Required Immunoproteasome Subunit Inhibition Profile for Anti-Inflammatory Efficacy and Clinical Candidate KZR-616 ((2 S,3 R)- N-((S)-3-(Cyclopent-1-en-1-yl)-1-((R)-2-methyloxiran-2-yl)-1-oxopropan-2-yl)-3-hydroxy-3-(4-methoxyphenyl)-2-((S)-2-(2-morpholinoacetamido)propanamido)propenamide)

Article abstract of DOI:10.1021/acs.jmedchem.8b01201

Selective immunoproteasome inhibition is a promising approach for treating autoimmune disorders, but optimal proteolytic active site subunit inhibition profiles remain unknown. We reveal here our design of peptide epoxyketone-based selective low molecular mass polypeptide-7 (LMP7) and multicatalytic endopeptidase complex subunit-1 (MECL-1) subunit inhibitors. Utilizing these and our previously disclosed low molecular mass polypeptide-2 (LMP2) inhibitor, we demonstrate a requirement of dual LMP7/LMP2 or LMP7/MECL-1 subunit inhibition profiles for potent cytokine expression inhibition and in vivo efficacy in an inflammatory disease model. These and additional findings toward optimized solubility led the design and selection of KZR-616 disclosed here and presently in clinical trials for treatment of rheumatic disease.

Full text of DOI:10.1021/acs.jmedchem.8b01201

Subscriber access provided by University of Sunderland
Article
A required immunoproteasome subunit inhibition profile for anti-inflammatory
efficacy and clinical candidate KZR-616 ((2S,3R)-N-((S)-3-(cyclopent-1-
en-1-yl)-1-((R)-2-methyloxiran-2-yl)-1-oxopropan-2-yl)-3-hydroxy-3-(4-
methoxyphenyl)-2-((S)-2-(2-morpholinoacetamido)propanamido)propenamide)
Henry W.B. Johnson, Eric Lowe, Janet L. Anderl, Andrea Fan, Tony Muchamuel,
Simeon Bowers, David Moebius, Christopher Kirk, and Dustin L. McMinn
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.8b01201 • Publication Date (Web): 31 Oct 2018
Downloaded from http://pubs.acs.org on November 1, 2018
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 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 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.
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 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
A required immunoproteasome subunit inhibition
profile for anti-inflammatory efficacy and clinical
candidate KZR-616 ((2S,3R)-N-((S)-3-(cyclopent-1-en-1-
yl)-1-((R)-2-methyloxiran-2-yl)-1-oxopropan-2-yl)-3-
hydroxy-3-(4-methoxyphenyl)-2-((S)-2-(2-
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
morpholinoacetamido)propanamido)propenamide)
Henry W.B. Johnson1, Eric Lowe1, Janet L. Anderl1, Andrea Fan1,
Tony Muchamuel1, Simeon Bowers2, David C. Moebius2,
Christopher Kirk1, & Dustin L. McMinn1,*
1Kezar Life Sciences, South San Francisco, California, 94080, USA
2Onyx Pharmaceuticals, An Amgen Subsidiary, South San Francisco, California 94080, USA
Material requests and correspondence should be addressed to:
Dustin L. McMinn, Kezar Life Sciences, 4000 Shoreline Ct, Suite 300, South San Francisco, California,
94080, USA, e-mail: dmcminn@kezarbio.com
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 49
1
2
3
4
Abstract
5
6
7
8
9
Selective immunoproteasome inhibition is a promising approach for treating autoimmune disorders, but
optimal proteolytic active site subunit inhibition profiles remain unknown. We reveal here our design of
peptide epoxyketone-based selective low molecular mass polypeptide-7 (LMP7) and multicatalytic
endopeptidase complex subunit-1 (MECL-1) subunit inhibitors. Utilizing these and our previously
disclosed low molecular mass polypeptide-2 (LMP2) inhibitor, we demonstrate a requirement of dual
LMP7/LMP2 or LMP7/MECL-1 subunit inhibition profiles for potent cytokine expression inhibition and in
vivo efficacy in an inflammatory disease model. These and additional findings toward optimized solubility
led the design and selection of KZR-616 disclosed here and presently in clinical trials for treatment of
rheumatic disease.
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

Page 3 of 49
Journal of Medicinal Chemistry
1
2
3
4
Introduction
5
6
7
8
9
Proteasomes control several cellular processes by mediating protein homeostasis through regulated
protein degradation1. Proteasome function is tuned between a ubiquitously expressed constitutive
proteasome and an immuno-optimized form, or immunoproteasome, largely confined to cells of
hematopoietic origin or induced elsewhere by inflammatory cytokines. These isoforms further differ by
subtle sequence differences in proteolytic subunits of their 20S core particles (Figure 1). The constitutive
proteasome contains 2 copies each of three proteolytic subunits within its 20S core, β1, β2 and β5, while
the immunoproteasome replaces these with low-molecular mass polypeptide-2 (LMP2 or β1i), multi-
catalytic endopeptidase complex-like 1 (MECL-1 or β2i), and low-molecular mass polypeptide-7 (LMP7 or
β5i), respectively. Sequence differences between analogous subunits render altered substrate
preferences and engender specialized roles for the immunoproteasome (antigen presentation, cytokine
regulation, etc.)2. Increased expression of the immunoproteasome is observed at sites of inflammation
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
in several autoimmune disorders3,4,5
.
Figure 1. Subunit composition of the proteasome 20S core particle β rings.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 49
1
2
3
4
5
6
7
8
9
Combined inhibition of both forms of the proteasome, specifically the β5 and LMP7 subunits, drives cell
death and has been an effective drug strategy against multiple myeloma6, but chronic administration in
non-life-threatening disorders, such as autoimmunity, is precluded by systemic toxicities including
anemia, thrombocytopenia, and neutropenia7. Nonetheless, bortezomib, a dual proteasome inhibitor,
demonstrated efficacy in mouse models of autoimmunity and resulted in improved clinical outcomes in
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
patients with systemic lupus erythematosus8,9,10 and other autoimmune disorders11,12
.
ONX 0914 (Figure 2), a peptide epoxyketone-based immunoproteasome-selective inhibitor, potently
blocks inflammatory cytokine production and alters pro-inflammatory T-cell plasticity without the
antiproliferative effects of non-selective proteasome inhibitors13. Moreover, ONX 0914 is efficacious in
animal models of autoimmune disease (rheumatoid arthritis, type-I diabetes, multiple sclerosis14, colitis15,
and lupus16) and results in no loss of normal immune function. At effective doses, ONX 0914 maximally
inhibited mouse LMP7 while sparing constitutive proteasome subunit activity13. Though these results
suggested LMP7 was the primary therapeutic target, efficacious doses were associated with 60%
inhibition of LMP2 and a low level of MECL-1 inhibition. Moreover, partial LMP2/MECL-1 inhibition by
ONX 0914 in stimulated T-cells enhanced in vitro cytokine inhibition13. Together, these observations
suggested the anti-inflammatory effects of ONX 0914 may be potentiated by LMP2 and/or MECL-1
inhibition, an implication which we sought to clarify through design and assessment of epoxyketone-
based selective inhibitors to each of the immunoproteasome subunits.
Results and Discussion
Chemistry. The generation of LMP7 inhibitors 3-8 (Scheme 1) involved standard peptide coupling
chemistry and commenced with the coupling of the BOC protected amino acid with the appropriate keto-
epoxide TFA salt. The dipeptide building block was then deprotected and coupled to D or L-BOC protected
ACS Paragon Plus Environment

Page 5 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
alanine to furbish the BOC intermediate. After deprotection the TFA salt was carried into a final amide
coupling step.
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 1.a
O
R
R3
O
R3
O
i
O
R
BocHN
O
O
+
OH
BocHN
N
H2N
TFA
H
R3
O
R1
O
R
H
O
N
v
ii, iii, iv
H3N
O
N
H
O
O
F3C
R3
O
O
R1
O
H
O
N
R2
N
N
H
H
O
R
R
R1
R2
R3
Ph
Compound
indole
2-methylindene
2-methylindene
cyclohexyl
3
4
5
6
7
8
epi-Me
Me
Ph
Ph
Ph
4-methoxyphenyl
4-methoxyphenyl
4-methoxyphenyl
4-methoxyphenyl
4-methoxyphenyl
Me
trans-4-methylcyclohexanol
trans-4-methylcyclohexanol
trans-4-methylcyclohexanol
Me
c-butyl
cyclopentenyl
Me
Me
aReagents
and
conditions:
(i)
hydroxybenzotriazole,
2-(1H-benzotriazol-1-yl)-1,1,3,3-tetramethyluronium
hexafluorophosphate (HBTU), N-ethyl-N-isopropyl-propan-2-amine, DMF, 0 oC -> rt (ii) TFA, DCM (iii) (tert-
butoxycarbonyl)-L-alanine or (tert-butoxycarbonyl)-D-alanine, 1-[bis(dimethylamino)methylene]-1H-1,2,3-triazolo[4,5-
o
b]pyridinium 3-oxide hexafluorophosphate (HATU), N-ethyl-N-isopropyl-propan-2-amine, DMF, 0 C -> rt (iv) TFA,
DCM (v) R2-COOH, HATU, N-ethyl-N-isopropyl-propan-2-amine, DMF, 0 oC -> rt.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 49
1
2
3
4
5
6
7
8
9
The synthesis of MECL-1 inhibitors (Scheme 2) started with a coupling of BOC-leucine to L-phenylalanine
benzyl ester followed by a deprotection to furnish the TFA salt (13). The N-cap was synthesized by
hydrolyzing methyl 2-(azepan-1-yl)acetate (14) to provide the corresponding acid (15). Coupling of 13 and
15 yielded the dipeptide for which the benzyl ester was cleaved by hydrogenation to yield 16. Final
coupling of the corresponding keto-epoxide TFA salt using HATU furnished the MECL-1 inhibitors 9 and
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
Scheme 2.a
TFA
H2N
O
i, ii
H
N
OH
O
OBn
BocHN
O
O
N
O
O
iv, v
Ph
H
13
N
N
N
OH
H
O
O
Ph
iii
OH
OMe
16
N
15
14
R
O
Compound
R
vi
O
O
H
N
O
N
phenyl
CH2-4-methylphenyl
9
10
N
N
H
H
O
Ph
aReagents and conditions: (i) L-phenylalanine benzyl ester, 1-[bis(dimethylamino)methylene]-1H-1,2,3-triazolo[4,5-
o
b]pyridinium 3-oxide hexafluorophosphate (HATU), N-ethyl-N-isopropyl-propan-2-amine, DMF, 0 C -> rt (84%) (ii)
TFA, DCM, 0 oC -> rt (quant.) (iii) methyl 2-(azepan-1-yl)acetate, LiOH-H2O (iv) HATU, N-ethyl-N-isopropyl-propan-2-
amine, DMF, 0 oC -> rt (90% over 2 steps) (v) Pd/C (10%), methanol, H2 (quant.) (vi) (S)-2-Amino-1-((R)-2-methyloxiran-
2-yl)-3-phenylpropan-1-one TFA salt or 24, HATU, N-ethyl-N-isopropyl-propan-2-amine, DMF, 0 oC -> rt
ACS Paragon Plus Environment

Page 7 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
The custom amino acids required for P1 keto-epoxide building blocks involved two different routes. The
cyclobutyl amino acid started with cyclobutylmethanol and in two steps was converted to the
corresponding iodide (Scheme 3, 17). Condensation with diethyl 2-acetamidomalonate provided the
corresponding diester (18) which was hydrolyzed to provide racemic 19. An enzymatic resolution with L-
acylase provided the S-enantiomer of the amino acid which was directly BOC protected to yield the amino
acid 20.
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 3.a
iv
i, ii
iii
COOEt
COOEt
OH
I
NHAc
18
17
O
O
v, vi
OH
OH
NHBoc
20
NHAc
19
aReagents and conditions: (i) methanesulfonyl chloride, cyclobutylmethanol, triethylamine, DCM, 0 oC -> rt (ii) sodium
iodide, methanesulfonate, acetone (iii) potassium tert-butoxide, diethyl 2-acetamidomalonate, DMF (iv) ethanol, NaOH
(34% over 4 steps) (v) L-acylase, water, 37 oC (vi) BOC2O, NaOH (47% over 2 steps)
Scheme 4.a
O
ii
O
iii
i
OH
NHBoc
23
O
OTf
21
OMe
NHBoc
22
aReagents and conditions: (i) trifluoromethanesulfonic anhydride, sodium carbonate, DCM, -20 oC -> rt (73%) (ii) methyl
(R)-2-((tert-butoxycarbonyl)amino)-3-iodopropanoate, Zn dust, trimethylsilyl chloride, [1,1′-
Bis(diphenylphosphino)ferrocene]dichloropalladium(II), DMF, 0 oC -> 50 oC (72%) (iii) LiOH, water, methanol (95%).
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 49
1
2
3
4
5
6
7
8
9
An alternate route towards the cyclopentenyl amino acid was also employed (Scheme 4). Cyclopentanone
was treated with triflic anhydride and sodium carbonate to provide 21. The crude vinyl triflate was carried
directly into a Negishi coupling with di-protected alkyl iodide. The fully protected amino acid 22 was then
hydrolyzed to provide 23.
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
Synthesis of the keto-epoxides involved a 4-step sequence starting from the requisite BOC protected
amino acids (Scheme 5). Weinreb amide formation followed by addition of isopropenylmagnesium
bromide provided the enone in high yield. Addition of bleach resulted in rapid conversion to the epoxide.
Removal of the BOC group yielded the labile amine TFA salt which was carried forward directly without
further purification.
Scheme 5.a
O
O
O
R
O
i
iii
ii
O
R
R
OH
R
N
O
Me
NHBoc
NHBoc
BocHN
NHBoc
R
Compound
R
TFA
O
CH2-4-methylphenyl
cyclobutyl
cyclopentenyl
24
25
26
iv
H2N
O
aReagents and conditions: (i) N,O-Dimethylhydroxylamine hydrochloride, ethylchloroformate, N-methylmorpholine,
THF, DCM, 0 oC -> rt (ii) isopropenylmagnesium bromide, THF, 0 oC (iii) sodium hypochlorite, DMF, -20 oC -> 0 oC (iv)
TFA, DCM.
The synthesis of 11 and KZR-616 (12) employed the same P3/N-cap building block (Scheme 6).
Morpholinoacetic acid was first coupled to benzylalanine to provide the protected building block 28 which
was then hydrogenated to yield the acid 29.
ACS Paragon Plus Environment

Page 9 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Scheme 7 outlines the remaining construction of 11 and KZR-616 (12). Condensation of anisaldehyde and
glycine via an Erlenmeyer condensation provided the P2 amino acid (30) as a racemic mixture with high
diastereoselectivity which was then converted to the racemic methyl ester (31). Isolation of the desired
enantiomer by chiral HPLC separation followed by BOC protection provided 33. A protecting group switch
to benzyl followed by BOC protection furnished the amine intermediate. The synthesis of 11 commenced
with commercially available (S)-2-((tert-butoxycarbonyl)amino)-3-(4-methoxyphenyl)propanoic acid and
followed the same route to completion as KZR-616 (12). Coupling of the P2 building block with 29 using
HATU furnished the benzyl ester which was then deprotected by hydrogenation. The acid was coupled
with the P1 keto-epoxide building block 26 to provide the final compounds.
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 6.a
O
O
i
ii
O
O
O
O
N
N
N
N
COOBn
N
COOH
OH
H
H
29
28
aReagents and conditions: (i) benzyl L-alaninate, 4-(4,6-dimethoxy-1,3,5-triazin-2-yl)-4-methyl-morpholinium chloride
(DMTMM), N-methyl morpholine, DMF, DCM, (50%) (ii) hydrogen, Pd/C (10%), THF, (86%)
Design and Evaluation. Our design of subunit selective inhibitors utilized a human homology model
evolved of the murine crystal structure of the ONX 0914-proteasome isoform complexes.17 This model
was developed to elucidate per subunit the optimal recognition of non-primed selectivity pockets (S1, S2,
S3) by the amino acid sidechains/features (P1, P2, and P3, respectively) of epoxyketone peptidomimetic
18
inhibitors (Figure 1a). Peptide epoxyketones are well-established, exquisitely chemo-selective and
irreversible covalent inhibitors of N-terminal threonine proteases, a class of proteases to which the
proteasomal subunits are nearly exclusive.19 We therefore chose to keep compound designs consistent
to this chemical class and thus pre-engineer selectivity against other hydrolases.20
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 49
1
2
3
4
Scheme 7.a
5
6
7
8
OMe
iii, iv
OMe
OMe
ii
O
i
HO
HO
HO
H
9
MeOOC
NH2
30-(rac)
NH2
31-(rac)
MeO
MeOOC
NHBoc
HOOC
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
33
O
N
O
O
MeO
ix
H
N
viii
v, vi, vii
N
OBn
R
H
O
R
H2N
COOBn
OMe
O
O
O
O
O
O
x
H
H
O
N
N
N
N
N
OH
N
N
H
H
H
O
O
O
R
R
OMe
OMe
Compound
11
KZR-616 (12)
R
H
OH
aReagents and conditions: (i)glycine, potassium hydroxide (23%) (ii) thionyl chloride, methanol, 0 C -> rt (59%) (iii)
chiral separation (45%) (iv) tert-butoxycarbonyl anhydride, THF (quant.) (v) lithium hydroxide, methanol, THF (65%)
(vi) benzyl bromide, cesium carbonate, DMF, 0 oC -> rt (66%) (vii) TFA, DCM, 0 oC -> rt (viii) (2-morpholinoacetyl)-L-
alanine (29), HATU, N-ethyl-N-isopropyl-propan-2-amine (DIEA), DMF, 0 oC -> rt (54% over 2 steps) (ix) hydrogen, Pd/C
(10%), THF (x) 26, 1-[bis(dimethylamino)methylene]-1H-1,2,3-triazolo[4,5-b]pyridinium 3-oxide hexafluorophosphate
(HATU), N-ethyl-N-isopropyl-propan-2-amine, DMF.
o
We’ve recently reported using this homology model to design KZR-504, a highly selective peptide
epoxyketone inhibitor of LMP2, which was also employed in this study21. At the outset of our work, PR-
924 was a known, selective inhibitor of human LMP7 (Table 1)22. Our model and recent co-crystal
structures in humanized yeast proteasome23 suggested selectivity for LMP7 arose from interactions
between PR-924’s indene terminus and the LMP7 S3 pocket formed at neighboring β6. In contrast to ONX
ACS Paragon Plus Environment

Page 11 of 49
Journal of Medicinal Chemistry
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 2. Modeling of selective inhibitors within the human LMP7/β6 binding site.(a) ONX 0914 and PR-924 registered to
selectivity pockets. (b) Different binding modes are observed for PR-924 (orange) and ONX 0914 (blue) (van der Waals surfaces).
(c) PR-924 (orange) and the P3 epimer, 3 (green). (d) PR-924 modeled with Ala28 and Glu131 residues highlighted (yellow). In
human β5 (magenta) and mouse (not shown for clarity), an H-bonding interaction between Glu131 and Ser28 is observed. (e)
Cutaway of the S3 pocket with the cyclohexyl P3 analogue, 5, and the trans cyclohexanol, 6, superimposed.
0914 binding, a change in stereochemistry at P3 places PR-924’s terminal carboxamide deep within the
S3 pocket (Figure 2b). The S3 pockets of other constitutive and immunoproteasome subunits are less
accommodating to either bulky or hydrophobic functionality and thus uniformly reflect lower PR-924
affinity. The epimer of PR-924, 3 (Table 1), lacks selectivity for LMP7. Modeling places its methyl indene
tail outside the S3 pocket and thus supports P3 stereochemistry as a driver for selectivity (Figure 2c). The
use of PR-924 as a subunit selective probe in vivo was precluded by suboptimal solubility and
compromised mouse LMP7/ β5 selectivity (Table 1). Replacing the Trp-P2 substituent of PR-924 with
methyltyrosine in 4 provided 2-fold increased solubility and preserved human LMP7/β5 selectivity. Ala28
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 49
1
2
3
4
5
6
7
8
9
of the human LMP7 S3 pocket is replaced with serine in β5 and in murine β5 and LMP7 (Figure 2d). Ser28
constricts the S3 pocket by forming a hydrogen-bond with a conserved Glu131 of the neighboring β6
subunit. Accordingly, reducing the size of the P3 ligand from methyl indene to cyclohexane (5) lessened
human LMP7/β5 selectivity (Table 1, Figure 2e). However, addition of a 4-hydroxyl group to the cyclohexyl
terminus in 6 increased human LMP7/β5 selectivity to rival that of PR-924, this while imparting > 150-fold
increased solubility (Table 1, Figure 2e). Like PR-924, 6 lacked selectivity for mouse LMP7. Reducing the
P1-ligand size to cyclobutane in 7 resulted in significantly decreased β5 potency, presumably through loss
of optimal S1 pocket occupancy. This loss in β5 potency translated into significantly increased mouse
LMP7/β5 selectivity. The subtlety of this P1/mouse-S1 interaction is highlighted with 8 where replacing
the P1 sidechain with cyclopentenyl provided a 7-fold selectivity window for mouse LMP7 over mouse β5.
Based on an overall favorable in vitro profile, 8 was chosen as our LMP7 selective probe. The challenge of
designing selective MECL-1 inhibitors has been highlighted elsewhere24 and likely stems from a near
identical substrate binding channel between MECL-1 and 2.17 A MECL-1 inhibitor screen of our
epoxyketone-peptide library brought to our attention 9 (Table 1). Like immunoproteasome selective
inhibitor ONX 0914, 9 contains a solubilizing amine at its amino-terminus and phenylalanine scaffolds at
P1 and P2. Our modeling suggested increased MECL-1 inhibition of 9 vs. ONX 0914 can be rationalized by
Leu-P3 providing an optimized hydrophobic surface for interaction with Met131 of the S3 pocket
(Figure3a). Modeling ONX 0914 to MECL-1 also revealed incomplete MECL-1 S1/P1 occupancy (Figure
3b). Moreover, the S1 pocket of MECL-1 is larger and more hydrophobic than 2 by consequence of two
residue changes (T52A and D53E). With 10, increased size and reach of the hydrophobic P1 substituent
provided improvements to MECL-1 inhibition and selectivity over 2 (Table 1). Although proteasome
active site subunits bear distinguishing features outside the S1 pocket, it is largely differences within this
pocket which drive selective inhibition of human MECL-1 by 10 (Figure 4). However, designing a cross-
species selective MECL-1 inhibitor was problematic. Going from human to mouse, alanine at position 52
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

Page 13 of 49
Journal of Medicinal Chemistry
1
2
3
4
Table 1. Subunit IC50’s and Solubility Measurements of Peptide Epoxyketones
5
6
7
Subunit-Specific Cell Lysate IC50 (µM)a
MECL-1 β2 LMP7 β5
Solubility
(µg/mL)b
ID
Structure
Human
LMP2
Mouse
LMP2
LMP7
β5
β1
β1
MECL-1
β2
8
9
O
O
H
O
O
N
0.039 ± 0.422 ±
0.902 ± 0.927 ± 0.098 ± 0.401 ±
0.699 ± 0.913 ±
ONX
0914
0.287 ±
0.061
0.328 ±
0.055
N
N
H
N
>12.7
>25
<0.5
H
0.009
0.090
0.230
0.264
0.033
0.095
0.034
0.082
O
O
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
OMe
O
O
H
N
O
2.21 ±
0.744
0.039 ±
0.007
1.67 ±
0.507
0.75 ±
0.05
N
H
N
PR-924
>25.0
N/D
>25.0
>25.0
0.573
1.36
4.78
N/D
N/D
N/D
N/D
N/D
3.96
0.415
N/D
N/D
N/D
N/D
N/D
N/D
N/D
>250
N/D
N/D
H
O
O
NH
O
O
H
O
N
1.85 ±
0.05
N
N
H
H
3
4
5
6
7
8
9
0.038
0.054
N/D
N/D
N/D
3.53
N/D
N/D
N/D
N/D
N/D
0.05
0.047
N/D
N/D
N/D
N/D
N/D
N/D
0.159
N/D
N/D
N/D
N/D
N/D
N/D
0.141
O
O
NH
O
O
H
O
N
0.076 ±
0.001
3.62±
1.30
0.764 ±
0.001
2.34 ±
1.08
1.40 ±
0.20
N
N
H
N/D
H
O
O
OMe
O
O
H
O
N
12.1 ±
3.5
N
H
N
0.029
0.185
N/D
N/D
N/D
0.141
0.118
H
O
O
OMe
O
O
H
O
N
240.0 ±
2.8
0.023 ± 0.782 ±
0.002
0.262 ± 0.507 ±
0.025
N
N
H
>25.0
N/D
>25.0
N/D
>25.0
N/D
H
O
O
0.243
0.009
HO
OMe
O
O
H
O
N
979.1 ±
15.1
N
N
H
0.262
6.63
0.556
6.81
H
O
O
HO
OMe
O
O
H
N
O
491.3 ±
0.7
0.034 ±
0.001
2.67 ±
0.277
1.85 ±
0.169
0.374 ±
0.017
2.65 ±
0.587
N
H
N
>25.0
>250
>25.0
>25.0
H
O
O
HO
OMe
O
O
H
N
O
0.097 ± 0.081 ±
0.025 0.013
N
0.049
0.039
0.415
>25.0
0.056
0.038
nd
N
H
N
H
O
O
0.346 ± 0.486 ±
0.008 0.068
0.071 ± 0.413 ±
0.053
0.378 ±
0.009
0.305 ± 0.573 ±
0.041
1.92 ±
0.214
O
O
10
11
>25.0
36.421
>10.6
>25.0
>25.0
7.0 ± 0.4
H
N
O
0.115
0.102
N
N
N
H
H
O
O
O
O
O
H
N
O
N
N
N
H
0.083 ± 0.926 ±
0.015 0.160
0.110 ± 0.694 ±
0.014 0.066
2730.8 ±
151.6
H
0.223
1.437
1.176
N/D
N/D
N/D
N/D
O
O
OMe
O
O
O
H
O
KZR-
616
(12)
N
N
N
H
N
0.039 ± 0.688 ± 0.131 ±
0.005 0.160 0.013
0.623 ± 0.604 ± 0.057 ± 0.483 ± 0.179 ±
0.040 0.044 0.018 0.092 0.003
7072.1 ±
518.8
H
>25
0.522
0.857
O
O
HO
OMe
aUtilizing MOLT-4 (human T cell leukemia) or A20 (mouse B-cell lymphoma) cellular lysate, a proteasome constitutive/immunoproteasome subunit ELISA (ProCISE) was employed for
quantitative assessment of subunit-specific activity. Data are reported as the mean ± SD ( n ≥ 2).
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 49
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 3. Modeling of ONX 0914 in blue and 10 in yellow. (a) Van der Waals surfaces depicted with a cutaway of the S3 pocket.
Met131 is depicted in magenta. (b) Van der Waals surfaces depicted with a cutaway of the S1 pocket. (c) Human MECL-1. In the
binding site, there is only a single residue difference (A52T) from human to mouse. Ala52 is depicted in magenta.
is replaced by a threonine which creates a smaller, more polar S1 pocket in mouse MECL-1 relative to the
human ortholog (Figure 3c). Relative to human, this difference triggered 4-fold reduced mouse MECL-1
inhibition and precluded advancement of 10 to our disease model. Further efforts to rescue mouse MECL-
1 potency within the 10 chemotype proved challenging.
Human leukemia cells (MOLT-4) expressing both forms of the proteasome were used to determine full
subunit-selective inhibitory concentrations of ONX 0914, 8, 10, and KZR-504 using proteasome
constitutive/immunoproteasome subunit ELISA (ProCISE)22, 25 (Supporting Information, Figure 1). From
these data, we chose individual concentrations to test in peripheral blood mononuclear cells (PBMC),
which express essentially only the immunoproteasome. We found that that immunoproteasome subunit
inhibition in human PBMC reflected MOLT-4 inhibition upon individual and combination compound
exposure (Figure 5a, 5b). This enabled assessment of inflammatory cytokine expression in LPS stimulated
and anti- CD3/CD28 stimulated PBMC upon discrete and combined subunit inhibition (Figure 5a, 5b).
Consistent with previous reports13, 16, 26, ONX 0914 potently reduced cytokine expression in stimulated
PBMC (Figure 5a). By comparison, selective LMP7 inhibition by 8 in LPS stimulated PBMC failed to
significantly reduce TNF and IL-6 and only partially inhibited the p40 subunit of IL-12/23. TNF and IFN-
expression in anti-CD3/CD28 stimulated PBMC was inhibited by 8 but significantly less so than by ONX
ACS Paragon Plus Environment

Page 15 of 49
Journal of Medicinal Chemistry
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. Modeling of 10 within proteasome catalytic binding sites. Van der Waals surfaces depicted with 10 in yellow and, for
comparison, ONX 0914 is in blue. (a) 10 modelled with human MECL-1. (b) Cutaway of the S1 pocket of the human β2 binding site
with 10 and ONX 0914 overlaid. Key residues distinguishing β2 from MECL-1, Thr52 and Asp53, are depicted in magenta. (c)
Cutaway of the S1 pocket of the human LMP7 binding site with 10 and ONX 0914 overlaid. Met45 is depicted in magenta. A steric
clash between Met45 and the P1 substituent of 10 prevents this ligand from binding efficiently to LMP7. (d) Cutaway of the S1
pocket of the human β5 binding site with 10 and ONX 0914 overlaid Met45 is depicted in magenta. Like LMP7, a steric clash
between Met45 and the P1 substituent of 10 prevents this ligand from binding efficiently to 5. (e) Cutaway of the S1 pocket of
the human β5 binding site with 10 and ONX 0914 overlaid. Residues different from MECL-1 are depicted in magenta: G45L and
E53Q; C31F is not shown for clarity. The LMP2 pocket is shallower than MECL-1. (f) Cutaway of the S1 pocket of the human β1
binding site with 10 and ONX 0914 overlaid. A key residue difference from MECL-1 is a glycine to arginine switch at position 45
(shown in magenta) making the β1 pocket significantly shallower than MECL-1.
0914 treatment. Neither KZR-504 nor 10 significantly reduced cytokine expression in LPS or anti-
CD3/CD28 stimulated cells. When either KZR-504 or 10 was combined with 8, the cytokine inhibition
profile matched that of ONX 0914 (Figure 5b). However, the combination of KZR-504 and 10 did not result
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 49
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 5. Effect of LMP7, LMP2, and MECL-1 selective subunit inhibitors alone and in combination. Active site occupancy was
measured using ProCISE and activity (probe occupancy) for LMP7, LMP2, and MECL-1 subunits was normalized to DMSO-treated
controls. PBMC were treated with either 250 nM ONX 0914, 125 nM 8, 500 nM KZR-504, 250 nM 10 alone (a), or in combination
(b) for 1 h and then stimulated with 1µg/ml LPS or antibodies against CD3 and CD28 for 24 h. Supernatants were analyzed for
TNFα, IL-6, the p40 subunit of IL-12/23, and IFNγ via electrochemiluminescent ELISA (Meso Scale Discovery). Data are presented
as mean relative level ± SD (n = 6) compared to DMSO-treated controls. * = P<0.05; ** = P<0.01; *** = P<0.001 vs. DMSO controls
by Friedman test followed by Dunn’s post-hoc comparison of absolute cytokine levels. (c) BALB/c mice were co-administered
(i.v.) 8 at 20 mg/kg and KZR-504 at 5 mg/kg and in combination. Kidney and splenocyte (erythrocyte-depleted) samples were
taken 1 h after dosing and the activity of LMP7, LMP2, MECL-1 (splenocytes) and β5 (kidney) was measured by ProCISE. Data
were normalized to the average activity of vehicle-treated animals and are presented as the average relative activity + SD (n = 3).
BALB/c mice received 1.75 mg of a cocktail of 5 antibodies against type II collagen on day 0 followed by 25 μg of LPS on day 3.
On day 4, when disease symptoms were present in all mice, animals were randomized into 4 groups and were treated i.v. with
vehicle, 8 at 20 mg/kg, KZR-504 at 5 mg/kg or the combination of 8 at 20 mg/kg and KZR-504 at 5 mg/kg. Dosing was repeated
on days 6, 8, 11, 13, and 15 and clinical scores (0 – 4/paw; n = 7/group) were followed until day 18. Data, presented as the mean
ACS Paragon Plus Environment

Page 17 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
clinical score + SEM, are from one experiment of two performed with similar results. **** = P<0.0001 by two-way ANOVA
followed by Bonferroni post-hoc comparison at the end of study.
6
7
8
9
in significant inhibition of cytokine secretion under either stimulation condition. These data indicate that
LMP7 inhibition is necessary but not sufficient for the full cytokine inhibition profile of ONX 0914.
Inhibition of all 3 subunits of the immunoproteasome, achieved with the combination of 8, KZR-504, and
10, resulted in the greatest effect on cytokine secretion, though a 30% reduction in cell viability was noted
(data not shown).
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 mediates anti-inflammatory responses in mouse models of rheumatoid arthritis at doses
resulting in inhibition of multiple immunoproteasome subunits13. Peptide epoxyketones show rapid
pharmacokinetic clearance across species (typical half-life (t1/2) < 15 min),27 yet like other covalent
inhibitors, their pharmacodynamic effects outlast compound pharmacokinetics and reflect turn-over of
covalently inhibited enzyme in pervaded tissues28. Pharmacokinetics of compounds 8 and KZR-504 upon
intravenous administration to BALB/c mice are also comparably rapid (t1/2 < 7 min, clearance > 73
mL/min/kg, Supporting Information, Table 1). To test the pharmacodynamic effect of individual and
combined subunit inhibition in vivo, 8, KZR-504, or their combination were administered to BALB/c mice,
and subunit inhibition profiles in tissues (spleen and kidney) were evaluated at 1 hour, a time point easily
capturing exposure and clearance for both compounds. Subunit selective inhibitory doses were
established for 8 and KZR-504 at 5 and 20 mg/kg, respectively. Like in vitro results in human PBMC, subunit
inhibition profiles in combination reflected the individual compound administration effects (Figure 5c).
Next, subunit selective inhibitory doses were used to explore the therapeutic effect of individual and
combined subunit inhibition in mice with active collagen antibody induced arthritis. Single agent
treatment of diseased mice with KZR-504 and 8 had a negligible impact on disease progression, indicating
that individual subunit inhibition is insufficient to inhibit inflammation in vivo. In contrast, the
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 49
1
2
3
4
5
6
7
8
9
combination treatment with KZR-504 and 8 resulted in significant reduction of disease burden to levels
similar to that of ONX 0914 treatment alone. Based on body weights and cage-side observations, we did
not detect a change in tolerability between any treatment group (data not shown). Although ONX 0914
possesses an efficacious immunoproteasome subunit inhibition repertoire, its suboptimal solubility
precluded further clinical development (Table 1). Upon replacement of the planar, hydrophobic P1
phenylalanine sidechain with saturated or semi-saturated ring systems, unformulated solubility
improvements were observed amongst the cyclohexanol bearing compounds (6 vs 7 and 8, Table 1).
Amongst these compounds, the P1 cyclopentene of KZR 329 offered improved solubility and optimal LMP7
selectivity. Incorporation of this feature into the ONX 0914 scaffold resulted in a desired subunit inhibition
profile with vastly improved solubility (11). Toward furthering solubility, our modeling indicated that an
R-hydroxyl group substitution at the beta position of the P2 methyltyrosine sidechain would be well
tolerated and result in hydrogen bonding with Ser21. Merging of these features provided KZR-616 (12,
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 6. Modeling of KZR-616 in the LMP7/β6 binding site. Similar binding modes are observed for ONX 0914 (blue) and KZR-616
(purple); van der Waals surfaces shown. A productive H-bonding interaction takes place between KZR-616 and backbone carbonyl
of Ser21 (yellow). An increase in binding affinity and solubility is realized with the addition of the P2 alcohol functionality.
ACS Paragon Plus Environment

Page 19 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Figure 6), a compound with 2.5-fold improved solubility relative to 11. Relative to ONX 0914, KZR-616
maintained LMP7 and LMP2 selective inhibition (Figure 7a), a comparable cytokine inhibition profile
(Figure 7a, 7b), and similar rapid pharmacokinetics (Supporting Information, Table 1). Moreover, efficacy
of KZR-616 was comparable to ONX 0914 in the anti-collagen antibody induced arthritis (CAIA) model at
half the dosage (Fig 7b) while subcutaneous administration gave comparable results at similar dosages
(data not shown). As expected of compounds from the epoxyketone chemical class, KZR-616 showed no
inhibition at 10 M against a broad selectivity panel of 20 serine, metallo-, cysteine and aspartyl proteases
and 11 hydrolases (Supporting Information, Table 2 and Screening Panel Methods). Based in part on these
results, KZR-616 was selected as a clinical candidate for treatment of autoimmune disease.
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
Conclusion
This is the first study to describe and employ a library of immunoproteasome subunit-specific inhibitors
to probe the role of individual subunits in regulation of cytokine secretion and inflammation. These results
demonstrate that while necessary, inhibition of LMP7 alone was insufficient to inhibit cytokine expression
in activated immune effector cells or prevent disease progression in a mouse model of autoimmunity.
Thus, previously published effects of ONX 0914 likely reflect the combined inhibition of multiple
immunoproteasome subunits. To demonstrate this, we see that a combination of LMP7 and MECL-1
and/or LMP2 inhibitors offered an optimal cytokine inhibition profile in vitro, which, in the case of dual
LMP7 and LMP2 inhibition, was predictive of efficacy in a mouse model of inflammatory arthritis. In PBMC
exposed to fully inhibitory concentrations of all 3 subunit-selective inhibitors, greater cytokine inhibition
was achieved relative to ONX 0914 but was also associated with a loss of viability in the cells (data not
shown). It is noteworthy that dual inhibitors KZR-616 and ONX 0914 achieved activity in the mouse
arthritis model via complete LMP7 inhibition and ~40% LMP2 inhibition, suggesting that only partial
inhibition of a secondary immunoproteasome subunit is required for a multi-cytokine inhibitory effect. In
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 49
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 7. Effect of KZR-616 (a) Active site ELISA analysis of MOLT-4 cells following proteasome inhibition with KZR-616 and
corresponding cytokine inhibition in stimulated PBMC at selective concentration (250 nM). (b) top panel: BALB/c mice were
administered (i.v.) KZR-616 at 5 mg/kg. Kidney and splenocyte (erythrocyte-depleted) samples were taken 1 h after dosing and
the activity of LMP7, LMP2, MECL-1 (splenocytes) and β5 (kidney) was measured by ProCISE. Data were normalized to the average
activity of vehicle-treated animals and are presented as the average relative activity + SD (n = 3). Lower panel: BALB/c mice
received 1.75 mg of a cocktail of 5 antibodies against type II collagen on day 0 followed by 25 μg of LPS on day 3. On day 4, when
disease symptoms were present in all mice, animals were randomized into 5 groups and were treated i.v. with vehicle, KZR-616,
or ONX 0914 at dosages shown. Dosing was repeated on days 6, 8, 11, and 13 and clinical scores (0 – 4/paw; n = 7/group) were
ACS Paragon Plus Environment

Page 21 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
followed until day 15. Data, presented as the mean clinical score + SEM, are from one experiment of two performed with similar
results. *=P<0.0133, ***= P<0.001, **** = P<0.0001 by two-way ANOVA followed by Bonferroni post-hoc comparison at the end
of study.
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
addition to providing further clarity for the necessary immunoproteasome subunit inhibition profile for
treatment of autoimmune disorders, these findings and discoveries toward improved physicochemical
properties have been applied to our clinical candidate, KZR-616, now in clinical trials for treatment of
rheumatic diseases.29
Experimental Section
General procedures
Cells
MOLT-4 (human T-cell leukemia) and A20 (mouse B-cell lymphoma) cell lines (ATCC, Manassas, VA) were
cultured in RPMI 1640 media (with 300 mg/L L-glutamine) supplemented with 10% heat-inactivated fetal
bovine serum (FBS), 0.15% (w/v) sodium bicarbonate, 0.45% (w/v) glucose, 10 mM HEPES, 1 mM sodium
pyruvate, 100 IU penicillin, and 100 µg/mL streptomycin (Corning Mediatech, Manassas, VA). A20 media
was also supplemented with 0.05 mM β-mercaptoethanol (Sigma-Aldrich, St. Louis, MO). Cryopreserved
peripheral blood mononuclear cells (PBMCs) from normal healthy volunteers were purchased from
AllCells (Alameda, CA) and cultured in RPMI 1640 media (with 300 mg/L L-glutamine) supplemented with
5% heat-inactivated FBS, 100 IU penicillin, and 100 µg/mL streptomycin (Corning Mediatech, Manassas,
VA). Cellular lysate was generated by 15-30 minute incubation of freeze-thawed cell pellets on ice, in twice
the pellet volume of hypotonic lysis buffer (20 mM Tris pH 8, 5 mM EDTA), with occasional vortexing.
Uncleared lysate was then centrifuged at 4°C at 18,000-18,500 x g for 15 minutes, and the supernatant
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 49
1
2
3
4
5
6
7
8
9
(cleared lysate) collected for protein quantitation via bicinchoninic acid (BCA) assay (Thermo Scientific
Pierce, Rockford, IL) against a bovine serum albumin (BSA) standard curve and subsequent proteasome
activity analysis. Lysate was kept on ice during all subsequent handling or stored between analyses at
-80°C.
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
Proteasome active site ELISA (ProCISE)22
An
enzyme-linked
immunosorbent
assay
(ELISA)-based
technique,
the
proteasome
constitutive/immunoproteasome subunit ELISA (ProCISE) assay, was utilized for quantitative assessment
of subunit-specific activity as previously described [1]. For cell lysate analysis, test compounds were
serially diluted in DMSO at 100X concentration, then diluted to 10X in hypotonic lysis buffer (20 mM Tris
pH 8, 5 mM EDTA). MOLT-4 or A20 cell lysate was treated for 1 hour at 25°C with compound at a final 1X
concentration (1 mg/mL final lysate concentration). For MOLT-4 whole (live) cell analysis, compounds
were serially diluted in DMSO at 400X concentration, then diluted to 5X in cell growth media. Compounds
were added to live MOLT-4 cells suspended at ~2 million cells/mL and incubated for 1 hour at 37°C, 5%
CO2 at a final 1X concentration. Whole cells were then washed 3 times in phosphate buffered saline (PBS),
frozen at -80°C, thawed, and processed to lysate. For PBMC whole (live) cell analysis, compounds were
diluted in DMSO at 2000X concentration, then diluted to 2X in cell growth media. Compounds were added
to live PBMCs suspended at 1 million cells/mL and incubated for 1 hour at 37°C, 5% CO2 at a final 1X
concentration. Whole cells were then washed 4 times in phosphate buffered saline (PBS), frozen at -80°C,
thawed, and processed to lysate. Treated cell lysate (1 mg/mL) was incubated with a biotinylated
proteasome active site binding probe (Nanosyn, Santa Clara, CA; 5 µM final) for 2 hours at 25°C. Following
this, lysate (0.1 mg/mL final) was denatured in guanidine hydrochloride (Sigma-Aldrich, St. Louis, MO; 5.6
M final), and subunits bound to probe were isolated with streptavidin-conjugated sepharose beads (GE
Healthcare Bio-Sciences, Uppsala, Sweden; 1:20 dilution of original stock volume) via 1 hour incubation in
0.65 µm filter plates (EMD Millipore, Billerica, MA) at 25°C with vigorous shaking. Multiple rinses in wash
ACS Paragon Plus Environment

Page 23 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
buffer (1X PBS, 0.1% Tween 20, 1% BSA; Teknova, Hollister, CA) were utilized to remove denaturant, and
individual subunits were probed overnight at 4°C (with shaking) with subunit-specific primary antibodies
(Covance, Denver, PA; Enzo Life Sciences, Farmingdale, NY; Santa Cruz Biotechnology, Santa Cruz, CA).
Following primary antibody removal and washing, samples were incubated for 2 hours with HRP-
conjugated secondary antibodies (Jackson ImmunoResearch, West Grove, PA; Invitrogen, Carlsbad, CA).
Finally, secondary antibody was eliminated and a chemiluminescent substrate (Thermo Scientific Pierce,
Rockford, IL) was used to generate signal associated with HRP binding, which was detected on a plate
reader (Tecan, Männedorf, Switzerland). Luminescent signal was normalized to protein content, then,
percent activity calculated relative to DMSO-treated controls to generate IC50 curves.
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
PBMC stimulation and cytokine analysis
PBMCs (250,000 cells/well) were plated in culture media in 96-well round bottom plates (Corning,
Corning, NY). An equal volume of compounds diluted in media were then added (final DMSO
concentration of 0.1%). After 1 hour incubation at 37˚C, 5% CO2, plates were washed 4 times with culture
media. Following compound washout, PBMCs were stimulated with either lipopolysaccharide (LPS,
Escherichia coli, O111:B4, EMD Millipore, Billerica, MA) at 1 µg/ml or antibodies to CD3 (coated overnight
at 2 µg/mL in PBS then washed out, OKT3, Thermo Fisher Scientific, San Diego, CA) and CD28 (2 µg/mL in
media, BD Biosciences, San Jose, CA) for 24 hours. Supernatants were then collected and analyzed for
TNFα, IL-6, IL-12/23p40, and IFNγ by Meso Scale Discovery electrochemiluminescent immunoassay
detection (MSD).
Mice
BALB/c mice (H-2d) were purchased from Taconic Labs. All experiments were done under protocols
approved by an institutional animal care and use committee.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 49
1
2
3
4
5
6
7
Arthritis model
8
9
Anti-collagen antibody induced arthritis (CAIA) was induced in 7-8 week old female BALB/c mice (kept on
breeder chow) by an intravenous (i.v.) administration of 1.75 mg of a cocktail of 5 antibodies against type
II collagen (Chondrex, Redmond, WA) followed by intraperitoneal challenge with 25 µg LPS on day 3.
Treatment was initiated after clinical signs of arthritis were observed (day 4). Paws were scored for disease
severity on a 0 (no disease) – 4 (maximal swelling) scoring system and summed for individual animal
scores. For ANOVA, Bonferroni post hoc analysis was used to compare treatment groups. All statistical
analyses were performed using GraphPad Prism Software (version 7.01, La Jolla, CA). Statistical
significance was achieved when P was less than 0.05. 8, KZR-504, 10 and ONX 0914 were all formulated in
an aqueous solution of 10% (w/v) sulfobutylether-β-cyclodextrin and 10 mM sodium citrate (pH 3.5) and
administered to mice as a single i.v. bolus. For proteasome activity determination, whole blood (sodium
heparin anti-coagulant) and tissue samples (kidney and spleen) were collected 1 hour after administration
and processed as described below for protein quantitation and ProCISE analysis.
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
Mouse tissue pharmacodynamics
Following collection 1 hour post-compound administration, ~500 µL whole blood samples were washed 3
times in 5 mL PBS, and the resultant blood pellet frozen at -80°C. Kidney was immediately frozen on dry
ice and stored at -80°C. Spleen was passed through a 100 µm cell strainer, underwent red blood cell lysis
(Pharm Lyse, BD Biosciences, San Jose, CA), and was rinsed in PBS before freezing of the splenocyte pellet
at -80°C. Frozen tissue samples were later thawed on ice in preparation for lysate collection. Hypotonic
lysis buffer (20 mM Tris pH 8, 5 mM EDTA; approximately twice the volume of cell pellet/tissue) was added
to each sample. Whole blood and splenocyte pellets were lysed for 15-30 minutes on ice, with occasional
vortexing. A 5 mm stainless steel bead (QIAGEN, Hilden, Germany) was added to each kidney sample in a
ACS Paragon Plus Environment

Page 25 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
2 mL microcentrifuge tube. Solid tissue was bead-homogenized on a TissueLyser (QIAGEN, Hilden,
Germany) for 2 minutes at 20 Hz. All pellet and tissue samples were then centrifuged at 18,000-18,500 x
g at 4°C for 15 minutes. Supernatant (lysate) was collected and kept on ice during subsequent handling.
Tissue lysate was diluted in lysis buffer for protein quantification via BCA assay (Thermo Scientific Pierce,
Rockford, IL) against a BSA standard curve. Following determination of protein concentration, lysate was
diluted to appropriate concentrations for ProCISE probe binding and plate loading: 10 mg/mL (blood), 5
mg/mL (kidney), and 1 mg/mL (spleen). Inhibition of subunit-specific proteasome active sites was
quantified utilizing the ProCISE assay as described above. Luminescent signal was normalized to protein
content and protein-normalized background signal from DMSO-probed control samples was subtracted.
Percent activity was then calculated relative to vehicle-treated animals.
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
Mouse Pharmacokinetics
Pharmacokinetics of 8, KZR-504, KZR-616, and ONX 0914 were evaluated in 7-8 week old female BALB/c
(Charles Rivers Laboratory) mice administered at 2 mg/kg each intravenously in a vehicle of
25%PEG300/WFI. Mice were euthanized (CO2 narcosis) at 2, 5, 10, 20, 30, 60 and 120 minutes post-dose.
Blood was collected via cardiac puncture in sodium heparin containing tubes and processed to plasma
(n=3/time point). Acetonitrile was used to precipitate plasma proteins and drug concentrations were
determined by LC-MS/MS. Noncompartmental analysis was applied to the mean plasma concentrations
at each time point. Nominal times and target doses were used in PK analysis. Reported pharmacodynamic
parameters were determined by fit using Phoenix WinNonlin (Version 6.4, Pharsight Inc., S. Louis, MO).
Generation of homology models
Protein homology models were generated for the active sites of human subunits based on the mouse
crystal structures, PDB 3UNB. The human sequences were aligned to the appropriate mouse structural
subunits, structures were cleaned and parameterized with the AMBER12/EHT force field, and a homology
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 49
1
2
3
4
5
6
7
8
model was built following standard protocols in MOE 2013.08 from Chemical Computing Group
(Molecular Operating Environment, 2013.08; Chemical Computing Group Inc., 1010 Sherbooke St. West,
Suite #910, Montreal, QC, Canada, H3A 2R7, 2015).
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
Generation of ligand binding models
Starting ligand structures were imported as SMILES into MOE 2013.08 using standard protocols for Wash,
Parameterization, and low mode 3D conformer search with the AMBER12/EHT force field in MOE 2013.08.
Ligands were modeled within the active sites using a 2-step process. First they were aligned to crystal
ligand (ONX 0914) conformer position in the appropriate proteasome subsite from the 3UNB structure
using standard MOE process for rigid and flexible alignment. The initial poses were optimized in homology
model binding site using MOE LigX in-site minimization, where protein atoms >10 Å from ligands were
fixed and inert, protein atoms >8 Å were fixed, protein site < 8 Å was subject to 20 kcal/0.5 Å tether, and
the ligand was completely free to move. Figures were created using Pymol v. 1.5.0.4. (The PyMol
Molecular Graphics System, Version 1.5.0.4, Schrodinger, LLC.)
Peptidase/Hydrolase Panel
Results were part of a broad Cerep Panlabs selectivity panel against KZR-616 maleate salt (10 M)
performed in replicate at Eurofins Panlabs Tawain, Ltd. Methods of the 31 assays are detailed detailed in
Supporting Information.
Chemical Synthesis
General Procedures of Chemical Synthesis
All reagents and solvents employed were purchased commercially and used without further purification
unless otherwise indicated. All tested compounds were >95% pure as assayed by analytical HPLC using an
Agilent 1260 Infinity system with an Atlantis T3 150X4.6mm column at a 1.0 mL/min flow rate with a
ACS Paragon Plus Environment

Page 27 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
gradient of 0-95% acetonitrile/water (both solvents containing 0.1% aqueous TFA) for 20 min. H NMR
spectra were recorded on either a JEOL ECX (400 MHz) or a Bruker AVANCE III (400 MHz) spectrometer.
Chemical shifts are provided in ppm (δ) relative to DMSO-d6 or CDCl3 as an internal standard. LC-MS
analysis was performed on an Agilent 6120 Single Quad MS with a 1260 Infinity HPLC and EclipsePlus C18,
2.1X50mm column at a 0.3 mL/min flow rate with a gradient of 0-95% acetonitrile/water (both solvents
containing 0.1% aqueous TFA) for 3 min. Flash column chromatography was carried out using an ISCO
Combiflash Rf-200 system with a 200-360 variable wavelength detector and a gradient of 0-10% methanol
in dichloromethane/ethyl acetate (3:1) unless otherwise noted. Synthesis of peptide epoxyketones in this
work and related art have also been published elsewhere30. PR-924 (IPSI) is published elsewhere31. (S)-2-
Amino-1-((R)-2-methyloxiran-2-yl)-3-phenylpropan-1-one TFA salt is published elsewhere32.
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
N-((S)-1-(((S)-3-(1H-Indol-3-yl)-1-(((S)-1-((R)-2-methyloxiran-2-yl)-1-oxo-3-phenylpropan-2-yl)amino)-1-
oxopropan-2-yl)amino)-1-oxopropan-2-yl)-3-methyl-1H-indene-2-carboxamide (3). The title compound
was prepared in analogy to 4. 1H NMR (400 MHz, CDCl3) δ 7.97 – 7.89 (s, 1H), 7.67 – 7.58 (m, 1H), 7.45 (m,
2H), 7.41 – 7.31 (m, 2H), 7.24 – 7.18 (m, 1H), 7.19 – 7.04 (m, 5H), 6.92 (d, J = 2.4 Hz, 1H), 6.89 – 6.82 (m,
2H), 6.72 (d, J = 7.6 Hz, 1H), 6.31 (d, J = 7.6 Hz, 1H), 6.08 (d, J = 6.8 Hz, 1H), 4.70 (m, 2H), 4.55 (m, 1H), 3.40
(dd, J = 22.2, 2.4 Hz, 1H), 3.35 – 3.26 (m, 2H), 3.26 – 3.17 (m, 1H), 3.07 (dd, J = 14.6, 7.4 Hz, 1H), 2.99 (dd,
J = 13.9, 4.9 Hz, 1H), 2.85 (d, J = 4.9 Hz, 1H), 2.60 (dd, J = 13.9, 8.2 Hz, 1H), 2.45 (t, J = 2.3 Hz, 3H), 1.45 (s,
3H), 1.39 (d, J = 7.0 Hz, 3H); 13C NMR (125 MHz, CDCl3): δ 207.1, 172,3. 170.8, 166.0, 148.2, 145.4, 142.1,
136.1, 135.8, 131.2, 129.2, 128.4, 127.4, 127.3, 126.9, 126.8, 123.8, 123.6, 122.2, 120.9, 119.7, 118.6,
111.3, 109.9, 59.1, 53.6, 52.6, 52.4, 49.0, 38.0, 37.0, 27.6, 18.3, 16.4, 12.3. HRMS (m/z): [M]+ calcd. for
C37H39N4O5, 619.2920; found, 619.2934.
N-((R)-1-(((S)-3-(4-Methoxyphenyl)-1-(((S)-1-((R)-2-methyloxiran-2-yl)-1-oxo-3-phenylpropan-2-
yl)amino)-1-oxopropan-2-yl)amino)-1-oxopropan-2-yl)-3-methyl-1H-indene-2-carboxamide (4). To (S)-
2-((tert-butoxycarbonyl)amino)-3-(4-methoxyphenyl)propanoic acid (10.0 g, 33.9 mmol) in DMF (10 mL)
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 49
1
2
3
4
5
6
7
8
9
at 0 °C was added HOBt (4.81 g, 37.3 mmol) and HBTU ((14.1 g, 37.3 mmol). The mixture was stirred for 5
min to the dissolve solids at which time (S)-2-amino-1-((R)-2-methyloxiran-2-yl)-3-phenylpropan-1-one
TFA salt (10.2 g, 33.9 mmol) and DIEA (17.4 mL, 0.101 mol) was added. The reaction mixture was stirred
at ambient temperature for 30 min then quenched with sodium bicarbonate (sat.), extracted with ethyl
acetate (2×), dried with sodium sulfate, filtered, and concentrated. Purification by column
chromatography (0-60% ethyl acetate/heptane) provided tert-butyl ((S)-3-(4-methoxyphenyl)-1-(((S)-1-
((R)-2-methyloxiran-2-yl)-1-oxo-3-phenylpropan-2-yl)amino)-1-oxopropan-2-yl)carbamate (13.4 g, 82%)
as an colorless amorphous solid. MS (EI) for C27H34N2O6, found 483.3 (MH+).
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 tert-butyl ((S)-3-(4-methoxyphenyl)-1-(((S)-1-((R)-2-methyloxiran-2-yl)-1-oxo-3-phenylpropan-2-
yl)amino)-1-oxopropan-2-yl)carbamate (1.00 g, 2.07 mmol) was added DCM (5 mL) and TFA (5 mL). The
reaction mixture was stirred for 15 min at ambient temperature then concentrated and carried forward
without further purification. (S)-2-amino-3-(4-methoxyphenyl)-N-((S)-1-((R)-2-methyloxiran-2-yl)-1-oxo-
3-phenylpropan-2-yl)propanamide TFA salt was immediately carried forward into the subsequent step
(quant. yield). MS (EI) for C22H26N2O4, found 383.2 (MH+).
To
(S)-2-amino-3-(4-methoxyphenyl)-N-((S)-1-((R)-2-methyloxiran-2-yl)-1-oxo-3-phenylpropan-2-
yl)propanamide TFA salt (2.07 mmol) was added (R)-2-((tert-butoxycarbonyl)amino)propanoic acid (782
mg, 4.14 mmol), HATU (1.82 g, 4.77 mmol), and DMF (7 mL). The mixture was cooled to 0 °C and DIEA
(3.54 mL, 20.7 mmol) was added. The reaction mixture was stirred at ambient temperature for 30 min
then quenched with sodium bicarbonate (sat.), extracted with ethyl acetate (2×), dried with sodium
sulfate, filtered, and concentrated. Purification by column chromatography (0-80% ethyl
acetate/heptane) provided tert-butyl ((S)-1-(((S)-3-(4-methoxyphenyl)-1-(((S)-1-((R)-2-methyloxiran-2-yl)-
1-oxo-3-phenylpropan-2-yl)amino)-1-oxopropan-2-yl)amino)-1-oxopropan-2-yl)carbamate (897 mg, 89%)
as a colorless solid. MS (EI) for C26H37N3O6, found 488.4 (MH+).
ACS Paragon Plus Environment

Page 29 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
To tert-butyl ((S)-1-(((S)-3-(4-methoxyphenyl)-1-(((S)-1-((R)-2-methyloxiran-2-yl)-1-oxo-3-phenylpropan-
2-yl)amino)-1-oxopropan-2-yl)amino)-1-oxopropan-2-yl)carbamate (190 mg, 0.412 mmol) was added
DCM (2 mL) and TFA (2 mL). The reaction mixture was stirred for 15 min at ambient temperature then
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
concentrated
and
crude
(S)-2-((R)-2-aminopropanamido)-3-(4-methoxyphenyl)-N-((S)-1-((R)-2-
methyloxiran-2-yl)-1-oxo-3-phenylpropan-2-yl)propanamide TFA salt was carried forward without further
purification. MS (EI) for C27H31F3N3O7, found 470.3 (MH+).
(S)-2-((R)-2-aminopropanamido)-3-(4-methoxyphenyl)-N-((S)-1-((R)-2-methyloxiran-2-yl)-1-oxo-3-
phenylpropan-2-yl)propanamide TFA salt (2.02 g, 3.67 mmol, 1.00) was dissolved in DMF (10 mL) and
cooled to 0 °C. N-ethyl-N-isopropyl-propan-2-amine (1.50 g, 11.6 mmol, 3.00) was added and the solution
was stirred for 5 minutes. 3-Methyl-1H-indene-2-carboxylic acid (0.741 g, 4.25 mmol, 1.10) was added,
followed by HATU (1.62 g, 4.25 mmol, 1.10) over 2 minutes. The resulting solution was stirred at 0 °C for
10 minutes. Brine was added, and the resulting solution was extracted with ethyl acetate. The organic
phase washed with brine (3×) and was subsequently dried over sodium sulfate, filtered, and the filtrate
was concentrated under vacuum. The residue was purified by column chromatography, (0-100%
heptane/ethyl acetate) to provide 4 (0.13 g, 0.22 mmol, 62% yield). 1H NMR (400 MHz, DMSO-d6) δ 8.42
(d, J = 7.2 Hz, 1H), 8.01 (d, J = 8.7 Hz, 1H), 7.71 (d, J = 7.2 Hz, 1H), 7.57 – 7.44 (m, 2H), 7.40 – 7.13 (m, 4H),
7.07 (d, J = 8.6 Hz, 2H), 6.81 – 6.59 (m, 2H), 4.65 – 4.55 (m, 1H), 4.47 (td, J = 9.3, 4.0 Hz, 1H), 4.35 (m, 1H),
3.66 (m, 2H), 3.62 (s, 3H), 3.22 (d, J = 5.3 Hz, 1H), 3.01 – 2.89 (m, 3H), 2.77 (dd, J = 13.9, 9.2 Hz, 1H), 2.71
– 2.56 (m, 1H), 2.39 (t, J = 2.3 Hz, 3H), 1.34 (s, 3H), 1.09 (d, J = 7.1 Hz, 3H); 13C NMR (125 MHz, CDCl3): δ
208.3, 172.7, 171.9, 165.6, 158.3, 145.8, 145.7, 142.7, 137.7, 133.7, 130.8, 129.9, 129.6, 128.9, 127.3,
127.1 (2), 124.3, 121.0, 113.8, 59.6, 55.4, 53.8, 53.2, 52.1, 49.2, 38.7, 37.2, 36.0, 18.5, 16.8, 12.4. HRMS
(m/z): [M]+ calcd. for C36H40N3O6, 610.2917; found, 610.2929.
N-((R)-1-(((S)-3-(4-methoxyphenyl)-1-(((S)-1-((R)-2-methyloxiran-2-yl)-1-oxo-3-phenylpropan-2-
yl)amino)-1-oxopropan-2-yl)amino)-1-oxopropan-2-yl)cyclohexanecarboxamide
(5).
The
title
ACS Paragon Plus Environment