Elucidation of the active conformation of vancomycin dimers with antibacterial activity against vancomycin-resistant bacteria

Article abstract of DOI:10.1002/chem.201201211

Covalently linked vancomycin dimers have attracted a great deal of attention among researchers because of their enhanced antibacterial activity against vancomycin-resistant strains. However, the lack of a clear insight into the mechanisms of action of these dimers hampers rational optimization of their antibacterial potency. Here, we describe the synthesis and antibacterial activity of novel vancomycin dimers with a constrained molecular conformation achieved by two tethers between vancomycin units. Conformational restriction is a useful strategy for studying the relationship between the molecular topology and biological activity of compounds. In this study, two vancomycin units were linked at three distinct positions of the glycopeptide (vancosamine residue (V), C terminus (C), and N terminus (N)) to form two types of novel vancomycin cyclic dimers. Active NC-VV-linked dimers with a stable conformation as indicated by molecular mechanics calculations selectively suppressed the peptidoglycan polymerization reaction of vancomycin-resistant Staphylococcus aureus in vitro. In addition, double-disk diffusion tests indicated that the antibacterial activity of these dimers against vancomycin-resistant enterococci might arise from the inhibition of enzymes responsible for peptidoglycan polymerization. These findings provide a new insight into the biological targets of vancomycin dimers and the conformational requirements for efficient antibacterial activity against vancomycin-resistant strains. Squashing superbugs: Conformationally constrained vancomycin dimers that inhibit the peptidoglycan synthesis of vancomycin-resistant bacteria were prepared (see scheme). The potent antibacterial activity of the dimers was suggested to arise from their direct action on transglycosylase enzymes. Copyright

Full text of DOI:10.1002/chem.201201211

FULL PAPER
DOI: 10.1002/chem.201201211
Elucidation of the Active Conformation of Vancomycin Dimers with
Antibacterial Activity against Vancomycin-Resistant Bacteria
Jun Nakamura,[a] Hidenori Yamashiro,[b] Sayaka Hayashi,[a] Mami Yamamoto,[a]
Kenji Miura,[b] Shu Xu,[a] Takayuki Doi,[c] Hideki Maki,[b] Osamu Yoshida,[b] and
Hirokazu Arimoto*[a]
Abstract: Covalently linked vancomy-
cin dimers have attracted a great deal
of attention among researchers because
of their enhanced antibacterial activity
against vancomycin-resistant strains.
However, the lack of a clear insight
into the mechanisms of action of these
dimers hampers rational optimization
of their antibacterial potency. Here, we
describe the synthesis and antibacterial
activity of novel vancomycin dimers
with a constrained molecular confor-
mation achieved by two tethers be-
tween vancomycin units. Conforma-
tional restriction is a useful strategy for
studying the relationship between the
molecular topology and biological ac-
tivity of compounds. In this study, two
vancomycin units were linked at three
distinct positions of the glycopeptide
(vancosamine residue (V), C terminus
(C), and N terminus (N)) to form two
types of novel vancomycin cyclic
dimers. Active NC-VV-linked dimers
with a stable conformation as indicated
by molecular mechanics calculations
selectively suppressed the peptidogly-
can polymerization reaction of vanco-
mycin-resistant Staphylococcus aureus
in vitro. In addition, double-disk diffu-
sion tests indicated that the antibacteri-
al activity of these dimers against van-
comycin-resistant enterococci might
arise from the inhibition of enzymes re-
sponsible for peptidoglycan polymeri-
zation. These findings provide a new
insight into the biological targets of
vancomycin dimers and the conforma-
tional requirements for efficient anti-
bacterial activity against vancomycin-
resistant strains.
Keywords: antibiotics · conforma-
tional restriction · dimers · struc-
ture–activity relationships · vanco-
mycin
Introduction
fore not effective against VRE and VRSA.[5] Covalently
linked vancomycin dimers have been extensively studied
over the past decade for their antibacterial properties
(Figure 1)[6–8] against vancomycin-resistant bacteria. As our
recent studies have suggested that monomeric and dimeric
analogues of vancomycin might have different inhibitory ac-
tions on cell-wall biosynthesis in vancomycin-resistant
strains,[9,10] it is believed that further investigation of vanco-
mycin dimers might lead to a new class of antibacterial
agents. As the molecular mode of action of dimers is largely
unknown, it would be of great interest to shed light on how
the two vancomycin units cooperatively contribute to the
optimal interaction with target molecules and induce anti-
bacterial activity. All reported vancomycin dimers have a
linker that joins two vancomycin units, and thus have con-
siderable conformational flexibility. This makes it difficult to
predict the geometry of each unit required for activity.
Conformational restriction is a useful strategy for studying
the relationship between the molecular topology and biolog-
ical activity of compounds. In general, dimeric ligands with
carefully designed length and flexibility of the linker have
enhanced affinity for target molecules. Indeed, flexibility of
the linker allows the dimeric ligand to fit easily into the
binding site during complex formation (Figure 2A).
Vancomycin-resistant enterococci (VREs)[1] and -Staphylo-
coccus aureus (VRSA)[2] are major nosocomial pathogens
worldwide, and as such there is an urgent need for novel an-
tibiotics effective against VRE/VRSA infections.[3,4] As
VRE and VRSA have a d-alanyl-d-lactate (d-Ala-d-Lac)
terminal in their modified cell-wall precursors instead of a
d-Ala-d-Ala terminal (vancomycin binding motif), vanco-
mycin cannot bind to these modified precursors and is there-
[a] Dr. J. Nakamura, S. Hayashi, M. Yamamoto, Dr. S. Xu,
Prof. Dr. H. Arimoto
Graduate School of Life Sciences
Tohoku University, 2-1-1 Katahira
Aoba-ku, Sendai 980-8577 (Japan)
Fax : (+81)0-22-217-6204
E-mail: arimoto@biochem.tohoku.ac.jp
[b] H. Yamashiro, K. Miura, Dr. H. Maki, O. Yoshida
Discovery Research Laboratories
Shionogi & Co., Ltd., 3-1-1 Futaba-cho
Toyonaka, Osaka 561-0825 (Japan)
[c] Prof. Dr. T. Doi
Graduate School of Pharmaceutical Sciences
Tohoku University, 6-3 Aza-aoba
Aramaki, Aoba-ku, Sendai 980-8578 (Japan)
However, it should be noted that the high degree of con-
formational freedom of flexible dimers in solution becomes
highly constrained upon complex formation, thereby making
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201201211.
Chem. Eur. J. 2012, 00, 0 – 0
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
1
&
ÞÞ
These are not the final page numbers!

is required to optimize the biological activity constrained
dimers, especially when the target molecule is unknown or
the detailed three-dimensional structure of the binding site
is not available. Here, we describe the synthesis and antibac-
terial activity of novel vancomycin dimers with constrained
molecular conformation achieved by two tethers between
vancomycin units (cyclic dimers). These analogues provide
important new information on the topological requirements
for optimum dimer-target interaction and efficient antibacte-
rial activity against vancomycin-resistant strains.
Results and Discussion
Synthesis of vancomycin constrained dimers: In this study,
two vancomycin units were linked at three distinct positions
of the glycopeptide (vancosamine residue (V), C terminus
(C), and N terminus (N)) to form two types of novel vanco-
mycin cyclic dimers (Schemes 1 and 2). To minimize the syn-
thetic steps required, an extensive protection–deprotection
strategy was avoided. However, macrocyclization of the
dimers remained challenging. We have already shown that
oxidative coupling of two ortho-aminophenols is a mild but
powerful strategy to connect two vancomycin units by a
phenoxazone linker without protecting their functional
groups.[8] Thus, in this study, we first tethered two vancomy-
cin units by an alkyl chain, and then constructed a more
challenging second VV link by oxidative coupling. Synthesis
of the CC-VV-linked dimers 5 and 6 commenced with re-
ductive alkylation of the vancosamine nitrogen atom, which
afforded compound 2.[8] Two units of compound 2 were con-
nected by means of the C6- or C10-n-alkylenediamine by
using O-(benzotriazol-1-yl)-N,N,N’,N’-tetramethyluronium
hexafluorophosphate (HBTU).[12]
Figure 1. Structure of vancomycin and VV-linked vancomycin dimer 1.[8]
Vancomycin units are joined by a phenoxazone linker.
Palladium-catalyzed reduction of the CC-linked dimer 3
or 4 liberated two aminophenol moieties at the sugar resi-
due. These aminophenols were connected at their terminals
by oxidation using p-benzoquinone.[8] The resulting CC-VV-
linked cyclic dimers 5 and 6 had a 56- and a 60-membered
macrocycle, respectively, in the molecules. Synthesis of the
NC-VV-linked compound 12 began with modification of 2
at the carboxylate terminal with N,N-dimethyl-1,3-propane-
diamine to give intermediate 7. Reductive alkylation of 7 at
the N-terminal by an N-Fmoc-protected (Fmoc=9-fluore-
nylmethoxycarbonyl) aliphatic aldehyde, followed by depro-
tection of the Fmoc group by piperidine to give 8. Conden-
sation of 8 at its least-hindered primary amine with 2 by
using (benzotriazol-1-yloxy)tripyrrolidinophosphonium hex-
afluorophosphate (PyBOP)[7] gave the NC-linked dimer 10.
Finally, the NC-VV-linked cyclic dimer 12 that contained a
56-membered macrocycle was successfully synthesized by
means of a reduction–oxidation sequence as mentioned for
the syntheses of 5 and 6. Another NC-VV-linked dimer
(compound 13) was also prepared by a similar procedure.
As the phenoxazone linker is asymmetrical, dimers 5, 6, 12,
and 13 were obtained as mixtures of diastereomeric isomers.
These isomers could be separated only for 12 (Figure 3) and
Figure 2. Interaction of dimeric ligand with a macromolecular target.
A) A dimeric ligand with one linker can adjust its conformation during
complex formation. B) Ligands with two linkers can form a complex with
the target only when two units are properly fixed.
the dimer susceptible to a substantial entropic penalty upon
binding.[11] Such an entropic penalty can be minimized by
imposing appropriate structural constraints on the ligand
(Figure 2B, top). However, constrained cyclic dimers with
inappropriately folded conformation (Figure 2B, bottom)
fail to bind to their target molecules. Thus, substantial effort
&
2
&
www.chemeurj.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 0000, 00, 0 – 0
ÝÝ
These are not the final page numbers!

Vancomycin Dimers
FULL PAPER
Scheme 1. Synthesis of CC-VV-linked cyclic dimers (DIPEA=N,N-diisopropylethylamine).
and 12-2; 13-1 and 13-2 exhibit-
ed similar antibacterial proper-
ties.
Activity of NC-VV dimers
against VREs and VRSA: The
antibacterial activity of the pre-
pared constrained vancomycin
dimers against selected strains
is shown in Table 1. Because all
the dimers listed in the Table 1
have a common phenoxazone
linker, the observed differences
in antibacterial activity are due
to the second tethering by a
polymethylene chain. The NC-
VV-cyclic dimers 12 and 13 in-
Figure 3. Isomer of the NC-VV-linked dimer 12.
13 by high-performance liquid chromatography (HPLC).
This is probably because dimers 5 and 6 have pseudo-meso
symmetry and properties that resemble those of the diaster-
eomers. As shown in the table below, diastereomers 12-1
hibited the growth of vancomycin-susceptible and -resistant
strains with a potency similar to that of the VV-linked dimer
1. However, the CC-VV-linked dimers 5 and 6 did not inhib-
it growth of vancomycin-resistant E. faecium or S. aureus, al-
Chem. Eur. J. 2012, 00, 0 – 0
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
&
3
&
ÞÞ
These are not the final page numbers!

H. Arimoto et al.
Scheme 2. Synthesis of NC-VV-linked cyclic dimers (HOBt=1-hydroxybenzotriazole).
though they had certain activity against vancomycin-suscep-
tible strains. This finding emphasizes the importance of the
conformational constraint in the loss of anti-VREs/VRSA
activity of compound 5. The constraint in 5 probably prohib-
ited vancomycin units from being in the three-dimensional
arrangement required for anti-VREs/VRSA activity. The
moderate but slightly stronger anti-VREs/VRSA activity of
6 than that of 5 might be due to the longer linker length of
6, which allows increased conformational freedom at the C
terminals (a similar linker length/activity relationship was
seen between vancosamine terminals; see Table S1 in the
Supporting Information). As shown in Figure 2B, anti-
VREs/VRSA activity requires simultaneous dimeric interac-
tion of appropriately oriented ligands with the target mole-
cule.
&
4
&
www.chemeurj.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 0000, 00, 0 – 0
ÝÝ
These are not the final page numbers!

Vancomycin Dimers
FULL PAPER
Table 1. Antibacterial activity of the synthesized vancomycin dimers against selected strains.
vancomycin
((9.6Æ
1.3) mgmLÀ1) in d-Ala-d-Ala
substrate. These findings under-
line the importance of molecu-
lar topology in vancomycin
dimersꢁ inhibition of peptido-
glycan polymerization in bacte-
rial resistant models.
[a]
MIC [mgmLÀ1
S. aureus[d]
VR[h]
]
Compound
Linker
type
S. aureus[b]
E. faecium[c]
VS
E. faecium[e]
VR
E. faecium[f]
VR
VS[g]
A
N
ACHTUNGTRENNUNG
vancomycin
1
5
6
12-1
12-2
13-1
13-2
–
VV
1
2
8
4
2
2
2
4
1
0.5
2
1
1
1
2
2
>64
4
>64
16
2
>64
4
>64
32
4
CC-VV
CC-VV
NC-VV
NC-VV
NC-VV
NC-VV
>64
8
2
8
2
4
4
8
8
8
8
8
Externally added d-Ala-d-Lac
does not antagonize the anti-
VRE activity of vancomycin
dimers in the double-disk diffu-
[a] Minimum inhibitory concentrations were determined by the broth microdilution method. [b] RN4220.
[c] SR16972. [d] HIP11983. [e] SR7940. [f] SR23598. [g] Vancomycin-susceptible. [h] Vancomycin-resistant.
Vancomycin dimersꢀ inhibition of peptidoglycan synthesis by
d-Ala-d-Lac-containing precursors: The molecular mecha-
nism by which VREs or VRSA develop resistance to vanco-
mycin is well documented.[13,14] The d-Ala-d-Ala terminal of
bacterial cell-wall intermediates is replaced with d-Ala-d-
Lac in these resistant strains. Because vancomycin binds
weakly to the d-Ala-d-Lac terminal, it does not affect the
cell-wall biosynthesis of VREs and VRSA.
Studies to elucidate the effects of dimers on bacterial cell-
wall synthesis have been conducted only recently. In this
context, we have previously reported the first experimental
proof that covalently linked vancomycin dimers selectively
suppress cell-wall synthesis in vancomycin-resistant bacteria
by inhibiting peptidoglycan polymerization catalyzed by
Figure 4. Late stage of bacterial peptidoglycan biosynthesis. Lipid inter-
mediate (LI) synthetic steps catalyzed by a sequence of enzyme.
transglycosylase
(penicillin-binding
protein
2,
or
sion test: We next focused on whether the NC-VV dimer re-
quires binding to the peptidoglycan precursor for anti-VRE/
VRSA activity.
PBP2).[9,15,16] In this study, we shifted our attention to the ef-
ficacy of the constrained dimers (5 and 12). As shown in
Table 2, the NC-VV dimers 12-1 and 12-2, which exhibited
potent antibacterial activity against VREs and VRSA, inhib-
ited polymerization of peptidoglycan (Figure 4, PG step)
more potently ((92Æ38) and (120Æ23) mgmLÀ1) than the
VV-linked dimer 1 ((400Æ220) mgmLÀ1). This enhanced in-
hibitory activity may be due to a reduced entropic cost of
association between 12 and its target molecule as a result of
conformational restriction. The CC-VV-linked dimer 5,
which exhibited antibacterial activity only against vancomy-
cin susceptible strains, did not inhibit cell-wall synthesis in
the resistant d-Ala-d-Lac model (>950 mgmLÀ1). However,
dimer 5 strongly suppressed the peptidoglycan polymeriza-
tion ((9.4Æ1.6) mgmLÀ1) with a potency similar to that of
Vancomycin dimers were originally designed to potently
bind d-Ala-d-Lac residues in cell-wall intermediates of re-
sistant bacteria through dimeric interactions.[12,17,18] As de-
scribed above, the vancomycin dimers prepared in this study
showed good activity against VRE/VRSA. However, this
antibacterial activity might not have necessarily originated
from binding to the d-Ala-d-Lac residue. Accordingly, it is
fair to assume that d-Ala-d-Lac is just a hypothetical target
of covalently linked vancomycin dimers.[19]
To investigate whether the vancomycin dimers substantial-
ly enhanced the binding affinity for d-Ala-d-Lac, we carried
out a double-disk diffusion test (Figure 5).
Disks that contained vancomycin, linezolid, compound 1,
and compound 12 were placed on a lawn of either
vancomycin-susceptible E. faecium (SR16972) or
Table 2. Inhibitory activity of test compounds against cell-wall synthesis in an in vitro
assay.
vancomycin-resistant E. faecium (vanA, SR7940).
The effect of d-Ala-d-Ala-containing tripeptide or
d-Ala-d-Lac-containing tridepsipeptide on inhibi-
tion zones was evaluated. Competitive binding of
the externally added tripeptide or depsipeptide
with drugs can be visualized as a distortion in the
usually circular inhibition zone, if bacterial cell-wall
intermediates (d-Ala-d-Ala- or d-Ala-d-Lac termi-
nal) are the binding target of the drugs. The anti-
bacterial activity of linezolid is due to the inhibition
IC80 [mgmLÀ1
]
[a]
Compound
Linker
type
Susceptible model
Resistant model
LI steps
LI steps
PG step
PG step
vancomycin
–
240Æ8.2
130Æ57
>720
9.6Æ1.3
6.9Æ3.0
9.4Æ1.6
12Æ2
3900Æ120
>1024
>1024
770Æ240
400Æ220
>950
1
VV
5
12-1
12-2
CC-VV
NC-VV
NC-VV
260Æ8
380Æ120
92Æ38
300Æ46
17Æ0.8
670Æ110
120Æ23
[a] 80% inhibitory concentration.
Chem. Eur. J. 2012, 00, 0 – 0
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
&
5
&
ÞÞ
These are not the final page numbers!

H. Arimoto et al.
Molecular topology of constrained dimer 12: A pharmaco-
phore is a simple model that describes distances between
important atoms or groups of atoms in a ligand that are re-
quired for it to interact with a specific biological macromo-
lecule. Once the pharmacophore for a specific receptor is
defined, it can be used to search chemical libraries for com-
pounds that fit the pharmacophore. Although the binding
target of NC-VV-linked dimer remains to be elucidated, mo-
lecular topology analysis of the constrained dimer could pro-
vide useful insight into the substructures involved in interac-
tions with the target molecule in vancomycin-resistant bac-
teria.
To obtain information about the molecular topology of
active dimers, a stable conformation of the cyclic dimer 12
was investigated in silico using MacroModel 9.1.[21] After
drawing the structure of compound 12, its potential energy
was minimized and used for the subsequent conformational
search. All conformers within 24 kcalmolÀ1 of the structure
with the lowest energy could be classified into a group that
was defined on the basis of the topology of the aglycon moi-
eties of vancomycin units. The most stable generated confor-
mation of cyclic dimer 12 is depicted in Figure 6.[22] In this
conformer, two vancomycin units exist in a “back-to-back”
configuration with an antiparallel b structure forming at the
“back” of each unit. This constrained structure highlighted
the importance of the back-to-back arrangement in the anti-
Figure 5. Double-disk diffusion test performed using vancomycin-suscep-
tible E. faecium SR16972 (A and C) and vancomycin-resistant E. faecium
SR7940 (B and D). Distortion of drugs-inhibition zones by d-Ala-d-Ala-
containing tripeptide (Ac2-l-Lys-d-Ala-d-Ala, 100 mg, A and B) and d-
Ala-d-Lac-containing tridepsipeptide (Ac2-l-Lys-d-Ala-d-Lac, 100 mg, C
and D). Key: V, vancomycin (10 mg per disk); L, linezolid (10 mg per
disk); 1, vancomycin dimer 1 (50 mg per disk); 12, vancomycin cyclic
dimer 12–1 (80 mg per disk); AA, d-Ala-d-Ala-containing tripeptide; AL,
d-Ala-d-Lac-containing tridepsipeptide. See Table 1 for the antibacterial
activity of test compounds (1, 12, and vancomycin) against E. faecium
SR16972 and SR7940. In general, the size of the inhibition zone in the
disk diffusion test does not always correlate with minimum inhibitory
concentration (MIC) values obtained by the broth microdilution method
because of differences in the physical properties of compounds.
of protein synthesis,[20] and thus it is not affected by the ex-
ternal addition of the tripeptide or the depsipeptide (used as
negative control). Although all vancomycin derivatives
showed clear distortion of the zones with external addition
of the d-Ala-d-Ala-containing peptide, they did not show
any detectable distortion with addition of the d-Ala-d-Lac-
containing depsipeptide. If vancomycin dimers 1 and 12 ex-
hibit anti-VRE activity by binding to the d-Ala-d-Lac termi-
nal of bacterial cell-wall intermediates, the external addition
of d-Ala-d-Lac-containing depsipeptide should have com-
petitively inhibited this activity (Figure 5C,D). Overall, this
double-disk diffusion test indicates that the binding target of
vancomycin dimers is not the d-Ala-d-Lac terminal of the
cell-wall intermediates of VRE/VRSA. As dimers 1 and 12
inhibit peptidoglycan polymerization catalyzed by transgly-
cosylase (penicillin-binding protein 2 of S. aureus) without
binding to the d-Ala-d-Lac-containing peptidoglycan precur-
sor, a plausible anti-VRSA mechanism might be direct inter-
action of these dimers with the enzyme (see Figure 4).
Figure 6. Stereoview of A) conformationally stable NC-VV-linked dimer
12, and B) its chemical structure. Red and blue: two vancomycin units;
green: phenoxazone bridge in the VV position, and alkyl chain.
&
6
&
www.chemeurj.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 0000, 00, 0 – 0
ÝÝ
These are not the final page numbers!

Vancomycin Dimers
FULL PAPER
sylase enzymes. These findings
would promote the rational mo-
lecular design of better antibac-
terial agents.
Experimental Section
General methods: Reagents and sol-
vents were used as purchased from
commercial suppliers. All reactions
were carried out at room temperature
unless otherwise noted. Mass spectra
were obtained using a JEOL LMS-700
mass spectrometer (FAB MS) and a
Bruker Daltonics microTOF-TSfocus
(LC-ESI-TOF MS). LC was per-
formed using an Agilent 1100 Series,
with an Imtakt UNISON UK 3C8
column (150ꢂ2 mm) under the follow-
Figure 7. Des-N-methyl leucyl-NC-VV-linked cyclic dimer 14.
ing
conditions:
flow
rate
of
bacterial activity 12. The conformation of 12 also suggests
that the VV-linked dimer 1 (with one tether) is in the back-
to-back conformation, which is a target-bound state. This in-
terpretation can be supported by the fact that vancomycin
and the related glycopeptide antibiotics (without tethering)
have a propensity to form a back-to-back dimer through hy-
drogen bonds between aglycons in an antiparallel arrange-
ment in the crystal (Figure S1 in the Supporting informa-
tion).[23,24] In the back-to-back conformation of 12, leucine
residues and carboxylate terminals of vancomycin are proxi-
mate to each other. This proximity might favor simultaneous
binding (Figure 2B) of an N-terminal and a C-terminal of
vancomycin units with the target molecule in vancomycin-
resistant bacteria.
To gain further insight into the dimer–target interaction,
an N-methyl leucine residue of dimer 12 was removed by
Edman degradation to give 14 (Figure 7). This removal sig-
nificantly diminished the antibacterial activity against vanco-
mycin-resistant strains (minimum inhibitory concentration
(MIC) for S. aureus HIP11983>64 mgmLÀ1; see Table S2 in
the Supporting information). Collectively, these findings
stress the involvement of the N-methyl leucine residue at
the uncrosslinked terminal in the dimer–target interaction
required for anti-VRSA/VRE activity.
0.2 mLminÀ1, 65% to 95% acetonitrile/water containing 0.1% formic
acid, duration of 5 min, temperature of 408C, and UV detection at l=
214, 254, 280, and 190–400 nm. Monoisotopic mass was used for calcula-
tion of exact mass (C 12.0000, H 1.0078, O 15.9949, N 14.0031, Cl
35.9689). NMR spectra were recorded using
a JEOL ECA600 at
600 MHz, Varian Unity INOVA 500 at 500 MHz, and an INOVA 600 in-
strument at 600 MHz. The 1H NMR spectra are relative to the residual
solvent peak of tetramethylsilane (d=0 ppm), CHCl3 (d=7.26 ppm),
CHD2OD (d=3.31 ppm), (CHD2)ACTHNUTRGNE(UNG CD3)SO (d=2.5 ppm), and water (d=
3.33 ppm in [D6]DMSO); 13C NMR spectra are relative to the residual
solvent peak of CHCl3 (d=77.16 ppm). TLC was performed using Merck
silica gel 60 F254 precoated plates (0.25 mm). Flash chromatography was
carried out using 60–230 mesh silica gel (Kanto Chemical, silica gel 60N).
Analytical reversed-phase HPLC was performed using a 150ꢂ4.6 mm
column (Nacalai Tesque, Cosmosil 5C18-AR-II) under the following con-
ditions: flow rate of 1.0 mLminÀ1, 15 to 100% acetonitrile/water contain-
ing 0.1% trifluoroacetic acid (TFA), duration of 10 min, temperature of
308C, and UV detection at l=280 nm.
Compound 5: Compound 3 (59 mg, 0.017 mmol) in methanol (3 mL) was
reduced for 26 h with hydrogen in the presence of 10% Pd/C catalyst
(60 mg) to give the corresponding aminophenol intermediate. The Pd/C
catalyst was removed by filtration by using a membrane filter (Millipore,
pore size 1.0 mm, JAWP04700) and rinsed with methanol (12 mL) at 08C
(ice–water bath). A solution of p-benzoquinone (11 mg, 0.098 mmol) in
methanol (3 mL) was added to the filtrate, and the reaction solution was
stirred at 08C for an additional 1.5 h in the dark. The resulting crude
sample was purified by reverse-phase HPLC (Nacalai Tesque, Cosmosil
5C18 AR-II-waters, 250ꢂ10 mm, acetonitrile/water containing 0.1% TFA,
flow rate 4 mLminÀ1) and lyophilized overnight to yield compound 5 as
an orange solid (4.1 mg, 1.27 mmol, 7.5% yield, >99% purity). tR =
3.4 min; 1H NMR (600 MHz, [D6]DMSO and one drop of D2O, 40 8C):
d=9.01 (s, 1H), 8.65 (s, 1H), 8.59 (s, 1H), 8.54 (s, 1H), 8.44 (s, 1H), 8.41
(s, 1H), 7.88 (s, 1H), 7.84 (d, J=7.2 Hz, 1H), 7.74 (s, 1H), 7.68 (d, J=
7.2 Hz, 1H), 7.62–7.51 (comp, 4H), 7.49 (d, J=7.6 Hz, 2H), 7.37 (d, J=
7.9 Hz, 1H), 7.33 (d, J=7.9 Hz, 1H), 7.28 (t, J=8.6 Hz, 1H), 7.21 (t, J=
8.6 Hz, 1H), 7.12 (d, J=7.9 Hz, 2H), 6.76 (t, J=8.6 Hz, 2H), 6.72 (t, J=
8.6 Hz, 2H), 6.36 (s, 2H), 6.25 (s, 1H), 6.22 (s, 1H), 5.73–5.7 (comp, 2H),
5.64 (t, J=11 Hz, 1H), 5.48 (q, J=6.5 Hz, 1H), 5.25 (s, 1H), 5.22 (s, 1H),
5.17 (s, 1H), 4.98 (s, 1H), 4.9 (s, 1H), 4.67 (s, 2H), 4.46 (s, 2H), 4.39 (s,
1H), 4.34 (t, J=4.4 Hz, 1H), 4.28 (s, 1H), 4.22 (s, 2H), 3.3 (t, J=8.6 Hz,
2H), 3.16 (s, 3H), 2.64 (s, 2H), 2.62 (s, 1H), 2.28 (s, 3H), 2.15 (d, J=
9.7 Hz, 1H), 1.87 (d, J=13 Hz, 1H), 1.69–1.57 (comp, 8H), 1.49 (m, 1H),
1.4 (s, 1H), 1,32–1.06 (comp, 8H), 0.91 (d, J=4.9 Hz, 6H), 0.87 ppm (d,
Conclusion
By using a mild phenoxazone-based macrocyclization, we
have prepared conformationally constrained vancomycin
dimers that exhibit excellent inhibitory activity on the pepti-
doglycan synthesis of vancomycin-resistant bacteria. Evalua-
tions of this antibacterial activity along with molecular me-
chanics calculations allowed us to identify the active confor-
mation of vancomycin dimers necessary to ensure activity
against vancomycin-resistant stains. We have also showed in
this study that the potent antibacterial activity of vancomy-
cin dimers is likely to arise from their action on transglyco-
J=5.5 Hz, 6H); HRMS (LC-ESI): m/z calcd for C152H172Cl4N22O48
:
1606.5224 [M+2H]2+; found: 1606.5167.
Chem. Eur. J. 2012, 00, 0 – 0
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
&
7
&
ÞÞ
These are not the final page numbers!

H. Arimoto et al.
Compound 12: Compound 10 (0.40 g, 0.11 mmol) was dissolved in metha-
nol (18 mL). Pd/C catalyst (0.30 g) was added under an argon atmos-
phere, which was subsequently replaced by a hydrogen atmosphere. The
mixture was then reduced for 18 h in the dark to give the corresponding
aminophenol intermediate. The palladium catalyst was removed by filtra-
tion by use of a Celite filter (Kanto Chemical, Hyflo Super-Cel), rinsed
with water (82 mL), and the resulting filtrate was treated with a freshly
prepared solution of p-benzoquinone (36 mg, 0.33 mmol) in methanol
(10 mL). The reaction mixture was stirred at room temperature for
70 min in the dark, at which point methanol was removed under vacuum,
and the aqueous solution was frozen at À788C (dry ice–methanol bath)
for 2 h, and then lyophilized overnight. The resulting crude solid was pu-
rified by reverse-phase column chromatography (Yamazen, Ultra Pack
ODS-S-50B, 300 mmꢂ26 mm, 50 mm particle size, 15, 20, 25, 30, 35, 50,
80% methanol/water containing 0.1% TFA), and then lyophilized over-
night to yield compound 12-1 and 12-2 (TFA salts) as a red orange solids
(12-1: 38 mg, 11% yield, 95% purity, tR =2.9 min; 12-2: 45 mg, 13%
yield, 94% purity, tR =3.1 min). The counterion was changed as described
below. Toyo Pearl DEAE-650m (2 mL, 40–90 mm particle size) was first
preactivated by the washing the resin with 0.1m aqueous HCl (2 mL), dis-
tilled water, 0.5m aqueous NaOH (4 mL), distilled water, 50% aqueous
methanol (10 mL), methanol (20 mL), distilled water, 0.1m acetic acid
(8 mL), and distilled water (20 mL). A solution of compound 12-1 and
12-2 TFA salts (30 mg each) in methanol was then uploaded onto the
resin and eluted with methanol. Aqueous HCl (0.9 mL, 0.1m) was added
dropwise into the resulting solution, and the mixture was concentrated
under vacuum to remove methanol and lyophilized to yield compound
Semiquantitative enzyme inhibition assay: The inhibitory activity of the
synthetic compounds against enzyme reactions involved in bacterial cell-
wall biosynthesis in S. aureus was evaluated according to the procedure
of Miura et al.[10]
Double-disk diffusion test: After preincubation at 378C overnight in
brain–heart infusion broth (BHIB), the suspension of the test strains (E.
faecium SR16972 or SR7940) was diluted to an optical density of 0.1 at
660 nm (OD660
) and incubated at 378C for 3 h (SR16972) or 6 h
(SR7940). The bacterial suspension in exponential growth phase was
then diluted to an OD660 of 0.01. A portion of this suspension was spread
on a brain-heart infusion agar (BHIA) plate. After air-drying, paper
disks (Advantech, 6 mm in diameter, thin type) that contained test com-
pounds were put on the plate and incubated at 378C for 24 h.
Molecular mechanics calculations: Conformational analysis of the com-
pounds was performed by molecular mechanics calculations with Macro-
Model 9.1.[21] A three-dimensional structure of the cyclic dimer 12 was
built using the crystal structure of vancomycin obtained from the Protein
Data Bank (ID: 1fvm).[24] The potential energy of the cyclic compound
was minimized in the OPLS_2005 force field with water solvation until a
0.05 gradient convergence criterion was met. In the minimized structure,
two vancomycin units located away from each other were in a conforma-
tion recognized as “extended” topology. The global minimum conforma-
tion of 12 was calculated by the Low-Mode Conformational Search[25]
using the OPLS_2005 force field. Any structure with energy greater than
24 kcalmolÀ1 above the current global minimum was discarded. The max-
imum number of minimization with truncated Newton conjugate gradient
(TNCG) method was 500. This process generated 4973 conformers that
lay greater than 24 kcalmolÀ1 above the global minimum. All conformers
could be classified into a group defined based on the topology of the
aglycon moieties of vancomycin units. Molecules in this group folded
their two vancomycin units in a back-to-back manner as shown in Fig-
ure 6A. This conformation was stabilized in “folded” topology by eight
hydrogen bonds between two vancomycin units like the self-associated
state of noncovalently formed vancomycin dimer (Figure S1 in the Sup-
porting Information).[23]
12-1 and 12-2 (HCl salts) as
a red solid. Compound 12-1 (22 mg,
6.67 mmol, 96% purity): tR =3.8 min; 1H NMR (600 MHz, CD3OD,
208C): d=9.43 (s, 1H), 9.42 (s, 1H), 9.37 (s, 1H), 9.03 (s, 2H), 8.66 (s,
1H), 8.61 (s, 1H), 8.56 (s, 1H), 8.44 (s, 1H), 7.88 (t, J=7.8 Hz, 2H), 7.85
(comp, 8H), 7.75–7.52 (comp, 28H), 7.47 (d, J=8.4 Hz, 1H), 7.39–7.37
(comp, 4H), 7.34 (d, J=8.4 Hz, 2H), 7.31 (d, J=9 Hz, 2H), 7.13–7.06
(comp, 8H), 7.04 (t, J=7.8 Hz, 2H), 6.92 (d, J=7.8 Hz, 1H), 6.87 (t, J=
8.4 Hz, 4H), 6.71 (d, 7.8 Hz, 2H), 6.64–6.56 (comp, 10H), 6.51 (d, J=
8.4 Hz, 2H), 6.43–6.42 (comp, 5H), 6.4 (s, 2H), 6.39 (s, 1H), 6.32 (s, 2H),
6.28 (s, 2H), 6.26 (s, 1H), 6.23 (comp, 4H), 6.04 (d, J=7.8 Hz, 1H), 5.76
(s, 1H), 5.7 (s, 1H), 5.59 (s, 4H), 5.53 (s, 2H), 5.51 (s, 1H), 5.5 (s, 2H),
5.46 (s, 2H), 5.44 (s, 4H), 5.4–5.36 (comp, 8H), 5.34 (d, J=7.8 Hz, 1H),
5.24 (s, 2H), 5.19 (d, J=4.8 Hz, 1H), 4.81 (d, J=9 Hz, 1H), 4.68 (t, J=
6 Hz, 2H), 4.63 (s, 1H), 4.59 (dd, J=10.9, 4.3 Hz, 2H), 4.52 (t, J=6 Hz,
2H), 4.35 (d, J=11 Hz, 1H), 4.3 (d, J=11 Hz, 1H), 4.24–4.13 (comp,
10H), 4.07–3.99 (comp, 4H), 3.96–3.93 (comp, 2H), 3.91–3.81 (comp,
6H), 3.74 (t, J=8.4 Hz, 2H), 3.69 (t, J=8.4 Hz, 2H), 3.65–3.55 (comp,
6H), 3.49–3.32 (comp, 10H), 3.22–3.18 (comp, 8H), 3.11 (s, 3H), 3.07 (s,
3H), 3.01 (s, 3H), 2.96 (s, 6H), 2.93 (d, J=9 Hz, 6H), 2.88 (s, 3H), 2.81
(s, 3H), 2.78 (s, 3H), 2.7 (s, 1H), 2.68 (s, 1H), 2.43–2.31 (comp, 4H),
2.26–2.21 (comp, 4H), 2.09 (s, 2H), 2.04 (br, 10H), 1.91 (s, 3H), 1.88–1.84
(comp, 3H), 1.8 (s, 3H), 1.74 (d, J=8.4 Hz, 6H), 1.67 (d, J=6 Hz, 3H),
1.61 (m, 3H), 1.49–1.39 (comp, 8H), 1.29 (d, J=4.8 Hz, 3H), 1.25 (d, J=
6 Hz, 6H), 1.11 (d, J=6 Hz, 3H), 1.02 (d, J=6 Hz, 3H), 0.98 (d, J=
4.8 Hz, 6H), 0.89 (d, J=5.4 Hz, 3H), 0.82 ppm (d, J=5.4 Hz, 3H);
Acknowledgements
This study was supported in part by Grants-in-Aid for Scientific Re-
search, MEXT (nos. 17035039, 18032010, and 21310136), Japan; by the
Uehara Memorial Foundation; and by the Mochida Foundation. We are
grateful to Prof. Makoto Sasaki (Tohoku University) for use of LC-MS
to provide high- and low-resolution mass spectra.
[1] A. H. C. Uttley, C. H. Collins, J. Naidoo, R. C. George, Lancet 1988,
1, 57–58.
[2] D. M. Sievert, M. L. Boulton, G. Stoltman, D. Johnson, M. G. Sto-
bierski, F. P. Downes, P. A. Somsel, J. T. Rudrik, W. Brown, W.
Hafeez, T. Lundstrom, E. Flanagan, R. Johnson, J. Mitchell, S.
Chang, Morbid. Mortal. Weekly Rep. 2002, 51, 565–567.
[3] C. M. Crane, J. G. Pierce, S. S. F. Leung, J. Tirado-Rives, W. L. Jor-
gensen, D. L. Boger, J. Med. Chem. 2010, 53, 7229–7235.
[4] P.-A. Ashford, S. P. Bew, Chem. Soc. Rev. 2012, 41, 957–978.
[5] D. Kahne, C. Leimkuhler, W. Lu, C. Walsh, Chem. Rev. 2005, 105,
425–448.
[6] K. C. Nicolaou, R. Hughes, S. Y. Cho, N. Winssinger, C. Smethurst,
H. Labischinski, R. Endermann, Angew. Chem. 2000, 112, 3981–
3986; Angew. Chem. Int. Ed. 2000, 39, 3823–3828.
[7] J. H. Griffin, M. S. Linsell, M. B. Nodwell, Q. Chen, J. L. Pace, K. L.
Quast, K. M. Krause, L. Farrington, T. X. Wu, D. L. Higgins, T. E.
Jenkins, B. G. Christensen, J. K. Judice, J. Am. Chem. Soc. 2003, 125,
6517–6531.
HRMS (FAB): m/z calcd for
C157H182Cl4N23O48:
3297.1262 [M+H]+;
found: 3297.1284. Compound 12-2 (19 mg, 5.76 mmol, 98% purity): tR =
3.9 min; 1H NMR (600 MHz, CD3OD, 208C): d=9.48 (s, 2H), 7.84 (s,
6H), 7.76–7.63 (comp, 12H), 7.55 (t, J=7.8 Hz, 2H), 7.49 (d, J=7.8 Hz,
1H), 7.35 (s, 2H), 7.26 (d, J=7.8 Hz, 1H), 7.14–7.04 (comp, 10H), 7 (d,
J=7.8 Hz, 1H), 6.88–6.81 (comp, 6H), 6.62 (s, 4H), 6.59 (d, J=6 Hz,
4H), 6.51 (s, 1H), 6.47 (d, J=10.2 Hz, 4H), 6.41 (s, 3H), 6.35–6.29
(comp, 6H), 6.23 (s, 2H), 5.73 (s, 4H), 5.54–5.36 (comp, 22H), 5.25 (s,
1H), 5.19 (s, 2H), 4.36 (s, 2H), 4.24–4.07 (comp, 6H), 4.02 (s, 2H), 3.98–
3.84 (comp, 2H), 3.74 (d, J=8.4 Hz, 2H), 3.65–3.59 (comp, 6H), 3.41 (s,
6H), 3.4 (s, 3H), 3.18 (s, 3H), 3.04 (s, 3H), 2.99 (s, 1H), 2.95 (br, 2H),
2.92 (s, 6H), 2.88 (s, 3H), 2.78 (s, 6H), 2.2 (s, 1H), 2.16 (s, 3H), 2.1 (s,
2H), 2.04–2.01 (comp, 3H), 1.85 (s, 3H), 1.81 (s, 3H), 1.76–1.7 (comp,
4H), 1.57 (s, 2H), 1.47 (br, 4H), 1.28 (br, 4H), 1.1 (s, 3H), 1.03 (s, 3H),
0.97 (s, 6H), 0.89 (d, J=6 Hz, 3H), 0.82 ppm (s, 3H); HRMS (FAB): m/z
calcd for C157H182Cl4N23O48: 3297.1262 [M+H]+; found: 3297.1316.
[8] J. Lu, O. Yoshida, S. Hayashi, H. Arimoto, Chem. Commun. 2007,
251–253.
&
8
&
www.chemeurj.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 0000, 00, 0 – 0
ÝÝ
These are not the final page numbers!

Vancomycin Dimers
FULL PAPER
[9] O. Yoshida, J. Nakamura, H. Yamashiro, K. Miura, S. Hayashi, K.
Umetsu, S. Xu, H. Maki, H. Arimoto, Med. Chem. Commun. 2011,
2, 278–282.
[10] K. Miura, H. Yamashiro, K. Uotani, S. Kojima, T. Yutsudo, J. Lu, O.
Yoshida, Y. Yamano, H. Maki, H. Arimoto, Antimicrob. Agents Che-
mother. 2010, 54, 960–962.
[11] M. Mammen, S.-K. Choi, G. M. Whitesides, Angew. Chem. 1998,
110, 2908–2953; Angew. Chem. Int. Ed. 1998, 37, 2754–2794.
[12] U. N. Sundram, J. H. Griffin, J. Am. Chem. Soc. 1995, 60, 1102–
1103.
[13] S. Handwerger, M. J. Pucci, K. J. Volk, J. Liu, A. S. Lee, J. Bacteriol.
1992, 174, 5982–5984.
[18] K. C. Nicolaou, R. Hughes, S. K. Cho, N. Winssinger, H. Labischin-
ski, R. Endermann, Chem. Eur. J. 2001, 7, 3824–3843.
[19] R. K. Jain, J. Trias, J. A. Ellman, J. Am. Chem. Soc. 2003, 125, 8740–
8741.
[20] C. M. Perry, B. Jarvis, Drugs 2001, 61, 525–551.
[21] MacroModel, Version 9.1, Schrçdinger, LLC, New York, 2005.
[22] A stable conformation of CC-VV dimer 5 substantially different
from that of 12 was also investigated (see the Supporting Informa-
tion, Figure S2).
[23] U. Gerhard, J. P. Mackay, R. A. Maplestone, D. H. Williams, J. Am.
Chem. Soc. 1993, 115, 232–237.
[24] Y. Nitanai, T. Kikuchi, K. Kakoi, S. Hanamaki, I. Fujisawa, K. Aoki,
J. Mol. Biol. 2009, 385, 1422–1432.
[14] C. T. Walsh, S. L. Fisher, I. S. Park, M. Prahalad, Z. Wu, Chem. Biol.
1996, 3, 21–28.
[25] P. Labute, J. Chem. Inf. Model. 2010, 50, 792–800.
[15] H. F. Chambers, C. Miick, Antimicrob. Agents Chemother. 1992, 36,
656–661.
[16] J.-M. Ghuysen, Int. J. Antimicrob. Agents 1997, 8, 45–60.
[17] J. Rao, L. Yan, J. Lahiri, G. M. Whitesides, R. M. Weis, H. S.
Warren, Chem. Biol. 1999, 6, 353–359.
Received: April 10, 2012
Published online: && &&, 0000
Chem. Eur. J. 2012, 00, 0 – 0
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
&
9
&
ÞÞ
These are not the final page numbers!

H. Arimoto et al.
Squashing superbugs: Conformation-
ally constrained vancomycin dimers
that inhibit the peptidoglycan synthesis
of vancomycin-resistant bacteria were
prepared (see scheme). The potent
antibacterial activity of the dimers was
suggested to arise from their direct
action on transglycosylase enzymes.
Vancomycin
J. Nakamura, H. Yamashiro,
S. Hayashi, M. Yamamoto, K. Miura,
S. Xu, T. Doi, H. Maki, O. Yoshida,
H. Arimoto* .................. &&&& — &&&&
Elucidation of the Active Conforma-
tion of Vancomycin Dimers with Anti-
bacterial Activity against Vancomycin-
Resistant Bacteria
&
10
&
www.chemeurj.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 0000, 00, 0 – 0
ÝÝ
These are not the final page numbers!