Hydantoin based inhibitors of MMP13 - Discovery of AZD6605

Article abstract of DOI:10.1016/j.bmcl.2013.05.089

Piperidine ether and aryl piperazine hydantoins are reported as potent inhibitors of MMP13. A medicinal chemistry campaign focused on replacing the reverse hydroxamate zinc binding group associated with historical inhibitors with a hydantoin zinc binding

Full text of DOI:10.1016/j.bmcl.2013.05.089

Bioorganic & Medicinal Chemistry Letters 23 (2013) 4705–4712
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Hydantoin based inhibitors of MMP13—Discovery of AZD6605
b
b
b
b
b
Chris De Savi a,b, , David Waterson , Andrew Pape , Scott Lamont , Elma Hadley , Mark Mills ,
⇑
Ken M. Page b, Jonathan Bowyer b, Rose A. Maciewicz b
a Oncology Innovative Medicines, AstraZeneca R&D Boston, 35 Gatehouse Drive, Waltham, MA 02451, United States
b Respiratory and Inflammation Innovative Medicines, AstraZeneca R&D Molndal, SE-43813 Molndal, Sweden
a r t i c l e i n f o
a b s t r a c t
Article history:
Piperidine ether and aryl piperazine hydantoins are reported as potent inhibitors of MMP13. A medicinal
chemistry campaign focused on replacing the reverse hydroxamate zinc binding group associated with
historical inhibitors with a hydantoin zinc binding group then optimising MMP13 potency, solubility
and DMPK properties whilst maintaining good selectivity over MMP14. A number of high quality candi-
dates were progressed and following rat and dog safety evaluation, AZD6605 (3m) was identified as a
candidate drug.
Received 7 April 2013
Revised 24 May 2013
Accepted 28 May 2013
Available online 10 June 2013
Keywords:
MMP13
Ó 2013 Elsevier Ltd. All rights reserved.
Hydantoin
Zinc binder
Cyp P450
Lead optimisation
Osteoarthritis (OA) is a heterogeneous group of conditions
sharing common pathologic and radiologic features.1 OA is a
non-overtly inflammatory degenerative joint disease with a wide
range of presentations, from asymptomatic to seriously disabling.
It is extremely common and age-related and is often accompanied
by clinical manifestations such as use-related joint pain and loss of
joint movement. OA is one of the last remaining poorly treated
chronic conditions, which has huge health economic impacts.2,3
The only treatments are pain relief (e.g., NSAIDs and COX-II)4 and
surgery.
It is postulated that inhibition of MMP13 will give rise to bene-
ficial clinical effects in OA by blocking the destruction of cartilage,
preventing deterioration of joint integrity and improving joint
mobility.
The objective of the following medicinal chemistry programme
was to identify orally active inhibitors of MMP13 for the treatment
of OA. The candidate drug target profile called for a compound with
potency/DMPK to support once a day oral dosing in man and also
be structurally different from reverse hydroxamate and hydroxa-
mic acid based MMP13 inhibitors.5,10–12
The activity of matrix metalloproteinases (MMPs), a family of
24 zinc-dependent neutral endopeptidases, has been implicated
in both physiological and pathological tissue remodelling,5 as they
are collectively capable of degrading essentially all of the compo-
nents of the extracellular matrix. One of the MMPs, MMP13, also
known as Collagenase 3, is thought to have a major role in degrad-
ing type II collagen,6 which provides cartilage with its structural
integrity. In adults, MMP13 is expressed only in pathological
tissue. Matrix metalloproteinase inhibitors (MMPi’s) have been
studied for many years as possible drugs for prevention of cartilage
degradation but their clinical use has been hampered by severe
musculoskeletal side effects commonly characterized by joint stiff-
ness and pain.7,8 It is still not known which metalloproteinases
contribute to the fibrodysplasia but it is likely to be due to the
combined inhibition of a combination of several critical MMPs
such as MMP9, MMP14.9
Reverse hydroxamate and hydroxamic acid based MMP13
inhibitors have been widely reported in the literature.12 Previous
reverse hydroxamate inhibitors 1a and 2 from AstraZeneca have
shown potent inhibition of cytochrome P450 isoform 3A4
(Cyp3A4)13,14 and also time dependent inhibition (Fig. 1).15 We
have previously reported that the addition of a methyl group adja-
cent to the reverse hydroxamate moiety 1b has been a successful
strategy to reduce Cyp3A4 activity for aggrecanase inhibitors.13,14
Unfortunately, this change also led to significant drops in MMP13
potency for these inhibitors.
Alternative zinc binding groups to the reverse hydroxamate,
such as hydantoin (3a), have been used in-house to deliver potent
MMP12 compounds for the treatment of COPD with significantly
reduced Cyp3A4 liability.16 Hydantoin 3a had an attractive
in vitro profile: moderate MMP13 potency (IC50 21 nM), clean cyto-
chrome P450 isoform profile (>40
2C19 isoforms), moderate solubility (23
microsomes (Clint <10 L/min/mg in human, rat and dog).
l
M in 3A4, 2D6, 1A2, 2C9 and
l
M) and stability in liver
l
⇑
Corresponding author. Tel.: +1 781 839 4682.
E-mail address: chris.desavi2@astrazeneca.com (C. De Savi).
0960-894X/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.bmcl.2013.05.089

4706
C. De Savi et al. / Bioorg. Med. Chem. Lett. 23 (2013) 4705–4712
Cl
O
F
O
S
F3C
N
N
S
N
Cl
R
N
N
O
N
O
N
N
O
OH
2
OH
O
1a R =H
1b R = Me
O
MMP-13
Cyp 3A4
IC50 0.8 nM
IC50 1 µM
MMP-13 IC50 Cyp 3A4 IC50
O
1a
1b
1.7 nM
60 nM
<0.1 µM
10 µM
O
HN
O
N
S
NH
O
O
3a
F3C
MMP-12
MMP-13
Cyp 3A4
IC50 6 nM
IC50 21 nM
IC50 >40 µM
Figure 1. Historical AstraZeneca MMP13 and MMP12 compounds.
We chose to use the MMP12 compound as a start point and
modify the pendant aryl ether to optimize MMP13 potency and
other properties including aqueous solubility and MMP14 selectiv-
ity9 (Table 1).
Interestingly, the fully fluorinated matched pair compound 3n
maintained good MMP13 potency but led to significant increases
in MMP14 selectivity (>1000Â). The tetrafluorinated ether 3l gave
a balance of both excellent MMP13 potency and good MMP14
selectivity (550Â) and reinforced the importance of fluoro substi-
tution to MMP13 potency. The corresponding des-fluorinated
matched pair compound 3o was 60-fold less potent (LLE 7.1 vs
5.4). Replacing OEt (3e) with CH2OMe (3p) led to reductions in
MMP13 potency. A wide variety of substituted benzyl and hetero-
cyclic ethers were also prepared to explore MMP13 SAR, MMP14
selectivity and solubility (Table 2). Although substituted benzyl
ethers gave excellent MMP13 potency (4a–4g), this change also re-
sulted in poor solubility. F-benzyl ether 4f had superior MMP14
selectivity (>2000Â) compared to all other benzyl ethers tested.
Meta-substituted CN derivative 4d maintained very good MMP13
Previous SAR and X-ray structural information from the reverse
hydroxamate series13,14 showed that ortho substitution on the ter-
minal aryl group was not tolerated and led to significant reductions
in MMP13 potency; the 2,4-dichloro phenyl ether 3c is 600Â less
potent compared to 4-Cl phenyl ether 3b (7 lM and 12 nM, respec-
tively). Similarly, meta substitution also resulted in a significant
reduction in MMP13 potency (3i vs 3h, 1.6 M and 8.5 nM). Due
l
to the good potency associated with para-ethyl ether 3e, a wide
variety of alkyl substituted ethers were prepared. A modest in-
crease in MMP13 potency was achieved when replacing OEt with
OCH2CF3 but with no significant change in MMP14 selectivity.
Table 1
MMP13, MMP12, MMP14 potency and solubility data of piperidine ethers 3a–r
O
1
O
R
N
2
4
S
3
H
N
O
O
N
H
O
b
b
d
Compd
Ra
MMP-13 IC50bnM
MMP-12 IC50 (nM)
MMP-14 IC50 (nM)
cLogP
LLEc
Sol
(lM)
3a
3b
3c
3d
3e
3f
3g
3h
3i
3j
3k
3l
3m
3n
3o
3p
3q
3r
CF3
Cl
2,4-Di-Cl
F
OEt
21
5.9
8.9
1.8
1.5
1.8
0.9
1.3
1.8
0.07
1.5
1.5
1.5
2.5
1.8
2.6
2.5
1.8
0.43
À1.2
1.5
5.9
6.5
3.4
6.6
6.5
5.4
7.2
6.6
4.3
6.8
6.1
7.1
5.7
5.6
5.4
6.3
7.2
6.2
23
12
11.5
6980
28.9
18.2
59.1
53.8
8.5
1617
5.4
2.6
1.2
4.7
9.8
71.2
178
1051
8445
23.7
2
3.3
52
15
18
2817
9501
OCF3
OSO2Me
OCH2CF3
3-OCH2CF3
OCH2CH2CF3
OCH2CF2CF3
OCH2CF2CHF2
OCF2CHF2
OCF2CF3
On-Pr
1950
34
3121
4735
553
2220
>10,000
>10,000
110
4.4
32
6.8
<2
CH2OMe
CH2SO2Me
CH2-4-pyridyl
a
b
c
All substituents are para-substituted (4-position) unless otherwise indicated.
For general assay procedures see Ref. 13; values mean of n = 2 experiments.
LLE (Lipophilic Ligand Efficiency)17 values are calculated as pIC50-clogP.
d
Thermodynamic solubility (lM) in 0.1 M phosphate buffer pH 7.4 at constant temperature (25 °C) for 24 h.

C. De Savi et al. / Bioorg. Med. Chem. Lett. 23 (2013) 4705–4712
4707
potency but additionally had improved MMP14 selectivity com-
pared to para CN benzyl ether 4c. Methyl sulfone 4b had the high-
est LLE but the poorest MMP14 selectivity in this subset of
compounds. Due to the overall poor solubility associated with
the benzyl ethers, a variety of pyridyl and other heterocyclic com-
pounds were made to see if improvements in solubility could be
achieved whilst maintaining target potency. The unsubstituted
pyridyls (4h–j) were potent MMP13 inhibitors (high LLE) with
incremental increases in MMP14 selectivity achieved as the nitro-
gen was moved from para to meta to ortho positions. Although the
clogP of these compounds had decreased, only modest increases in
solubility were achieved relative to the phenyl analogs.
Heterocycles 4k and 4l maintained good MMP13 potency and
good LLE, however thiadiazole4l showed reduced MMP14 selectiv-
ity. Trifluoromethyl substituted furan 4m displayed excellent
MMP13 potency and superior MMP14 selectivity (ꢀ2200Â). The
most lipophilic ligand efficient compounds were 4l, 4b and 4h
(LLE 8.1, 8.0 and 7.7, respectively).
A second hydantoin series, piperazine pyrimidine benzyl ethers
5, was also investigated in the lead optimization campaign (Ta-
ble 3). Due to the poor solubility associated with the piperidine
ethers, design in this area focused primarily on low clogP hetero-
cyclic analogues.
The primary advantage of this series was the excellent MMP14
selectivity: all the compounds tested showed at least >1000-fold
selectivity ratios (vs MMP13), with isoxazole 5k being very selec-
tive over MMP14 (26,000Â). The most potent piperazine pyrimi-
dine heterocycles were 5j, 5k, 5l and 5m, all showing sub 10 nM
MMP13 potency and excellent MMP14 selectivity ratios. Isoxazole
5k and thiazole 5m were also the most ligand lipophilic efficient
compounds (LLE 8.4 and 8.1, respectively). Pyrazole 5e and imidaz-
ole 5f were not potent MMP13 inhibitors, however rearranging the
methyl substitution pattern from 1,5-dimethyl pyrazole (5e) to
Nonetheless, 2-pyridyl 4j had a good overall profile; excellent
MMP14 selectivity (ꢀ1500Â) with acceptable solubility (20
lM)
and no inhibition of cytochrome P450 (IC50 >10 M; 3A4, 2D6,
l
1A2, 2C9 and 2C19 isoforms). The pharmacokinetic profiles in rat
and dog showed low to moderate clearance (Clp = 17 and 3 mL/
min/kg, respectively) and good half lives (t1/2 = 2.5 and 2.3 h,
respectively) given the low volume of distribution (Vdss = 0.8 and
0.6 L/kg). Oral bioavailability was moderate to good in rat and
dog (F% = 35 and 51, respectively).
Table 2
MMP13, MMP12, MMP14 potency and solubility data of piperidine ethers 4a–m
O
O
N
S
H
R
O
O
N
O
N
H
O
LLEc
6.4
Sol (
lM)
b
b
Compd
R
MMP-13 IC50 (nM)
1.4
MMP-14 IC50 (nM)
cLogP
4a
671
2.5
<0.8
O
S
4b
4c
1.4
62
0.85
1.9
8.0
7.2
3.1
1.8
O
NC
NC
0.69
116
4d
1.8
915
1.9
6.8
4.4
O
4e
4f
2.7
3.6
1080
7731
2.4
2.6
6.2
5.8
1.2
NV
F
F
4g
0.88
764
2.6
6.4
NV
4h
4i
N
N
1.9
2.9
450
0.99
0.99
7.7
7.5
130
8.4
1282
N
4j
4.7
5.3
1.6
6897
4090
350
0.99
0.83
0.73
7.3
7.5
8.1
20
87
28
N
4k
4l
S
N
S
N
4m
0.82
1858
2.6
6.5
O
F3C
NV = no value.
b
See footnotes from Table 1.
See footnotes from Table 1.
c

4708
C. De Savi et al. / Bioorg. Med. Chem. Lett. 23 (2013) 4705–4712
Table 3
MMP-13, MMP-14, solubility and selected in vitro/in vivo rat PK of aryl piperazine pyrimidine ethers 5a–m
O
R
N
N
N
O
N
S
H
N
O
O
N
H
O
Rat Cle (mL/
min/kg)
t1/2
(h)
Ff
(%)
a
a
b
c
d
Compd
R
MMP-13 IC50
(nM)
MMP-14 IC50
(nM)
cLogP LogD Sol
M)
Human Clint
min/mg)
(
lL/
Rat Clint
(l
L/min/
(
l
106 cells)
5a
5b
F
14.0
3.3
>99,999
53,584
1.7
0.9
2.1
1.6
16
66
16
<2
nt
4.8
1.9
7.5
nt
NC
6.7
18.1
51
O
S
5c
13.8
>10,000
-0.1
1.1
1
nv
O
N
N
5d
5e
544.6
1080
-0.9
-0.1
560
N
N
N
N
5f
73297329
-0.6
-0.1
N
O
N
5g
47.3
0.8
32
S
5h
5i
31.5
17.7
-0.2
-0.2
<0.5
0.7
49
13
N
N
>100,000
>10,000
<2
nt
2.8
nt
13.4
9.4
0.67
3.3
37
nt
O
N
5j
2.8
0.6
1.9
3.1
F3C
O
N
5k
5l
3.8
5.1
8.3
98,907
33,882
>10,000
0.3
1.6
-0.4
<2
7.3
16
<2
3.6
28
F3C
NC
O
2.3
1.2
11
14
6.3
2.6
100
6
S
5m
97.3
0.23
N
a
IC50’s were derived from triplicate measurements whose standard errors were normally <5% in a given assay. Assay to assay variability was within twofold based on the
results of a standard compound; nt = not tested.
b
Thermodynamic solubility (
l
M) in 0.1 M phosphate buffer pH 7.4 at constant temperature (25 °C) for 24 h.
c
d
e
f
Human microsome metabolism intrinsic clearance Clint L/min/mg). Clint reported is mean of three separate experiments; nv = no value.
For procedure see Ref. 18 Clint reported is mean of three separate experiments.
Compounds dosed at 2 mg/kg iv, n = 2 animals.
(l
Compounds dosed at 3–10 mg/kg p.o., n = 4 animals.
1,3-dimethyl pyrazole (5g) resulted in a significant increase in
MMP13 potency (1080 nM and 47 nM, respectively).
erate clearance with relatively high volumes leading to prolonged
elimination half-lives (t1/2 = 5.8, 7.5 and 5.4 h in mouse, rat and
dog, respectively) coupled with excellent oral absorption in all
pre-clinical species dosed (F% 50–100).
The single X-ray crystal structure of compound 5b in MMP13
overlaid with a historical reverse hydroxamate 6 clearly shows
the transposition of both the sulfonamide and pyrimidine motifs
of the molecules (Fig. 2).19 Both compounds are able to form H-
bonds between the sulfonamide oxygens and backbone residues
Leu185 and Ala186. The CN benzyl motif extends further into the
MMP13 S1’ pocket versus the trifluoroethyl ether. Both zinc
binding groups also overlay well, however the N-hydroxyforma-
mide group of 6 chelates Zn2+ in a bidentate manner whereas the
The aryl piperazine hydantoins were relatively stable in both
human microsomes and rat hepatocytes and a selected number
of examples (5a, 5b, 5i, 5j, 5l and 5m) were dosed in vivo to rat.
These selected compounds all showed low to moderate clearance
with Vdss generally within the range of 0.5–2 L/kg leading to mod-
erate iv half-lives, albeit thiazole 5m was highly cleared in rat
which led to poor bioavailability. Compounds 5b and 5l showed
particularly acceptable t1/2 (7.5 and 2.6 h, respectively) which war-
ranted both oral profiling and further investigation in other spe-
cies. Meta cyano benzyl hydantoin 5b showed further good
pharmacokinetic properties in both mouse and dog showing mod-

C. De Savi et al. / Bioorg. Med. Chem. Lett. 23 (2013) 4705–4712
4709
O
H
N
HN
O
O
N
S
O
N
N
5b
N
O
CN
O
N
O
S
N
N
F
HO
O
N
6
N
O
F
F
F
Figure 2. Overlay of bound compounds hydantoin 5b (cyan) and representative reverse hydroxamate 6 (orange) from MMP13 crystal structure complexes. Key hydrogen
bond interactions with active site residues of MMP13 are highlighted in yellow. The catalytic zinc ion (blue) is shown as a sphere.
Figure 3. (a) MMP13 pIC50 versus core substructure (red—piperidine benzyl ethers 4; blue—aryl piperazine benzyl ethers 5); lines attaching direct matched pairs. (b) MMP14
pIC50 versus core substructure (<indicate ‘less than’ out of range measurements in MMP14 pIC50 biochemical assay).
hydantoin 5b forms a monodentate coordination with the Zn2+ ion
through the acidic N, with the additional NH forming a H-bond
with the carbonyl of Ala186.
more, it has good physicochemical properties and excellent ADME
parameters. 3m has very low in vivo clearance in both rat and dog
(0.4 and 0.2 mL/min/kg, respectively) and shows no turnover in
either human microsomes or hepatocytes (additionally no turn-
over is seen in dog hepatocytes). The predicted dose to man (pred.
DTM) assuming that free blood levels at Cmin equal to 3Â MMP13
IC50 is 0.04 mg/kg/day based on a predicted human clearance
0.07 mL/min/kg, Vdss 0.76 L/kg (both estimated using rat/dog
allometry) and Fu 0.05.22 PB-PK modelling gives a pred. DTM
0.06 mg/kg/day.23 The low dose to man is primarily driven by a
A number of benzyl ethers used in the piperazine pyrimidines 5
had been previously used in the piperidine benzyl ethers 4, en-
abling a Matched Molecular Pair Analysis.20 On average, the
MMP-13 potency of the piperazine pyrimidine benzyl ethers 5
was reduced by approximately 0.5 log unit compared to their
piperidine ether analogues (Fig. 3a, mean difference in MMP13
pIC50 = 0.9, standard error of mean = 0.20, N = 9 pairs). Nonetheless
the aryl piperazines were significantly more selective versus
MMP14 (Fig. 3b, mean difference in MMP14 pIC50 >1.8, standard
error of mean = 0.25, N = 9 pairs), presumably due to the benzyl
ether extending deeper into the MMP13 S1’ pocket and causing
negative interactions with the shorter MMP14 S1’ pocket.21
Due to the excellent MMP13 potency, acceptable cross-MMP
selectivity profile and in vitro stability observed in both human
microsomes and hepatocytes, compounds 3m and 5l were selected
for further advanced in vivo profiling (Table 4). Piperidine 3m is a
potent inhibitor of MMP13 (IC50 4.7 nM). It exhibits a 25- to 100-
fold selectivity against MMP3, 8 and 15; 150 to 550-fold selectivity
against MMP16, 14, 19 and ADAM-17; and >2000-fold selectivity
against MMP1, 7, ADAM-9, ADAM-10 and ADAM-TS4.13,14 Further-
long predicted half-life in man (t 122 h). In contrast, furan 5l
½
has a higher pDTM 0.54 mg/kg/day based on a predicted human
clearance of 0.7 mL/min/kg, Vdss 0.9 L/kg, Fu 0.04 and a shorter pre-
dicted half-life in man (t 16 h). Furan 5l has a significantly im-
½
proved MMP14 selectivity profile. Neither compound inhibits the
five most common human cytochrome P450 isoforms at 10
or below, nor is there time dependent inhibition of human P450
3A4 at 50 M.
lM
l
The synthesis of compound 3m is concisely described in
Scheme 1.27,28 Tetrafluoroethoxy phenol 9 was prepared in high
yield from the commercially available bromide 8 via a lithium
halogen exchange reaction followed by quenching the intermedi-
ate boronate ester with H2O2. The resultant phenol was subjected

4710
C. De Savi et al. / Bioorg. Med. Chem. Lett. 23 (2013) 4705–4712
Table 4
Profiles of advanced compounds 3m and 5l28
O
O
F3C
N
O
F
O
O
N
N
N
F
O
F
F
O
O
S
N
H
N
S
O
H
N
O
O
O
N
H
N
H
Parameter
3m
5l
a
MMP13 IC50
4.7 nM
430
>1989/0.6/60/50/547
2
5.1 nM
14871
>1809/32/47/180/7
2.3
MMP 14 selectivitya
MMP 1/2/3/8/19 selectivitya
LogD7.4
FASIF solubilityb
(
l
M)
14.9
26.5
Dose solubility25 (q.d./b.i.d.)
Hu/rat/dog %free
389/194
5/3/5
<2/<2
2545/1018
6/8/8
7/<2
Hu Micsc/Heps Clint
d
e
Cyp Inhib IC50
>10
40
l
M (5/5)
>10
44
lM (5/5)
26
hERG IC50
lM
l
M
Caco-2 (@10
l
M)f
21 Â 10À6 AB/52 Â 10À6 BA
0.4 mL/min/kg/16 h/0.6 L/kg
0.2 mL/min/kg/35 h/0.6 L/kg
64/40
9 Â 10À6 AB/52 Â 10À6 BA
6.3 mL/min/kg/2.6 h/1.3 L/kg
1.3 mL/min/kg/10.6 h/1.2 L/kg
100/51
Rat PKg Cl/ t /Vdss
½
Dog PKh Cl/t /Vdss
½
Rat/Dog/F%
Human pred. PK Cl/t /Vdss
Human pred. DTM Dose (mg/day)
0.07 mL/min/kg/122 h/0.76 L/kg
2.8 q.d.
0.7 mL/min/kg/16 h/0.9 L/kg
38 q.d.
½
Human pred. Free Cmax
(lM)
0.37
0.34
a
For general assay procedures see Ref. 13 IC50 values were derived from at least six measurements whose standard errors were normally <5% in a given assay. Assay to
assay variability was within twofold based on the results of a standard compound.
b
FASIF—Fasted simulated intestinal fluid solubility (lM). See Ref. 24
c
d
e
f
Human microsome metabolism intrinsic clearance Clint
Human hepatocyte metabolism intrinsic clearance Clint
(
l
L/min/mg microsomal protein). Clint reported is mean of six separate experiments.
(
l
L/min/106 cells). Clint reported is mean of six separate experiments.
Inhibition of cytochrome P450 isoforms: 1A2, 2C9, 2C19, 2D6 and 3A4.
Apparent permeability (Papp cm/s  10À6) in human colon carcinoma (CACO2) cells at 10
l
HO
Cl
M/pH7.4. AB—apical to basolateral direction; BA—basolateral to apical direction
(efflux).
g
Compounds dosed at 2 mg/kg iv, n = 2 animals and 5 mg/kg p.o., n = 4 animals.
Compounds dosed at 0.5 mg/kg iv, n = 4 animals and 1 mg/kg p.o., n = 4 animals.
h
F
F
F
b
a
O
Br
O
F
O
c
N
boc
O
OH
F
F
F
F
F
N
boc
F
F
F
10
9
8
O
F
O
O
F
N
d
F
S
O
O
F
O
NH
H
N
O
S
F
O
O
F
F
O
S
HN
O
F
12
NH
N
11
3m
H
O
Scheme 1. Reagents and conditions: (a) n-Buli and triisopropyl borate, À50 to À20 °C then H2O2, 85%; (b) triphenylphosphine, DTAD, DCM, 64%; (c) 4 M HCl, dioxane, 70%; (d)
DIPEA, THF, À10 °C to rt, 51%.
to a Mitsunobu reaction with commercially available N-Boc pipe-
ridinol to deliver compound 10 which was then deprotected with
HCl to deliver piperidine 11 in quantitative yield. The coupling of
chirally pure sulfonyl chloride hydantoin 1228 and piperidine 11
in the presence of DIPEA successfully furnished compound 3m in
51% yield after chromatography as a white solid.
The effect of compound 3m on cartilage loss was evaluated by
magnetic resonance imaging (MRI) in the Dunkin Hartley guinea
pig model of spontaneous osteoarthritis.29 The guinea pig sponta-
neous model of osteoarthritis is considered to mimic aspects of
articular cartilage degradation in humans.30 Briefly, guinea pigs
knee were imaged in vivo by high-resolution three-dimensional
(3D) MRI at 9 months of age. Animals were randomised into either
vehicle (20% DMSO, 80% PEG, n = 6), compound 3m (11 mg/kg/day,
n = 7), or the MMP13 reference compound31(0.06 mg/kg/day in
20% DMSO, 60% PEG, 20% water, n = 6) previously shown to be effi-
cacious in this study. Compounds were administered continuously
by subcutaneous osmotic mini-pump implanted 7 days post initial
MRI imaging. A second MRI image was obtained approximately
12 weeks later and image analysis performed to assess quantita-
tive volumetric changes of the medial tibial cartilage. PK analysis
confirmed that 3m reached sufficient exposure to provide free
cover over guinea pig MMP13 between 1- and 5.6-fold IC50, respec-
tively, at termination. MRI results, shown in Figure 4 demonstrate
vehicle treated animals lost ꢀ23% medial tibial cartilage volume. In
contrast, both reference compound31 and 3m treatment resulted in

C. De Savi et al. / Bioorg. Med. Chem. Lett. 23 (2013) 4705–4712
4711
G. Cumming and David A. Scott for technical suggestions and
proof-reading the manuscript.
Inhibition of Cartilage Loss
30
25
20
15
10
5
(n)=6
Supplementary data
Supplementary data (experimental procedures and character-
isation data for compounds 3m, 5b and 5l) associated with this
article can be found, in the online version, at http://dx.doi.org/
10.1016/j.bmcl.2013.05.089.
76.3% Inhib
(p<0.001)
References and notes
(n)=6
104% Inhib
1. Altman, R.; Asch, E.; Bloch, D.; Bole, D.; Borenstein, K.; Brandt, W.; Christy, W.;
Cooke, T. D.; Greenwald, R.; Hochberg, M.; Howell, D.; Kaplan, D.; Koopman,
W.; Longley, S.; Mankin, H.; McShane, D. J.; Medsger, T.; Meenan, R.; Mikkelsen,
W.; Moskowitz, R.; Murphy, W.; Rothschild, B.; Segal, M.; Sokoloff, L.; Wolfe, F.
Arthritis Rheum. 1986, 29, 1039.
(p<0.01)
(n)=7
0
2. Helmick, C. G.; Felson, D. T.; Lawrence, R. C.; Gabriel, S.; Hirsch, R.; KentKwoh,
C.; Liang, M. H.; MaraditKremers, H.; Mayes, M. D.; Merkel, P. A.; Pillemer, S. R.;
Reveille, J. D.; Stone, J. H. Arthritis Rheum. 2008, 58, 15.
3. Symmons, D. P.; Mathers, C. D.; Pfleger, B. ‘The Global Burden of Osteoarthritis
in the Year 2000’ GBD 2000 Working Paper, World Health Organisation,
Vehicle
-5
Reference
Compound 3m
0.06mg/kg/day 11mg/kg/day
Geneva.
Accessed March 2013.
http://www.who.int/healthinfo/statistics/bod_osteoarthritis.pdf.
Figure 4. % Cartilage loss versus compound: Effect of compound 3m (and MMP13
reference compound31) on MRI assessed medial tibial cartilage volume loss in the
guinea pig model of spontaneous OA.
4. Conaghan, P.; Birrel, F.; Burke, M.; Cumming, J.; Dickson, J.; Dieppe, P.; Doherty,
M.; Dziedzic, K.; Francis, R.; Grant, R.; Kell, C.; Latimer, N.; MacGregor, A.;
Naisby, C.; O’Mahony, R.; Oliver, S.; Richards, A.; Underwood, M. Osteoarthritis:
National Clinical Guideline for Care and Management in Adults; NICE, Royal
College of Physicians: London, 2008.
a statistically significant inhibition of cartilage loss (76% and 104%,
respectively).
5. Hu, J.; Van den Steen, P. E.; Sang, Q.-X. A.; Opdenakker, G. Nat. Rev. Drug Disc.
2007, 6, 480.
Compound 3m is free of hERG activity up to 50 lM (hERG index
6. Mitchell, P. G.; Magna, H. A.; Reeves, L. M.; Lopresti-Morrow, L. L.; Yocum, S. A.;
Rosner, P. J.; Geoghegan, K. F.; Hambor, J. E. J. Clin. Invest. 1996, 97, 761.
7. Bigg, H. F.; Rowan, A. D. Curr. Opin. Pharmacol. 2001, 1, 314.
8. Brown, P. D. Expert Opin. Investig. Drug 2000, 9, 2167.
9. Peterson, T. Cardiovasc. Res. 2006, 69, 677.
10. Whittaker, M.; Floyd, C. D.; Brown, P.; Gearing, A. J. H. Chem. Rev. 1999, 99,
2735.
11. Jacobsen, J. A.; Major Jourden, J. L.; Miller, M. T.; Cohen, S. M. Biochim. Biophys.
Acta 2010, 1803, 72.
12. Li, N.-G.; Shi, Z.-H.; Tang, Y.-P.; Wang, Z.-J.; Song, S.-L.; Qian, L.-H.; Qian, D.-W.;
Duan, J.-A. Curr. Med. Chem. 2011, 18, 977.
13. De Savi, C.; Pape, A.; Cumming, J. G.; Ting, A.; Smith, P. D.; Burrows, J. N.; Mills,
M.; Davies, C.; Lamont, S.; Milne, D.; Cook, C.; Moore, P.; Sawyer, Y.; Gerhardt,
S. Bioorg. Med. Chem. Lett. 2011, 21, 1376.
14. De Savi, C.; Pape, A.; Cumming, J. G.; Ting, A.; Smith, P. D.; Tart, J.; Page, K.;
Davies, C.; Lamont, S.; Milne, D.; Moore, P.; Sawyer, Y. Bioorg. Med. Chem. Lett.
2011, 21, 3301.
ꢀ2778) and was found not to be mutagenic in SOS/umu, Ames and
mouse lymphoma screens. Furthermore, no significant activities
were detected at ꢀ100Â free Cmax against 141 additional targets
(unrelated proteases, ion channels, receptors and transporters) in
wider secondary pharmacology screening. Due to the potent
in vivo activity and benign in vitro safety profile, 3m was advanced
into rat 28 day toxicity studies. In rat, there were no mortalities,
significant clinical observations, effects on body weight, food and
water consumption, or significant changes in clinical chemistry
or hematological parameters. Interestingly, it did not cause any
changes to the connective tissue, including fibrodysplasia (the ex-
pected significant change) in this study.
In summary, this lead optimization program delivered several
high quality hydantoins which showed potent MMP13 potency,
good MMP14 selectivity, low in vitro human Clint, which when cou-
pled with low pre-clinical in vivo clearance, confidently suggests
low in vivo human clearance. There is also no cytochrome P450
or time dependent inhibition commonly associated with the his-
torical reverse hydroxamate series. Several strategies were em-
ployed to deliver compounds with acceptable overall solubility
profiles without compromising excellent DMPK properties and
ultimately leading to advanced candidates supporting once-a-day
oral dosing regimens in man.
In conclusion, several novel series of MMP13 hydantoin inhibi-
tors were discovered by replacing the zinc binding reverse
hydroxamate with a hydantoin group. Lead optimisation from an
MMP12 hydantoin hit was undertaken via a variety of modifica-
tions made to the P1’ group leading to three discrete hydantoin
sub-series. Compound 3m was progressed into development for
the treatment of osteoarthritis by AstraZeneca as AZD6605.32
15. Riley, R.; Grime, K.; Weaver, R. Expert Opin. Drug Metab. Toxicol. 2007, 3, 51.
16. (a) Eriksson, A.; Lepisto, M.; Lundkvist, M.; Munck Af Rosenschold, M.;
Zlatoidsky, P. WO 02/074767, 2002; Chem. Abstr.2002, 137, 263031.; (b)
Eriksson, A.; Lepisto, M.; Lundkvist, M.; Munck Af Rosenschold, M.; Zlatoidsky,
P. WO 02/074750, 2002; Chem. Abstr.2002, 137, 247696.
17. Leeson, P. D.; Springthorpe, B. Nat. Rev. Drug Disc. 2007, 6, 881.
18. Houston, J. B.; Carlile, D. J. Drug Metab. Rev. 1997, 29, 891.
19. The X-ray crystal ligand–protein structures of both compounds 5b (PDB
ID = 4JPA) and 6 (PDB ID = 4JP4) have been deposited into the RCSB Protein
data bank (http://www.pdb.org/pdb/home/home.do).
20. Leach, A. G.; Jones, H. D.; Cosgrove, D. A.; Kenny, P. W.; Ruston, L.; MacFaul, P.;
Wood, J. M.; Colclough, N.; Law, B. J. Med. Chem. 2006, 49, 6672.
21. (a) De Savi, C.; Morley, A. D.; Ting, A.; Nash, I.; Karabelas, K.; Wood, C. M.;
James, M.; Norris, S. J.; Karoutchi, G.; Rankine, N.; Hamlin, G.; MacFaul, P. A.;
Ryan, D.; Baker, S. V.; Hargreaves, D.; Gerhardt, S. Bioorg. Med. Chem. Lett. 2011,
21, 4215; (b) De Savi, C.; Morley, A. D.; Ting, A.; Nash, I.; Karoutchi, G.; Page, K.;
Gerhardt, S. Bioorg. Med. Chem. Lett. 2011, 22, 271.
22. McGinnity, D.; Collington, J.; Austin, R.; Riley, R. Curr. Drug Metab. 2007, 8, 463.
23. Willmann, S.; Schmitt, W.; Keldenich, J.; Lippert, J.; Dressman, J. B. J. Med. Chem.
2004, 47, 4022.
24. (a) Galia, E.; Nicolaides, E.; Hörter, D.; Löbenberg, R.; Reppas, C.; Dressman, J. B.
Pharm. Res. 1998, 15, 698; (b) Dressman, J. B.; Amidon, G. L.; Reppas, C.; Shah, V.
P. Pharm. Res. 1998, 15, 11; (c) Dressman, J. B.; Reppas, C. Eur. J. Pharm. Sci.
2000, 11, S73.
25. Rinaki, E.; Valsami, G.; Macheras, P. Pharm. Res. 1917, 2003, 20.
26. Bridgland-Taylor, M. H.; Hargreaves, A. C.; Easter, A.; Orme, A.; Harmer, A.;
Henthorn, D. C.; Ding, M.; Davis, A.; Small, B. G.; Heapy, C. G.; Abi-Gerges, N.;
Paulsson, F.; Jacobson, I.; Schroeder, K.; Neagle, B.; Albertson, N.; Hammond, T.
G.; Sullivan, M.; Sullivan, E.; Valentin, J.-P.; Pollard, C. E. J. Pharmacol. Toxicol.
Methods 2006, 54, 189.
27. Full experimental for the preparation and analytical data of compounds 3m, 5b
and 5l can be found in Supplementary data.
28. Waterson, D.; Persson, D. J. WO 06/064218, 2006; Chem. Abstr.2006, 145,
83331.
Acknowledgments
We would like to thank Liz Flavell who produced the MMP13
protein, David Hargreaves for crystallization and depositing struc-
tures into RCSB protein data bank, Stefan Gerhardt for crystallogra-
phy determination and Attilla Ting for modeling structures and
general suggestions for the paper. And final thanks to both John

4712
C. De Savi et al. / Bioorg. Med. Chem. Lett. 23 (2013) 4705–4712
29. Tessier, J. J.; Bowyer, J.; Brownrigg, N. J.; Peers, I. S.; Checkley, D. R.; Westwood,
R. F.; Waterton, J. C.; Maciewicz, R. A. Osteoarthritis Cartilage 2003, 11, 845.
30. Bendele, A. M.; Hulman, J. F. Arthritis Rheum. 1988, 31, 561.
31. MMP13 historical internal reference compound: N-hydroxy-N-[(1S)-3-(2-
pyrimidinyl)-1-[[[4-[5-(2,2,2-trifluoroethoxy)-2-pyridinyl]-1-
32. AZD6605 was made available free of charge to UK medical researchers in a
recent agreement between the Medical Research Council (MRC) and
AstraZeneca, http://www.pharmatopics.com/2011/12/astrazeneca-grants-uk-
academia-free-access-to-its-patented-compounds/. Accessed March 2013.
piperazinyl]sulfonyl]methyl]propyl]-formamide).