Design of selective Cathepsin inhibitors

Article abstract of DOI:10.1016/j.bmcl.2009.06.090

A number of molecular recognition features have been exploited in structure-based design of selective Cathepsin inhibitors.

Full text of DOI:10.1016/j.bmcl.2009.06.090

Bioorganic & Medicinal Chemistry Letters 19 (2009) 4622–4625
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Design of selective Cathepsin inhibitors
a
Paul A. Bethel a, Stefan Gerhardt a, Emma V. Jones a, Peter W. Kenny a, , Galith I. Karoutchi ,
*
Andrew D. Morley a, , Keith Oldham , Neil Rankine , Martin Augustin , Stephan Krapp ,
a
a
b
b
*
Hannes Simader b, Stefan Steinbacher b
a RIRA, AstraZeneca, Mereside, Alderley Park, Macclesfield, SK10 4TG, UK
b Proteros Biostructures, Am Klopferspitz 19, D-82152 Martinsried, Germany
a r t i c l e i n f o
a b s t r a c t
Article history:
A number of molecular recognition features have been exploited in structure-based design of selective
Cathepsin inhibitors.
Received 30 April 2009
Revised 20 June 2009
Accepted 23 June 2009
Available online 27 June 2009
Ó 2009 Elsevier Ltd. All rights reserved.
Keywords:
Cathepsin
Cathepsin B
Cathepsin K
Cathepsin L
Cathepsin L2
Cathepsin S
Cathepsin V
Cysteine protease
Nitrile
Covalent inhibitor
Selectity
Protein structure
X-ray crystallography
Structure-based design
Hydrogen bonding
Hot spot
Electrostatic
Hydrogen bond
O
O
Cysteine Cathepsins are implicated in a number of diseases
including cancer, osteoarthritis, osteoporosis, autoimmune disor-
ders and viral infection.1 Selectivity is an important consideration
in design of inhibitors of this class of protease, especially given that
many of these feature an electrophilic warhead, such as a nitrile,
that interacts covalently with the active site cysteine. For instance,
gene knockout studies suggest that Cathepsins B (CatB) and L2
(CatL2) should be considered key anti-targets in optimization of
Cathepsin L (CatL) inhibitors.2–4
N
X
Ph
NH
NH
N
H
H
N
CN
N
CN
O
O
Cl
2 : X = Me 3 : X = Cl
1
Compounds 1–3 were described recently as CatL inhibitors and
the binding mode of 2 has been characterized by X-ray crystallog-
raphy.4 Interaction with Leu69 appears to be a factor in the im-
proved selectivity profiles of 2 and 3 over 1. The equivalent
residue in a number of the other Cathepsins is an aromatic amino
acid. This selectivity was unexpected at the outset because the
molecular surface in the region of the Leu69 side chain is convex.
One rationale for observed selectivity is that Leu69 interacts with
a concave region of the ligand and is held in what we have de-
scribed4 as a ‘molecular pincer’.
* Corresponding authors. Tel.: +44 1625 514396 (P.W.K.).
E-mail addresses: pwk.pub.2008@gmail.com (P.W. Kenny), andy.morley@astra
zeneca.com (A.D. Morley).
0960-894X/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2009.06.090

P. A. Bethel et al. / Bioorg. Med. Chem. Lett. 19 (2009) 4622–4625
4623
highly complementary to the aligned donors. Repulsion between
the aligned nitrogen lone pairs enhances the hydrogen bond accep-
tor strength of each nitrogen atom.13–15 Inspection of the crystal
structure for the complex of 2 and CatL suggested that linking from
C3 of the phenyl ring would provide the best access to the amide
donors. A number of analogs of 2, C3-substituted with heteroaro-
matic rings, were synthesized.
The esters 5 were prepared from commercially available 3-
bromophenylalanine (4). Subsequent hydrolysis followed by amide
coupling with aminoacetonitrile generated the key 3-bromophe-
nylalanine derivatives 6 (Scheme 1). Compounds 7–13 and 15 were
prepared in low to moderate yield by Suzuki coupling with various
commercially available heterocyclic boronic acid pinacol esters.
Protecting groups were present in two of these reagents and these
were removed using standard methods to give 14 and 16.
The boronic acid/esters required for synthesis of 20–22 using
the route described in Scheme 1 were not readily available. These
compounds were prepared via Suzuki coupling of the commer-
cially available heteroaryl triflates/bromides with the boronic acid
pinacol ester 19 (Scheme 2). This racemic intermediate was ob-
tained from 17 by conversion to triflate 18 followed by cross-cou-
pling reaction with bis(pinacolato)diboron.
Compound 23 was prepared as previously described from com-
mercially available (S)-m-tyrosine. Subsequent conversion to tri-
flate followed by hydroxycarbonylation using molybdenum
hexacarbonyl generated the key acid intermediates 24. Standard
HATU coupling with various hydrazides either commercially avail-
able (or prepared) followed by cyclodehydration with either Bur-
gess reagent or Lawesson’s reagent afforded oxadiazoles/
thiazdiazoles 25. Subsequent ester hydrolysis followed by amide
coupling with aminoacetonitrile generated 26–32. Demethylation
of compound 25a with boron tribromide, followed by hydrolysis
and amide formation afforded 33. Formic acid mediated Boc depro-
tection of intermediates 34 and 36 afforded compounds 35 and 37
(Scheme 3).
Cathepsin inhibition profiles are presented in Table 1 for a num-
ber of phenylalanine derivatives with heteroaromatic substituents
at C3. Compounds with 3-pyridazinyl (21) 1,3,4-oxadiazolyl (26,
28), or 1,3,4-thiadiazolyl (27, 29) substituents at C3 were consis-
tently more potent than the corresponding 3-chloro analogues (1,
3) against CatS and CatL2. Compounds 21 and 27 are particularly
potent CatS inhibitors with good selectivity over other family
members, and would be useful start points for a program against
this target.19
Different trends were observed for CatL potency and the pres-
ence of adjacent hydrogen bond acceptors appears to less benefi-
cial than for CatS or CatL2. For example, the CatL pIC50 of 21 is
0.7 units less than that of 1 and 29 is slightly more potent (0.4
units) than 3 against CatL. This suggests that Leu69 may restrict
Figure 1. Crystal structure of 2 bound to Cathepsin L showing molecular surface of
the protein and the locations of key residues.
Other residues that might be targeted to improve selectivity are
Asp71 and Asp114. The corresponding residues in CatL2 are hydro-
phobic amino acids (Ala71; Val114). The crystal structure of 2
bound to CatL (Fig. 1) shows the side chains of Asp71 and
Asp114 to be in mutual contact, suggesting that at least one of
these residues is protonated under crystallization conditions. Pro-
vided that one of these aspartates remains anionic, achieving con-
tact between this region of the protein and a cationic group in an
inhibitor is likely to result in improved selectivity over CatL2.
It is important that the spacer that links this cationic group to
the rest of the molecule makes contact with the surface of the pro-
tein. The backbone amide hydrogen bond donors of Met70 and
Asp71 are of particular interest in this regard. These molecular rec-
ognition elements sit next to each other at the bottom of a pocket
where the molecular surface of the protein is concave. Surface con-
cavity5 and solvent enclosure6 in a binding pocket facilitate dis-
placement of water molecules by ligand. These characteristics
suggest that this region of the protein may be able to function as
a hot spot7,8 for ligand recognition. The presence of exposed polar
atoms deep within a binding pocket contributes to the druggabili-
ty9 of
a target protein especially when fragment-based ap-
proaches10,11 are used. Exploiting these two hydrogen bond
donors was seen as a means to anchor ligands in order to make
more effective contact with the Asp71 and Asp114 residues. This
approach was predicted to generate good selectivity with respect
to CatB as the amino acid corresponding to Met70 is proline, which
lacks a backbone donor.
The amide donors of Met70 and Asp71 reinforce each other be-
cause they are close together and mutually aligned.12,13 An adja-
cent pair of hydrogen bond acceptor nitrogen atoms in
a
heteroaromatic ring is a molecular recognition feature that is
O
O
O
NH2
R
NH
R
NH
R
NH
OH
H
H
O
N
CN
N
CN
a, b
c, d
e
Pyrz =
O
N
O
O
N
O
Br
6
Br
Br
X
4
5
thp
N
N
N
N
N
N
N
N
11 : R = Ph, X =
12 : R = Pyrz; X =
7 : R = Ph, X =
8 : R = Ph, X =
9 : R = Ph, X =
13 : R = Ph, X =
14 : R = Ph, X =
15 : R = Ph, X =
16 : R = Ph, X =
N
N
boc
g
f
N
N
N
N
10 : R = Ph,
X
N
NH
N
N
H
Scheme 1. Reagents and conditions: (a) 10 M HCl/MeOH; (b) acid, HATU, DIPEA, DMF; (c) LiOH, MeOH/H2O/THF; (d) H2NCH2CN, HATU, DIPEA, DMF; (e) het-boronic acid
pinacol ester, 1,1 bis(di-tert-butylphospino)ferrocene palladium dichloride, Na2CO3, dioxane/H2O, 100 °C, MW; (f) formic acid; (g) 1.25 M HCl/EtOH.

4624
P. A. Bethel et al. / Bioorg. Med. Chem. Lett. 19 (2009) 4622–4625
O
O
O
O
Ph
NH
Ph
NH
Ph
NH
Ph
NH
20 : X=
H
H
H
H
N
CN
N
N
CN
N
CN
CN
N
N
b
a
c
O
O
O
O
21 : X=
22 : X=
N
O
B
HO
TfO
X
19
18
O
17
N
N
Scheme 2. Reagents and conditions: (a) PhNTf2, K2CO3, THF, 120 °C, MW; (b) bis(pinacolato)diboron, 1,1 bis(di-tert-butylphospino)ferrocene palladium dichloride, KOAc,
DME, 85 °C; (c) het-OTf/Br, Pd(OAc)2, K3PO4, S-Phos, THF/H2O, 150 °C, MW.
O
O
O
O
R
NH
R
NH
R
NH
R
NH
H
a, b
c, d
O
O
e, f
N
CN
O
O
O
O
O
26 : R = Ph; X = O;Y = H
27 : R = Ph; X = S; Y = H
28 : R = Pyrz; X = O;Y = H
29 : R = Pyrz; X = S; Y = H
30 : R = Pyrz; X = O;Y = Me
31 : R = Pyrz; X = S; Y = Me
32 : R = Pyrz; X = O;Y = CH2 OMe
HO
N
N
N
Y
HO
N
Y
24
X
25
O
X
23
e,f
e,f
25a : R = Pyrz; X = O;Y = CH 2OMe
25b : R = Pyrz; X = O; Y = CH2OH
g
33 : R = Pyrz; X = O;Y = CH2 OH
N
34 : R = Pyrz; X = O;Y = CH2 NHBOC
35 : R = Pyrz; X = O;Y = CH2 NH2
36 : R = Pyrz; X = O;Y = CH2 CH2 NHBOC
37 : R = Pyrz; X = O;Y = CH2 CH2 NH2
N
h
Pyrz
h
Scheme 3. Reagents and conditions: (a) PhNTf2, K2CO3, THF, 120 °C, MW; (b) Mo(CO)6, Pd(OAc)2, dppf, K2CO3, dioxane, 140 °C, MW; (c) hydrazide, HATU, DIPEA, DCM, rt; (d)
burgess reagent, 70 °C, MW or Lawesson’s reagent, 40 °C, THF; (e) 2.5 N aq NaOH, MeOH/H2O, 75 °C; (f) H2NCH2CN, HATU, DIPEA, DMF; (g) BBr3, DCM, rt; (h) formic acid, rt.
interaction of the heterocyclic nitrogen atoms with the backbone
donors to a greater extent than the corresponding Phe residues
in CatS and CatL2. Subtle SAR was observed for these structural
types. For example, 30 is 0.5 log units more potent than 28 against
CatL but 31 is 1.4 log units less potent than 29.
The methoxy (32) and hydroxyl (33) analogues of 30 showed
inhibition profiles that were very similar to that observed for 30.
Compounds 35 and 37, each of which presents a pendant amino
group that is largely protonated under assay conditions, are 0.4
and 0.6 log units more potent than 30 against CatL. Both CatL2
and CatS potencies are lower for the cationic compounds than for
30, resulting in better selectivity profiles. Compound 35 shows
greater selectivity
pIC50 = À1.3). Compound 37 shows a better overall selectivity
profile than 35 with a marginally greater bias against CatL2
pIC50 = À1.9) than CatS ( pIC50 = À1.7).
The crystal structure20 of 35 bound to CatL shows contact be-
(D
pIC50 = À2.0) against CatS than CatL2
(D
(
D
D
tween the cationic amino substituent and the interacting Asp71-
Asp141 pair (Fig. 2). There are four protein molecules in the asym-
metric unit and contact distances between oxadiazole nitrogen
atoms and backbone amide hydrogen atoms lie in the range
2.21 Å to 2.52 Å for Met70 and 2.45 Å to 2.78 Å for Asp71. These
hydrogen bonds are relatively long and the greater CatS potency
of 21, 26 and 27 may reflect hydrogen bond geometries that are
closer to ideal values when these compounds are bound to CatS.
Table 1
Cathepsin pIC50 values
a
a
a
a
a
Compd Cat L pIC50
Cat L2 pIC50
Cat B pIC50
Cat K pIC50
Cat S pIC50
1
2
3
7
8
6.6
7.9
7.9
6.3
5.5
5.8
5.7
5.4
7.8
6.1
5.8
5.5
5.9
<5.1
6.3
6.7
7.8
8.3
8.3
6.9
8.4
8.2
8.7
8.9
6.1
6.4
6.7
5.8
5.5
5.5
5.4
<5.1
6.8
5.8
5.4
<5
5.3
5.4
5.2
5.2
5.1
<5
5.1
<5
<5
5.1
5.3
5.2
5.2
<5
<5
<5
<5
<5
<5
<5
<5
<5
5.5
5.6
5.5
<5
<5
<5
<5.1
<5.1
5.2
<5
<5
<5
6.9
6.1
6.0
7.0
6.5
6.6
6.9
6.5
6.5
7.1
6.5
6.2
8.1
6.1
7.7
8.6
7.2
7.8
7.1
6.5
7.1
7.2
6.7
7.2
9
10
11
12
14
16
20
21
22
26
27
28
29
30
31
32
33
35
37
7.2
<5
<5.2
<5
6.8
7.6
7.7
8.2
8.2
8.5
8.7
8.1
7.4
7.0
5.9
6.2
6.9
6.9
7.2
5.5
6.9
7.0
6.0
6.8
Figure 2. Crystal structure of 35 bound to Cathepsin L showing molecular surface of
the protein and locations of key residues. The hydrogen bonds between the
oxadiazole nitrogen acceptors and the backbone amide donors of Met70 and Asp71
are shown in turquoise.
5.4
5.2
a
Values are means of at least three experiments; see Refs. 16–18.

P. A. Bethel et al. / Bioorg. Med. Chem. Lett. 19 (2009) 4622–4625
4625
This variability in the contributions of these hydrogen bonds to po-
References and notes
tency suggests that this region of the protein surface can be used to
manipulate Cathepsin selectivity.
1. Vasiljeva, O.; Reinheckel, T.; Peters, C.; Turk, D.; Turk, V.; Turk, B. Curr. Pharm.
Des. 2007, 13, 387.
The CatS and CatL2 pIC50 values for 20 and 21 have implications
beyond Cathepsin inhibition. Aza-substitution of 20 leading to 21,
which adds a hydrogen bond acceptor, leads to a 1.9 log unit in-
crease in CatS potency. The CatL2 pIC50 difference for these com-
pounds appears to be even larger (>2.2) although 20 is too weak
an inhibitor to be detected in the assay. It has been suggested23
that a neutral–neutral hydrogen bond can contribute no more than
a factor of 15-fold (1.2 log units) to binding affinity. This is rather
less than either of the differences in CatS (1.9) and CatL2 (>2.2)
pIC50 values measured for 20 and 21 which suggests that the upper
limit to the contribution of a neutral–neutral hydrogen bond to po-
tency may be higher than was previously thought. Caution is
needed when interpreting these results because 20 and 21 are both
racemic and the structure of 21 bound to CatS is not currently
available. However, the increase in potency resulting from isosteric
introduction of a hydrogen bond acceptor into an inhibitor mole-
cule is compelling evidence that the acceptor forms a hydrogen
bond with the protein. It is also worth noting that while 21 is 0.7
units less potent than 1 against CatL it is still 0.4 units more potent
than 20 against this enzyme.
2. Felbor, U.; Kessler, B.; Mothes, W.; Ploegh, H. L.; Goebel, H. H.; Bronson, R. T.;
Olsen, B. R. Proc. Nat. Acad. Sci. 2002, 99, 7883.
3. Hagemann, S.; Günther, T.; Dennemarker, J.; Lohmüller, T.; Brömme, D.; Schüle,
R.; Peters, C.; Reinheckel, T. Eur. J. Cell Biol. 2004, 83, 775.
4. Asaad, N.; Bethel, P. A.; Coulson, M. D.; Dawson, J. E.; Ford, S. J.; Gerhardt, S.; Grist,
M.; Hamlin, G. A.;James, M. J.; Jones, E. V.; Karoutchi, G. I.; Kenny, P. W.;Morley, A.
D.; Oldham, K.; Rankine, N.; Ryan, D.; Wells, S. L.; Wood, L.; Augustin, M.; Krapp,
S.; Simader, H.; Steinbacher, S. Bioorg. Med. Chem. Lett., 2009, 19, 4280.
5. Honig, B.; Nicholls, A. Science 1995, 268, 1144.
6. Young, T.; Abel, R.; Kim, B.; Berne, B. J.; Friesner, R. A. Proc. Nat. Acad. Sci. 2007,
104, 808.
7. Bogan, A. A.; Thorn, K. S. J. Mol. Biol. 1998, 280, 1.
8. DeLano, W. L. Curr. Opin. Struct. Biol. 2002, 12, 14.
9. Hajduk, P. J.; Huth, J. R.; Fesik, S. W. J. Med. Chem. 2005, 48, 2518.
10. Erlanson, D. A.; McDowell, R. S.; O’Brien, T. J. Med. Chem. 2004, 47, 3463.
11. Congreve, M.; Chessari, G.; Tisi, D.; Woodhead, A. J. J. Med. Chem. 2008, 51,
3661.
12. There are four protein molecules in the asymmetric unit of the crystal structure
of the complex of Cathepsin L with 2. The ranges for the H–H distance and
angle between the NH vectors are 2.77 Å to 2.84 Å and 25° to 27°, respectively.
13. Abraham, M. H.; Duce, P. D.; Prior, D. V.; Barratt, D. G.; Morris, J. J.; Taylor, P. J. J.
Chem. Soc., Perkin Trans. 2 1989, 1355.
14. Kenny, P. W. J. Chem. Soc., Perkin Trans. 2 1994, 199.
15. Jorgensen, W. L.; Pranata, J. J. Am. Chem. Soc. 1990, 112, 2008.
16. pIC50 = Àlog(IC50/M).
17. Values of pIC50 for inhibition of Cathepsins L, L2, S, K and B were determined
from dose dependent inhibition of cleavage of fluorogenic, AMC-tagged,
peptide substrates. See also Ref. 18.
In summary, we have exploited differences between protein
structures to modulate Cathepsin selectivity. We have also demon-
strated that an aligned pair of hydrogen bond donors can function as
a hot spot. Finallywe suggest that a proposed upper limit for the con-
tribution of a neutral–neutral hydrogen bond is underestimated.
18. Hulkower, K. I.; Butler, C. C.; Linebaugh, B. E.; Klaus, J. L.; Keppler, D.; Giranda,
V. L.; Sloane, B. F. Eur. J. Biochem. 2000, 267, 4165.
19. Gupta, S.; Singh, R. K.; Dastidar, S.; Ray, A. Exp. Opin. Ther. Targets 2008, 12, 291.
20. Diffraction data for compound 35 were collected at 100 K on beamline PX at
the Swiss Light Source (SLS, Villigen, Switzerland). The structure was solved by
molecular replacement and refined to a final resolution of 2.33 Å and R-factor
of 25.0% using the CCP4 (Ref. 21) and Coot (Ref. 22) software packages. This
structure has been deposited in the RSB (home.rcsb.org) Protein Data Bank
(www.rcsb.org/pdb/home/home.do) with reference code 3HWN.
21. Collaborative Computational Project, Number 4; (http://www.ccp4.ac.uk) Acta
Crystallogr. 1994, D50, 760.
Acknowledgments
A number of pIC50 values were measured by Jack Dawson and
Linda Wood. Lyn Rosenbrier Ribeiro and Helen Sawney shared as-
say development expertise.
22. Coot: http://www.ysbl.york.ac.uk/~emsley/coot.
23. Davis, A. M.; Teague, S. J. Angew. Chem., Int. Ed. 1999, 38, 736.