Use of papain as a model for the structure-based design of cathepsin K inhibitors: Crystal structures of two papain-inhibitor complexes demonstrate binding to S'-subsites

Article abstract of DOI:10.1021/jm980249f

Papain has been used as a surrogate enzyme in a drug design effort to obtain potent and selective inhibitors of cathepsin K, a new member of the papain superfamily of cysteine proteases that is selectively and highly expressed in osteoclasts and is implic

Full text of DOI:10.1021/jm980249f

J . Med. Chem. 1998, 41, 4567-4576
4567
Use of P a p a in a s a Mod el for th e Str u ctu r e-Ba sed Design of Ca th ep sin K
In h ibitor s: Cr ysta l Str u ctu r es of Tw o P a p a in -In h ibitor Com p lexes
Dem on str a te Bin d in g to S′-Su bsites
J udith M. LaLonde,*,† Baoguang Zhao,† Ward W. Smith,† Cheryl A. J anson,‡ Renee L. DesJ arlais,§
Thaddeus A. Tomaszek,| Thomas J . Carr, Scott K. Thompson, Hye-J a Oh, Dennis S. Yamashita,
Daniel F. Veber, and Sherin S. Abdel-Meguid*,†
Departments of Structural Biology, Protein Biochemistry, Physical and Structural Chemistry, Molecular Recognition, and
Medicinal Chemistry, SmithKline Beecham Pharmaceuticals, 709 Swedeland Road, King of Prussia, Pennsylvania 19406
Received April 23, 1998
Papain has been used as a surrogate enzyme in a drug design effort to obtain potent and
selective inhibitors of cathepsin K, a new member of the papain superfamily of cysteine
proteases that is selectively and highly expressed in osteoclasts and is implicated in bone
resorption. Here we report the crystal structures of two papain-inhibitor complexes and the
rational design of novel cathepsin K inhibitors. Unlike previously known crystal structures of
papain-inhibitor complexes, our papain structures show ligand binding extending deep within
the S′-subsites. The two inhibitor complexes, carbobenzyloxyleucinyl-leucinyl-leucinal and
carbobenzyloxy-L-leucinyl-L-leucinyl methoxymethyl ketone, were refined to 2.2- and 2.5-Å
resolution with R-factors of 0.190 and 0.217, respectively. The S′-subsite interactions with
the inhibitors are dominated by an aromatic-aromatic stacking and an oxygen-aromatic ring
edge interaction. The knowledge of S′-subsite interactions led to a design strategy for an
inhibitor spanning both subsites and yielded a novel, symmetric inhibitor selective for cathepsin
K. Simultaneous exploitation of both S- and S′-sites provides a general strategy for the design
of cysteine protease inhibitors having high specificity to their target enzymes.
In tr od u ction
Papain shares sequence and structural homology with
other mammalian cysteine proteases. Several of these
cysteine proteases are implicated in diseases caused by
abnormal levels of proteolytic activity. For example,
cathepsin K is selectively expressed in human osteo-
clasts3 and is a target for treatment of diseases involving
excessive bone loss such as osteoporosis. Cathepsin B
is involved in intracellular protein degradation and is
implicated in muscular dystrophy4 and metastasis.5 The
high degree of sequence homology between papain and
cathepsin K (46% sequence identity) suggests that
papain can serve as a suitable model for designing
disease-specific cysteine protease inhibitors. Moreover,
in the presence of structure-activity relationship (SAR)
data on cathepsin K, papain can serve as a model for
elucidating components important in selectivity.
Recent successes in the rational design of potent,
selective HIV-1 protease inhibitors and the subsequent
demonstration that they are highly effective drugs in
the treatment of AIDS have validated the important role
of rational design in the drug discovery process. Knowl-
edge of the three-dimensional structure of HIV protease,
particularly with bound ligands, was critical for the
rapid progress in the design of HIV-1 protease inhibi-
tors. Equally important was the vast knowledge gained
earlier from other aspartyl proteases such as renin and
pepsin. Here, we will describe the use of papain as a
model for the structure-based design of cathepsin K
inhibitors that are potential drugs for the treatment of
osteoporosis. The use of papain as a drug design target
proved instrumental in obtaining inhibitors for cathe-
psin K at a time when large quantities and structural
information of cathepsin K were unavailable.
Papain is a plant cysteine protease isolated from
papaya latex. It is a 212-amino acid polypeptide with
well-known biological and structural properties.1 Its
catalytic mechanism involves an active site triad,
consisting of Cys25, His159, and Asn175, that resides
in a cleft between two domains. The reactive thiol of
Cys25 hydrolyzes peptide bonds by formation of a
covalent acyl enzyme intermediate with the carbonyl
carbon of the scissile bond.2
Thus far, a number of inhibitors for papain or papain
family members have been identified and the three-
dimensional structures of their complexes with papain
have been elucidated. These include the peptide alde-
hyde6 (leupeptin), the chloromethyl ketone,7 and the
epoxide8,9 inhibitors of papain. In these structures, the
inhibitors were found to bind to Cys25 through a
reactive group, with the inhibitors interacting exclu-
sively with the “nonprime” side of the active site
(S-subsites). Two other papain-inhibitor structures,
succinyl-p-nitroanilide10 and NR-p-tolylsulfonyl-L-lysine
chloromethyl ketone,11 show limited penetration into the
S′-subsite (extending only into the S1′-subsite). Also,
a crystal structure of cathepsin B with an irreversible
epoxide inhibitor bound was recently published.12 The
inhibitor binds within the S′-side of the active site, but
* To whom reprint requests should be addressed.
† Department of Structural Biology.
‡ Department of Protein Biochemistry.
§ Department of Physical and Structural Chemistry.
| Department of Molecular Recognition.
Department of Medicinal Chemistry.
10.1021/jm980249f CCC: $15.00 © 1998 American Chemical Society
Published on Web 10/21/1998

4568 J ournal of Medicinal Chemistry, 1998, Vol. 41, No. 23
LaLonde et al.
(1)
larger inhibitor and makes more contacts with residues
comprising the active site cavity. Small differences in
peptide bond conformation occur in order to avoid bad
contacts or form closer contacts with the inhibitor. For
example, in the aldehyde 1 complex, the carbonyl of
Ser21 is rotated away relative to its position in meth-
oxymethyl ketone 2 to avoid a bad contact with the
second leucyl group of the inhibitor (see Figures 5 and
6). Main-chain differences at Gly65 and Asp158 occur
in the aldehyde 1 complex to enhance van der Waals
contacts with the inhibitor. These are minor changes
and do not affect the overall structure of the enzyme.
Most of the differences in side-chain positions for the
two enzyme-inhibitor complexes occur in regions with
high thermal motion and are not a consequence of
inhibitor binding.
(2)
Furthermore, the papain structure in the aldehyde 1
and methoxymethyl ketone 2 complexes is nearly iden-
tical to its structure for the leupeptin-papain complex,6
a S-subsite inhibitor-papain complex. The rms differ-
ences between protein atoms in the aldehyde 1 and
methoxymethyl ketone 2 complexes with the leupeptin-
papain complex for main-chain and side-chain atoms are
0.36 and 1.46 Å and 0.41 and 1.41 Å, respectively. Only
small differences occur in the conformation of main-
chain and side-chain atoms that form the S- and
S′-subsites between papain, papain-S-subsite inhibitor
complex, and papain-S′-subsite inhibitor complex.
Therefore, no change in protein conformation is ob-
served upon inhibitor binding to either the S- or S′-side
of the active site.
F igu r e 1. Depiction of the aldehyde 1 and methoxymethyl
ketone 2 inhibitors.
cathepsin B has a significant insertion on the S′-side
compared to the other papain family members including
cathepsin K. Therefore, the structure was of little use
in designing cathepsin K inhibitors. The previously
known inhibitor-papain complexes have been used to
extrapolate the interactions of the S-portion of the
peptide substrate complex. However, they only reveal
limited information on inhibitor specificity for the S′-
subsite. Here, we report the structures of two papain-
inhibitor complexes where the inhibitors bind exten-
sively into the S′-subsites. The inhibitors carbobenzyloxy-
leucinyl-leucinyl-leucinal (abbreviated aldehyde 1) and
carbobenzyloxy-L-leucinyl-L-leucinyl methoxymethyl ke-
tone (abbreviated methoxymethyl ketone 2), depicted in
Figure 1, serve as a starting point for investigating both
S- and S′-subsite selectivity for cysteine proteases.
P a p a in -In h ibitor In ter a ction s. The electron den-
sities of the bound inhibitors are shown in Figure 2. In
both cases the density of the inhibitors is easily distin-
guished from side-chain density in the binding cleft.
Methoxymethyl ketone 2 binds noncovalently to papain
(kobs/I ) 26 M-1 s-1), while it appears that aldehyde 1
binds covalently (kobs/I ) 430 000 M-1 s-1).
The substrate binding pocket of papain lies in a cleft
between two domains. Domain 1 consists of residues
1-11 and 113-207, while domain 2 consists of residues
12-112 and 208-212. The binding pocket of papain has
been defined by Berger and Schechter13 in terms of
seven subsites designated S4 to S1 and S1′ to S3′. Each
subsite accommodates one amino acid of the peptide
substrate. The four S-subsites (S4 to S1) are located
on the acyl side of the scissile bond of the bound
substrate, while the three S′-subsites (S1′ to S3′) are
located on the amino side. The primary specificity of
papain is for binding hydrophobic residues such as Phe
in the S2-subsite.1 Both inhibitors bind in the groove
between papain’s two domains, along the S′-subsites
(Figure 3). This is in contrast to binding along the
S-subsites by other papain inhibitors for which crystal
structures are known, including the chloromethyl ke-
tone,7 epoxysuccinyl,8,9 and leupeptin6 peptide inhibi-
tors. The peptide groups of aldehyde 1 and methoxy-
methyl ketone 2 bind in the opposite direction to that
postulated for a peptide substrate.7 In addition, the
binding of the inhibitors in the cleft is such that one
edge of the benzene ring of the Cbz group is entirely
buried and the other edge is exposed to solvent.
Resu lts a n d Discu ssion
Over a ll Str u ctu r e of th e P a p a in -In h ibitor Com -
p lexes. The structures of papain in the two papain-
inhibitor complexes are virtually identical to the struc-
ture of native papain. The rms differences between
native papain and the aldehyde 1 and methoxymethyl
ketone 2 complexes for main-chain atoms and side-chain
atoms are 0.14 and 2.11 Å and 0.30 and 1.57 Å,
respectively. Most of the differences observed in side-
chain positions occur on the surface of the protein and
involve atoms with high thermal motion. For the
methoxymethyl ketone 2 complex, differences with the
native structure occur at Gly180 and Trp181, where the
peptide bond is rotated to allow the carbonyl group to
interact with the inhibitor Cbz ring (see below).
The structural differences between the two inhibitor
complexes are also small. The rms differences between
protein atoms in the aldehyde 1 and methoxymethyl
ketone 2 complexes with papain are 0.28 and 1.45 Å for
main-chain and side-chain atoms, respectively. Some
differences in main-chain positions occur as a result of
inhibitor binding to the S′-subsites. Aldehyde 1 is a
Ald eh yd e 1 Bin d in g In ter a ction s. The amino acid
residues that form contacts with the inhibitor are shown

Papain as Model for Design of Cathepsin K Inhibitors
J ournal of Medicinal Chemistry, 1998, Vol. 41, No. 23 4569
F igu r e 2. Omit electron density maps around the aldehyde 1 (top) and methoxymethyl ketone 2 (bottom) inhibitors. The maps
were calculated with coefficients |Fo - Fc| omitting the inhibitor positions from phasing.
in Figure 4. On the basis of the reactivity of the
aldehyde group toward thiols and the observation of the
“omit” electron density maps, we presume that the
aldehyde 1 is covalently bound to Cys25 as a hemimer-
captal. This covalent bond occurs at the analogous
position to the P1 carbonyl carbon in a peptide sub-
strate. The inhibitor contacts (less than 4.0 Å) 11
protein residues: Gln19, Gly20, Ser21, Gly23, Asn64,
Gly65, Gln142, Asp158, His159, Trp177, and Trp181
and 2 water molecules (Wat412 and Wat480). These
interactions include hydrophobic, electrostatic, and
hydrogen bonding, but most interesting are the stacking
interactions between the aromatic ring of the Cbz group
and Trp177 and Trp181. Figure 5 shows interactions
between the inhibitor and the protein in the active site.
Aromatic stacking interactions in proteins have been
defined by Burley and Petsko as pairs of aromatic
residues with phenyl ring centroid separations between
4.5 and 7.0 Å and dihedral angles between aromatic
planes within 30-90°.14 These separations and angles
between the inhibitor’s phenyl ring and Trp177 and
Trp181 are 5.7 Å and 33° and 7.0 Å and 50°, respec-
tively. On the basis of the Burley and Petsko defini-
tion,14 the tryptophan-Cbz stacking interactions can be
characterized as weak. The stacking interaction with
Trp181 appears stronger (closer to perpendicular) than
that for Trp177. The position of the phenyl ring is
further stabilized by a favorable electrostatic interaction
with the carbonyl oxygen of Gln142 (3.3 Å).15 The
carbonyl oxygen faces the phenyl ring so that the partial
negative charge on the carbonyl oxygen is directed
toward the partial positive charge at the edge of the

4570 J ournal of Medicinal Chemistry, 1998, Vol. 41, No. 23
LaLonde et al.
F igu r e 3. Folding diagram of papain showing the inhibitor
bound in the cleft between two domains. The ball-and-stick
model of aldehyde 1 bound in the S′-subsite is shown in yellow.
For comparison, leupeptin bound in the S-subsite is drawn in
red. Figure was prepared using MOLSCRIPT.23
phenyl ring. A second favorable oxygen-aromatic ring
edge interaction is formed between Wat480 and the Cbz
ring (3.5 Å). Hydrophobic contacts are also observed
between the CH2 group of the Cbz and Trp177.
There are four carbonyl oxygens in the inhibitor: the
first is the hemimercaptal oxygen, the second and third
are peptidyl amide oxygens, and the fourth is a car-
bamate carbonyl oxygen. Only the first peptidyl amide
oxygen is hydrogen-bonded to the protein (Figures 5 and
6); it interacts with the imidazole nitrogen of His159
and Wat412. The hemimercaptal oxygen forms two
hydrogen bonds: one to the amide group of Gln19 (a
hydrogen bond between the amide nitrogen of Gln19
and the carbonyl oxygen of the substrate is postulated
to stabilize the tetrahedral intermediate by forming the
so-called oxyanion hole10) and a second to the backbone
amide nitrogen from Cys25.
The three Leu side chains (Leu1, Leu2, Leu3) of the
inhibitor are partially buried forming a number of van
der Waals contacts in their respective subsites. Leu1
forms van der Waals contacts with Asn64, Gly65, and
Asp158. This side chain is bent, allowing it to fall into
the S1-subsite rather than extend into the S1′-subsite.
Thus, it mimics a P1 rather than a P1′ residue. Leu2
is actually the P1′ side chain; it makes van der Waals
contacts with Gly20, Ser21, and Cys22 in the S1′-
subsite. Leu3 forms van der Waals contacts with the
amide group of Gln142 in the S2′-subsite.
Meth oxym eth yl Keton e 2 Bin d in g In ter a ction s.
Methoxymethyl ketone 2 makes contacts with seven
residues (Gln19, Gly23, Gly65, Gln142, His159, Trp177,
Trp181) as well as with a water molecule (Wat511)
along the S′-side, Figure 4. As observed with aldehyde
1, the aromatic ring of the Cbz in this case is also
stacked with Trp177 and Trp181, but the interactions
are weaker. The centroid distances between phenyl
groups and dihedral angles are 8.0 Å and 24° and 7.9 Å
and 59° for Trp177 and Trp181, respectively. The
distances are greater than those observed with aldehyde
1, but the angles are approximately the same. Again
better stacking (closer to perpendicular) is observed with
the Cbz to Trp181 than to Trp177. The position of the
aromatic ring is stabilized by an electrostatic interaction
with the main-chain carbonyl oxygen of Gly180 (3.0 Å).
F igu r e 4. Diagram of residues that form contacts with the
inhibitors aldhehyde 1 (top) and ketone 2 (bottom). Distances
(in Å) are shown as dashed lines.
There are two peptidyl amide oxygens and one car-
bamate carbonyl oxygen in the inhibitor. The second
peptidyl amide oxygen forms a hydrogen bond to the
aromatic ring of Trp177. The carbamate carbonyl
oxygen hydrogen bonds to Wat511. The carboxymethyl
oxygen is also involved in hydrogen bonding. It forms
a weak hydrogen bond to the amido nitrogen on Gln142
and a second hydrogen bond to Wat511. In addition,
Wat511 forms a hydrogen bond to the amido group of
Gln142. Therefore, Wat511 stabilizes the bent confor-
mation of the inhibitor by hydrogen bonding to the
carbamate carbonyl oxygen near the Cbz group, the
carboxymethyl oxygen, and a protein side chain.
Methoxymethyl ketone 2 binds noncovalently. As a
result it does not bind in the S1-subsite, and the first
Leu side chain, Leu1, does not make the same set of
contacts as P1′ Leu2 of aldehyde 1 (Figure 6); instead
it forms a hydrophobic contact with His159. The
inhibitor is forced into a bent conformation by contacts
formed with Gln142 of the S2′-subsite as noted above.
Leu2 forms hydrophobic contacts with Trp177 in the

Papain as Model for Design of Cathepsin K Inhibitors
J ournal of Medicinal Chemistry, 1998, Vol. 41, No. 23 4571
F igu r e 5. Stereodiagram of aldehyde 1 (top) and methoxymethyl ketone 2 (bottom) bound in the S′-subsites of papain. Residues
forming interactions with the inhibitor are drawn in green. The inhibitors are drawn in yellow. Hydrogen bonds and oxygen-
aromatic ring edge interactions are shown with dashed lines. Figure was prepared using MOLSCRIPT.23
S3′-subsite. The Cbz moiety forms contacts with Trp177
and Trp181 in the S3′-subsite as does aldehyde 1.
the second peptidyl amide oxygen for ketone 2. There
is the common use of water molecules for inhibitor
binding in the cleft. Both inhibitors use a water
molecule for hydrogen bonding, but to a different
carbonyl of the inhibitor. These are the first peptidyl
oxygen for aldehyde 1 and the carbamate carbonyl
oxygen adjacent to the Cbz group in ketone 2.
Water molecules are observed in the active site of
most of the other reported papain-inhibitor complexes
as well as in native papain. Water 511 observed in the
ketone 2 complex is not observed in other papain
structures including the 1.6-Å structure of papain.16
Both of the waters associated with aldehyde 1 are
observed in other papain-inhibitor complexes. There-
fore, no clear pattern of hydration can be associated with
inhibitor binding in the S′-subsite.
Com p a r ison of Bin d in g Mod es for t h e Tw o
In h ibitor s. Only minor differences in protein side-
chain conformations are observed between the two
inhibitor complexes, and no major difference occurs in
the S′-subsite upon inhibitor binding, Figure 6. As
described above the conformations of both inhibitors are
stabilized by aromatic-aromatic interactions, specific
hydrogen bonds to peptidyl amide oxygens, and hydro-
phobic contacts from leucine side chains. Antiparallel
â-sheet type hydrogen bonding between the inhibitor
and the enzyme is not observed with these two inhibi-
tors. Moreover, each inhibitor exhibits a different
hydrogen-bonding pattern of the peptidyl amide oxy-
gens. A hydrogen bond is formed between the His159
indole ring and the first peptidyl amide oxygen for
aldehyde 1 and between the Trp177 aromatic ring and
In comparing the position of the two inhibitors, the
benzyl ring of methoxymethyl ketone 2 is translated 2.6

4572 J ournal of Medicinal Chemistry, 1998, Vol. 41, No. 23
LaLonde et al.
F igu r e 6. Superposition of aldehyde 1 (blue) and methoxymethyl ketone 2 (red) in the active site. Hydrogen bonds and oxygen-
aromatic ring edge interactions are shown with blue and red dashed lines for aldehyde 1 and methoxymethyl ketone 2, respectively.
Figure was prepared using MOLSCRIPT.23
Å farther away from the catalytic cysteine than the
benzyl ring of aldehyde 1 (Figure 6). Despite this
translation, both Cbz moieties are stacked with Trp177
and Trp181. Cbz groups of both inhibitors form an
oxygen-aromatic ring edge interaction. However, two
different main-chain carbonyls are used, one from
Gly180 for methoxymethyl ketone 2 and the other from
Gln142 for aldehyde 1. The carbonyl oxygens from
Gly180 and Gln142 approach the inhibitors from op-
posite sides of their aromatic rings (Figure 6). The
differences in the position and interactions of aldehyde
1 and ketone 2 despite the similarity in the active site
regions indicate that the binding mode is determined
by chemical functionality of the inhibitor. The fact that
aldehyde 1 binds covalently whereas ketone 2 binds
noncovalently may be the sole reason that the two
inhibitors have different positions in the S′-subsite.
Su bsite Sp ecificity. Binding of aldehyde 1 and
methoxymethyl ketone 2 inhibitors deep within the S′-
subsites of papain is a novel observation. The presence
of a Cbz moiety and its aromatic interactions with
tryptophans and carbonyl oxygens are important com-
ponents of binding specificity. However, the presence
of the Cbz group alone is not sufficient to explain the
specificity for the S′-subsite. Two other papain-inhibi-
tor structures also contain the Cbz group. These are
the chloromethyl ketone inhibitors Cbz-Phe-Ala-CK and
Cbz-Gly-Phe-Gly-CK.7 These chloromethyl ketone in-
hibitors bind covalently, along the S-subsites with their
Cbz groups forming interactions with Tyr61 and Tyr67.
The Phe side chain of Cbz-Phe-Ala-CK and Cbz-Gly-
Phe-Gly-CK sits in the S2 pocket where primary sub-
strate specificity is determined.
S atom of Cys25 and the first carbonyl group of the
inhibitor, whereas the inhibitor Cbz-Gly-Phe-Gly-CK
contains an extra methylene group between the first
carbonyl of the inhibitor and the S atom of Cys25.
Although crystallized under different conditions, differ-
ences in the crystal do not account for the subsite
preference, since none of the inhibitors or their active
site residues are involved in crystal contacts. Further-
more, there are no major differences in side-chain
conformation in the residues comprising the S- and S′-
subsites between the chloromethyl ketone inhibitor-
papain complexes and the aldehyde 1-papain and
ketone 2-papain complexes. Factors that may explain
subsite preference would include the presence of a CH2
spacer in Cbz-Gly-Phe-Gly-CK and the difference in
side-chain composition. Examination of the Cbz-Gly-
Phe-Gly-CK-papain complex shows that the extra CH2
group forms interactions with H159. In aldehyde 1,
H159 forms a hydrogen bond to the first peptidyl amide
oxygen instead, Figure 5. In Cbz-Gly-Phe-Gly-CK the
CH2 group prevents H159 from interacting with the first
peptidyl amide oxygen. Instead, its first peptidyl amide
oxygen hydrogen bonds to Gly66 in the S-subsite. In
regards to side-chain composition, a model of aldehyde
1 bound in the S-subsite shows that the larger Leu side
chain restricts the position of the Cbz group. Thus,
aldehyde 1 bound in the S-subsite would have dimin-
ished interactions with Tyr61 and Tyr67 in comparison
to Cbz-Gly-Phe-Gly-CK. Although aldehyde 1 appears
to fit in the S-subsite, Cbz-Gly-Phe-Gly-CK is not
accommodated as easily in the S′-subsite because of the
large Phe residue. Therefore, in the absence of a Phe
side chain that would bind to the P2 pocket, a Cbz-
containing inhibitor may bind to either subsite. The
explanation for S′-subsite preference may be a combina-
tion of the factors discussed above as well as the
aromatic interactions formed between the Cbz and
tryptophan side chains and the main-chain carbonyls.
The S-subsite inhibitor Cbz-Gly-Phe-Gly-CK is similar
to aldehyde 1 in that both bind covalently, contain three
peptide residues, and are approximately the same
length (15.5 Å). However, aldehyde 1 is bound to Cys25
as a hemimercaptal with the covalent bond between the

Papain as Model for Design of Cathepsin K Inhibitors
J ournal of Medicinal Chemistry, 1998, Vol. 41, No. 23 4573
F igu r e 7. (A) Design of the diacylaminomethyl ketone from leupeptin and aldehyde 1. (B) Compound 3, diacylaminomethyl
ketone, and compound 4, diacyl carbohydrazide.
Design of Ca th ep sin K In h ibitor s. Prior to the
structure determination of cathepsin K,17,18 papain was
chosen as a surrogate for the structure-based design of
cathepsin K inhibitors. Papain serves as a suitable
model since it has 46% amino acid sequence identity to
cathepsin K. Papain was also selected because it is
commercially available in large quantities and its
structure is known. To ensure that the novel binding
mode observed in the crystal structure of the aldehyde
1-papain complex was not an artifact of crystallization,
crystals of the leupeptin-papain complex were produced
under identical conditions. The structure of the leu-
peptin-papain complex (data not shown) was nearly
identical to its previously reported structure.6 Further-
more, it was isomorphous with the aldehyde 1-papain
complex except that leupeptin was bound on the S-side
of the active site as observed in the earlier reported
structure.6
By overlaying the structures containing leupeptin in
the S-subsite and aldehyde 1 in the S′-subsite, a
hypothetical 1,3-bis(acylaminomethyl) ketone was de-
signed that spanned both subsites. This resulting
inhibitor was simplified by (1) removing the two side
chains on either side of the ketone, (2) shortening the
structure by one amino acid on both sides, and (3)
targeting a symmetrical structure, ketone 3, by replac-
ing the acetyl group found in leupeptin with a Cbz group
(Figure 7A).19 Modeling suggested that even though the
aromatic-aromatic interaction with Trp188 (Trp181 in
papain) and the Cbz group would be eliminated by

4574 J ournal of Medicinal Chemistry, 1998, Vol. 41, No. 23
LaLonde et al.
shortening the inhibitor structure, it appeared that a
better interaction with Trp184 (Trp177 in papain) could
compensate. Furthermore, the high potency of the
dipeptide aldehyde, Cbz-Leu-Leu-H, against cathepsin
K supported this hypothesis.19
dehyde inhibitors and molecular modeling provided
pivotal insights for the design of novel, selective inhibi-
tors of cathepsin K. The use of papain as a surrogate
for cathepsin K illustrates the utility and limitations
of employing surrogate enzymes in the design of pro-
tease inhibitors.
Ketone 3 was found to be a potent inhibitor of
cathepsin K. A subsequent crystallographic structure
determination of ketone 3 bound to cathepsin K19
revealed that (1) the covalent inhibitor spans both the
S- and S′-subsites, (2) one of the leucine side chains fills
the hydrophobic S2 pocket, (3) aromatic-aromatic
interactions are formed between the two Cbz groups and
Trp184 (face-to-face, centroid distance and angle of 5.5
Å and 23°) and Tyr67 (face-to-edge, centroid distance
and angle of 5.5 Å and 83°), and (4) aromatic ring edge
interactions are observed between the Cbz groups and
the amide carbonyl oxygens of Asn18 (3.02 Å) and
weakly with Leu160 (3.12 Å).
Exp er im en ta l Section
In h ibitor s. Aldehyde 1 was prepared as previoiusly re-
ported:24 1H NMR (CDCl3) δ 9.43 (s, 1H), 7.28 (m, 5H), 6.68
(d, 1H), 6.47 (d, 1H), 5.12 (d, 1H), 5.02 (s, 2H), 4.39 (m, 2H),
4.08 (m, 1H), 1.70-1.32 (m, 9H), 0.81 (m, 18H). Anal.
(C26H41N3O5‚0.5H2O) C, H, N.
Ketone 2 was synthesized by following the method of J ones
et al.25,26 and is described below:
Cbz-Leu -Leu r-Br om om eth yl Keton e. Isobutyl chloro-
formate (1.37 mL, 10.6 mmol) was added to a solution of Cbz-
Leu-Leu-OH (4.0 g, 10.6 mmol; Bachem) and N-methylmor-
pholine (1.16 mL, 10.6 mmol) in THF (20 mL) at -40 °C, and
the reaction was stirred for 15 min. The reaction mixture was
then filtered; then a solution of diazomethane (from 5.9 g of
1-methyl-3-nitro-1-nitrosoguanidine and 18 mL of 40% aque-
ous KOH in 150 mL of diethyl ether) was added slowly. The
reaction was sealed and was maintained at 0 °C in a refrigera-
tor overnight. The reaction mixture was then treated with
30% HBr in AcOH (7 mL) and was stirred for 5 min. The
reaction mixture was diluted with EtOAc (50 mL) and then
was extracted with 15% aqueous citric acid, then saturated
sodium bicarbonate (with CO2 evolution), and then brine. The
combined organics were dried with anhydrous sodium sulfate,
filtered, and concentrated in vacuo to give the title compound
as a white solid (3.2 g, 67%): 1H NMR (CDCl3, 400 MHz) δ
7.4-7.3 (m, 5H), 6.74 (d, J ) 7.6 Hz, 1H), 5.34 (d, J ) 7.6 Hz,
1H), 5.10 (s, 2H), 4.80-4.75 (m, 1H), 4.22 (brs, 1H), 4.05 (AB,
J AB ) 16.0 Hz, ∆δAB ) 0.045, 2H), 1.7-1.6 (m, 4H), 1.55-1.47
(m, 2H), 0.95 (d, J ) 6.8 Hz, 3H), 0.91 (3d, J ) 6.4 Hz, 9H); IR
(thin film) 1716, 1700, 1670 cm-1. Anal. (C21H31BrN2O4) C,
H, N, Br.
Interestingly, the potency of inhibition by ketone 3 is
much weaker against papain (K
> 10 µM) than
i,app
against cathepsin K (K i,app of 22 nM).19 One difference
in the structures of cathepsin K and papain is near the
S-subsite where cathepsin K contains a 2-amino acid
residue insertion in the loop from residues 151 to 161.
Nonetheless, overlays of the papain and cathepsin
K-ketone 3 crystal structures show no obvious steric
hindrance between ketone 3 and papain side chains. The
difference in potency of ketone 3 in cathepsin K versus
papain can also be rationalized from the two observed
binding modes of Cbz groups on the S-side of cathepsin
K. In the ketone 3-cathepsin K crystal structure, the
Cbz group on the S-side is oriented between Tyr67 and
Leu160; however, in the crystal structure of cathepsin
K and (Cbz-LeuNHNH)2CO (4), an isosteric compound
of the diacyl carbohydrazide class of inhibitors20 (Figure
7B), the Cbz binds between Tyr 67 and Asp 61, the S3-
subsite. Modeling indicates that ketone 3 is not able
to bind to papain as in the (Cbz-LeuNHNH)2CO (4)-
cathepsin K complex,20 because a steric clash would
occur between the Cbz group and Tyr61 which is
replaced by the smaller Asp61 in cathepsin K. In the
cocrystal of cathepsin K-ketone 3, this S3-subsite is
occupied by Lys44 from an adjacent molecule in the
crystal lattice. The two binding modes are observed in
different crystal forms for the two complexes. Therefore,
understanding the large difference in inhibitor potency
of ketone 3 in papain versus cathepsin K may be aided
by additional structural studies of ketone 3 bound to
papain.
Cbz-Leu -Leu r-Hyd r oxym eth yl Keton e. Benzoylformic
acid (1.1 g, 7.1 mmol) was added to a solution of Cbz-Leu-Leu
R-bromomethyl ketone (2.7 g, 5.9 mmol) in DMF (20 mL) at
room temperature. Then solid potassium fluoride (0.52 g, 8.9
mmol) was added, and the reaction mixture was stirred
overnight. The reaction mixture was then diluted with EtOAc
(100 mL), and was poured into water (50 mL). The combined
organics were extracted with brine, dried with magnesium
sulfate, filtered, concentrated in vacuo, and then dissolved in
THF (200 mL). A solution of potassium bicarbonate (200 mL,
200 mmol, 1 M) was added, and the reaction mixture was
stirred vigorously overnight. The reaction mixture was diluted
with EtOAc (200 mL) and then was extracted with water and
then brine. The combined organics were dried with anhydrous
magnesium sulfate, filtered, and concentrated in vacuo to give
the title compound as a white solid (3.2 g, 100%): 1H NMR
(CDCl3, 400 MHz) δ 7.35-7.31 (m, 5H), 6.74 (d, J ) 7.6 Hz,
1H), 5.35 (d, J ) 7.6 Hz, 1H), 5.10 (s, 2H), 4.7-4.6 (m, 1H),
4.36 (s, 2H), 4.25-4.18 (m, 1H), 3.2 (brs, 1H), 1.7-1.43 (m,
6H), 0.93 (d, J ) 6.0 Hz, 3H), 0.91 (2d, J ) 6.0 Hz, 6H), 0.90
Con clu sion s
(d, J ) 6.0 Hz, 3H); IR (thin film) 3307, 1716, 1699, 1652 cm-1
.
Binding of the aldehyde (1) and methoxymethyl
ketone (2) inhibitors deep within the S′-subsite is unique
for papain, although it has been reported for inhibitor
binding to cathepsin B. The directionality of binding
appears to be influenced by aromatic-aromatic stacking
and oxygen-aromatic ring edge interactions. This
observation has inspired a drug design strategy using
papain as a surrogate for cathepsin K. With the
knowledge of S′-subsite interactions in papain, numer-
ous potent, selective cathepsin K inhibitors were de-
signed that span both sides of the active site.19,20
Ultimately, the papain structures combined with infor-
mation from structure-activity studies on peptide al-
Cbz-Leu -Leu r-Meth oxym eth yl Keton e. Methyl iodide
(0.32 mL, 5.1 mmol) was added to a solution of Cbz-Leu-Leu
R-hydroxymethyl ketone (0.4 g, 1.0 mmol) and silver(I) oxide
(0.46 g, 2.0 mmol) in methylene chloride (30 mL), and the
reaction mixture was refluxed for 7 h. The reaction was
incomplete; therefore, additional methyl iodide (1.0 mL, 15.5
mmol) was added, and the reaction mixture was stirred at
room temperature over 2 days. The reaction mixture was
filtered, concentrated in vacuo, and then chromatographed
(silica gel, 25% EtOAc-hexanes) to give the title compound
as a white solid (0.17 g, 42%): 1H NMR (CDCl3, 400 MHz) δ
7.35-7.24 (m, 5H), 6.35-6.33 (d, 1H, J ) 7.74 Hz), 5.06 (s,
2H), 4.81-4.75 (m, 1H), 4.16 (brs, 1H), 4.12 (AB, J AB ) 10.3
Hz, ∆δAB ) 0.097, 2H), 3.40 (s, 3H), 1.60-1.30 (m, 6H), 0.92

Papain as Model for Design of Cathepsin K Inhibitors
J ournal of Medicinal Chemistry, 1998, Vol. 41, No. 23 4575
factor refinement as implemented in XPLOR.21 The final
native model consisted of 1694 atoms, including 64 water
molecules, with an R-factor of 0.218 for data from 10 to 2.5 Å
and an average temperature factor of 24.2 Å2. The phases
from native papain were used in calculating difference Fourier
maps (coefficients |2Fo - Fc| and |Fo - Fc|) into whose density
the inhibitor atoms were built. The molecular graphics
program CHAIN22 was used for model building and map
inspection in subsequent cycles of simulated annealing and
positional and temperature factor refinement of the papain-
inhibitor complexes. Refinement statistics for the papain-
inhibitor complexes are shown in Table 2.
Ta ble 1. Data Collection Statistics
aldehyde
1-papain
ketone
2-papain
cell (Å)
a ) 100.54
b ) 50.74
c ) 62.28
6
a ) 101.00
b ) 50.99
c ) 62.48
2
2.5 Å
41190
11412
13.7
no. of crystals
resolution limit
no. of observations
no. of reflections
2.2
62560
14458
8.78
R
sym(I)
no. refls, I/σ I > 2
redundancy
12955
4.3
9152
3.6
91
In the final model, several of the side-chain positions have
weak electron density and are disordered beyond either the
C-â or C-γ atom. These disordered side chains are listed in
Table 3 and have their occupancies set to 0 in the coordinate
file. The coordinates for the aldehyde 1-papain and meth-
oxymethyl ketone 2-papain complexes have been submitted
to the PDB and given codes 1BP4 and 1BQI, respectively.
% complete
95
Ta ble 2. Final Refinement Statistics
aldehyde
ketone
2-papain
1-papain
resolution Å
10-2.2
0.284/0.291
0.240/0.190
10
12800
1772
8-2.5
0.316/0.311
0.306/0.217
9
8679
1727
44
26.4
48.7
0.85
initial R-free/R-factor
final R-free/R-factor
no. of macrocycles
no. of reflections
no. of atoms
Refer en ces
(1) Baker, E. N.; Drenth, J . The Thiol Proteases: Structure and
Mechanism. In Biological Macromolecules and Assemblies, Vol.
3. Active Sites of Enzymes; J urnak, F. A., McPherson, A., Eds.;
J ohn Wiley and Sons: New York, 1987; pp 313-368.
(2) Storer, A. C.; Menard, R. Catalytic Mechanism in Papain Family
of Cysteine Peptidases. Methods Enzymol. 1994, 244, 487-500.
(3) Drake, F. H.; Dodds, R.; J ames I.; Connor J .; Debouck, C.;
Richardson, S.; Lee, E.; Rieman, D.; Barthlow, R.; Hastings, G.;
Gowen, M. Cathepsin K, But Not Cathepsins B, L, or S, Is
Abundantly Expressed in Human Osteoclasts. J . Biol. Chem.
1996, 271, 12511-12516.
no. of solvent
88
22.7
43.3
0.65
average overall B-factor (Å2)
inhibitor B-factor (Å2)
inhibitor occupancy
rms deviations
bonds (Å)
angles (deg)
dihedrals (deg)
impropers (deg)
0.010
1.529
24.743
1.387
0.011
1.595
24.859
1.441
(4) Katunuma, N.; Kominami, E. Abnormal Expression of Lysosomal
Cysteine Proteases in Muscle Wasting Diseases. Rev. Physiol.
Biochem. Pharmacol. 1987, 108, 1-20.
(5) Sloane, B. F. Cathepsin B and Cystatins: Evidence for a Role
in Cancer Progression. Semin. Cancer Biol. 1990, 1, 137-152.
(6) Schroeder, E.; Phillips, C.; Garman, E.; Harlos, K.; Crawford,
C. X-ray Crystallographic Structure of a Papain-Leupeptin
Complex. FEBS Lett. 1993, 315, 38-42.
(7) Drenth, J .; Kalk, K. H.; Swen, H. M. Binding of Chloromethyl
Ketone Substrate Analogues to Crystalline Papain. Biochemistry
1976, 15, 3731-3738.
Ta ble 3. Disordered Side Chains
aldehyde 1
methoxymethyl ketone 2
R59, N77, R98, R111, K139,
K145, K156
R59, Q73, N77, R98, K145
(8) Varughese, K. I.; Ahmed, F. R.; Carey, P. R.; Hasnain, S.; Huber,
P.; Storer, A. C. Crystal Structure of a Papain-E-64-Complex.
Biochemistry 1989, 28, 1220-1332.
(9) Yamamoto, D.; Matsumoto, K.; Ohishi, H.; Ishida, T.; Inoue, M.;
Kitamura, K.; Mizuno, H. Refined X-ray Structure of Papain
E-64-c Complex at 2.1 Å Resolution. J . Biol. Chem. 1991, 266,
14771-14777.
(10) Yamamoto, A.; Tomoo, K.; Doi, M.; Ohishi, H.; Inoue, M.; Ishida,
T.; Yamamoto, D.; Tsuboi, S.; Okamoto, H.; Okada, Y. Crystal
Structure of Papain-Succinyl-Gln-Val-Val-Ala-Ala-p-Nitroanilide
complex at 1.7 Å resolution: Noncovalent Binding Mode of a
Common Sequence of Endogenous Thiol Protease Inhibitors.
Biochemistry 1992, 31, 11305-11309.
(11) Drenth, J .; J ansonius, J . N.; Koekoek, R.; Swen, H. M.; Wolthers,
B. G. Structure of Papain. Nature 1968, 218, 929-932.
(12) Turk, D.; Podobnik, M.; Popovic, T.; Katunuma, N.; Bode, W.;
Huber, R.; Turk, V. Crystal Stucture of Cathepsin B Inhibited
with CA030 at 2.1 Angstroms Resolution: A Basis for the Design
of Specific Epoxysuccinyl Inhibitors Biochemistry 1995, 34,
4791-4797.
(13) Berger, A.; Schechter, I. Mapping the Active Site of Papain With
Aid of Peptide Substrates and Inhibitors. Philos. Trans. R. Soc.
London 1970, B257, 249-264.
(14) Burley, S. K.; Petsko, G. A. Aromatic-Aromatic Interaction: A
Mechanism of Protein Structure Stablization. Science 1985, 229,
23-28.
(15) Deber, C. M.; J oshua, H. Side Chain Interactions in Aromatic
Dipeptides. Biopolymers 1972, 11, 2493-2503.
(d, 9H, J ) 6.2 Hz), 0.90 (d, 3H, J ) 6.47 Hz); IR (KBr) 3301,
1731, 1694, 1646 cm-1. Anal. (C22H34N2O5) C, H, N.
Cr ysta lliza tion . Papain from papaya latex was purchased
from ICN Biomedicals (#100924) as a suspension. Prior to
crystallization the suspension buffer was removed by centrifu-
gation. A concentrated solution (approximately 35 mg/mL) of
papain was prepared in water. Papain was crystallized by
hanging drop vapor diffusion method with 0.1 M Tris HCl, pH
8.5, 0.5 M trisodium citrate, and 20% PEG 600. Crystals of
the enzyme-inhibitor complexes were prepared by either
cocrystallization (methoxymethyl ketone 2) or by soaking the
crystal (aldehyde 1). Solutions of inhibitors were prepared in
DMSO to give approximately 0.03 M. For cocrystallization of
methoxymethyl ketone 2, a 5-fold molar excess of the inhibitor
was added to the enzyme solution and allowed to stand
overnight before setting up crystals. The aldehyde 1 complex
crystals were formed by soaking a native crystal with a 5-fold
molar excess of the inhibitor in the crystallization droplet.
X-r a y Meth od s a n d Refin em en t. The native enzyme and
inhibitor complexes crystallized in space group C2 with one
molecule in the asymmetric unit. Data were collected on a
Siemens area detector. The cell dimensions and data collection
statistics are listed in Table 1.
The structure of the native papain was solved first by
molecular replacement using the program XPLOR,21 using the
previously solved structure of papain (code 1PIP)8 as a starting
model. The cross-rotation function was calculated using data
from 10 to 4 Å with a radius of integration of 38 Å resulting
in the highest peak at 3.8σ. The translation search was
computed using data from 10 to 3.5 Å and gave a peak at 32σ.
The resulting model yielded an R-factor of 0.33. Several cycles
of refinement of the native papain were conducted using rigid
body, simulated annealing, and positional and temperature
(16) Pickersgill, R. W.; Harris, G. W.; Garman, E. Structure of
Monoclinic Papain at 1.60 Å Resolution. Acta Crystallogr., Sect.
B 1992, 48, 59-67.
(17) Zhao, B.; J anson, C. A.; Amegadzie, B. Y.; D’Alessio, K.; Griffin,
C.; Hanning, C. R.; J ones, C.; Kurdyla, J .; McQueney, M.; Qiu,
X.; Smith, W. W.; Abdel-Meguid, S. S. Crystal Structure of
Human Osteoclast Cathepsin K Complex With E-64. Nature
Struct. Biol. 1997, 4, 109-111.
(18) McGrath, M. E.; Klaus, J . L.; Barnes, M. G.; Bromme, D. Crystal
Structure of Human Cathepsin K Complexed with A Potent
Inhibitor. Nature Struct. Biol. 1997, 4, 105-109.

4576 J ournal of Medicinal Chemistry, 1998, Vol. 41, No. 23
LaLonde et al.
(19) Yamashita, D. S.; Smith, W. W.; Zhao, B.; J anson, C. A.;
Tomaszek, T. A.; Bossard, M. J .; Levy, M. A.; Marquis, R. W.;
Oh, H.-J .; Ru, Y.; Carr, T. J .; Thompson, S. K.; Ijames, C. F.;
Carr, S. A.; McQueney, M.; D’Alessio, K. J .; Amegadzie, B. Y.;
Hanning, C. R.; Abdel-Meguid, S. S.; DesJ arlais, R. L.; Gleason,
J . G.; Veber, D. F. Structure And Design of Potent and Selective
Cathepsin K Inhibitors J . Am. Chem. Soc. 1997, 119, 11351-
11352.
(20) Thompson, S. K.; Halbert, S. M.; Bossard, M. J .; Tomaszek, T.
A.; Levy, M. A.; Meek, T. D.; Zhao, B.; Smith, W. W.; J anson, C.
A.; D′Alessio, K. J .; McQueney, M.; Abdel-Meguid, S. S.; Des-
J arlais, R. L.; Briand, J .; Sarkar, S. K.; Huddleston, M.; Ijames,
C.; Carr, S.; Garnes, K. T.; Shu, A.; Heys, J . R.; Bradbeer, J .;
Zembryki, D.; Lee-Rykaczewski, L.; J ames, I. E.; Lark, M.;
Drake, F. H.; Gowen, M.; Gleason, J . G.; Veber, D. F. Design of
Potent and Selective Human Cathepsin K Inhibitors Which Span
the Active Site. Proc. Natl. Acad. Sci. U.S.A. 1997, 94, 14249-
14254.
(22) Sack, J . S. CHAIN - A Crystallographic Modeling Program. J .
Mol. Graph. 1988, 6, 224-225.
(23) Kraulis, P. J . MOLSCRIPT: A Program to Produce Both
Detailed and Schematic Plots of Protein Structures. Appl.
Crystallogr. 1991, 24, 946-950.
(24) Votta, B.; Levy, M. A.; Badger, A.; Bradbeer, J .; Dodds, R. A.;
J ames, I. E.; Thompson, S.; Bossard, M. J .; Carr, T.; Connor, J .
R.; Tomaszek, T. A.; Szewczuk, L.; Drake, F. H.; Veber, D. F.;
Gowen, M. Peptide Aldehyde Inhibitors of Cathepsin K Inhibit
Bone Resorption Both In Vitro and In Vivo. J . Bone Miner. Res.
1997, 12, 1396-1406.
(25) J ones, D. M.; Atrash, B.; Teger-Nilsson, A. C.; Gyzander, E.;
Deinum, J .; Szelke, M. Design and synthesis of thrombin
inhibitors. Lett. Pept. Sci. 1995, 2, 147-154.
(26) J ones, D. M.; Atrash, B.; Teger-Nilsson, A. C.; Gyzander, E.;
Deinum, J .; Szelke, M. Inhibitors of thrombin containing argi-
nine ketone derivatives. In Pept. 1994, Proc. Eur. Pept. Symp.,
23rd; Maia, H. L. S., Ed.; ESCOM: Leiden, 1995; pp 899-900.
(21) Brunger, A. T.; Kuriyan, J .; Karplus, M. Crystallographic R
Factor Refinement by Molecular Dynamics. Science 1987, 235,
458-460.
J M980249F