.
Angewandte
Communications
moiety might be diminished by hydration effects, thereby
considerably decreasing its electrophilicity, and altered the
ketoaldehyde to an a-ketoamide (6, Figure 1A).[14] We
coupled 1 with phenylisocyanide and the resulting a-hydroxy-
amide was oxidized using IBX[15] (see the Supporting
Information). Compound 6 contains an extended planar p-
electron system stretching from the phenyl ring through the
amide bond to the a-keto moiety; it conserves the bivalent
reactivity and reversibility of 2, but with minimal hydration of
both carbonyl units (see the Supporting Information). More-
over, a-ketoamides provide the opportunity to also partially
exploit the primed site of the proteasomal substrate binding
channel by protruding with their C-terminal moiety towards
the S1’ specificity pocket. To date, all applied peptidic CP
modulators exclusively bind to the nonprimed site of the
proteasomal binding channel.[16] Hence, the endoproteolytic
character of the proteasome has originated the conception
that compounds occupying both the nonprimed and primed
substrate binding channel display enhanced selectivity and
inhibitory potential. Intriguingly, the binding strength of 6 was
significantly improved to as low as 70 nm and is only exceeded
by the irreversible inhibitors 3 and 4.[17]
Encouraged by these results, we set out to elucidate the
molecular reasons for the vastly differing activities of the CP
inhibitors and to analyze the binding profile to the Thr1Og at
the atomic level. For systematic comparison the peptidic
scaffold of all compounds was kept identical, thus allowing us
to solely analyze the isolated influence of the respective
electrophilic head moieties. We therefore determined yeast
CP:inhibitor complex structures of all compounds with final
resolutions better than 3.0 ꢀ and final Rfree values below 25%
(Figure 2, Tables ST1 and ST2). All yeast CP crystals were
soaked in a 2 mm solution of the respective inhibitor for 24 h.
Data sets were recorded at the synchrotron facility of the SLS
(Paul-Scherrer-Institut, Villingen, Switzerland) and data
processing as well as structural refinement were performed
as described previously.[18]
Inspection of the respective electron density maps illus-
trates that each ligand occupies at least the most favored
substrate binding channel of the b5 subunit by forming
identical antiparallel b-sheet structures. Interestingly, the
highly reactive boronic acid 4 and the least potent vinyl-
sulfone 5 bind quite specifically to the ChTL substrate binding
channel, hence, contradicting hypotheses that a selective
binding mode is determined by the reactivity of the electro-
phile. In contrast, the carbonyl compounds 1 and 2 are found
in all active sites, whereas the epoxyketone 3 omits b1 and the
ketoamide 6 displays a binding preference for the b5 subunit.
These different specificities came as a surprise, as usually only
the chemical nature of the amino acid backbone of the
inhibitor is thought to generate subunit selectivity. It has been
shown, however, that the blockage of several subunits
generates a higher level of cytotoxicity against cancer cells,
which can be exploited in oncological treatments. In contrast,
a more subunit-specific mode of binding is desirable for the
attenuation of the immunological branch of proteasomal
signaling.
Figure 1. Analysis of head groups by in vitro experiments. A) Electro-
philic warheads were coupled to a Z-Leu-Leu-Leu backbone, which led
to different subuntit selectivities. B) IC50 measurements of the ChTL
activity of the proteasome after addition of a dilution series of the
proteasome inhibitors. The highly reactive epoxyketone 3 and the
boronic acid 4 are the strongest inhibiting compounds, whereas the
ketoamide 6 is the most potent reversible inhibitor, hereby surpassing
reactive aldehydes, which are both of similar, intermediate potencies.
an a-ketoaldehyde (2), an a’,b’-epoxyketone (3), a boronic
acid (4), and a vinylsulfone (5) (Figure 1A). Apart from the
commercially available compound 5, all inhibitors were
synthesized by a combination of standard peptide chemistry
and distinct synthesis routes to prepare the electrophilic
warhead. In the case of 3 and 4, the functional moieties were
produced separately and fused to the peptide scaffold in
a convergent synthesis.[10] The aldehyde compound 1, in
contrast, was constructed on a previously generated leucinol
C-terminus by oxidation using 2-iodoxybenzoic acid (IBX).[11]
Compound 2 was produced from 1 by a Grignard reaction
with iodoform followed by an oxidation of the resulting
b-diiodoalcohol, which was subsequently hydrolyzed to yield
the ketoaldehyde[12] (see the Supporting Information).
After the procurement of all the CP inhibitors, we
performed IC50 measurements of the most affected chymo-
trypsin-like activity (ChTL), which is conferred by the
catalytically active Thr1 nucleophile within the b5 subunit
(Figure 1B).[13] All substances inhibited the ChTL hydrolytic
site with the strongest activities displayed by 3 (15 nm) and 4
(27 nm), whereas the rather unreactive vinylsulfone 5 marked
the lower end of the spectrum with an IC50 value of 780 nm.
Surprisingly, the aldehydes 1 and 2 displayed almost identical
inhibitory strengths of 270 and 244 nm, respectively. Despite
the entropically and enthalpically favorable contribution
from the Schiff base condensation reaction, the bivalency
does not endow 2 with a higher inhibitory potential. Hence,
we wondered whether the reactivity of the terminal aldehyde
We thus analyzed the different binding mechanisms of the
compounds to the CP. As expected, the boronic acid 4 and the
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2014, 53, 1679 –1683