Structure-activity relationship studies of salinosporamide A (NPI-0052), a novel marine derived proteasome inhibitor

Article abstract of DOI:10.1021/jm048995+

Salinosporamide A (1, NPI-0052) is a potent proteasome inhibitor in development for treating cancer. In this study, a series of analogues was assayed for cytotoxicity, proteasome inhibition, and inhibition of NF-κB activation. Marked reductions in potency

Full text of DOI:10.1021/jm048995+

3684
J. Med. Chem. 2005, 48, 3684-3687
with the activity of NF-κB by proteasome inhibition
results in enhanced chemosensitivity and increased
apoptosis in cancer cells.8 Salinosporamide A inhibits
all three catalytic activities of the proteasome,4,5 NF-
κB activation (vide infra), and the growth of human
tumor cell lines.4,5
Structure-Activity Relationship Studies
of Salinosporamide A (NPI-0052), a Novel
Marine Derived Proteasome Inhibitor
Venkat R. Macherla,† Scott S. Mitchell,†
Rama Rao Manam,† Katherine A. Reed,†
Ta-Hsiang Chao,† Benjamin Nicholson,†
Gordafaried Deyanat-Yazdi,† Bao Mai,†
Paul R. Jensen,‡ William F. Fenical,‡
Saskia T. C. Neuteboom,† Kin S. Lam,†
Michael A. Palladino,† and Barbara C. M. Potts*,†
A number of important proteasome inhibitors are of
microbial origin, including lactacystin,9 the precursor
to clasto-lactacystin â-lactone.10 Also known as “omu-
ralide” (2),11 this â-lactone forms a covalent adduct with
the catalytic N-terminal Thr1Oγ residue of the â5-
subunit of the 20S proteasome.12,13 The â-lactone-γ-
lactam bicyclic ring system of omuralide is present in
salinosporamide A; however, each molecule is uniquely
substituted such that 1 bears a methyl group and a
cyclohex-2-enyl moiety at C-3 and C-5, respectively,
while 2 bears a hydrogen and an isopropyl group. In
addition, the chloroethyl group alpha to the lactam
carbonyl of 1 is replaced with a methyl group in 2. While
the â-lactone is recognized as a key pharmacophore of
omuralide, the enhanced potency of salinosporamide A
suggests that certain substituents of the bicyclic ring
system are also critical; thus, their significance was
evaluated in a SAR program utilizing analogues gener-
ated through fermentation or derivatization of 1. The
SAR was based on proteasome mediated activities,
including inhibition of CT-L, T-L, and CA-L activities
of the rabbit 20S proteasome, inhibition of NF-κB
activation in HEK293 cells, and cytotoxicity against the
MM cell line RPMI 8226.
Nereus Pharmaceuticals, Inc., 10480 Wateridge Circle,
San Diego, California 92121, and Center for Marine
Biotechnology and Biomedicine, Scripps Institution of
Oceanography, University of California, San Diego,
La Jolla, California 92093
Received December 10, 2004
Abstract: Salinosporamide A (1, NPI-0052) is a potent pro-
teasome inhibitor in development for treating cancer. In this
study, a series of analogues was assayed for cytotoxicity,
proteasome inhibition, and inhibition of NF-κB activation.
Marked reductions in potency in cell-based assays accompa-
nied replacement of the chloroethyl group with unhalogenated
substituents. Halogen exchange and cyclohexene ring epoxi-
dation were well tolerated, while some stereochemical modi-
fications significantly attenuated activity. These findings
provide insights into structure-activity relationships within
this novel series.
The proteasome, a multicatalytic proteolytic complex
that regulates protein degradation in cells, is receiving
considerable attention as a target for treating cancer.1
This approach was validated by FDA approval of the
proteasome inhibitor bortezomib (Velcade) for the treat-
ment of multiple myeloma (MM).2 Emerging clinical
resistance to bortezomib3 suggests the need for alterna-
tive therapies from structurally unique proteasome
inhibitors such as salinosporamide A (1, NPI-0052), a
product of a marine actinomycete Salinispora tropica
that exhibits nM potency against the 20S proteasome4,5
and overcomes bortezomib-resistance in MM cells.5
The 26S proteasome contains one or two 19S regula-
tory subunit caps and a 20S proteolytic core. Three
distinct catalytic functions within the 20S proteasome
[chymotrypsin-like (CT-L), trypsin-like (T-L), and PGPH
or caspase-like (CA-L)] regulate the proteolysis of many
substrates, including damaged proteins and cell signal-
ing molecules. The 20S proteasome also regulates the
activity of the transcription factor NF-κB.1 NF-κB
promotes cell survival by regulating genes encoding cell-
adhesion molecules, proinflammatory cytokines, and
antiapoptotic proteins.6 In its inactive form, NF-κB
complexes with its inhibitor IκB, but upon stimulation,
IκB is degraded by the proteasome, leading to NF-κB
activation.1 NF-κB is constitutively active in many
malignancies including MM, thereby enhancing MM cell
survival and resistance to cytotoxic agents.7 Interference
Prior to studying the substituents of the bicyclic ring
system, the result of hydrolyzing the â-lactone ring was
evaluated. When aqueous solutions of 1 were monitored
by LC-MS, a species with a molecular ion (m/z 332, [M
+ H]+) consistent with the â-lactone hydrolysis product
3 was observed. This product is analogous to the
hydroxy acid reportedly formed from omuralide in
aqueous solution.10 Uniquely, however, â-lactone hy-
drolysis of 1 was followed by intramolecular nucleophilic
addition to the chloroethyl group, forming cyclic ether
4 (Scheme 1). Compound 4 was independently obtained
by base hydrolysis of 1 (0.5 N LiOH in THF/H2O;
complete conversion in 10 min) or by very slow conver-
sion under acidic conditions. While the instability of 3
precluded isolation in sufficient purity for assay, 4 was
inactive in all assays at the highest concentrations
tested (IC50 > 20 µM). These data suggest that, as with
omuralide, the â-lactone is a critical pharmacophore of
1, but further implicate the chloroethyl group as a
reactive “trigger”.
As noted, the â-lactone-γ-lactam bicyclic ring system
is common to both 1 and 2, yet 1 exhibits superior
potency across key bioassays (Table 1). It was therefore
of interest to modify each substituent (R1 (cyclohex-2-
enyl); R2 (chloroethyl); R3 (methyl), and C-5(OH); For-
mula I) and assess activity. Several naturally occurring
“R2 analogues” were obtained from “S. tropica” fermen-
tation extracts or via directed biosynthesis of related
secondary metabolites, including methyl analogue 5,
ethyl analogue 6 (salinosporamide B), bromoethyl ana-
logue 7, and C-2 epimer 8. In addition, compound 9, in
* Corresponding author. Phone: 858-200-8324. Fax: 858-587-4088.
E-mail: bpotts@nereuspharm.com.
† Nereus Pharmaceuticals, Inc.
‡ Scripps Institution of Oceanography.
10.1021/jm048995+ CCC: $30.25 © 2005 American Chemical Society
Published on Web 05/05/2005

Letters
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 11 3685
Chart 1
corresponding ketone 18 as a mixture of rotamers. The
C-5 epimer of 1 could be prepared by sodium borohy-
dride reduction of the ketone; however, the sensitivity
of the â-lactone ring to hydrolysis precluded the use of
polar protic solvents. A procedure reported for reduction
of keto-omuralide was considered,12 and by modifying
the temperature, the desired 5(R) diastereomer was
obtained. Specifically, treatment with sodium borohy-
dride in 1,2-dimethoxyethane/1% H2O at -78 °C gave
5(R) isomer 19 in 90% ee, whereas reactions at higher
temperatures gave increasing amounts of 1.
The rank order of the active compounds in each assay
was established based on their IC50 values. In terms of
cytotoxicity against RPMI 8226 cells, the following
activities were defined: 1 = 7 = 10 > 15 > 14 > 16 >
8 = 9 > 2 > 13 = 6 = 11 > 5. The rank order of the
active compounds based upon their inhibitory effect of
NF-κB-mediated luciferase reporter gene activity was:
1 = 7 = 10 > 15 > 14 > 16 > 5 g 6 g 11 g 2 > 13 >
12 > 9 g 8. Thus, cytotoxicity was well correlated with
inhibition of NF-κB activation, particularly with respect
to the most potent compounds in the series (1, 7, 10,
15, 14, and 16). The four most potent compounds as
determined against the three protease-like activities of
rabbit 20S proteasomes were 1, 7, 10, and 15, which
were also the most potent compounds in the cell-based
assays. As discussed below, however, several potent
proteasome inhibitors in the series were only weakly
cytotoxic. The results are summarized in Table 1 and
were evaluated according to the functional group modi-
fied (Formula I) to help define the preferred substitution
pattern for these novel proteasome inhibitors.
Scheme 1. Proposed Mechanism of Degradation of 1 in
Aqueous Solution
which the methyl substituent at the C-3 ring junction
(R3) is replaced with an ethyl group, was obtained.
Although these analogues provided preliminary SAR
data, a more systematic study was warranted, and the
analogue series was expanded through derivatization
of the natural products. The presence of the potentially
labile â-lactone ring presented unique challenges given
the necessity to preserve this functionality in the
derivatives.
Modifications to the cyclohexene ring (R1) resulted in
varying degrees of attenuated biological activity. Cy-
clohexyl analogue 14, in which the double bond and one
chiral center are removed, was 3- to 12-fold less potent
than 1. Epoxidation gave two analogues with unique
stereochemistry, including 15, the most potent analogue
in the R1 series. This epoxide was only 2- to 4-fold less
potent than 1 across most assays and comparable with
respect to inhibition of CA-L activity. In contrast,
epoxide 16 was 25- to 40-fold less potent than 1 across
most assays and over 7-fold less potent in its inhibition
of CA-L activity. Halohydrin 17 weakly inhibited CT-L
activity with an IC50 value in the µM range, and no
other activity was observed. The decreased potency of
16 and 17 indicate that steric bulk on this face of the
cyclohexane ring is not well tolerated.
The only analogue that reported on the impact of R3
was 9, in which the ring junction methyl is replaced with
an ethyl group. This simple modification attenuated the
inhibition of CT-L activity by nearly 3 log units, while
cytotoxicity and inhibition of NF-κB activation and T-L
activities decreased by over 2 logs. The weak cytotoxicity
and poor inhibition of NF-κB activation by 9 may stem
from poor proteasome inhibition and an unfavorable
steric effect and/or hydrophobic interaction. With re-
spect to C-5(OH), epimerization (19) resulted in com-
plete loss of activity. Oxidation of 1 to the corresponding
ketone 18 also resulted in a dramatic loss in activity,
with retention of only µM inhibition of the CT-L and
T-L activities and no other activity observed over the
concentration range tested. These results are similar
to those reported for epi- and keto-omuralide, for which
To complement the analogues available through
fermentation, four additional R2 analogues were ob-
tained as follows. Treatment of 1 with sodium iodide in
acetone yielded iodoethyl analogue 10 after slow conver-
sion at room temperature. This derivative allowed for
more complete evaluation of halogen substitution and
served as an excellent substrate for nucleophilic dis-
placement reactions to produce additional analogues.
Reaction of 10 with sodium azide in DMSO gave desired
azido derivative 11. Hydroxyethyl analogue 12 was
obtained after briefly exposing 10 to a sodium hydroxide
solution in acetone, which resulted in a complex mixture
of products from which 12 was isolated in low yield.
Finally, to extend the analogue series characterized by
a hydrocarbon substituent at C-2 (i.e., 5, R2 ) methyl;
6, R2 ) ethyl), propyl analogue 13 was prepared by
addition of 10 to a lithium dimethylcuprate (Gilman
reagent) solution in THF at -78°C.
The impact of modification to the cyclohexene ring
(R1) was of particular interest in light of SAR reports
on omuralide, which indicated that an isopropyl group
at this position was optimal for proteasome inhibition
and that replacement with a phenyl ring resulted in
complete loss of activity.11 Modification of the cyclohex-
ene ring of 1 was first achieved through catalytic
hydrogenation (Pd/C) to produce cyclohexyl analogue 14.
Other derivatizations included mCPBA epoxidation of
1, which gave rise to diastereomers 15 and 16 in a 9:1
ratio. Subsequent epoxide ring opening of 15 with HCl
(5% in CH3CN) gave chlorohydrin 17. To assess the
significance of C-5(OH), the secondary alcohol was
oxidized using Dess-Martin periodinane to obtain the

3686 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 11
Letters
Table 1. Biological Activities (IC50 (nM)) of 1 and Its Analogues and Omuralide (2)a
a IC50 values represent the mean ( standard deviation of three or more experiments except where indicated *, the average of two
experiments is shown. ** IC50 value determined in a single experiment.
very weak or no inhibition of the CT-L activity of the
proteasome were reported.11
icity. Omuralide is also a more potent inhibitor of CT-L
activity than its C-2 epimer.11
Modifications to R2 provided particularly insightful
SAR trends. Substitution of chlorine (1) with bromine
(7) or iodine (10) resulted in equipotent analogues across
all assays. The halogenated analogues were 10- to 80-
fold more potent inhibitors of T-L activity than other
analogues in the R2 series when the 2S stereochemistry
was retained. With respect to inhibition of CT-L activity,
most 2S R2 analogues exhibited single-digit nM IC50
values. The presence of a heteroatom at the 3-position
of the R2 chain (11 and 12) was well tolerated in this
assay. In contrast, substitution with a hydrocarbon at
this position (13; R2 ) propyl) resulted in a 9-fold
reduction in potency. While omuralide analogues re-
portedly follow the trend: propyl > ethyl = butyl >
methyl (omuralide) > H with respect to inhibition of
CT-L activity,11,14 suggesting an optimal hydrocarbon
chain length of three, salinosporamide A analogues did
not follow this pattern. Methyl analogue 5 was more
potent than ethyl and propyl analogues 6 and 13, and
was ∼7-fold more potent than omuralide, which also
contains a methyl group at C-2, demonstrating the
collective impact of the cyclohexyl and ring junction
methyl substituents on inhibition of CT-L activity. In
terms of stereochemistry, C-2 epimer 8 was over 2-log
units less potent against the CT-L and T-L activities of
the proteasome than 1 and exhibited only µM cytotox-
Perhaps most striking with respect to R2 analogues
was the observation that replacement of halogenated
ethyl groups (1, 7, and 10) with methyl (5), ethyl (6),
propyl (13), azidoethyl (11), or hydroxyethyl (12) re-
sulted in a dramatic reduction in potency in cell-based
assays. In fact, a ∼3-log decrease in potency was
observed with respect to cytotoxicity, while a ∼2-log
decrease in potency was observed with respect to
inhibition of NF-κB activation. Several factors may
contribute to the disparity between potent inhibition of
purified proteasomes and weak activity in cell-based
assays. (i) Proteasome-mediated events, such as inhibi-
tion of NF-κB activation and cytotoxicity, may be
dependent upon inhibition of specific proteasomal ac-
tivities. Weakly cytotoxic R2 analogues were less potent
inhibitors of T-L activity (IC50 210 nM (11) to 1.1 µM
(13)) than their halogenated counterparts (IC50 13 to
21 nM). This rationale is challenged, however, by R1
analogue 14, which inhibits T-L activity with an IC50
value of 250 nM but retains nM potency in cell-based
assays. (ii) Analogues of 1 may exhibit differences in
their abilities to permeate the cell membrane and/or
inhibit the proteasome once inside the cell. Experimen-
tal evidence of cell permeability was obtained by expos-
ing RPMI 8226 cells to 1, 2, and 5, lysing the cells, and
measuring inhibition of CT-L activity. Inhibition was

Letters
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 11 3687
observed in each case, indicating that all compounds
cross the cell membrane.15 Interestingly, while 85%
inhibition of CT-L activity was achieved upon exposure
to 1 at a concentration of 5 nM, the same concentrations
of 2 and 5 resulted in only 8% and 31% inhibition,
respectively. Higher levels of inhibition (80% and 93%
for 2; 73% and 87% for 5) could only be achieved by
increasing the concentrations to 1 µM and 10 µM,
respectively, values of similar magnitude to those
required to achieve cytotoxicity. (Increased serum con-
centrations had little effect on the cytotoxicity of com-
pounds 1 and 5 (data not shown; compound 2 not
tested)). (iii) It is tempting to consider the possibility
that the role of the R2 halogen is mechanistic. As noted
previously, the â-lactone ring of 1 is hydrolyzed in
aqueous solutions with subsequent formation of cyclic
ether 4 (Scheme 1). An analogous cyclic ether may exist
in the drug-enzyme complex as a stabilized entity
covalently bound to the catalytic threonine. Given that
â-lactone 2 can be generated from its corresponding
thioester lactacystin (Chart 1),10 it is conceivable that
an ester, such as the one that tethers 2 to the protea-
some, could be similarly cleaved through regeneration
of the â-lactone ring, potentially allowing 2 to be
eliminated from its binding site. The cyclic ether form
of 1 would preclude regeneration of the â-lactone and
the potential for subsequent elimination. Alternative to
cyclic ether formation is the possibility that a second
nucleophile on the enzyme displaces chlorine to form a
two-point covalent drug-enzyme adduct. This would not
be unprecedented, as the X-ray crystal structure of
epoxomicin complexed with the 20S proteasome re-
vealed a two-point addition product.13 Although the
leaving group ability of the halogen provides potential
alkylating power, the observation that both 1 (data not
shown) and 212 irreversibly inhibit the proteasome in
vitro suggests that the â-lactone ring and other struc-
tural features common to 1 and 2 are sufficient for
irreversible inactivation. While the structure of salino-
sporamide A is well optimized, analogues with alterna-
tive functional groups at R2 may provide further insight
into the significance of the substituent at this position.
References
(1) Adams J. Proteasome Inhibitors in Cancer Therapy; Humana
Press: Totowa, NJ, 2004.
(2) Richarson, P. G.; Barlogie, B.; Berenson, J.; Singhal, S.; Jagan-
nath, S.; Irwin, D.; Rajkumar, S. V.; Srkalovic, G.; Alsina, M.;
Alexanian, R.; Seigel, D.; Orlowski, R. Z.; Kuter, D.; Limentani,
S. A.; Lee, S.; Hideshima, T.; Esseltine, D. L.; Kauffman, M.;
Adams, J.; Schenkein, D. P.; Anderson, K. C. A Phase 2 Study
of Bortezomib in Relapsed, Refractory Myeloma. N. Engl. J. Med.
2003, 348, 2609-2617.
(3) Chauhan, D.; Li, G.; Shringarpure, R.; Podar, K.; Ohtake, Y.;
Hideshima, T.; Anderson, K. C. Blockade of Hsp27 Overcomes
Bortezomib/Proteasome Inhibitor PS-341 Resistance in Lym-
phoma Cells. Cancer Res. 2003, 63, 6174-6177.
(4) Feling, R. H.; Buchanan, G. O.; Mincer, T. J.; Kauffman, C. A.;
Jensen, P. R.; Fenical, W. F. Salinosporamide A: A Highly
Cytotoxic Proteasome Inhibitor from a Novel Microbial Source,
a Marine Bacterium of the New Genus Salinospora., Angew.
Chem., Int. Ed. 2003, 42, 355-357.
(5) Chauhan, D.; Catley, L.; Li, G.; Podar, K.; Hideshima, T.;
Mitsiades, C.; Ovaa, H.; Berkers, C.; Munshi, N.; Chao, T.;
Nicholson, B.; Neuteboom, S. T.; Richardson, P.; Palladino, M.;
Anderson, K. C. A Novel Non-Peptide and Orally Active Pro-
teasome Inhibitor Overcomes Bortezomib-Resistance in Multiple
Myeloma Cells. Cancer Cell, submitted.
(6) Wang, C. Y.; Mayo, M. W.; Korneluk, R. G.; Goeddel, D. V.;
Baldwin, A. S., Jr. NF-kappaB antiapoptosis: induction of
TRAF1 and TRAF2 and c-IAP1 and c-IAP2 to suppress caspase-8
activation. Science 1998, 281, 1680-1683.
(7) Mitsiades, N.; Mitsiades, C. S.; Poulaki, V.; Chauhan, D.;
Richardson, P. G.; Hideshima. T.; Munshi, N. C.; Treon, S. P.;
Anderson, K. C. Apoptotic signaling induced by immunomodu-
latory thalidomide analogues in human multiple myeloma
cells: therapeutic implications. Blood 2002, 99, 4525-4530.
(8) Cusack, J. C., Jr.; Liu, R.; Houston, M.; Abendroth, K.; Elliott,
P. J.; Adams, J.; Baldwin, A. S., Jr. Enhanced chemosensitivity
to CPT-11 with proteasome inhibitor PS-341: implications for
systemic nuclear factor-kappaB inhibition. Cancer Res. 2001, 61,
3535-3540.
(9) Omura, S.; Matsuzaki, K.; Fujimoto, T.; Kosuge, K.; Furuya, T.;
Fujita, S.; Nakagawa, A. Structure of Lactacystin, a New
Microbial Metabolite which Induces Differentiation of Neuro-
blastoma Cells. J. Antibiot. 1991, 44, 117-118.
(10) Dick, L. R.; Cruikshank, A. A.; Grenier, L.; Melandri, F. D.;
Nunes, S. L.; Stein, R. L. Mechanistic Studies on the Inactivation
of the Proteasome by Lactacystin. J. Biol. Chem. 1996, 271,
7273-7276.
(11) Corey, E. J.; Li, W. Z. Total Synthesis and Biological Activity of
Lactacystin, Omuralide, and Analogues. Chem. Pharm. Bull.
1999, 47, 1-10.
(12) Fenteany, G.; Standaert, R. F.; Lane, W. S.; Choi, S.; Corey, E.
J.; Schreiber, S. L. Inhibition of Proteasome Activities and
Subunit-Specific Amino-Terminal Threonine Modification by
Lactacystin. Science 1995, 268, 726-731.
(13) Groll, M.; Huber, R. Inhibitors of the Eukaryotic 20S Proteasome
Core Particle: A Structural Approach. Biochim. Biophys. Acta
2004, 1695, 33-44.
(14) Soucy, F.; Grenier, L.; Behnke, M. L.; Destree, A. T.; McCormack,
T. A.; Adams, J.; Plamondon, L. A Novel and Efficient Synthesis
of a Highly Active Analogue of clasto-Lactacystin â-lactone. J.
Am. Chem. Soc. 1999, 121, 9967-9976.
Acknowledgment. We sincerely thank Dr. Samuel
Danishefsky of the Sloan-Kettering Institute for Cancer
Research for helpful discussions.
(15) Omuralide has previously been shown to cross the cell mem-
brane. Dick, L. R.; Cruikshank, A. A.; Destree, A. T.; Grenier,
L.; McCormack, T. A.; Melandri, F. D.; Nunes, S. L.; Palombella,
V. J.; Parent, L. A.; Plamondon, L.; Stein, R. L. Mechanistic
Studies on the Inactivation of the Proteasome by Lactacystin in
Cultured Cells. J. Biol. Chem. 1997, 272, 182-188.
Supporting Information Available: Experimental de-
tails on biological assays, compound preparation, spectral
characterization of compounds 4 and 10-19 (including 1H
NMR spectra), and HPLC purity data for compounds 1 and
4-19 are available free of charge via the Internet at http://
pubs.acs.org.
JM048995+