Isoxazole-3-hydroxamic acid derivatives as peptide deformylase inhibitors and potential antibacterial agents

Article abstract of DOI:10.1016/j.bmcl.2004.09.087

The synthesis and initial in vitro evaluation of isoxazole-3-hydroxamic acid derivatives as a new class of PDF inhibitors are reported. A series of isoxazole-3-hydroxamic acid derivatives has been identified as a new class of small, nonpeptidic inhibitors of peptide deformylase (PDF). The synthesis, enzyme inhibition and preliminary investigation of the binding mode of this potential antibacterial compounds are reported.

Full text of DOI:10.1016/j.bmcl.2004.09.087

Bioorganic & Medicinal Chemistry Letters 14 (2004) 5997–6000
Isoxazole-3-hydroxamic acid derivatives as peptide
deformylase inhibitors and potential antibacterial agents
a,
Patrizia Calı, Lars Nærum, Seema Mukhija and Anders Hjelmencrantz
a
b
a
*
´
aCombio A/S, Vesterbrogade 188, DK-1800 Frederiksberg C, Denmark
bArpida Ltd, Dammstrasse 36, 4142Mu¨nchenstein, Switzerland
Received 2 June 2004; revised 29 August 2004; accepted 30 September 2004
Available online 18 October 2004
Abstract—A series of isoxazole-3-hydroxamic acid derivatives has been identified as a new class of small, nonpeptidic inhibitors of
peptide deformylase (PDF). The synthesis, enzyme inhibition and preliminary investigation of the binding mode of this potential
antibacterial compounds are reported.
Ó 2004 Elsevier Ltd. All rights reserved.
S1'
In recent years, there has been a growing interest in
developing bacterial peptide deformylase (PDF) inhibi-
tors as novel antibiotics. PDF is an iron-containing met-
alloenzyme, which catalyses the removal of the N-formyl
group from the terminal methionine residue of nascent
proteins.1 PDF is essential for both Gram-positive and
Gram-negative bacteria, since deformylation is a neces-
sary step to complete protein biosynthesis and matura-
tion. Due to its wide distribution in bacteria and its
absence in mammalian cells, PDF represents an attrac-
tive target for the discovery of broad spectrum antibac-
terial drugs.2
S3'
P3'
OH
Metal
O
O
O
H
H
X
N
HO
N
N
N
H
H
O
P2'
O
S2'
Generic PDF inhibitor
Actinonin (1)
OH
N
O
O
H
S
R
H
N
HO
N
N
H
N
O
O
N
S
O
Many PDF inhibitors have been reported in recent
years.3,4 Most of the compounds with sufficient potency
and antibacterial activity share a common structure, as
exemplified in Figure 1. Two structural features are con-
stantly recurrent and seem to account for most of the
binding energy; a metal chelating group X, most com-
monly a hydroxamate or N-formylhydroxylamine, and
a n-alkyl (usually n-butyl) residue at P10 mimicking the
methionine side chain and fitting into the deep S10
hydrophobic pocket in the PDF active site. Examples
are the natural occurring antibiotic actinonin (1)5 and
the reverse hydroxamate from British Biotech BB-
83698 (2),6 currently in phase I clinical trials (Fig. 1).
However, since most of the known inhibitors still have
significant peptide characteristics, there are some con-
N
BB-83698 (2)
3
Figure 1. Structure of a generic PDF inhibitor compared to actinonin
(1), to BB-83698 (2) and to the new series of isoxazole-3-hydroxamic
acid derivatives (3).
cerns about their selectivity and in vivo metabolic stabil-
ity. Here we wish to report the synthesis and preliminary
in vitro evaluation of a new series of nonpeptidic PDF
inhibitors having an isoxazole-3-hydroxamic acid as
central core (3). The binding mode of this class of inhib-
itors is investigated by comparison with the crystal
structure of the actinonin–PDF complex.
The general route for the synthesis of the isoxazole-3-
hydroxamic acids (3) is outlined in Scheme 1. Radical
bromination of ethyl-5-methylisoxazole-3-carboxylate
Keywords: Peptide deformylase; Inhibitors; Antibacterial.
*
Corresponding author. Tel.: +45 70222365; fax: +45 70222367;
e-mail: pca@combio.biz
0960-894X/$ - see front matter Ó 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2004.09.087

´
P. Calı et al. / Bioorg. Med. Chem. Lett. 14 (2004) 5997–6000
5998
Table 1. PDF enzyme inhibitory activity of isoxazole-3-hydroxamic
acids 3a–q, 8a–e and 10a compared to actinonin 1
CH3
Br
a
b
EtO
EtO
EtO
O
O
Compds
R
E. coli
S. aureus
N
N
N
34%
65-90%
IC50, (lM)a IC50 (lM)a
O
O
4
5
R
R
S
1
0.0007
20.0
0.002
4.0
2.3
S
3a
3b
3c
3d
3e
3f
4-(CH3)Ph
3-(Cl)Ph
4-(Cl)Ph
c
HOHN
9.8
14.5
6.5
O
O
N
1.3
2.2
65-96%
O
O
3-(OCH3)Ph
2-(i-Pr)Ph
6a-q
3a-q
14.0
31.3
29.5
62.5
4.4
7.6
24.0
4-(OCH3)Ph
2,4,6-(CH3)Ph
2-(Cl)Ph
Scheme 1. Reagents and conditions: (a) NBS, BPO, CCl4, 95°C, 4h,
34%; (b) RSH, K2CO3, DMF, 60°C, 1h; (c) NH2OH, KOH, MeOH,
rt, 2–24h.
3g
3h
3i
37.7
3.8
2.0
4.9
2.0
1.5
9.3
8.7
42.0
12.7
2.2
1.4
1.7
0.8
6.5
3.4
0.9
3,4-(Cl)Ph
2-(Br)Ph
3j
60.0
7.0
3k
3l
3-(Br)Ph
4-(F)Ph
(4) was obtained by refluxing with N-bromosuccinimide
and catalytic amounts (10%) of benzoylperoxide as rad-
ical initiator.7 Selective bromination of the benzylic
position over the isoxazole ring could be achieved by
using equimolar amount of NBS. The intermediate ben-
zylic bromide 5 was alkylated with an array of aromatic,
heteroaromatic and benzylic thiols, in the presence of
K2CO3. All reactions were completed after 1h affording
the desired thioethers 6a–q in 65–96% yield after purifi-
cation by column chromatography. The reaction of the
ethyl esters 6a–q with a methanolic solution of hydrox-
ylamine hydrochloride and KOH yielded the desired
hydroxamic acids 3a–q.8 All final products were purified
by preparative HPLC and characterised by 1H, 13C
NMR and LCMS analysis.
20.7
25.3
22.0
20.3
3.4
3m
3n
3o
3p
3q
8a
8b
8c
8d
8e
10a
2-(CF3)Ph
4-(CF3)Ph
4-(NHCOCH3)Ph
Benzothiophen-2-ylmethyl
3-(CF3)-benzyl
4-(CH3)Ph
3-(Cl)Ph
7.6
11.5
6.8
4-(Cl)Ph
7.5
3-(OCH3)Ph
2-(i-Pr)Ph
4-(CH3)Ph
26.3
10.8
12.0
a Values are means of three experiments.
critical for good inhibitory activity,11 this functionality
was conserved in all derivatives synthesised and tested.12
In spite of the high similarity of the binding sites gener-
ally reported,13 significant differences in the inhibitory
activity were observed with S. aureus and E. coli PDF
enzymes, the former being generally more sensitive to
changes in the substitution pattern of the inhibitors.
A few oxidised analogues (8a–e and 10a) were synthe-
sised as represented in Scheme 2. The reaction of the thi-
oethers 6a–e with m-chloroperbenzoic acid yielded the
sulfonyl derivatives 7a–e.9 When intermediate 6a was re-
acted with NaBO3 at rt for 1h, the mono-oxidised prod-
ucts 9a was selectively obtained. The synthesis of the
final hydroxamic acids 8a–e and 10a was performed as
previously described in a methanolic solution of hydrox-
ylamine hydrochloride and KOH.
In order to investigate the binding mode of our inhibi-
tors and rationalise the results, compound 3b was mini-
mised and positioned into the crystal structure of the
PDF–actinonin complex (PDB access code, 1G2A).4f
The result of our molecular modelling study is illus-
trated in Figure 2.14 The hydroxamic acid moiety of
3b can be positioned favourably for chelating the Ni
atom in the active site, in a similar fashion to actinonin.
The isoxazole ring orientation allowed for a favourable
H-bond between the oxygen and the backbone NH of
Ile44, similarly to the P10 carbonyl in actinonin. The dis-
tances between the actinonin-carbonyl and isoxazole-
The final hydroxamic acids were evaluated for their in
vitro inhibitory activity on Escherichia coli and Sta-
phylococcus aureus PDF enzymes.10 The activities of
the compounds compared to actinonin 1 are summa-
rised in Table 1. Since the bidentate, metal-chelating
hydroxamic acid moiety was previously shown to be
R
R
S
O
S
O
O
O
˚
oxygen with the NH of Ile44 are 1.8 and 2.2A, respec-
a
EtO
b
HOHN
O
O
6a-e
tively. The best fit was obtained when 3b assumed a bent
conformation, positioning the thioaromatic residue and
the meta-chloro substituent into the hydrophobic S10
pocket. As the n-pentyl group in the actinonin–PDF
complex, the thioaromatic group is engaged in extensive
hydrophobic interactions in the S10 pocket with enzyme
side chains. A similar binding mode was also observed
with the potent PDF inhibitors by Hoffmann-La Roche
(magenta in Figure 2), 2,2-dioxo-1,4-dihydro-
benzo(1,2,6)thiadiazinyl hydroxamic acid, where
small lipophilic groups (F, Cl, Br) were found to ideally
fit into the S10 pocket and increase the activity.4b Since a
N
N
83-97%
70-84%
O
O
7a-e
8a-e
R
R
S
S
O
O
EtO
HOHN
d
c
O
N
O
N
79%
87%
O
O
9a
10a
Scheme 2. Reagents and conditions: (a) m-CPBA, CH2Cl2, rt, 1–3h;
(b) NH2OH, KOH, MeOH, rt, 2–24h; (c) NaBO34H2O, AcOH, rt; 1h;
(d) NH2OH, KOH, MeOH, rt, 2h.

´
P. Calı et al. / Bioorg. Med. Chem. Lett. 14 (2004) 5997–6000
5999
more active with S. aureus PDF. ortho-Substitution with
small lipophilic groups (Cl, Br, CF3) and alkyl chains
(i-propyl) led to significantly less active inhibitors of E.
coli enzyme (e.g., 3e, 3h, 3j, 3m), though the loss of activity
appeared less pronounced in the case of S. aureus PDF.
para-Substitution (3a, 3c, 3f, 3l, 3n, 3o) generally re-
sulted in poor inhibition of E. coli PDF. More dramatic
is the effect of para-substitution with the S. aureus PDF
enzyme, particularly with bulky and polar substituents.
Activity comparable to the corresponding meta-substi-
tuted series, was observed with S. aureus PDF enzyme
in the case of small lipophilic substituents, like fluorine
(3l) or chlorine (3c), suggesting that these groups can
still be accommodated at the end of the S10 pocket with-
out significant distortion from the optimal interaction.
Higher IC50 values were observed with more bulky sub-
stituents like methyl (3a), methoxy (3f), N-acetyl groups
(3o). Since the methionine pocket has proven to be very
sensitive to the length of the hydrophobic chain, it can
be speculated that these derivatives bind the PDF en-
zyme with an extended conformation, locating the thio-
ether side chain away from the S10 pocket. Oxidation of
the thioether of derivatives 3a–e to sulfoxide (10a) and
sulfone (8a–e) only slightly improved the potency of
the compounds, resulting in IC50 values 1.5–3 times lower
than the corresponding un-oxidised compounds. Since
for derivatives 3a–e the sulfone oxygen is placed in close
proximity to the P20 carbonyl of actinonin, a similar
interaction with the backbone NH of Gly89 could be
envisaged and account for the observed difference in
activity.
Figure 2. Molecular model predicting the binding mode of inhibitor 3b
in comparison with the actinonin–PDF complex. Actinonin is shown
in gold, inhibitor 3b is in white and the Hoffmann-La Roche
inhibitor4b is in magenta. The surface of the enzyme (S10 pocket) is
represented in gold and the Ni2+ ion is represented as a yellow ball.
The backbone NH of Ile44 is also highlighted.
tight binding in the S10 pocket has proven crucial in
order to obtain potent inhibitors, it can be concluded that
the nature and the position of the thioaromatic substit-
uents will play a central role in determining the affinity.
As suggested from the modelling study, neither the P20
nor the P30 regions are addressed by compound 3b. This
could account for the much lower activity observed in
the isoxazole derivatives series compared to the refer-
ence actinonin 1.
In addition, the whole-cell activity was assessed for the
most active compounds in the series (Table 2).15 None
of the inhibitors tested showed any antibacterial activity
against wild type or resistant strains of both E. coli and
S. aureus, as observed for actinonin 1. However, while
actinonin displayed a strong antibacterial activity in mu-
tant of E. coli EC-DC216 and EC-AH547, respectively, a
hypersensitive and a strain with reduced activity of ef-
flux pumps, the isoxazole derivatives showed only poor
MIC values.
From the modelling study we expected that small lipo-
philic groups (Cl, Br, CF3) in the meta-position will be
filling the methionine pocket optimally. In accordance
with our expectations, meta-substitution (e.g., 3b, 3d,
3i, 3k) appeared generally favoured over ortho- and
para- substitution, both with E. coli and S. aureus
PDF enzymes. Derivatives 3i, with a 3,4-dichloro sub-
stituent, was one of the most potent inhibitor of E. coli
PDF enzyme with an IC50 value of 4.4lM, and 2 times
These results suggest that the poor antibacterial activity
in Gram negatives derives from a combination of poor
penetration of the bacterial cell wall and recognition
Table 2. In vitro antibacterial activity of selected isoxazole-3-hydroxamic acids
Compds
R
MIC (lg/mL)a
SA-ATCC-25923b
SA-101b
EC-ATCC-25922c
EC-DC2c
EC-AH547d
1
64
>128
>128
>128
128
128
>128
>128
>128
128
128
>128
128
0.5
32
0.5
64
3b
3c
3d
3i
3-(Cl)Ph
4-(Cl)Ph
32
64
3-(OCH3)Ph
3,4-(Cl)Ph
3-(Br)Ph
>128
>128
>128
128
128
32
128
16
3k
3p
128
128
>128
128
64
32
64
64
Benzothiophen-2-ylmethyl
a MIC = minimum inhibitory concentration.15
b SA-ATCC-25923 and SA-101 are, respectively, the wild type and a multi resistant strain of S. aureus.
c EC-ATCC-25922 and EC-DC2 are, respectively, the wild type and a hyperpermeable strain of E. coli.
d EC-AH547 is a tolC mutant strain of E. coli with reduced activity of efflux pumps.

´
P. Calı et al. / Bioorg. Med. Chem. Lett. 14 (2004) 5997–6000
6000
Cramer, J. A.; Margolis, P. S.; Wang, W.; Koehn, J.; Wu,
C.; Lopez, S.; Withers, G., III; Gu, H.; Dunn, E.;
Kulathila, R.; Pan, S. H.; Porter, W. L.; Jacobs, J.; Trias,
J.; Patel, D. V.; Weidmann, B.; White, R. J.; Yuan, Z.
Antimicrob. Agents Chemother. 2002, 46, 2752–2764; (e)
Takayama, W.; Shirasaki, Y.; Sakai, Y.; Nakajima, E.;
Fujita, S.; Sakamoto-Mizukani, K.; Inoue, J. Bioorg. Med.
Chem. Lett. 2003, 13, 3273–3276; (f) Clemens, J. M.;
Beckett, R. P.; Brown, A.; Catlin, G.; Lobell, M.; Palan,
S.; Thomas, W.; Whittaker, M.; Wood, S.; Salama, S.;
Baker, P. J.; Rodgers, H. F.; Barynin, V.; Rice, D. W.;
Hunter, M. G. Antimicrob. Agents Chemother. 2001, 45,
563–570; (g) East, S. P.; Beckett, R. P.; Brookings, D. C.;
Clemens, J. M.; Doel, S.; Keavey, K.; Pain, G.; Smith, H.
K.; Thomas, W.; Thompson, A. J.; Todd, R. S.; Whit-
taker, M. Bioorg. Med. Chem. Lett. 2004, 14, 59–62; (h)
Hu, X.; Nguyen, K. T.; Verlinde, C. L. M. J.; Hol, W. G.
J.; Pei, D. J. Med. Chem. 2003, 46, 3771–3774.
by an efficient efflux system of the cell, similarly to com-
pound 1. However, in the case of the isoxazole-3-
hydroxamic acids the lack of antibacterial activity might
also be related to the only moderate inhibitory activity
of PDF enzyme. In fact, more potent inhibitors are gen-
erally required in order to observe significant antibacte-
rial activity.
In conclusion, we have identified a new class of nonpept-
idic PDF inhibitors. According to our initial molecular
model, three key interactions occur between the inhibi-
tors and the enzyme binding site: bidentate coordination
to the metal centre, H-bond with Ile44 and hydrophobic
interaction at P10 position. The compounds studied dis-
play inhibitory activity on the enzymes from E. coli and
S. aureus, though they exhibit weak antibacterial activ-
ity only in hypersensitive or efflux-defective Gram-nega-
tive strains. As suggested by the molecular model,
neither the P20 nor the P30 regions seem addressed by
the studied isoxazole derivatives. Therefore, the intro-
duction of interactions in the P20 and P30 regions might
improve both potency and antibacterial activity.
5. Chen, D. Z.; Patel, D. V.; Hackbarth, C. J.; Wang, W.;
Dreyer, J.; Young, D. C.; Margolis, P. S.; Wu, C.; Ni, Z.
J.; Trias, J.; White, R. J.; Yuan, Z. Biochemistry 2000, 39,
1256–1262.
6. (a) Lofland, D.; Difuntorum, S.; Waller, A.; Clements, J.
M.; Weaver, M. K.; Karlowsky, J. A.; Johnson, K.
Antimicrob. Chemother. 2004, 53, 664–668; (b) Azoulay-
´
Dupuis, E.; Mohler, J.; Bedos, J. P. Antimicrob. Agents
Chemother. 2004, 48, 80–85.
Acknowledgements
7. Tashiro, M.; Yamato, T. J. Org. Chem. 1985, 50, 2939–
2942.
8. Hauser, C. R.; Renfrow, W. B., Jr. Org. Synth. 1939, 19,
15–17.
The authors thank all the colleagues who have contrib-
uted to the project and Dr. Peter Schneider for helpful
advice.
´
´
9. Guzman, A.; Romero, M.; Talamas, F. X. J. Org. Chem.
1996, 61, 2470–2483.
References and notes
10. PDF inhibition assay, optimised to microtiter plate
format, was performed as previously described by Groche,
D.; Becker, A.; Schlichting, I.; Kabsch, W.; Schultz, S.;
Wagner, A. F. Biochem. Biophys. Res. Commun. 1998, 246,
342–346.
1. (a) Adams, J. M. J. Mol. Biol. 1968, 5, 571–589; (b)
Adams, J. M.; Capecchi, M. Proc. Natl. Acad. Sci. U.S.A.
1966, 55, 147–155.
2. (a) Yuan, Z.; Trias, J.; White, R. J. Drug Discovery Today
2001, 6, 954–961; (b) Clemens, J. M.; Ayscough, A. P.;
Keavey, K.; East, S. P. Curr. Med. Chem. Anti-infective
Agents 2002, 1, 239–249; (c) Pei, D. Emerging Ther.
Targets 2001, 5, 23–40.
3. For examples of thiol inhibitors, see: (a) Meinnel, T.;
Patiny, L.; Ragusa, S.; Blanquet, S. Biochemistry 1999, 38,
4287–4295; (b) Huntington, K. M.; Yi, T.; Wei, Y.; Pei, D.
Biochemistry 2000, 39, 4543–4551; (c) Wei, Y.; Yi, T. Y.;
Huntingtom, K. M.; Chaudhury, C.; Pei, D. J. Comb.
Chem. 2000, 2, 650–657.
11. Smith, H. K.; Beckett, R. P.; Clemens, J. M.; Doel, S.;
East, S. P.; Launchbury, S. B.; Pratt, L. M.; Spavold, Z.
M.; Thomas, W.; Todd, R. S.; Whittaker, M. Bioorg. Med.
Chem. Lett. 2002, 12, 3595–3599.
12. The corresponding isoxazole-3-carboxylic acids of com-
pounds 3a–q were also synthesised and resulted in no
inhibitory activity with PDF enzyme (data not reported).
13. For X-ray of PDF enzymes and comparison of binding
sites in different bacterial strains, see: (a) Guilloteau, J. P.;
Mathieu, M.; Giglione, C.; Blanc, V.; Dupuy, A.;
Chevrier, M.; Gil, P.; Famechon, A.; Meinnel, T.; Mikol,
V. J. Mol. Biol. 2002, 320, 951–962; (b) Kreusch, A.;
Spraggon, G.; Lee, C. C.; Klock, H.; McMullan, D.; Ng,
K.; Shin, T.; Vincent, J.; Warner, I.; Ericson, C.; Lesley, S.
A. J. Mol. Biol. 2003, 330, 309–321.
4. For examples of hydroxamate and reverse hydroxamate
inhibitors, see: (a) Apfel, C.; Banner, D. W.; Bur, D.;
Dietz, M.; Hirata, T.; Hubschwerlen, C.; Locher, H.;
´
Page, M. G. P.; Pirson, W.; Rosse, G.; Specklin, J. L. J.
Med. Chem. 2000, 43, 2324–2331; (b) Apfel, C.; Banner,
D. W.; Bur, D.; Dietz, M.; Hubschwerlen, C.; Locher, H.;
Marlin, F.; Masciadri, R.; Pirson, W.; Stalder, H. J. Med.
Chem. 2001, 44, 1847–1852; (c) Thorarensen, A.; Douglas,
M. R., Jr.; Rohrer, D. C.; Vosters, A. F.; Yem, A. W.;
Marshall, V. D.; Lynn, J. C.; Bohanon, M. J.; Tomich, P.
K.; Zurenko, G. E.; Sweeney, M. T.; Jensen, R. M.;
Nielsen, J. W.; Seest, E. P.; Dolak, L. A. Bioorg. Med.
Chem. Lett. 2002, 11, 1355–1358; (d) Hackbart, C. J.;
Chen, D. Z.; Lewis, J. G.; Clark, K.; Mangold, J. B.;
14. Modelling study was performed using a combined FlexX
and Maximin approach within Sybyl 6.8, Tripos Inc.,
USA.
15. Antibacterial activities were determined in liquid culture
according to NCCLS standard conditions: Methods for
Dilution Susceptibility Tests for Bacteria That Grow
Aerobically; Approved Standard-Fifth Edition, 2000
NCCLS M7-A5, Vol. 20, No. 2.
16. Richmond, M. H.; Clark, D. C.; Wotton, S. Antimicrob.
Agents Chemother. 1976, 10, 215–218.