The design and synthesis of novel N-hydroxyformamide inhibitors of ADAM-TS4 for the treatment of osteoarthritis

Article abstract of DOI:10.1016/j.bmcl.2011.01.036

Two series of N-hydroxyformamide inhibitors of ADAM-TS4 were identified from screening compounds previously synthesised as inhibitors of matrix metalloproteinase-13 (collagenase-3). Understanding of the binding mode of this class of compound using ADAM-TS1 as a structural surrogate has led to the discovery of potent and very selective inhibitors with favourable DMPK properties. Synthesis, structure-activity relationships, and strategies to improve selectivity and lower in vivo metabolic clearance are described.

Full text of DOI:10.1016/j.bmcl.2011.01.036

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Bioorganic & Medicinal Chemistry Letters 21 (2011) 1376–1381
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
The design and synthesis of novel N-hydroxyformamide inhibitors
of ADAM-TS4 for the treatment of osteoarthritis
Chris De Savi a, , Andrew Pape a, John G. Cumming a, Attilla Ting a, Peter D. Smith a, Jeremy N. Burrows a,
⇑
Mark Mills a, Chris Davies a, Scott Lamont a, David Milne a, Calum Cook a, Peter Moore a,
Yvonne Sawyer a, Stefan Gerhardt b
a Respiratory and Inflammation Research Area, AstraZeneca, Mereside, Alderley Park, Macclesfield, Cheshire SK10 4TG, UK
b Global Structural Chemistry, AstraZeneca, Mereside, Alderley Park, Macclesfield, Cheshire SK10 4TG, UK
a r t i c l e i n f o
a b s t r a c t
Article history:
Two series of N-hydroxyformamide inhibitors of ADAM-TS4 were identified from screening compounds
previously synthesised as inhibitors of matrix metalloproteinase-13 (collagenase-3). Understanding of
the binding mode of this class of compound using ADAM-TS1 as a structural surrogate has led to the dis-
covery of potent and very selective inhibitors with favourable DMPK properties. Synthesis, structure–
activity relationships, and strategies to improve selectivity and lower in vivo metabolic clearance are
described.
Received 17 December 2010
Revised 8 January 2011
Accepted 10 January 2011
Available online 18 January 2011
Keywords:
ADAM-TS4
Ó 2011 Elsevier Ltd. All rights reserved.
MMP-13
Reverse hydroxamates
Aggrecanases
Osteoarthritis
Cyp inhibition
Osteoarthritis (OA) is a chronic disease that results in progres-
sive cartilage degradation which clinically leads to pain and joint
dysfunction.1,2 Currently, this disease affects more than 20 million
patients in the United States and this number is expected to double
by 2020, making OA one of the most common health problems in
this country. It is estimated that up to 90% of people older than
75 are affected by OA and of these almost all have OA in the hands
and a third in the knee. Current pharmacological treatment for
arthritis is limited to steroidal and non-steroidal anti-inflamma-
tory drugs (NSAIDS). Although these drugs provide symptomatic
relief for pain and inflammation in arthritis, they have failed to
modify the progression of the disease.3,4 There is still a massive un-
met medical need to develop disease modifying agents that reduce
or reverse the cartilage destruction associated with OA. Cartilage is
primarily constituted of chondrocytes and an extracellular matrix
that consists of aggrecan, collagen and water. Collagen primarily
renders tensile strength whilst aggrecan is responsible for provid-
ing the elasticity and compressibility associated with cartilage.5
Aggrecan and type-II collagen are degraded by proteolytic enzymes
that have elevated activity in osteoarthritic joints. ‘Aggrecanase-1’
(ADAM-TS4)6 and ‘Aggrecanase-2’ (ADAM-TS5)7 have been identi-
fied as the most likely members of the ADAMTS (a disintegrin and
metalloproteinase with thrombospondin) family to be responsible
for cartilage aggrecan degradation and are attractive drug targets
for intervention in OA.8,9 ADAM-TS5 is responsible for disease pro-
gression in a surgically-induced model of OA in mice10 nonetheless
the relative individual contributions of ADAM-TS4 and ADAM-TS5
proteases to cartilage pathology in human disease is not fully
understood.11 Collagenases are the primary enzymes associated
with type-II collagen cleavage and MMP-13 (collagenase-3) prefer-
entially cleaves type-II collagen. Matrix metalloproteinase inhibi-
tors (MMPi’s) have been studied for many years as possible drugs
for prevention of cartilage degradation but their clinical use has
been limited by severe musculoskeletal side effects commonly
characterized by joint stiffness and pain.12,13 Histopathalogical
findings attributed to fibrodysplasia have been seen in dogs follow-
ing inhibition with a selective MMPi.14 Furthermore, it is widely
believed that aggrecan cleavage precedes collagen cleavage in the
disease state15 which justifies the need to discover potent inhibi-
tors of aggrecanases void of MMP activity.16
⇑
Corresponding author. Tel.: +44 1509644844.
E-mail address: chris.desavi2@astrazeneca.com (C. De Savi).
0960-894X/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2011.01.036

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
C. De Savi et al. / Bioorg. Med. Chem. Lett. 21 (2011) 1376–1381
1377
were transformed into a sub-series which were free of cytochrome
P450 3A4 activity.
F
O
S
N
F
N
N
The ADME profile of di-substituted benzyloxypiperidine 4 was
evaluated further and high metabolic liability was identified in
both in vitro and in vivo systems; microsomal clearance in both
human and rat was high, which was also observed in vivo in the
rat with measured clearance (134 mL/min/kg) significantly in ex-
cess of hepatic liver blood flow. As a consequence short half life
and low oral bioavailability were also observed. Non-radiolabelled
metabolite identification studies suggested the major route of
metabolism occurred via oxidation at the benzylic carbon atom.
Furthermore cysteine conjugation adducts were also identified
which indicated the presence of reactive metabolites. To address
this, alternative linkers were proposed to replace the methylene-
oxy group, using crystal structure information and modelling to
rank the ideas (Table 2).18,19
A focused set of compounds were designed to improve the
ADME properties of the benzyloxypiperidine series and CH2CH2
was identified as the optimal replacement linker. The new linker
was tolerated when either piperidine or piperazine provided the
central ring. Compounds 5 and 6 provide the direct piperazine
and piperidine matched pair to compound 4. Both compounds
demonstrate maintained or improved aggrecanase potency, signif-
icantly reduced clearance, both in vitro and in vivo and acceptable
bioavailability.
N
O
N
Me
OH
O
1
Cl
F
O
S
N
Cl
O
N
N
O
N
Me
OH
O
2
Two related series of N-hydroxyformamide inhibitors of
ADAM-TS4, ‘phenyl piperazines’ represented by compound 1 and
‘benzyloxypiperidines’ represented by compound 2, were identi-
fied from screening compounds previously synthesized as poten-
tial MMP-13 inhibitors.17 Compound 1 was a potent inhibitor of
ADAM-TS4 with acceptable physical, DMPK properties and hERG
activity, but showed moderate inhibition of cytochrome P450 iso-
form 3A4. However the MMP selectivity profile was poor and
moreover, SAR and structural studies (Table 1)18,19 suggested that
gaining the required selectivity within this sub-series would be
very challenging. In contrast, benzyloxypiperidine 2 had poorer
physical properties and hERG inhibition but a much better MMP
selectivity profile. There appeared to be scope to address these is-
sues by making changes to the substitution on the benzyl ring but
all compounds suffered from potent cytochrome P450 isoform 3A4
inhibitory activity leading to concerns about potential drug–drug
interaction.20
Previous attempts to generate crystal structures of our early
inhibitors bound to ADAM-TS4 and ADAM-TS5 were unsuccessful,
however
a high quality structure of the surrogate enzyme
ADAM-TS1, co-crystallised with inhibitors was obtained.22 This
surrogate system was used in conjunction with homology model-
ing of the ADAM-TS4 enzyme to (i) understand binding mode
and key interactions, (ii) guide optimisation of MMP selectivity
within the series, and (iii) rank proposed scaffold changes for likely
ADAM-TS4 activity.
The focus of compound design now turned to modifying the
substituents associated with the terminal phenyl ring of compound
6 to ultimately enhance cross-MMP isoform selectivity (Table 3).
The single X-ray crystal structure of compound 4, bound to the ac-
tive site of ADAM-TS1 (Fig. 1a)23 together with ADAM-TS4 homol-
ogy model was used as the basis of a structure based design
approach. Three key interactions were identified: (i) both oxygen
atoms of the N-hydroxyformamide formed a chelate with the
Zn2+ ion and additionally the formyl oxygen formed a hydrogen
We hypothesised that increasing steric bulk around the
N-hydroxyformamide moiety might disrupt its interaction with
the cytochrome P450 haem group.21 The fully substituted ana-
logues, 3 and 4 were designed to test the hypothesis. The results
show the addition of a methyl group adjacent to the N-hydroxyfor-
mamide does indeed reduce cytochrome P450 3A4 inhibitory
activity (1.4 lM vs >10 lM, compounds 1 and 3, respectively). A
more dramatic improvement was observed for the benzyloxypi-
peridine series; highly potent cytochrome P450 3A4 inhibitors
Table 1
SAR for phenyl piperazine and benzyloxypiperidine compounds
Cl
F
O
S
N
F
R
O
S
N
Cl
F
N
N
R
O
N
N
O
N
O
N
Me
OH
N
Me
OH
O
O
2 R = H
4 R = Me
1 R = H
3 R = Me
a
Compd
R
ADAM-TS4 IC50 (nM)
MMP-1 IC50 (nM)
MMP-13 IC50 (nM)
MMP-14 IC50 (nM)
Cyp 3A4 IC50 (nM)
hERG IC50 (lM)
1
2
3
4
H
H
Me
Me
1.4
1
7.3
3.5
>1100
9700
980
1.7
1.6
60
11
1.4
26
5
21
8
180
7.6
430
<0.1
>10
>10
>10,000
550
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.

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
1378
C. De Savi et al. / Bioorg. Med. Chem. Lett. 21 (2011) 1376–1381
Table 2
Rat in vitro and in vivo data on alternative linkers
Cl
F
O
S
O
N
Cl
A
B
N
N
N
OH
O
L/min/106 cells)
Clc (mL/min/kg)
t1/2 (h)
Fd (%)
a
b
Compd
A
B
ADAM-TS4 IC50 (nM)
Hu Clint
(
lL/min/mg)
Rat Clint
(l
4
5
6
7
O
C
C
C
N
C
C
3.5
3.8
1.1
2200
107
54
6.9
nt
56.9
10.2
7.6
130
14
20
nt
0.3
5.6
14.9
nt
0
30
38
nt
N
nt
a
b
c
Clint reported is mean of two separate experiments; nt = not tested.
For procedure see Ref. 34.
Compounds dosed at 2 mg/kg iv, n = 2 animals.
Compounds dosed at 3–10 mg/kg po, n = 4 animals.
d
Table 3
Aggrecanase, MMP and ADAM inhibitory activities of piperidine ether analogs of 6
F
N
O
S
O
R
N
N
N
H
O
O
a
Compd
R
ADAM-TS4 IC50
(nM)
MMP-1 IC50
(nM)
MMP-13 IC50
(nM)
MMP-14 IC50
(nM)
MMP-2 IC50
(nM)
TACE IC50
(nM)
c log P
8a
2-Chloro-4-
0.48
>100,000
260
900
170
nt
4.9
(trifluoromethyl)phenyl
4-Chloro-2-methyl-phenyl
2-Bromo-4-fluoro-phenyl
2-Chloro-5-fluoro-phenyl
4-Fluoro-2-
8b
8c
8d
8e
0.36
0.48
2.8
>8500
>8000
>9600
9700
370
2300
570
360
390
56
530
1700
500
nt
nt
nt
19
4.5
4.3
4.2
4.3
0.52
16,000
3600
1900
(trifluoromethyl)phenyl
3,5-Dimethylisoxazol-4-yl
4-Fluoro-2-methyl-phenyl
2-Methyl-4-
8f
8g
8h
0.69
0.18
0.49
>19,000
>8100
>10,000
>8100
530
2300
>8900
160
4600
>6300
120
350
250
21
420
1.5
3.9
4.6
(trifluoromethyl)phenyl
2-Cyclopropyl-4-
8i
0.94
>10,000
17,000
18,000
>9100
3200
5.1
(trifluoromethyl)phenyl
2,5-Dimethylphenyl
8j
8k
8l
8m
8n
8o
0.48
0.43
1.9
0.26
0.68
4.2
>10,000
>100,000
55,000
nt
>100,000
>70,000
23,000
14,000
27,000
4600
>4500
>16,000
4200
4200
22
230
1400
nt
3000
27,000
4.3
2.8
2.8
2.9
2.8
3.6
2,5-Dimethyl-4-pyridyl
2,5-Dimethyl-3-pyridyl
3,5-Dimethylisothiazol-4-yl
4,6-Dimethyl-3-pyridyl
3-Chloro-5-(trifluoromethyl)-2-
pyridyl
17,000
>8000
2600
>3700
>18,000
12,000
35,000
2400
2200
2800
8p
3-Methyl-5-(trifluoromethyl)-2-
pyridyl
2.1
>79,000
6100
40,000
2500
24,000
3.3
8q
8r
2-Chloro-4-methylsulfonyl-phenyl
2,5-Dimethylpyrazol-3-yl
0.57
5.4
>70,000
>10,000
350
6700
>3600
29,000
180
26,000
5500
2800
2.5
1.5
a
See footnote from Table 1.
bond interaction with Glu150, (ii) one of the sulfonyl group oxy-
gens formed a hydrogen bond to the backbone NH of Leu118,
and (iii) the 2,4-dichlorophenyl group was involved in a hydropho-
bic interaction with the residues forming the S10 pocket. A further
noteworthy feature of the observed binding model was the appar-
ent stacked interaction between the fluoropyrimidinylmethyl P1
group and N-hydroxyformamide; the plane of the pyrimidine ring
stacked parallel to the plane of the N-hydroxyformamide group.
Further examination of the S10 region of the crystal complex also
indicated that the ortho-chloro atom associated with the 2,4-di-
chloro benzyl group occupied a small defined pocket of hydropho-
bic space. This particular finding was of significant interest as
overlay with the single crystal X-ray structure of compound 4 in
MMP-13 showed that this pocket did not appear to be present in
MMP-13. This observation supported the exploration of further
ortho -substituents in combination with a para-substituent, to
define the SAR related to a potential increase in MMP selectivity.
The importance of an ortho-substituted benzyl ether resulting in

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
C. De Savi et al. / Bioorg. Med. Chem. Lett. 21 (2011) 1376–1381
1379
inhibitors and very selective across all measured MMPs (>5000-
fold with respect to ADAM-TS4). The 2,5-dimethyl substitution
pattern in particular inferred excellent MMP-13 selectivity as well
as MMP-14 and MMP-1 (8j, 8k and 8l). Generally, potency followed
lipophilicity and the SAR was parallel between piperazine (5) and
piperidine (6) sub-series, with the piperidine analogues showing
two log units greater potency for most matched pairs (Fig. 2). Inter-
estingly, although the pyridyl analogues (8k, 8l and 8n) followed
the established c log P relationship, a number of the five-mem-
bered heterocycles showed a left-shifted relationship, that is, in-
creased potency combined with reduced lipophilicity.
Due to the excellent selectivity profiles associated with 8f it was
taken forward for further in vitro and in vivo testing. The pharma-
cokinetic properties26 in rat and dog showed low to moderate
clearance (Clp = 11 and 20 mL/min/kg, respectively) and excellent
half live (t1/2 = 11 and 11.3 h) driven by good volume of distribu-
tion (Vdss = 7.4 and 9.2 L/kg). As a consequence of reduced lipophil-
icity, compound 8f also demonstrated low plasma protein binding
in rat (fu = 0.22) and dog (fu = 0.25). Oral bioavailability was moder-
ate in dog (F% = 20) and mouse (F% = 34) and poor in rat (F% = 3.5%).
However, upon higher oral dosing (50 and 100 mg/kg in DMSO/HP-
b-CD formulation) AUC and bioavailability levels in rat increased
significantly. Caco-2 data suggested good permeability with little
efflux (Papp A:B = 5.3 ꢀ 10ꢁ6; B:A = 10.9 ꢀ 10ꢁ6). Compound 8f
also showed potent inhibition of IL-1 induced aggrecan
(IC50 = 310 nM, n = 3) degradation in bovine nasal cartilage ex-
plants.27 In human osteoarthritic cartilage, compound 8f inhibited
IL-1 induced aggrecan degradation with an IC50 = 243 nM.28
The synthesis of compound 8f is shown in Scheme 1. Chloro-
pyrimidine 10 was synthesized via a Zn-mediated reduction using
known literature methods from dichloropyrimidine 9.29 Com-
pound 10 was subjected to a Pd mediated cyanation30 followed
by MeMgBr Grignard addition to the resultant nitrile to deliver
methyl ketone 11 in 74% yield.
Ketone 11 smoothly underwent a Horner–Wadsworth–Em-
mons reaction to selectively deliver the cis a,b-unsaturated ketone
12 in 45% yield. The first chiral centre was introduced utilising an
asymmetric hydrogenation using the 2S,5S (COD)Rh(EtDuPhos)
catalyst with excellent control of stereochemistry to deliver 1331
in high yield (ꢂ88% ee, 90% yield). A Peterson reaction joined the
chiral methyl ketone 13 with the pre-elaborated methanesulfonyl-
piperazine 14,32 and the resultant product was reacted in concert
with hydroxylamine to generate the second, fully substituted chi-
ral centre. Fortuitously the desired diastereoisomer was formed
preferentially (2.2:1 major/minor diastereomeric ratio). Formyla-
tion of the hindered hydroxylamine 16 was achieved at room tem-
perature following pre-formation of the O-acetyl derivative. After
cleavage of the O-acetyl during work-up the desired diastereomer
8f33 crystallised with very high purity and diastereoisomeric ex-
cess. The route gave 4.9% overall yield with no chiral separation
and has been used to synthesise >200 g of compound 8f for early
safety profiling.
In summary, we have discovered a very potent series of novel
N-hydroxyformamide inhibitors of ADAM-TS4 with excellent
selectivity over related metalloproteinases. The key issue of inhibi-
tion of cytochrome P450 was addressed by sterically hindering the
metal-chelating group. Metabolic instability of the benzyloxy lin-
ker was addressed by replacement with an ethyl linker. Subse-
quent optimisation of the P10 aryl group led to maintenance of
potency with concomitant reduction of lipophilicity. A structure
based design approach was supported with single crystal X-ray
structures, using ADAM-TS1 as a structural surrogate. 3,5-Dimethl-
oxazole 8f showed excellent potency and selectivity combined
with favourable DMPK properties and was advanced into in vivo
safety studies and the guinea pig spontaneous OA PD model.
Figure 1. (a) ADAM-TS1 in complex with bound compound 4. Key hydrogen bond
interactions with active site residues of ADAM-TS1 are highlighted in yellow. The
catalytic zinc ion (purple) is shown as a sphere. Final electron density map (blue
mesh) is contoured at 1.0 s. (b) Overlay of bound compounds 4 and 8f from ADAM-
TS1 crystal structure complexes.
potent aggrecanase activity had also been noted on a series of
5-hydroxypipercolic hydroxamates from Noe et al.24,25
The overlaid conformations of bound compounds 4 and 8f
(Fig. 1b)23 from ADAM-TS1 complexes gave the project team con-
fidence both that the linker change would be tolerated and that
the ortho methyl substituent associated with 8f would access the
exclusive ADAM- TS1 pocket.
All piperidine ether analogs (8a–r) exhibited excellent
ADAM-TS4 potency in the picomolar to nanomolar range with
excellent selectivity achieved compared with MMP-1. The majority
of compounds also showed good selectivity compared with other
MMPs. Small substituents such as F in the meta and para positions
on the pendant aryl P1 group (compounds 8d and 8g) did not
exhibit significant selectivity compared with MMP-14. The ortho
Cl substituted compounds generally showed less selectivity
towards both MMP-13 and MMP-2 (compounds 8a, 8b, 8d and
8q). This interesting observation is further reinforced by compar-
ing the nature of substituent in the ortho position where the para
-substituent is fixed (8a, 8h and 8i), and selectivity improves sig-
nificantly for the ortho cyclopropyl variant (8i). Five membered
heterocycles (8f and 8m) in which both the ortho positions are
flanked with methyl groups were both very potent ADAM-TS4