C1 and N5 derivatives of cerpegin: Synthesis of a new series based on structure-activity relationships to optimize their inhibitory effect on 20S proteasome

Article abstract of DOI:10.1016/j.bmcl.2013.02.079

Thirty-two new derivatives of cerpegin (1,1,5-trimethylfuro[3,4-c]pyridine- 3,4-dione) were designed and synthesized in high yield by a new method, combining several C¹ and N⁵ substituents. All compounds were tested for their inhibitory effect on the CT-L, T-L and PA proteolytic activities of a purified mammalian 20S proteasome. Only one molecule inhibited both CT-L and PA activities. Sixteen molecules specifically inhibited PA at the micromolar range, out of which fourteen had IC₅₀ values around 5 μM and two had IC₅₀ values closer to 2 μM. Except in one case, neither calpain I nor cathepsin B was inhibited. In silico docking suggests a unique mode of binding of the most efficient compounds to the β1 catalytic site (PA activity) in relation to the chemical nature of C¹ substituents.

Full text of DOI:10.1016/j.bmcl.2013.02.079

Bioorganic & Medicinal Chemistry Letters 23 (2013) 2696–2703
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
C1 and N5 derivatives of cerpegin: Synthesis of a new series based
on structure–activity relationships to optimize their inhibitory effect
on 20S proteasome
Anna Hovhannisyan b, , The Hien Pham a, , Dominique Bouvier c, Lixian Qin a, Gagik Melikyan b,
Michèle Reboud-Ravaux a, Michelle Bouvier-Durand a,
⇑
a Enzymologie Moléculaire et Fonctionnelle, UR4-Université de Paris 6, UPMC-Sorbonne Universités, Case Courrier 256, 7 Quai Saint Bernard, F 75252 Paris Cedex 05, France
b Department of Organic Chemistry, Yerevan State University, A. Manoogian Str. 1, 0025 Yerevan, Armenia
c Atelier de Bioinformatique, Université de Paris 6, UPMC-Sorbonne Universités, Case Courrier 1202, 4 Place Jussieu, F 75252 Paris Cedex 05, France
a r t i c l e i n f o
a b s t r a c t
Article history:
Thirty-two new derivatives of cerpegin (1,1,5-trimethylfuro[3,4-c]pyridine-3,4-dione) were designed and
synthesized in high yield by a new method, combining several C1 and N5 substituents. All compounds
were tested for their inhibitory effect on the CT-L, T-L and PA proteolytic activities of a purified mamma-
lian 20S proteasome. Only one molecule inhibited both CT-L and PA activities. Sixteen molecules specif-
Received 23 December 2012
Revised 15 February 2013
Accepted 18 February 2013
Available online 27 February 2013
ically inhibited PA at the micromolar range, out of which fourteen had IC50 values around 5
lM and two
had IC50 values closer to 2 M. Except in one case, neither calpain I nor cathepsin B was inhibited. In silico
l
Keywords:
Alkaloid cerpegin
docking suggests a unique mode of binding of the most efficient compounds to the b1 catalytic site (PA
activity) in relation to the chemical nature of C1 substituents.
Furo[3,4-c]pyridine-3,4-dione
Benzenesulfanilamide
Proteasome inhibitors
Selective inhibition of PA activity
In silico docking
Ó 2013 Elsevier Ltd. All rights reserved.
A number of naturally occurring molecules and their derivatives
have been shown to inhibit the proteolytic activity of proteasome
(reviewed in1–3). The natural alkaloid cerpegin enters this case. In-
deed, the cerpegin scaffold (1,1,5-trimethylfuro[3,4-c]pyridine-3,4-
dione) has recently been used to develop series of N5 derivatives
with good performances as selective inhibitors of the post-acid
activity (PA) of the 20S proteasome.4 The 20S proteasome is the
multiproteolytic core of the more complex 26S proteasome, work-
ing downstream the ubiquitinylation/deubiquitinylation steps nec-
essary for the recognition of cell substrates.1–3,5–9 The 20S complex
(post-acid or caspase-like) are found in the so-called b1 subunits
(each b1 subunit belonging to a separate heptameric b ring), two
chymotrypsin-like sites (CT-L, located on b5 subunits) and two
trypsin-like sites (T-L, located on b2 subunits) The six catalytic sub-
units represent the main drug targets of the proteasome. Each type
of catalytic site possesses a threonine residue at position one capa-
ble of nucleophilic attack on the substrates. However, the activities
can be distinguished in vitro by their specificity towards amino
acids adjacent to the scissile peptide bonds of specific synthetic
substrates. In brief, cleavage occurs on the C-side of mainly acidic
(PA site), basic (T-L site) or hydrophobic (CT-L site) residues.10
Three similar types of catalytic subunits (b1i, b2i and b5i) with
similar activities arise in eukaryotic cells after induction by biolog-
ical stimuli. They are assembled in a 20S core particle in the so-
called immunoproteasome in place of the constitutively expressed
b1, b2 and b5 subunits. In this work, only the constitutive fraction
was used.
comprises four stacked heptameric rings (a1-7, b1-7, b1-7 and a1-
7) forming a cylindrical structure. The different proteolytic activi-
ties are confined to the two inner b rings: two PA active sites
Abbreviations: DMF/DMA, dimethylformamide/dimethylacetal; DMSO, dimethyl
sulfoxide; 20S, 20S proteasome catalytic core; CT-L, chymotrypsin-like; T-L,
trypsin-like; PA, post-acid or caspase-like; AMC, 7-amino-4-methyl-coumarin;
Ò-NA, Ò-naphthylamide; PDB, Protein Data Bank.
Covalent inhibitors including peptide mimetics and non-pep-
tidic molecules have been first described and represent the largest
class of molecules known to target the 20S catalytic core.2,11–13
Non-covalent inhibitors have been more recently developed.13–19
Dipeptide boronate bortezomib (Velcade or PS341) belongs to
the class of covalent inhibitors. It was the first to be used as a
⇑
Corresponding author. Tel.: +33 1 44 27 69 52; fax: +33 1 44 27 51 40.
E-mail addresses: annahovh@gmail.com (A. Hovhannisyan), bsthehien@yahoo.-
com (T.H. Pham), dbouvier@snv.jussieu.fr (D. Bouvier), lixian.qin@free.fr (L. Qin),
gagikmelikyan@yahoo.com (G. Melikyan), Michele.Reboud@upmc.fr (M. Reboud-
Ravaux), mbouvier@snv.jussieu.fr (M. Bouvier-Durand).
These authors contributed equally to this work.
0960-894X/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.bmcl.2013.02.079

A. Hovhannisyan et al. / Bioorg. Med. Chem. Lett. 23 (2013) 2696–2703
2697
prescription drug to treat multiple myeloma and mantle lym-
phoma diseases.20,21 Acquired resistance to, and adverse effects
of the chemotherapy making use of bortezomib represent practical
difficulties.22 Thus, second-generation inhibitors have been devel-
oped: CEP-18770 and MLN9708, two other peptide-boronates, car-
filzomib, a tetrapeptide epoxyketone, and salinosporamide A, a b-
lactone.23 All gained intensive clinical studies and carfilzomib
(Kyprolis or PR-171) has been recently approved by FDA for the
treatment of multiple myeloma (MM).
of N5 derivatives with a dimethyl at C1 has been developed in the
laboratory.4,25 A number of them specifically inhibit PA activity at
the micromolar range (7 molecules have an IC50 of about 5–6 l
M).4
Within this range, seven N5 substituents were identified as good
candidates to promote PA inhibition by the compounds to which
they belong (Fig. 1A).
For practical reasons, out of the seven N5 substituents associ-
ated with acceptable IC50 values, only four (Fig. 1A, 10–40) were re-
tained in the present study to serve as a basis for the synthesis of
new cerpegin derivatives (Fig. 1B) with the aim to further optimize
inhibition. A selection of R1, R2 substituents at C1 (including spiro-
compounds) was done (Fig. 1B). We report here two new series of
cerpegin derivatives (here generically called compounds III) classi-
fied according to the nature of their N5 moiety: purely aromatic or
aliphatic displaying or not an aromatic moiety (substituted
aliphatic).
The natural product salinosporamide A (SalA) is a potent inhib-
itor of all three proteasome activities whose lactone is attacked by
the nucleophilic Thr1 side chain, leading to formation of a highly
stable acyl-enzyme.24 The
c-lactone ring of cerpegin (Fig. 1A, left)
is also susceptible to nucleophilic attack by Thr1 and a first series
The first series (1–15) mainly combined the aromatic benzene-
sulfonamides at N5 with either aliphatic or aliphatic/aromatic R1,
or spiro-fused R1/R2 substituents (Fig. 1B). Our interest towards
benzenesulfonamides comes from their interesting inhibitory ef-
fect on 20S proteasome as previously reported4 (Fig. 1A) and from
the published properties of sulfonamide-substituted heterocyclic
systems.26 Moreover, sulfone-containing heterocycles have suc-
cessfully been tested for proteasome inhibition.27 The second ser-
ies of derivatives (16–32) combined one of the three other most
efficient substituents at position N5 (Fig. 1B) with the R1 or R1/R2
substituents described for the first series (Fig. 1B).
The new furo[3,4-c]pyridine-3,4-dione derivatives substituted
at N5 and C1 positions with aromatic and/or aliphatic substituents
on both the pyridinone and furanone rings were prepared as out-
lined in Scheme 1.
The starting lactones I are mostly commercially available and
could be synthesized by condensation of the corresponding ketoal-
cohols with diethyl malonate in the presence of catalytic amount
of sodium ethylate.28
A
R
R = (CH2)2C6H4SO2NH2
R = CH2CH(OH)CH2N(C2H5)2
R = CH2CH(OH)C6H5
N5
O
R = CH2CH(OH)CH2O H
1
O
O
4
5
N
derivatives of cerpegin
Cerpegin : R = CH3
B
R
N5
O
R1
R2
1
O
O
Second series of compounds III
(16 - 32)
First series of compounds III
(1 - 15)
Aliphatic R
Aromatic R
SpecialinteresthasalsobeenpaidtocompoundswithR1/R2 alicy-
clic (cyclohexane) and heterocyclic (tetrahydropyran) spiro-struc-
tures (Fig. 1B, bottom right). Such spiro-structures were obtained
by the same method,28 starting with 1-(1-hydroxy-cyclohexyl)-eth-
anone and 1-(4-hydroxy-2,2-dimethyl-tetrahydropyran-4-yl)-eth-
anone, respectively. In the case of 1-(4-hydroxy-2,2-dimethyl-
tetrahydropyran-4-yl)-ethanone new spiro-lactone Ih was obtained
(Supplementary data).
R = C6H4SO2NH2,
R = C6H4SO2NHR
R = CH2CH(OH)CH2N(C2H5)2 ,
R = CH2CH(OH)C6H5,
R = CH2CH(OH)CH2((OH),
or R = (**)
(
see Table 1),
or R = (*)
Combined with
or
Spiro-fused R1/R2
Separate R1/R2
R1 = CH3,
R
N
R1 = CH2CH3,
O
O
N
R1 = CH2CH(CH3)2,
R1 = CH2C6H5,
or R1 = (CH2)2C6H5
O
O
O
R1
R2
R1
R2
R1
R2
i
ii / iii
O
O
O
O
O
O
O
R2 = CH3
Ia-h
IIa-h
III 1-32
R1, R2
=
Me, Me
(a)
(b)
(c)
(d)
(e)
(f)
Et, Me
Figure 1. General presentation of the C1 and N5 derivatives of cerpegin synthesized
in this study: (A) structure of cerpegin and formulae of the main N5 substituents (R)
already described in4 for 1,1-dimethyl cerpegin derivatives. Primed numbers on the
right are used in the text to refer to these 1,1-dimethyl derivatives. (B) New
derivatives of cerpegin (compounds III) combining some of the most performant R
substituents as defined in4 at position N5 and different R1 or spiro-fused R1/R2
substituents at position C1. By convention, in the case of separate R1/R2, R1 is the
variable substituent and R2 is always a methyl group. In the case of spiro-fused R1/
R2, R1 and R2 are forming a cyclohexyl ring or a tetrahydropyran ring. (⁄) Another N5
i-Bu, Me
Ph, Me
Bz, Me
PhEt, Me
R1/R2
(CH2)5
=
(g)
(h)
(CH2)2O(CH2)2
Scheme 1. Synthesis of new furo[3,4-c]pyridine-3,4-dione compounds III, 1–32.
Reagents and conditions: (i) for compound I, dimethylformamide dimethylacetal
(DMF/DMA), xylene, reflux 3 h, 75–89% yield; (ii) for compounds III, 1–15: II,
aromatic amines (1:1 molar ratio), AcOH, reflux 15 h. (iii) for compounds III, 16–32:
II, NH2R, xylene, reflux 15 h, typical yields: 75–90%.
substituent was included in this series: 5-amino-3-methyl-1-phenylpyrazole. (⁄⁄
)
Other N5 substituents were included in this series: 2-(4-chloro)-phenethyl (23 and
24), 3-morpholinopropyl (25–29), 2-hydroxy-3-morpholinopropyl (30), thiourea
(31), as well as 1-phenyl-substituted cerpegin (32).

2698
A. Hovhannisyan et al. / Bioorg. Med. Chem. Lett. 23 (2013) 2696–2703
Condensation of lactones I with DMF/DMA by refluxing in anhy-
Compounds 7–9 with the respective N-4-acetyl, benzoyl, carb-
drous xylene (Scheme 1, i) led to the corresponding dimethylami-
novinyl derivatives II. Starting from II, two distinct pathways
yielded two series of N5-substituted furopyridine structures (com-
pounds III) in one step by interaction with either aromatic
(Scheme 1, ii) or aliphatic (Scheme 1, iii) amines. Experimental de-
tails for the synthesis of II are given in Supplementary data. The
first series of compounds III, mainly comprised of N5-benzenesul-
fonamide derivatives (1–14, see Table 1), was synthesized with
high yield from II (Scheme 1, ii) and 4-aminobenzenesulfonamides
in glacial acetic acid (Table 1 and Supplementary data).
Compounds 1–14 correspond to the first R entry in Figure 1B
(left). Compounds 1–6 are derivatives of sulfanilamide with the
same N5-substituent. Compounds 1–5 have different R1 substitu-
ents so as to make a decision about the influence of R1 on the inves-
tigated activities. Compound 6 possesses unique spiro-fused R1/R2.
Compounds 7–14 have two methyl groups as R1, R2 and can be
viewed as N-substituted sulfonamide derivatives of compound 1.
For comparison, compound 15 with N5 aromatic heterocyclic
substituent was obtained by condensation with 5-amino-3-
methyl-1-phenylpyrazole in glacial acetic acid (Table 1). For the
second series of compounds III (16–32, see Table 2), the cyclocon-
densation and insertion of an aliphatic-substituted group at N5 of
the pyridinone ring was performed in the xylene ambience follow-
ing step iii (Scheme 1).
amidoyl benzenesulfonamide group in N5 position were at least
as efficient as 3, 2 and 1 (IC50 around 5–6 lM). A dramatic loss of
effect was observed for compounds 10–14 with the respective
N-4-pyridinyl-, pyrimidinyl-, (1-phenyl-pyrazolyl-), (3,4-dimethy-
lisoxazolyl-), thiazolyl-benzenesulfonamide, that is, heterocyclic-
substituted benzenesulfonamides at N5, with practically no
inhibition with 100 lM compounds 12 and 14. Compound 15 is
not a sulfonamide but bears the aromatic 3-methyl-1-phenyl-1H-
pyrazol-4-yl substituent and was not inhibitor on the three activi-
ties of the 20S proteasome. Neither calpain I nor cathepsin B was
inhibited by these compounds (1–15).
Results obtained with the second series of compounds III (16–
32) are summed up in Table 2.
Once again, none of these new molecules were inhibitors of CT-
L or T-L activities, except for molecule 24 which caused mild CT-L
inhibition (IC50 = 22.5 lM). A number of these molecules (16–22),
designed after compounds 20, 30 and 40 (Fig. 1A), inhibited only
PA within an acceptable micromolar range. Compared with 1,1-di-
methyl derivatives of cerpegin, a significant optimization of inhibi-
tion (2- to 4-fold) was obtained only when R1 was a benzyl or a
phenethyl group. For example, molecule 19, which combines the
3-diethylamino-2-hydroxypropyl substituent at N5 with one phen-
ethyl at C1 (IC50 = 2.5
lM) was 2.5 times more efficient on PA than
the 1,1-dimethyl homolog. The same was true of the comparison
between compound 27 of this study (3-morpholinopropyl
N5-group) and its dimethyl homolog (optimization ꢀ4-fold as
compared to compound 10 in4). In the same way, associating the
2-(4-chloro-phenethyl) N5-group with a benzyl at C1 yielded a
molecule (24), which not only inhibited PA with higher efficiency
(optimization ꢀ3-fold relative to the dimethyl counterpart) but
also inhibited CT-L. No significant improvement of inhibition was
associated with the other types of substituents at C1, including
cases where these substituents were replaced by a cyclohexyl ring
(16 and 21), a tetrahydropyran ring (17) or when R1 was simply a
phenyl group (18 and 26).
The most frequently used substituents were chosen on the
basis of our previous results4 and correspond to R entries in Fig-
ure 1B (right). In spite of their low efficiency when associated
with a dimethyl at C1,4 other N5 groups were also included in
this second series to check for possible improvements with lar-
ger R1 substitutions. These were: 2-(4-chloro)-phenethyl (23
and 24), 3-morpholinopropyl (25–29), 2-hydroxy-3-morpholino-
propyl (30), thiourea (31), as well as 1-phenyl-substituted cerpe-
gin (32).
The combined effects of R1, R2 and R substitutions of com-
pounds III on the inhibition of the activities of the proteasomal
20S catalytic core (CT-L, T-L and PA) were investigated. For practi-
cal reasons, the three proteolytic activities were assayed using 20S
proteasome fractions purified from rabbit erythrocytes. We have
recently reported that the results remain quantitatively the same
when using 20S proteasome fractions from human cells.4 The de-
tailed protocol used to measure the three peptidase activities has
previously been described29 and is summed up thereafter.30 A sys-
tematic search for the selectivity of the inhibitions was performed
using two other proteases: cathepsin B, purified from human hepa-
tocytes and calpain I, purified from human erythrocytes, the activ-
ities of which were measured as described.30 Results were
expressed as IC50 values, which correspond to the inhibitor concen-
tration causing 50% inhibition, relative to controls (without
inhibitor).
As observed for compounds 1–15, none of the compounds stud-
ied in this series (16–32) worked on purified fractions of cathepsin
B and only one (24) worked on calpain I (IC50 = 29.6 lM).
In summary, for these two series of compounds III (1–15) and
(16–32), the inhibitory effects of N5 derivatives of cerpegin upon
PA activity can be optimized by providing them with a large,
hydrophobic and flexible R1 such as the phenethyl group. The spiro
groups initially introduced to assess the effect of ligand stiffening
did not improve PA inhibition relative to a methyl group.
In silico docking experiments were performed to illustrate the
binding mechanisms of compounds 24, 5 and 19 to the proteasome
active sites (Fig. 2). The catalytic chains from the bovine 20S pro-
teasome and their cognate partners were used as targets.31,32 The
choice of this model was made on the basis of the very strong ami-
no acid sequence identity between human and bovine b-type sub-
units.33 For example, the bovine b1 precursor protein is 95%
identical to the human one and the identity reaches 98.5% when
considering only the mature subunit, after precursor autolysis.
Moreover, the few non-identical residues are located at the inter-
faces of b1 with its neighbors and are thus absent from the sub-
strate binding sites. The AudoDock Vina program34 was used for
docking calculations,35 with default parameters. In cases where
asymmetric carbons were present, all corresponding configura-
tions were docked.
Results obtained with the first series of compounds III (1–15)
for the three proteolytic activities of the 20S proteasome are
summed up in Table 1.
In this series designed after the 1,1-dimethylated compound 10
(Fig. 1A), the N5-substituents are mainly benzenesulfonamides (1–
6) and N-derivatives thereof (7–14). As for compound 10 from
which they derive, these compounds III did not inhibit CT-L nor
T-L activities. Only PA activity was inhibited at the micromolar
range for a limited number of compounds (1–9). The best PA inhi-
bition was obtained with compound 5 (IC50 = 2.8
lM), with a phen-
ethyl as R1. Compound 4 with a benzyl group as R1 was a little less
Visual Molecular Dynamics (VMD)36 was used to prepare
molecular pictures.
We observed very similar poses of the inhibitors in the first
ranks (1–3) of results for all the combinations of N5 and C1 substi-
tutions subjected to docking analysis. As already noted with other
derivatives of cerpegin,4 the bicyclic furopyridinone ring fitted into
efficient (IC50 = 3.6 lM). The gradual impact of the chemical nature
of R1 on inhibition in the series from 1 to 6 can be summed up as
follows: 5 (phenethyl) > 4 (benzyl) > 3 (isobutyl) ꢀ 2 (ethyl) > 1
(methyl) ꢀ 6 (cyclohexyl) and is discussed below in the context
of docking results.

A. Hovhannisyan et al. / Bioorg. Med. Chem. Lett. 23 (2013) 2696–2703
2699
Table 1
Inhibition of 20S proteasome by C1 and N5 derivatives of cerpegin (compounds III, 1–15): effects of diverse N5 aromatic substituents (mainly benzenesulfonamides) and C1
substituents
R
N5
R
N5
O
O
Compounds
III
Yield
(%)
Mp
IC50 (μM) or (%)
inhibition at
100 μM
R1
R2
(oC)
1
O
O
O
O
Spiro-Fused R1/ R2
R
R1
R2
CT-L
T-L
PA
O
S
O
84.0
87.2
65.8
67.8
74.4
95.5
68,7
>270
272
ni
ni
ni
ni
ni
ni
ni
ni
6.1 0.3
4.9 0.2
4.6 0.7
3.6 0.2
2.8 0.2
6.6 0.6
5.7 0.3
NH2
NH2
NH2
NH2
NH2
NH2
1
2
3
4
5
6
7
O
S
O
ni
ni
ni
ni
ni
ni
O
S
O
216-8
>260
193
O
S
O
O
S
O
O
S
O
>300
205
O
O
S
O
N
H
O
O
S
O
60.5
52.1
278
310
ni
ni
ni
ni
4.5 0.3
5.9 0.4
N
H
8
9
NH
O
S
O
N
H
NH2
O
S
O
H
N
73.0
88.2
>260
>260
ni
ni
ni
ni
(64%)
(51%)
10
11
N
O
S
O
N
N
H
N
O
S
O
N
89.3
44.7
170
147
ni
ni
ni
ni
(28%)
(51%)
N
H
N
12
13
O
S
O
H
N
N
O
O
S
O
N
H
N
66.1
99.0
>250
245
ni
ni
ni
ni
(22%)
14
15
S
N
N
ni
10.4 0.5
5.5 0.1
ni
ni
ni
ni
4
Cerpegin
1’
23
0.0026
0.021
0.46
SalA
Yield (%) and melting points are indicated for each compound. Inhibitions of 20S proteasome from rabbit erythrocytes are expressed as IC50 values or as % inhibition at 100 lM
(italics in parenthesis). ni = no inhibition. When R1 and R2 are different the synthesis led to racemates. Cerpegin and compound 10 (from4) and salinosporamide A (SalA)23 are
given as reference inhibitors.
a slit in the active site between Thr residues 1 and 21 (Fig. 2A–F),
which appeared as the main hydrogen bond donors in the interac-
tion (dotted lines in Fig. 2B). This placed the inhibitors alongside
the peptide chain of a sequence from residue 46 to residue 49
(Ser-Gly-Ser-Ala in PA). In all cases, the free end of the N5 substitu-
ent was close to the aromatic ring of Tyr114 of the b2 subunit.

2700
A. Hovhannisyan et al. / Bioorg. Med. Chem. Lett. 23 (2013) 2696–2703
Table 2
Inhibition of 20S proteasome by C1 and N5 derivatives of cerpegin (compounds III, 16–32): effects of diverse N5 aliphatic substituents and C1 substituents
R
R
N5
N5
O
O
R1
R2
1
Compounds
III
Yield
(%)
Mp
IC50 (μM) or % inhibition
O
O
(oC)
at
O
O
100 μM
R
R1
R2
Spiro-fused R1/ R2
CT-L
T-L
PA
N
N
89.5
72.1
183
ni
ni
6.7 0.3
OH
OH
16
17
O
175.6
ni
ni
4.1 0.3
N
N
76.2
71.1
54.5
137
95-7
184
ni
ni
ni
ni
ni
ni
4.4 0.2
2.5 0.06
4.8 0.5
OH
OH
18*
19*
20*
OH
57.2
77.9
225
224
ni
ni
ni
ni
11.9 0.4
8.1 0.2
21
22
OH
OH
OH
Cl
Cl
89.4
71.6
83.9
85.1
151
191
143
167
ni
22.5 1.2
ni
ni
ni
ni
ni
12.7 0.5
7 0.5
23
24
25
26
N
N
N
N
O
(31.8%)
56.4 1.8
O
O
O
ni
82.6
92.4
137
175
ni
ni
ni
ni
19.6 1.6
54.7 2.3
27
28
O
N
N
O
O
80.9
168
ni
ni
46.9 3.4
29
83.7
65.5
175
270
ni
ni
ni
(30.7%)
OH
S
30
31
(23.8%)
ni
N
H
NH2
91.4
214
ni
ni
36 4.2
32
4
10.4 0.5
6,.1 0.7
ni
ni
ni
ni
Cerpegin
2’
23
0.0026
0.021
0.46
SalA
Yield (%) and melting points are indicated for each compound. Inhibitions of 20S proteasome from rabbit erythrocytes are expressed as IC50 values or as % inhibition at 100 lM
(italics in parenthesis). ni = no inhibition. Cerpegin and compound 20 (from4) and salinosporamide A (SalA)23 are given as reference inhibitors.
a
Compounds 18–20 were produced as mixtures of diastereoisomers (see Supplementary data).
When the N5 substituent comprised one (Fig. 2C and D) or several
(Fig 2E and F) polar group(s), the poses predicted the possibility of
hydrogen bond formation with the carbonyl oxygen of Thr21 in b1
and/or the side chain alcohol of Ser118 in b2.

A. Hovhannisyan et al. / Bioorg. Med. Chem. Lett. 23 (2013) 2696–2703
2701
A
B
β7 Y30
G129
S130
M116
O
+
4.2 Å
O
H3N
O
N
H
M95
T1
β1
3.3 Å
4.0 Å
O
3.5 Å
O
N
S
HN
OH
O
T21
H
O
3.0 Å
O
H2N
O
S118
β2
Y114
β2
C
D
β7 Y30
G129
S130
3.8Å
O
+
NH3
O
M95
N
H
T1
4.2 Å
3.7 Å
O
O
3.8 Å
O
N
N
β1
HN
OH
2.8 Å
O
OH
T21
3.9 Å
OH
S118
β2
E
Y114
β2
F
Figure 2. In silico docking of compounds 24 (A and B), 5 (C and D) and 19 (E and F) in the PA active site of bovine 20S proteasome. For all three compounds illustrated, the
absolute configuration of the asymmetric carbon C1 is S and that of the alcohol carbon in 19 is R. (A, C, E) Protein chains are shown as solvent accessible surfaces and the
residues and atoms relevant to an enzyme activity are labelled. Inhibitors are displayed as stick models colored by atom type. Each docking mode is shown as six
representative poses from the most populated and lowest-energy cluster. (B) Compound 24 is shown with transparent molecular surface and relevant amino-acids in the b1,
b2 and b7 subunits are displayed as sticks. The dotted lines between the ligand and T1 and T21 represent putative hydrogen bonds. (D, F) Distances reported for the b1
subunit to atoms in 5 and 19, respectively, were measured from the lowest energy pose. Putative hydrogen bonds are shown as blue dashed lines and the predicted
nucleophilic attack on the lactone carbon, by a dashed arrow. Note: the crystal structure of the eukaryotic bovine proteasome (PDB ID: 1IRU)31 was obtained from the Protein
Data Bank (PDB).32

2702
A. Hovhannisyan et al. / Bioorg. Med. Chem. Lett. 23 (2013) 2696–2703
Table 3
Comparison of calculated KD values for compounds 5, 19 and 24, with opposite configurations S and R at C1 and for their respective 1,1-dimethylated analogs 1, 20 and 214
Calculated KD for compounds 5, 19 and 24a
Calculated KD for the 1,1-dimethylated analogs 1, 20, 214
Compound
KD
(
lM)
Compound
KD
(lM)
Compound
KD (lM)
5 S
6–7
20–25
10
5 R
20–24
200
10
1
20–25
170
30
19 Sb
24 S
19 Rb
24 R
20
214
KD values were computed from the binding energies given by AutoDock Vina.
a
The substituents at C1 are a methyl and a phenethyl for compounds 5 and 19, a methyl and a benzyl for 24.
The asymmetric carbon of the hydroxyl group of the N5 substituent is R.
b
The R1 substituents lied in a rather hydrophobic groove running
the inhibitory power of such molecules while preserving their
specificity towards PA. Moreover, it will be necessary to develop
protocols for the purification of the most efficient stereoisomers
where asymmetric carbons are involved. Subunit-selective inhibi-
tors of proteasome are still poorly documented. One example to
into Tyr30 in subunit b7 and lined by the peptide groups of Gly129
and Ser130 on one side and by the side chains of Met residues 95
and 116 of b1 on the other side (Fig. 2). This groove corresponds to
the primed substrate binding or specificity channel of the PA
activity.37
date is a urea-containing peptide epoxyketone derived from
Substitutions at C1 made this carbon asymmetric and two con-
figurations had to be considered.
syringolins.
The C1 and N5 derivatives of cerpegin could consti-
39
tute a convenient set of tools to further analyze the respective roles
of proteasome subunits in cell physiology and pathologies involv-
ing the proteasome.
The absolute configuration at C1 is S for all the molecules shown
in Figure 2. This configuration corresponded to the best docking re-
sults in terms of rank and calculated KD (Table 3) for compounds 5
(Fig. 2C and D) and 19 (Fig. 2E and F). The opposite configuration
(R) gave rise to much higher KD values due to the fact that the R1
substituent was excluded from the binding site (illustrated in
Fig. S1A and B). In contrast, the two stereoisomers of 24 adopted
very similar poses (illustrated in Fig. S1C and D) and the computed
KD values were the same (Table 3). Table 3 also compares the bind-
ing affinities of 1-monosubstituted cerpegin derivatives 5, 19 and
24 with those of their respective 1,1-dimethyl analogs, which
adopted similar position in the b1 binding site (illustrated in
Fig. S2A for 24). KD values were always higher for the dimethyl
compounds than for the monomethyl ones and in some cases
(see for examples 5 and 19) this effect depended on the stereo-
chemistry of the substitution. The consequences of this stereo-
chemistry varied among compounds (Table 3), ranging from no
effect for 24 to a 10-fold increase in KD for 19 and an intermediate
threefold increase in KD for 5. Thus, docking calculations repro-
duced the optimization associated with C1 substitutions already
revealed by in vitro assays (IC50). In addition, the Ki values calcu-
lated from experimental IC50 using the Cheng–Prusoff equation38
were of the same order of magnitude as the KD reported in Table 3.
In the present study none of the tested compounds were pre-
dicted to interact with the S1 pocket. Therefore, the molecular ba-
sis for the PA specificity of these compounds should be searched
elsewhere. We have already pointed out4 the possible involvement
of a tyrosine residue (Tyr114) in the contiguous b2 subunit, which
most often makes favorable contacts with N5 substituents and is
absent from the catalytic site of the two other proteolytic activities.
The present study also suggests that the R1 substituent could play
some part in PA specificity by binding to the primed specificity
channel of the b1 active site, which is the most hydrophobic of
the three b primed binding channels. The binding of 24 to the b5
active site (CT-L) is shown in Figure S3 for comparison.
Acknowledgments
A.H. and T.H.P. received financial supports from the Yerevan
University (A.H.) and from the Vietnamian Ministery for Education
(T.H.P.). This study was also supported by the Universities of Yere-
van (G.M.), Pierre et Marie Curie (UPMC, Paris 6) (M.B.-D., M.R.-R.,
L.Q. and D.B.) and by a Grant from the ‘French Association Against
Myopathies’.
Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.bmcl.2013.02.
079.
References and notes
1. Grawert, M. A.; Groll, M. Chem. Commun. (Camb.) 2012, 48, 1364.
2. Kisselev, A. F.; van der Linden, W. A.; Overkleeft, H. S. Chem. Biol. 2012, 19, 99.
3. Genin, E.; Reboud-Ravaux, M.; Vidal, J. Curr. Top. Med. Chem. 2010, 10, 232.
4. Pham, T. H.; Hovhannisyan, A.; Bouvier, D.; Tian, L.; Reboud-Ravaux, M.;
Melikyan, G.; Bouvier-Durand, M. Bioorg. Med. Chem. Lett. 2012, 22, 3822.
5. Coux, O.; Tanaka, K.; Goldberg, A. L. Annu. Rev. Biochem. 1996, 65, 801.
6. Hershko, A.; Ciechanover, A. Ann. Rev. Biochem. 1998, 67, 425.
7. Borissenko, L.; Groll, M. Chem. Rev. 2007, 107, 687.
8. Marques, A. J.; Palanimurugan, R.; Matias, A. C.; Ramos, P. C.; Dohmen, R. J.
Chem. Rev. 2009, 109, 1509.
9. Weissman, A. M.; Shabek, N.; Ciechanover, A. Nat. Rev. Mol. Cell Biol. 2011, 12,
605.
10. Groll, M.; Heinemeyer, W.; Jâger, S.; Ullrich, T.; Bochtler, M.; Wolf, D. H.; Huber,
R. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 10976.
11. Kisselev, A. F.; Goldberg, A. L. Chem. Biol. 2001, 8, 739.
12. Zhang, S.; Shi, Y., et al J. Mol. Model. 2009, 15, 1481.
13. Beck, P.; Dubiella, C.; Groll, M. Biol. Chem. 2012, 393, 1101.
14. Koguchi, Y.; Kohno, J.; Nishio, K.; Takahashi, T.; Okuda, T.; Ohnuki, S.;
Komatsubara J. Antibiot. 2000, 53, 105.
15. Groll, M.; Goetz, M.; Kaiser, M.; Weyher, E.; Moroder, L. Chem. Biol. 2006, 13,
607.
In conclusion, the two series of cerpegin derivatives synthesized
in this study have been characterized for their selectivity towards
20S proteasome. All molecules but one were inefficient in inhibit-
ing the activity of both calpain I and cathepsin B. Sixteen molecules
specifically inhibited PA activity within the micromolar range, out
16. Basse, N.; Piguel, S.; Papapostolou, D.; Ferrier-Berthelot, A.; Richy, N.; Pagano,
M.; Sarthou, P.; Sobczak-Thepot, J.; Vidal, J. J. Med. Chem. 2007, 50, 2842.
17. Groll, M.; Gallastegui, N.; Maréchal, X.; Le Ravalec, V.; Basse, N.; Richy, N.;
Genin, E.; Huber, R.; Moroder, L.; Vidal, J.; Reboud-Ravaux, M. ChemMedChem
2010, 1701.
18. Blackburn, C.; Gigstad, K. M.; Hales, P.; Garcia, K. M.; Barrett, M.; Liu, J. X.;
Soucy, T. A.; Sappal, D. S.; Bump, N.; Olhava, E. J.; Fleming, P.; Dick, L. R.; Tsu, C.;
Sintchak, M. D.; Blank, J. L. Biochem. J. 2010, 430, 461.
of which 14 had IC50 values around 5
to 2
lM and two IC50 values closer
l
M. Interestingly, the latter two molecules have a phenethyl-
substituted C1 and this substitution gives rise to inhibitors of high-
er potency relative to the 1,1-dimethylated analogs, depending on
the steoreochemistry. These results open new ways for enhancing
19. Marechal, X.; Pujol, A.; Richy, N.; Genin, E.; Basse, N.; Reboud-Ravaux, M.;
Vidal, J. Eur. J. Med Chem. 2012, 52, 322.
20. Adams, J. Nat. Rev. Cancer 2004, 4, 349.

A. Hovhannisyan et al. / Bioorg. Med. Chem. Lett. 23 (2013) 2696–2703
2703
nH
nH
21. Adams, J. Cancer Cell 2003, 5, 417.
22. Orlowski, R. Z.; Kuhn, D. J. Clin. Cancer Res. 2008, 14, 1649.
ðIC þ ½Iꢁ Þ where nH is the Hill number. The KM values of the fluorogenic
50
substrates in our experimental conditions were: 30
77 (Z-LLE-bNA) and 26 (Boc-LRR-AMC). For calpain
cathepsin B activities the respective Suc-LLVY-AMC and Z-RR-AMC substrates
were used. Buffers and temperatures used were as follows: 50 mM Tris, 100
5
l
M (Suc-LLVY-AMC),
23. Macherla, V. R.; Mitchell, S. S.; Manam, R. R.; Reed, K. A.; Chao, T. H.; Nicholson,
B.; Deyanat-Yadzi, G.; Mai, B.; Jensen, P. R.; Fenical, W. F.; Neuteboom, S. T. C.;
Lam, K. S.; Palladino, M. A.; Potts, B. C. M. J. Med. Chem. 2005, 48, 3684.
24. Groll, M.; Huber, R.; Potts, B. C. J. Am. Chem. Soc. 2006, 128, 5136.
25. Communication by Melikyan, G.; Hovhannisyan, A.; Avetisyan, A. at the
Congress of the French Chemical Society «La Chimie du Futur. Le Futur de la
Chimie» SFC 07, Paris, July 16–18, 2007.
4
l
M
6
lM
I
and
lM
CaCl2 and 10 mM DTT (pH 7.2) at 25 °C for calpain I, and 352 mM KH2PO4, 48 mM
Na2HPO4, 1 mM EDTA, 1 mM DTT (pH 6) at 37 °C for cathepsin B. In all assays,
emitted fluorescence was measured using black 96-well microplates and a BMG
Fluostar microplate reader.
26. Haque, W.; Dinkur, J. U.S. Patent No. US 7,459,468 B2, 2008.
31. Unno, M.; Mizushima, T.; Morimoto, Y.; Tomisugi, Y.; Tanaka, K.; Yasuoka, N.;
Tsukihara, T. Structure 2002, 10, 609.
32. Berman, H. M.; Westbrook, J.; Feng, Z.; Gilliland, G.; Bhat, T. N.; Weissig, H.;
Shindyalov, I. N.; Bourne, P. E. Nucleic Acids Res. 2000, 28, 235.
33. UniProt Consortium, Nucleic Acids Res., 2012, 40, D71.
27. Basse, N.; Montes, M.; Marechal, X.; Qin, L.; Bouvier-Durand, M.; Genin, E.;
Vidal, J.; Villoutreix, B. O.; Reboud-Ravaux, M. J. Med. Chem. 2010, 53, 509.
28. Avetisyan, A. A.; Tadevosyan, G. E.; Mangasaryan, C. A.; Matsoyan, S. G.;
Dangyan, M. T. J. Org. Chim. (Rus.) 1970, 5, 962.
29. Formicola, L.; Maréchal, X.; Basse, N.; Bouvier-Durand, M.; Bonnet-Delpon, D.;
Milcent, T.; Reboud-Ravaux, M.; Ongeri, S. Bioorg. Med. Chem. Lett. 2009, 19, 83.
30. CT-L, PA and T-L activities were determined using the respective synthetic
substrates Suc-LLVY-AMC, Z-LLE-bNA and Boc-LRR-AMC. All substrates and
inhibitors were previously dissolved in dimethylsulfoxide (DMSO) whose final
concentration was 2% (v/v). After a pre-incubation period of 15 min with the
34. Trott, O.; Olson, A. J. J. Comput. Chem. 2010, 31, 455.
35. The docking space was limited to a 20 ꢂ 20 ꢂ 20 Å box centered on the N
terminal threonine of the catalytic subunits. Docking modes were selected on
the basis of the distance (3–4 Å) between the carbonyl carbon of the compound
lactone and the hydroxyl oxygen of the N terminal threonine of each subunit.
36. Humphrey, W.; Dalke, A.; Schulten, K. J. Mol. Graphics 1996, 14, 33.
37. Groll, M.; Larionov, O. V.; Huber, R.; de Meijere, A. Proc. Natl. Acad. Sci. U.S.A.
2006, 103, 4576.
38. Cheng, Y.; Prusoff, W. H. Biochem. Pharmacol. 1973, 22, 3099.
39. Van der Linden, W. A.; Willems, L. I.; Shabaneh, T. B.; Li, N.; Ruben, M.; Florea, I.
B.; Van der Marel, G. A.; Kaiser, M.; Kisselev, A. F.; Overkleeft, H. S. Org. Biomol.
Chem. 2012, 10, 181.
enzyme in the absence (control) or presence of test compounds (0.1–200 lM),
reactions were followed for 45 min at 37 °C. The buffers (pH 7.5) were 20 mM
Tris, 1 mM DTT, 10% glycerol, 0.02% (w/v) SDS for CT-L and PA activities, and
20 mM Tris, 1 mM DTT, 10% glycerol for T-L activity. The inhibitor
concentrations giving 50% inhibition (IC50 values) were obtained by plotting
the percent inhibition against inhibitor concentration to equation:
nH
50
nH
nH
% inhibition ¼ 100½Iꢁ=ðIC þ ½Iꢁ Þ, or equation % inhibition ¼ 100½Iꢁ
=