
Streptomyces rimosus
C1950
N
N
N
N
OH
F
F
(6a)
(10)
Scheme 3.

4512
P. J. Beswick et al. / Bioorg. Med. Chem. Lett. 19 (2009) 4509–4514
Table 3
Structures and COX-2 inhibition data for phase 1 modifications of compound 6a
O
O
S
F
R
N
N
R1
R2
(11)
a
No.
R
R1
R2b
COX-2 IC50 (nM)
57
Organism
10c
H
—
Streptomyces rimosus C1950
OH
11a
11b
11c
11dc
11e
11fc
11g
11hc
11i
H
—
—
—
—
O
43
Mortierella isabellina 4278E
Beauvaria bassiana C2592
Streptomyces sp. 4789E
OH
OH
H
NSEb
NSE
NSE
717
NSE
NSE
NSE
NSE
NSE
OH
H
OH
Beauvaria bassiana C2592
Streptomyces sp. 4789E
OH
H
O
H
—
—
—
—
—
Streptomyces mashuensis 5221I
Streptomyces mashuensis 5221I
Streptomyces armentosus C2162
Streptomyces griseus 4321E
Streptomyces armentosus C2162
OH
OH
OH
H
OH
OH
U
U
U
OH
OMe
H
OH
OH
11j
H
OH
a
IC50 values for inhibition of PGE2 produced by arachidonic acid stimulated COS cells stably expressing human COX-2 as described in Ref. 7.
b
c
No effect observed up to a concentration of 100
Relative or undefined stereochemistry.
lM.

P. J. Beswick et al. / Bioorg. Med. Chem. Lett. 19 (2009) 4509–4514
4513
Table 4
Structures of phase 2 modifications
O
O
S
R
N
N
R1
F
(12)
No.
R
R1
Organism
OH
O
HO
OH
O
12a
H
Streptomyces sp. 4789E
HO
O
CH3
O
O
12ba
12ca
12da
H
H
H
Actinomyces sp. 3986E
OH
HO
O
OH
OH
O
O
Streptomyces rimosus C1950
OH
HO
O
OH
O
O
Streptomyces rimosus C1950
HO
OH
OH
OH
O
O
OH
HO
O
12ea
H
Streptomyces eurythermus 5014I
OH
OH
O
HO
HO
CH3
O
12fa
H
Streptomyces sp. 4789E
(continued on next page)

4514
P. J. Beswick et al. / Bioorg. Med. Chem. Lett. 19 (2009) 4509–4514
Table 4 (continued)
No.
R
R1
Organism
OH
O
OH
HO
O
12ga
OH
Streptomyces griseus 4321E
OH
O
HO
O
12ha
H
Streptomyces mashuensis 5221I
O
HO
HO
HO
a
Relative or undefined stereochemistry.
preparative HPLC and characterised by a combination of electro-
spray MS, 1H NMR spectroscopy and, when necessary, 2D NMR
data.
properties, however such species have potential utility in the
identification of in vivo metabolites arising from future pharmaco-
dynamic studies with lead molecules from this series.
In addition to the target compound 10, an array of additional
analogues was identified (represented by general structures: 11
and 12. Many of these compounds would have been extremely
challenging to prepare and included hydroxylation of the benzyl
group (11a, 11b, 11i and 11j)) and both C and N oxidation of the
core pyrimidine ring (11c, 11e and 11g) together with some chem-
ically very interesting transformations of the benzyl group (11d,
11f and 11h). Modified compounds can be divided into so-called
phase 1 (Table 3) and phase 2 (Table 4) products. It was often
the case that multiple micro-organisms produced the same prod-
uct however only representative species are shown.
In summary we have prepared a novel series of COX2 inhibitors,
early examples of this series were highly potent and selective but
devoid of in vivo efficacy. We hypothesised that the high lipophil-
icity of these molecules was contributing to an unfavourable
pharmacokinetic profile and focussed on preparing more hydro-
philic compounds, as apart of this exercise we employed a
biotransformations approach which identified a highly potent
and selective inhibitor with an excellent in vivo profile.
References and notes
1. Gans, K. R.; Galbraith, W.; Roman, R. J.; Haber, S. B.; Kerr, J. S.; Schmidt, W. K.;
Smith, C.; Hewes, W. E.; Ackerman, N. R. J. Pharmacol. Exp. Ther. 1990, 254, 180.
2. Seibert, K.; Zhang, Y.; Leahy, K.; Hauser, S.; Maferrer, J.; Perkins, W.; Lee, L.;
Isakson, P. Proc. Natl. Acad. Sci. 1994, 91, 12013.
3. Beswick, P.; Bingham, S.; Bountra, C.; Brown, T.; Browning, K.; Campbell, I.;
Chessell, I.; Clayton, N.; Collins, S.; Corfield, J.; Guntrip, S.; Haslam, C.; Lambeth, P.;
Lucas, F.; Mathews, N.; Murkit, G.; Naylor, A.; Pegg, N.; Pickup, E.; Player, H.; Price,
H.; Stevens, A.; Stratton, S.; Wiseman, J. Bioorg. Med. Chem. Lett. 2004, 14, 5445.
4. Roberts, S. M. Biotransformations: Preparative Organic Chemistry; Academic Press,
1989.
5. Rosazza, J. P.; Smith, R. V. Arch. Biochem. Biophys. 1974, 161, 551.
6. Hall, R. M.; Dawson, M. J.; Jones, C. A.; Roberts, A. D.; Sidebottom, P. J.; Stead, P.;
Taylor, N. L. J. Antiobiot. 2001, 54, 948.
7. Bingham, S.; Beswick, P. J.; Bountra, C.; Brown, T.; Campbell, I. B.; Chessell, I. P.;
Clayton, N.; Collins, S. D.; Davey, P. T.; Goodland, H.; Gray, N.; Haslam, C.;
Hatcher, J. P.; Hunter, A. J.; Lucas, F.; Murkitt, G.; Naylor, A.; Pickup, E.; Sargent,
B.; Summerfield, S. G.; Stevens, A.; Stratton, S. C.; Wiseman, J. J. Pharmacol. Exp.
Ther. 2005, 312, 1161.
Encouragingly the target compound 10 was a highly potent and
selective COX-2 inhibitor (no activity at COX-1 up to 100
lM) and
was progressed into
a model inflammatory pain (Freund’s
Complete Adjuvant (FCA)—induced arthritis model).7 When
administered orally the compound showed similar activity to stan-
dard compounds in the same model (ED50 = 1.6 mg/kg).
Whilst the majority of the phase 1 products showed little or no
activity in the COX-2 enzyme inhibition assay we believe the is due
the increased polarity this observation is consistent with previ-
ously observed SAR amongst COX-2 inhibitors which demonstrates
that the enzyme is generally not tolerant of polar ligands.
Data for the phase 2 products is shown in Table 4. None of the
compounds had any significant activity in the COX-2 inhibition
assay, again this is consistent with their physico-chemical