The biological targets of acivicin inspired 3-chloro- and 3-bromodihydroisoxazole scaffolds

Article abstract of DOI:10.1039/c0cc02825h

Target analysis of acivicin derived 3-halodihydroisoxazoles scaffolds in living non-pathogenic and pathogenic bacteria.

Full text of DOI:10.1039/c0cc02825h

View Online / Journal Homepage / Table of Contents for this issue
COMMUNICATION
www.rsc.org/chemcomm | ChemComm
The biological targets of acivicin inspired 3-chloro- and
3-bromodihydroisoxazole scaffoldsw
¨
Ronald Orth, Thomas Bottcher and Stephan A. Sieber*
Received 27th July 2010, Accepted 22nd September 2010
DOI: 10.1039/c0cc02825h
Target analysis of acivicin derived 3-halodihydroisoxazoles
scaffolds in living non-pathogenic and pathogenic bacteria.
Nature’s repertoire of bioactive compounds has provided a
rich source of potent drugs against many human diseases1 and
a diverse molecular toolkit for biochemical research. Due to
the tremendous success of protein reactive natural products
such as beta-lactams,2 beta-lactones3,4 and Michael acceptor5
systems to inhibit key enzymes involved in viability and
pathogenesis we here extend our efforts to unravel the targets
of additional privileged electrophilic natural products. We
report the synthesis of acivicin inspired 3-chloro- and 3-bromo-
dihydroisoxazole probes and their application in target profiling
in non-pathogenic bacteria as well as in pathogenic bacteria
such as S. aureus and multiresistant S. aureus (MRSA).
Acivicin is produced by Streptomyces sviceus6 and was first
structurally characterized by Hanaka et al. in 1973.6 It is
reported to be a glutamine antimetabolite antibiotic7 that in
addition exhibits anti-tumor activity. Although no detailed
target analysis has been carried out so far, in vitro analysis
showed that several important enzymes such as L-glutamine-
dependent amidotransferases,8 glutamate synthase (GltS)9 as
well as g-glutamyl transpeptidase (g-GT) could be inhibited
by a covalent modification of the active site cysteine and
threonine residues.10
Fig. 1 Identification of target proteins via ABPP. Modification of
cysteine by the 3-chlorodihydroisoxazole probe.17
reactive compounds that are prone to react with enzyme active
sites containing residues of elevated nucleophilicity such as
cysteines and serines. Nucleophilic attack via an addition–
elimination mechanism (Fig. 1) most likely leads to a halogen
displacement and a resulting stable covalent bond.17 As this
intrinsic reactivity of the 3-halodihydroxyisoxazole is a crucial
parameter for their inhibition capability we utilized a bromo
substituent (probes 1 to 4) in addition to the naturally occurring
chloro substituent (probes 5 to 8) to investigate and compare
protein reactivity profiles (Fig. 1).
We equipped the core scaffold with a comprehensive set of
diverse substituents ranging from unsubstituted to small aliphatic
(methyl and ethyl) and aromatic (phenyl) residues to explore
different target protein selectivities.
As a total proteome analysis of the acivicin derived privileged
core structure 3-halo-dihydroisoxazoles in living bacteria will
likely reveal a more complete picture of the full complement of
dedicated targets, we designed and synthesized structurally
diverse 3-chloro- and 3-bromo-dihydroisoxazole probes for
activity based protein profiling (ABPP).11–13 All probes con-
tain an alkyne handle as a benign tag for the modification with
rhodamine and biotin azides via the Huisgen cycloaddition
(click chemistry, CC)14–16 after cell penetration and lysis to
visualize and identify target proteins via fluorescent SDS-gel
analysis and mass spectrometry (MS), respectively (Fig. 1).
In order to explore the protein binding preferences of
3-halodihyroisoxazoles we designed several probes which were
fine tuned towards binding and reactivity by substitution of
the dihydroisoxazole ring with chloro- and bromo substituents
in 3 position and structurally diverse residues in 4 and 5
positions. Halogen substituted dihydroisoxazoles are protein
The synthesis of probes 1 to 8 started by an established
procedure. Glyoxylic acid reacted with hydroxyl amine to give
glyoxylic acid aldoxime18 that was subsequently converted to
dibromo- or dichloroformaldoxime by treatment with Br2 or
N-chloro succinimide, respectively (Scheme 1 and Scheme S1,
ESIw).19 Both formaldoximes were used immediately for the
cycloaddition reaction, in which they reacted with unsaturated
alcohols to give a 3-halo-5-(hydroxymethyl)-dihydroisoxazole.
Although synthetically very useful, this syn-cycloaddition
was not enantio- and regioselective and generated racemic
mixtures of 3-halodihydroisoxazoles for compounds 14 and 15
as well as cis-substituted regioisomers (variation of substitution
in ring positions 4 and 5) for compounds 16–21. As the
preferences of dedicated targets were unknown, we rationalized
Center for Integrated Protein Science Munich CIPSM
Department of Chemistry, Technische Universitat Munchen,
Lichtenbergstr. 4, 85747 Garching, Germany.
E-mail: Stephan.sieber@tum.de; Fax: +49 289 13302;
Tel: +49 289 13319
w Electronic supplementary information (ESI) available: Synthetic
procedures, characterization, biological methods and supplementary
figures. See DOI: 10.1039/c0cc02825h
,
¨
¨
Scheme 1 Synthesis of the probes 1 and 5. For other probes please
refer to ESI.w (a) NH2OH, H2O, 16 h, rt, (b) for dibromo: Br2, H2O,
3 h, 0 1C to rt, (c) for dichloro: N-chloro succinimide, DME, 10 min,
110 1C to rt, (d) allyl alcohol, K2CO3, EtOAc, 24 h, rt, (e) hexynoic
acid chloride, DCM, TEA, 16 h, rt.
c
This journal is The Royal Society of Chemistry 2010
Chem. Commun., 2010, 46, 8475–8477 8475

View Online
that for initial screens multiplexed regioisomeric mixtures of
our probes would likely increase the chance of productive
binding in proteome profiling studies. If specific binding is
observed for a regioisomeric mixture, deconvolution by HPLC
based separation and subsequent individual analysis will
reveal the binding preferences of each individual compound
with their dedicated targets. All 3-halo-5-(hydroxymethyl)-
dihydroisoxazoles were easily converted to their corresponding
activity probes by esterification with hex-5-ynoic acid chloride
in high yield (ESIw).
We initiated our proteome screening with 3-chloro- and
3-bromo-dihydroisoxazol probes (1 to 8) against intact cells of
B. licheniformis, P. putida, L. welshimeri as well as pathogenic
S. aureus and a resistant MRSA strain. These initial screens
were carried out with regioisomeric probe mixtures (2–4 and 6–8)
to speed up the discovery process and identify promising probe
scaffolds and ring substitutions which, in case of promising
candidates, were subsequently separated and again investi-
gated by repeated proteome profiling to unravel the structure
of the active species (see below). Cells were grown into
stationary phase and incubated with varying probe concentra-
tions for 2 h. Upon incubation, cells were lyzed and a
fluorescent rhodamine azide attached via CC. Subsequent
SDS-gel analysis and fluorescent scanning revealed a distinct
set of specific protein bands in all proteomes investigated,
emphasizing that the probes display suitable properties for
in situ studies. A concentration of 150 mM and an incubation
time of 2 h turned out to be optimal for saturated labeling
(Fig. 2a and b and Fig. S1a,b, ESIw).
5 (isomer 3a) or in 4 position (isomer 3b) exhibited a binding
preference for a protein of lower molecular weight (Fig. 2a).
Interestingly, probe mixture 2 with a shorter methyl side chain
in 5 (isomer 2a) or in 4 position (isomer 2b) revealed again a
different pattern that represents a combination of the other
probes, emphasizing that the targeted enzymes exhibit very
distinct active site geometries.
Phenyl-substituted probe isomers 4 showed significantly less
labeling events in the majority of proteomes (Fig. 2 and Fig. S1
and S2, ESIw). Comparison of all proteome labeling profiles
further revealed that chloride substituted dihydroisoxazoles led
to a slightly weaker labeling compared to the bromide substi-
tuted counterparts, but not to a change in target preferences.
Target identification was carried out by mass spectrometry
(MS) using a quantitative enrichment strategy for labeled
enzymes (ESIw).3 Among the identified enzymes there was a
surprisingly large fraction of six members of the dehydrogenase
(DH) enzyme family including pyrroline-5-carboxylate DH
(PCDH), aldehyde DH (ADH) and bifunctional aldehyde-
CoA/alcohol dehydrogenase (AADH) emphasizing that the
probe structures and reactivities are well suited to interact with
this important enzyme class (Table S1, ESIw). As the substrate
of PCDH, pyrroline-5-carboxylate (Fig. S3, ESIw), is struc-
turally related to probes 1 and 5, the observed labeling of this
enzyme by probe 1 seems to be based on similar recognition
elements.
To confirm the results of MS by additional independent
experiments, we recombinantly expressed several of our observed
targets and labeled them subsequently with the corresponding
probes. In fact, in all cases the probes labeled their targets in
an activity dependent manner as confirmed by heat denatura-
tion controls (Fig. 2c and Fig. S1c, ESIw).
To investigate the mode of binding in more detail, we
incubated AADH, a so far uncharacterized DH, with probe
1 and applied the labeled enzyme to MS analysis after trypsin
digestion to search for the peptide fragment and residue with
the attached probe. Among the peptide fragments, one sequence
occurred in which Cys 255 was covalently modified with the
dihydroisoxazole scaffold (Fig. S4 and Table S2, ESIw) sup-
porting our initial hypothesis that nucleophilic residues attach
to the electrophilic 3-halo-dihydroisoxazole ring by an addition–
elimination mechanism. The observed acylation of cysteine
corresponds very well to the oxidation mechanism of many
aldehyde dehydrogenases in which a cysteine residue attacks the
aldehyde moiety in the first step of catalysis.20
Interestingly, a comparison of the labeling profiles of
compounds with different ring substitutions revealed a highly
diverse and distinct pattern of enzyme binding preferences
(Fig. 2a and Fig. S1a and S2, ESIw).
In S. aureus for example, probe 1 with an unsubstituted
scaffold revealed the labeling of two different enzymes, while
in contrast probe isomers 3 with an ethyl substituent in
All identified enzymes are important for the primary meta-
bolism of bacteria and involved in multiple cellular functions
such as glutamate and proline metabolism in case of PCDH21
or alcohol metabolism for energy production in case of
AADH22 (Table S1, ESIw). Moreover, FabH, the only identi-
fied enzyme that does not belong to the dehydrogenase enzyme
family, is an important acyltransferase that is crucial for
bacterial fatty acid biosynthesis.23
Fig. 2 (a) In situ labeling (fluorescent images) of non-pathogenic
S. aureus NCTC 8325 using probes 1 to
8 (1 h incubation).
Aldehyde dehydrogenase (AldA, 53.6 kDa), aldehyde dehydrogenase
(ADH, 51.7 kDa) and 3-oxoacyl-[acyl-carrier-protein] synthase III
(FabH, 33.9 kDa) were detected in S. aureus NCTC 8325. (b) Titration
of probes 1 and 3 in in situ labeling experiments. (c) In vitro labeling of
recombinant, overexpressed target enzymes (+: positive control using
native proteome, n: not induced, i: induced, À: heat control using
induced proteome).
All previous screens revealed that unsubstituted (probe 1) as
well as methyl (probes 2a + b) and ethyl (probes 3a + b)
substituted bromine dihydroisoxazole probes represent a minimal
probe set for an almost complete coverage of all observed
biological targets. As this initial screen was designed to rapidly
identify promising scaffold decorations by the application of
c
8476 Chem. Commun., 2010, 46, 8475–8477
This journal is The Royal Society of Chemistry 2010

View Online
structural similarity of probe 1 with acivicin, which was
reported to be an irreversible inhibitor of g-GT,10 no labeling
of the purified recombinant enzyme could be observed by
probe 1 at concentrations at which recombinant AADH
and PCDH showed strong signals (Fig. S7 and S8, ESIw).
This emphasizes that small variations at the dihydroisoxazole
scaffold may significantly alter target preferences.
In conclusion we designed and synthesized a library of
natural product inspired 3-chloro- and 3-bromodihydroisoxazole
probes for bacterial proteome analysis and obtained a unique
preference of the probes to label and inhibit members of the
DH enzyme family via an addition–elimination mechanism on
a nucleophilic cysteine residue. The probe selectivity for a
certain enzyme target is remarkable as small structural
changes e.g. methyl vs. ethyl substitution already discriminate
binding. Although the probes did not reveal antibiotic effects,
their application for the functional analysis and annotation
of uncharacterized DH enzymes is a useful addition to
the proteomic tool box and enlarges our knowledge on how
nature utilizes electrophilic scaffolds for directed protein
inhibition.
Fig. 3 Labeling studies by isolated isomers. (a) Structural composi-
tion of individual isomers that constitute a minimal probe set for an
almost complete target coverage. (b) Labeling profiles of the individual
isomers with several bacterial proteomes (MW = molecular weight
marker).
multiplexed regioisomeric samples, we next deconvoluted this
information in order to unravel the labeling preferences of the
most valuable individual isomers 2(a + b) and 3(a + b)
(Fig. 3a). Preparative HPLC was applied to separate regio-
isomers (racemic) which were subsequently characterized by
2D NMR (COSY, NOESY, HMQC and HMBC). In brief,
probes 2a/3a were assigned as methyl substituted in 5 position
(ESIw) and vice versa probe 2b/3b are methyl substituted in
4 position. All probes were cis configured.
We thank Dr Markus Gerhard (MRI) for providing g-GT.
We gratefully acknowledge funding by the Deutsche
Forschungsgemeinschaft (DFG), a DFG grant (SFB 749)
and by the Center for Integrated Protein Science Munich
CIPSM.
Subsequent labeling studies with bacteria under in situ con-
ditions revealed indeed a remarkable selectivity of individual
probes for certain enzyme targets depending on the ring
substitution pattern (Fig. 3b). For instance, in S. aureus probe
2a with the methyl group in 5 position labels the FabH enzyme
but once the methyl group exchanges with the hexynoic ester
group (probe 2b) ADH is labeled instead. A similar pattern
can be obtained in P. putida as well as in L. welshimeri where
FabH is predominantly labeled by 2a and 3a but AADH
exclusively by 2b.
Notes and references
1 J. Clardy and C. Walsh, Nature, 2004, 432, 829–837.
2 I. Staub and S. A. Sieber, J. Am. Chem. Soc., 2009, 131, 6271–6276.
3 T. Bottcher and S. A. Sieber, Angew. Chem., Int. Ed., 2008, 47,
¨
4600.
4 T. Bottcher and S. A. Sieber, J. Am. Chem. Soc., 2008, 130,
¨
14400.
5 M. Pitscheider and S. A. Sieber, Chem. Commun., 2009, 3741.
6 L. J. Hanka, D. G. Martin and G. L. Neil, Cancer Chemother.
Rep., 1973, 57, 141.
7 H. N. Jayaram, D. A. Cooney, J. A. Ryan, G. Neil, R. L. Dion and
V. H. Bono, Cancer Chemother. Rep., 1975, 59, 481.
8 R. H. Earhart and G. L. Neil, Adv. Enzyme Regul., 1985, 24, 179.
9 J. Y. Tso, S. G. Bower and H. Zalkin, J. Biol. Chem., 1980, 255,
6734.
This unique selectivity emphasizes the great utility of the
dihydroisoxazole scaffold as a tool for proteome profiling
experiments in order to identify biologically relevant targets
and study their activity and regulation in the context of
complex proteomic mixtures.
10 C. M. Mc, M. E. Callender and J. F. Lawlis, Science, 1951, 113,
202.
11 M. J. Evans and B. F. Cravatt, Chem. Rev., 2006, 106, 3279.
12 M. Fonovic and M. Bogyo, Expert Rev. Proteomics, 2008, 5, 721.
Since we were able to confirm the covalent binding mode of
probe 1 with AADH we continued to investigate the capability
of this probe to inhibit enzyme activity. AADH showed
activity in a substrate turnover assay with acetaldehyde and
NAD+ in which enzyme catalyzed NAD+ reduction to
NADH/H+ could be followed via UV-vis at 340 nm. Upon
addition of the best AADH labeling probe 1, the enzyme was
effectively inhibited with an IC50 of 11 mM (Fig. S5, ESIw). All
other probes, which exhibited less or little labeling of AADH
such as probes 3 and 4, also showed only weak inhibition
properties with IC50s > 100 mM (Fig. S6, ESIw).
The good cell permeability, active site labeling and inhibition
of several enzymes raised the question if the compounds
exhibit any antibacterial effect. Although the targeted enzymes
are involved in important cellular processes we were not able
to detect any antibacterial activity of probes 1, 2, 3, 5 and 6 up
to a concentration of 500 mM in S. aureus, Listeria welshimeri,
Bacillus licheniformis and Pseudomonas putida. Despite the
13 T. Bottcher, M. Pitscheider and S. A. Sieber, Angew. Chem., Int.
¨
Ed., 2010, 49, 2680.
14 V. V. Rostovtsev, J. G. Green, V. V. Fokin and K. B. Sharpless,
Angew. Chem., Int. Ed., 2002, 41, 2596.
15 C. W. Tornoe, C. Christensen and M. Meldal, J. Org. Chem., 2002,
67, 3057.
16 A. E. Speers, G. C. Adam and B. F. Cravatt, J. Am. Chem. Soc.,
2003, 125, 4686.
17 S. V. Chittur, T. J. Klem, C. M. Shafer and V. J. Davisson,
Biochemistry, 2001, 40, 876.
18 H. Wieland, Ber. Dtsch. Chem. Ges., 1910, 43(3), 3362.
19 K. Halling, K. B. G. Thomsen and I. Torssell, Liebigs Ann. Chem.,
1989, 958.
20 Z. J. Liu, et al., Nat. Struct. Biol., 1997, 4, 317.
21 E. Inagaki, N. Ohshima, H. Takahashi, C. Kuroishi, S. Yokoyama
and T. H. Tahirov, J. Mol. Biol., 2006, 362, 490.
22 I. Bruchhaus, Biochem. J., 1994, 303, 743.
23 A. C. Price, K. H. Choi, R. J. Heath, Z. Li, S. W. White and
C. O. Rock, J. Biol. Chem., 2001, 276, 6551.
c
This journal is The Royal Society of Chemistry 2010
Chem. Commun., 2010, 46, 8475–8477 8477