Enzyme inhibition by hydroamination: Design and mechanism of a hybrid carmaphycin-syringolin enone proteasome inhibitor

Article abstract of DOI:10.1016/j.chembiol.2014.04.010

Hydroamination reactions involving the addition of an amine to an inactivated alkene are entropically prohibited and require strong chemical catalysts. While this synthetic process is efficient at generating substituted amines, there is no equivalent in small molecule-mediated enzyme inhibition. We report an unusual mechanism of proteasome inhibition that involves a hydroamination reaction of alkene derivatives of the epoxyketone natural product carmaphycin. We show that the carmaphycin enone first forms a hemiketal intermediate with the catalytic Thr1 residue of the proteasome before cyclization by an unanticipated intramolecular alkene hydroamination reaction, resulting in a stable six-membered morpholine ring. The carmaphycin enone electrophile, which does not undergo a 1,4-Michael addition as previously observed with vinyl sulfone and α,β-unsaturated amide-based inhibitors, is partially reversible and gives insight into the design of proteasome inhibitors for cancer chemotherapy.

Full text of DOI:10.1016/j.chembiol.2014.04.010

Chemistry & Biology
Article
Enzyme Inhibition by Hydroamination: Design
and Mechanism of a Hybrid Carmaphycin-Syringolin
Enone Proteasome Inhibitor
Daniela B.B. Trivella,1,2,3,8 Alban R. Pereira,1,8 Martin L. Stein,4,8 Yusuke Kasai,1 Tara Byrum,1 Frederick A. Valeriote,5
1,7,
1,7,
Dean J. Tantillo,6 Michael Groll,4, William H. Gerwick,
and Bradley S. Moore
*
*
*
1Center for Marine Biotechnology and Biomedicine, Scripps Institution of Oceanography, University of California at San Diego, La Jolla,
CA 92093-0212, USA
2Institute of Chemistry, University of Campinas, Campinas SP 13083-970, Brazil
3Brazilian Biosciences National Laboratory, National Center for Research in Energy and Materials, Campinas SP 13083-970, Brazil
4
¨
¨
Center for Integrated Protein Science, Department Chemie, Lehrstuhl fu¨ r Biochemie, Technische Universitat Munchen, Garching 85747,
Germany
5Henry Ford Health System, Department of Internal Medicine, Josephine Ford Cancer Center, 440 Burroughs, Detroit, MI 48202, USA
6Department of Chemistry, University of California at Davis, 1 Shields Avenue, Davis, CA 95616, USA
7Skaggs School of Pharmacy and Pharmaceutical Sciences, University of California at San Diego, La Jolla, CA 92093-0736, USA
8Co-first author
*Correspondence: michael.groll@tum.de (M.G.), wgerwick@ucsd.edu (W.H.G.), bsmoore@ucsd.edu (B.S.M.)
http://dx.doi.org/10.1016/j.chembiol.2014.04.010
SUMMARY
been reported. The entropic penalty for a biochemical hydroami-
nation reaction, however, may conceivably be overcome by pre-
Hydroamination reactions involving the addition of organization in an enzyme reactive site. In this work, we report
enzyme inhibition mediated by hydroamination using protea-
some inhibitors derived from natural products. The scaffolds of
an amine to an inactivated alkene are entropically
prohibited and require strong chemical catalysts.
While this synthetic process is efficient at generating
substituted amines, there is no equivalent in small
molecule-mediated enzyme inhibition. We report an
unusual mechanism of proteasome inhibition that
involves a hydroamination reaction of alkene deriva-
tives of the epoxyketone natural product carmaphy-
these small molecules interact tightly with the protein, which
facilitates hydroamination by the enzyme N-terminal amine.
The proteasome functions as the central hub of nonlysosomal
cellular proteolysis where it mediates a number of key processes
such as cell cycle control, cell differentiation, immune response,
amino acid recycling, and apoptosis (Goldberg, 2007; Murata
et al., 2009). These biological processes can thus be manipu-
lated through the addition of small molecules that selectively
target the proteolytically active b subunits of the proteasome
cin. We show that the carmaphycin enone first forms
a hemiketal intermediate with the catalytic Thr1 resi-
due of the proteasome before cyclization by an (Kisselev et al., 2012; Moore et al., 2008; Borissenko and Groll,
2007). Due to the importance of the proteasome to malignant
cells and the immune process, it is considered a biological target
unanticipated intramolecular alkene hydroamination
reaction, resulting in a stable six-membered morpho-
line ring. The carmaphycin enone electrophile, which
does not undergo a 1,4-Michael addition as previ-
ously observed with vinyl sulfone and a,b-unsatu-
rated amide-based inhibitors, is partially reversible
and gives insight into the design of proteasome in-
hibitors for cancer chemotherapy.
of high interest for pharmaceutical development. Two protea-
some inhibitors, the epoxyketone carfilzomib (Kyprolis) and the
peptide boronate bortezomib (Velcade), are now used clinically
as anticancer agents and others are in development.
Several proteasome inhibitors have been reported from natu-
ral and synthetic sources and include both noncovalent and
covalent inhibitors (Kisselev et al., 2012). The covalent protea-
some inhibitors can display reversible or irreversible inhibition
INTRODUCTION
profiles and present, in most cases, a peptidic core and an elec-
trophilic warhead. The peptidic core is responsible for forming a
Hydroamination of unactivated alkenes is a challenging process stable antiparallel b sheet with the enzyme, which in turn posi-
because such reactions are generally not very exothermic and tions the warhead in ideal geometry for covalent attachment of
are entropically disfavored (Beller et al., 2004; Hultzsch, 2005). the Thr1 proteasome catalytic residue. The Thr1 side chain oxy-
In general, these reactions require protonation of the alkene gen (Thr1Og) is the nucleophile responsible for the attack on
p-bond, leading to a carbocation intermediate that is then electrophilic substrates, including the natural peptidic substrate
attacked by the amine nucleophile (Beller et al., 2004). Whereas and several classes of inhibitor electrophiles, thus forming cova-
this process can be promoted by alkali, transition, or rare earth lent adducts (Kisselev et al., 2012). Taking advantage of the
metals, as well as by Lewis or Brønsted acids and bases (Beller inhibitor stability and warhead positioning conferred by the
et al., 2004; Schlummer and Hartwig, 2002; Hultzsch, 2005), no peptidic core of proteasome inhibitors, we used the recently
equivalent in small molecule-mediated enzyme inhibition has discovered natural product carmaphycin (1; Pereira et al.,
782 Chemistry & Biology 21, 782–791, June 19, 2014 ª2014 Elsevier Ltd All rights reserved

Chemistry & Biology
Enzyme Inhibition by Hydroamination
Table 1. Inhibitory Activity of Carmaphycin and Analogs as
Measured in Human Cell Assays and with the Purified Yeast 20S
Proteasome
ChTL (b5)a
TL (b2)b
H-460c
HCT-116d
19.4 0.1
10.7 2.3
727.4 167.1
1
3
4
1.5 0.2
46.2 6.0
112.4 7.0
>50,000
16.4 2.7
1.2 0.1
19.6 2.4
164.5 5.6
1,667.0 82.0
IC50 values (nM) are presented.
aPurified yeast proteasome, Suc-LLVY-Amc is a peptide substrate to
specifically determine ChTL activity.
bPurified yeast proteasome, Ac-LRR-Amc is a peptide substrate to
specifically determine TL activity.
cH-460 human lung cancer cell line.
dHCT-116 human colon cancer cell line.
placed this residue with N,N-dimethylglutamine. Based on alter-
ations to our original total synthesis (Pereira et al., 2012) of 1, we
prepared epoxyketone 3, enones 4 and 5, and ketone 6. These
synthetic analogs along with natural product 1 were used in
this study to explore and compare the relative inhibitory effect
of different ketone derivatives in the carmaphycin series. The
complete synthetic procedures are reported in the Supplemental
Experimental Procedures and Experimental Procedures.
Figure 1. Natural Product Proteasome Inhibitors
Carmaphycin A (1), syringolin A (2), and derivatives 3–6.
2012) and its derivatives (Figure 1) to challenge the Thr1 nucleo-
phile in interacting with enone electrophiles.
Structurally, 1 features a leucine-derived a0,b0-epoxyketone
warhead (the P1 residue) directly connected to a methionine sulf-
oxide (the P2 residue), which in turn is connected to a valine (the
P3 residue) and an alkyl chain terminal tail (Figure 1). a0,b0-epoxy-
ketones, as exemplified in 1, the bacterial natural product epox-
omicin (Groll et al., 2000; Meng et al., 1999), and its recently US
Food and Drug Administration (FDA)-approved derivative carfil-
zomib (Molineaux, 2012) are potent, selective, and irreversible
proteasome inhibitors. Epoxyketone warheads form stable mor-
pholine derivatives with the active site Thr1 residues in the six
proteolytic sites of the 20S proteasome core particle (Groll
et al., 2000; Meng et al., 1999). The warhead carbonyl and
epoxide undergo two successive nucleophilic attacks performed
by Thr1Og and Thr1N, respectively (Groll et al., 2000).
Activity Assays
We first interrogated the chymotrypsin-like (ChTL) inhibitory
activity of the carmaphycin analog 3 and measured a similar
potency in comparison with the natural compound 1 (1.2 nM
versus 1.5 nM with the purified proteasome, respectively), sug-
gesting the functional similarity of their P2 residues (Table 1).
We next analyzed the carmaphycin/syringolin chimera 4, which
also displayed high potency with nanomolar inhibition of the
purified proteasome, albeit with a 100-fold loss in activity
(164 nM). The proteasome trypsin-like activity (TL, b2 subunit)
was also measured, showing preferable inhibitor binding to
ChTL over TL for the three inhibitors tested. Overall, the relative
potencies of ChTL inhibition matched the cell toxicity properties
of these new derivatives, with compound 3 being of highest
potency, whereas 4 displayed decreased activity yet still had
effects in the nanomolar range (Table 1).
Another class of proteasome inhibitor warheads of interest are
a,b-unsaturated systems, such as a,b-unsaturated amides, as
seen in the proteasome inhibitor and plant pathogen virulence
factor syringolin A (Groll et al., 2008; 2), and vinylsulfones (Kisse-
lev et al., 2012). These undergo 1,4-Michael addition with the
Thr1Og nucleophile, forming a one-step irreversible covalent
adduct with Thr1Og.
We further showed the importance of the alkene or epoxide
groups in the warhead of these inhibitors by preparing and
testing saturated ketone 6. Compound 6 was not active in the
enzyme nor in the cell-based assays, even at concentration
as high as 1 mM.
We thus hypothesized that replacing the epoxyketone warhead
in 1 with a complementary a,b-unsaturated carbonyl as in 2
would probe the plasticity of the proteasome and change the
nature of the chemical reactions between the inhibitor and the
enzyme. Herein we report an unusual mechanism of proteasome
inhibition that involves a hydroamination reaction of alkene
derivatives of the carmaphycin class of proteasome inhibitors.
To more fully explore the consequence of replacing the epox-
yketone in 3 with the enone warhead in 4, we separately incu-
bated 1, 3 and 4 with the yeast and human 20S proteasomes
to measure their relative binding properties (Figures S1 and S2
available online). Compounds 1 and 3 showed a clear reversible
mode of binding after 2 hr of incubation with the ligand (Figures
S1A and S1D). However, the reversibility was less evident after
6 hr (Figures S1B and S1E) and after 24 hr was irreversible (Fig-
ures S1C and S1F). This inhibition profile suggests a two-step
mechanism of interaction with the proteasome, much like that
observed for the prototype epoxomicin (Groll et al., 2000;
RESULTS
Synthesis of Carmaphycin Derivatives
Due to the unstable redox properties of the methionine sulfoxide Meng et al., 1999), and is highly anticipated for epoxyketone
at the P2 residue position of natural carmaphycin A (1), we re- warheads. Epoxyketones undergo a fast reversible inhibition
Chemistry & Biology 21, 782–791, June 19, 2014 ª2014 Elsevier Ltd All rights reserved 783

Chemistry & Biology
Enzyme Inhibition by Hydroamination
Figure 2. Crystal Structures of the Yeast 20S Proteasome Binding the Natural Product Carmaphycin A (1) and Derivatives Containing
Epoxyketone and Enone Warheads
(A–H) General view of the ChTL catalytic unit binding 1 (A) and detailed view of 1 (B) and 3 (C) epoxyketone warheads. The enone chimera 4 is shown in (D) and a
detailed view of the enone warhead bound to the b2 (TL) and b5 (ChTL) subunits are presented in (E) and (F), respectively. Ligands are contoured by the 2Fo-Fc
omit maps at 1s (blue net). A hemiketal adduct was trapped at the TL subunit binding compound 4 (E), unequivocally showing a 1,2 addition reaction of the
Thr1Og to the enone carbonyl. In contrast, a cyclic adduct was found at the ChTL subunit (F). Differentiation between six (morpholine) and seven (oxazepine)
membered ring adducts was possible by the use of enone 5 derivative, which contains an extra methyl group attached to the enone beta carbon. The super-
position of morpholine (green) with oxazepine (light gray) modeled adducts is shown in (G) and a detailed view of 5 binding the ChTL subunit is shown in (H). The
Fo-Fc difference map calculated from experimental data after modeling the oxazepine ring is contoured at 3s (G). Positive and negative peaks are presented in
green and red nets, respectively. Extra electron density near the wrong N-Cb bond (red net) and missing density at the extra methyl position for morpholine ring
possibility (green net) were found, indicating the correctness of the 6-membered morpholine ring as the final product. Sequence numbering is based on the
primary sequence alignment of the respective yeast subunit with that of Thermoplasma (Lo¨ we et al., 1995). Figures were generated with the Pymol software
(Schroedinger). Oxygen, nitrogen, and sulfur atoms are shown in red, blue, and yellow, respectively.
step, followed by a second reaction leading to irreversible inhibi- the binding data of enone 4 is similar to that observed for
tion of the proteasome.
epoxyketone inhibitors 1 and 3, showing a first reversible and a
Enone 4 also exhibited this inhibition profile in the ChTL second irreversible reaction step with the proteasome ChTL sub-
subunit, thus suggesting a similar stepwise reaction with the unit, thereby suggesting the likelihood of a 1,2-addition on the
proteasome (Figures S1G–S1I). Compound 4 had a prolonged ketone of compound 4 as the first reversible step.
reversible phase, with reversible profiles at 2 and 6 hr of inhibitor
incubation with the enzyme, indicating a less favored irreversible Crystal Structures
reaction step when compared to epoxyketones. We further To characterize the binding modes of the carmaphycin deriva-
measured the proteasome TL activity, showing that epoxyke- tives, compounds 1, 3, and 4 were crystallized with the yeast
˚
tones 1 and 3 are irreversible, whereas enone 4 is a reversible in- 20S proteasome and diffraction data collected to 2.8 A resolu-
hibitor of this proteasome catalytic unit (Figures S1J–S1L).
Taken together, these results suggest a different mode of pro- Rfree values below 24.3% and root-mean-square deviation
tion (Figures 2 and S3). All three structures were refined to final
ꢀ
˚
teasome binding for the a,b-unsaturated carbonyl system in bond and angle values less than 0.006 A and 1.0 , respectively
compound 4. Although one might anticipate the enone moiety (Table S1).
of 4 to function as a one-step covalent irreversible inhibitor
Using inhibitor concentrations in the millimolar range for crys-
undergoing a 1,4-Michael reaction such as compound 2 (Groll tal soaking, compounds 1, 3, and 4 targeted all three active sites
et al., 2008), the reversible nature of the first step of inhibition of the proteasome through covalent linkages, establishing that
observed for compound 4 precludes the possibility of a 1,4-addi- they are catalytically active in the crystalline state (Figure S3A).
tion, which would lead to an irreversible adduct. Furthermore, The backbones of bound inhibitors were well defined as
784 Chemistry & Biology 21, 782–791, June 19, 2014 ª2014 Elsevier Ltd All rights reserved

Chemistry & Biology
Enzyme Inhibition by Hydroamination
Table 2. Interaction of the Proteasome Catalytic Center with the
Natural Substrate and Inhibitors
Substrate/Inhibitor
Natural substrate
Compound 1
Electrophile
Carbonyl (C)
Carbonyl (C)
b carbon (Cb)
Carbonyl (C)
Reaction
Nucleophile
1,2-addition
1,2-addition
1,4-addition
1,2-addition
Og from Thr1
Og from Thr1
Og from Thr1
Og from Thr1
Compound 2
Compound 4
Reaction types, nucleophiles, and electrophiles involved are described.
In contrast, enone 4 adopts two different ligand states in the
crystal structure that give insight into the observed partial revers-
ibility of 4 (Figures 2D–2F). The electron densities calculated for
the TL subunit in the 2FO-FC omit as well as in the FO-FC differ-
ence maps clearly displayed the attachment of 4 to Thr1Og via
a hemiketal linkage involving the carbonyl group of the ligand
(Figure 2E). The resulting alkoxide is stabilized in the oxyanion
hole, interacting with the main chain of Gly47NH, similar to that
observed in an epoxyketone-derived intermediate recently re-
ported (Huber et al., 2012). The preference for a 1,2-addition
over a 1,4-addition to the a,b-unsaturated system of 4 (Figure 2E)
might be a consequence of warhead accommodation at the
proteasome reactive center (Figure 3 and Table 2). Compound
4 positions the carbonyl of the enone in a position for nucleo-
philic attack by the Thr1Og nucleophile, thus enabling the
1,2-addition reaction. This scenario is different with 2, which
positions the b carbon of its a,b-unsaturated system at this reac-
Figure 3. Detail of the Reactive Center of the Proteasome Catalytic
Subunits as Found in the Crystallographic Structures of 2 and 4 in
Complex with the 20S Proteasome
The Thr1Og is the general nucleophile, but the electrophile can vary depending
upon warhead accommodation at the reactive center. The a,b-unsaturated
amide and carbonyl systems of proteasome inhibitors 2 (pale yellow; Protein
Data Bank ID 2ZCY) and 4 (green) are superposed at the TL (b2) catalytic unit.
The Cb of 2 is in prone position for an irreversible 1,4 addition type nucleophilic
attack by Thr1Og. On the other hand, the carbonyl of 4 and the carbonyl of the
natural peptide substrate (not shown) are positioned to undergo reversible 1,2
addition. Figures were generated with the Pymol software (Schroedinger).
Oxygen, nitrogen, and sulfur atoms are shown in red, blue, and yellow,
respectively.
antiparallel b sheets with contacts formed to the main chain tive site, thus facilitating a 1,4-addition (Groll et al., 2008).
atoms of residues Gly47N, Gly47O, Thr21N, Thr21O, and Structural inspection of the ChTL and caspase-like sites, on
Ala49N as well as the side chain of Asp126 (Figure 2). The P1 the other hand, revealed that enone 4 binds in an alternative
leucine side chain was in proximity to the structurally rearranged mode at these catalytic centers (Figure 2F). Notably, 4 forms a
Met45, as well as Ala20, Val31, and Ala49 in the S1 pocket. The cyclic adduct with the Thr1 residue involving both the ketone
methionine sulfoxide (via 1) and N,N-dimethylglutamine (via 3 and alkene functional groups. Inspection of the diffraction data
˚
and 4) P2 side chains were equally aligned and did not form at 2.8 A resolution and molecular fitting, however, did not un-
any direct interaction with protein residues. The P3 valine residue equivocally differentiate whether the cyclic product involved a
was effectively stabilized in the S3 pocket by three alanine resi- 6-membered morpholine ring or a 7-membered oxazepine ring.
dues (Ala20, Ala22, and Ala27), whereas a hydrophobic cluster Both adducts are structurally distinct and would involve different
comprised of Pro104, Tyr106, Pro127, and Val128 accommo- mechanistic paths. In a first hypothesis, the hemiketal intermedi-
dated the hexanoate tail in the ChTL (b5) and TL (b2) sites. This ate of 4, as observed in the TL subunit (Figure 2E), would
aliphatic group was exposed to solvent in the b1 subunit undergo an unprecedented hydroamination reaction at the
(caspase-like activity), where it adopts a random arrangement alkene residue with Thr1N. Depending on the regiochemistry of
(Figure S3A). the reaction, two outcomes are plausible and would generate
Structural refinement and electron-density map calculations either morpholine or oxazepine adducts.
revealed that the epoxyketone inhibitors 1 and 3 bind to the
Formation of the oxazepine, however, could alternatively be
Thr1 residue in a manner highly similar to that previously achieved by a different reaction mechanism first involving a
observed for epoxomicin (Groll et al., 2000; Huber et al., 2012) 1,4-addition of Thr1N to the enone followed by addition of the
and its derivatives (Figure S3B). Unambiguously, a morpholine Thr1Og to the ketone (Figure S10). In this scenario the first cova-
ring was formed between the catalytic Thr1 residue and the lent adduct would represent the irreversible product derived
epoxyketone warheads (Figures 2A–2C), as a result of a two- from 1,4-addition and the proteasome nucleophile would
step reaction between the protein and the inhibitors. As previ- change from Thr1Og to Thr1N. Considering that Thr1Og is the
ously reported (Huber et al., 2012), the 1,2-addition of Thr1Og general nucleophile reported in all proteasome covalent interac-
to the carbonyl carbon of the a0,b0-epoxyketone first occurs to tions to date, including the innate proteolytic mechanism (re-
form
a hydrolysable hemiketal adduct. Subsequently, the viewed in Kisselev et al., 2012), and that Thr1N is protonated
epoxide group is prearranged for a nucleophilic attack by the under physiological conditions, it is unlikely that Thr1N should
Thr1 free N-terminal amine nitrogen. This reaction occurs at act as the initial nucleophile. Furthermore, our reversibility exper-
the C2 position of the epoxide and results in the final morpholine iments clearly show that 4 undergoes a stepwise reaction with
ring product (Figures 2A–2C), which is stabilized by interactions the proteasome, in which the first step is reversible (Figures
with Lys33Nε and Ser129Og.
S1G–S1I). While this orthogonal mechanism is not consistent
Chemistry & Biology 21, 782–791, June 19, 2014 ª2014 Elsevier Ltd All rights reserved 785

Chemistry & Biology
Enzyme Inhibition by Hydroamination
investigated the regiochemistry of the ChTL site alkene hydroa-
mination of 4 and 5, by calculating the relative energies of both
reaction products and transition state structures for formation
of their carbocation precursors. These calculations indicate
that 4 has an inherent thermodynamic and kinetic preference
for morpholine over oxazepine ring formation (Figure 4). The mor-
pholine product is predicted to be approximately 8 kcal/mol
lower in free energy than the oxazepine product (Figures 4A
and 4B). In addition, proton transfer (the first step in the hydroa-
mination reaction) to the alkene CH2 group is predicted to be
favored over proton transfer to the more substituted carbon by
5 kcal/mol (Figure S9), as expected.
These theoretical calculations are in agreement with the
experimental results, leading to a consistent mechanistic model
in which the enone warhead in 4 undergoes an initial 1,2-addi-
tion, followed by hydroamination of the alkene, which occurs in
a Markovnikov sense, leading to a morpholine ring as the final
product.
DISCUSSION
Figure 4. Quantum Chemical Calculations of the Regiochemistry of
the ChTL Site Alkene Hydroamination of Inhibitors 4 and 5
(A and B) Computed structures (distances are approximate) of models for (A)
the morpholine product and (B) a possible alternative oxazepine product and
their computed relative free energies (kcal/mol).
As with the well-studied epoxyketone pharmacophore first
described with the proteasome inhibitor epoxomicin (Groll
et al., 2000), we show here that the enone functional group
supports a two-step reaction leading to a cyclic adduct
formation. The ketone groups of both electrophilic inhibitors
are first attacked by the proteasome Thr1Og, followed by a sec-
ond attack of the Thr1N on either the epoxy or alkene groups
(C) Model system used to examine proton transfer to carbon a versus
carbon b. The transition state structure for proton transfer to Cb is favored by
5 kcal/mol over that for transfer to Ca.
with our biochemical observations and with previously observed (Figure 5).
proteasome biochemistry, our structural data did not unequivo-
cally rule it out.
Based on the complementary results from our quantum chem-
ical calculations and the crystallographic data obtained with
To address this issue, we designed compound 5 as a struc- compound 5, we demonstrated that a morpholine ring, as
tural probe to help discern between the six- and seven- observed in epoxyketone proteasome inhibitors, is also gener-
membered ring formation scenarios. Compound 5 contains an ated in the case of the enone pharmacophore. The general ligand
additional carbon atom at the g position of the enone. If the binding in the proteasome-inhibitor complexes of epoxyketones
cyclized product was morpholine-based, then an ethyl side 1 and 3 and enones 4 and 5 is identical (Figure S3), and markedly
chain would be expected. On the other hand, an oxazepine- different to that observed in the 2-derived proteasome complex
based adduct would be reflected by two adjacent methyl (Groll et al., 2008; Figure 3). Consequently, the proteasome inhi-
groups. The crystal structure of the proteasome bound to 5, bition mechanism suggested for enones 4 and 5 is more similar
˚
obtained at 2.5 A resolution (Table S1), revealed a cyclized to that verified for its parent scaffold-containing compounds 1
adduct occupying proteasome binding pockets similar to those and 3 than to that for its parent warhead-containing compounds
observed with analogs 1, 3, and 4 (Figure 2H). Importantly, we as in 2 and vinylsulfones (Kisselev et al., 2012; Figures 3 and 5).
clearly saw the telltale ethyl side chain in the 2FO-FC omit map
Contemplating the reaction mechanism of the proteasome
of 5 binding the proteasome (Figure 2H) and, further inspection with inhibitor 4, we considered several scenarios to dissect the
of the FO-FC difference maps was consistent only with a morpho- mechanism of enzyme inhibition by this enone compound. First,
line cyclic product (Figure 2G). These data definitely exclude a the hemiketal intermediate is formed following the classical pro-
1,4-addition mechanism and reveal that the regiochemistry of teasome 1,2-addition mechanism involving the Thr1Og nucleo-
the hydroamination reaction proceeds in a Markovnikov sense phile and the inhibitor carbonyl, which is ideally positioned at
in generating the morpholine ring as final product.
the proteasome reactive center (Figure 3). After the formation
of the hemiketal intermediate, the alkene residue is no longer
activated. Thus, attack by the Thr1 amino group may proceed
Quantum Chemical Calculations
To further evaluate the reactivity of enone 4 with the proteasome, either directly with the isolated olefin (Figure 5C) or indirectly
we modeled transition state structures for 1,2- and 1,4-addition (Figure 5D).
as the first reaction step (Figures S4–S6), in the absence of the
While the aminolysis of epoxides is known to occur in water
surrounding active site, using density functional theory calcula- under mild conditions, the direct hydroamination of an inacti-
tions (M06-2X/6-31+G(d,p); see Experimental Procedures and vated alkene (Figure 5C), although thermodynamically feasible,
Supplemental Experimental Procedures for details). The data generally requires a catalyst (Hultzsch, 2005). The geometric re-
indicate that the 1,2-addition (Figure S5) is favored over the strictions conferred by the hemiketal intermediate, however,
1,4-addition (Figure S6) by approximately 1 kcal/mol. We further should favor hydroamination, which is otherwise prohibited in
786 Chemistry & Biology 21, 782–791, June 19, 2014 ª2014 Elsevier Ltd All rights reserved

Chemistry & Biology
Enzyme Inhibition by Hydroamination
Figure 5. Proposed Reaction Mechanisms
Involving the Proteasome Active Site Resi-
due Thr1 and Inhibitors Containing Different
Reactive Functional Groups
(A–D) Inhibitory reaction mechanisms are shown
for (A) the epoxyketone warhead in epoxomicin
(Huber et al., 2012, Wei et al., 2012) and 1, (B) the
natural a,b-unsaturated-amide system present in
2 (Groll et al., 2008), and (C and D) the enone
warhead in the synthetic derivatives 4 and 5.
epoxide intermediate would then react
with the free amine from Thr1 to complete
the reaction and form the morpholine
adduct. This mechanistic pathway is sup-
ported by quantum chemical calculations
that predict barrierless formation of the
epoxide intermediate (in the gas phase),
and consequently, align the inhibitory
mechanisms of the epoxyketone and
enone functional groups.
Irrespective of the mechanistic route,
the energetic barrier for the final step in
the morpholine ring formation from the
enone is expected to be greater than
that of the epoxyketone, which may
explain the decreased potency (Table 1)
and partial reversibility of derivative 4
(Figure S1). Interestingly, when the func-
tional group adjacent to the ketone is
removed as in acylketone 6, proteasome
inhibition potency and cell activity are
both abolished. This observation further
suggests that the reactive epoxide and
alkene functional groups are key to avoid-
ing rapid hydrolysis of the initially formed
hemiketal adduct. The nature of the
warhead functional group and viability of
the second reaction step to form an irre-
versible final product strongly contributes
to the potency of these agents (Table 1).
solution. Direct hydroamination of the inactivated alkene may be Although derivatives of 4 have not yet been observed in nature,
facilitated by proton transfer from the free Thr1 amine to the C-C recent biosynthetic studies of the epoxyketone proteasome in-
double bound, because the free amine is predicted to be proton- hibitors epoxomicin and eponemycin suggest the biosynthetic
ated. The geometric restrictions conferred by the hemiketal bond intermediacy of enone intermediates in those natural products
and the overall ligand and warhead positioning in the enzyme that may foretell their future discovery (Schorn et al., 2014).
binding pocket further reduce the entropic penalty for amine The modulated reactivity observed with compound 4 is some-
approach to the alkene p-bond. In addition, the Ser129Og side what reminiscent of the potent proteasome inhibitor fluoro-
˚
chain is approximately 3 A from the Thr1 amine nitrogen atom salinosporamide, which also reacts with the proteasome in a
in both cyclized and hemiketal adducts. Consequently, the two-step reaction mechanism involving a fast and reversible
placement of this residue at this proximate position may increase attachment to Thr1 followed by a slow and irreversible fluoride
´
the nucleophilicity of the amine, thereby facilitating amine addi- displacement reaction (Eustaquio and Moore, 2008; Groll et al.,
tion to the alkene (Schlummer and Hartwig, 2002). 2009). A major difference between these two inhibitors involves
An alternative mechanism involving an indirect hydroamina- the nature of the reversibility of the initial proteasome adducts.
tion reaction is also plausible (Figure 5D). Here the hemiketal in- In the case of fluorosalinosporamide, the b-lactone warhead is
termediate may first undergo an intramolecular rearrangement in destroyed upon hydrolysis, whereas in compound 4, the reverse
which the generated alkoxide reacts with the adjacent olefin, the reaction of the hemiketal intermediate would restore the reactive
latter requiring activation by an acidic proton. This newly formed enone warhead.
Chemistry & Biology 21, 782–791, June 19, 2014 ª2014 Elsevier Ltd All rights reserved 787

Chemistry & Biology
Enzyme Inhibition by Hydroamination
Figure 6. Synthetic Scheme Leading to Carmaphycin Analogs 3–6
The FDA approval of carfilzomib for treatment of multiple SIGNIFICANCE
myeloma in 2012 sparks hope for a broader application of pro-
teasome inhibitors as pharmaceutical agents, especially with re- The proteasome is a validated biochemical target for cancer
gard to their unexploited potency for treatment of other cancer chemotherapy and is inhibited by natural products and
subtypes and immunological diseases. Although carfilzomib synthetic molecules, following reversible or irreversible
shows decreased side effects compared to bortezomib, it still reaction mechanisms. The proteasome Thr1 catalytic resi-
suffers from its highly reactive warhead and toxic effects with due actively participates in the inhibitory mechanisms
prolonged use (Arastu-Kapur et al., 2011). Increasing the revers- through its side chain oxygen (Thr1Og) and its terminal
ibility window with the newly established enone mechanism of amine (Thr1N) nucleophiles that attack the inhibitor’s
action may be a promising approach to further improvements warhead electrophiles. We here report the crystal structures
in this inhibitor class.
of the yeast 20S proteasome bound to carmaphycin and its
In summary, our data reveal that the a,b-unsaturated derivatives containing epoxyketone and enone warheads,
carbonyl systems in compounds 4 and 5 bind in an inverted together with cell-based and mechanism-based assays.
orientation with respect to warhead positioning reported for 2, These data, along with quantum chemistry calculations,
thereby allowing for the initial attack on the carbonyl versus reveal previously unknown proteasome inhibitory biochem-
the conjugated alkene (Figures 3 and 5). This inversion of the istry showing that the enone compounds are partially
innate chemical properties of the Michael-system can be reversible and display promising biological properties. The
ascribed to reactive group positioning for nucleophilic attack proposed mechanism for the newly established enone pro-
by Thr1Og (Figure 3). We demonstrated that the remaining teasome inhibitor electrophile involves a two-step reaction,
alkene derived from the enone warhead of compounds 4 or 5 involving a 1,2-addition of Thr1Og to the ketone, followed by
can undergo a second nucleophilic attack by proteasome a hydroamination reaction of the inhibitor’s inactivated
Thr1N, leading to a cyclic morpholine adduct. Although this alkene by Thr1N. This is an example of enzyme inhibition
second reaction is slower, it is ultimately essential for inhibitor by hydroamination, and the suggested mechanism is well
potency. We anticipate that alkene hydroamination could be supported by the data presented. We were able to trap the
exploited for enzyme inhibitor design in situations where a reaction intermediate for the enone inhibitor in crystal struc-
nucleophile and an acid are in proximity and under ideal geo- tures, showing that warhead positioning in the enzyme
metric constrains to the target alkene, allowing for an otherwise active center is essential for defining the nature of chemical
entropically unfavorable reaction to yield a stable enzyme interactions with the Thr1Og nucleophile. In the carmaphy-
adduct.
cin derivatives series, the carbonyl moiety of epoxyketone
788 Chemistry & Biology 21, 782–791, June 19, 2014 ª2014 Elsevier Ltd All rights reserved

Chemistry & Biology
Enzyme Inhibition by Hydroamination
et al., 2010; Tidgewell et al., 2010) with cell viability being determined by
MTT reduction.
and enone warheads is ideally positioned for 1,2-addition of
Thr1Og, resulting in a reversible hemiketal adduct as the
first reaction product. The second reaction step involves
the covalent attachment of Thr1N to the epoxide or to the
alkene, following intramolecular aminolysis or hydroamina-
tion, respectively. Hydroamination is slower than aminolysis
and represents a new chemical reaction for enzyme inhibi-
tion that allows for controlling the irreversibility of protea-
some inhibitor binding.
Reversibility Experiments
Yeast or human 20S proteasomes (10 nM) were incubated with 1 mM of inhib-
itor 4 or 0.01 mM of 1 or 3 for 2, 6, or 24 hr (yeast) or 1 hr (human) in Tris 100 mM
pH 7.0. The proteolytic activity was measured with fluorescence, using ChTL
(Suc-LLVY) or TL (Ac-LRR) 7-amino-4-methylcoumarin chromophoric sub-
strates after 10-fold dilution of the protein and inhibitor. A control group main-
tained at the same experimental conditions but with constant initial inhibitor
concentration ([4] = 1 mM or [1] or [3] = 0.01 mM) was performed for comparison.
Protein activity in groups maintained at the same experimental condition,
however in the absence of the inhibitor (DMSO control) was monitored to
guarantee protein integrity during the different time points (Figures S1G–S1I
in black). ChTL activity was stable up to 24 hr. However, TL catalytic activity
was unstable in the incubation buffer after the first 4 hr; therefore, only data
derived within the first 2 hr of incubation are reported for this latter enzyme
activity.
EXPERIMENTAL PROCEDURES
Chemicals
All chemicals used in the present work were of reagent grade quality.
Crystal Structures
The 20S proteasome was purified from Saccharomyces cerevisiae as
described previously (Gallastegui and Groll, 2012; Groll et al., 1997; Groll
and Huber, 2005). Crystals were grown in hanging drops at 24ꢀC, using a pro-
tein solution at 40 mg/ml in 10 mM morpholino-ethane-sulphonic acid (MES)
pH 7.5 and EDTA (1 mM). Drops contained 1 ml of protein and 1 ml of the reser-
voir (30 mM of magnesium acetate, 100 mM of MES [pH 7.2] and 12% of MPD)
solutions (Groll et al., 1997; Groll and Huber, 2005). Crystals were soaked with
compounds 1, 3, 4, 5, or 6 at final concentrations ranging from 2 to 10 mM for
24 hr. Crystals were then transferred to a cryoprotecting buffer (30% 2-methyl-
2,4-pentanediol, 20 mM magnesium acetate, 100 mM Tris [pH 6.9]) before
cooling in liquid nitrogen.
Computational Details
All calculations were carried out using the M06-2X/6-31+G(d,p) method
(Hohenstein et al., 2008; Zhao and Truhlar, 2008) with a continuum treatment
for solvation (the SMD approach [Marenich et al., 2009], using chloroform, i.e.,
a solvent with a dielectric constant in the range of estimates for enzyme active
sites), as implemented in Gaussian 09 (Gaussian). Explicit active site residues
were not included in the calculations, so the preferences described here reflect
inherent reactivity. Conformations of computed structures represent produc-
tive conformations for formation of the morpholine and oxazepine products.
Structural drawings were produced using Ball & Stick (Mu¨ ller et al., 2004).
Computations on proton transfer did not include solvent and were carried
out using the model system shown in Figure 4C in the main text. N–H and
Data collections were carried out at 100 K in a stream of liquid nitrogen gas
(Oxford Cryo Systems). Crystals formed in the P21 space group with cell
ꢀ
˚
˚
˚
dimensions of about a = 135 A, b = 301 A, c = 144 A and b = 113 (Table
˚
Ca/b–H distances were frozen to 1.30 A for the two systems shown in Figure S9
˚
˚
S1). Data to 2.5 A were collected using synchrotron radiation with l = 1.0 A
at the X06SA-beamline in SLS/Villingen/Switzerland. X-ray intensities and
data reduction were evaluated using the XDS program package (Kabsch,
2010). Conventional crystallographic rigid body, positional, and temperature
factor refinements were carried out with PHENIX (Adams et al., 2010) using
the yeast 20S proteasome structure as starting model (Protein Date Bank
accession code 3UN8; Huber et al., 2012). Cycles of maximum likelihood mini-
mization, solvent modeling, and anisotropic correction were conducted in
PHENIX (Adams et al., 2010), using grouped B-factor and noncrystallographic
symmetry refinement. Real space refinement and inhibitor modeling were con-
ducted using Coot (Emsley and Cowtan, 2004). Ligand construction, geometry
file elaboration, and covalent ligand-protein linkage were carried out using
JLigand (CCP4, 1994) and Avogadro software. The last step of refinement
was performed in Refmac5 (Murshudov et al., 1997), using noncrystallo-
graphic symmetry and TLS refinement.
and the remainder of each system was allowed to fully relax. The 5 kcal/mol
energy difference discussed in the main text corresponds to the difference
in electronic energies between these two optimized structures.
Synthesis
Compounds 7, 9, 12, and 15 were synthesized as previously described (Per-
eira et al., 2012) and compounds 3–6 were synthesized according to Figure 6.
Detailed synthetic procedures are reported in the Supplemental Experimental
Procedures.
Optical rotations were measured on a JASCO P-2000 polarimeter. UV and
infrared spectra were recorded on a Beckman DU800 spectrophotometer
and on a Nicolet 100 FT-IR spectrometer, respectively. 1H, 13C, and two-
dimensional nuclear magnetic resonance spectra were collected at a 1H reso-
nance frequency of either 400 MHz (Varian Mercury), 500 MHz (Varian VX500),
or 600 MHz (Bruker Avance III equipped with 1.7 mm and 5 mm TCI cryo-
probes). Chemical shifts were calibrated internally to the residual signal of
the solvent in which the sample was dissolved (CDCl3, dH 7.26, dC 77.0).
High-resolution mass spectra were obtained on a ThermoFinnigan MAT900XL
mass spectrometer or an Agilent Technologies 6530 Accurate-Mass Q-time-
of-flight liquid chromatography/mass spectrometer. High-performance liquid
chromatography was carried out using a dual Waters 515 pump system equip-
ped with a Waters 996 photodiode array detector. Vacuum and flash chro-
matographic separations were performed using type H (10–40 mm, Aldrich)
silica and silica gel 60 (40–63 mm, EMD), respectively. Merck thin-layer chro-
matography (TLC) sheets (silica gel 60 F254) were used for analytical TLC
(aluminum-supported, layer thickness 200 mm) and preparative TLC (glass-
supported, layer thickness 250 mm). All chemical reagents were obtained
from Aldrich in an analytical or higher grade and were used as received unless
stated otherwise. Solvents were acquired as high-performance liquid chroma-
tography grade. All reactions were performed under dry nitrogen using glass-
ware previously oven dried (150ꢀC), unless otherwise specified. Glassware
was allowed to reach room temperature under a flow of inert gas. Likewise,
glass syringes and stainless steel needles, used to handle anhydrous reagents
and solvents, were oven dried, cooled in a desiccator, and flushed with inert
gas prior to use. Anhydrous THF was purchased from Aldrich or distilled
from sodium/benzophenone; CH2Cl2 was distilled from CaH2.
Omit maps were calculated using the program Omit (Bhat, 1988).
Inhibition Measurements
Inhibition assays of the purified 20S proteasome core particle from
S. cerevisiae was conducted as previously reported (Pereira et al., 2012),
with minor modifications. One nanomolar of proteasome was incubated with
different inhibitor concentrations in Tris 25 mM pH 7.5; SDS 0.03%; EDTA
0.5 mM for 15 min at 37ꢀC in 96-well plates, in a final reaction volume of
40 mL. Ten microliters of specific fluorogenic proteasome substrate Suc-
LLVY-Amc (CTL substrate) or Ac-LRR-Amc (TL substrate) at 200 mM were
added, resulting in a final substrate concentration of 40 mM. Samples were
incubated for 15 min at 37ꢀC and then the fluorescence in each well was
measured. Fluorescence was normalized to the control conducted in the
same conditions, however without inhibitor (related to 100% of enzyme activ-
ity), and plotted on a graph of inhibitor concentration versus remaining enzyme
activity. The experimental data were fitted using the logistic four parameters
equation in GraphPad software version 5 (GraphPad Prism).
Cell Assays
Cytotoxicity to H-460 human lung cancer and HCT-116 human colon cancer
cell lines was determined as previously reported (Mevers et al., 2011; Gross
Chemistry & Biology 21, 782–791, June 19, 2014 ª2014 Elsevier Ltd All rights reserved 789

Chemistry & Biology
Enzyme Inhibition by Hydroamination
¨
Groll, M., Ditzel, L., Lowe, J., Stock, D., Bochtler, M., Bartunik, H.D., and
ACCESSION NUMBERS
˚
Huber, R. (1997). Structure of 20S proteasome from yeast at 2.4 A resolution.
The Protein Data Bank accession numbers for the proteasome binding com-
Nature 386, 463–471.
pounds 1, 3, 4, and 5 reported in this paper are 4HRD, 4HRC, 4HNP, and 4TLC.
Groll, M., Kim, K.B., Kairies, N., Huber, R., and Crews, C.M. (2000). Crystal
structure of epoxomicin:20S proteasome reveals a molecular basis for selec-
tivity of a,b-epoxyketone proteasome inhibitors. J. Am. Chem. Soc. 122, 1237–
1238.
SUPPLEMENTAL INFORMATION
Supplemental Information includes Supplemental Experimental Procedures,
30 figures, and 1 table and can be found with this article online at http://dx.
doi.org/10.1016/j.chembiol.2014.04.010.
Groll, M., Schellenberg, B., Bachmann, A.S., Archer, C.R., Huber, R., Powell,
T.K., Lindow, S., Kaiser, M., and Dudler, R. (2008). A plant pathogen virulence
factor inhibits the eukaryotic proteasome by a novel mechanism. Nature 452,
755–758.
AUTHOR CONTRIBUTIONS
Groll, M., McArthur, K.A., Macherla, V.R., Manam, R.R., and Potts, B.C. (2009).
Snapshots of the fluorosalinosporamide/20S complex offer mechanistic in-
sights for fine tuning proteasome inhibition. J. Med. Chem. 52, 5420–5428.
D.B.B.T. conducted the inhibition assays; D.B.B.T. and M.L.S. the mechanistic
assays; D.B.B.T., M.L.S., and M.G. obtained the proteasome crystals,
collected the diffraction data and solved the structures; W.H.G. and A.R.P. de-
signed the compounds and with Y.K. synthesized the compounds; D.J.T.
carried out the in silico experiments; T.B. and F.A.V. performed the cell exper-
iments; B.S.M., W.H.G., A.R.P., and D.B.B.T. planned the experiments; all
authors analyzed and discussed the results; B.S.M., D.B.B.T., and W.H.G. pre-
pared the article with input from all authors.
Gross, H., McPhail, K.L., Goeger, D.E., Valeriote, F.A., and Gerwick, W.H.
(2010). Two cytotoxic stereoisomers of malyngamide C, 8-epi-malyngamide
C and 8-O-acetyl-8-epi-malyngamide C, from the marine cyanobacterium
Lyngbya majuscula. Phytochemistry 71, 1729–1735.
Hohenstein, E.G., Chill, S.T., and Sherrill, C.D. (2008). Assessment of the per-
formance of the M05-2X and M06-2X exchange-correlation functionals for
noncovalent interactions in biomolecules. J. Chem. Theory Comput. 4,
1996–2000.
ACKNOWLEDGMENTS
Huber, E.M., Basler, M., Schwab, R., Heinemeyer, W., Kirk, C.J., Groettrup,
M., and Groll, M. (2012). Immuno- and constitutive proteasome crystal
structures reveal differences in substrate and inhibitor specificity. Cell 148,
727–738.
This research was generously supported by the NIH (CA127622 to B.S.M. and
CA100851 to W.H.G. and F.A.V.), the German-Israeli Foundation for Scientific
Research and Development (GIF 23 1102/2010 to M.G.), and the Sao Paulo
Research Foundation (FAPESP 2011/21358-5 to D.B.B.T.).
Hultzsch, K.C. (2005). Catalytic asymmetric hydroamination of non-activated
olefins. Org. Biomol. Chem. 3, 1819–1824.
Kabsch, W. (2010). Xds. Acta Crystallogr. D Biol. Crystallogr. 66, 125–132.
Received: December 7, 2013
Revised: March 13, 2014
Accepted: April 22, 2014
Published: June 12, 2014
Kisselev, A.F., van der Linden, W.A., and Overkleeft, H.S. (2012). Proteasome
inhibitors: an expanding army attacking a unique target. Chem. Biol. 19,
99–115.
Lo¨ we, J., Stock, D., Jap, B., Zwickl, P., Baumeister, W., and Huber, R. (1995).
REFERENCES
Crystal structure of the 20S proteasome from the archaeon T. acidophilum at
˚
3.4 A resolution. Science 268, 533–539.
´
Adams, P.D., Afonine, P.V., Bunkoczi, G., Chen, V.B., Davis, I.W., Echols, N.,
Marenich, A.V., Cramer, C.J., and Truhlar, D.G. (2009). Performance of SM6,
SM8, and SMD on the SAMPL1 test set for the prediction of small-molecule
solvation free energies. J. Phys. Chem. B 113, 4538–4543.
Headd, J.J., Hung, L.W., Kapral, G.J., Grosse-Kunstleve, R.W., et al. (2010).
PHENIX: a comprehensive Python-based system for macromolecular struc-
ture solution. Acta Crystallogr. D Biol. Crystallogr. 66, 213–221.
Meng, L., Mohan, R., Kwok, B.H., Elofsson, M., Sin, N., and Crews, C.M.
(1999). Epoxomicin, a potent and selective proteasome inhibitor, exhibits
in vivo antiinflammatory activity. Proc. Natl. Acad. Sci. USA 96, 10403–10408.
Arastu-Kapur, S., Anderl, J.L., Kraus, M., Parlati, F., Shenk, K.D., Lee, S.J.,
Muchamuel, T., Bennett, M.K., Driessen, C., Ball, A.J., and Kirk, C.J. (2011).
Nonproteasomal targets of the proteasome inhibitors bortezomib and carfilzo-
mib: a link to clinical adverse events. Clin. Cancer Res. 17, 2734–2743.
Mevers, E., Liu, W.T., Engene, N., Mohimani, H., Byrum, T., Pevzner, P.A.,
Dorrestein, P.C., Spadafora, C., and Gerwick, W.H. (2011). Cytotoxic veragua-
mides, alkynyl bromide-containing cyclic depsipeptides from the marine
cyanobacterium cf. Oscillatoria margaritifera. J. Nat. Prod. 74, 928–936.
Beller, M., Seayad, J., Tillack, A., and Jiao, H. (2004). Catalytic Markovnikov
and anti-Markovnikov functionalization of alkenes and alkynes: recent devel-
opments and trends. Angew. Chem. Int. Ed. Engl. 43, 3368–3398.
Molineaux, S.M. (2012). Molecular pathways: targeting proteasomal protein
Bhat, T. (1988). Calculation of an OMIT map. J. Appl. Cryst. 21, 279–281.
degradation in cancer. Clin. Cancer Res. 18, 15–20.
Borissenko, L., and Groll, M. (2007). 20S proteasome and its inhibitors: crys-
Moore, B.S., Eusta´ quio, A.S., and McGlinchey, R.P. (2008). Advances in and
tallographic knowledge for drug development. Chem. Rev. 107, 687–717.
applications of proteasome inhibitors. Curr. Opin. Chem. Biol. 12, 434–440.
CCP4; Collaborative Computational Project, Number 4 (1994). The CCP4
Mu¨ ller, N., Falk, A., and Gsaller, G. (2004). Ball & Stick V.4.0a12, Molecular
Graphics Application for MacOS Computers. (Linz: Johannes Kepler
University).
suite: programs for protein crystallography. Acta Crystallogr.
D Biol.
Crystallogr. 50, 760–763.
Emsley, P., and Cowtan, K. (2004). Coot: model-building tools for molecular
Murata, S., Yashiroda, H., and Tanaka, K. (2009). Molecular mechanisms of
graphics. Acta Crystallogr. D Biol. Crystallogr. 60, 2126–2132.
proteasome assembly. Nat. Rev. Mol. Cell Biol. 10, 104–115.
Eusta´ quio, A.S., and Moore, B.S. (2008). Mutasynthesis of fluorosalinospora-
mide, a potent and reversible inhibitor of the proteasome. Angew. Chem. Int.
Ed. Engl. 47, 3936–3938.
Murshudov, G.N., Vagin, A.A., and Dodson, E.J. (1997). Refinement of macro-
molecular structures by the maximum-likelihood method. Acta Crystallogr.
D Biol. Crystallogr. 53, 240–255.
Gallastegui, N., and Groll, M. (2012). Analysing properties of proteasome inhib-
itors using kinetic and X-ray crystallographic studies. Methods Mol. Biol. 832,
373–390.
Pereira, A.R., Kale, A.J., Fenley, A.T., Byrum, T., Debonsi, H.M., Gilson, M.K.,
Valeriote, F.A., Moore, B.S., and Gerwick, W.H. (2012). The carmaphycins:
new proteasome inhibitors exhibiting an a,b-epoxyketone warhead from a
marine cyanobacterium. ChemBioChem 13, 810–817.
Goldberg, A.L. (2007). Functions of the proteasome: from protein degradation
and immune surveillance to cancer therapy. Biochem. Soc. Trans. 35, 12–17.
Schlummer, B., and Hartwig, J.F. (2002). Brønsted acid-catalyzed intramolec-
ular hydroamination of protected alkenylamines. Synthesis of pyrrolidines and
piperidines. Org. Lett. 4, 1471–1474.
Groll, M., and Huber, R. (2005). Purification, crystallization, and X-ray analysis
of the yeast 20S proteasome. Methods Enzymol. 398, 329–336.
790 Chemistry & Biology 21, 782–791, June 19, 2014 ª2014 Elsevier Ltd All rights reserved

Chemistry & Biology
Enzyme Inhibition by Hydroamination
Schorn, M., Zettler, J., Noel, J.P., Dorrestein, P.C., Moore, B.S., and
Kaysser, L. (2014). Genetic basis for the biosynthesis of the pharmaceutically
important class of epoxyketone proteasome inhibitors. ACS Chem. Biol. 9,
301–309.
Wei, D., Lei, B., Tang, M., and Zhan, C.G. (2012). Fundamental reaction
pathway and free energy profile for inhibition of proteasome by Epoxomicin.
J. Am. Chem. Soc. 134, 10436–10450.
Tidgewell, K., Engene, N., Byrum, T., Media, J., Doi, T., Valeriote, F.A., and
Zhao, Y., and Truhlar, D.G. (2008). The M06 suite of density functionals for
main group thermochemistry, kinetics, noncovalent interactions, excited
states, and transition elements: Two new functionals and systematic testing
of four M06 functionals and twelve other functionals. Theor. Chem. Acc.
120, 215–241.
Gerwick, W.H. (2010). Evolved diversification of
product pathway: apratoxins F and G, two cytotoxic cyclic depsipeptides
from Palmyra collection of Lyngbya bouillonii. ChemBioChem 11,
1458–1466.
a modular natural
a
Chemistry & Biology 21, 782–791, June 19, 2014 ª2014 Elsevier Ltd All rights reserved 791