Angewandte
Communications
Chemie
compounds were tested for the inhibition
of bacterial growth against a panel of 16
different strains, including major pathogens
such as S. aureus, multiresistant S. aureus
(MRSA), and S. Typhimurium (Table S2).
Interestingly, type III AEBQs FM205
and FM243 as well as type I AEHC FM255
inhibited the growth of Gram-positive
strains, including MRSA, with minimal
inhibitory concentrations (MICs) of
10 mm. Contrary, analogues of these com-
pounds with an acetal group (12a–12h,
Scheme 1) did not show growth inhibition
(up to 300 mm), suggesting that the carbonyl
or reduced hydroxy moieties are essential
for the antibiotic activity. Strikingly,
AEBQs FM233 and FM239 inhibited the
growth of S. Typhimurium LT2 and TA100
and S. enteritidis with MICs of about 30 mm
while the corresponding controls lacking
the epoxide (Figure 1B and Scheme S1)
did not affect viability. Satisfyingly, FM233
displayed only low toxicity (< 15% apop-
totic cells) in apoptosis assays against A549
cells at concentrations up to 150 mm (Fig-
Figure 1. A) Representative aminoepoxycyclohexenone natural products: LL-C10037a (1),
asukamycin A1 (2), epoxyquinomycin C (3), and cetoniacytone A (4) belong to the AEHC
subclass (type I/II). Epoxyquinomycin B (5) and G-7063-2 (6) are members of the AEBQ
subclass (type III).[4,8] B) Structures of diverse type I–III aminoepoxycyclohexenone ABPP
probes and their corresponding epoxide-lacking benzoquinone controls.
ure S1). However, as the metabolic activity was impaired in
human cells (Figure S2 and Table S3), we utilized a more
complex and representative model of toxicity evaluation
based on a whole organism. The nematode Caenorhabditis e-
legans[9] tolerated high doses of FM233 as well as the inactive
control FM375 (highest tested concentration: 500 mm) over
a period of five days without any signs of toxicity or
developmental defects. Moreover, normal production of L1
larvae and eggs was observed whereas doxorubicin (100 mm),
a toxic control, led to growth arrest (Figure S3). Contrary to
extensive investigations of AEHC and AEBQ targets in
human cells, the bacterial mode of action of this compound
class has remained elusive.[10] Thus we applied our probe
molecules in activity-based protein profiling (ABPP)[11]
experiments and incubated several Gram-positive and
Gram-negative strains with probe molecules in situ. The
cells were lyzed, the probe-tagged proteins were clicked to
rhodamine azide and separated by SDS-PAGE, and labeled
proteins were then visualized by fluorescence scanning (Fig-
ure 2A). Whereas all S. aureus strains exhibited largely
comparable labeling patterns of discrete cytosolic protein
bands (Figure S4), a closer inspection of the Salmonella
specific targets revealed striking differences between the
antibiotic AEBQs (FM239 and FM233) and their correspond-
ing inactive controls lacking the epoxide moiety (FM257 and
FM375; Figure S5). For example, FM233 labeled several
distinct proteins (e.g., a prominent band at 30 kDa) that are
absent in the FM375 treated sample, suggesting that the
epoxide addresses targets essential for bacterial viability
(Figure 2A,B). In contrast, these probes did not reveal
pronounced labeling in S. aureus and exhibited no MIC,
suggesting a species-specific mode of action (Figure S6).
Owing to the challenging discovery of druggable targets
especially in Gram-negative strains, we selected antibiotic
Scheme 1. Synthesis of type I and II AEHCs and type III AEBQs.[7]
Reagents and conditions: a) 1. Boc2O, THF, 708C, 16 h, 99%; 2.
iodobenzene dipivalinate, MeOH, 08C!RT, 16 h, 84%; 3. Boc2O,
THF, 708C, 16 h, 68%; b) 1. H2O2, NaOH, THF, RT, 16 h, 53%; 2.
trifluoroacetic acid, CH2Cl2, RT, 2 h, 96%; c) synthesized acids or acid
chlorides 11a–11e, lithium tert-butoxide, THF, ꢀ108C!RT, 4–16 h,
30–73%; d) BF3·OEt3, CH2Cl2, ꢀ208C!RT, 14 h, 68–93%; e) NaBH-
(OAc)3, MeOH, 08C!RT, 2 h, 53%; f) 1. LiBHEt3, THF, ꢀ788C!RT,
3 h; 2. montmorillonite K10, CH2Cl2, RT, 16 h, 70% over two steps.
aniline (7) with iodobenzene dipivalinate (a; Scheme 1),
followed by epoxidation using aqueous hydrogen peroxide
solution and subsequent deprotection with trifluoroacetic
acid (b). The Michael acceptor–epoxide key fragment 9 was
further coupled to various alkyne-tagged acids or acid
chlorides (see the Supporting Information, Section S9.1) by
amide bond formation (c). Next, the acetal of the partially
protected quinones (12a–12h) was removed with boron
trifluoride diethyl etherate leading to type III AEBQs;
subsequent reduction gave access to type II AEHCs. Upon
inverting these two steps (d, e), type I AEHCs were generated
(f). Benzoquinone analogues lacking the epoxide moiety
(control, Figure 1B) served as reference compounds for
phenotypic studies and labeling experiments (Scheme S1
and Table S1). Prior to proteomic target identification, all
2
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2016, 55, 1 – 7
These are not the final page numbers!