4-Quinolones as noncovalent inhibitors of high molecular mass penicillin-binding proteins

Article abstract of DOI:10.1021/ml3001006

Penicillin-binding proteins (PBPs) are important bacterial enzymes that carry out the final steps of bacterial cell wall assembly. Their DD-transpeptidase activity accomplishes the essential peptide cross-linking step of the cell wall. To date, all attemp

Full text of DOI:10.1021/ml3001006

Letter
pubs.acs.org/acsmedchemlett
4-Quinolones as Noncovalent Inhibitors of High Molecular Mass
Penicillin-Binding Proteins
Abbas G. Shilabin,‡ Liudmila Dzhekieva,‡ Pushpa Misra,§ B. Jayaram,§ and R. F. Pratt*,‡
‡Department of Chemistry, Wesleyan University, Middletown, Connecticut 06459, United States
§Department of Chemistry, IIT Delhi, Hauz Khas, New Delhi 110016, India
S
* Supporting Information
ABSTRACT: Penicillin-binding proteins (PBPs) are important bacterial enzymes that carry
out the final steps of bacterial cell wall assembly. Their DD-transpeptidase activity
accomplishes the essential peptide cross-linking step of the cell wall. To date, all attempts
to discover effective inhibitors of PBPs, apart from β-lactams, have not led to new antibiotics.
Therefore, the need for new classes of efficient inhibitors of these enzymes remains. Guided by
a computational fragment-based docking procedure, carried out on Escherichia coli PBP5, we
have designed and synthesized a series of 4-quinolones as potential inhibitors of PBPs. We
describe their binding to the PBPs of E. coli and Bacillus subtilis. Notably, these compounds
bind quite tightly to the essential high molecular mass PBPs.
KEYWORDS: Penicillin-binding protein, DD-peptidase, noncovalent inhibitor, 4-quinolone
acterial resistance to β-lactam antibiotics continues to
B
rise.1−4 The targets of these antibiotics are the penicillin-
binding proteins (PBPs) or DD-peptidases, which catalyze the
final peptide cross-linking steps in peptidoglycan synthesis.5,6
There are a number of variants of these enzymes, which appear
to possess somewhat different activities, including trans-
peptidase, carboxypeptidase, and endopeptidase.7,8
Beyond β-lactams, few effective DD-peptidase inhibitors have
been discovered, and, of these, none have been developed into
new antibiotics. Direct substrate (and their derived transition
states) analogues have not proved successful.8−10 Thus we, and
others,11−14 following a variety of approaches, have sought new
leads. We describe here some results of a computational,
fragment-based approach. We were particularly interested in
exploring small molecules that interacted with the conserved,
polar residues of the DD-peptidase active site.5,6 A platform
that strongly interacted with those residues could then be
expanded for particular applications. We initially focused
especially on neutral ligands since the thermodynamics of ion
binding are complicated.
An initial lead that interested us was the 4-quinolone 1,
composed from fragments and docked15,16 into the active site
of Escherichia coli PBP5 [this enzyme was chosen because a
crystal structure is available17 and substrates are available for
direct assay in vitro8], as shown in Figure 1. We were impressed
by the spanning of the active site by a 4-quinolone and
envisioned the prospect of further interactions with oxyalkyl
substituents. We therefore obtained the 4-quinolones 2−10 and
assessed their potential as ligands of E. coli and Bacillus subtilis
PBPs. Strikingly, we found that they bind quite tightly to the
essential high molecular mass PBPs such as E. coli PBP 1a/1b,
2, and 3.
Figure 1. Computational docking of 1 into the E. coli PBP5 active site.
The 4-quinolones were either purchased (2 and 3) or
synthesized (Scheme 1). The classical Conrad−Limpach
cyclization of an aniline and β-ketoester is a standard synthetic
route for the preparation of 4-quinolones18 and seemed
appropriate for our purpose. To synthesize 4-quinolones 4−8
with side chain functionality at C-2 and C-3, we developed
shorter and more efficient syntheses than those previously
reported (synthetic details may be found in the Supporting
Information).
Following the Conrad−Limpach reaction, equimolar quanti-
ties of aniline and β-ketoesters 11b,c were reacted in a
desiccator over concentrated sulfuric acid to yield correspond-
ing Schiff bases 12b,c. Thermal cyclization of 12b,c in
DowTherm A resulted in the construction of quinolone-2-
Received: April 24, 2012
Accepted: May 28, 2012
Published: June 4, 2012
© 2012 American Chemical Society
592
dx.doi.org/10.1021/ml3001006 | ACS Med. Chem. Lett. 2012, 3, 592−595

ACS Medicinal Chemistry Letters
Letter
a
Scheme 1. Synthesis of 4-Quinolones
a
Reagents and conditions: (a) desiccator over H2SO4 (conc), rt, 2 days; (b) DowTherm A, 250 °C, 20 min; (c) NaOH (2 N), rt, 15 h; (d)
BH3·SMe2, THF, rt, 15 h; (e) MeCN, reflux, 2 h.
carboxylic acid ethyl esters 13b,c. Reduction of esters 13a−c by
the borane-dimethyl sulfide (BH3·SMe2) complex provided the
corresponding alcohols 4−6 with varied length side chains.
By a similar procedure to that employed to obtain 4−6,
treatment of aniline with ethyl-α-formyl succinate (14) led to
the synthesis of quinolone-3-carboxylic acid ethyl ester 16. This
was followed by hydrolysis of 16 to acid 7 or reduction with the
borane−dimethyl sulfide complex to alcohol 8.
To access the hydroxy-4-quinolones 9 and 10, Meldrum’s
acid and 3-hydroxyaniline were subjected to a condensation
reaction in the presence of trimethyl orthoformate to form the
corresponding arylaminomethylene derivative 17. Thermal ring
closure of the Meldrum’s acid derivative afforded a mixture of
5- and 7-hydroxy-4-quinolone analogues in low yield. The
challenging separation of the isomers was performed by
repetitive RP-HPLC on a C-18 analytical column. All products
were characterized as described in the Supporting Information
and were at least 95% pure by NMR estimation.
The dissociation constants (Ki) of these molecules from their
PBP complexes were measured by their competition with the
irreversible reaction of the fluorescent penicillin Bocillin FL
with membrane-bound PBPs of E. coli and B. subtilis, as
described previously.10 Some typical experimental data is
shown in Figure 2. The resulting Ki values from the E. coli
measurements are presented as Table 1.
The results show that the parent molecule, 4-quinolone (2),
at concentrations up to 1 mM, does not bind measurably to any
E. coli PBP. The derivatives 3−10, however, do bind to the
various PBPs to varying degrees. All compounds bound most
tightly to PBP4, which is a nonessential low molecular mass
class C PBP with a distinctive substrate-binding site.8,19 The
weakest binding was to the low molecular mass class A PBPs 5
and 6. Notably, after PBP4, the tightest binding is to the high
molecular mass PBPs 1a/1b, 2, and 3. The latter enzymes in
bacteria are generally essential to bacterial survival and are the
β-lactam antibiotic targets. Of the various 4-quinolones tested,
4 was generally the tightest binding, with Ki values of around 30
Figure 2. (A) Extent of fluorescence labeling of E. coli PBPs by
Bocillin FL (20 μM) in the presence of various concentrations of the
4-quinolone 6. (B) Extent of Bocillin FL (20 μM) labeling of
membrane-bound E. coli PBP2 as a function of the concentrations of 6
( ) and 7 ( ). In each case, the ordinate is normalized to the
observed initial intensity. Other details of these experiments are found
in the Supporting Information.
□
●
593
dx.doi.org/10.1021/ml3001006 | ACS Med. Chem. Lett. 2012, 3, 592−595

ACS Medicinal Chemistry Letters
Letter
Table 1. Dissociation Constants of 4-Quinolones from E. coli PBPs in Membrane Preparations
a
4-quinolone
Ki (μM)
b
b
PBP1a/1b
PBP2
PBP3
PBP4
NB
37
PBP5/6
c
2
NB
NB
NB
NB
NB
d
3
470 50
210 50
950 230
7
e
e
4
27
37
49
4
9
6
26 4 (64 10)
27
7
4.8 2.6
7.0 2.0
6.1 0.7
220 80 (NB)
5
26
50
8
5
120 80
120 50
NB
520 160
NB
6
7
300 50
250 30
29
5
NB
8
38
8
27
4
70 20
NB
3.1 0.8
NB
NB
9
NB
NB
89
NB
10
510 40
3
NB
55 10
NB
a
b
The uncertainties quoted are from fitting the data to eq 1. Within the limits cited, these values could be reproduced in separate experiments. These
c
d
were not resolved on the gel. NB: no binding observed at 1 mM ligand. The large uncertainties of the Ki values of PBP3 reflect the faintness of this
band on the gel (see Figure 1). Experiments carried out with purified enzymes in solution.
e
μM for each of the high molecular mass enzymes. The position
of the hydroxyl group in 4 seems moderately important, as
judged by the generally lesser effectiveness of 5, 6, and 8. All of
the alkanols are, however, much more effective than the parent
quinolone 2. The carboxylates, 3 and 7, are distinctly less
potent, presumably reflecting the cost of dehydration and/or
inserting a charge into a more hydrophobic medium than bulk
water. Aromatic ring hydroxylation (9 and 10) gave only
modest improvement over the case of 2.
The association of 4 with two purified enzymes, E. coli PBP2
and PBP5, in solution was also determined. As reported in
Table 1, binding by the membrane-associated enzymes appears
somewhat stronger. This could arise from differences in the
structure/dynamics of the membrane-associated holoenzymes
from those of the truncated solubilized enzymes in solution, or
perhaps from sequestration of the neutral 4 to the membrane.
That the 4-quinolones 3−10 should bind to PBPs other than
the PBP5 template is not unexpected because the active site
components to which the quinolones were predicted to bind
(Figure 1) are conserved in all of the PBPs and, for catalysis, are
arranged in a very similar fashion. It seems likely that the
additional polar functionality of 3−10, not present in 2,
interacts favorably with some elements of the diverse set of
active site functional groups shown in Figure 1.
PBPs in membranes (Supporting Information, Figure S1), they
do not have any antibacterial activity against the Gram-positive
B. subtilis, which lacks an outer membrane.
This research has shown that small suitably functionalized 4-
quinolones are quite effective inhibitors of the high molecular
mass PBPs of E. coli and B. subtilis. Although the present
molecules appear to have no antimicrobial activity themselves,
it is possible that more elaborated derivatives may have such
activity as well as stronger affinity for the PBPs. The search for
such compounds is in progress. Structural studies would also be
informative.
ASSOCIATED CONTENT
■
S
* Supporting Information
Detailed experimental proceduresorganic syntheses, charac-
terization of products, bacteria and membrane preparation, and
determination of binding constantsand a figure demonstrat-
ing the reaction of 4-quinolones with B. subtilis PBPs. This
material is available free of charge via the Internet at http://
pubs.acs.org.
AUTHOR INFORMATION
■
Corresponding Author
A further notable result was that 3−10 did not have
antibiotic activity against E. coli (MIC > 1 mM). The direct
conclusion is that they are unable to reach their target enzymes
in vivo. This conclusion was confirmed by experiments to
directly measure the binding of 3−10 to the target PBPs in
vivo; none was observed. This may suggest that the outer
membrane of E. coli blocked their access. This explanation is
made less likely, however, by the additional observation that,
although 3−10 also bind to B. subtilis high molecular mass
*E-mail: rpratt@wesleyan.edu.
Funding
This research was supported by National Institutes of Health
Grant AI-17986 (RFP) and the Department of Biotechnology,
Government of India (B.J.).
Notes
The authors declare no competing financial interest.
594
dx.doi.org/10.1021/ml3001006 | ACS Med. Chem. Lett. 2012, 3, 592−595

ACS Medicinal Chemistry Letters
Letter
(19) Sauvage, E.; Duez, C.; Herman, R.; Kerff, F.; Perrella, S.;
ABBREVIATIONS
■
̀
Anderson, J. W.; Adediran, S. A.; Pratt, R. F.; Frere, J.-M.; Charlier, P.
MIC, minimum inhibitory concentration; PBP, penicillin-
binding protein; RP-HPLC, reverse phase high pressure liquid
chromatography; THF, tetrahydrofuran
Crystal structure of the Bacillus subtilis penicillin-binding protein 4a
and its complex with peptidoglycan-mimetic peptide. J. Mol. Biol.
2007, 371, 528−539.
REFERENCES
■
(1) Sykes, R. The 2009 Garrod lecture: The evolution of
antimicrobial resistance: a Darwinian perspective. J. Antimicrob. Ther.
2010, 65, 1842−1852.
(2) Llarrull, L. I.; Testero, S. A.; Fisher, J. F.; Mobashery, S. The
future of β-lactams. Curr. Opin. Microbiol. 2010, 13, 651−657.
(3) Walsh, T. R. Emerging Carbapenamases: A global perspective.
Int. J. Antimicrob. Agents 2010, 36 (Suppl 3), S8−S14.
(4) Cornaglia, G.; Giamarellou, H.; Rossolini, G. M. Metallo-β-
lactamases: A last frontier for β-lactams. Lancet Infect. Dis. 2011, 11,
381−393.
(5) Macheboeuf, P.; Conteras-Martel, C.; Job, V.; Dideberg, O.;
Dessen, A. Penicillin binding proteins: Key players in bacterial cell
cycle and drug resistance processes. FEBS Microbiol. Rev. 2006, 30,
673−691.
(6) Sauvage, E.; Kerff, F.; Terrak, M.; Ayala, J. A.; Charlier, P. The
penicillin-binding proteins: Structure and role in peptidoglycan
biosynthesis. FEMS Microbiol. Rev. 2008, 32, 234−258.
(7) Popham, D. L.; Young, K. D. Role of penicillin-binding proteins
in bacterial cell morphogenesis. Curr. Opin. Microbiol. 2003, 6, 594−
599.
(8) Nemmara, V. V.; Dzhekieva, L.; Sarkar, K. S.; Adediran, S. A.;
Duez, C.; Nicholas, R. A.; Pratt, R. F. Substrate specificity of low-
molecular mass bacterial DD-peptidases. Biochemistry 2011, 50,
10091−10101.
(9) Kumar, I.; Josephine, H. J.; Pratt, R. F. Reactions of
peptidoglycan-mimetic β-lactams with penicillin-binding proteins in
vivo and in membranes. ACS Chem. Biol. 2007, 2, 620−624.
(10) Dzhekieva, L.; Kumar, I.; Pratt, R. F. Inhibition of bacterial DD-
peptidases (penicillin-binding proteins) in membranes and in vivo by
peptidoglycan-mimetic boronic acids. Biochemistry 2012, 51, 2804−
2811.
(11) Miguet, L.; Zervosen, A.; Gerards, T.; Pasha, F. A.; Luxen, A.;
Disteche-Nguyen, M.; Thomas, A. Discovery of new inhibitors of
̀
Streptococcus pneumoniae penicillin-binding protein (PBP) 2x by
structure-based virtual screening. J. Med. Chem. 2009, 52, 5926−5936.
(12) Turk, S.; Verlaine, O.; Gerards, T.; Zivec, M.; Humljan, J.; Sosic,
I.; Amoroso, A.; Zervosen, A.; Luxen, A.; Joris, B.; Gobec, S. New non-
covalent inhibitors of penicillin-binding proteins from penicillin-
resistant bacteria. PLoS ONE 2011, 6, e19418.
(13) Contreras-Martell, C.; Amoroso, A.; Woon, E. C. Y.; Zervosen,
A.; Inglis, S.; Martins, A.; Verlaine, O.; Rydzik, A. M.; Job, V.; Luxen,
A.; Joris, B.; Schofield, C. J.; Dessen, A. Structure-guided design of cell
wall biosynthesis inhibitors that overcome β-lactam resistance in
Staphylococcus aureus (MRSA). ACS Chem. Biol. 2011, 6, 941−951.
(14) Woon, E. C. Y.; Zervosen, A.; Sauvage, E.; Simmonds, K. J.;
Zivec, M.; Inglis, S. R.; Fishwick, C. W. G.; Gobec, S.; Charlier, P.;
Luxen, A.; Schofield, C. J. Structure guided development of potent
reversibly binding penicillin binding protein inhibitors. ACS Med.
Chem. Lett. 2011, 2, 219−223.
(15) Jayaram, B.; Latha, N.; Jain, T.; Sharma, P.; Ghandimathi, A.;
Pandey, V. S. Sanjeevini: A Comprehensive active site directed lead
design software. Indian J. Chem. 2006, 45A, 21−33.
(16) Latha, N.; Jain, T.; Sharma, P.; Jayaram, B. A free energy based
computational pathway from chemical templates to lead compounds: a
case study of COX-2 inhibitors. J. Biomol. Struct. Dyn. 2004, 6, 791−
804.
(17) Nicholas, R. A.; Krings, S.; Tomberg, J.; Nicola, G.; Davies, C.
Crystal structures of wild-type penicillin-binding protein 5 from
Escherichia coli. J. Biol. Chem. 2003, 278, 52826−52833.
(18) Boteva, A. A.; Kasnykh, O. P. The methods of synthesis,
modification, and biological activity of 4-quinolones (review). Chem.
Hetercycl. Compd. 2009, 45, 757−785.
595
dx.doi.org/10.1021/ml3001006 | ACS Med. Chem. Lett. 2012, 3, 592−595