α-Keto Phenylamides as P1′-Extended Proteasome Inhibitors

Article abstract of DOI:10.1002/cmdc.201402244

The major challenge for proteasome inhibitor design lies in achieving high selectivity for, and activity against, the target, which requires specific interactions with the active site. Novel ligands aim to overcome off-target-related side effects such as peripheral neuropathy, which is frequently observed in cancer patients treated with the FDA-approved proteasome inhibitors bortezomib (1) or carfilzomib (2). A systematic comparison of electrophilic headgroups recently identified the class of α-keto amides as promising for next generation drug development. On the basis of crystallographic knowledge, we were able to develop a structure-activity relationship (SAR)-based approach for rational ligand design using an electronic parameter (Hammett's σ) and in silico molecular modeling. This resulted in the tripeptidic α-keto phenylamide BSc4999 [(S)-3-(benzyloxycarbonyl-(S)-leucyl-(S)-leucylamino)-5-methyl-2-oxo-N-(2,4-dimethylphenyl)hexanamide, 6a], a highly potent (IC₅₀=38 nM), cell-permeable, and slowly reversible covalent inhibitor which targets both the primed and non-primed sites of the proteasome's substrate binding channel as a special criterion for selectivity. The improved inhibition potency and selectivity of this new α-keto phenylamide makes it a promising candidate for targeting a wider range of tumor subtypes than commercially available proteasome inhibitors and presents a new candidate for future studies.

Full text of DOI:10.1002/cmdc.201402244

CHEMMEDCHEM
FULL PAPERS
DOI: 10.1002/cmdc.201402244
a-Keto Phenylamides as P1’-Extended Proteasome
Inhibitors
[a]
[a]
[b]
[c]
[c]
Constantin Voss, Christoph Scholz, Sabine Knorr, Philipp Beck, Martin L. Stein,
[a]
[d]
[d]
[c]
[b]
Andrea Zall, Ulrike Kuckelkorn, Peter-Michael Kloetzel, Michael Groll, Kay Hamacher,
[a]
and Boris Schmidt*
The major challenge for proteasome inhibitor design lies in
achieving high selectivity for, and activity against, the target,
which requires specific interactions with the active site. Novel
ligands aim to overcome off-target-related side effects such as
peripheral neuropathy, which is frequently observed in cancer
patients treated with the FDA-approved proteasome inhibitors
bortezomib (1) or carfilzomib (2). A systematic comparison of
electrophilic headgroups recently identified the class of a-keto
amides as promising for next generation drug development.
On the basis of crystallographic knowledge, we were able to
develop a structure–activity relationship (SAR)-based approach
for rational ligand design using an electronic parameter (Ham-
mett’s s) and in silico molecular modeling. This resulted in the
tripeptidic a-keto phenylamide BSc4999 [(S)-3-(benzyloxycar-
bonyl-(S)-leucyl-(S)-leucylamino)-5-methyl-2-oxo-N-(2,4-dime-
thylphenyl)hexanamide, 6a], a highly potent (IC =38 nm),
cell-permeable, and slowly reversible covalent inhibitor which
targets both the primed and non-primed sites of the protea-
some’s substrate binding channel as a special criterion for se-
lectivity. The improved inhibition potency and selectivity of
this new a-keto phenylamide makes it a promising candidate
for targeting a wider range of tumor subtypes than commer-
cially available proteasome inhibitors and presents a new can-
didate for future studies.
50
Introduction
Intracellular protein degradation via the ubiquitin–proteasome
system is crucial for protein homeostasis within eukaryotic or-
ganisms. The 26S proteasome is a sophisticated multicatalytic
molecular degradation machinery consisting of a proteolytic
ates the primed and a non-primed regions of this channel,
[
2]
starting from the cleavage site of a peptide substrate.
However, fluorogenic assays with eukaryotic cells revealed
that each b-ring only harbors three catalytic active sites with
distinct substrate preferences: b1 cleaves after acidic residues
and therefore mimics a caspase-like (CL) activity, b2 displays
trypsin-like (TL) activity, while b5 typifies chymotrypsin-like
(ChTL) activity.
20S core particle (CP) surrounded by two regulatory 19S caps,
which are responsible for the recognition of ubiquitin marked
substrates, their unfolding, and transport to the inner CP. The
20S proteasome appears as an elongated hollow cylinder form-
ing the multicatalytic center: four heptameric rings form a pile
The mode of action of each active site follows a uniform
mechanism employing an N-terminal threonine (Thr1) hydro-
lyzing the substrate’s scissile peptide bond by nucleophilic
[
1]
following an a b b a stoichiometry. The binding channels of
7
7 7 7
the proteasome active sites within the inner b-rings bear differ-
ent subpockets that are essential for the specific binding of de-
fined substrate peptides. A systematic nomenclature differenti-
g
attack of its hydroxy group (Thr1O ). The N terminus (Thr1N)
acts via a water molecule and coordinates the proton shuttle
and cleavage of the acyl–enzyme intermediate by the release
of defined oligopeptides with a length distribution between
[
a] C. Voss, C. Scholz, Dr. A. Zall, Prof. Dr. B. Schmidt
Clemens Schçpf Institute for Organic Chemistry & Biochemistry
Technische Universitꢀt Darmstadt
[3]
three and 25 amino acids. Due to the central role of the CP
in antigen processing, cell cycle control, cell signaling, and pro-
tein quality control, this protease represents an important
target in the fields of cell biology, structural biology, and me-
dicinal chemistry. Numerous specific and nonspecific inhibitors
Alarich Weiss-Str. 4-8, 64287 Darmstadt (Germany)
E-mail: schmidt_boris@t-online.de
[b] S. Knorr, Prof. Dr. K. Hamacher
Computational Biology & Simulation, Technische Universitꢀt Darmstadt
Schnittspahnstr. 10, 64287 Darmstadt (Germany)
[4]
have been developed to target proteasome activities. This
led to the approval of the boronic acid bortezomib (1) by the
US Food and Drug Administration (FDA) in 2003 and, most re-
cently, the epoxyketone carfilzomib (2) for the treatment of
[c] P. Beck, Dr. M. L. Stein, Prof. Dr. M. Groll
Center of Integrated Protein Science at the Department of Chemistry
Chair of Biochemistry, Technische Universitꢀt Mꢁnchen
Lichtenbergstr. 4, 85747 Garching (Germany)
[5]
multiple myeloma in 2012 (Figure 1).
[d] Dr. U. Kuckelkorn, Prof. Dr. P.-M. Kloetzel
So far, the majority of known natural or synthetic inhibitors
of the proteasome addresses the non-primed site of the bind-
ing channel. However, the primed regions show significant var-
iations amongst the active sites and thus can be used as selec-
Institute of Biochemistry CCM, Charitꢂ Universitꢀtsmedizin Berlin
Charitꢂplatz 1/Virchowweg 6, 10117 Berlin (Germany)
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cmdc.201402244.
ꢀ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 1 – 9
1
ÞÞ
These are not the final page numbers!

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
Results
[10a]
In a previous work,
BSc2189
(
5, Figure 2A) was identified as
a promising potent proteasome
inhibitor (IC : 72 nm) that exhib-
50
its high selectivity for the b5
subunit. Diminished proteolysis
in the cytosolic fraction, as well
as in the protease inhibitor mix-
ture-pretreated lysate (Complete,
Roche Applied Science), con-
firmed inhibitory activity exclu-
sively to the CP and distinguish-
ed it from inhibition of most cy-
tosolic serine/aspartate proteas-
es. X-ray data from crystallization
of the compound in complex
Figure 1. Structures of FDA-approved proteasome inhibitors 1 and 2, as well as 3, which binds to the primed site
of the proteasome substrate binding channel, and its natural product counterpart 4.
tivity criteria for structure-guided inhibitor design. As an exam-
ple, the b-lactone proteasome inhibitor homobelactosin C (3,
Figure 1) shows high b5-selectivity by occupying the primed
site of the ChTL substrate binding channel. Interestingly, this
natural product displays high antitumor activity, with nanomo-
with the yeast 20S proteasome revealed hemiacetal formation
in the a-position of the phenylamide moiety, due to nucleo-
g
[10c]
philic attack of the ligand a-ketone by Thr1O (Figure 2A).
The observation of the amide terminus occupying the S1’
cavity of the binding channel is in accordance with a previous
[6]
[10b]
lar IC50 values in human pancreoma and colon cancer cells.
work from Chatterjee et al.
X-ray data revealed the ligand’s
Crystallographic analysis of this ligand in complex with the
yeast 20S proteasome identified that its high selectivity was
primed site residue (P1’) extending into the hollow inner side
of the cylindrical CP. The attack by Thr1O on the ligand’s “si”
g
[
7]
caused by an unexpected mode of action. Thus, 3 and its
natural counterpart, belactosin A (4, Figure 1), turned out to be
promising leads for the design of new and highly selective pro-
face enables hydrogen bond formation between the hemiace-
tal and Thr1N (Figure 2B). Occupation of the oxyanion hole
formed by Gly47N with the terminal amide carbonyl group ad-
ditionally stabilizes the ligands orientation on the target. How-
ever, the most notable feature of 5 is its aromatic C-terminal
phenylamide moiety that resides in an almost perfectly planar
fashion in the S1’ subpocket of the binding channel (Fig-
ure 2D). This rigidity of the P1’ residue of the ligand contrib-
utes to the driving force for its high-affinity binding, as de-
creased degrees of freedom in the unbound state restrict the
entropic penalty upon binding.
[
8]
teasome inhibitors. Even though acyl–ester formation of the
g
b-lactone headgroup with Thr1O follows a reversible mecha-
nism, the ligand orientation blocks access to a water molecule,
preventing deacetylation. The resulting long half-life of b-lac-
tone-induced inhibition is similar to the activity of irreversibly
binding ligands. This imposes limits for deep solid tissue pene-
tration, restricting these inhibitors to the treatment of non-
[
9]
solid tumors.
Recently, the class of reversible and potent peptidic a-keto
amides has been shown to exploit both the primed and the
non-primed sites, thereby compensating for their moderate
chemical reactivity and leading to a strong proteasome bind-
As a proof of concept, we investigated the variation in the
electron density of the aromatic system by introducing sub-
stituents at the para position of the phenyl moiety in P1’. Both
electron donating (6b; 6c) and electron withdrawing groups
(6d; 6e) were analyzed in this study. We converted the respec-
tive anilines (7a–d) to formamides (8a–d, Scheme 1) to pro-
vide access to functionalized derivatives of 5. N-formylation
was performed according to a published catalyst- and solvent-
[
10]
ing preference. It turned out that the a-keto phenylamide
headgroup harbors a potential motif for ligand optimization,
due to its unique orientation in the CP active site cavity with
the terminal phenylamide moiety projecting into the primed
site of the binding channel. Furthermore, it is the a-keto phe-
nylamide residue that accounts for the ligand’s inhibitory po-
tency.
[11]
free procedure, which was adopted for anilines 7a–d. The
subsequent conversion into isonitriles 9a–e was followed by
a dehydration step, using phosphorus oxychloride and triethyl-
[12]
Therefore a-keto phenylamide CP inhibitors are promising
candidates for preclinical studies of a wider range of tumor
subtypes as currently targeted by bortezomib (1) and carfilzo-
mib (2). This work describes the development of an inhibitor
based on the a-keto phenylamide headgroup, that engages in
additional primed site interactions and thus exhibits improved
potency.
amine in dichloromethane.
Subsequent conversion was
done directly after a quick workup of hydrolytically labile com-
pounds 9a–e. The tripeptidic aldehyde Cbz-Leu-Leu-Leu-al and
trifluoroacetic acid then underwent a multicomponent Passeri-
ni reaction with isonitriles 9a–e to form a-hydroxyphenyla-
mides 10a–e. Oxidation with 2-iodoxybenzoic acid (IBX) pro-
vided the desired tripeptidic a-keto amides 6a–e in moderate
yields, as displayed in Scheme 1.
ꢀ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 0000, 00, 1 – 9
&
2
&
ÞÞ
These are not the final page numbers!

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
Figure 2. A) Binding mechanism of a-keto phenylamides to the b5 subunit of the 20S proteasome. Reversible hemiacetal formation by nucleophilic attack of
the catalytically active Thr1 (green) with the ketone moiety of the inhibitor (black). B)–E) Crystal structures of BSc2189 (5, PDB code: 4NO8) and BSc4999 (6a,
PDB code: 4R02) in complex with the yeast CP. B,C) 2D images show interactions as grey dashed lines between the inhibitor and relevant subpocket residues
of the proteasome (blue/grey). D,E) 3D images show the ligands as stick models (yellow) and the proteasome subunits in ribbon/loop representation (blue/
grey). The dihedral angles (Ccarbonyl–N–Cphenyl–Cphenyl) of the ligand phenyl amide moieties are indicated by red arrows.
A structure–activity relationship study was established for
our ensemble of initial derivatives 6b–e, using the tabulated
Hammett’s constants (s), a parameter expressing the electronic
constitution of an aromatic reaction center in dependency of
Accordingly, we observed that increased electron density of
the aromatic system, which is induced by electron donating
substituents, resulted in a stronger double bond character of
the nitrogen–phenyl bond (NÀCphenyl) and, thus, to increased ri-
gidity. We assume this to be the major feature of an entropical-
ly favored binding process, thus corresponding to the im-
proved biological activity. Nevertheless, all derivatives carrying
functional groups in the para position were less active than
the unsubstituted phenylamide lead (5). For illustration pur-
poses, we have included 5 in Figure 3. Hence, we aimed to
characterize the intermolecular forces within the primed site of
the b5 subunit. The SAR study suggested that the electron do-
nating groups play a key role in the affinity of the inhibitor,
contrary to the withdrawing groups. As the considered elec-
[13]
the nature of a substituent at a meta or para position. Our
hypothesis is that the binding process to the CP predominant-
ly depends on the electron density of the aromatic C-termini,
modified by different para substituents in 6b–e. Therefore, we
performed a hypothesis test in which we plotted logarithmic
IC50 values against the respective s constants. As a result, we
identified a linear correlation supporting our hypothesis: bio-
logical data were consistent with the SAR for all para-substitut-
ed compounds (6b–e), demonstrating a loss in activity with
higher s constants (Figure 3).
ꢀ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 0000, 00, 1 – 9
&
3
&
ÞÞ
These are not the final page numbers!

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
a conformational database span-
ning different dihedral angles
between the plane of the re-
spective aromatic system and
the amide moiety was generated
by keeping the peptidic back-
bone in its “native conforma-
tion”, based on the blueprint of
the CP–5 X-ray structure. Next,
simulations starting with iterated
dihedral angles in 458 steps
were carried out to determine
relative energy minima, allowing
us to rank the ligands according
to their respective energies (see
Supporting Information). The
2,4-dimethyl substitution dis-
played the lowest overall energy
score, suggesting the synthesis
of ligand 6a. In agreement with
our modeling studies, 6a turned
out to be the most potent inhib-
itor of the 20S proteasome
within this series. Blocking the
2 3 3 2 2
Scheme 1. Synthesis of target compounds 6a–e. Reagents and conditions: a) HCO H, 608C; b) POCl , NEt , CH Cl ,
0
8C; c) 1. Cbz-Leu-Leu-Leu-al, pyridine, CH
2
Cl
2
, À108C, 2. TFA, 08C; d) IBX, DMSO, RT. Percent yields are shown;
for 6a–e, yields were calculated over two reaction steps, as 10a–e were not isolated.
ChTL activity in vitro resulted in an IC50 value of 38 nm. More-
over, this compound is selective for the ChTL substrate binding
channel, as the CL and the TL activities remained unaffected
Table 1).
(
Table 1. Activities of peptidic a-keto phenylamides 6a–e against the
[
a]
three catalytic activities of isolated proteasomes.
[
b]
Compd IC50 [nm]
ChTL (b5)
TL (b2)
CL (b1)
6
6
6
6
6
a
b
c
d
e
38Æ19
118 Æ40
142Æ95
1608Æ53
326Æ35
>3000
NA
>9000
NA
NA
NA
NA
NA
NA
NA
Figure 3. Plot of logarithmic IC50 values of inhibitors 6b–e against respective
Hammett’s constants s (*). Note the significant linear correlation
[a] See Scheme 1 for compound structures. [b] Data, normalized to con-
trols, are the means of two independent experiments, each performed in
triplicate (n=2); in vitro data for lead compound 5 (IC
(
pꢀ0.0409) between log(IC50) and s. For illustrative purposes, we also added
a single data point for 5 (~). Significance tests were performed based on
our starting hypothesis for exclusively para-substituted derivatives 6b–e.
Linear correlation: log(IC50)=0.85584s+2.48202.
0,b5
=72 nm) were
5
[
10a]
published previously.
NA: not affected (IC50 >10 mm).
tron donating groups of 6b and 6c predominantly act as hy-
drogen bond acceptors, water molecules are coordinated by
each of the ligands and cause a decrease in their binding pref-
erence by compensating for the benefit of high electron densi-
ty within the aromatic system. We therefore propose that
methyl groups donate electron density into the aromatic
system by s-conjugation, but they do not participate in hydro-
gen bonding. We performed a docking study to evaluate the
most promising compound of the five possible ortho- and
para-substituted methylphenylamide derivatives. Covalent
docking with flexible side chains of the receptor was realized
Crystallographic data of 6a in complex with the yeast 20S
proteasome revealed that the dimethylated aromatic system is
twisted out of the amide plane, with a dihedral angle of 27.68,
in contrast with the co-planar keto amide moiety of lead 5
upon CP binding (Figure 2D,E). This conformational change is
interpreted as an on-target effect due to steric and enthalpic
interactions within the active site, which compensates for the
electronically favored periplanar shape. Note, these findings
are in agreement with our predictions, as a hindered rotation
of the C
–N–Cphenyl–Cphenyl plane prior to bond formation
carbonyl
g
with Thr1O is crucial for the entropically favored binding of
the ligand.
[14]
with conflexdock within the MOE2012.10 software.
First,
ꢀ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 0000, 00, 1 – 9
&
4
&
ÞÞ
These are not the final page numbers!

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
Finally, we performed a HeLa cell-based assay to investigate
cell permeability of the synthesized compounds in vivo (Sup-
porting Information). We obtained promising inhibition activi-
ties, thus confirming cell penetration of 5 and 6a–e. In addi-
tion, the reversibility of the a-keto phenylamide binding mode
was addressed by a dialysis experiment with 5 and the most
potent inhibitor, 6a (Supporting Information). After inhibitor
treatment, the proteasome activity recurred significantly after
Conclusions
Identification and biological evaluation of the tripeptidic a-
keto phenylamide 5 was the major subject of prior work, in-
cluding binding mode elucidation, IC50 and LD50 determination,
and selectivity profiling of the distinct proteasome active sites
[10a,c]
and most cytosolic serine/aspartate proteases.
The compa-
rative study of different headgroup inhibitors, all bearing the
Cbz-Leu-Leu-Leu-backbone, indicated ketoamide 5 as the most
promising candidate to be investigated in cellular models of
chemo- and immunosuppressive therapies. The aim of this
study was to improve both potency and ligand efficiency of 5
while avoiding significant enlargement, as this may result in
too lipophilic and thus undruggable compounds. Our SAR ap-
proach identified structure 6a and provided improved ligand
efficiency by marginal modifications of lead structure 5. Biolog-
ical evaluation of 6a showed that the beneficial properties of
the a-keto phenyl amide moiety could be conserved, while
gaining improved potency against the b5 subunit. The S1’ oc-
cupation of the ligand as an additional selectivity criterion may
diminish off-target side effects in humans, such as peripheral
neuropathy. This disabling neuropathy is frequently observed
after bortezomib treatment and even with the second genera-
tion drug carfilzomib. Furthermore, the reversible binding
mode of hemiketal formation is likely to enable penetration of
deeper solid tissues and allow cells to recuperate unless they
are sufficiently damaged. The CP inhibitor 6a is a promising
drug candidate to address a wider range of tumor subtypes
than targeted by irreversible commercial drugs, owing to its
strong inhibition potency (IC =38 nm), and thus has emerged
72 h, as expected for hemiacetal formation of the ligand with
the active sites.
Discussion
[
10b]
Chatterjee et al.
first reported a-keto amides as P1’-extend-
ed proteasome inhibitors in 1999, yet crystallographic evalua-
tion of the binding mode to the active site of the CP enabled
[
10c]
detailed SAR studies only recently.
Focusing on the elec-
tronic situation of the ligand’s aromatic C-terminus and its
effect on binding to the CP, we plotted tabulated Hammett’s
constants from aromatic para substituents against IC50 values
of the synthesized a-keto phenylamides 6b–e. We found a cor-
relation between selected substituents of inhibitors and their
biological activities, which depend on the strength of the elec-
tron donating group. Though the lead structure (5) was still
the most active inhibitor, compared with 6b–e, it is the polar
nature of the para-NMe or para-OMe groups that diminishes
2
the biological activity. However, in contrast to nonpolar ligand
interactions with lipophilic protein surfaces, the influence of
hydrogen bond formation in the thermodynamic ligand bind-
ing process is more complex and still needs further experimen-
50
as a target for further investigations.
[
15]
tal characterization. We thus speculate that a polar group in
P1’ may induce water binding and create a hydrogen bond Experimental Section
network that entropically hinders the ligand binding process
Synthetic procedures
rather than stabilizes the on-target structure. By introducing
methyl groups, which are nonpolar s-donors, we were able to
confirm our conclusions from the SAR studies. Subsequently,
an optimal substitution pattern of methyl groups was identi-
fied by a molecular docking approach, resulting in the 2,4-di-
methylated derivative 6a. With 6a having an IC50 value of
This section contains the experimental description of new com-
pounds synthesized in this work: 6a–e. Experimental descriptions
of synthesized substrate compounds already similarly described in
the literature (8a–d, 9a–d) can be found in the Supporting Infor-
mation. Compound 5 was synthesized by a previously published
[10a]
procedure.
The aldehyde substrate Cbz-Leu-Leu-Leucinal was
38 nm, we could demonstrate a significant enhancement of
synthesized by standardized peptide coupling from commercial
enantiopure Cbz-Leu-Leu-OH and (S)-leucinol. Oxidation with IBX
gave the desired aldehyde. IBX was synthesized following a pub-
ligand interaction and inhibitory activity relative to 5 (IC =
50
72 nm). Remarkably, analysis of the crystal structure of 6a in
[16]
complex with the yeast proteasome revealed a dihedral twist
of the planar conjugated p-system, which is due to increased
hydrophobic interactions with the P1’ binding site. This devia-
tion from planarity stands in contrast with the rigid and planar
arrangement of 5 in complex with the CP. Addressing the
primed and non-primed regions of the binding channel differ-
entiates a-keto phenylamides from the majority of known pep-
tidic proteasome inhibitors, thus providing access to highly
potent, reversible, and specific small molecule inhibitors. Fur-
thermore, our data enable targeted ligand design in both di-
rections of the substrate binding channel, using the a-keto
phenylamide moiety as a superb linker between the primed
and non-primed sites.
lished procedure. All chemicals that were purchased as reagent
grade from commercial suppliers were used without further purifi-
cation.
1
13
General methods: H NMR and C NMR spectra were recorded
with a Bruker AC 300 (300 MHz) or AC 500 (500 MHz) spectrometer.
Chemical shifts are reported in d (ppm), adjusted to the central
line of the deuterated solvent (MeOD, CDCl , [D ]DMSO). High reso-
3
6
lution mass spectrometry (HRMS) was performed with an Agi-
lent 1290 Infinity HPLC system coupled to an Agilent G6530A
QTOF MS system. HPLC analysis was performed with an Agi-
lent 1100 system. The purity of the final compounds was deter-
mined by UV detection (l=254 nm). The chromatographic
method employed the following: Zorbax Eclipse XDB-C18 column,
4.6ꢁ150 mm; mobile phase A: H O (0.1% TFA), mobile phase B:
2
ꢀ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 0000, 00, 1 – 9
&
5
&
ÞÞ
These are not the final page numbers!

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
À1
+
acetonitrile; flow rate: 1 mLmin ; gradient elution: 30 to 100% B
6.47 min; HRMS calcd for ([C H N O ]H) m/z: 625.3596, found:
34
48
4
7
over 15 min. According to this method, the purities for all com-
pounds that were evaluated in biological assays were ꢁ95%. Thin-
layer chromatography was carried out using aluminum sheets pre-
coated with silica gel 60 F254 (0.2 mm; E. Merck). Chromatographic
spots were visualized by UV and/or by spraying with a methanolic
solution of vanillin/H SO or aqueous KMnO solution, followed by
625.3611.
(S)-3-(Benzyloxycarbonyl-(S)-leucyl-(S)-leucylamino)-5-methyl-2-
oxo-N-(4-cyanophenyl)hexanamide (6d): Yield: 12% (17 mg), col-
orless oil: H NMR (300 MHz, CDCl , 300 K): d=8.86 (1H, s), 7.76
1
3
(2H, m), 7.65 (2H, m), 7.35 (5H, m), 6.94 (1H, m), 6.39 (1H, m), 5.28
2
4
4
(1H, m), 5.12 (2H, m), 4.46 (1H, m), 4.14 (1H, m), 1.73 (9H, m),
heating. Silica gel chromatography was carried out using Merck
silica gel 60 (0.063–0.2 mm).
13
0
.94 ppm (18H, m); C NMR (125 MHz, CDCl , 300 K): d=196.0,
3
1
72.6, 172.1, 157.5, 140.4, 136.0, 133.5, 128.8, 128.6, 128.2, 120.1,
1
18.6, 108.6, 67.5, 54.0, 52.8, 51.6, 41.1, 40.3, 39.9, 25.4–24.9, 23.3–
a-Keto phenylamides (6a–e) from peptidic aldehydes and
phenyl isonitriles via a Passerini reaction and subsequent oxida-
tion of intermediate a-hydroxy phenylamides (10a–e): The pep-
2
5
6
1.6 ppm; HPLC: t =7.75 min, HPLC (intermediate 10d): t =
R
R
+
.05 min; HRMS calcd for ([C H N O ]H) m/z: 620.3448, found:
20.3457.
34 45
5
6
tidic aldehyde (1.0 equiv), phenyl isonitrile (1.5 equiv), and pyridine
À1
(
4.0 equiv) were dissolved in dry CH Cl2 (2 mLmmol aldehyde)
2
(S)-3-(Benzyloxycarbonyl-(S)-leucyl-(S)-leucylamino)-5-methyl-2-
and cooled to À108C. Trifluoroacetic acid (2.0 equiv) was added
oxo-N-(4-bromophenyl)hexanamide (6e): Yield: 41% (218 mg),
white solid: H NMR (500 MHz, CDCl , 300 K): d=8.95 (1H, s), 7.52–
dropwise, and the reaction mixture was allowed to stir for 2 h at
1
3
0
8C. After stirring for an additional 72 h at room temperature,
7
.05 (9H, m), 5.75 (1H, s), 5.29 (1H, m), 5.12–5.01 (2H, m), 4.60
completion of the reaction was monitored by HPLC. CH Cl was
2
2
(
1H, m), 4.31 (1H, m), 1.57–1.26 (9H, m), 0.95–0.66 ppm (18H, m);
added, and the mixture was washed with 0.1N aqueous HCl (3ꢁ)
13
C NMR (125 MHz, CDCl , 300 K): d=196.7, 172.4, 172.1, 157.4,
3
and aqueous saturated NaHCO (3ꢁ). The organic layer was then
3
1
5
2
56.3, 136.2, 135.5, 132.1, 128.5, 128.2, 128.0, 121.5, 118.1, 67.1,
3.6, 53.1, 51.1, 51.4, 41.5, 40.7, 39.8, 25.2, 24.7, 23.7, 23.1, 22.8,
2.5, 22.4, 22.2, 21.4 ppm; HPLC: t =7.68 min, HPLC (intermediate
dried over Na SO , filtered, and concentrated under reduced pres-
2
4
sure. The resulting colorless oil and IBX (1.5 equiv) were dissolved
R
À1
in DMSO (2 mLmmol aldehyde) and stirred for 12 h at room tem-
+
1
6
0e): tR =6.05 min; HRMS calcd for ([C H BrN O ]H)
m/z:
33
45
4
6
perature. After addition of CH Cl , the mixture was washed with
2
2
73.2595, found: 673.2605.
water (3ꢁ), aqueous saturated NaHCO (3ꢁ), and brine (3ꢁ). The
3
organic layer was then dried over Na SO , filtered, and concentrat-
2
4
ed under reduced pressure. Purification was done via liquid chro-
matography.
Biological and structural analysis
Inhibition assay of purified 20S proteasome: 100 ng of constitutive
0S proteasomes (isolated from human red blood cells) were incu-
bated with defined concentrations of inhibitors 6a–e for 15 min at
room temperature. Equal volumes of a protease substrate solution
were added (final concentration: 50 mm) and incubated at 378C for
(
S)-3-(Benzyloxycarbonyl-(S)-leucyl-(S)-leucylamino)-5-methyl-2-
2
oxo-N-(2,4-dimethylphenyl)hexanamide (6a): Yield: 21% (35 mg),
colorless oil: H NMR (300 MHz, CDCl , 300 K): d=8.57 (1H, s), 7.85
1
3
(
1H, d, J=8.7 Hz), 7.32 (5H, m), 7.05 (1H, d, J=7.4 Hz), 7.01 (2H,
m), 6.81 (1H, m), 5.49 (1H, m), 5.40 (1H, m), 5.08 (2H, m), 4.54 (1H,
m), 4.24 (1H, m), 2.28 (3H, s), 2.24 (3H, s), 1.58 (9H, m), 0.92 ppm
1
h. Proteasome activity was recorded by the release of the fluoro-
genic AMC group from the protease substrate at 360 nm excitation
and 460 nm emission, (LLE-AMC, VGR-AMC, and LLVY-AMC were
used to analyze the different cleavage properties of the protea-
some). Data, normalized to controls, represent the means of two
independent experiments, each performed in triplicate (n=2). Stat-
istical analysis was performed following the methods of Cumming
1
3
(
18H, m); C NMR (125 MHz, CDCl , 300 K): d=196.9, 172.6, 171.9,
3
1
1
2
4
6
56.8, 156.3, 136.2, 135.6, 131.7, 131.4, 128.8, 128.7, 128.3, 128.1,
27.5, 121.8, 67.2, 53.7, 53.1, 51.7, 41.4, 40.7, 40.2, 25.4–24.8, 23.3–
1.5 ppm; HPLC: t =7.16 min, HPLC (intermediate 10a): t =
R
R
+
.98 min; HRMS calcd for ([C H N O ]Na) m/z: 645.3627, found:
35
50
4
6
45.3629.
[17]
et al.
(
S)-3-(Benzyloxycarbonyl-(S)-leucyl-(S)-leucylamino)-5-methyl-2-
5
Intracellular inhibition of proteasomes: HeLa cells (2ꢁ10 cells per
oxo-N-(4-N,N-dimethylaminophenyl)hexanamide (6b): Yield: 16%
well) were seeded in 96-well microtiter plates and cultured in RPMI
supplemented with 10% FCS, 2 mm glutamine, and penicillin–
1
(
20 mg), colorless oil: H NMR (500 MHz, CDCl , 300 K): d=8.55
3
(
1H, s), 7.50 (2H, d, J=9.0 Hz), 7.33 (5H, m), 6.87 (1H, m), 6.75 (2H,
À1
À1
streptomycin (100 UmL penicillin, 100 mgmL streptomycin). In-
hibitors were added as a tenfold stock to adjust the indicated con-
centrations and were incubated overnight at 378C under 5% CO2.
The supernatant was removed, and cells were washed with cold
phosphate-buffered saline and lysed in 100 mL of Tris (20 mm),
EDTA (1 mm), and 0.1% NP-40. The proteasome activity was mea-
sured in 25 mL of the lysate with a final concentration of 50 mm
Suc-LLVY-AMC. The assays were incubated for 1 h at 378C. Protea-
some activity was estimated at 460 nm emission (excitation at
m), 6.46 (1H, m), 5.39 (1H, m), 5.23 (1H, m), 5.10 (2H, m), 4.49 (1H,
m) 4.18 (1H, m), 2.94 (6H, s), 1.65 (6H, m), 1.50 (3H, m), 0.90 ppm
1
3
(
18H, m); C NMR (125 MHz, CDCl , 300 K): d=197.0, 172.4, 171.6,
3
1
5
56.4, 156.3, 136.2, 128.8, 128.7, 128.4, 128.2, 121.4, 113.2, 67.4,
3.8, 53.2, 51.8, 41.4, 41.0, 40.7, 40.5, 25.4–24.9, 23.4–21.5 ppm;
HPLC: t =7.28 min, HPLC (intermediate 10b): t =5.91 min; HRMS
calcd for ([C H N O ]H) m/z: 638.3916, found: 638.3917.
R
R
+
3
5
51
5
6
(S)-3-(Benzyloxycarbonyl-(S)-leucyl-(S)-leucylamino)-5-methyl-2-
3
60 nm). Experimental data (see Supporting Information), normal-
ized to controls, represent the means of triplicate experiments (n=
). Statistical analysis was performed following the methods of
oxo-N-(4-methoxyphenyl)hexanamide (6c): Yield: 24% (21 mg),
colorless oil: H NMR (500 MHz, CDCl , 300 K): d=8.26 (1H, s), 7.55
1
3
1
(
2H, d, J=9.1 Hz), 7.33 (5H, m), 6.88 (2H, d, J=9.1 Hz), 6.87 (1H,
[17]
Cumming et al.
m), 6.53 (1H, d, J=8.2 Hz), 5.38 (1H, m), 5.29 (1H, m), 5.10 (2H, m),
4
0
1
1
2
.49 (1H, m), 4.19 (1H, m), 3.79 (3H, s), 1.66 (6H, m), 1.49 (3H, m),
Crystallization and structure determination: Crystals of the yeast CP
13
[1a,18]
.92 ppm (18H, m); C NMR (125 MHz, CDCl , 300 K): d=196.9,
were grown in hanging drops at 208C as described previously.
The protein concentration used for crystallization was 40 mgmL
3
À1
72.8, 172.5, 171.8, 157.3, 156.7, 136.2, 130.0, 128.7, 128.4, 128.2,
21.6, 114.5, 67.4, 55.8, 53.8, 53.1, 51.7, 41.3, 40.7, 40.3, 25.4, 24.9,
3.4–21.3 ppm; HPLC: t =7.89 min, HPLC (intermediate 10c): t =
in Tris/HCl (20 mm, pH 7.5) and EDTA (1 mm). Drops contained 1 mL
of protein and 1 mL of the reservoir solution (30 mm magnesium
R
R
ꢀ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 0000, 00, 1 – 9
&
6
&
ÞÞ
These are not the final page numbers!

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
acetate, 100 mm morpholino-ethane-sulfonic acid (pH 7.2), and
bases were docked into the active site and energy-minimized
using the Amber12EHT force field, which is parameterized for pro-
teins and nucleic acids using Amber and parameterized for small
1
2
0% (w/v) 2-methyl-2,4-pentanediol). Crystals appeared after
days and were then soaked with inhibitors in DMSO at final con-
[24]
centrations of 2 mm for at least 24 h. Droplets were then comple-
mented with a cryoprotecting buffer (30% (w/v) 2-methyl-2,4-pen-
tanediol, 20 mm magnesium acetate, 100 mm morpholino-ethane-
sulfonic acid, pH 6.9) and supercooled in a stream of liquid nitro-
gen gas at 100 K (Oxford Cryo Systems). A dataset was collected
from the CP–6a complex (PDB code: 4R02) at 2.5 ꢂ and cell param-
eters of a=137 ꢂ, b=301 ꢂ, c=146 ꢂ, and b=1138 in the P21
space group using synchrotron radiation (l=1.0 ꢂ) at the X06SA-
beamline (Swiss Light Source, Villingen, Switzerland). X-ray intensi-
molecules using 2D Extended Hꢃckel Theory.
All non-anchor
atoms belonging to the ligand were free to move during the mini-
mization. The pocket atoms of the receptor were tethered to allow
movement. All other atoms were fixed. To estimate the binding
free energy score, the London dG scoring function was used. For
visualization of the procedure, see the Supporting Information.
Supporting Information
[19]
ties were assessed with the program XDS, while data reduction
Experimental procedures and characterization data for substrate
compounds 8a–d and 9a–d; Inhibition of isolated 20S protea-
somes by compounds 6a–e; Activity curves for intracellular inhibi-
tion of proteasomes by compounds 6a–e; X-ray data collection
and refinement statistics; Reversibility of proteasome inhibition by
compounds 5 and 6a, as determined by dialysis and proteasome
activity; Molecular docking specification; NMR spectra of com-
pounds 6a–e.
[19]
was carried out with XSCALE (Table ST1). Electron density was
improved by averaging and back-transforming the reflections
1
0 times over the twofold noncrystallographic symmetry axis using
[20]
the program package MAIN. Conventional crystallographic rigid
body, positional, and temperature factor refinements were carried
out with CNS using the yeast CP structure as a starting model
(
PDB code: 1RYP) (Table ST1), while model building was performed
with the program MAIN.
Analysis of the reversibility of proteasome inhibition via dialysis: Red
blood cell proteasome solution (500 mL, 2 mg proteasome) in
Acknowledgements
2
0 mm Tris, pH 7.2, 0.5 mm EDTA, and 0.1 mm acid (TAE) was sup-
plemented with 25 mL of 0.1 mm inhibitor 5 or 6a, or with 25 mL of
% DMSO in water. The final inhibitor concentration was 5 mm.
This work was supported by the Deutsche Forschungsgemein-
schaft (DFG), Germany (GRK1657, projects 2E and 3C), and by the
Hans und Ilse Breuer Stiftung.
1
After pre-incubation for 30 min at room temperature, the samples
were transferred to 0.5 mL dialysis tubes and dialyzed against 1 L
TAE buffer at room temperature for 72 h. The buffer was ex-
changed after 6, 22, and 48 h. Duplicates of 25 mL (50 ng of protea-
somes) were used for estimation of activity with 100 mm of Suc-
LLVY-AMC at 460 nm emission (excitation 360 nm) at indicated
time points. Experimental data (see Supporting Information) repre-
sent the means of duplicate experiments (n=1). Statistical analysis
Keywords: 20S proteasome · cancer · drug development ·
Passerini reaction · ubiquitin
[1] a) M. Groll, L. Ditzel, J. Lowe, D. Stock, M. Bochtler, H. D. Bartunik, R.
Huber, Nature 1997, 386, 463–471; b) J. Lçwe, D. Stock, B. Jap, P.
Zwickl, W. Baumeister, R. Huber, Science 1995, 268, 533–539.
[17]
was laid out following Cumming et al.
[
2] I. Schechter, A. Berger, Biochem. Biophys. Res. Commun. 1967, 27, 157–
62.
1
[
3] a) M. Groll, R. Huber, Int. J. Biochem. Cell Biol. 2003, 35, 606–616;
b) A. K. Nussbaum, T. P. Dick, W. Keilholz, M. Schirle, S. Stevanovic, K.
Dietz, W. Heinemeyer, M. Groll, D. H. Wolf, R. Huber, H. G. Rammensee,
H. Schild, Proc. Natl. Acad. Sci. USA 1998, 95, 12504–12509.
Molecular docking
Preparation of protein–ligand structures: Only the b5 and b6 subu-
nits of the complex were prepared; all others were removed from
the system. All unbound water molecules were removed from the
complexes, because they showed no significant role in ligand bind-
ing. Hydrogens were added to the X-ray structures of the CP–5
[
[
4] A. F. Kisselev, W. A. van der Linden, H. S. Overkleeft, Chem. Biol. 2012,
1
9, 99–115.
5] a) D. Chen, M. Frezza, S. Schmitt, J. Kanwar, Q. P. Dou, Curr. Cancer Drug
Targets 2011, 11, 239–253; b) J. L. Thompson, Ann. Pharmacother. 2013,
[14]
complex using the Protonate 3D function within MOE2012.10.
4
7, 56–62.
The Protonate 3D application assigns protonation states from a dis-
crete collection of states by optimizing the titration free energy of
all titratable groups in the context of an all-atom model of a macro-
[6] A. Asai, T. Tsujita, S. V. Sharma, Y. Yamashita, S. Akinaga, M. Funakoshi, H.
Kobayashi, T. Mizukami, Biochem. Pharmacol. 2004, 67, 227–234.
[7] M. Groll, O. V. Larionov, R. Huber, A. de Meijere, Proc. Natl. Acad. Sci. USA
2006, 103, 4576–4579.
[21]
molecular structure (including ligands and solvent). The general-
ized Born/volume integral electrostatics model is used for longer
[
8] a) V. S. Korotkov, A. Ludwig, O. V. Larionov, A. V. Lygin, M. Groll, A. de
Meijere, Org. Biomol. Chem. 2011, 9, 7791–7798; b) S. Kawamura, Y.
Unno, A. List, A. Mizuno, M. Tanaka, T. Sasaki, M. Arisawa, A. Asai, M.
Groll, S. Shuto, J. Med. Chem. 2013, 56, 3689–3700.
[22]
range interactions and solvation effects.
Preparation of the ligand database: All five possible ortho- and
para-substituted methyl-derivatives of 5 were derived from the X-
ray structure by manipulation with the Builder tool of MOE. For
each derivative, the dihedral angle spanning the phenyl moiety
and the amide carbonyl was altered in 458 steps using the Builder
[
9] a) P. G. Richardson, B. Barlogie, J. Berenson, S. Singhal, S. Jagannath, D.
Irwin, S. V. Rajkumar, G. Srkalovic, M. Alsina, R. Alexanian, D. Siegel, R. Z.
Orlowski, D. Kuter, S. A. Limentani, S. Lee, T. Hideshima, D.-L. Esseltine,
M. Kauffman, J. Adams, D. P. Schenkein, K. C. Anderson, New Engl. J.
Med. 2003, 348, 2609–2617; b) M. J. Williamson, M. D. Silva, J. Terkelsen,
R. Robertson, L. Yu, C. Xia, P. Hatsis, B. Bannerman, T. Babcock, Y. Cao, E.
Kupperman, Mol. Cancer Ther. 2009, 8, 3234–3243.
software, resulting in four different conformations for each C -sym-
2
metrical and eight conformations for each unsymmetrical aromatic
ring.
[10] a) H. A. Braun, S. Umbreen, M. Groll, U. Kuckelkorn, I. Mlynarczuk, M. E.
Wigand, I. Drung, P.-M. Kloetzel, B. Schmidt, J. Biol. Chem. 2005, 280,
Covalent docking of the ligand database: The customized script
28394–28401; b) S. Chatterjee, D. Dunn, S. Mallya, M. A. Ator, Bioorg.
[23]
Conflexdock was used for covalent docking of the ligands. The
Med. Chem. Lett. 1999, 9, 2603–2606; c) M. L. Stein, H. Cui, P. Beck, C.
Dubiella, C. Voss, A. Krꢃger, B. Schmidt, M. Groll, Angew. Chem. Int. Ed.
2014, 53, 1679–1683; Angew. Chem. 2014, 126, 1705–1709.
g
acetal carbon connected to Thr1O was assigned as the “anchor”
atom and fixed in position. Conformations from the prepared data-
ꢀ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 0000, 00, 1 – 9
&
7
&
ÞÞ
These are not the final page numbers!

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
[
[
[
11] K. P. Dhake, P. J. Tambade, R. S. Singhal, B. M. Bhanage, Green Chem. Lett.
[17] G. Cumming, F. Fidler, D. L. Vaux, J. Cell Biol. 2007, 177, 7–11.
Rev. 2011, 4, 151–157.
[18] M. Groll, R. Huber, Methods Enzymol. 2005, 398, 329–336.
[19] W. Kabsch, J. Appl. Crystallogr. 1993, 26, 795–800.
[20] D. Turk, PhD thesis, Technische Universitꢄt Mꢃnchen, 1992.
[21] P. Labute, Proteins Struct. Funct. Bioinf. 2009, 75, 187–205.
[22] P. Labute, J. Comput. Chem. 2008, 29, 1693–1698.
[23] Scientific Vector Language (SVL) source code provided by Chemical
Computing Group Inc., 1010 Sherbooke Street West, Suite #910, Mon-
treal, QC, H3A 2R7 (Canada), 2012.
12] R. B. Lacerda, C. K. de Lima, L. L. da Silva, N. C. Romeiro, A. L. Miranda,
E. J. Barreiro, C. A. Fraga, Bioorg. Med. Chem. 2009, 17, 74–84.
13] a) L. P. Hammett, J. Am. Chem. Soc. 1937, 59, 96–103; b) J. Shorter,
Chem. Unserer Zeit 1985, 19, 197–208; c) T. Yoshida, M. Shimizu, M.
Harada, S. Hitaoka, H. Chuman, Bioorg. Med. Chem. Lett. 2012, 22, 124–
128.
[
14] Molecular Operating Environment (MOE) 2012.10, Chemical Computing
Group Inc., 1010 Sherbooke Street West, Suite #910, Montreal, QC,
H3A 2R7 (Canada), 2012.
[24] R. Salomon-Ferrer, D. A. Case, R. C. Walker, Wiley Interdiscip. Rev. Comput.
Mol. Sci. 2013, 3, 198–210.
[
[
15] H. Kubinyi in Pharmacokinetic Optimization in Drug Research, Helvetica
Chimica Acta, Zꢃrich, 2007, pp. 513–524.
16] M. Frigerio, M. Santagostino, S. Sputore, J. Org. Chem. 1999, 64, 4537–
Received: June 12, 2014
4538.
Published online on && &&, 0000
ꢀ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 0000, 00, 1 – 9
&
8
&
ÞÞ
These are not the final page numbers!

FULL PAPERS
Prime time for proteasome inhibition:
Insights into the primed binding site of
the proteasome b5 subunit enabled tar-
geted lead optimization for new inhibi-
tors. Crystal structure analysis and mo-
lecular modeling allowed a structure–
activity relationship study to identify
a promising a-keto phenylamide-based
drug candidate with unique pharmaco-
kinetic properties.
C. Voss, C. Scholz, S. Knorr, P. Beck,
M. L. Stein, A. Zall, U. Kuckelkorn,
P.-M. Kloetzel, M. Groll, K. Hamacher,
B. Schmidt*
&& – &&
a-Keto Phenylamides as P1’-Extended
Proteasome Inhibitors
ꢀ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 0000, 00, 1 – 9
&
9
&
ÞÞ
These are not the final page numbers!