Hydantoins, triazolones, and imidazolones as selective non-hydroxamate inhibitors of tumor necrosis factor-α converting enzyme (TACE)

Article abstract of DOI:10.1016/j.bmcl.2007.02.076

We have discovered selective and potent inhibitors of TACE that replace the common hydroxamate zinc binding group with a hydantoin, triazolone, and imidazolone heterocycle. These novel heterocyclic inhibitors of a zinc metalloprotease were designed using

Full text of DOI:10.1016/j.bmcl.2007.02.076

Bioorganic & Medicinal Chemistry Letters 17 (2007) 2769–2774
Hydantoins, triazolones, and imidazolones as selective
non-hydroxamate inhibitors of tumor necrosis factor-a
converting enzyme (TACE)
James E. Sheppeck, II,* John L. Gilmore, Andrew Tebben, Chu-Biao Xue,
Rui-Qin Liu, Carl P. Decicco and James J.-W. Duan
Bristol-Myers Squibb Pharmaceutical Research Institute, PO Box 4000, Princeton, NJ 08543-4000, USA
Received 18 January 2007; revised 23 February 2007; accepted 26 February 2007
Available online 3 March 2007
Abstract—We have discovered selective and potent inhibitors of TACE that replace the common hydroxamate zinc binding group
with a hydantoin, triazolone, and imidazolone heterocycle. These novel heterocyclic inhibitors of a zinc metalloprotease were
designed using a pharmacophore model that we previously described while developing hydantoin and pyrimidinetrione (barbiturate)
inhibitors of TACE. The potency and binding orientation of these inhibitors is discussed and they are modeled into the X-ray crystal
structure of TACE and compared to hydroxamate and earlier hydantoin TACE inhibitors which share the same 4-[(2-methyl-4-
quinolinyl)methoxy]benzoyl P10 group.
Ó 2007 Elsevier Ltd. All rights reserved.
Tumor necrosis factor-a (TNF-a) is a cytokine that is
produced by activated monocytes and macrophages,
and plays a central role in immunity and inflammation.1
Binding proteins or antibodies that intercept TNF-a
(infliximab, adalimumab, etanercept) block its ability
to activate the TNF-a receptor and have become highly
effective treatments for rheumatoid arthritis, Crohn’s
disease, and psoriasis.2 One of the many strategies to
downregulate the activity of TNF-a is the use of small
molecules to inhibit its convertase, TNF-a converting
enzyme (TACE, ADAM-17). TACE is a membrane
bound zinc metalloprotease that is primarily responsible
for proteolytically processing the 26 kDa membrane-
bound form of pro-TNF-a to its 17 kDa soluble form
that is secreted from cells.3
but often suffer from intrinsically poor pharmacokinet-
ics or metabolic liabilities.5 Over the past several dec-
ades, numerous zinc-ligating, non-hydroxamate
chemotypes have been incorporated into small mole-
cules with the goal of finding potent and druglike inhib-
itors of zinc metalloproteases (Fig. 1).6
Because TACE inhibitors logically started from the bet-
ter studied matrix metalloprotease (MMP) inhibitors,
virtually all of them contain the ubiquitous hydrox-
amate and more recently, the reverse hydroxamate zinc
binding groups (ZBGs; Fig. 1).4 Inhibitors containing
these bidentate ligands generally have excellent potency
Keywords: TACE; TACE inhibitor; TNF-a; MMP; Matrix metallo-
protease; Non-hydroxamate.
*
Corresponding author. Tel.: +1 609 252 3211; fax: +1 609 252
3993; e-mail: jim.sheppeck@bms.com
Figure 1. Precedented ZBGs found in zinc metalloprotease inhibitors.
0960-894X/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2007.02.076

2770
J. E. Sheppeck et al. / Bioorg. Med. Chem. Lett. 17 (2007) 2769–2774
In our quest to find more ‘druglike’ inhibitors of TACE
and other zinc metalloproteases, we discovered that hyd-
antoins and barbiturates make for excellent replace-
ments of hydroxamates.7 Extensive SAR and modeling
of the hydantoins elucidated a pharmacophore that con-
sists of:
mational analysis and their adherence to the hydantoin
model.
Hydantoin 3 (Scheme 1) was synthesized starting from
2-hydroxymethylaniline by coupling it to our preferred
P10 group to form an amide. Dess–Martin periodinane
oxidation of the alcohol to the aldehyde followed by
(1) A cyclic urea that coordinates the active site zinc,
putatively in monodentate fashion via its carbonyl.
One of the urea N–H groups must have a low pKa
for hydrogen bonding to (or deprotonation by) a
conserved active site glutamate general base and
the other N–H is required for a H-bond to a prox-
imal Gly amide carbonyl.
(2) A linker that conformationally induces a U-turn
between the ZBG and the P10 group.
(3) An H-bond acceptor in the linker capable of inter-
acting with mainchain Gly and Leu amide N–H.
(4) An appropriately functionalized P10 group that
complements the metalloprotease S10 subsite.
Having validated this model with highly constrained
hydantoins and barbiturates, we sought to prospectively
identify new non-hydroxamate ZBGs using the above
criteria as structural prerequisites. With these design
constraints in place, both triazolones and imidazolones
(Fig. 2) can achieve the same hydrogen bonding compli-
mentarity as do the hydantoins and barbiturates in the
active site of TACE as their enol tautomers.
Scheme 1. Reagents and conditions: (a) 4-[(2-methyl-4-quinoli-
nyl)methoxy]benzoyl chloride, DCM; 10% NaHCO3 (88%); (b)
Dess–Martin periodinane, 1:1 DCM/DMF (100%); (c) (NH4)2CO3,
KCN, EtOH/H2O, 80 °C sealed tube (10%).
The immediate problem of transposing a hydantoin
TACE inhibitor into a triazolone or imidazolone was
the trajectory of the P10 side chain. Since one of the pre-
requisites for activity in the hydantoins is a U-shaped
turn to engage main chain Leu348 and Gly349 in
H-bonds and angle the P10 appropriately, an even
greater turn would be required for the sp2 triazolones/
imidazolones to compensate for the absent hydantoin
5-position chiral center. Since we had reliable models
of both hydroxamates in TACE (later confirmed by an
X-ray co-crystal structure8) and hydantoins in TACE,7b
all that remained to do was search for optimal bridging
groups between the triazolone/imidazolone ZBG and
the 4-[(2-methyl-4-quinolinyl)methoxy]phenyl P10 side
chain. Molecular modeling was used to prioritize the
extensive number of potential linkers based on confor-
Scheme 2. Reagents and conditions: (a) HCl, EtOH (84%); (b)
semicarbazideÆHCl, N-methylmorpholine, DMF, 120 °C, 12 h (54%);
(c) 5% Pd/C, H2, MeOH, THF, water (100%); (d) 4-[(2-methyl-4-
quinolinyl)methoxy]-benzoic acid, POCl3, pyridine (76%).
Scheme 3. Reagents and conditions: (a) 2-nitrobenzoyl chloride,
AlCl3, nitrobenzene (45%); (b) 5% Pd/C, H2, MeOH (98%); (c) 4-[(2-
methyl-4-quinolinyl)methoxy]-benzoic acid, POCl3, pyridine (6%).
Figure 2. Pharmacophore elements of heterocyclic ZBGs.

J. E. Sheppeck et al. / Bioorg. Med. Chem. Lett. 17 (2007) 2769–2774
2771
the Bucherer-Bergs reaction provided compound 3 in
low yield due to significant competing benzoin conden-
sation. Scheme 2 illustrates how triazolone 6 was
prepared. Pinner reaction with nitrile 4 gave an imidate
that condensed with semicarbazide to form triazolone 5.
Palladium-catalyzed hydrogenation of the nitro substi-
tuent followed by coupling to the P10 acid using the
method of Alves et al.9 gave product 6. Imidazolone 9
was prepared by treatment of commercially available
imidazol-2-one hydrogen sulfate with 2-nitrobenzoyl
chloride under Friedel–Crafts acylation conditions
(Scheme 3). Reduction of the nitro group by catalytic
hydrogenation and coupling to the P10 as before pro-
vided compound 9 albeit in very low yield due to the
unreactivity of the aniline. The synthesis of the remain-
ing compounds in Table 1 is straightforward.10
The pharmacophore model described in Figure 2 above
was built upon an extensive hydantoin SAR7a which
provided a logical starting point for triazolone and imi-
dazolone design. Hydantoins 10 and 11, containing ami-
do or sulfonamide side chains, showed good TACE
activity at 29 and 37 nM, respectively. Another active
compound was hydantoin 3 that contains a phenyl lin-
ker which makes a longer turn to the P10 group to give
a potent inhibitor at 14 nM.
Table 1. In vitro potency of selected hydantoins, triazolones, and imidazolones in pTACE, MMP-2, ꢀ3, ꢀ7, ꢀ12, and ꢀ13a
Compound
Structureb
pTACE
IC50, nM
MMP-2
Ki, nM
MMP-3
Ki, nM
MMP-7
Ki, nM
MMP-12
Ki, nM
MMP-13
Ki, nM
Me
H
O
Me
N
IK682
1
29
2050
>3333
>3333
>3333
>3333
2472
141
—
259
—
—
1417
N
HO
O
O
O
H
N
10c
>5025
>5025
782
—
P1'
N
H
N
H
O
O
O
S
O
H
N
11c
37
—
—
—
N
H
P1'
O
N
H
O
H
N
3
14
>4501
—
>6368
—
>5025
—
N
P1'
O
O
H
N
H
O
O
H
N
13c
>1000
439
162
>5025
4120
>5025
P1'
N
H
N
N
H
O
O
S
H
N
N
H
14
P1'
567
—
1000
—
>5025
—
O
O
N
N
N
H
Boc
N
(3R,4S)-7c
O
>3333
H
N
N
H
P1'
P1'
N
H
O
O
O
N
H
H
N
6
15
9
34
1600
9
>3333
>3333
>3333
>4501
>4501
>4501
>6368
>6368
>6368
3611
>6023
>6023
4314
>5025
4314
O
O
O
N
N
H
P1'
N
H
H
N
N
H
N
H
P1'
H
N
O
N
H
a pTACE IC50s are an average of two determinations; MMP Ki data are from a single determination. All compounds are racemic with the exception
of 7.
b The P10 group in all cases is 4-[(2-methyl-4-quinolinyl)methoxy]phenyl.
c These compounds were also >4946 nM for MMP-1 and >2128 nM for MMP-9.

2772
J. E. Sheppeck et al. / Bioorg. Med. Chem. Lett. 17 (2007) 2769–2774
Simply substituting a hydantoin ZBG such as 10 for a
triazolone 13 proved ineffective for the amide side chain
and models of the latter suggest that 13 is unable to
make an acute enough angle to access the S10 site. How-
ever, the same change in ZBG when the side chain incor-
porates a sulfonamide (14) can better achieve the
requisite turn and is 439 nM in pTACE albeit not as
active as the hydantoin analog 11 at 37 nM. A more
acute turn is achieved by the pyrrolidinyl linker rac-
cis-7 which is 3-fold more potent at 162 nM against
TACE. Lastly, a benzyl linker was appended to a triazo-
lone ZBG to make an even sharper albeit longer turn to
provide compound 6 with an IC50 of 34 nM. As previ-
ously observed with hydantoin and barbiturate ZBGs,
the 4-[(2-methyl-4-quinolinyl)methoxy]phenyl P10 group
confers excellent selectivity against other MMPs. It is
interesting to compare triazolone 6 with hydantoin 3
because the added methylene in 6 appears to compen-
sate for the chiral center in 3 to achieve a similar
presentation of the ZBG and P10 group (see Fig. 3b).
Based on the pharmacophore model in Figure 2, we also
wanted to explore an imidazolone. Transposition of the
triazolone 6 into the corresponding imidazolone 15 sur-
prisingly yielded an inactive compound.
Figure 3a shows a model of compound 9 superimposed
with the IK682-TACE crystal structure.12 The carbonyl
of the imidazolone is in close contact with zinc via a
monodentate interaction and strongly hydrogen bonded
to E406 while its N–H forms another hydrogen bond to
mainchain Gly349. The ketone-containing linker may
increase the potency of compound 9 by making an addi-
tional interaction to Leu348 in addition to the pKa
enhancement it confers. The amide side chain forms a
hydrogen bond with Leu348 analogous to the side
chains of the hydantoins such as 10, 11, and 3. Shown
in Figure 3b is an overlay of hydantoin 3, triazolone 6,
and imidazolone 9, and their shared interactions to zinc
and residues in the vicinity of the active site. Clearly vis-
ible is how each unique linking group bridges the ZBG
to the P10 in a low energy conformation. Each molecule
is able to engage Glu406 in a strong H-bond (or be
deprotonated by it) which serves to enhance its interac-
tion to the Zn, especially as an anion. Density functional
computational methods of TACE by Cross et al.13 cal-
culate the theoretical pKa of the active site Glu406 at
5.9. Furthermore, the pKa of a hydroxamic acid (8.9)
upon coordination to the active site Lewis acidic Zn is
reduced to 5.3.
Closer analysis reveals that the pKa of an imidazolone is
approximately 12, far less acidic than any of the other
ZBGs and unlikely to be deprotonated, even with the
assistance of the active site Lewis acidic zinc ion. We
reasoned that replacement of the imidazolone benzyl
linker with a benzoyl linker should have the effect of
lowering the imidazolone pKa by several log units and
further preorganizing the P10 side chain into the optimal
U-shape conformation to access S10. Indeed, compound
9 proved to be the most potent TACE compound in the
series at 9 nM and retained its excellent selectivity.11
Table 2 shows the aqueous pKa’s of ZBGs that we have
successfully incorporated into TACE inhibitors. If the
magnitude of the pKa reduction from the above analysis
holds for the non-hydroxamate ZBGs in Table 2, their
pKa would likewise decrease to approximately the pre-
dicted value of Glu406 and in some cases, below. Gener-
ally, our data suggest that the potency of the
heterocyclic ZBGs roughly correlates with their pKa.
The quest for inhibitors of zinc-dependent metallo-pro-
teases has been a fertile area of research for over 20
Figure 3. (a) Overlay of the modeled binding mode of compound 9, blue, and the crystal structure of IK682, yellow. Hydrogen bonds and metal
interactions between 9 and TACE are denoted by the dashed blue lines. (b) Modeled binding modes of compounds 3, cyan, 6, orange, and 9, blue.
Representative hydrogen bonds between 3 and TACE are indicated with dashed lines.

J. E. Sheppeck et al. / Bioorg. Med. Chem. Lett. 17 (2007) 2769–2774
2773
Table 2. Aqueous pKa’s of ZBGsa
6. (a) Michaelides, M. R.; Dellaria, J. F.; Gong, J.; Holms, J.
H.; Bouska, J. J.; Stacey, J.; Wada, C. K.; Heyman, H. R.;
Curtin, M. L.; Guo, Y.; Goodfellow, C. L.; Elmore, I. B.;
Albert, D. H.; Magoc, T. J.; Marcotte, P. A.; Morgan, D.
W.; Davidsen, S. K. Bioorg. Med. Chem. Lett. 2001, 11,
1553; (b) Schroder, J.; Henke, A.; Wenzel, H.; Brandstet-
ter, H.; Stammler, H. G.; Stammler, A.; Pfeiffer, W. D.;
Tschesche, H. J. Med. Chem. 2001, 44, 3231; (c) Grams,
F.; Brandstetter, H.; D’Alo, S.; Geppert, D.; Krell, H.-W.;
Leinert, H.; Livi, V.; Menta, E.; Oliva, A.; Zimmerman,
G. Biol. Chem. 2001, 382, 1277; (d) Miyata, K.; Kimura,
H.; Ishikawa, T.; Yamamoto, K. PCT Int. Appl.
WO0172718, 2001; (e) Kawamura, N.; Yamashita, T.;
Takizawa, M.; Yoshimura, K. JP2000143650, 2000.
7. (a) Sheppeck, J. E., II; Gilmore, J. L.; Yang, A.; Chen, X.-
T.; Xue, C.-B.; Roderick, J.; Liu, R.-Q.; Covington, M. B.;
Decicco, C. P.; Duan, J. J.-W. Bioorg. Med. Chem. Lett.
2007, 5. doi:10.1016/j.bmcl. 2006.11.089; (b) Sheppeck, J.
E., II; Tebben, A.; Gilmore, J. L.; Yang, A.; Wasserman,
Z.; Decicco, C. P.; Duan, J. J.-W. Bioorg. Med. Chem.
Lett. 2007. doi:10.1016/j.bmcl.2006.11.082; (c) Duan, J. J.-
W.; Lu, Z.; Wasserman, Z. R.; Liu, R.-Q.; Convington,
M. B.; Decicco, C. P. Bioorg. Med. Chem. Lett. 2005, 15,
2970; (d) Duan, J. J.-W.; Chen, L.; Lu, Z.; Jiang, B.;
Asakawa, N.; Sheppeck, J. E., II; Liu, R.-Q.; Covington,
M. B.; Pitts, W.; Kim, S.-H.; Decicco, C. P. Bioorg. Med.
Chem. Lett. 2006. doi:10.1016/j.bmcl.2006.09.048.
ZBG
pKa (aq)
Hydroxamic acid
Barbiturate
Hydantoin
8.9
7.0
9.0
8
1,3,4-Triazol-2-one
Imidazol-2-one
ꢁ12
—
4-Benzoyl-imidazol-2-one
1,3,4-Triazole-2-thioneb
7.0
a pKa’s were experimentally measured with the exception of triazolone
and imidazolone which were taken from comparably substituted
examples reported in Beilstein.
b An account of our triazolethione-based TACE inhibitors is forth-
coming in this journal.
years. Despite this massive effort however, there are no
marketed inhibitors of zinc-dependent metalloproteases
due to poor PK, lack of efficacy, and toxicity. Many
have attempted to develop druglike, non-hydroxamate
zinc metalloprotease inhibitors using ZBGs that have
a strong intrinsic Kd for zinc.14 Such an approach
may be handicapped by promiscuous chelation to met-
als other than zinc (e.g., hydroxamates bind Fe(III)
106- to 1011-fold stronger than Zn(II)).15 The heterocy-
cles described herein have compensated for the re-
duced, intrinsic Kd’s for Zn exhibited by the
bidentate hydroxamates using a functional group with
putatively weaker monodentate metal interactions sup-
plemented by additional hydrogen bonds in the vicinity
of the active site. The 4-[(2-methyl-4-quinolinyl)meth-
oxy]phenyl P10 group maintains good TACE selectivity
across series. Use of these more druglike ZBGs16 with
other P10 groups should find broad applicability to the
inhibition of zinc metalloproteases in the future.
8. Niu, X.; Umland, S.; Ingram, R.; Beyer, B. M.; Yan-Hui,
L.; Sun, J.; Lundell, D.; Orth, P. Arch. Biochem. Biophys.
2006, 451, 43.
9. Alves, L. C.; Almeida, P. C.; Franzoni, L. J.; Juliano, M.
A. Pept. Res. 1996, 9, 92.
10. (a) For the synthesis of 10, 11, 12, see Ref. 7a; (b)
Compounds 13 and 14 were prepared by coupling the P10
sulfonyl chloride or P10 acid chloride to glycine under
Schotten-Baumann conditions. The acid was coupled to
semicarbazideÆHCl (HATU, Et3N, DMF) followed by
treatment with 2 M NaOH overnight to give the triazolone;
(c) (3R,4S)-7 was prepared by coupling the b-amidopyrro-
lidinyl carboxylate (prepared in Duan, J.; Ott, G.; Chen,
L.; Lu, Z.; Maduskuie, T.P.; Voss, M.E.; Xue, C.-B.
WO2001/070673, 2001) with semicarbazide (HATU, Et3N,
DMF) followed by treating with 2 M NaOH overnight
followed by Boc deprotection with TFA/DCM; (d) Syn-
thesis of 15 started from 2-chloro-1-(2-nitrophenyl)etha-
none (Terrier, C.; Mary, A.; Thal, C. Tetrahedron 1994, 50,
6287) which was then treated with KOCN in water at rt for
12 h to form the imidazolone (29% yield). Reduction of the
nitro group (Pd/C, MeOH, H2; 100% yield) followed by
coupling to the P10 acid using the same method as used for
compounds 6 and 9 provided 15.
References and notes
1. (a) Aggarwal, B.; Kohr, W.; Hass, P.; Moffat, B.; Spencer,
S.; Henzel, W.; Bringman, T.; Nedwin, G.; Goeddel, D.;
Harkins, R. J. Biol. Chem. 1985, 260, 2345; (b) Bemel-
mans, M.; vanTits, L.; Buurman, W. Crit. Rev. Immunol.
1996, 16, 1.
2. (a) Braun, J.; van der Heijde, D. Exp. Opin. Invest. Drugs
2003, 12, 1097–1109; (b) Mikuls, T. R.; Moreland, L. W.
Exp. Opin. Pharm. 2001, 2, 75.
3. Black, R. A.; Rauch, C. T.; Kozlosky, C. J.; Peschon, J. J.;
Slack, J. L.; Wolfson, M. F.; Castner, B. J.; Stocking, K.
L.; Reddy, P.; Srinivasan, S.; Nelson, N.; Bolani, N.;
Schooley, K. A.; Gerhart, M.; Devis, R.; Fitzner, J. N.;
Johnson, R. S.; Paxton, R. J.; March, C. J.; Cerretti, D. P.
Nature 1997, 385, 729.
11. While it would be intriguing to see the activity of the benzoyl
linked triazolone version of 6, 5-ketotriazolones are virtu-
ally unknown in the literature and we were not successful in
making it. Also, it is unlikely that the potency enhancement
would mirror compound 9 (a vinylogous acylurea) in
magnitude because its pKa would not benefit as much from
resonance stabilization due to cross-conjugation.
4. (a) Moss, M. L.; White, J. M.; Lambert, M. H.; Andrews,
R. C. Drug Discovery Today 2001, 6, 417; (b) Skotnicki, J.
S.; Martin, J.; Levin, J. I. Curr. Opin. Drug Disc. Devel.
2003, 6, 742.
5. For an excellent review that discusses hydroxamate
shortcomings and non-hydroxamate inhibitors, see (a)
Breuer, E.; Trant, J.; Reich, R. Exp. Opin. Ther. Patents
2005, 15, 253; (b) Skiles, J. W.; Gonnella, N. C.; Jeng, A.
Y. Curr. Med. Chem. 2001, 8, 425; (c) Whittaker, M.;
Floyd, C. D.; Brown, P.; Gearing, A. J. H. Chem. Rev.
1999, 99, 2735.
12. Compounds were modeled in chain A of the TACE crystal
structure 2FV5. Waters were removed from the structure
except for those within the active site that formed >1
direct interactions with the protein. The complex was
minimized with the OPLSAA-2005 force field using the
GB/SA water model as implemented in Macromodel.
´
˚
IK682 and residues within 5 A were allowed to move
´
˚
freely, residues 5–10 A were constrained with a force
´
˚
constant of 200, and residues >10 A were frozen. The
structure was first subjected to 500 steps of SD minimi-
zation followed by TNCG minimization to a gradient

2774
J. E. Sheppeck et al. / Bioorg. Med. Chem. Lett. 17 (2007) 2769–2774
change of <0.05. The resulting minimized structure was
sterics and hydrogen bonding) selected manually. The
complex of the new ligand in TACE was then refined by
the minimization protocol described above.
13. Cross, J. B.; Duca, J. S.; Kaminski, J. J.; Madison, L. M.
J. Am. Chem. Soc. 2002, 124, 11004.
14. Puerta, D. T.; Lewis, J. A.; Cohen, S. M. J. Am. Chem.
Soc. 2004, 126, 8388, and Ref. 6.
15. Farkas, E.; Katzy, B. S. J. Biol. Inorg. Chem. 2004, 9,
307.
used as the starting template for ligand modeling. Mod-
eled ligands were constructed manually by removing the
hydroxamate moiety from IK682 and replacing it with the
appropriate functional group. The ZBGs were assumed to
be in the enol tautomeric form. A conformational search
was then carried out using the conformational search
module of Macromodel allowing only the newly con-
structed functional group to move while freezing the
remaining coordinates. The resulting conformations were
aligned onto IK682 in the TACE/IK682 minimized
complex and the best fitting conformation (as judged by
16. The hydantoin group is found in phenytoin, the barbitu-
rate is found in pentobarbital, and the triazolone is found
in aprepitant.