A detailed study of antibacterial 3-acyltetramic acids and 3-acylpiperidine-2,4-diones

Article abstract of DOI:10.1002/cmdc.201402093

Inspired by the core fragment of antibacterial natural products such as streptolydigin, 3-acyltetramic acids and 3-acylpiperidine-2,4-diones have been synthesised from the core heterocycle by direct acylation with the substituted carboxylic acids using a strategy which permits ready access to a structurally diverse compound library. The antibacterial activity of these systems has been established against a panel of Gram-positive and Gram-negative bacteria, with activity mostly against the former, which in some cases is very potent. Data consistent with modes of action against undecaprenylpyrophosphate synthase (UPPS) and/or RNA polymerase (RNAP) for a small subset of the library has been obtained. The most active compounds have been shown to exhibit binding at known binding sites of streptolydigin and myxopyronin at UPPS and RNAP. These systems offer potential for their antibacterial activity, and further demonstrate the use of natural products as biologically validated starting points for drug discovery.

Full text of DOI:10.1002/cmdc.201402093

CHEMMEDCHEM
FULL PAPERS
DOI: 10.1002/cmdc.201402093
A Detailed Study of Antibacterial 3-Acyltetramic Acids and
3-Acylpiperidine-2,4-diones
Yong-Chul Jeong,[a] Zsolt Bikadi,[b] Eszter Hazai,[b] and Mark G. Moloney*[a]
Inspired by the core fragment of antibacterial natural products
such as streptolydigin, 3-acyltetramic acids and 3-acylpiperi-
dine-2,4-diones have been synthesised from the core heterocy-
cle by direct acylation with the substituted carboxylic acids
using a strategy which permits ready access to a structurally
diverse compound library. The antibacterial activity of these
systems has been established against a panel of Gram-positive
and Gram-negative bacteria, with activity mostly against the
former, which in some cases is very potent. Data consistent
with modes of action against undecaprenylpyrophosphate syn-
thase (UPPS) and/or RNA polymerase (RNAP) for a small subset
of the library has been obtained. The most active compounds
have been shown to exhibit binding at known binding sites of
streptolydigin and myxopyronin at UPPS and RNAP. These sys-
tems offer potential for their antibacterial activity, and further
demonstrate the use of natural products as biologically validat-
ed starting points for drug discovery.
Introduction
The emergence of persistent
nosocomial and community ac-
quired infections, caused partic-
ularly by bacterial strains resist-
ant to current clinically effective
drugs such as methicillin-resist-
ant
Staphylococcus
aureus
(MRSA),
vancomycin-resistant
S. aureus (VRSA) and Enterococ-
cus (VRE) and multidrug resist-
ant Streptococcus pneumoniae
(MDRSP), requires the urgent
development of new antibacte-
rial drugs.[1,2] However, a crisis
era appears to have emerged in
which the speed of new anti-
Figure 1. Some naturally occurring and some synthetic 3-acyltetramic acids.
bacterial development is slower
than the emergence of resist-
ance.[3] Approaches using new or hybridised forms of existing
antibacterials, and combination therapies, have helped to re-
dress the balance, but the discovery of new classes of antibac-
terials is underexploited.[4–9] The identification of antibacterials
with novel modes of action (MOAs), or which have new mech-
anisms at existing targets, is of particular interest, not least be-
cause it might be expected to slow the emergence of resist-
ance development. Antibacterial drug development is now
perceived to be expensive and of high risk, at least compared
to other therapeutic areas, and few new antibacterial families
have been launched since 1970, linezolid and the topical agent
retapamulin being examples. However, small molecule natural
products and their analogues derived from diverted synthesis
provide major inspiration in drug discovery, and this is espe-
cially true for antibacterial agents.[10–12] Of interest are 3-acyl-
tetramic acids (3ATs), found as the core structure in many natu-
ral products which possess intrinsic antibacterial activity, in-
cluding for example magnesidin A (1a), reutericyclin (1b),
quorum sensing product (1c), melophlin G (1d) and streptoly-
digin (1e) (Figure 1).[13–19] In a search for more effective com-
pounds based on these natural products, the antibacterial ac-
tivity of some unnatural 3ATs has been examined.[20–27] We
have reported antibacterial bicyclic system 3AT 1 f, along with
[a] Dr. Y.-C. Jeong, Prof. M. G. Moloney
Chemistry Research Laboratory, University of Oxford
Mansfield Rd, Oxford, OX1 3TA (UK)
E-mail: mark.moloney@chem.ox.ac.uk
[b] Dr. Z. Bikadi, Dr. E. Hazai
Virtua Drug
4C Csalogany, Budapest, 1015 (Hungary)
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cmdc.201402093.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 1826 – 1837 1826

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
related 3-carboxamide tetramic acids, which exhibit bac-
terial RNA polymerase (RNAP) and undecaprenyl pyro-
phosphate synthase (UPPS) inhibitory activity.[28] The
characteristic core in this family, the N-substituted 3AT
system, has also been mimicked by the a ring expanded
3-carboxamide piperidine-2,4-dione 1g, which shows
antibacterial activity.[29] Since the structure–activity rela-
tionships (SARs) of such unnatural N-acyl 3ATs and
3-acylpiperidine-2,4-diones (3APs) have not been stud-
ied in great detail, we decided to systematically explore
the antibacterial activity of N-unsubstituted, N-alkyl and
N-acyl 3ATs and 3APs. We have established facile syn-
thetic methodologies both for core tetramic acid tem-
plates, and for their acylation at the 3-position, routes
which provide direct access to 3AT compound libraries
bearing a large range of diversity.[30,31] In this paper, we
report the antibacterial activity and SARs of 3ATs and
3APs, with three points of diversity, achieved by modi-
fied substitution at N(1), C(5) and the 3-acyl position.
Results and Discussion
Synthesis
In order to access a wide range of diversity at the N(1)
and C(5) positions, 17 tetramate and piperidinone tem-
plates were prepared (Figure 2), seven of which (tem-
plates 2a–c, 3b–d and 5a) have been reported previ-
ously; these were predominantly monocyclic ring sys-
tems, with the exception of bicyclic derivative 4.[30] Tet-
ramic acids 3a and 5b were accessed from 8a, itself
prepared from glycine ethyl ester hydrochloride, and
commercially
available
N-benzoyl-dl-leucine
8b
(Scheme 1A), by coupling with Meldrum’s acid followed
by thermal rearrangement. Tetramic acid 4, synthesised
from (S)-9 using the same method (Scheme 1B), was
Scheme 1. Synthesis of tetramic acids and piperidine-2,4-diones. Reagents and
conditions: a) Meldrum’s acid (1.1 equiv), DCC (1.1 equiv), DMAP (1.1 equiv),
CH2Cl2, RT b) EtOAc, reflux; c) acid chloride (1.1 equiv), Et3N (2.2 equiv), CH2Cl2,
08C–RT; d) LiOH (3.0 equiv), H2O/CH3OH/THF (2:1:1), 08C–RT. DCC=N,N’-dicyclo-
hexylcarbodiimide, DMAP=4-(dimethylamino)pyridine.
found to be optically active ([a]D22 =ꢀ15.7 cm3 gꢀ1 dmꢀ1, c=0.3
in MeOH), although the enantiomeric excess was not deter-
mined. Tetramic acids 6a–e were prepared as shown in
Scheme 1C; carboxylates 12b,d,e (either made by N-acylation
of l-methionine methyl ester hydrochloride 10 with the corre-
sponding acid chlorides to give 11a–c, followed by hydrolysis,
or in the case of N-acetyl-l-methionine (S)-12a and N-benzoyl-
dl-methionine (ꢁ)-12c were commercially available) were cou-
pled with Meldrum’s acid and then cyclised to give tetramic
acids 6a–e. Noteworthy was that tetramic acids 5a[30,32] and
6a,b,d,e, synthesised from the optically active parent carboxyl-
ic acids, were optically inactive, indicating the high lability of
the stereocentre on the C(5) position under these conditions.
Application of this methodology to the synthesis of piperidine-
Figure 2. 2,4-Dione templates.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 1826 – 1837 1827

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
2,4-diones 7a,b[33] gave yields which were similar to those of
the five-membered tetramic acids (Scheme 1D).
acter, leading to a variety of possible keto–enol structures, and
this is discussed in more detail below. Specifically, 3-acylpiperi-
dine-2,4-diones (3APs) 19a–f (Figure 4), 3-acyltetrameric acids
(3ATs) 20–25 (Figure 5 and Table 1) and 26–31 (Figure 6 and 7),
were accessed in this way, some of which have been reported
previously.[31]
For further derivatisation of the heterocyclic core of the tet-
ramate or piperidine 2,4-dione systems 2–7, 3-acylation with
the desired carboxylic acid was found to be the most conven-
ient (Scheme 2; see the Supporting Information for full details),
Tautomeric behaviour
3ATs 18 and 3APs 19 both exist as an equilibrium
between four tautomeric forms A–D in solution
(Figure 3).[31,34–36] For N-unsubstituted and N-alkyl
3ATs, the predominant forms were the CD external
tautomeric pair (about 70–99%), with internal tauto-
mer D being the main contributor, and the 5-substi-
tuted N-alkyl 3ATs reported herein showed similar
behaviour (see Figure S1 for HMBC NMR spectra of
representative analogues and the Supporting Infor-
mation for the ratio of external tautomers of all ana-
logues). By contrast, N-acyl 3ATs generally existed as
a similar ratio of the two external tautomeric pairs
(AB and CD), in which A and B equally contributed
to the external tautomer AB while D was the main
contributor to external tautomer CD.[31] However, of
Scheme 2. Synthesis of 3-acyltetramic acids and 3-acylpiperidine-2,4-diones. Reagents
and conditions: a) R3CO2H (1.1 equiv), DCC (1.1 equiv), DMAP (0.1 equiv), CH2Cl2, RT;
b) R3COCl (1.1 eq), Et3N (1.2 eq), CH2Cl2, RT; c) (CH3)2C(OH)CN (0.5 equiv), Et3N (2.0 equiv),
CH3CN, RT; d) DMAP (1.3 equiv), CH2Cl2, RT; e) R3CO2H (1.1 equiv), DCC (1.1 equiv), DMAP
(1.3 equiv), CH2Cl2, RT; f) trifluoroacetic acid, CH2Cl2, RT.
greater interest was that the tautomeric behavior of
3APs was different, and in this case, the major exter-
nal tautomer was assigned as the AB tautomer, with
B being the major contributor, while for the minor
although acylation of 5-unsubstituted tetramic acids via O-acy-
lation followed by acyl migration also provided indirect but ef-
fective access to modified tetramates.[31] Carboxylic acids for in-
troduction as the C-3 side chain were chosen on the basis of
systematic change of their structure, along with their hydro-
phobic/hydrophilic and hydrogen-bond donor/acceptor prop-
erties, and were generally commercially available; acids 16a–
c and 17 were easily prepared as shown in Scheme 3. This
gave a large range of tricarbonyl systems, typified by struc-
tures 18 and 19 (Figure 3). An immediate complication in
these systems that was encountered is the existence of exten-
sive tautomeric behaviour resulting from their tricarbonyl char-
external tautomer CD, D was the main contributor (Figure S1).
This tautomeric behaviour is supported by calculation of
ground state energies of simplified analogues 18 (n=0, R1 =
CH3C(O), R2 =H, R3 =Me) and 19 (n=1, R1 =CH3C(O), R2 =H,
R3 =Me) (Table 2). Unlike the tautomeric behaviour of
3ATs,[31,34–38] this is the first study of such behaviour for 3APs.
Scheme 3. Synthesis of carboxylic acid derivatives. Reagents and conditions:
a) RC(O)Cl (1.1 equiv), Et3N (2.2 equiv), CH2Cl2, 08C–RT; b) LiOH (3.0 equiv),
H2O/CH3OH/THF (2:1:1), 08C–RT; c) 2-amino-4,5-dimethylthiazole hydrochlo-
ride (1.1 equiv), Et3N (2.2 equiv), CH2Cl2, 08C–RT.
Figure 3. Tautomerisation in 3-acyltetramic acids and 3-acylpiperidine-2,4-
diones.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 1826 – 1837 1828

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
Figure 4. 3-Acylpiperidine-2,4-diones 19a–f (see Table 1 and 3).
Thus, for five-membered 18, the ground state energy of tauto-
mers B and D were similar (energy difference DE=0.17 kcal
molꢀ1), confirming the experimental ratio for external tauto-
mers AB and CD. In addition, the energy difference between
internal tautomers A and B (DE=1.42 kcalmolꢀ1) was smaller
than for internal tautomers C and D (DE=5.62 kcalmolꢀ1), sup-
porting the experimental ratio in each external tautomer (simi-
lar contribution of tautomers A and B, and main contribution
of tautomer D). Even though the energy difference was small
(DE=0.92 kcalmolꢀ1) in the calculation of six-membered 19,
the ground state energy of tautomers A and B, the major ex-
ternal tautomers as shown by NMR spectroscopy, were found
to be more stable than tautomers C and D. In addition, tauto-
mer B was more stable than tautomer A (DE=0.51 kcalmolꢀ1),
as observed by NMR spectroscopy.
Antibacterial activity
The activity of 104 different derivatives 19 and 20–31 against
clinically relevant bacteria (eight strains of Gram-positive and
five strains of Gram-negative bacteria) was screened, using
linezolid and ciprofloxacin as standards (Table 1 and 3). Al-
though it was found that antibacterial activity depended both
on the structural diversity of the substrate as well as the bacte-
rial strain, the ring size of the templates (five-membered tetra-
mic acids versus six-membered piperidine-2,4-diones) was not
found to be a critical factor for the intrinsic antibacterial activi-
ty, as has been reported for 3-carboxamides.[29] Significantly,
3-acyl analogues were more effective against Gram-positive
than Gram-negative bacteria. Amongst Gram-negative bacteria,
Figure 5. 3-Acyltetrameric acids 20–25 (see Table 1 and 3).
usually weaker than that against the efflux-negative strain (H4),
consistent with effective transport of 3-acyl analogues by
membrane efflux pumps. However, the 3-acyl analogues not
only showed excellent antibacterial activity against Gram-posi-
tive bacteria (MIC as low as ꢃ0.06 mgmLꢀ1), depending on
their substituents, but also retained a good level of activity
against all of the virulent Gram-positive strains, compared to
nonresistant S. aureus (S4). By contrast, the activity of ciproflox-
acin against MRSA and VRE decreased more than 50-fold for
these organisms. The SAR of the 3-acyl analogues reported in
this study are therefore similar to the bicyclic 3ATs described
previously.[28]
no 3-acyl analogues exhibited activity (MICꢂ32 mgmLꢀ1
)
against either Pseudomonas aeruginosa or efflux-positive and
-negative Escherichia coli strains, indicating that the inactivity is
likely to be a result of cell impermeability. In addition, the ac-
tivity against efflux-positive Haemophilus influenzae (H3) was
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 1826 – 1837 1829

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
The SAR of analogues
Table 1. In vitro antibiotic activity (MIC, mgmLꢀ1) of 3-acyl analogues 20–25 (for identity of templates, see Figure 2).[a–d]
possessing
n-alkyl
groups pendant on the
3-acyl moiety is shown
in Table 1. Amongst
these analogues, N-tert-
butoxycarbonyl (N-Boc)
derivatives 20i–l were
inactive against all bac-
terial strains, while the
activity of other ana-
logues was found to
critically depend on
chain length.[28,17] How-
ever, the optimal chain
length varied with the
identity of the individual
templates, and is proba-
bly related to lipophilici-
ty as described below
(see Table S1 in Support-
ing Information for the
Compd Template n+1
Antibacterial strains [MIC, mgmLꢀ1
]
S1
S26
S4
S2
E1
E2
P1
P9
P9B
H3
H4
[c]
[c]
[c]
[c]
[c]
[c]
[c]
[c]
[c]
[c]
[c]
[c]
[c]
[c]
[c]
[c]
[c]
[c]
[c]
[c]
[c]
[c]
[c]
[c]
[c]
[c]
[c]
[c]
[c]
[c]
[c]
[c]
20a
20b
20c
20d
20e
20 f
20g
20h
20i
2a
2b
2c
6
9
11
13
6
>64
8
>64[d]
16[d]
1[d]
–
–
–
–
–
–
–
–
>64
16
1
>64
16
>64
16
>64
64
64
–
–
–
–
–
–
–
–
–
>64
>64
>64
>64
4
>64
>64
>64
–
–
–
–
–
–
–
–
–
–
–
–
–
–
–
1
8
4
4
0.5
1
1[d]
4
4
0.5
64
4[d]
4
2
2
8
[c,d]
9
–
32
>64
>32
>64
>64[d]
>64[d]
8
4
0.5
16
0.12
0.12
8
8
11
13
1
32
16
>64
>32
>64
>64
>64
4
>64
>32
>64
>64
>64
2
>64
>32
>64
>64
>64
1
>64
16
>32
>64
–
>32
>64
>64
>64
2
>32
>64
>64
>64
1
>32
>64
–
>32 >32 >32
>64 >64 >64
20j
9
[c]
[c]
[c]
[c]
[c]
[c]
20k
20l
11
13
9
11
13
4
–
–
4
2
1
8
4
–
–
4
8
8
2
>64
>64
8
–
–
[c]
[c]
–
–
1
21a
21b
21c
21d
21e
21 f
21g
21h
21i
3a
3b
8
2
0.25
0.25
32
1
8
64
0.5
2
32
2
2
1
4
1
0.25
0.5
<0.06
0.12
4
2
16
32
8
8[d]
64[d]
2
1
32
4
2
16
2
<0.06
16
2
0.12 <0.06
8
1
>64
4
4
1
32
8
4
1
–
–
calculated
physical
6
8
2
[c]
[c]
[c]
[c]
[c]
[c]
properties). For exam-
ple, amongst 20a–d
(derived from 2a), unde-
canoyl 20c was the
most active while hexyl
20a was inactive, while
amongst 20e–h (de-
rived from 2b), hexyl
20e was the most
active and undecanoyl
20g was only mildly
active. Amongst 21a–g,
11
13
9
11
8
10
1
13
1
3
5
5
7
10
11
12
5
7
8
10
1
5
–
–
2
8
64
2
4
32
4
8
8
1
4
32
2
0.12
1
1
0.5
16
2
2
2
0.5
1
4
>64
>64
>64
>64 >64
32
32
–
–
[c]
[c]
3c
3d
4
0.25
1
16
8
4
32
8
21j
32
8
4
8
4
21k
22a
22b
23a
23b
23c
23d
23e
24a
24b
24c
24d
24e
24 f
24g
24h
24i
1
0.5
>32
4
>32
16
16
0.5
0.5
32
16
8
>32
4
>32
4
>32
16
8
>32
>32
16
2
>32
>32 >32 >32
64 >64 >64
4
4
2
1
5a
8
8
0.25
0.5
16
4
8
4
1
2
16
8
2
64
>64
>64
32
>64
8
4
2
<0.06
0.25
16
0.25
0.5
32
0.25 <0.06
0.25
16
0.25
0.12
[c]
0.12 <0.06 <0.06
–
5b
6a
16
8
16
4
16
2
8
4
4
4
16
4
0.5
0.25
0.5
2
0.5
8
4
1
2
2
0.5
0.5
0.5
0.25
2
4
4
1
8
2
1
<0.06 <0.06 <0.06 <0.06
0.5
2
2
each with
a different
2
4
0.12
2
0.25 <0.06 <0.06
2
4
1
4
2
>64
32
16
8
4
6b
6c
6d
2
2
8
2
2
2
chain length on the
N-acyl group, undecano-
yl 21b and hexyl 21e
were the most active in
each template system.
Acetyl 22a and 24h
were inactive while
acetyl 23a, the racemic
0.5
8
0.25
8
0.5
4
0.12
1
0.12
1
0.25
2
8
2
2
1
2
2
2
0.5
0.25
>64
>64
>64
1
>64
>64
>64
>64
>64
>64
>64 >64 >64
0.25
0.5
0.5
4
2
2
0.5
2
8
4
2
0.5
0.5
0.5
0.5
8
0.5
0.25
0.5
8
0.25
0.25
0.25
2
0.25
0.12
0.25
1
0.25
0.12
0.12
1
0.25
0.5
0.25
1
2
4
2
4
4
4
1
8
8
8
0.5
1
8
>64
>64
8
16
16
2
2
24j
7
1
1
8
4
4
1
4
8
0.5
0.5
8
24k
24l
9
7
9
9
11
13
9
64
2
6e
7a
24m
25a
25b
25c
25d
25e
linezolid
2
2
2
4
0.5
2
2
2
0.25
2
0.5
4
0.5
1
0.5
0.5
0.25
1
4
8
4
form
of
reutericy-
clin,[28,32] was active.
0.5
1
4
8
>64
64
>64
16
32
8
The antibacterial ac-
tivity of a further 56 an-
alogues, with a large va-
riety of 3-acyl groups,
7b
4
4
4
2
2
11
4
2
0.5
1
2
4
2
1
2
1
2
0.5
1
0.5
0.5
1
8
4
ciprofloxacin
0.12
0.12
16
1
32
1
0.5 ꢃ0.06
[a] Abbreviations: S1=S. aureus 1, ATCC13709 in vivo (methicillin sensitive); S26=S. aureus 26, ATCC25923 (vancomycin
susceptible); S4=S. aureus 4, Oxford; S2=S. aureus 2, MRSA in vivo (methicillin resistant); E1=E. faecalis 1, ATCC29212
VanS (vancomycin susceptible); E2=E. faecium 1, VanA (vancomycin resistant); P1=S. pneumonia 1, ATCC49619 (erythro-
mycin susceptible); P9=S. pneumonia 9, PenR (penicillin and erythromycin resistant); P9B=S. pneumonia 9 in presence
of 2.5% horse blood; H3=H. influenzae 3, ATCC31517 MMSA; H4=H. influenzae 4, LS2 Efflux knock-out. [b] All analogues
are inactive against E. coli 1, ATCC25922 (non Pathogenic strain), E. coli 50, Ec49 no efflux and P. aeruginosa 1, ATCC27853
(MIC>32 mgmLꢀ1). [c] Not determined. [d] Not determined against S. aureus 26 in presence of 10% serum, E. coli 50, and
P. aeruginosa 1.
was
also
assessed
(Figure 5–7 and Table 3,
and see also Table S2 in
the Supporting Informa-
tion for calculated physi-
cal properties). Again,
the activity was highly
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 1826 – 1837 1830

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
bridges on the 3-acyl moiety,
were inactive or only weakly
active against most strains
(ꢀ0.86< clogD7.4 <2.76), while
28b,d,h were active (2.48<
clogD7.4 <3.11). This tendency
was also found for aromatic ana-
logues of the other templates
(30a (clogD7.4 =4.71, active)
versus 30b (clogD7.4 =3.02, inac-
tive), and 17d,e (clogD7.4 =2.19
and 2.34, respectively, both
active) versus 17 f (clogD7.4
=
1.47, inactive). Interestingly, all
of 27a,b,i–k, 31d–g and 17a
(1.25< clogD7.4 <3.47), possess-
ing nonpolar cyclic alkyl groups
(cyclic hexyl, 2-norbonyl, and
1-adamantyl) with methylene
bridges, as well as cyclohexyl
27c,o,s, 31a and 17c (2.21<
clogD7.4 <3.25) with longer
chain bridges (ethylene, propyl-
ene and butylenes) showed
good antibacterial activity. By
contrast, 27 f,l–n,p,q, 30c and
31b,h (0.01< clogD7.4 <2.55), all
containing a polar atom in the
side chain, were inactive (espe-
cially against S. aureus) or less
active against most strains, even
though their lipophilicity is simi-
lar to the active n-alkyl ana-
logues discussed above.
In order to understand lipo-
philicity–activity
relationships,
plots of clogD7.4 against molecu-
lar surface area (MSA) of the
3-acyl analogues from this study
(104 examples), along with the
Figure 6. 3-Acyltetrameric acids 26–28 (see Table 1 and 3).
3-acyl and the 3-carboxamide
bicyclic TAs from our previous
study (152 examples) were
affected by lipophilicity, and varied with the templates:
made (Figure S2A–C).[9,28,39,40] In the SAR of MRSA and efflux-
negative H. influenzae (H4) (Figure S2A,C), clogD7.4 of the
active analogues is approximately proportional to MSA, with
active 3-acyls (red-filled circles) from the current study tending
to be smaller than the previously reported active 3-acyls (blue-
filled circles) and more lipophilic than previously reported 3-
carboxamides (green-filled circles). As noted above, these ana-
logues were readily transported by the efflux pump of H. influ-
enzae (Figure S2B,C). Of particular interest is that the efflux
pump efficiency is affected not only by lipophilicity (clogP>
4.0) but also molecular volume (MV>620 ꢁ3).[28] There is a simi-
lar trend in the plots of clogP against MSA (Figure S3), but no
strong tendency is found in the plot of polar surface area
(PSA) against MSA (Figure S4). In the plot of relative PSA (rel-
amongst the 3-alkylacyl analogues, 26c (clogD7.4 =0.40) com-
pared with 20c (clogD7.4 =2.22), 27d,e,g, (ꢀ2.65ꢃ clogD7.4
ꢀ1.75) compared with 21e (clogD7.4 =2.29), 27t,u (clogD7.4
2.88 and 2.43, respectively) compared with 21k (clogD7.4
3.76), and 17b (clogD7.4 =1.40) compared with 25b (clogD7.4
ꢃ
=
=
=
3.92) were all inactive. The clogD7.4 values of these are all sig-
nificantly lower (more than 2-log) than the corresponding
most active n-alkylacyl analogues in Table 1. By contrast, 26d,
27h and 31c (clogD7.4 =1.81, 3.01 and 1.88, respectively) were
all active; these possess similar lipophilicity to the correspond-
ing active n-alkylacyl 20e, 21e and 24a (clogD7.4 =2.55, 2.29
and 2.73, respectively). Similarly, analogues 28a,c,e–g,i,j, which
contained aromatic functional groups with various carbon
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 1826 – 1837 1831

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
Mode of action
As the MOA of reutericyclin
(1b)[15] and quorum sensing
product 1c[17] has been report-
ed to be depolarisation of the
bacterial cell membrane, the de-
polarisation ability at S. aureus
membrane (DEP) was evaluated
with selected 3-acyl analogues
(Table 4).[41] Of particular interest
is that this disruption ability
was found to be sensitive both
to the identity of the templates,
and the functionalities on N(1)
and C(5). In the assay, racemic
reutericyclin 23a exhibited dis-
ruption activity (IC50 =88.0 mm),
and the more lipophilic sub-
strates 23b,c, with longer side
chains, were even more active
(IC50 =49.5 and 19.6 mm respec-
tively). In addition, six-mem-
bered 25a,b,d (69.0ꢃIC50 ꢃ
Figure 7. 3-Acyltetrameric acids 29–31 (see Table 1 and 3).
94.5 mm), with the exception of
25e (IC50 >100 mm), were active,
but 5-unsubstituted (20e,g,h,
Table 2. Calculated energy of the ground state of 3AT 18 and 3AP 19.[a]
21e–g,i, 26d and 27a,c,i,k), 5-substituted (24c,e,i,j,m, 30a and
31 f,g,j) and bicyclic (22b) 3ATs did not exhibit such membrane
disruption capacity, with the exception of 5-substituted 24k
and 31e (IC50 =68.8 and 91.4 mm respectively). In order to de-
termine the selectivity between bacterial and mammalian
membranes, red blood cell lysis (RBCL) activity was also evalu-
ated at a concentration of 100 mm. For 3ATs, reutericyclin deriv-
atives 23a–c did not show RBCL activity, whereas N-branched
alkylacyl 24k and 31e disrupted both membranes. Amongst
N-acyl derivatives, RBCL ability of the analogues derived from
N-acyl leucines (23a–c and 30a, inactive) tended to be weaker
than that of the analogues derived from N-acyl methionines
(24e and 31 f, inactive; 24c,i–k,m and 31e,g,j, active). Howev-
er, 5-unsubstituted 20e,g,h, 21 f,g, 26d and 27a did not show
RBCL activity even though 21i and 27i,k showed such activity.
In the assay of 3APs, 25a,b selectively disrupted bacterial
membranes, while 25d disrupted both membranes.
Another possible MOA of 3ATs, namely inhibition of bacterial
RNAP,[19] was also examined for several selected analogues.
5-Substituted systems 24a,b, 31b and 24i,j all showed RNAP
inhibition activity in the range of 16 mmꢃIC50 ꢃ34 mm, and
these values are similar to the value of the known natural
product inhibitor streptolydigin (IC50 =38 mm in our assay[28]
and IC50 =7.5 mm in a previous report[42]). By contrast, 5-unsub-
stituted 27c,h and bicyclic 22b, bearing less-hindered groups
around N(1) and C(5), and six-membered 25a,b, were inactive
or only weakly active. In keeping with the observation that bi-
cyclic 1 f has been reported to be RNAP active,[28] 5-unsubsti-
tuted 27i,k (IC50 =3.1 and 6.7 mm, respectively) showed better
activity than 5-substituted 31d,e,f (IC50 >100, 11 and 40 mm, re-
[b,c]
Compd
Calcd relative energy (kcalmolꢀ1
)
Form A
Form B
Form C
Form D
3AT 18
3AP 19
+1.59
ꢀ0.40
+0.17
ꢀ0.91
+5.62
+0.01
0
0
[a] 3AT 18 (n=0, R1 =acetyl, R2 =H, R3 =Me), 3AP 19 (n=1, R1 =acetyl,
R2 =H, R3 =Me) in Figure 3. [b] The energy difference between each tau-
tomer related to tautomer D, [c] Calculated by using DFT B3LYP (6-31G*)
in Spartan 02.
PSA) against MSA in Figure S5, however, the active analogues
are seen to occur at a narrow range of rel-PSA (9%<rel-PSA<
15% for MRSA and 11%<rel-PSA<15% for H3). These ana-
logues possess a range of physicochemical properties tending
to higher PSA and rel-PSA compared with previously reported
unnatural 3AT libraries (Figure S2D–S5D).[28] This tendency re-
sults from the fact that our library has di- or tri-substitution on
N(1), C(5) and at the 3-acyl groups, whereas previous libraries
usually have only one or two points of diversity and are gener-
ally without polar atoms. However, these analogues possess
a narrower range of MSA and PSA compared with natural 3ATs,
and the library has a higher hit ratio (ratio of active analogues/
total analogues) than these previous libraries, probably as
a result of their greater diversity.
In addition, the antifungal activity against Candida albicans
of the 3-acyl analogues (fluconazole as a standard) was also
evaluated, but none were active (minimum inhibitory concen-
tration (MIC)ꢂ32 mgmLꢀ1, data not shown). This is evidence
for phenotypic selectivity against bacteria.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 1826 – 1837 1832

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
which showed mild antibacterial activity,
did not show either RNAP (IC50 >100 mm)
or UPPS (IC50 >10 mm) activities.
Table 3. In vitro antibiotic activity (MIC, mgmLꢀ1) of 3-acyl analogues 19 and 26–31 (for identity
of compounds, see Figure 5).[a–f]
Compd S1
S26
S4
S2
E1
E2
P1
P9
P9B H3 H4
[c]
[c]
[c]
[c]
[c]
[c]
[c]
[c]
[c]
[c]
[c]
[c]
[c]
26a
26b
26c
26d
27a
27b
27c
27h
27i
>64
>64[d]
>64[d]
>64[d]
8[d]
4
4
1
0.5
0.5
16
–
–
–
–
2
4
>64
>64
>64
8
>64
>64
>64
16
1
2
0.5
0.25
0.12
8
0.5
16
16
8
4
64
8
2
64
1
16
64
4
0.5
64
2
>64
>64
>64
16
>64
>64
>64
16
–
–
–
–
–
–
–
–
4
2
2
1
1
4
1
16
8
8
>64
>64
>64
4
–
–
–
–
1
Binding model
[c]
[c]
[c]
>64
>64
8
In order to gain insight to the possible
modes of substrate binding to RNAP and
UPPS, docking calculations with 3AP 25b
and 3AT 27i were carried out in which one
binding site of UPPS (PDB code: 1V7U)[43]
and two binding sites of RNAP (PDB code:
1ZYR with streptolydigin[19] and 3DXJ with
myxopyronin[44]) were considered, based
on the structural similarity of the com-
pound libraries with these ligands
(Table 5).[28,45] In order to determine wheth-
er there are specifically favoured tautomer-
ic forms for binding, calculation using each
tautomeric form was carried out. Although
binding energies of 3AP 25b to both of
the RNAP sites yielded substantial negative
docking energies, a nonspecific geometric
conformation was observed for all tauto-
mers, suggesting that 3AP 25b is unlikely
to specifically bind to the RNAP sites. In
the docking model for the UPPS site, how-
ever, tautomers B and C of 3AP 25b yield-
ed not only significant negative docking
energies but also a specific docking geom-
etry, while tautomers A and D had no spe-
cific geometric preference. Based on the
docking results, as well as from the analysis
of the tautomeric behaviour as described
above in which the tautomer C is unlikely
to be a significant contributor, tautomer B
of 3AP 25b is favoured to bind at the
UPPS site. These docking results are in
agreement with the experimental data
where 3AP 25b is strongly active at UPPS
(IC50 =0.6 mm) but only weakly active at
RNAP (IC50 =86 mm). On the other hand, all
the docking models of 3AT 27i yielded sig-
1
2
0.25
0.25
0.25
2
4
1
2
2
0.12
1
1
0.5
0.5
4
4
2
2
0.5
1
0.5
0.5
0.25
0.12 <0.06
0.25 <0.06 <0.06
0.5
0.25
16
2
>64
64
>64
4
>64
16
2
>64
8
>64
>64
64
2
64
2
16
>64
64
32
2
1
4
4
8
16
2
2
16
4
2
32
1
8
16
8
4
2
8
16
0.25
8
1
32
32
2
2
32
8
1
64
2
32
64
8
0.5
32
2
0.12
0.12
4
0.5
2
16
8
64
>64
>64
8
27j
27k
27l
16
16
2
4
1
8
16
0.5
0.5
16
2
0.5
64
0.5
16
32
4
0.5
32
1
2
>64
32
>64
4
>64
16
2
64
4
64
>64
16
1
64
2
16
2
>64 >64
>64 >64
>64 >64
16
8
27m
27n
27o
27r
27s
28b
28c
28d
28e
28 f
28g
28h
28i
30a
31a
31b
31c
31d
31e
31 f
31g
31i
0.5
0.5
8
16
4
>64 >64
16
4
>64 >64
32 16
>64 >64
>64 >64
>64 >64
32
2
64 >64
16
2
16
2
32
4
16
32
8
4
64
4
32
8
64
4
32
32
16
8
64
>64
0.25
32
0.5
16
32
4
4
4
0.5
32
1
>64 >64
4
64
2
32
8
8
8
8
16 >64
32 >64 >64
16 >64
>64
64
16
>64 >64
>64
64
16
4
0.5
0.25
2
1
16
>64
32
16
4
>64
32
8
32
16
8
64
32
1
1
0.5
>64
1
16
2
64
16
4
0.5
0.5
16
1
16
2
4
4
4
2
2
8
8
8
8
8
4
2
2
4
8
8
4
4
8
32
32
32
2
0.5
0.12
8
0.5
8
1
2
4
4
2
0.5
8
0.5
16
2
4
4
2
1
0.5
2
0.5
8
1
2
2
2
0.5
0.25
1
0.25
8
1
1
2
1
0.25
1
0.25
32
>64
>64
32
16
16
31j
19a
19c
19d
19e
16
2
4
2
2
4
2
4
4
4
4
8
16
>64
[a] Abbreviations: see footnote in Table 1. [b] All analogues are inactive against E. coli 1,
ATCC25922 (non Pathogenic strain), E. coli 50, Ec49 no efflux and P. aeruginosa 1, ATCC27853
(MIC>32 mgmLꢀ1). [c] Not determined. [d] Not determined against S. aureus 26 in presence of
10% serum, E. coli 50, and P. aeruginosa 1. [e] 27d,e,g,p,t,u, 28j, 29b, 30b,c and 31h were inac-
tive against all strains (MIC>32 mgmLꢀ1). [f] 27 f,q (MIC; 32 mgmLꢀ1), 28a, 29a and 17b,f (MIC=
16 mgmLꢀ1) were active against H4 but inactive against the other strains (MIC>32 mgmLꢀ1).
spectively). However, dioxolanyl 30c, norbornyl 31g (similar
steric effect to the adamantyl group) and the less sterically hin-
dered bicyclic 29a were inactive or only weakly active. In addi-
tion, phenyl 28h also showed RNAP inhibition activity (IC50 =
33 mm), while other phenyl derivatives 28c,g and 31j were in-
active or weakly active.
nificant negative docking energies in which the binding
energy at 1V7U was lower than at 1ZYR and 3DXJ. Further-
more, the low number of specific geometric conformations,
with the exception of tautomer C to 1ZYR and tautomer A to
3DXJ, support the experimental observation that 3AT 27i is
strongly active at RNAP and UPPS (IC50 =3.1 mm and 0.4 mm, re-
spectively). These results are of interest and suggest that 3AT
27i might be able to bind to the three active sites (multi-tar-
geting ability); if true, this may have significance for the pre-
vention of rapid target-based endogenous resistance,[46] as dis-
cussed further below.
An alternative possible MOA, inhibition of undecaprenyl py-
rophosphate synthase (UPPS), was also evaluated for three an-
alogues. Similarly to 3-carboxamide 1g reported in a Novartis
study (IC50 =0.04 mm),[29] six-membered 3-acyl 25b (only weakly
active against RNAP, IC50 =86 mm) was found to be strongly
UPPS active (IC50 =0.6 mm). Notably, monocyclic 27i showed
dual MOA ability as an RNAP (IC50 =3.1 mm) and UPPS (IC50 =
0.4 mm) inhibitor, similar to bicyclic 1 f.[28] However, 3AT 28g,
The docking model of tautomer B of 3-acyl derivatives 25b
and 27i at the UPPS site is presented in Figure S6 (see the
Supporting Information). In these models, the phenyl group in
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 1826 – 1837 1833

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
3AP 25b and the adamantyl
group in 3AT 27i form hydro-
phobic interactions with the
amino acids Leu85, Leu88 and
Phe89. The flexible alkyl chains
in 3-acyls 25b and 27i also
form hydrophobic interactions
with Ile24, Met25 and Ala69. In
addition, the oxygen on the 3-
acyl group in 3AP 25b interacts
with Asn28. Analyzing the ge-
ometry of tautomer B of 27i
bound at the streptolydigin
binding site (1ZYR) of RNAP re-
veals that the molecule is inter-
calated between two helices
(Figure S6C). The adamantyl
group is in hydrophobic contact
with Phe1241 and the hexyl
chain forms hydrophobic inter-
actions with Leu1094, Leu1256
and Ile1260, even though these
amino acid residues do not in-
teract with streptolydigin (Fig-
ure S7A). The oxygen of the
3-acyl moiety in 3AT 27i also
accepts a hydrogen bond from
Arg1087. Moreover, Thr1237
and Asp1090 are in polar inter-
action with the oxygen of the
tetramic acid ring. Of particular
interest is that the binding ge-
ometry of monocyclic 3AT 27i
is different to bicyclic 3AT 1 f, in
which the adamantyl group in
monocyclic 27i is positioned
near the binding pocket of the
sugar pendant on the tetramic
acid scaffold of streptolydigin,
while that of bicyclic 1 f is posi-
tioned near the binding pocket
of the bicyclic pendant in the
streptolol moiety of streptolydi-
Table 4. Pharmacological properties of selected tetramic acids.[a–f]
Compd
RNAP
IC50 [mm]
DEP
IC50 [mm]
RBCL
(IC50)[d]
HEK293
LD50 [mm]
PMBC
LD50 [mm]
SI
SOL
[mm]
PPB
[%]
[c]
[c]
[c]
[c]
[c]
20e
20g
20h
21b
21e
21 f
21g
21i
22b
23a
23b
23c
24b
24c
24e
24i
–
–
–
–
–
–
–
–
>100
>100
>100
Inactive
Inactive
Inactive
>90.9
10.1
>90.9
30.3
90.9
30.3
>90.9
30.3
90.9
90.9
90.9
30.3
30.3
30.3
>90.9
10.1
10.1
10.1
30.3
>90.9
>90.9
90.9
90.9
>90.9
90.9
30.3
30.3
10.1
30.3
30.3
>90.9
30.3
90.9
10.1
90.9
90.9
90.9
90.9
90.9
30.3
10.1
90.9
>90.9
90.9
90.9
30.3
90.9
30.3
>90.9
30.3
30.3
90.9
90.9
>90.9
>90.9
90.9
90.9
90.9
90.9
90.9
90.9
30.3
90.9
90.9
90.9
90.9
>90.9
30.3
30.3
90.9
6.7
–
–
–
–
–
–
[c]
[c]
[c]
<0.23
[c]
[c]
–
[c]
[c]
[c]
–
–
–
11
7.3
1.5
>300
–
99.9
[c]
[c]
[c]
[c]
-c]
[c]
[c]
>100
>100
>100
>100
>100
88.0[e]
49.5
–
–
–
–
–
–
–
–
[c]
[c]
Inactive
Inactive
Active
–
[c]
[c]
<0.06
2.9
-
[c]
[c]
–
[c]
>100
Active
8.6
4.0
38–75
–
[c]
[c]
[c]
[c]
[c]
–
–
–
25
–
–
33
19
–
Inactive
Inactive
Inactive
[c]
[c]
137
25
–
–
[c]
[c]
19.6
–
[c]
[c]
–
12
6.2
>300
>300
>300
>300
>300
>300
>300
>300
>300
150–300
19–38
–
93.6
99.9
[c]
>100
>100
>100
>100
68.8
Active
Inactive
Active
Active
Active
[c]
[c]
>75
<8.3
<8.9
<9.5
6.9
>8.4
>73
9.1
–
100
24j
100
–
–
[c]
[c]
[c]
24k
24m
25a
25b
25d
25e
26d
27a
27c
27h
27i
27k
28h
30a
31e
31 f
31g
31j
[c]
–
>100
83.6
Active
>100
Inactive
Inactive
Active
98.9
86 (0.6)[b]
94.4
69.0
>100
>100
>100
>100
–
–
–
–
[c]
[c]
[c]
–
–
–
–
[c]
[c]
Active
9.8
3.3
16
11
[c]
[c]
[c]
[c]
Inactive
Inactive
[c]
[c]
–
–
[c]
>100
>100
–
–
>200
150–300
67–200
>300
>300
>300
>300
>300
>300
>300
>300
100
[c]
[c]
–
21
91.9
99.9
3.1(0.4)[b]
6.7
33
>100
>100
Active
Active
<16
5.7
[c]
–
[c]
[c]
–
–
6.4
19
3.2
21
100
[c]
[c]
[c]
[c]
[c]
[c]
[c]
–
11
40
>100
91.4
>100
>100
>100
Inactive
Active
Inactive
Active
–
–
–
–
–
–
>100
91
<9.1
<11
9.7
Active
10.1
90.9
[c]
[c]
[c]
19d
–
–
–
[a] Abbreviations: RNAP=in vitro activity against E. coli RNAP; DEP=in vitro activity in depolarisation of
S. aureus membrane; RBCL=in vitro mammalian red blood cell membrane lysis activity; HEK293=in vitro toxic-
ities against human embryonic kidney 293 cells; PMBC=in vitro toxicities human peripheral blood cells; SI=
selectivity index (LD50 of one of lower value between HEK293 and PMBC divided by MIC against S. aureus 2,
MRSA in vivo after converting the unit of LD50 from mm to mgmLꢀ1); SOL=aqueous solubility at pH 7.4 (water
with 2% DMSO); PPB=ratio of plasma protein binding. [b] In vitro activity against S. pneumonia UPPS (IC50).
[c] Not determined, [d] Active: IC50 <100 mm; inactive: IC50 >100 mm. [e] Reported in our previous paper.[28]
[f] UPPS inhibition activity of 28g (>10 mm) and RNAP inhibition activity of 24a (16 mm), 30c (85 mm), 31b
(34 mm) and 28c,g, 29a, 31d (>100 mm), were also determined.
gin.[19,28] On the other hand, analysis of the geometry of
tautomer B of 27i-RNAP complex at the myxopyronin bind-
ing site (3DXJ) reveals two hydrogen bonds as the driving
force of the complexation. Thus, the hydroxyl group of the
3-acyl moiety accepts a hydrogen bond from the backbone
amide of Gly620, while the carbonyl oxygen on C(4)-posi-
tion accepts a hydrogen bond from the backbone amide of
Glu1034. In addition, the adamantyl group is in hydropho-
bic interaction with Leu619 and Ile1467. However, in this
case, the hexyl chain does not form favourable hydropho-
bic interactions, while the alkyl chain, which contains a car-
bamate, forms a polar interaction (Figure S7B). This result
supports the observation that the shorter chain analogue
27k is active at RNAP (IC50 =6.7 mm).
Table 5. Docking results of 3AP 25b and 3AT 27i.
UPPS (1V7U)
RNAP (1ZYR[a]
x[d] E [kcalmolꢀ1
)
RNAP (3DXJ[b]
)
[c]
[c]
[c]
Compd Tautomer E [kcalmolꢀ1
]
]
x[d] E [kcalmolꢀ1
]
x[d]
[e]
[e]
[e]
3AP
25b
A
B
C
D
A
B
ꢀ8.05
ꢀ9.27
ꢀ8.09
ꢀ8.96
ꢀ9.18
ꢀ9.69
ꢀ9.43
ꢀ9.14
–
1
1
–
2
3
1
4
ꢀ6.08
ꢀ7.03
ꢀ6.44
ꢀ7.28
ꢀ7.49
ꢀ8.41
ꢀ6.73
ꢀ7.84
–
–
–
–
1
1
–
2
ꢀ7.96
ꢀ7.32
ꢀ8.27
ꢀ7.11
ꢀ8.13
ꢀ8.20
ꢀ7.65
ꢀ8.01
–
–
–
–
[e]
[e]
[e]
[e]
[e]
[e]
[e]
[e]
3AT
27i
–
3
1
1
[e]
C
D
[a] Active site of streptolydigin. [b] Active site of myxopyronin. [c] E=docking
energy. [d] x=number of conformations. [e] Nonspecific number of confor-
mations.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 1826 – 1837 1834

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
Toxicity
against S. aureus Sa26, and only 24b, 25a,b and 27i showed
some weak activity (16ꢃMICꢃ64, data not shown). For the
same reason, 24b and 27i were inactive in a mouse thigh in-
fected by septicaemia in vivo assay (data not shown). Figure S8
presents the relationships of the plasma protein binding (PPB)
effect against various physicochemical properties. The PPB
effect correlates with clogP and rel-PSA in which only those
analogues possessing lower than a clogP value of 1.8 and
more than 15% of rel-PSA are likely to show low PPB affinity.
The toxicities against human embryonic kidney 293 cells
(HEK293) and human peripheral blood cells (PMBC) (Table 4)
depended on both the identity of the templates, as well as the
functionalities on N(1), C(5)) and the 3-acyl group, although
a clear SAR pattern could not be discerned. 3ATs 24e and 31 f,
both derived from 6b, were less toxic than counterparts 24b,c
and 31e, derived from the less lipophilic 6a. N-Hexyl 20e,g,h
and 26d all exhibited lower toxicities (LD50 ꢂ90.9 mm) against
both cell lines, with the exception of 20g (HEK293 LD50 =
10.1 mm) whereas LD50 values of 3ATs 24i–k and 31g,j derived
from 6d against HEK293 and PMBC were 10.1 and 30.3 mm, re-
spectively. One exception was 24k against PMBC (LD50 =
90.9 mm). However, 3ATs 27c,h,i,k and 28h derived from 3a,b
(except 27a) tended to be more toxic against HEK293 than
PMBC. Of particular interest is that LD50 values of six-mem-
bered 25a,b,d,e and 17d against both cell lines were
ꢂ90.9 mm. In addition, reutericyclins 23a,b showed low toxici-
ties (LD50 =90.9 mm) against both cell lines, while the longer
chain 23c (as compared to 23a,b) led to increased toxicity
(LD50 =30.3 mm). This dependence of toxicity on chain length
was also observed in the other analogues: lengthening of the
n-alkyl chain on N(1) and the 3-acyl group increased PMBC tox-
icity (21e!21 f!21g and 24b!24c; 21b!21 f) although
again no clear relation between the chain length and HEK293
toxicity could be found. In order to judge the balance between
antibacterial activity and toxicity, a selectivity index (SI) was
calculated in which the ratio of MIC value against S. aureus 2
(MRSA strain) against the lower LD50 value of HEK293 and
PMBC was taken. 3ATs 24e and 31 f derived from 6b and reu-
tericyclins 23b,c (with the exception of 23a, SI=4.0) exhibited
high selectivities (13.7<SI<75). Other analogues 20e, 21e,
22b, 27a and 30a possessing low toxicity and good antibacte-
rial activity (LD50 =90.9 mm, MIC=2–4 mgmLꢀ1) also provided
high SI values (6.7–19). However, SI values >10 for 21b, 24b
and 27c,h possessing mild toxicity and good antibacterial ac-
tivity were also found. Even though 24i–k, 27i and 31g,j were
found to be toxic (LD50 =10.1 mm), their SI values are favoura-
ble (8.3<SI<16) because of their excellent antibacterial prop-
erties (MIC=0.5 mgmLꢀ1), and these SI values are similar to
3APs 25a,d,e and 19d (8.4<SI<9.8) (with the exception of
25b, SI=73). By contrast, highly toxic 20g and 21g also pos-
sessing weak antibacterial activity had low SI values (SI=0.23
and 0.06, respectively).
Analogues showing
a large MIC difference (>15ꢂ) are
bunched in regions of high clogP (>3.2) and low rel-PSA
(<13.0%).[47,48]
Conclusions
A series of 3ATs and 3APs bearing a large range of substituents
on the N(1), C(5) and 3-acyl groups has been prepared, based
on several monocyclic core tetramate templates. The direct
3-acylation promoted by 1.3 equivalent of 4-dimethylamino-
pyridine (DMAP) offers the easiest way to incorporate the re-
quired 3-acyl groups at a late synthetic stage and thereby pro-
vides ready access to diversely substituted libraries. Their tau-
tomeric behaviour is mainly affected by ring size, with tetramic
acids and piperidine-2,4-diones exhibiting very different behav-
iour, in which the identity of N-substitution (rather than C(5)-
or 3-acyl substitution) dominates. Thus, N-unsubstituted and
N-alkyl 3ATs favour tautomer D, N-acyl 3ATs favour tautomers
AB and D, and N-acyl 3APs favour tautomer B.
3ATs and 3APs exhibited good antibacterial activity against
various Gram-positive strains including MRSA, VRE and MDRSP,
along with activity against Gram-negative H. influenzae. A
strong relationship between antibacterial activity and lipophi-
licity (clogD7.4 and clogP) rather than polar surface area (PSA)
was found. This SAR study provides a guide for the design of
new antibacterial 3ATs. Limited mode of action (MOA) informa-
tion at two targets (RNAP, UPPS) has been developed, and
5-substitution of 3ATs appears to be necessary for bacterial
membrane disrupting ability (e.g., 24a–c,k and 31e; IC50 up to
19.6 mm), while the adamantyl pendant on the 3-acyl position
with appropriate functionality on N(1) and C(5) provides for
RNAP inhibition activity (e.g., 27i,k and 31e,f; IC50 up to
3.1 mm). This indicates that MOAs should be capable of modu-
lation by incorporation of appropriate functionality. Notewor-
thy is that 27i and 27k, which exhibited low-micromolar inhib-
ition of RNAP, and 24b, 24i and 24j only moderate RNAP in-
hibition, were strongly active against the panel of bacteria
(Table 2), while 27c, 27h, 31g, 31j, 31 f (Table 3) which were
essentially inactive at RNAP nonetheless also showed potent
phenotypic antibacterial activity. This RNAP activity is perhaps
unsurprising, given the known activity of the natural product
parent, streptolydigin, at RNAP for example.[49–51] Moreover,
none of these compounds exhibited depolarisation activity
(Table 4). This shows that, although a subset of the library may
exhibit RNAP activity, this activity is neither necessary nor suffi-
cient for antibacterial activity, and this outcome points to
other as yet unidentified MOAs. This could include UPPS activi-
ty, since both 25b and 27i showed potent activity at this
enzyme, but is clear that extensive investigation of these com-
Other biological activities
In order to profile general pharmacological properties, selected
analogues were subjected to ADME assays. The 3-acyl ana-
logues were generally soluble in water at pH 7.4 (>150 mm),
with the exception of 22b and 25e (<100 mm). Unfortunately,
high affinity towards plasma protein binding (91.9–100%) of
many substrates interrupted their antibacterial activity, and
almost all MIC values against Streptococcus pneumonia 9 (P9B)
were shifted to higher values in the presence of 2.5% blood.
MIC values were also shifted in the presence of 10% serum
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 1826 – 1837 1835

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
pounds is required in order to fully delineate their MOA. Toxici-
ty against HEK293 and PMBC weakly depended on the func-
tionalities of the 3-acyl analogues rather than their lipophilicity.
Unfortunately, this activity was generally lost in the presence
of serum, consistent with strong plasma protein binding of
these systems.
products as lead structures with biologically validated starting
points for drug discovery, as has been promoted by
Newman,[64] Waldemann[65–67] and Quinn.[68,69]
Experimental Section
Tautomeric behaviour: 3-Acyltetramic acids (3ATs) 18 and 3-acylpi-
peridine-2,4-diones (3APs) 19 both exist as an equilibrium between
four tautomeric forms A–D in solution (Figure 3).[31,34–36] Each set of
external tautomers (AB and CD) are usually easy to distinguish,
since their interconversion rate is slow on the nuclear magnetic
resonance (NMR) time scale. By contrast, the internal tautomers of
each set (A 1 B and C 1 D) are observed as broadened peaks be-
cause of their rapid interconversion. After assigning each external
tautomer by analysis of HMBC NMR spectra, their ratio could be
readily determined from the intensity of separate peaks (usually
Of all the systems examined, those of interest appear to be
3ATs 23b, 27h,l and 31g which may be promising for topical
use (Figure 8), since these were uniformly active against
S. aureus and S. pneumonia (MICꢃ0.5 mgmLꢀ1), organisms as-
sociated with skin, ear and respiratory tract infections.[52–54]
Moreover, such topical applications may not be so affected by
plasma protein binding affinity. Also, the possibility of multiple
targeting abilities of 3AT 27i at RNAP (possessing two active
sites) and UPPS, and with IC50 values amenable for practical ap-
plication (3.1 and 0.4 mm, respectively) is of interest.[28,55] Impor-
tantly, these targets are related to different metabolic functions
in bacteria, namely cell wall (UPPS) and RNA synthesis (RNAP),
which have not been exploited by current antibacterial agents.
By contrast, previously reported multiple targeting antibacteri-
als with novel MOAs have been related only to single metabol-
ic functions (e.g., platencin for fatty acid biosynthesis (FabF
and FabH)[56] and thiazolidin-4-one-based inhibitor for peptido-
glycan biosynthesis (MurD and MurE ligases[57])). Moreover, the
depolarisation selectivity for bacterial membranes is important,
especially since this activity is known to be targeted by some
natural products (e.g., telavancin[58] and ceragenin[59]) and has
been highlighted as being underdeveloped[58] and may provide
opportunities for further development, especially in the treat-
ment of slow growing organisms.[60,61] Furthermore, recent
work has suggested that appropriate chemical modification of
the tetramate/piperidinedione ring system may ameliorate
serum binding activity, and this offers significant opportunity
for future development.[62]
1
the C(5) proton) in the H NMR spectra. The ratio of internal tauto-
mers was roughly estimated from the chemical shifts of the car-
1
bonyl carbons in the 13C NMR spectra. In the H NMR and 13C NMR
spectra of 3AP 19a, two sets of signals were observed in a ratio of
about 20:80. In order to assign these signals, the HMBC spectrum
was acquired and the correlations for the main ring established as
shown in Figure 6, in which the major external tautomer was as-
signed as the AB tautomer (in the HMBC correlation, resonances
for the free carbonyl on C(2) in the external tautomer AB (about
165 ppm) and the free carbonyl on C(4) in external tautomer CD
(about 190 ppm), which are not affected by the ratio of each inter-
nal tautomer, were indicative). The hydrogen-bonded carbonyl on
C(4) in the external tautomer AB (about 195 ppm) and the hydro-
gen-bonded carbonyl on C(2) in the external tautomer CD (about
175 ppm) indicated that tautomers B and D were the main contrib-
utors to each external tautomer.
Calculation of physicochemical properties: The energies in equi-
librium geometry at ground state were carried out with density
functional B3LYP (6-31G*) in Spartan 02 in each tautomeric isomer.
In the calculation of physicochemical properties, tautomer D was
selected. MSA, PSA, clogP, clogD7.4, hydrogen-bond donor count,
hydrogen-bond acceptor count and rotatable bond count were ob-
tained from MarvinSketch version 5.3.8. (www.chemaxon.org), in
which clogP and clogD7.4 by using VG method and MSA by using
van der Waals method were performed and sulfur atoms were ex-
cluded in the calculation of PSA. Rel-PSA were calculated from
(PSA/MSA)ꢂ100.
The tetramates and piperidinediones reported herein pro-
vide synthetically accessible but unusual systems worthy of
further investigation, for which suitable optimisation might
provide a novel approach for the treatment of bacterial infec-
tions.[63] These results further demonstrate the use of natural
Docking study: Docking calculations were performed by using the
DockingServer methodology as described in our previous paper.[28]
In the course of docking calculations, 100 independent docking
runs were carried out, and the results were clustered according to
docking geometry. A conformation was considered to be relevant
if the cluster had more than 10 members, and the lowest docking
energy of the given conformation was within 1 kcalmolꢀ1 of the
lowest energy hit. Low numbers of alternative conformations im-
plies potentially specific binding, whereas if the frequency of each
conformation is below 10%, there is no significantly favoured com-
plex geometry at the binding site, suggesting no specific inhibition
at the binding site.
Biological testing: Bacterial RNA polymerase (RNAP) and undecap-
renyl pyrophosphate synthase (UPPS) enzyme assay, minimum in-
hibitory concentration (MIC) determination for bacteria and Candi-
da albicans strains, cytotoxicity test in HEK293 and PBMC cell cul-
tures, solubility assay, plasma protein binding (PPB) assay, bacterial
cell membrane depolarisation and red blood cell lysis were per-
formed as described previously.[28]
Figure 8. Chemical systems with optimal properties.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 1826 – 1837 1836

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
[32] U. Marquardt, D. Schmid, G. Jung, Synlett 2000, 1131–1132.
[33] J. Marin, C. Didierjean, A. Aubry, J.-R. Casimir, J.-P. Briand, G. Guichard, J.
Org. Chem. 2004, 69, 130–141.
Acknowledgements
We are particularly grateful for valuable input by Drs. Phil Dud-
field and John Lowther, and for funding by Galapagos SASU
(France).
[34] K. Saito, T. Yamaguchi, J. Chem. Soc. Perkin Trans. 2 1979, 1605–1609.
[35] M. J. Nolte, P. S. Steyn, P. L. Wessels, J. Chem. Soc. Perkin Trans. 1 1980,
1057–1065.
[36] M. Petroliagi, O. I. Markopoulou, J. Chem. Soc. Perkin Trans. 1 1997,
3543–3548.
[37] H. B. Broughton, P. R. Woodward, J. Comput.-Aided Mol. Des. 1990, 4,
147–153.
[38] C.-K. Skylaris, O. I. Marlopoulou, A. Detsi, J. Markopoulos, Chem. Phys.
2003, 293, 355–363.
[39] R. O’Shea, H. E. Moser, J. Med. Chem. 2008, 51, 2871–2878.
[40] C. A. Holloway, C. J. Matthews, Y.-C. Jeong, M. G. Moloney, C. F. Robert,
M. Yaqoob, Chem. Biol. Drug Des. 2011, 78, 229–235.
[41] D. R. Gentry, I. Wilding, J. M. Johnson, D. Chen, K. Remlinger, C. Richards,
S. Neill, M. Zalacain, S. F. Rittenhouse, M. N. Gwynn, J. Microbiol. Methods
2010, 83, 254–256.
[42] E. Sarubbi, F. Monti, E. Corti, A. Miele, E. Selva, Eur. J. Biochem. 2004,
271, 3146–3154.
[43] S.-Y. Chang, T.-P. Ko, A.-C. Chen, A. H.-J. Wang, P.-H. Liang, Protein Sci.
2004, 13, 971–978.
[44] J. Mukhopadhyay, K. Das, S. Ismail, D. Koppstein, M. Jang, B. Hudson, S.
Sarafianos, S. Tuske, J. Patel, R. Jansen, H. Irschik, E. Arnold, R. H.
Ebright, Cell 2008, 135, 295–307.
Keywords: antibacterial agents · drug discovery · synthesis ·
tetramates
[1] S. R. Palumbi, Science 2001, 293, 1786–1790.
[2] R. J. L. Willems, W. P. Hanage, D. E. Bessen, E. J. Feil, FEMS Microbiol. Rev.
2011, 35, 872–900.
[3] U. Theuretzbacher, J. W. Mouton, Curr. Opin. Pharmacol. 2011, 11, 429–
432.
[4] M. A. Fischbach, C. T. Walsh, Science 2009, 325, 1089–1093.
[5] M. A. Fischbach, Curr. Opin. Microbiol. 2011, 14, 519–523.
[6] V. Pokrovskaya, T. Baasov, Expert Opin. Drug Discovery 2010, 5, 883–902.
[7] S. Donadio, S. Maffioli, P. Monciardini, M. Sosio, D. Jabes, J. Antibiot.
2010, 63, 423–430.
[8] M. N. Gwynn, A. Portnoy, S. F. Rittenhouse, D. J. Payne, Ann. N. Y. Acad.
Sci. 2010, 1213, 5–19.
[9] D. J. Payne, M. N. Gwynn, D. J. Holmes, D. L. Pompliano, Nat. Rev. Drug
Discovery 2007, 6, 29–40.
[10] F. von Nussbaum, M. Brands, B. Hinzen, S. Weigand, D. Hꢃbich, Angew.
Chem. 2006, 118, 5194–5254; Angew. Chem. Int. Ed. 2006, 45, 5072–
5129.
[11] F. E. Koehn, G. T. Carter, Nat. Rev. Drug Discovery 2005, 4, 206–220.
[12] S. Danishefsky, Nat. Prod. Rep. 2010, 27, 1114 – 1116.
[13] B. L. Royles, Chem. Rev. 1995, 95, 1981–2001.
[45] Z. Bikadi, E. Hazai, J. Cheminf. 2009, 1, 15.
[46] L. L. Silver, Nat. Rev. Drug Discovery 2007, 6, 41–55.
[47] A. Thorarensen, J. Li, B. D. Wakefield, D. L. Romero, K. R. Marotti, M. T.
Sweeney, G. E. Zurenko, R. W. Sarver, Bioorg. Med. Chem. Lett. 2007, 17,
3113–3116.
[48] G. Colmenarejo, A. A. Pedraglio, J.-L. Lavandera, J. Med. Chem. 2001, 44,
4370–4378.
[14] N. Imamura, K. Adachi, H. Sano, J. Antibiot. 1994, 47, 257–261.
[15] M. G. Gꢃnzle, R. F. Vogel, Appl. Environ. Microbiol. 2003, 69, 1305–1307.
[16] G. F. Kaufmann, R. Sartorio, S.-H. Lee, C. J. Rogers, M. M. Meijler, J. A.
Moss, B. Clapham, A. P. Brogan, T. J. Dickerson, K. D. Janda, Proc. Natl.
Acad. Sci. USA 2005, 102, 309–314.
[49] R. Mariani, S. I. Maffioli, Curr. Med. Chem. 2009, 16, 430–454.
[50] K. von der Helm, J. S. Krakow, Nat. New Biol. 1972, 235, 82–83.
[51] D. G. Vassylyev, M. N. Vassylyeva, J. Zhang, M. Palangat, I. Artsimovitch,
R. Landick, Nature 2007, 448, 163–168.
[52] R. Dagan, Vaccine 2000, 19, S9–S16.
[17] C. A. Lowery, J. Park, C. Gloeckner, M. M. Meijler, R. S. Mueller, H. I. Bosh-
off, R. L. Ulrich, C. E. Barry, D. H. Bartlett, V. V. Kravchenko, G. F. Kauf-
mann, K. D. Janda, J. Am. Chem. Soc. 2009, 131, 14473–14479.
[18] C.-Y. Wang, B.-G. Wang, S. Wiryowidagdo, V. Wray, R. Soest, K. G. Steube,
H.-S. Guan, P. Proksch, R. Ebel, J. Nat. Prod. 2003, 66, 51–56.
[19] S. Tuske, S. G. Sarafianos, X. Wang, B. Hudson, E. Sineva, J. Mukhopad-
hyay, J. J. Birktoft, O. Leroy, S. Ismail, A. D. Clark, C. Dharia, A. Napoli, O.
Laptenko, J. Lee, S. Borukhov, R. H. Ebright, E. Arnold, Cell 2005, 122,
541–552.
[20] R. Yendapally, J. G. Hurdle, E. I. Carson, R. B. Lee, R. E. Lee, J. Med. Chem.
2008, 51, 1487–1491.
[21] J. G. Hurdle, R. Yendapally, D. Sun, R. E. Lee, Antimicrob. Agents Chemo-
ther. 2009, 53, 4028–4031.
[22] C. Ueda, K. Tateda, M. Horikawa, S. Kimura, Y. Ishii, K. Nomura, K.
Yamada, T. Suematsu, Y. Inoue, M. Ishiguro, S. Miyairi, K. Yamaguchi, An-
timicrob. Agents Chemother. 2010, 54, 683–688.
[23] S. V. Pronin, A. Martinez, K. Kuznedelov, K. Severinov, H. A. Shuman, S. A.
Kozmin, J. Am. Chem. Soc. 2011, 133, 12172–12184.
[24] K. Katsuo, M. Kimura, T. Kinuta, N. Takai, K. Tanaka, Chem. Pharm. Bull.
1984, 32, 4197–4204.
[25] K. Katsuo, T. Kitaguchi, Y. Takata, K. Tanaka, Chem. Pharm. Bull. 1980, 28,
2494–2502.
[26] T. Rosen, P. B. Fernandes, M. A. Marovich, L. Shen, J. Mao, A. G. Pernet, J.
Med. Chem. 1989, 32, 1062–1069.
[53] S. McChlery, G. Ramage, J. Bagg, Periodontology 2000, 2009, 49, 151–
165.
[54] A. Fogo, N. Kemp, R. Morris-Jones, BMJ 2011, 343, d5343.
[55] J. Cheng, J. A. Thanassi, C. L. Thoma, B. J. Bradbury, M. Deshpande, M. J.
Pucci, Antimicrob. Agents Chemother. 2007, 51, 2445–2453.
[56] J. Wang, S. Kodali, S. H. Lee, A. Galgoci, R. Painter, K. Dorso, F. Racine, M.
Motyl, L. Hernandez, E. Tinney, S. L. Colletti, K. Herath, R. Cummings, O.
Salazar, I. Gonzalez, A. Basilio, F. Vicente, O. Genilloud, F. Pelaez, H. Jaya-
suriya, K. Young, D. F. Cully, S. B. Singh, Proc. Natl. Acad. Sci. USA 2007,
104, 7612–7616.
ˇ
ˇ ´
[57] T. Tomasic, R. Sink, N. Zidar, A. Fic, C. C. Martel, A. Dessen, D. Patin, D.
ˇ ˇ
Blanot, M. M. Premru, S. Gobec, A. Zega, D. Kikelj, L. P. Masic, ACS Med.
Chem. Lett. 2012, 3, 626–630.
[58] J. G. Hurdle, A. J. O’Neill, I. Chopra, R. E. Lee, Nat. Rev. Microbiol. 2011, 9,
62–75.
[59] C. S. Lunde, et al., Antimicrob. Agents Chemother. 2009, 53, 3375–3383.
[60] D. M. McGrath, et al., Proc. Natl. Acad. Sci. USA 2013, 110, 3477–3482.
[61] R. F. Epand, J. E. Pollard, J. O. Wright, P. B. Savage, R. M. Epand, Antimi-
crob. Agents Chemother. 2010, 54, 3708–3713.
[62] T. D. Claridge, M. Gianella-Borradori, M. G. Moloney, unpublished results.
[63] E. C. Rivers, R. L. Mancera, Drug Discovery Today 2008, 13, 1090–1098.
[64] D. J. Newman, G. M. Cragg, J. Nat. Prod. 2012, 75, 311–335.
[65] M. Yoshida, C. Hedberg, M. Kaiser, H. Waldmann, Chem. Commun. 2009,
2926–2928.
[66] M. A. Koch, H. Waldmann, Drug Discovery Today 2005, 10, 471–483.
[67] R. Balamurugan, F. J. Dekkerab, H. Waldmann, Mol. BioSyst. 2005, 1, 36–
45.
[27] A. Schlenk, R. Diestel, F. Sasse, R. Schobert, Chem. Eur. J. 2010, 16,
2599–2604.
[28] Y.-C. Jeong, M. Anwar, Z. Bikadi, E. Hazai, M. G. Moloney, Chem. Sci.
2013, 4, 1008–1015.
[29] S. Peukert, Y. Sun, R. Zhang, B. Hurley, M. Sabio, X. Shen, C. Gray, J.
Dzink-Fox, J. Tao, R. Cebula, S. Wattanasin, Bioorg. Med. Chem. Lett.
2008, 18, 1840–1844.
[68] M. Pascolutti, R. J. Quinn, Drug Discovery Today 2014, 19, 215–221.
[69] R. J. Quinn, et al., J. Nat. Prod. 2008, 71, 464–468.
[30] Y.-C. Jeong, M. G. Moloney, Synlett 2009, 2487–2491.
[31] Y.-C. Jeong, M. G. Moloney, J. Org. Chem. 2011, 76, 1342–1354.
Received: March 31, 2014
Published online on May 18, 2014
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 1826 – 1837 1837