Discovery of biphenylketones as dual modulators of inflammation and bone loss

Article abstract of DOI:10.1016/j.bmcl.2010.07.055

Biphenylketones were identified as novel inhibitors of NFκB activation. Structure-activity studies led to the identification of compound 4c, which had good potency against osteoclasts (IC₅₀ = 0.8 μM), showed oral activity, and was able to compl

Full text of DOI:10.1016/j.bmcl.2010.07.055

Bioorganic & Medicinal Chemistry Letters 20 (2010) 5548–5551
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Discovery of biphenylketones as dual modulators of inflammation and bone loss
b
b
b
Iain R. Greig a, , Emmanuel Coste , Stuart H. Ralston , Rob J. van’t Hof
*
a School of Medical Sciences, Institute of Medical Sciences, University of Aberdeen, Foresterhill, Aberdeen AB25 2ZD, United Kingdom
b Rheumatic Diseases Unit, University of Edinburgh, Molecular Medicine Centre, Edinburgh EH4 2XU, United Kingdom
a r t i c l e i n f o
a b s t r a c t
Article history:
Biphenylketones were identified as novel inhibitors of NFjB activation. Structure–activity studies led to
the identification of compound 4c, which had good potency against osteoclasts (IC50 = 0.8 lM), showed
oral activity, and was able to completely prevent inflammation and bone loss in vivo.
Received 16 May 2010
Revised 13 July 2010
Accepted 14 July 2010
Available online 18 July 2010
Ó 2010 Elsevier Ltd. All rights reserved.
Keywords:
Inflammation
Bone loss
Rheumatoid arthritis
TNF
NF
RANKL
a
j
B
Rheumatoid arthritis (RA) is a chronic inflammatory disease
destruction in in vivo models for postmenopausal osteoporosis
and RA.
Our previous studies have shown biphenylesters such as 1
(Fig. 1) to be active in the prevention of bone loss and to induce
characterized by painful swelling of joints, stiffness, loss of move-
ment and the destruction of articular cartilage and bone. Whilst
the ultimate cause of RA is still not fully understood, various pro-
inflammatory cytokines are known to be instrumental in orches-
trating the complex signal transduction pathways which initiate
and propagate the inflammatory response and tissue destruction.1
apoptosis in cells of the osteoclast lineage by inhibition of NFjB-
and MAP-kinase signalling.2 However, as a labile ester, this was
not a suitable candidate for development. Our initial studies there-
fore focused on the related biphenylsulfonamides, such as 2
(Fig. 1). These were up to 100-fold more potent in vitro than the
esters, but proved of no better potency in our studies on inhibition
of RANKL-induced signalling,5 possibly due to poor cell penetra-
tion. Therefore, we decided to investigate other closely related
classes of compounds, including the biphenylketones.
The most important is believed to be TNFa and the advent of anti-
TNF therapies has revolutionalized the treatment of RA. These
therapies include neutralizing antibodies such as infliximab and
adalimumab, and decoy receptors such as etanercept. However,
all current therapies are expensive biological agents and there is
an urgent requirement to identify inexpensive small molecules to
act against these targets.
Our initial reported synthesis of biphenylketone 4a involved
attack of the lithium salt of a dithiane-protected biphenyl on a si-
lyl-protected tosyl-butylalcohol6 and was not suitable for the prep-
aration of a larger number of derivatives. Our proposed revised
synthesis of biphenylketones involved preparation of 1-(biphe-
nyl)hex-5-en-1-one as an intermediate: using attack of a Grignard,
prepared from 5-bromopentene, on a Weinreb amide. Initial at-
tempts to convert this to the alcohol, via hydroboration, led to the
reduction of the ketone and formation of 3a; testing demonstrated
that this and 4a were equipotent in a number of cell-based assays.
A further complication of RA is the associated bone loss, mediated
through the RANKL signalling pathway, which leads to increased
osteoclast differentiation and activity. We have previously reported
a series of anti-resorptive compounds which prevented bone loss by
inhibition of RANKL signalling.2–4 As both TNF
up-regulation of NF
a
and RANKL cause
j
B, we chose to further investigate the potential
of this class of compounds, to determine if their mode of action
shared a common signalling intermediate that would also make
them active as anti-inflammatory agents. Through these studies
we have identified a novel class of simple small molecules which
are highly active in the prevention of inflammation and bone
O
O
H
S N
O
OH
OH
O
1
2
* Corresponding author. Tel.: +44 1224 555823.
E-mail address: i.greig@abdn.ac.uk (I.R. Greig).
Figure 1. Structures of biphenylester 1 and biphenylsulfonamide 2.
0960-894X/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2010.07.055

I. R. Greig et al. / Bioorg. Med. Chem. Lett. 20 (2010) 5548–5551
5549
As these diols can be readily prepared by attack of the Grignard on an
both ends of the chain gave
a
slight reduction in potency
aldehyde, this suggested a more accessible series of compounds that
could be used as a surrogate system for rapid generation of an SAR
model.
Accordingly, 1-(4-bromophenyl)hexane-1,6-diol was prepared
from 4-bromobenzaldehyde and 5-bromopentene followed by
hydroboration, and was then coupled to various phenylboronic
acids by standard Suzuki coupling methods7 to give a range of
substituted biphenyls for biological testing and SAR generation
(Scheme 1).
These studies on reduced ketones allowed us to identify struc-
tures of interest, for which it was then necessary to synthesize
the corresponding ketone. These were prepared from the respec-
tive 1-(4-bromophenyl)hex-5-en-1-one, and then selectively hyd-
roborated using 9-BBN and sodium perborate8 (Scheme 2).
Hydroboration usually gave a mixture of starting material and ke-
tone when performed under the given conditions.
As in our previous studies on ester and sulfonamide deriva-
tives,3,4 we performed initial in vitro screening of candidate com-
pounds by measuring survival of the macrophage cell line J774.
The unsubstituted derivatives, 3a and 4a were tested in this assay
and found to be moderately potent inhibitors of J774 survival, with
(IC50 = 30 lM), and therefore did further pursue this line of inves-
tigation for their biphenylketone analogues.
To test whether the substitution pattern for the aromatic rings
was important to the activity of the drug, we prepared a range of
derivatives with a variety of substituents.
It was apparent that the pattern generally matched that previ-
ously found for the sulfonamides and esters,3,4 but with a much
smaller substituent dependency (Table 1) and a shallow SAR. Sub-
stitution at the 20 or 40 positions generally had little effect on po-
tency, whilst substitution at the 30 position caused a loss of
activity. Only a 20,40-dihalo substitution pattern gave an increase
in potency: we therefore focused on preparing similar derivatives
for the ketone series, compounds 4a–m. These had potencies clo-
sely matching those of the reduced ketone analogues, compounds
3a–x.
We next investigated the effects of adding substituents onto the
inner ring, and found that again only small increases in potency
could be obtained by addition of methyl or halogen groups
(Table 1).
Studies carried out to probe the effects of further structural
changes were unable to provide any improvements in potency.
Partial rigidification of the alkyl chain by incorporation of an aryl
IC50 values of approximately 22
lM, comparable to the original
ester and sulfonamide, compounds 1 (18
l
M) and 2 (18 M). With
l
the ester derivatives we had found that a butanediol side chain
gave the most potent compounds, and a similar feature was found
for the biphenylketones, where a 5 atom separation between the
ketone/reduced ketone moiety and the terminal hydroxyl resulted
in optimal potency; a one-carbon reduction or extension (3y and
3z, respectively, not shown) resulted in a large decrease in potency
Table 1
In vitro viability assay results for compounds 3a–3x and 4a–4m
OH
O
OH
OH
R1
R1
3
4
R2
R2
(IC50 >75 lM and 50 lM, respectively). We had previously con-
ducted studies on the esters to investigate the effects of chain
R1
R2
IC50, l
Ma
J774
OC
OB
branching and found that addition of a methyl group at either or
3a
3b
3c
3d
3e
3f
3g
3h
3i
3j
3k
3l
3m
3n
3o
3p
3q
3r
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
2-Me
3-Cl
3-F
2-F
H
H
H
2-Me
3-Me
3-Cl
H
22
20
22
18
8
24
>100
22
40
15
24
>100
17
8
13
14
14
28
70
20
8
8
>100
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
>100
—
—
—
>100
>100
>100
—
—
>100
—
—
—
—
—
—
—
4-Br
2-F
4-F
2-F, 4-F
2-Me
3-Me
4-Me
2-OH
4-OH
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
1.3
—
—
—
9
OH
O
a
Br
Br
H
R2
R2
OH
b
OH
Br
R2
4-SMe
3-Cl
4-Cl
OH
c
OH
R1
2-Cl, 4-Cl
4-CF3
4-OCF3
4-OEt
4-NMe2
2-OMe, 4-F
2-Me, 4-F
2-F, 4-F
4-F
R2
3a-x
Scheme 1. Reagents and conditions: (a) 5-bromopentene, Mg, THF (6 h); (b) 0.5 M
NaBH4, BF3ꢀOEt2, diglyme (0 °C, 1 h), 3 M NaOH, 30% H2O2, H2O (rt, 0.5 h), K2CO3;
(c) substituted benzeneboronic acid, (PPh3)4Pd, ethanol, toluene, 2 M Na2CO3
(reflux, 3 h).
3s
3t
3u
3v
3w
3x
4a
4b
4c
4d
4e
4f
4g
4h
4i
4j
4k
4l
8
4-F
H
H
14
17
22
6
8
8
O
O
OMe
a,b
Br
OH
Br
N
Me
R2
R2
2-F, 4-F
2-Cl, 4-Cl
2-F, 4-F
2-F, 4-F
2-F, 4-F
3-F,4-F
2-Me, 4-F
2-OMe, 4-OMe
3-OMe, 4-OMe
2-CF3,4-CF3
2-Cl, 4-F
3-Me
2
0.8
1.5
—
3
—
—
13
14
—
—
—
O
O
d
c
OH
4
7
Br
Br
R2
R2
20
20
50
60
23
15
100
O
H
H
H
H
H
H
e
OH
R1
4a-m
R2
Scheme 2. Reagents and conditions: (a) SOCl2, toluene, (reflux, 18 h); (b)
HN(Me)OMe, pyridine, CHCl3 (18 h); (c) 5-bromopentene, Mg, THF (3 h); (d)
0.5 M 9-BBN in THF (rt, 4 h), NaBO3 (rt, 1 h), H2O (50 °C, 4 h); (e) substituted
benzeneboronic acid, (PPh3)4Pd, ethanol, toluene, 2 M Na2CO3, (reflux, 3 h).
4m
a
Values are means of three experiments as determined by alamar blue viability
assay. J774 = J774 macrophages, OC = osteoclasts, OB = osteoblasts.

5550
I. R. Greig et al. / Bioorg. Med. Chem. Lett. 20 (2010) 5548–5551
ring was expected to provide more information about the active
conformation and ideally give an increase in potency, by reduction
of the number of inactive conformations available to the molecule.
Compounds were prepared by the same methodology as described
for the alkyl derivatives, using 3-bromostyrene to form the Grig-
10. Studies of this compound in the J774 macrophage assay were
complicated by low solubility, and did not permit calculation of
an IC50. However, as compound 10 did not inhibit TNFa signalling,
any activity is presumed not to be against the same target as com-
pounds 3 and 4.
nard: unusually, either of the rigid derivatives 5a (6
(6 M) (Fig. 2) showed any significant difference in potency from
the parent compound 4c. Likewise rigidification of the biphenyl
system, by replacement with a phenanthrene (IC50 = 25 M) or flu-
orene (33 M)) had little effect on potency.
l
M) or 5b
These findings broadly agree with earlier studies,3,4 but the very
flat SAR gives little scope for optimization and the potency is some-
what below that normally desirable for a lead compound, albeit in a
cell-based assay, rather than as assessed by direct inhibition of a
molecular target. A greater improvement in potency was seen in
viability studies on genuine osteoclasts,4 which are dependent on
RANKL-induced signalling for their survival. Compounds 4a–4d
l
l
l
We also investigated the effects of replacement of each of the
phenyl rings with pyridyls. As these are electron-poor, the results
shown in Table 1 led us to expect that these might be well toler-
ated in either ring. However, introduction of a heteroatom in the
and 4f have respective potencies of 9, 2, 0.8, 1.5 and 3 lM
(Table 1). These potencies are still below that which would be ideal
in a lead series, but do have a large therapeutic window, as shown by
their lack of activity against osteoblasts (Table 1), suggesting that
they may have low general cytotoxicity and may have good selectiv-
ity for the putative biological target.
outer ring by way of a 2-pyridyl (40
lM), 3-pyridyl (40 lM), 4-pyr-
idyl (15 M), did not give any meaningful increases in potency.
l
Derivatives with inner ring pyridyls were prepared, using
6-bromonicotinic acid and 5-bromopicolinic acid as starting mate-
rials and the same methodology as shown in Scheme 2, to give
compounds 6a and 6b (Fig. 3). In both cases the addition of the ring
Further use of these compounds as probes in mechanism of
action studies, demonstrated their unusual activity as inhibitors
nitrogen caused a reduction in potency (6a = 55
l
M, 6b = 60
l
M).
of NF
and TNF-induced signalling in mouse macrophages at 10
(Fig 5), whilst compounds 4m and 10 showed no effect even at
50 M. Our proposed mechanism of action is that these com-
pounds inhibit recruitment of components of the signalling
complex to the RANK and TNF receptors, and thus suppress
NF B- and MAP-kinase signalling in osteoclasts and macrophages.
jB signalling. Compounds 4c and 4d inhibited both RANKL-
Attempts to prepare the ketone analogue of 6b were unsuccessful,
with the carbonyl undergoing reduction under the given hydrobo-
ration conditions.
The ketone was reacted with hydroxylamine or O-methylhydr-
oxylamine to give the oximes 7a and 7b (Fig. 4): neither of which
lM
l
a
differed in potency from the parent compound (10
increase in potency (4 M) was seen when the ketone was alkyl-
ated with methylmagnesium bromide to give compound 8a and
a small decrease (18 M) upon similar addition of an ethyl group
lM). A small
j
l
Supporting evidence and further details will be presented in a
separate publication.
l
In the absence of a fully validated molecular target, we chose to
study these compounds directly in our models for osteoporosis and
rheumatoid arthritis. For these studies, the ketone derivatives were
preferred as they lack the complication of a chiral centre.
A range of biphenylketones was selected and tested for their
oral anti-resorptive activity and potential as a treatment for osteo-
porosis, using the ovariectomy-induced bone loss model as de-
scribed previously.4 The results are shown in Figure 6 and show
that 4c was the most effective of the compounds chosen at preven-
tion of bone loss. In the untreated mice, a drop in trabecular BMD
of 34% was seen in comparison with sham controls, in contrast to a
21% decrease in mice treated with 4a, a 10% decrease with 4b, a 3%
decrease with 4f and a 6% increase on BMD in mice treated with 4c.
From this study, we selected compound 4c (ABD328)9 to be fur-
ther investigated for anti-inflammatory activity and for potential
as an anti-arthritic agent, using the collagen-induced arthritis
model as described previously.10 Treatment with compound 4c
(10 mg/kg/day, orally or ip, delivered in corn oil) was started when
joint inflammation became apparent in the first animals (in this
experiment 15 days after injection of collagen), and the experi-
8b (not shown, prepared by a method analogous to Scheme 1,
using 40-bromopropiophenone as the starting material). The cen-
tral hydroxyl group was methylated prior to hydroboration to give
compound 9, which gave a reduction in potency (30 lM). This
suggests a relatively unselective binding interaction and that many
groups capable of acting as hydrogen bond acceptors can interact.
One exception to this is the amide, which we have previously
shown to have poor potency (>100 lM) against macrophages and
osteoclasts.4 Finally, the hydrogen bond acceptor was completely
removed by hydrogenation of compound 3a, to give a derivative
Cl
Cl
OH
O
OH
OH
Cl
Cl
5a
5b
Figure 2. Structures of rigidified derivatives 5a and 5b.
Cl
Cl
O
OH
N
N
OH
OH
Cl
Cl
6a
6b
A 30
25
20
15
10
5
B 2.5
Figure 3. Structures of derivatives 6a and 6b bearing a pyridyl group.
2.0
1.5
OH
F
F
F
1.0
0.5
0
N
Me OH
OH
OH
OH
F
F
F
8a
7a
0
OMe
Vehicle
Vehicle
4c
4d
4c
4d
OH
9
10
Figure 5. Compounds 4c and 4d inhibit both TNF
phosphorylation of B. Mouse macrophages were pretreated for 1 h with
compounds (10 M) or vehicle, stimulated with cytokines for 5 min and
phosphorylation assessed using Western blotting.
a- (A) and RANKL- (B) induced
I
j
Figure 4. Structures of compounds 7a, 8a, 9 and 10 in which the carbonyl has been
further modified.
l
IjB

I. R. Greig et al. / Bioorg. Med. Chem. Lett. 20 (2010) 5548–5551
5551
In a separate study, we also compared compound 4d and its sul-
fonamide analogue 2a (ABD295,3 not shown). In this study the
control group had a final total score of 29, 4d fully prevented the
onset of inflammation with a score of 3, whilst 2a had very little
effect (total score of 17, but a smaller difference from control on
previous days). The results (Fig. 8, shown in Supplementary data)
demonstrate that such sulfonamides are significantly less active
20
10
0
in this model, as matched by their lack of activity in the NFjB-
-10
-20
-30
-40
and MAP-kinase activation assays.5 The results also highlight that
the anti-inflammatory properties may be mediated by a different
mechanism to that responsible for the anti-resorptive properties
in closely related compounds, and that the differing activities are
not separated by the use of macrophages and osteoclasts alone.
However, they do still provide a useful primary screen, especially
when combined with a suitable assay which measures activation
Sham
Ovx
Ovx+ 4a
Ovx+ 4b
Ovx+ 4f
Ovx+ 4c
of the TNF-NF
jB signalling pathway.
11
These results demonstrate that biphenylketones such as 4c
Figure 6. Effects of biphenylketones on ovariectomy-induced bone loss. Each of 4a,
4b, 4c and 4f were administered orally to ovariectomized mice (Ovx) at 20 mg/kg/
day by gavage. Data are expressed as percentage change from sham operated
control animals SD, n = 6–8 per group. BV/TV: bone volume as percentage of total
tissue volume.
show great potential as dual modulators in the treatment of rheu-
matoid arthritis and associated bone loss. Whilst these compounds
are perhaps not ideal in terms of a modest in vitro potency and
structure not obviously drug-like, these compounds do represent
an invaluable starting point in the search for effective small mole-
cule anti-inflammatory agents.
50
45
40
35
30
25
20
15
10
5
Control
4c i.p.
4c oral
Supplementary data
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.bmcl.2010.07.055.
References and notes
1. Smolen, J. S.; Steiner, G. Nat. Rev. Drug Disc. 2003, 2, 473.
2. Idris, A. I.; Mrak, E.; Greig, I. R.; Guidobono, F.; Ralston, S. H.; van‘t Hof, R. J.
Biochem. Biophys. Res. Commun. 2008, 371, 94.
3. Greig, I. R.; Idris, A. I.; Ralston, S. H.; van‘t Hof, R. J. J. Med. Chem. 2006, 49, 7487.
4. van‘t Hof, R. J.; Idris, A. I.; Ridge, S. A.; Dunford, J.; Greig, I. R.; Ralston, S. H. J.
Bone Miner. Res. 2004, 19, 1651.
0
0
3
6
9
12
15
18
21
Time After Onset of Arthritis (days)
5. van‘t Hof, R. J. Unpublished results.
6. Greig, I. R.; Mohamed, A. I.; Ralston, S. H.; van‘t Hof, R. J. GB Patent:
WO2004098582, 2003.
7. O’Brien, P. M.; Ortwine, D. F.; Pavlovsky, A. G.; Picard, J. A.; Sliskovic, D. R.; Roth,
B. D.; Dyer, R. D.; Johnson, L. L.; Man, C. F.; Hallak, H. J. Med. Chem. 2000, 43, 156.
8. Fleming, I.; Maiti, P.; Ramarao, C. Org. Biomol. Chem. 2003, 1, 3989.
9. Greig, I.R.; Ralston, S. H.; van‘t Hof, R. J. U.S. Patent Application, US2008221220,
2008.
Figure 7. Graph showing the sum of the joint inflammation scores for each study
group over a 21-day period following the first signs of inflammation in an arthritis
model: (j) control group, corn oil ip; (h) 4c (10 mg/kg/day, ip); (}) 4c (10 mg/kg/
day, orally); n = 10 animals per group.
ment was terminated 3 weeks later. The animals were scored for
joint inflammation at least three times per week using the scoring
system whereby 0 = normal, 1 = mild inflammation of individual
digits, 2 = moderate redness and swelling and 3 = severe redness
and swelling of entire paw. For each animal, the scores for all four
joints were added (maximal score = 12). Figure 7 shows that the
untreated study group has a final total joint score of 47, whereas
the group treated with 4c orally has a score of 13 and the group
treated with 4c ip has a score of 5.
10. Leung, B. P.; Sattar, N.; Crilly, A.; Prach, M.; McCarey, D. W.; Payne, H.; Madhok,
R.; Campbell, C.; Gracie, J. A.; Liew, F. Y.; McInnes, I. B. J. Immunol. 2003, 170,
1524.
11. Data for 4c (ABD328): White solid; mp 63–65 °C (Et2O/petrol); 1H NMR (CDCl3):
d 1.45 (2H, m), 1.60 (2H, m), 1.80 (2H, m), 3.01 (2H, t, J = 6.5 Hz), 3.68 (2H, t,
J = 6.5 Hz), 7.28 (2H, d, J = 8.2 Hz), 7.30 (1H, s), 7.50 (2H, m) and 8.01 (2H, d,
J = 8.2 Hz); 13C NMR (CDCl3): d 23.9, 25.5, 32.5, 38.6, 62.7, 127.4, 128.0, 129.7,
129.9, 131.9, 133.1, 134.5, 136.3, 138.0, 142.8 and 199.9. Anal. (C18H18Cl2O2): C,
H; MS, m/z: calcd, 337.24; found, 337.28, 339.30 (M, M+2).