Structure-based optimisation of 2-aminobenzylstatine derivatives: Potent and selective inhibitors of the chymotrypsin-like activity of the human 20S proteasome

Article abstract of DOI:10.1016/S0960-894X(02)00178-6

We have identified 2-aminobenzylstatine derivatives that inhibit non-covalently the chymotrypsin-like activity of the human 20S proteasome. A structure-based optimisation approach has allowed us to improve the potency of this structural class of proteasome inhibitors from micromolar to nanomolar level. The new derivatives showed good selectivity against the trypsin-like and post-glutamyl-peptide hydrolytic activities of this enzyme.

Full text of DOI:10.1016/S0960-894X(02)00178-6

Bioorganic & Medicinal Chemistry Letters 12 (2002) 1331–1334
Structure-Based Optimisation of 2-Aminobenzylstatine
Derivatives: Potent and Selective Inhibitors of the Chymotrypsin-
Like Activity of the Human 20S Proteasome
Pascal Furet,* Patricia Imbach,* Peter Fuerst, Marc Lang, Maria Noorani,
Johann Zimmermann and Carlos Garcıa-Echeverrıa*
Oncology Research, Novartis Pharma Inc., CH-4002 Basel, Switzerland
Received 6 February 2002; accepted 22 March 2002
Abstract—We have identified 2-aminobenzylstatine derivatives that inhibit non-covalently the chymotrypsin-like activity of the
human 20S proteasome. A structure-based optimisation approach has allowed us to improve the potency of this structural class of
proteasome inhibitors from micromolar to nanomolar level. The new derivatives showed good selectivity against the trypsin-like
and post-glutamyl-peptide hydrolytic activities of this enzyme. # 2002 Elsevier Science Ltd. All rights reserved.
Introduction
of the 2-aminobenzylstatine core structure which is
assumed to bind in the S3 pocket of the enzyme. We then
combined the best substituents obtained for this part of
the inhibitors with a modification of the N-terminal group
aimed at increasing the strength of an aromatic stacking
interaction with an accessory pocket of the active site.
The 20S proteasome, which is the catalytic core of the
26S proteasome,1 is a threonine protease that exhibits at
least three distinct peptidase activities: chymotrypsin-like,
trypsin-like, and post-glutamyl-peptide hydrolytic activ-
ities.2 Our specific target in the search of novel cytotoxic
and anti-proliferative agents is the chymotrypsin-like
activity of the 20S proteasome. Modulation of this enzy-
matic activity by b-subunit-specific proteasome inhibitors
may convey an anti-tumor effect by induction of cell
cycle arrest and apoptosis in tumor cells.3,4
Synthesis and Biological Assays
Compounds 1–19 were prepared according to known
methods.8ꢀ10 The general route for the syntheses of
2-heterosubstituted derivatives of 4-amino-3-hydroxy-
5-phenylpentanoic acid is illustrated for compound 18
in Scheme 1. The ability of these compounds to inhibit
the 20S proteasome was determined in vitro using pur-
ified human 20S proteasome11 and the following
fluorogenic peptides as substrates: Suc-Leu-Leu-Val-
Tyr-AMC (substrate for chymotrypsin-like assay), Boc-
Leu-Arg-Arg-AMC (substrate for trypsin-like assay)
and Z-Leu-Leu-Glu-AMC (substrate for post-gluta-
myl-peptide hydrolytic-like assay). Fluorescence exci-
tation/emission wavelengths were 355 nm/460 nm for
7-amido-4-methyl-coumarin (AMC). The rates of
hydrolysis were monitored by the fluorescence increase
and the initial linear portions of curves were used to cal-
culate the IC50 values (Table 1).
Recently, we have described the identification and the
initial optimisation of a series of 2-aminobenzylstatine
derivatives that inhibit non-covalently the chymo-
trypsin-like activity of the 20S proteasome in an in vitro
enzyme assay (e.g., compound 1, Table 1).5,6 This new
structural class shows good selectivity over the trypsin-
like and post-glutamyl-peptide hydrolytic activities of
the 20S proteasome (for compound 1, IC50>20 mM).
In this letter, we report further efforts to improve the in
vitro 20S proteasome inhibition of this class of com-
pounds by creating new productive interactions with the
X and HC5 subunits of the enzyme.7 Following a mod-
ular approach, we first optimised the binding interac-
tions through diverse substitutions of the aromatic ring
Results and Discussion
*Corresponding authors. E-mail: pascal.furet@pharma.novartis.com
(P. Furet); patricia.imbach@pharma.novartis.com (P. Imbach);
carlos.garcia-echeverria@pharma.novartis.com (C. Garcıa-Echeverrıa).
We have identified by high throughput screening (HTS)
of our in-house compound archive 2-aminobenzylstatine
0960-894X/02/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved.
PII: S0960-894X(02)00178-6

1332
P. Furet et al. / Bioorg. Med. Chem. Lett. 12 (2002) 1331–1334
derivatives (e.g., compound 1, Table 1) that inhibit non-
covalently the chymotrypsin-like activity of the 20S
proteasome in an in vitro assay. This discovery repre-
sents a completely new avenue for medicinal chemistry
because early inhibitors of this enzyme exert their inhi-
bitory activity via adduct formation.12 The inherent
drawbacks of these classical protease inhibitors promp-
ted us to focus our efforts on the optimisation of the
potency of the new non-covalent inhibitors. To accom-
plish this, we have exploited a structural model of the
human proteasome in complex with compound 1.6
To start with, we evaluated the SARs surrounding the
substitutions on the aromatic ring of the 2-amino-
benzylstatine core structure. The structures of such
compounds (2–14) and their IC50 values are shown in
Table 1. According to our model, the benzylamino
group in position 2 of the statine moiety interacts with
the S3 pocket of the enzyme. This pocket is formed by
the side chains of subunit HC5 residues Tyr 135, Ser
151, Asp 153, Ser 157, Gln 159, Asp 161 and Lys 164 on
the one hand and the side chains of subunit X residues
Ala 20, Ala 22, Ala 27, Ser 28, Val 31 and Ala 49 on the
other hand.13 Two hydrogen bond interactions between
this part of the inhibitor and residues of the S3 pocket
are assumed: the first one between the nitrogen atom
and Asp 153 and the second one between the 4-methoxy
substituent of the benzylic group and Ser 151 located at
the bottom of the pocket (Fig. 1). Consistent with this
hypothesis, a substantial drop in activity was observed
with compound 2 which lacks the 4-methoxy sub-
stituent. Examination of the model suggested that two
other residues of the S3 pocket, Ser 157 and Tyr 135,
could be targeted for hydrogen bonding. In modeling
experiments,14 a methoxy substituent introduced either
in position 2 or 3 of the phenyl ring could accept a
hydrogen bond from Ser 157 while Tyr 135 could
donate a hydrogen bond to a methoxy substituent
Scheme 1. Synthesis of compound 18. Conditions: (a) (COCl)2,
DMSO, CH2Cl2, E tN, ꢀ55 ꢁC rt; (b) Ph3P=CHCO2Et, toluene,
ꢂ
3
80 ꢁC; (c)ꢁ MCPBA, CH2Cl2, rt; (d) 3,4,5-trimethoxybenzylamine,
.
EtOH, 70 C; (e) 1 N LiOH, THF, rt; (f) HCl (S)-Val-(2-hydroxy-4-
methoxy)benzylamine, TPTU, DIEA, DMF, rt; (g) 4 N HCl in diox-
ane, rt; (h) Na-Boc-L-Tleu-OH, TPTU, DIEA, DMA, rt; (i) 4 N HCl
in dioxane, rt; (j) 1-Naphthylacetic acid, TPTU, DIEA, DMA, rt.
Table 1. Inhibition of the chymotrypsin-like activity of the 20S pro-
teasome by 2-aminobenzylstatine derivativesa
Entry
R1
R2
IC50 (mM)
1
2
3
4
5
6
7
8
CO2CH2Ph
CO2CH2Ph
CO2CH2Ph
CO2CH2Ph
CO2CH2Ph
CO2CH2Ph
CO2CH2Ph
CO2CH2Ph
CO2CH2Ph
CO2CH2Ph
CO2CH2Ph
CO2CH2Ph
CO2CH2Ph
CO2CH2Ph
4-OCH3
H
2-OCH3
0.9
>20
>20
1.50
1.60
0.15
0.19
0.20
0.10
0.05
>20
1.60
>20
0.6
3-OCH3
2,4-(OCH3)2
3,4-(OCH3)2
3,5-(OCH3)2
2,4,5-(OCH3)3
2,3,4-(OCH3)3
3,4,5-(OCH3)3
4-C6H5
4-CH(CH3)2
4-OC6H5
4-N(CH3)2
4-OCH3
9
10
11
12
13
14
15
16
17
18
19
C(O)CH2(1-Naphthalene)
C(O)CH2(1-Naphthalene)
C(O)CH2(1-Naphthalene)
C(O)CH2(1-Naphthalene)
C(O)CH2(1-Naphthalene)
0.100
0.029
0.018
0.007
0.160
2,4,5-(OCH3)3
2,3,4-(OCH3)3
3,4,5-(OCH3)3
4-N(CH3)2
aThe IC50 value is the concentration of inhibitor at which the rate of
the chymotrypsin-like activity of the 20S proteasome catalyzed
hydrolysis of the substrate Suc-Leu-Leu-Val-Tyr-AMC is reduced at
50%.
Figure 1. Model of 1 (yellow) bound to the proteasome X/HC5 site.
View of the S3 pocket. Hydrogen bonds are shown in magenta.

P. Furet et al. / Bioorg. Med. Chem. Lett. 12 (2002) 1331–1334
1333
placed in position 5. Hence, compounds 3 and 4 were
synthesized, followed by compounds 5–10 in which
combinations of methoxy substituents simultaneously
targeting two or three of residues Ser 151, Ser 157 and
Tyr 135 were incorporated. The relative potencies of
compounds 4, 6, 7 and 10 show that this strategy was
quite successful with the 3-methoxy substitution. The
gain in potency obtained with 4 compared to 2 and with
7 compared to 4 suggests that the targeted hydrogen
bonds with Ser 157 and Tyr 135 were achieved. In
agreement with our concept, the most potent compound
of the series 10 resulted from the addition of a third
methoxy group in position 4 targeting Ser 151. In con-
trast, no beneficial effect was observed with the 2-methoxy
substitution. In the mono-substituted variation, com-
pound 3 is not more active than the unsubstituted com-
pound 2 while the di- and tri-substituted derivatives 5, 8
and 9 are equipotent to their analogues lacking the
2-substituent: 1 and 6. A careful examination of the
model suggests that in order to form a hydrogen bond
with Ser 157, the 2-methoxy group has to come to a short
distance of the C-terminal phenolic moiety of the inhibitor
which occupies the S1 pocket. Thus, in this case an intra-
molecular steric hindrance may cancel out any beneficial
effect obtained by forming a hydrogen bond with Ser 157.
Figure 2. Model of proteasome-compound 18 complex. Hydrogen
bonds are indicated as magenta lines.
The modifications introduced in compound 1 had no
effect on its 20S proteasome speficity profile. The highly
potent derivatives 16–18 still show good selectivity over
the trypsin-like and post-glutamyl-peptide hydrolytic
activities of the 20S proteasome (for all compounds,
IC50>20 mM).
In summary, our model of the human 20S proteasome
in conjunction with a modular chemistry approach has
allowed us to improve the proteasome chymotrypsin-
like inhibitory activity of the 2-aminobenzylstatine
compound class from micromolar down to nanomolar
potency. The reported compounds are the most potent
non-covalent inhibitors of the human proteasome
described to date.15 They open a new avenue for further
investigation of the proteasome as a therapeutic target
in oncology.
Parallel to our efforts to establish hydrogen bond inter-
actions, we explored the possibility to enhance potency
by creating additional van der Waals interactions in the
S3 pocket. The hydrocarbon parts of the side chains of
Gln 159 and Lys 164 were the most appropriate protein
atoms to target in this respect according to the model.
Compounds 11–14 were synthesized following this con-
cept. For analogues 11 and 13, we were counting on
some flexibility of the residues of the bottom of the
pocket since in the model their bulky 4-substituents
were causing a few steric clashes. The lack of activity of
these analogues put in perspective with the substantial
potency of 12, which bears a smaller isopropyl group,
clearly defines the tolerance of the S3 pocket in terms of
the size of the substituent in position 4. The higher
potency of 14 compared to 12 can be ascribed to the
ability of the dimethylamino susbstituent to form a
hydrogen bond with Ser 151 in addition to making
hydrophobic contacts with Gln 159 and Lys 164.
Acknowledgements
We thank D. Arz, R. Wille, V. von Arx, V. Huy Luu, J. M.
Groell, W. Beck and E. Boss for technical assistance.
References and Notes
1. Tanaka, K. J. Biochem. 1998, 123, 195.
2. Orlowski, M.; Cardozo, C.; Michaud, C. Biochemistry
1993, 32, 1563.
A key finding of our earlier work was the fact that
replacement of the benzyloxycarbonyl group of com-
pound 1 by a group derived from the coupling of
naphthalen-1-yl-acetic acid at the N-terminus improved
potency by one order of magnitude (see compound 15 in
Table 1).5 This modification was designed to increase
the strength of a postulated stacking interaction of the
N-terminal group with residue Tyr 133 of subunit HC5
which forms a small accessory hydrophobic pocket
together with Tyr 33 and Pro 131.6 Consistent with this
finding, we observed an increase in potency of similar
magnitude when the same modification was introduced
in compounds 8–10 and 14 to give derivatives 16–18 and
19. Thus, with an IC50 value of 7 nM, 18 turned out to
be the most potent compound of the series reported in
this study. A model of its complex with the X/HC5 site
of the proteasome is shown in Figure 2.
3. For recent reviews on this target and 20S proteasome inhi-
bitors, see: (a) Rivett, A. J.; Gardner, R. C. J. Pept. Sci. 2000,
6, 478. (b) Dou, Q. P.; Nam, S. Exp. Opin. Ther. Pat. 2000, 10,
1263. (c) Adams, J.; Palombella, V. J.; Elliott, P. J. Invest. New
Drugs 2000, 18, 109. (d) Murray, R. Z.; Norbury, C. Anti-
Cancer Drugs 2000, 11, 407. (e) Dou, Q. P.; Li, B. Drug Resist.
Updates 1999, 2, 215. (f) Elliott, P. J.; Adams, J. Curr. Opin.
Drug. Disc. Dev. 1999, 2, 484. (g) Kalogeris, T.; Gray, L.;
Laroux, F. S.; Cockrell, A.; Fuseler, J.; Conner, E. M.; Brand,
S.; Grisham, M. B. Exp. Opin. Invest. Drugs 1999, 8, 1397. (h)
Lee, D. H.; Goldberg, A. L. Trends Cell Biol. 1998, 8, 397.
4. A dipeptide boronic acid analogue (PS-341) is currently
under clinical evaluation in advanced cancer patients. For
additional data on this compound, see: Adams, J.; Palombella,
V. J.; Sausville, E. A.; Johnson, J.; Destree, A.; Lazarus, D. D.;
Maas, J.; Pien, C. S.; Prakash, S.; Elliott, P. J. Cancer Res.
1999, 59, 2615.

1334
P. Furet et al. / Bioorg. Med. Chem. Lett. 12 (2002) 1331–1334
5. Garcıa-Echeverrıa, C.; Imbach, P.; France, D.; Furst, P.;
Lang, M.; Noorani, M.; Scholz, D.; Zimmermann, J.; Furet,
P. Bioorg. Med. Chem. Lett. 2001, 11, 1317.
6. Furet, P.; Imbach, P.; Furst, P.; Lang, M.; Noorani, M.;
Zimmermann, J.; Garcıa-Echeverrıa, C. Bioorg. Med. Chem.
Lett. 2001, 11, 1321.
12. To date, several classes of 20S proteasome inhibitors have
been described: peptide aldehydes, peptide a-keto aldehydes
(glyoxals), peptide a0,b0-epoxyketones, peptide vinyl sulfones,
peptide indanones, peptide boronic acids (e.g., PS-341, see ref
4), bifunctional inhibitors or natural products (e.g., lactacys-
tin, epoxomicin, TMC-95A-D or UCK 14A2).
7. The active site of the chymotrypsin-type activity of the
human 20S proteasome is formed by the association of these
two b-subunits.
8. Billich, A.; Scholz, D.; Charpiot, B.; Gstach, H.; Lehr, P.;
Peichi, P.; Rosenwirth, B. Antiv. Chem. Chem. 1995, 6, 327.
9. Lehr, P.; Billich, A.; Charpiot, B.; Ettmayer, P.; Scholz, D.;
Rosenwirth, B.; Gstach, H. J. Med. Chem. 1996, 39, 2060.
10. Scholz, D.; Billich, A.; Charpiot, B.; Ettmayer, P.; Lehr,
P.; Rosenwirth, B.; Schreiner, E.; Gstach, H. J. Med. Chem.
1994, 37, 3079.
13. The amino acid numbering in the sequence of subunit X
starts at the N-terminal catalytic threonine (Thr 1). For HC5,
we use the numbering of entry P20618 of the Swiss-Prot data-
base.
14. These modeling experiments were performed in Macro-
model using the AMBER*/H2O/GBSA force field for energy-
minimisation. Macromodel: Mohamadi, F.; Richards, N. G. J.;
Guida, W. C.; Liskamp, R.; Lipton, M.; Caufield, C.; Chang, G.;
Hendrickson, T.; Still, W. C. J. Comput. Chem. 1990, 11, 440.
15. Very recently, it has been demonstrated by X-ray crystal-
lography that the natural product TMC-95A is also a non-
covalent inhibitor of the proteasome: Groll, M.; Koguchi, Y.;
Huber, R.; Kohno, J. J. Mol. Biol. 2001, 311, 543.
11. The 20S proteasome was purified in-house from human
red blood cells by Fast Protein Liquid Chromatography-sys-
tem (Amersham/Pharmacia, Uppsala/Sweden).