Structure-Based Design of β5c Selective Inhibitors of Human Constitutive Proteasomes

Article abstract of DOI:10.1021/acs.jmedchem.6b00705

This work reports the development of highly potent and selective inhibitors of the β5c catalytic activity of human constitutive proteasomes. The work describes the design principles, large hydrophobic P3 residue and small hydrophobic P1 residue, that led to the synthesis of a panel of peptide epoxyketones; their evaluation and the selection of the most promising compounds for further analyses. Structure-activity relationships detail how in a logical order the β1c/i, β2c/i, and β5i activities became resistant to inhibition as compounds were diversified stepwise. The most effective compounds were obtained as a mixture of cis- and trans-biscyclohexyl isomers, and enantioselective synthesis resolved this issue. Studies on yeast proteasome structures complexed with some of the compounds provide a rationale for the potency and specificity. Substitution of the N-terminus in the most potent compound for a more soluble equivalent led to a cell-permeable molecule that selectively and efficiently blocks β5c in cells expressing both constitutive proteasomes and immunoproteasomes.

Full text of DOI:10.1021/acs.jmedchem.6b00705

Subscriber access provided by CARLETON UNIVERSITY
Article
Structure-Based Design of #5c Selective
Inhibitors of Human Constitutive Proteasomes
Bo-Tao Xin, Gerjan de Bruin, Eva M. Huber, Andrej Besse, Bogdan I. Florea, Dmitri V. Filippov, Gijsbert A. van
der Marel, Alexei F Kisselev, Mario van der Stelt, Christoph Driessen, Michael Groll, and Herman S. Overkleeft
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.6b00705 • Publication Date (Web): 20 Jul 2016
Downloaded from http://pubs.acs.org on July 23, 2016
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Journal of Medicinal Chemistry is published by the American Chemical Society. 1155
Sixteenth Street N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 36
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
StructureꢀBased Design of β5c Selective Inhibitors of Human Constitutive Proteasomes
8
9
Bo-Tao Xin†#, Gerjan de Bruin†#, Eva M. Huber§, Andrej Besse‡, Bogdan I. Florea†, Dmitri V.
Filippov†, Gijsbert A. van der Marel†, Alexei F. Kisselev&, Mario van der Stelt†, Christoph
Driessen‡, Michael Groll§ and Herman S. Overkleeft†*
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
†Gorlaeus Laboratories, Leiden Institute of Chemistry, Einsteinweg 55, 2333 CC Leiden, The
Netherlands
§Center for Integrated Protein Science at the Department Chemie, Lehrstuhl für Biochemie,
Technische Universität München, 85748 Garching, Germany
&Department of Pharmacology and Toxicology and Norris Cotton Cancer Center, Geisel Cancer
School of Medicine at Dartmouth, 1 Medical Center Drive HB7936, Lebanon, New Hampshire
03756, United States
‡Department of Hematology and Oncology, Kantonsspital St. Gallen, 9007 St. Gallen,
Switzerland
Abstract
This work reports the development of highly potent and selective inhibitors of the β5c catalytic
activity of human constitutive proteasomes. The work details the design principles – large
hydrophobic P3 residue, small hydrophobic P1 residue – that led to the synthesis of a panel of
peptide epoxyketones; their evaluation and the selection of the most promising compounds for
ACS Paragon Plus Environment
1

Journal of Medicinal Chemistry
Page 2 of 36
1
2
3
4
5
6
7
8
9
further analyses. Structureꢀactivityꢀrelationships detail how in a logical order the β1c/i, β2c/i and
β5i activities became resistant to inhibition as compounds were diversified stepwise. The most
effective compounds were obtained as a mixture of cisꢀ and transꢀbiscyclohexyl isomers and
enantioselective synthesis resolved this issue. Studies on yeast proteasome structures complexed
to some of the compounds provide a rationale for the potency and specificity. Substitution of the
Nꢀterminus in the most potent compound for a more soluble equivalent led to a cellꢀpermeable
molecule that selectively and completely blocks β5c in cells expressing both constitutive
proteasomes and immunoproteasomes.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Introduction
The 20S proteasome core particle (CP) is the major cytosolic and nuclear protein degradation
machinery in eukaryotes.1 In vertebrates, three distinct CPs exist that differ in their tissue
distribution, active site composition, substrate preferences and in the signatures of the
oligopeptide pools they produce. Constitutive proteasome core particles (cCPs), expressed in all
mammalian tissues, contain β1c, β2c and β5c catalytic subunits, with β1c cleaving preferentially
after acidic residues, β2c after basic residues and β5c after hydrophobic residues.2 Hematopoetic
cells constitutively express immunoproteasome core particles (iCPs), and other tissues may do so
in an interferonꢀγ (INFꢀγ)ꢀinducible manner.3 In the iCP, β1c, β2c and β5c are substituted by β1i
(LMP2), β2i (MECLꢀ1) and β5i (LMP7), respectively. The substrate preferences of iCPs differ
from those of cCPs, most prominently in the respective β1 subunits: whereas β1c accepts both
acidic (Asp) and hydrophobic (Leu) residues, β1i has evolved to cleave preferentially after
hydrophobic residues.4 As a consequence, iCP expressing cells produce oligopeptide pools in
which basic and hydrophobic Cꢀterminal residues are prevalent – oligopeptides that may bind
well to major histocompatibility complex class I (MHCI) molecules.5 For this reason, iCPs are
ACS Paragon Plus Environment
2

Page 3 of 36
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
thought to have evolved to enhance the peptide repertoire for MHCIꢀmediated antigen
presentation, however cCPs also make a significant contribution to the production of MHCI
peptides. Cortical thymic epithelial cells that mediate positive CD8+ T cell selection finally
express thymoproteasome core particles (tCPs), essentially iCP particles in which β5i is
substituted for β5t with an apparent loss of bias for cleavage after hydrophobic amino acid
residues.6 In total, vertebrate tissue may express up to seven catalytic activities, distributed over
three distinct 20S core particles. Hybrid CPs composed of both constitutive proteasome and
immunoproteasome catalytic activities exist as well7 and may yield oligopeptide pools distinct
from those produced by pure cCPs or iCPs.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ONX-0914 (51)
LARGE P1
Apparent IC50 values ( M)
5i 0.0057
5c 0.054
1c/ 1i/ 2c/ 2i > 1.0
O
small P3
O
O
H
N
O
N
N
H
N
H
O
O
O
LARGE P1
LU-015i (52)
small P3
Apparent IC50 values ( M)
5i 0.016
1c/ 1i/ 2c/ 2i/ 5c > 1.0
O
O
H
N
O
N
H
N
H
O
O
NH
LARGE P3
LU-005c (29)
Apparent IC50 values ( M)
5c 0.020
small P1
1c/ 1i/ 2c/ 2i/ 5i > 1.0
O
O
H
O
N
N
H
N
H
N3
O
O
ACS Paragon Plus Environment
3

Journal of Medicinal Chemistry
Page 4 of 36
1
2
3
4
5
6
Figure 1. Structures and activities of the β5iꢀselective inhibitor ONXꢀ0914 (51); the β5iꢀ
selective inhibitor LUꢀ015i (52); and the β5cꢀselective inhibitor LUꢀ005c (29).
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Compounds able to selectively inhibit a single 20S proteasome catalytic activity are
indispensable tools to study proteasome activity in the context of global protein turnover, to
determine substrate preferences, and to establish the role of the individual activities in MHCI
antigen processing. Moreover, proteasomes are the primary target of clinical drugs such as
bortezomib,8 carfilzomib9 and ixazomib10 (with more clinical candidates in various stages of
development) for the treatment of multiple myeloma and mantle cell lymphoma. All these
compounds block several catalytic subunits of both cCPs and iCPs, thereby causing strong
cytotoxicity. Subunitꢀselective inhibitors could be used to establish the optimal combination of
catalytic activities both with respect to drug efficacy and (limited) side effects. Selective
proteasome inhibitors have been the subject of extensive studies in recent years in several
research groups both in academia and industry.11 We recently reported12 the development of an
activityꢀbased protein profiling (ABPP) assay which allows inspection of β1c/β1i, β2c/β2i and
β5c/β5i activities in a single experiment. With these probes cCP and iCP subunits, which are
expressed simultaneously by human cell lines and primary tumor tissues can be resolved and
visualized on a simple SDS PAGE. Together with the ABPP assay we disclosed a set of
inhibitors selective for each of the six catalytic activities of human cCPs and iCPs, including 52,
selective for β5i, and 29, selective for β5c (Figure 1). The development of 52 was subject of an
earlier report13 from our laboratories. Here, we describe our results on the development of β5cꢀ
selective inhibitors. We show how screening of a focused library followed by modulating
compound solubility led to the identification of 29. Furthermore, we provide routes of synthesis
ACS Paragon Plus Environment
4

Page 5 of 36
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
towards enantiopure analogues of 29 with respect to the bisꢀcyclohexyl moiety at P3. A crystal
structure of yeast CP complexed with one of our β5c specific inhibitors gives detailed insights
into the mode of action and binding at the molecular level.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Results
Our previous work13 on the development of 52 started with perusal of the crystal structures of
murine cCP and iCP.14 The complex structures bound to the β5i inhibitor 51 (Figure 1) revealed
that the S1 pocket of β5i was more suited to accommodate large hydrophobic residues than the
corresponding S1 site of β5c and that the S3 pocket in β5i was due to Ser27 less spacious than
the β5c counterpart (Figure 2).14 Based on these observations, we created a series of peptide
epoxyketone inhibitors varying in size at P1 and in this way identified 52, which has an
improved selectivity (though at the cost of a slightly lowered activity) for β5i over β5c when
compared to 5115 as well as to PRꢀ924,15a,16 another β5iꢀselective compound. With the aim to
identify β5c selective compounds we synthesized the hydrophobic triꢀ and tetrapeptide
epoxyketones 1ꢀ30 (Table 1 and Supporting Information Table S1) that feature, either a leucine
(1ꢀ24) or an alanine (25ꢀ30) at the P1 position, a phenylalanine or methyltyrosine at P2 and an
azidophenylalanine or a morpholine at P4. A variety of bulky, aromatic and aliphatic amino acids
were installed at P3 with the aim to identify compounds that still would fit in the more spacious
S3 site of β5c but not the S3 site of β5i. The focused compound library was prepared according
to our synthetic methodology11b,17 entailing the synthesis of leucineꢀ and alanine epoxyketone as
well as solutionꢀphase assembly of the peptide epoxyketones from the corresponding alphaꢀ
amino acids (see experimental and Supporting Information). Cyclohexylꢀhomoalanine (in 15),
methylcyclohexylalanine (in 16), methoxycyclohexylalanine (in 17), the decalineꢀalanines (in 18
ACS Paragon Plus Environment
5

Journal of Medicinal Chemistry
Page 6 of 36
1
2
and 19),18 and bisꢀcyclohexylalanine (in 20, 29 and 30) were prepared from the corresponding
aromatic amino acids by hydrogenation in the presence of 5% rhodium on activated alumina (see
for an example the transformation of 31 to 32, Scheme 1. See also Supporting Information).19
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 2. Crystal structures of the mouse cCP and iCP either without (PDB code: 3UNE for
mouse cCP and 3UNH for mouse iCP) or in complex with 51 (ONXꢀ0914, PDB code: 3UNB for
mouse cCP and 3UNF for mouse iCP) revealed14 that the S1 pocket of subunit β5i is larger than
that of β5c (upper row of pictures). This difference in size was attributed to distinct
conformations of Met45 (in green). By contrast, the S3 pocket of β5i is smaller compared to β5c
due to Ser27 (in magenta), which in β5c is substituted for Ala. These structural differences
served as the starting point for the design of β5cꢀspecific ligands as described here.
a
OH
OH
H2N
31
BocHN
32
O
O
Scheme 1. Reagents and conditions: a) Rh/Al, H2, MeOH/HCl (1 M, aq) 30:1, then Boc2O, TEA,
THF, H2O.
ACS Paragon Plus Environment
6

Page 7 of 36
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Amino acids with chiral carbons emerging after hydrogenolysis were obtained as stereoisomeric
mixtures, incorporated in the peptide epoxyketones as such and the resulting diastereomers tested
as mixtures. All compounds were evaluated on their proteasome inhibition profile in extracts
from Raji cells (a human Bꢀcell lymphoma cell line that expresses both cCPs and iCPs) applying
our established ABPP assay.12 Compounds were tested at 0.01, 0.1, 1.0 and 10.0 ꢀM final
concentrations, resulting in distinct IC50 values as summarized in Table 1.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Table 1. Chemical structures of compounds 1ꢀ30 with their respective inhibitory activity against
β5c and β5i. High β5i/β5c ratio indicates selectivity for β5c.
Apparent IC50
(nM)
Ratio
Compoud
1
R4
R3
R2
R1
β5i
β5c
<10
β5i/β5c
>10
n.d*
n.d
2
3
4
5
6
7
<10
<10
<10
<10
<10
<10
<10
<10
<10
<10
<10
<10
n.d
n.d*
n.d
n.d*
n.d*
ACS Paragon Plus Environment
7

Journal of Medicinal Chemistry
Page 8 of 36
1
2
3
4
5
6
7
8
9
8
9
<10
<10
<10
n.d*
n.d*
<10
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
10
11
<10
<10
<10
<10
n.d*
n.d*
12
<10
<10
n.d*
13
14
72
50
33
2
<10
>5
15
35
36
1
16
17
18
19
18
31
15
25
<10
<10
<10
<10
>2
>3
>1.5*
>2.5*
20
21
1230
33
19
65
<10
>3*
ACS Paragon Plus Environment
8

Page 9 of 36
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
22
23
24
25
26
68
18
4
2
274
638
119
853
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
H
H
>10000 >10000
n.d.
>175
>10000
134
57
<10
853
33
27
28
29
H
H
H
>13
>12
>10000
>10000
>303
30
H
>10000
1230
>8
n.d. not determined
*Note: At 0.01 µM final concentration of 1 56% of β5i activity and 44% of β5c remained.
Values for compound 4 at 0.01 µM final concentration: 15% β5i activity and 6% β5c activity; for
6: 49% β5i and 16% β5c; for 7: 28% β5i and 12% β5c; for 8: 49% β5i and 14% β5c; for 9: 28%
β5i and 9% β5c; for 10: 36% β5i and 30% β5c; for 11: 23% β5i and 9% β5c; for 12: 44% β5i and
23% β5c; for 18: 40% β5c; for 19: 36% β5c; for 21: 24% β5c.
Compounds 1ꢀ12, all featuring a P1ꢀleucine and an aromatic residue at P3, are low nanomolar
inhibitors of both β5c and β5i. Discrimination between the two chymotryptic sites starts to
ACS Paragon Plus Environment
9

Journal of Medicinal Chemistry
Page 10 of 36
1
2
3
4
5
6
7
8
9
emerge within the compound series 13ꢀ24 with a leucine at P1 and an aliphatic, bulky moiety at
P3. Whereas adamantylalanine at P3 (24) is detrimental for both β5c and β5i inhibition, all other
bulky aliphatic P3ꢀgroups appear less well accommodated by β5i than β5c. This effect is most
pronounced in bisꢀcyclohexylalanine derivative 20, with an apparent IC50 for β5i of 1230 nM and
for β5c of 19 nM. While the presence of this steric residue decreased β5c inhibition slightly, it is
much more severe for β5i, thus introducing a superb possibility to achieve specificity. This
discrimination further increases when going from leucine to alanine at P1, as in 29 (which we
designated as LUꢀ005c, Figure 1), implementing a β5i/β5c ratio of over two orders of magnitude,
while having an IC50 value of 33 nM for β5c.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
To establish the IC50 values more accurately and also to obtain insights into the coꢀinhibition of
the β1c/β1i/β2c/β2i activities, we selected 6 compounds (2, 3, 5, 16, 17 and 20) and tested these
together with 29 in our competitive ABPP assay using Raji cell extracts at a wider range of final
concentrations (Table 2 and Figure S3). The obtained results agree with the trend from our initial
screen (Table 1 and Figure S2) and allow us to draw structureꢀactivity relationships. Compounds
2, 3, 5, 16, 17 and 20 all inhibit both β5c and β5i at low nanomolar concentrations and most are
crossꢀreactive towards β2c and β2i (Table 2). Increasing the size at P3 from Phe (2) to homoꢀPhe
(3) results in loss of activity against β1c, but not β1i. Further increasing the bulk to a
methylcyclohexylꢀalanine (16) results in loss of β1i activity as well, yielding a compound highly
active against the β2c/β2i/β5c/β5i catalytic subunits. Increasing the bulk to bisꢀcyclohexylalanine
(20) abolishes activity against β2c/β2i and leads to an inhibitor with considerable specificity for
β5c over β5i. Compound 29 differs from 20 in that the P1 site features the alanine side chain, as
opposed to leucine. This modification led to the desired β5c specificity.
ACS Paragon Plus Environment
10

Page 11 of 36
Journal of Medicinal Chemistry
1
2
3
4
Table 2. Apparent IC50 (µM) and pIC50 values of compounds 2, 3, 5, 16, 17, 20 and 29 against
5
6
the six catalytic sites from human cCPs and iCPs.
7
8
9
Apparent IC50 (µM) and pIC50 values
10
11
β5c
β5i
β2c
β2i
β1c
β1i
Compound
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
IC50
0.0021 8.67±0.06 0.0055 8.26±0.08 0.067 7.18±0.08 0.013 7.89±0.10 >100
0.003 8.53±0.06 0.0051 8.29±0.06 0.060 7.22±0.06 0.045 7.34±0.12 >100
0.0021 8.68±0.04
0.002 8.70±0.09
0.0021 8.68±0.10
0.017 7.77±0.07
0.0747 7.13±0.12
pIC50
IC50
pIC50
IC50
pIC50
IC50
pIC50
IC50
pIC50
IC50
pIC50
2
3
<4.0
<4.0
0.12
0.16
0.13
>100
0.89
>100
>100
6.91±0.08
6.79±0.12
6.88±0.13
<4.0
5
00078
0.011
0.022
0.439
16.7
8.11±0.08
0.11
6.95±0.10 0.029 7.54±0.09
0.18
6.75±0.22
<4.0
16
17
20
29
7.96±0.12 0.046 7.34±0.15 0.045 7.35±0.15 >100
7.67±0.09 0.066 7.18±0.11 0.063 7.20±0.14 >100
<4.0
6.05±0.28
<4.0
6.36±0.17
4.78±0.13
>100
>100
<4.0
<4.0
>100
>100
<4.0
<4.0
>100
>100
<4.0
<4.0
<4.0
Having identified the P1ꢀAla/P3ꢀbisꢀcyclohexylꢀAla (with Phe at P2) as the optimal sequence for
β5c selectivity, we decided to establish which of the two stereoisomers (cis or trans with respect
to the 1,4ꢀdisubstituted inner cyclohexane ring) present in compound 29 is the most potent and
most selective. For this, we designed a route of synthesis that would deliver both diastereomers
of 1,4ꢀbisꢀcyclohexylꢀLꢀalanine 32 (Scheme 1) in enantiomerically pure form.
ACS Paragon Plus Environment
11

Journal of Medicinal Chemistry
Page 12 of 36
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Scheme 2. Reagents and conditions: a) i) 2M NaOH; ii) C2H5OH, conc. H2SO4; b) i) Pd/C,
C2H5OH, H2; ii) NaOH; iii) silica gel column chromatography; c) i) LiAlH4/Et2O; ii)
TsCl/TEA/DCM; iii) NaCN/DMF; d) i) KOH/ethylene glycol; ii) N,Oꢀdimethylhydroxylamine
hydrochloride, HCTU/DiPEA/DCM; e) i) LiAlH4/Et2O; ii) 40/CuSO4/DCM; f) Et2AlCN/iꢀ
PrOH/THF; g) i) 6M HCl, reflux; ii) Boc2O/TEA/THF/H2O; h) Rh/Al, H2/MeOH.
The synthesis of transꢀbiscyclohexylꢀLꢀalanine 43 is depicted in Scheme 2 and commences with
hydrolysis of α,βꢀunsaturated nitrile 33, prepared according to the published procedure,19 to the
ACS Paragon Plus Environment
12

Page 13 of 36
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
corresponding carboxylate, which was esterified to give ethyl ester 34 (79% yield, two steps).
Hydrogenolysis of the alkene in 34 (H2, Pd/C) followed by saponification of the ethyl ester gave
the diastereomeric mixture of transꢀ and cisꢀ1,4ꢀdisubstituted cyclohexanes 35 and 36, which
was separated by silica gel flash column chromatography to provide 35 and 36 in yields of 29%
and 28%, respectively. The carboxylic acid in trans isomer 35 was reduced to the primary
alcohol, which in two steps (Oꢀtosylation followed by treatment with sodium cyanide, 76% yield,
three steps) afforded nitrile 37. Hydrolysis of the nitrile and condensation of the resulting acid
with N,Oꢀdimethylhydroxylamine yielded Weinreb amide 38 (96% yield), which was reduced to
the aldehyde and in situ subjected to transamination with sulfinamine20 40 to provide
sulfimimide 39. In the key step of the sequence, compound 39 was treated with Et2AlCN in a
mixture of iꢀPrOH and THF to give, in an asymmetric Strecker reaction, in good enantiomeric
excess cyanide 41 (de ≥ 96%). The stereoselectivity obtained in the asymmetric Strecker reaction
matches those observed by Cordi and coꢀworkers, who first reported21 the use of Sꢀsulfimimides
in this reaction and observed the predominant formation of Sꢀconfigured products as well.
Hydrolysis of the cyanide to the carboxylate and the sulfinimide to the amine, and ensuing Nꢀ
Boc protection afforded transꢀ4ꢀphenylꢀLꢀcyclohexylalanine 42 (47% over the three steps).
Hydrogenation of the phenyl moiety in 42 to the corresponding cyclohexyl (43, H2, Ru/Al,
quantitative yield) concluded the synthesis scheme. In a similar fashion and efficiency, cisꢀ1,4ꢀ
disubstituted cyclohexane 36 was transformed into the corresponding cyclohexylalanine
derivatives 44 and 45 (see Supporting Information for details).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The diastereomeric cyclohexylalanine derivatives 43 and 45, as well as their unsaturated
precursors 42 and 44, were introduced at P3 of the N3ꢀPheꢀXꢀPheꢀAlaꢀepoxyketone (EK)
sequence to yield the four structurally closely related peptide epoxyketones 46ꢀ49 (Figure 3).
ACS Paragon Plus Environment
13

Journal of Medicinal Chemistry
Page 14 of 36
1
2
3
4
5
6
7
8
9
These compounds were subjected to our competitive ABPP assay and the results compared to the
data obtained for the diastereomeric mixture of compounds 29 (Figure 3, Figure S4 and Table
S2). All four compounds proved to be potent and selective β5c inhibitors. As can be seen, the
transꢀsubstituted derivatives 48 and 49 are both the most potent and the most selective inhibitors
of the series (48: β5i/β5c = 2253, 49: β5i/β5c = 2694; in comparison, 46: β5i/β5c = 165, 47:
β5i/β5c = 30). When we assessed compound 49 in living RPMIꢀ8226 cells, though, no
proteasome inhibition was observed. Likely, compound 49 is too hydrophobic to cross the cell
membrane. The same probably applies to the stereoisomers of 49 (46, 47 and 48), which we did
not evaluate but which are equally apolar. To overcome this shortcoming (and as we had done
for the mixture of diastereomers 29 in our previous report12) we substituted the azidoPhe Nꢀ
terminal cap in 49 for the morpholinoacetylꢀLeu cap to obtain the peptide epoxyketone 50.
Though compound 50 is slightly less potent (IC50 for β5c 28 nM), it still possesses significant
selectivity (β5i/β5c = 814) and inhibits β5c in RPMIꢀ8226 cells with an IC50 for β5c of 274 nM,
whereas the IC50 values for all other subunits are above 100 ꢀM (Figure S5 and Table S2, S3).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
14

Page 15 of 36
Journal of Medicinal Chemistry
1
2
3
4
Figure 3. Apparent IC50 values of compounds 29 and 46ꢀ50 against the six catalytic sites from
5
6
human cCPs and iCPs.
7
8
9
Further evaluation of compound 50 was performed on AMOꢀ1 multiple myeloma cells
expressing ubiquitinated green fluorescent protein (UbꢀGFP)22. This tool allows direct
assessment of the accumulation of polyubiquitinated proteasome substrate protein upon
functional proteasome inhibition. We observed that, although β5c is effectively inhibited, this
does not result in UbꢀGFP accumulation (Figure 4), demonstrating that a functional loss in β5c
activity is entirely compensated for. Presumably, β5i – either alone or in combination with other
unaffected catalytic sites compensates for the loss of β5c activity in the turnover of UbꢀGFP. The
ability of different proteasomal subunits to compensate for one nonꢀfunctional subunit has been
shown previously.23 At 100 ꢁM final concentration of compound 50, β5i is partially coꢀinhibited,
and this leads to an increase in GFPꢀmediated fluorescence. In comparison, the β5c/β5i selective
inhibitor, NCꢀ005,24 causes accumulation of UbꢀGFP at 3 ꢁM. The maximal level of UbꢀGFP
accumulation was established by complete inhibition of all proteasome activities using a
combination of NCꢀ005 (blocking β5c/β5i), NCꢀ00124 (blocking β1c/β1i) and LUꢀ10225
(blocking β2c/β2i). Compound 50 has an in situ proteasome inhibition profile in AMOꢀ1 Ubꢀ
GFP cells similar to that observed in RPMIꢀ8226 cells (see SI Figure S6ꢀ10), and inhibitor
concentrations used in the UbꢀGFP accumulation experiments were based on this observation.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
15

Journal of Medicinal Chemistry
Page 16 of 36
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 4. AMOꢀ1 multiple myeloma cells expressing UbꢀGFP were treated with indicated
concentrations of compound 50 and β5c/β5i selective inhibitor NCꢀ005 for 1 h, washed with PBS
and cultivated for additional 12 h in medium before measuring the median fluorescence intensity
(MFI) of GFP. Complete proteasome inhibition (PanPI) can be achieved by combining 5 ꢁM
NCꢀ005 (β5c/β5i), 5 ꢁM NCꢀ001 (β1c/β1i) and 3 ꢁM LUꢀ102 (β2c/β2i).
With the aim to obtain insight in the observed subunit specificities, we set out to obtain crystal
structures of the most potent compounds complexed to proteasomes. Crystal soakings were only
successful for compounds 15 (P3 homocyclohexylalanine; P1 leucine) and 27 (P3
biphenylalanine; P1 alanine) (Tables S4 and S5). Due to their high hydrophobicity most
compounds precipitated under the crystallization condition of the yeast 20S proteasome core
particle (yCP). Notably, contrast to inhibition assays that use only low concentrations of protein
and inhibitor, the crystal drop is set up with a proteasome concentration of 40 mg/ml and
requires mM ranges of ligands to achieve full occupancy of the compounds at the active sites.
The obtained crystal structure of compound 15 shows that the P3ꢀcyclohexylꢀhomoalanine
residue of the ligand protrudes deeply into the S3 pocket of the yβ5/yβ6 substrate binding
channel and undergoes hydrophobic interactions with the amino acid side chains of Ala20, Ala27
and Val31 of yβ5 as well as the hydrophobic atoms of the side chain of Asp114 and Glu120 of
ACS Paragon Plus Environment
16

Page 17 of 36
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
subunit yβ6 (Figure 5A). The P1ꢀLeu moiety, known to be suitable for β5i and β5c, likely causes
the observed nonꢀselectivity of 15 between β5i and β5c (Table 1). Substituting the P3ꢀ
cyclohexylꢀhomoalanine residue of 15 for a bisꢀcyclohexylalanine derivative (20) significantly
increases β5cꢀselectivity (by a factor of 65). Most likely, Ser27 (β5i) shaping the S3 pocket of
the β5i/β6 substrate binding channel, sterically hinders binding of the hydrophobic bisꢀ
cyclohexylꢀhomoalanine, while Ala27 (β5c) favors the interaction with this Pꢀsite, thereby
promoting β5cꢀinhibition. In agreement with this, compound 20 is 35 times less active towards
β5i than compound 15 but slightly more potent for β5c (1.9 fold; Table 1). Substitution of the
P1ꢀLeu residue of inhibitor 20 for a P1ꢀAla residue yields compound 29, a very potent β5cꢀ
selective inhibitor. Due to its insolubility, we could not determine a crystal structure of 29,
however, structural data could be obtained on 27, a derivative of 29 featuring a P3ꢀbiphenyl
moiety instead of the saturated bisꢀcyclohexylꢀhomoalanine (Figure 5B). The P3ꢀbiphenyl
moiety of 27 adopts a different conformation than the cyclohexylꢀhomoalanine of 15 and
occupies a distinct wellꢀdefined cavity at the interface of the subunits yβ5 and yβ6. Ile100 and
Gly127 of subunit yβ6 form the bottom of this pocket, while His98 and Arg125 laterally restrict
it. Additional hydrophobic interactions are provided by Ala27 of yβ5 and the hydrophobic atoms
of the side chain of Asp114 (yβ6). Notably, the chemical character of the interacting yβ6
residues is conserved in mouse and human β6 (Figures 5B, S11A and S11B). As previously
noted for AcꢀLAAꢀEK,11b the P1ꢀAla residue of 27 does not change the conformation of Met45
(Figure 5C). Structural superpositions of 27 with AcꢀLAAꢀEK as well as carfilzomib bound to
subunit β5 of yeast26 and human proteasomes27 reveal that the ligands’ P1 and P2 residues align
well, whereas the P3 sites and the Nꢀcaps are more variable in their conformation (Figure 5D).
Altogether, the obtained structural data suggest that the overall binding mode of β5c selective
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
17

Journal of Medicinal Chemistry
Page 18 of 36
1
2
3
4
5
6
7
8
9
ligands like AcꢀLAAꢀEK and 27 is similar to that of nonꢀselective compounds like 15 and
carfilzomib. Thus, selectivity for either β5i or β5c is mainly gained by shape complementarity of
the P1 and P3 residues with the S1 and S3 pockets, respectively.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 5. Yeast 20S proteasomes in complex with the epoxyketones 15 (PDB code: 5JHS) and
27 (PDB code: 5JHR). (A, B) Compound 15 (A) and compound 27 (B) bound to the yβ5/yβ6
active site. The 2FOꢀFC electron density map for the ligand and Thr1 (black) is shown as a blue
mesh contoured at 1σ. Hydrogen bonds are indicated by black dotted lines. Amino acids are
labelled by the oneꢀletter code and numbered according to the sequence alignment to the
proteasomal β subunit of Thermoplasma acidophilum. (C) Structural superposition of the
compounds 15 and 27 bound to subunit yβ5. In contrast to 15, binding of 27 does not trigger
conformational changes of the Met45 side chain. (D) Structural superposition of the ligands 27,
ACS Paragon Plus Environment
18

Page 19 of 36
Journal of Medicinal Chemistry
1
2
3
4
5
6
AcꢀLAAꢀEK and carfilzomib bound to either yβ5 or the human constitutive β5c subunit depicts
their identical binding modes.
7
8
9
Discussion and Conclusions
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
This work describes the development and evaluation of a set of potent and highly selective
inhibitors of the β5c catalytic activity of human proteasomes. As we did before in our work on
β5i and β1iꢀspecific inhibitors,13 we sought to achieve selectivity, not so much by optimizing
affinity for the target subunit, as by introducing penalties for the most closely related subunits –
here β5i. The most selective compound in the described series, compound 49, inhibits β5c in the
nanomolar range and over three orders of magnitude more potently than β5i. The most prominent
distinguishing feature in 49 is its P3 substituent, being a steric bisꢀcyclohexylalanine residue. In
our previous work12 we introduced this residue as a mixture of stereoisomers. Here, we describe
a stereoselective synthesis of both enantiomers, which proceeds through their partially
unsaturated analogues. The route of synthesis we employed – making use of Strecker chemistry
starting from chiral sulfimimides – along with the synthesis of the four structurally related amino
acids 42ꢀ45 – may be of relevance for the construction of peptideꢀbased materials other than
those reported in the present work. The crystallographical insights confirm our design, which we
based on the murine cCP and iCP structures and the notion that β5c would prefer a large P3
residue accompanied by a small P1 residue.14
Compound 49 appeared not able to reach proteasomes within living cells, and we attribute this to
the fact that the molecule is too hydrophobic. With the aim to create a cellꢀpermeable analogue
we substituted the Nꢀterminal indene cap in 49 for a morpholinoꢀleucine, arriving at compound
50. Though this agent turned out to be somewhat less active and selective, it is cell permeable.
Therefore, compound 50 is the most effective compound described to date for specific chemical
ACS Paragon Plus Environment
19

Journal of Medicinal Chemistry
Page 20 of 36
1
2
3
4
5
6
7
8
9
knock down of β5c in cells. We demonstrate that selective inhibition of β5c in UbꢀGFPꢀ
transfected AMOꢀ1 cells (a multiple myeloma cell line expressing cCPs and iCPs in about equal
amounts) did not lead to accumulation of the fluorescent proteasome substrate. This result
underscores the findings in literature that inhibition of β5c alone is not sufficient to cause cell
death.28 In conclusion, we believe that our set of β5c inhibitors will become a highly valuable
tool for fundamental and applied biological as well as biomedical research on proteasomes in
relation to oncology and immunology. The β5cꢀselective inhibitors reported here may assist in
the establishment of the optimal cCP/iCP subunit inhibition regime, in cell lines and specifically
also in primary tumors, and with the aim to improve the therapeutic window. Furthermore, our
β5c selective inhibitors may be of use in the assessment, in a chemical ligandomics approach, of
the contribution of β5c to the production of MHCI peptide repertoires.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Experimental Section
General procedures
All reagents were of commercial grade and used as received unless indicated otherwise. The
1
purity of all tested compounds is >95% on the basis of LCꢀMS and NMR. H and 13C NMR
spectra were recorded on a Bruker AVꢀ400 (400 MHz), AVꢀ600 (600 MHz) or AVꢀ850 (850
MHz) spectrometer. Chemical shifts are given in ppm (δ) relative to CD3OD or CDCl3 as
1
internal standard. Coupling constants are given in Hz and peak assignments are based on 2D H
13
13
COSY and C HSQC NMR experiments. All presented C APT spectra are proton decoupled.
LCꢀMS analysis was performed on a Finnigan Surveyor HPLC system with a Gemini C18 50 ×
4.60 mm column (detection at 200−600 nm) coupled to a Finnigan LCQ Advantage Max mass
spectrometer with ESI. Methods used are: 15 min (0ꢀ0.5 min: 10% MeCN; 0.5ꢀ10.5 min: 10% to
ACS Paragon Plus Environment
20

Page 21 of 36
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
90% MeCN; 10.5ꢀ12.5 min: 90% MeCN; 12.5ꢀ15 min: 90% to 10% MeCN) or 12.5 min (0ꢀ0.5
min: 10% MeCN; 0.5ꢀ8.5 min: 10% to 90% MeCN; 8.5ꢀ10.5 min: 90% MeCN; 10.5ꢀ12.5 min:
90% to 10% MeCN). HRMS was recorded on a LTQ Orbitrap (ThermoFinnigan). For reverse
phase HPLC purification, an automated Gilson HPLC system equipped with a C18 semiprep
column (Phenomenex Gemini C18, 5ꢁm 250×10 mm) and a GX281 fraction collector was used.
General procedure for azide couplings
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Compounds 1ꢀ30 and 46ꢀ50 were prepared via azide coupling of properly protected tripeptide
hydrazide and properly deprotected Leu epoxyketone amines and Ala epoxyketone amines. The
appropriate hydrazide was dissolved in 1:1 DMF:DCM (v/v) and cooled to ꢀ30 °C. tBuONO (1.1
eq.) and HCl (4N solution in 1,4ꢀdioxane, 2.8 eq.) were added and the mixture was stirred for 3 h
at ꢀ30 °C after which TLC analysis (10% MeOH/DCM, v/v) showed complete consumption of
the starting material. The epoxyketone was added as a free amine to the reaction mixture as a
solution in DMF with 5.0 eq. of DiPEA. The mixture was allowed to warm to r.t. slowly
overnight. The mixture was diluted with EtOAc and extracted with H2O (3x) and brine. The
organic layer was dried over MgSO4 and purified by flash column chromatography (1ꢀ5%
MeOH in DCM) or reverse phase HPLC. Peptide hydrazides were prepared by hydrazinolysis of
peptide methyl esters synthesized by standard procedures of solution peptide chemistry as
described in the Supporting Information.
N3PheꢀBiChaꢀPheꢀLeuꢀEK (20). The tripeptide hydrazide N3PheꢀBiChaꢀPheꢀNHNH2 was
prepared according to literature procedures12 and the title compound was prepared according to
the general procedure for azide coupling on a 50 ꢁmol scale. Purification by HPLC (85%ꢀ100%
MeCNꢀH2O) yielded the title compound (7.5 mg, 10.3 ꢁmol, 21%). 1H NMR (600 MHz, MeOD)
δ 7.38ꢀ7.09 (m, 10H), 4.66ꢀ4.62 (m, 1H), 4.57ꢀ4.54 (m, 1H), 4.44ꢀ4.36 (m, 1H), 4.10ꢀ4.07 (m,
ACS Paragon Plus Environment
21

Journal of Medicinal Chemistry
Page 22 of 36
1
2
3
4
1H), 3.23 (t, J = 4.5 Hz, 1H), 3.14ꢀ3.10 (m, 2H), 3.00ꢀ2.83 (m, 3H), 1.84ꢀ1.64 (m, 8H), 1.52ꢀ0.88
5
6
13
(m, 27H). C NMR (150 MHz, MeOD) δ 209.33, 209.31, 173.92, 173.88, 173.19, 173.17,
7
8
9
171.45, 171.39, 138.22, 138.20, 137.89, 137.84, 130.43, 130.42, 130.40, 129.61, 129.59, 129.42,
128.02, 127.72, 65.40, 59.99, 59.98, 55.50, 52.97, 52.94, 52.56, 51.44, 44.75, 44.67, 40.51,
40.41, 40.37, 38.81, 38.75, 38.69, 38.66, 35.49, 34.93, 33.73, 31.70, 31.44, 31.23, 30.93, 30.79,
29.40, 27.96, 27.89, 27.87, 26.64, 26.35, 26.23, 23.79, 21.62, 16.94. LCꢀMS (linear gradient 10
→ 90% MeCN/H2O, 0.1% TFA, 12.5 min): Rt (min): 11.43 (ESIꢀMS (m/z): 727.33, (M+H+)).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
HRMS calculated for C42H58N6O5 727.45415 [M+H]+; found 727.45465.
N3PheꢀcisꢀCha(4ꢀPhe)ꢀPheꢀAlaꢀEK (46). The title compound was prepared according to the
general procedure for azide coupling on a 50 ꢁmol scale. Purification by HPLC (70%ꢀ85%
MeCNꢀH2O) yielded the title compound (6.9 mg, 10.2 ꢁmol, 20%). 1H NMR (400 MHz, CDCl3)
δ 7.37ꢀ7.13 (m, 17H), 6.71 (d, J = 7.7 Hz, 1H), 6.52 (d, J = 7.9 Hz, 1H), 6.39 (d, J = 7.0 Hz, 1H),
4.61 (q, J = 7.2 Hz, 1H), 4.52ꢀ4.45 (m, 1H), 4.32ꢀ4.26 (m, 1H), 4.13ꢀ4.10 (m, 1H), 3.30ꢀ3.26 (m,
1H), 3.17 (d, J = 5.0 Hz, 1H), 3.10ꢀ3.00 (m, 3H), 2.89 (d, J = 4.9 Hz, 1H), 2.56ꢀ2.50 (m, 1H),
1.97ꢀ1.82 (m, 1H), 1.68ꢀ1.39 (m, 14H), 1.24 (d, J = 7.0 Hz, 3H). 13C NMR (100 MHz, CDCl3) δ
207.82, 171.36, 170.13, 168.86, 147.05, 136.48, 135.79, 129.68, 129.41, 128.81, 128.75, 128.44,
127.47, 127.19, 127.00, 126.02, 65.16, 59.02, 54.31, 52.54, 51.83, 48.18, 43.44, 38.31, 37.94,
33.60, 30.61, 29.43, 29.16, 28.87, 28.61, 17.32, 16.92. LCꢀMS (linear gradient 10 → 90%
MeCN/H2O, 0.1% TFA, 12.5 min): Rt (min): 9.40 (ESIꢀMS (m/z): 678.93, (M+H+)). HRMS
calculated for C39H46N6O5 679.36025 [M+H]+; found 679.36047.
N3PheꢀcisꢀBiChaꢀPheꢀAlaꢀEK (47). This compound was prepared according to the general
procedure for azide coupling on a 50 ꢁmol scale. Purification by HPLC (70%ꢀ85% MeCNꢀH2O)
yielded the title compound (16.1 mg, 23.5 ꢁmol, 47%). 1H NMR (400 MHz, CDCl3) δ 7.35ꢀ7.13
ACS Paragon Plus Environment
22

Page 23 of 36
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
(m, 10H), 6.66 (d, J = 7.7 Hz, 1H), 6.50 (d, J = 7.8 Hz, 1H), 6.39 (d, J = 7.0 Hz, 1H), 4.63ꢀ4.57
(m, 1H), 4.51ꢀ4.44 (m, 1H), 4.29ꢀ4.23 (m, 1H), 4.10ꢀ4.07 (m, 1H), 3.27 (dd, J = 14.1, 4.1 Hz,
1H), 3.17 (d, J = 4.0 Hz, 1H), 3.10ꢀ2.93 (m, 3H), 2.89 (d, J = 4.9 Hz, 1H), 1.77ꢀ1.62 (m, 6H),
1.51 (s, 3H), 1.42ꢀ1.01 (m, 18H), 0.94ꢀ0.76 (m, 2H). 13C NMR (100 MHz, CDCl3) δ 207.66,
171.33, 170.03, 168.70, 136.36, 135.73, 129.50, 129.28, 128.68, 128.60, 127.32, 127.03, 65.05,
58.88, 54.12, 52.40, 51.60, 48.01, 41.65, 40.42, 38.23, 37.80, 34.85, 30.68, 30.50, 29.89, 28.78,
26.74, 25.37, 25.20, 17.17, 16.78. LCꢀMS (linear gradient 10 → 90% MeCN/H2O, 0.1% TFA,
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
12.5 min): Rt (min): 10.53 (ESIꢀMS (m/z): 685.13, (M+H+)). HRMS calculated for C39H52N6O5
685.40720 [M+H]+; found 685.40722.
N3PheꢀtransꢀCha(4ꢀPhe)ꢀPheꢀAlaꢀEK (48). This compound was prepared according to the
general procedure for azide coupling on a 50 ꢁmol scale. Purification by HPLC (70%ꢀ85%
MeCNꢀH2O) yielded the title compound (5.8 mg, 8.5 ꢁmol, 17%). 1H NMR (850 MHz, CDCl3) δ
7.35ꢀ7.27 (m, 7H), 7.26ꢀ7.17 (m, 8H), 6.65 (d, J = 7.6 Hz, 1H), 6.53 (d, J = 7.8 Hz, 1H), 6.35 (d,
J = 7.0 Hz, 1H), 4.65ꢀ4.60 (m, 1H), 4.50ꢀ4.47 (m, 1H), 4.37ꢀ4.34 (m, 1H), 4.10ꢀ4.08 (m, 1H),
3.31ꢀ3.28 (m, 1H), 3.16 (d, J = 4.9, 1H), 3.07 (d, J = 7.1 Hz, 2H), 3.04ꢀ3.02 (m, 1H), 2.90 (d, J =
4.9 Hz, 1H), 2.44ꢀ2.41 (m, 1H), 1.88ꢀ1.84 (m, 2H), 1.80ꢀ1.73 (m, 2H), 1.67ꢀ1.64 (m, 1H), 1.52
13
(s, 3H), 1.42ꢀ1.35 (m, 3H), 1.25 (d, J = 7.1 Hz, 3H), 1.10ꢀ0.97 (m, 3H). C NMR (213 MHz,
CDCl3) δ 207.77, 171.37, 170.14, 168.85, 147.40, 136.43, 135.87, 129.65, 129.40, 128.86,
128.78, 128.47, 127.51, 127.22, 126.89, 126.10, 65.18, 59.02, 54.27, 52.54, 51.20, 48.21, 44.28,
38.89, 38.39, 37.98, 34.00, 33.89, 33.72, 32.92, 17.37, 16.93. LCꢀMS (linear gradient 10 → 90%
MeCN/H2O, 0.1% TFA, 12.5 min): Rt (min): 9.39 (ESIꢀMS (m/z): 679.00, (M+H+)). HRMS
calculated for C39H46N6O5 679.36025 [M+H]+; found 679.36043.
ACS Paragon Plus Environment
23

Journal of Medicinal Chemistry
Page 24 of 36
1
2
3
4
5
6
7
8
9
N3PheꢀtransꢀBiChaꢀPheꢀAlaꢀEK (49). This compound was prepared according to the general
procedure for azide coupling on a 50 ꢁmol scale. Purification by HPLC (80%ꢀ95% MeCNꢀH2O)
yielded the title compound (11.3 mg, 16.5 ꢁmol, 33%). 1H NMR (400 MHz, CDCl3) δ 7.38ꢀ7.11
(m, 10H), 6.64 (d, J = 7.7 Hz, 1H), 6.51 (d, J = 7.8 Hz, 1H), 6.39 (d, J = 7.0 Hz, 1H), 4.66ꢀ4.58
(m, 1H), 4.51ꢀ4.44 (m, 1H), 4.36 – 4.30 (m, 1H), 4.13ꢀ4.05 (m, 1H), 3.28 (dd, J = 14.1, 4.1 Hz,
1H), 3.16 (d, J = 4.9 Hz, 1H), 3.09ꢀ2.95 (m, 3H), 2.89 (d, J = 4.9 Hz, 1H), 1.75ꢀ1.56 (m, 9H),
1.51 (s, 3H), 1.38ꢀ0.73 (m, 17H). 13C NMR (100 MHz, CDCl3) δ 207.75, 171.46, 170.16, 168.80,
136.50, 135.95, 129.61, 129.40, 128.85, 128.73, 127.48, 127.16, 65.22, 59.00, 54.24, 52.52,
51.29, 48.13, 43.39, 43.27, 39.08, 38.43, 37.99, 34.35, 33.81, 33.00, 30.36, 29.65, 26.98, 17.33,
16.91. LCꢀMS (linear gradient 10 → 90% MeCN/H2O, 0.1% TFA, 12.5 min): Rt (min): 10.53
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
(ESIꢀMS (m/z): 685.00, (M+H+)). HRMS calculated for C39H52N6O5 685.40720 [M+H]+; found
685.40740.
MorphꢀLeuꢀtransꢀBiChaꢀPheꢀAlaꢀEK TFA salt (50) This compound was prepared according
to the general procedure for azide coupling on a 50 ꢁmol scale. Purification by HPLC (50%ꢀ55%
MeCNꢀH2O) yielded the title compound (9.9 mg, 11.4ꢁmol, 23%). 1H NMR (400 MHz, MeOD)
δ 7.32ꢀ7.15 (m, 5H), 4.62ꢀ4.58(m, 1H), 4.49ꢀ4.32 (m, 3H), 3.95ꢀ3.74 (m, 6H), 3.20 (s, 6H), 2.98ꢀ
2.82 (m, 2H), 1.82ꢀ1.60 (m, 10H), 1.60ꢀ1.45 (m, 7H), 1.31ꢀ1.11 (m, 7H), 1.08ꢀ0.80 (m, 14H). 13C
NMR (100 MHz, MeOD) δ 209.33, 174.34, 174.24, 172.76, 138.24, 130.43, 129.38, 127.72,
65.45, 60.00, 55.27, 54.09, 53.25, 53.11, 52.61, 49.09, 49.00, 44.77, 44.65, 41.97, 40.44, 38.84,
35.49, 34.91, 33.64, 31.47, 31.43, 30.99, 30.89, 27.94, 25.94, 23.51, 21.80, 16.98, 16.48. LCꢀMS
(linear gradient 10 → 90% MeCN/H2O, 0.1% TFA, 15.0 min): Rt (min): 7.30 (ESIꢀMS (m/z):
752.6, (M+H+)). HRMS calculated for C42H65N5O7752.49568 [M+H]+; found 752.49595.
Biological analysis.
ACS Paragon Plus Environment
24

Page 25 of 36
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Competitive activity-based protein profiling assay in Cell Lysates. Lysates of Raji cells were
prepared by sonication in 3 volumes of lysis buffer containing 50 mM Tris pH 7.5, 1 mM DTT, 5
mM MgCl2, 250 mM sucrose, 2 mM ATP, and 0.05% (w/v) digitonin. Protein concentration was
determined by the Bradford assay. Cell lysates (diluted to 5 ꢁg of total protein in buffer
containing 50 mM Tris pH 7.5, 2 mM DTT, 5 mM MgCl2, 10% (v/v) glycerol, 2 mM ATP) were
exposed to the inhibitors for 1 h at 37 °C prior to incubation with cocktail ABPs for another 1 h,
followed by 3 min boiling with a reducing gelꢀloading buffer and fractionation on 12.5% SDS
PAGE. Inꢀgel detection of residual proteasome activity was performed in the wet gel slabs
directly on a ChemiDoc™ MP system using Cy2 settings to detect BODIPY(FL)ꢀLUꢀ112, Cy3
settings to detect BODIPY(TMR)ꢀNCꢀ005ꢀVS and Cy5 settings to detect Cy5ꢀNCꢀ001.
Intensities of bands were measured by fluorescent densitometry and normalized to the intensity
of bands in mockꢀtreated extracts. Average valuesof three independent experiments were plotted
against inhibitor concentrations in Graphpad Prism (in the initial screening(Table 1, Table S1),
experiments were conducted once). pIC50 values and associated standard deviations were
calculated by Graphpad Prism (note that we converted pIC50 values to IC50 for clarification,
however this method used allows the determination of the standard deviations only from the
pIC50 values.)
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Competitive activity-based protein profiling assay in living RPMI-8226 cells. RPMIꢀ8226 were
cultured in RPMIꢀ1640 media supplemented with 10% (v/v) fetal calf serum, GlutaMAX™,
penicillin/streptomycin in a 5% CO2 humidified incubator. An amount of (5ꢀ8) x105 cells/mL
cells was exposed to inhibitors for 1 h at 37 °C. Cells were harvested and washed twice with
PBS. Cell pellets were treated with lysis buffer (50 mM Tris pH 7.5, 2 mM DTT, 5 mM MgCl2,
10% (v/v) glycerol, 2 mM ATP, 0.05% (w/v) digitonin) on ice for 1 h, followed by
ACS Paragon Plus Environment
25

Journal of Medicinal Chemistry
Page 26 of 36
1
2
3
4
5
6
7
8
9
centrifugation at 14000 rpm for 15 min. Proteasome inhibition in the obtained cell lysates was
determined using the method described above. Intensities of bands were measured by fluorescent
densitometry and divided by the intensity of bands in mockꢀtreated extracts. Gels were stained
by Coomassie Brilliant Blue, which was used to correct for gel loading differences. Average
values of three independent experiments were plotted against inhibitor concentrations. IC50
(inhibitor concentrations giving 50% inhibition) values were calculated using GraphPad Prism
software.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Functional validation of compound 50 on AMO-1 multiple myeloma cells expressing Ub-GFP.
AMOꢀ1 multiple myeloma cell line was maintained in RPMIꢀ1640 medium supplemented with
10% (v/v) fetal calf serum and penicillin/streptomycin (all Sigma, USA). Cells were
electroporated using GenePulser II (Biorad, USA) with UbꢀG76VꢀGFP plasmid, which was a gift
from Nico Dantuma (Addgene plasmid #11941). This insert allows estimating the activity of the
proteasome, as GFP is tagged by Ub and thus immediately degraded. However, when the
proteasome is inhibited, the GFP substrate accumulates within the cell and increase of GFP
fluorescence can be observed. Cells carrying plasmid were selected using G418, 500 ꢁg/mL
(Gibco,ThermoFisher, USA) and subcloned using MethoCult (StemCell technologies, USA).
After selection, positive clones expressing UbꢀG76VꢀGFP were identified using FACS Canto II
(BD Biosciences, USA). [briefly: 5x105 cells were seeded and treated for 1h at 37 °C with
indicated concentrations of PI (50/NC005). Cells were harvested, washed with PBS to remove PI
and subsequently cultured in PIꢀfree medium for 12h under normal culture conditions (37 °C, 5%
CO2). Then, cells were washed in PBS and analyzed on FACS Canto II for MFI in FITC channel.
MFI was analyzed in FlowJo software.)] Upon treatment with 10 nM bortezomib for 1 h (pulse)
ACS Paragon Plus Environment
26

Page 27 of 36
Journal of Medicinal Chemistry
1
2
3
4
5
6
followed by a 12 h incubation (chase), the clone giving the highest GFP fluorescence was used
for further analyses.
7
8
9
Competitive activity-based protein profiling assay in living AMO-1 Ub-GFP cells. AMOꢀ1 Ubꢀ
GFP were cultured in RPMIꢀ1640 media (SigmaꢀAldrich, USA) supplemented with 10% (v/v)
fetal calf serum (SigmaꢀAldrich, USA), penicillin/streptomycin (Gibco/ThermoFisher Scientific,
USA) in a 5% CO2 humidified incubator. An amount of 2 x106 cells/mL cells was exposed to
indicated concentrations of proteasome inhibitors for 1 h at 37°C. Cells were harvested, washed
twice with PBS and lysed with lysis buffer (20 mM TrisꢀHCl, 10 mM EDTA, 300 mM NaCl, 2%
(v/v) TritonꢀX, 2 nM NꢀEthylmaleimide, Halt™ꢀProtease and Phosphatase Inhibitor Cocktail
(ThermoFisher Scientific, USA) on ice for 10 minutes, followed by centrifugation at 14000 rpm
for 15 min at 4 °C. For the tricolor labeling, protein load was adjusted to 20 µg using Bradfordꢀ
assay (Roti®ꢀNanoquant, ROTH, Germany) in 9 µl and cocktail ABPs (1 µl) was added to the
lysate in indicated final concentrations. Mixture was incubated for 1 h at 37 °C. Subsequently, 4
µl of loading buffer (Biorad, USA) was added and incubated for 2 min at 90 °C. Proteins were
separated using SDSꢀPAGE on 12% precast MOPC gels (NuPAGE® BisꢀTris Precast Gels;
ThermoFisher Scientific, USA). Pictures of proteasome subunits were acquired by FUSION
SOLOꢀS (Vilber Lourmat, Germany) equipped with fluorescence excitation/emission filters.
Data evaluation was performed using Vision software (Vilber Lourmat, Germany).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Crystallographic analysis. yCP crystals were grown by hanging drop vapor diffusion as
previously described.29 Inhibitor complex structures were obtained by incubating crystals in 5 µl
cryobuffer (20 mM magnesium acetate, 100 mM 2ꢀ(Nꢀmorpholino)ethanesulfonic acid, pH 6.8
and 30% (v/v) 2ꢀmethylꢀ2,4ꢀpentanediol) supplemented with 0.5 µl of inhibitor (50 mM in
DMSO) for at least 12 h. Diffraction data were collected at the beamline X06SA at the Paul
ACS Paragon Plus Environment
27

Journal of Medicinal Chemistry
Page 28 of 36
1
2
3
4
5
6
7
8
9
Scherrer Institute, SLS, Villigen, Switzerland (λ= 1.0 Å). Evaluation of reflection intensities and
data reduction were performed with the program package XDS.30 Molecular replacement was
carried out with the coordinates of the yeast 20S proteasome (PDB entry code: 5CZ431) by rigid
body refinements (REFMAC532). MAIN33 and COOT34 were used to build models.
Translation/Libration/Screw (TLS) refinements finally yielded excellent R factors as well as
r.m.s.d. bond and angle values. The coordinates, proven to have good stereochemistry from the
Ramachandran plots, were deposited in the RCSB Protein Data Bank under the accession codes
5JHS (yCP:15) and 5JHR (yCP:27) (Table S6).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ASSOCIATED CONTENT
Supporting Information
Assays of compounds 1ꢀ30 and 46ꢀ50 in Raji lysates, pIC50 values and standard errors of all
compounds in cell lysates and intact cells, structures of activityꢀbased probes, Xꢀray data table,
complete synthetic details and characterization of all compounds and synthetic intermediates,
NMR spectra and LCꢀMS traces of compounds 46, 47, 48, 49, 50. This material is available free
of charge via the Internet at http://pubs.acs.org.
Accession Codes
Structure factors and coordinates were deposited in the RCSB Protein Data Bank under the
accession codes: 5JHS (yCP:15) and 5JHR (yCP:27).
AUTHOR INFORMATION
Corresponding Author
ACS Paragon Plus Environment
28

Page 29 of 36
Journal of Medicinal Chemistry
1
2
3
4
Email: h.s.overkleeft@chem.leidenuniv.nl; tel: +31(0)715274307.
5
6
7
Author Contributions
8
9
BTX and GdB synthesized the compounds under the supervision of GvdM. BTX carried out and
analyzed the proteasome inhibition assays under the supervision of BIF and with support from
MvdS and AFK. EMH and MG generated the crystal structures. BA and CD conducted the
functional proteasome inhibition assays. BTX, EMH, MG and HSO wrote the manuscript. HSO
conceived and supervised the research project.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Funding Sources
This work was supported by the China Scholarship Council (CSC PhD fellowship, to BTX), the
Netherlands Organization for Scientific Research (NWOꢀCW, TOPꢀPUNT grant, to HSO) and
the Deutsche Forschungsgemeinschaft (DFG, grant GR1861/10ꢀ1, to MG).
Notes
We thank R. Feicht for the purification of yeast 20S proteasomes and the staff of the beamline
X06SA at the Paul Scherrer Institute, Swiss Light Source, Villigen (Switzerland) for assistance
during data collection.
ABBREVIATIONS
CP, proteasome core particle; cCPs, constitutive proteasome core particles; iCPs,
immunoproteasome core particles; tCPs, thymoproteasome core particles; yCP, yeast proteasome
core particles; MHCI, major histocompatibility complex class I; ABPP, activityꢀbased protein
profiling; EK, epoxyketone; UbꢀGFP, ubiquitinated green fluorescent protein; MFI, median
ACS Paragon Plus Environment
29