β-Sultams exhibit discrete binding preferences for diverse bacterial enzymes with nucleophilic residues

Article abstract of DOI:10.1039/c3cc46002a

β-Sultams are potent electrophiles that modify nucleophilic residues in selected enzyme active sites. We here identify and characterize some of the specific bacterial targets and show a unique inhibition of the azoreductase family.

Full text of DOI:10.1039/c3cc46002a

ChemComm
View Article Online
View Journal
COMMUNICATION
b-Sultams exhibit discrete binding preferences for
diverse bacterial enzymes with nucleophilic
residues†
Cite this:DOI: 10.1039/c3cc46002a
Received 6th August 2013,
Accepted 25th October 2013
Roman Kolb, Nina C. Bach and Stephan A. Sieber*
DOI: 10.1039/c3cc46002a
www.rsc.org/chemcomm
b-Sultams are potent electrophiles that modify nucleophilic resi-
dues in selected enzyme active sites. We here identify and char-
acterize some of the specific bacterial targets and show a unique
inhibition of the azoreductase family.
Four-membered rings such as b-lactams and b-lactones inhibit diverse
enzyme classes by covalently modifying nucleophilic active site resi-
dues.1,2 Interestingly, although b-lactams and b-lactones are structu-
rally related and present in many bioactive natural products they bind
and inhibit a complementary set of enzymes pointing to a fine-tuned
reactivity towards different active sites. Although not of natural origin
b-sultams represent a third class of four-membered ring scaffolds that
exhibit discrete reactivity towards nucleophiles.3–6 The structural
similarity to b-lactams raised interest in the use of these compounds
Fig. 1 Principle reaction of b-sultams with serine nucleophiles in proteins
and target identification via activity based protein profiling.
as novel antibacterials that overcome resistance development. Thus (Fig. 1 and Fig. S1, ESI†).11–15 We incubated these probe scaffolds with
the binding of b-sultams to b-lactamases and penicillin binding living Burkholderia cenocepacia, B. thailandensis, Listeria welshimeri and
proteins (PBPs), the dedicated targets of b-lactams, has been pre- L. monocytogenes and analyzed the corresponding target profiles via
viously studied.7–9 It was shown that the nucleophilic serine active site gel-based and gel-free mass spectrometry (MS). Interestingly, the
attacks the sulfonyl center and displaces the amine as a leaving group b-sultams under investigation revealed a discrete labeling pattern of
resulting in a stable covalent attachment and corresponding enzyme several different enzymes. One protein, an azoreductase which is an
inhibition (Fig. 1).9 Alternatively, it has been observed that the important enzyme class for azo dye removal in biotechnological
sulfonate ester can undergo an elimination reaction leading to applications, was found in several investigated bacterial systems.
dehydroalanine (Dha).7 b-Sultams exhibit a smaller intramolecular Subsequent activity studies with methyl red revealed that this enzyme
resonance stabilization compared to the b-lactam analogues that in is able to break azo bonds and gets inhibited by selected b-sultams via
combination with ring strain contribute to their elevated reactivity.4 an unprecedented labeling of a conserved threonine residue.
The proteome wide binding of b-lactams has been investigated
Based on previous studies of N-acylated b-sultams as peptidase
previously and a high preference of natural product derived com- inhibitors9,16,17 we designed and synthesized a comprehensive collec-
pounds for several PBPs was observed.2,10 Similar studies with related tion of diverse alkynylated derivatives (Scheme 1 and Fig. S1, ESI†).
b-sultam scaffolds have not been performed up to now. We These compounds thus serve as proteomic tools for the identification
thus designed a small collection of diverse b-sultams with differing of irreversibly bound protein targets in living bacterial cells and extend
side chain decorations as well as an alkyne handle which allows the initial work on isolated enzymes to the diversity of cellular proteins.
us to detect irreversibly bound targets of these compounds in
The collection of b-sultam probes can be divided into three
living bacterial systems via activity based protein profiling (ABPP) categories. The first category comprises molecules that are
N-acylated b-sultams lacking additional substituents at the core
structure (S02 and S05). The alkyne handle for target identifi-
cation is part of the acyl substituent. In the second category
the N-acyl chain varies and an additional phenone moiety is
Department of Chemistry, Center for Integrated Protein Science Munich (CIPSM),
¨
¨
Technische Universitat Muchen, Garching, Germany. E-mail: stephan.sieber@tum.de
† Electronic supplementary information (ESI) available: Synthesis, compound
characterization, cloning, and assays. See DOI: 10.1039/c3cc46002a
This journal is ©The Royal Society of Chemistry 2014
Chem. Commun.

View Article Online
Communication
ChemComm
Fig. 2 b-Sultam profiling. Fluorescent gel of labeled proteins in the
cytosolic fractions of B. thailandensis (A) and L. monocytogenes (B).
Identified protein targets are indicated on the right. For unabbreviated
forms of protein names please refer to Table S1 (ESI†).
Scheme 1 Synthesis of category 1, 2 and 3 b-sultams. For synthetic
details please refer to the ESI.†
attached in the 2-position (S07–S12). The alkyne is either incor-
In order to identify the proteins that prominently associate with
porated in the para position of the phenone or as a terminal part selected b-sultams we utilized a biotin and rhodamine functionalized
of the N-acyl chain. The third category has an amide or ester azide linker and enriched labeled proteins via avidin bead binding.22,23
substituent in the 3-position (S25 and S26). Details of the After SDS-PAGE analysis selected fluorescent protein bands were
synthesis are listed in Scheme 1; Schemes S1 and S2 (ESI†).
isolated, trypsinized and analyzed via HPLC-MS/MS. In addition, we
To test the reactivity of b-sultams with full proteomes we first performed gel-free MS experiments to get a comprehensive list of
incubated intact cells of B. thailandensis and L. monocytogenes with putative targets. Peptide masses were searched with the SEQUEST
various concentrations of probes S07 and S10, respectively. After 2 h algorithm against available databases and the corresponding protein
of incubation the cells were washed, lysed and treated with rhoda- hits are listed in Tables S1–S3 (ESI†). The prominent 50 kDa protein in
mine azide by a copper catalysed click chemistry reaction to append B. thailandensis was identified as adenosylhomocysteinase (AHC) and
the fluorescent tag to compound labeled proteins.18–21 Subsequent the lower bands around 22 kDa as thiol disulfide interchange proteins
separation of the proteome by SDS-PAGE and fluorescent scanning (DsbC and DsbA) and FMN-dependent NADH-azoreductase (FMND-B)
revealed discrete bands that differed in intensity depending on the (Fig. 2A). Interestingly, a related azoreductase (FMND-L) was also
compound concentration (Fig. S2, ESI†). Based on this pattern a identified in L. monocytogenes (Fig. 2B) emphasizing that selected
concentration of 100–200 mM was selected in order to obtain the best b-sultams prefer binding to this enzyme class. In addition, Lmo0898
signal to noise ratio. However, two low molecular weight proteins was identified as an uncharacterized protein, which exhibited sequence
could be visualized down to 20 mM in both bacterial strains similarity to a transcription associated protein for toxin expression.24
emphasizing high binding affinity. Profiling of the entire compound
The two azoreductases exhibit only 32% sequence identity suggest-
collection in situ with pathogenic B. cenocepacia and non-pathogenic ing that the enzymes may recognize different substrates. This is also
B. thailandensis revealed a highly diverse labeling pattern in the supported by the broader labeling of Listeria azoreductase by several
cytosol emphasizing that the compounds are cell-permeable and probes and the exclusive labeling of Burkholderia azoreductase only by
that the different structural decorations directly bind in diverse S07. Importantly, a comparison of the labeling pattern between
active sites (Fig. 2A, Fig. S3 and S5, ESI†). The membrane fraction pathogenic L. monocytogenes and non-pathogenic L. welshimeri reveals
showed a similar but weaker labeling (Fig. S4 and S6, ESI†).
that the azoreductase is predominantly present in the pathogenic
Although no differences in probe binding preferences between strain which may implicate a so far undiscovered role of this enzyme
the two Burkholderia strains could be observed the comparison in disease development (Fig. S5, ESI†). It was previously suggested
of the two Listeria strains revealed a pronounced labeling of that bacteria need azoreductases for the detoxification of quinones
additional proteins in pathogenic L. monocytogenes (Fig. 2B and and thus could be reclassified as NAD(P)H quinone oxidoreduc-
Fig. S5, ESI†). We focused our analysis on B. thailandensis as well as tases.25–27 However, the focus of azoreductase research so far has
L. monocytogenes in order to identify characteristic protein targets of been on its application as a biotechnological tool for the removal of
b-sultams. Interestingly, all members of the small sultam library azo dyes from waste water.27–30
exhibited binding to several proteins in either of the two bacterial
In order to verify the results of MS we cloned identified proteins
species. A significant labeling of proteins in the 21 to 60 kDa range into expression vectors and incubated cells that overexpressed the
of B. thailandensis was obtained for category 2 compound S07 protein with the corresponding probe molecules. Heat denatura-
with a short N-acetyl substitution. Category 2 compound S12 with a tion of the proteome prior to labeling resulted in a lack of binding
N-acetoxy side chain revealed a similar binding pattern but lacks suggesting that the molecules require an intact and folded enzyme
the 21 kDa protein. Profiling of the compound collection in for interaction (Fig. S7, ESI†).
L. monocytogenes revealed a strong preference of category 2 probes
S10, S11 and S12 for a 24 kDa protein.
To elucidate the mechanism of azoreductase specific labeling
we first determined how many molecules are attached per protein.
Chem. Commun.
This journal is ©The Royal Society of Chemistry 2014

View Article Online
ChemComm
Communication
previous literature reports (Table S5, ESI†).31 The b-sultam targets
discovered here are different to those observed by b-lactams and
b-lactones (Fig. S12, ESI†).2,23 One of these targets are azoreductases
that are important enzymes in biotechnological application for
the removal of azo dyes as well as in pathogenic bacteria with
a yet undefined role. Among other applications, the b-sultam
probes introduced here represent customized tools for the
discovery and study of azoreductase activity across different
pathogenic and non-pathogenic bacterial strains.
S.A.S. was supported by the Deutsche Forschungsgemeinschaft
(DFG), SFB749, SFB1035, FOR1406 and an ERC starting grant.
Notes and references
¨
1 T. Bottcher and S. A. Sieber, J. Am. Chem. Soc., 2008, 130,
Fig. 3 MS/MS analysis of the modified Thr122 in FMND-B treated with
S07. For all other spectra please refer to Fig. S10 (ESI†).
14400–14401.
2 I. Staub and S. A. Sieber, J. Am. Chem. Soc., 2008, 130, 13400–13409.
3 M. I. Page and A. P. Laws, Tetrahedron, 2000, 56, 5631–5638.
4 M. Barwick, T. Abu-Izneid and I. Novak, J. Phys. Chem. A, 2008, 112,
10993–10997.
5 P. S. Hinchliffe, J. M. Wood, A. M. Davis, R. P. Austin, R. P. Beckett
and M. I. Page, Org. Biomol. Chem., 2003, 1, 67–80.
6 W. Y. Tsang, N. Ahmed and M. I. Page, Org. Biomol. Chem., 2007, 5,
485–493.
7 W. Y. Tsang, N. Ahmed, P. S. Hinchliffe, J. M. Wood, L. P. Harding,
A. P. Laws and M. I. Page, J. Am. Chem. Soc., 2005, 127, 17556–17564.
8 W. Y. Tsang, N. Ahmed, L. P. Harding, K. Hemming, A. P. Laws and
M. I. Page, J. Am. Chem. Soc., 2005, 127, 8946–8947.
9 A. Llinas, N. Ahmed, M. Cordaro, A. P. Laws, J. M. Frere,
M. Delmarcelle, N. R. Silvaggi, J. A. Kelly and M. I. Page, Biochem-
istry, 2005, 44, 7738–7746.
Intact protein MS of FMND-B with compound S07 resulted in
the addition of one molecule to the enzyme (Fig. S8, ESI†). The
site of modification was determined by digestion of the protein
with trypsin or chymotrypsin after S07 labeling. The resulting
peptides were investigated by MS/MS fragmentation. Proteome
Discoverer (PD) and MaxQuant (MQ) software revealed the
selective modification of Thr122 in a sequence region that is
conserved in several azoreductases (Fig. 3 and Fig. S9, ESI†).
Based on the crystal structure of the related enzyme from E. coli
this residue is outside the active site and does not coordinate to the
10 I. Staub and S. A. Sieber, J. Am. Chem. Soc., 2009, 131, 6271–6276.
FMN cofactor.28 A direct or indirect role of this residue in catalysis 11 M. J. Evans and B. F. Cravatt, Chem. Rev., 2006, 106, 3279–3301.
12 M. Fonovic and M. Bogyo, Expert Rev. Proteomics, 2008, 5, 721–730.
13 P. P. Geurink, L. M. Prely, G. A. van der Marel, R. Bischoff and
has not been proposed so far. The same threonine (Thr128) was
modified in FMND-L with S10 emphasizing a conserved nucleo-
H. S. Overkleeft, Top. Curr. Chem., 2012, 324, 85–113.
philic reactivity across species (Fig. S10, ESI†). Interestingly, MS/MS 14 M. B. Nodwell and S. A. Sieber, Top. Curr. Chem., 2012, 324, 1–41.
15 W. P. Heal, T. H. Dang and E. W. Tate, Chem. Soc. Rev., 2011, 40, 246–257.
16 M. Beardsell, P. S. Hinchliffe, J. M. Wood, R. C. Wilmouth,
sequencing and binding site localization analysis with PD and MQ
revealed an additional Ser152 binding site (Fig. S10, ESI†). Although
C. J. Schofield and M. I. Page, Chem. Commun., 2001, 497–498.
both probes are members of category 2 the difference in the alkyl 17 M. I. Page, P. S. Hinchliffe, J. M. Wood, L. P. Harding and A. P. Laws,
Bioorg. Med. Chem. Lett., 2003, 13, 4489–4492.
18 A. E. Speers, G. C. Adam and B. F. Cravatt, J. Am. Chem. Soc., 2003,
chain (methyl vs. cyclohexyl) seems to significantly influence their
binding in the respective enzyme pockets.
125, 4686–4687.
We finally evaluated if binding of azoreductase directed probes 19 R. Huisgen, Proc. Chem. Soc., 1961, 357–396.
20 V. V. Rostovtsev, J. G. Green, V. V. Fokin and K. B. Sharpless, Angew.
Chem., Int. Ed., 2002, 41, 2596–2599.
21 C. W. Tornøe, C. Christensen and M. Meldal, J. Org. Chem., 2002, 67,
would lead to a corresponding inhibition of enzyme activity. This was
monitored by the enzyme catalyzed cleavage of methyl red. Probes S10
and S07 that both labeled the azoreductases in Listeria and Burkhol-
deria, respectively, revealed a concentration dependent inhibition with
IC50 values of about 30 mM in both cases (Fig. S11, ESI†). In contrast
3057–3064.
22 J. Eirich, J. L. Burkhart, A. Ullrich, G. C. Rudolf, A. Vollmar,
S. Zahler, U. Kazmaier and S. A. Sieber, Mol. BioSyst., 2012, 8,
2067–2075.
¨
probes S05 and S02 that did not label both enzymes also revealed no 23 T. Bottcher and S. A. Sieber, Angew. Chem., Int. Ed., 2008, 47,
4600–4603.
inhibition (Table S4, ESI†). Based on the important role of azoreduc-
tases for biotechnological applications as well as their yet undiscovered
24 S. J. Johnson, D. Close, H. Robinson, I. Vallet-Gely, S. L. Dove and
C. P. Hill, J. Mol. Biol., 2008, 377, 1460–1473.
role in many bacteria our compounds represent active site directed 25 G. Liu, J. Zhou, R. Jin, M. Zhou, J. Wang, H. Lu and Y. Qu, Appl.
Microbiol. Biotechnol., 2008, 80, 409–416.
26 G. Liu, J. Zhou, Q. S. Fu and J. Wang, J. Bacteriol., 2009, 191,
tools to study their activity and function in living cells.
b-Sultams have attracted attention in the past as substitutes for
6394–6400.
b-lactam antibiotics which suffer from bacterial resistance devel- 27 A. Ryan, C. J. Wang, N. Laurieri, I. Westwood and E. Sim, Protein
Cell, 2010, 1, 780–790.
28 K. Ito, M. Nakanishi, W. C. Lee, Y. Zhi, H. Sasaki, S. Zenno, K. Saigo,
opment. Studies with recombinant PBPs, elastase and b-lactamase
indeed suggested inhibition of these targets by covalent bind-
Y. Kitade and M. Tanokura, J. Biol. Chem., 2008, 283, 13889–13896.
ing.9,16,17 Our in situ cellular profiling did not indicate interaction 29 P. A. Ramalho, S. Paiva, A. Cavaco-Paulo, M. Casal, M. H. Cardoso
and M. T. Ramalho, Appl. Environ. Microbiol., 2005, 71, 3882–3888.
30 J. Maier, A. Kandelbauer, A. Erlacher, A. Cavaco-Paulo and
with these enzymes by a covalent interaction. Correspondingly, we
did not observe any antibacterial activity of all compounds used
G. M. Gubitz, Appl. Environ. Microbiol., 2004, 70, 837–844.
here up to a concentration of 1000 mM which is in line with 31 M. Mu¨ller and H.-H. Otto, Liebigs Ann. Chem., 1991, 171–178.
This journal is ©The Royal Society of Chemistry 2014
Chem. Commun.