Tobramycin and nebramine as pseudo-oligosaccharide scaffolds for the development of antimicrobial cationic amphiphiles

Article abstract of DOI:10.1002/chem.201406404

Antimicrobial cationic amphiphiles derived from aminoglycoside pseudo-oligosaccharide antibiotics interfere with the structure and function of bacterial membranes and offer a promising direction for the development of novel antibiotics. Herein, we report

Full text of DOI:10.1002/chem.201406404

DOI: 10.1002/chem.201406404
Full Paper
&
Antibiotics
Tobramycin and Nebramine as Pseudo-oligosaccharide Scaffolds
for the Development of Antimicrobial Cationic Amphiphiles
Yifat Berkov-Zrihen,[a] Ido M. Herzog,[a] Raphael I. Benhamou,[a] Mark Feldman,[a]
Kfir B. Steinbuch,[a] Pazit Shaul,[a] Shachar Lerer,[b] Avigdor Eldar,[b] and Micha Fridman*[a]
Abstract: Antimicrobial cationic amphiphiles derived from
aminoglycoside pseudo-oligosaccharide antibiotics interfere
with the structure and function of bacterial membranes and
offer a promising direction for the development of novel an-
tibiotics. Herein, we report the design and synthesis of cat-
ionic amphiphiles derived from the pseudo-trisaccharide
aminoglycoside tobramycin and its pseudo-disaccharide seg-
ment nebramine. Antimicrobial activity, membrane selectivi-
ty, mode of action, and structure–activity relationships were
studied. Several cationic amphiphiles showed marked anti-
microbial activity, and one amphiphilic nebramine derivative
proved effective against all of the tested strains of bacteria;
furthermore, against several of the tested strains, this com-
pound was well over an order of magnitude more potent
than the parent antibiotic tobramycin, the membrane-target-
ing antimicrobial peptide mixture gramicidin D, and the cat-
ionic lipopeptide polymyxin B, which are in clinical use.
Introduction
tive charge of the bacterial cell surface reduce affinity to CAPs,
and this type of resistance is prevalent amongst CAP-resistant
bacteria.[6,7] For example, the incorporation of d-alanyl units
into the negatively charged teichoic acids found in Gram-posi-
tive bacterial cell walls through an ester bond neutralizes the
cell wall charge due to the positively charged amine group of
the attached amino acid residue (Figure 1A).[8] In Gram-nega-
tive bacteria, a similar negative-charge reduction takes place
through the attachment of a positively charged 4-aminoarabi-
nopyranose to phosphate groups on the lipid A part of the lip-
opolysaccharides (LPS) that compose the outer leaflet of the
outer membrane (Figure 1B).[9,10] This reduction in the overall
negative charge results in CAP resistance. An additional mech-
anism of CAP resistance is based on molecular trapping: Bacte-
rial surface-associated and secreted proteins bind to CAPs and
prevent them from reaching the lipid bilayer and disrupting
the bacterial cell membrane.[6,11] A third type of resistance to
CAPs involves proteolytic degradation of the CAPs by prote-
ases or peptidases; however, this mechanism is ineffective
against cyclic CAPs or those rich in prolines or arginines.[12] Fi-
nally, although cationic antimicrobial peptides act on the cell
membrane, resistance to these antimicrobials is also associated
with the presence of active efflux proteins.[13]
Bacterial cell membranes offer important drug targets,[1] and
yet only a limited number of clinically approved antibiotics act
by affecting membrane structure and function.[2] Antimicrobial
agents capable of damaging the bacterial membrane should
maintain similar levels of efficacy during each bacterial cell
cycle stage and, hence, should prove useful for the treatment
of infections caused by slowly dividing bacteria. These infec-
tions are challenging to eradicate by using the currently avail-
able repertoire of antibiotics.[1] To date, cationic antimicrobial
peptides (CAPs) and non-cationic antimicrobial peptides are
the best characterized families of antimicrobial agents that act
by disrupting the structures and functions of bacterial mem-
branes. These peptides are produced by numerous organisms
from bacteria and fungi to mammals.[3,4] Natural antimicrobial
peptides such as the non-cationic peptide mixture gramici-
din D (GRM D) and the CAP gramicidin S have been approved
for clinical use and are frequently used for the treatment of ex-
ternal infections.[5]
Several mechanisms of resistance to CAPs have evolved in
bacteria. Chemical modifications that reduce the overall nega-
Inspired by the chemical structure of CAPs and in search of
strategies to overcome the limitations of resistance, safety, and
antimicrobial activity of CAPs, several types of synthetic antimi-
crobial cationic amphiphile have been developed in recent
years.[14–18] We and other groups have demonstrated that ami-
noglycoside antibiotics (AGs), which primarily act by perturb-
ing bacterial protein synthesis and are highly positively
charged under physiological conditions, can be used as scaf-
folds for the development of potent unnatural cationic antimi-
crobial amphiphiles.[19,20] Attachment of one or more hydro-
[a] Y. Berkov-Zrihen, I. M. Herzog, R. I. Benhamou, Dr. M. Feldman,
K. B. Steinbuch, P. Shaul, Dr. M. Fridman
School of Chemistry, Raymond & Beverly Sackler Faculty of Exact Sciences
Tel Aviv University, Ramat Aviv, Tel Aviv, 6997801 (Israel)
E-mail: mfridman@post.tau.ac.il
[b] Dr. S. Lerer, Dr. A. Eldar
Department of Molecular Microbiology and Biotechnology
Faculty of Life Sciences, Tel Aviv University
Ramat Aviv, Tel Aviv, 6997801 (Israel)
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/chem.201406404.
Chem. Eur. J. 2015, 21, 4340 – 4349
4340
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
phatic chains are structural parameters that affect the antimi-
crobial activities of these compounds (Figure 1C).[21–24]
The same structural parameters also affect the selectivity of
these compounds for bacterial rather than mammalian erythro-
cyte cell membranes. In general, all of the previously reported
AG-derived antimicrobial cationic amphiphiles exhibit potency
against a variety of Gram-positive bacteria; however, their effi-
cacy against Gram-negative strains was poor. An exception
was reported by Mingeot-Leclercq and co-workers. This group
developed several di- and tri-O-alkylated neamine-derived cat-
ionic amphiphiles (Figure 1D) that exhibit potent antimicrobial
activity against pathogenic Gram-negative bacteria with mini-
mum inhibitory concentration (MIC) values in the range 4–
16 mgmLꢀ1.[25,26]
We have sought to determine whether the antimicrobial-ac-
tivity spectrum against Gram-positive and Gram-negative bac-
teria can be associated with general molecular descriptors of
AG-derived antimicrobial cationic amphiphiles or whether the
activity depends on the specific AG scaffold. We prepared 23
cationic antimicrobial amphiphiles derived from tobramycin or
its pseudo-disaccharide segment nebramine (Figure 1B). Each
of the 23 amphiphilic AGs were biologically evaluated by
means of several biological assays, including antimicrobial-ac-
tivity tests and hemolysis tests, and the results were analyzed
to identify the determinants of antimicrobial activity and spe-
cificity.
Results and Discussion
Molecular design and synthesis
In designing the cationic amphiphiles for this study, we fo-
cused on varying the following parameters: the numbers of
hydrophobic residues, the types of hydrophobic residue, and
the hydrophobicity/hydrophilicity ratio. Tobramycin has five
amine and five alcohol groups, whereas its pseudo-disacchar-
ide fragment nebramine has only four amine and three alcohol
groups. Hence, etherification by similar aliphatic residues of all
five alcohol groups of tobramycin or all three alcohol groups
of nebramine will result in more hydrophobic cationic amphi-
philes in the case of tobramycin than nebramine. Tobramycin-
derived cationic antimicrobial amphiphiles 2a–i (Scheme 1A)
were prepared in three steps from the parent AG. Commercial-
ly available tobramycin was transformed into the penta-azido
Figure 1. A) Structure of teichoic acid segment bearing a d-alanyl modifica-
tion on a glycerol phosphate unit. B) LPS lipid A core structure modified by
a 4-amino-l-arabinose unit. C) Tobramycin and mono- and dialkylated tobra-
mycin derivatives that act as cationic antimicrobial amphiphiles. D) Neamine
and di- and tri-O-ether neamine-derived cationic antimicrobial amphiphiles
and the structure of nebramine.
tobramycin following
a
previously reported procedure
(Scheme 1A).[27] Etherification of all five alcohol groups of the
azide-protected tobramycin resulted in 1a–j in 68–99% yield
of the isolated products. Reduction of the azido groups of 1a–
i by using the Staudinger reaction gave the tobramycin-de-
rived antimicrobial cationic amphiphiles 2a–i in 77–98% yield
of the isolated products. The free amine form of 1j was not
tested due to the very poor solubility of the deprotected prod-
uct in aqueous media.
phobic residues to various AGs results in potent cationic anti-
microbial amphiphiles. We previously designed and synthe-
sized several subsets of cationic antimicrobial amphiphiles de-
rived from the AG tobramycin (1; Figure 1C) and linear aliphat-
ic chains. This pseudo-trisaccharide AG contains five amine
groups that are positively charged under physiological condi-
tions. Biological evaluation of amphiphilic tobramycin deriva-
tives indicated that the number and length of the linear ali-
phatic chains, the chemical group that connects the hydropho-
bic chains to the AG scaffold, and the positioning of the ali-
Heating 1a, 1c, and 1e–j to reflux in a 1.5m solution of
H2SO4 in methanol resulted in selective cleavage of the pro-
tected 3-deoxy-3-amino-d-glucose ring of the tobramycin
pseudo-trisaccharide to yield the corresponding pseudo-disac-
Chem. Eur. J. 2015, 21, 4340 – 4349
www.chemeurj.org
4341
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Scheme 1. Synthesis of tobramycin- and nebramine-derived cationic amphiphiles. Reagents and conditions: a) TfN3, ZnCl2, Et3N, H2O/MeOH/CH2Cl2 (3:10:3),
95%; b) alkyl bromide, alkyl chloride, or alkyl iodide; NaH; TBAI; DMF; 68–99%; c) 1.0m P(Me)3 in THF, H2O/THF (1:9), 0.1m NaOH; 77–99%; d) 1.5m H2SO4 in
MeOH, reflux, 40–96%; e) N,N-Bis-Boc-l-histidine N-hydroxysuccinimide ester, K2CO3, MeOH; 75%; f) 95% TFA, quantitative yield. TBAI=tetrabutylammonium
iodide, TFA=trifluoroacetic acid.
charide nebramine derivatives 3a, 3c, and 3e–j in 40–96%
yield of the isolated products (Scheme 1B).[28] Reduction of the
azide groups of compounds 3a, 3c, and 3e–j using the Stau-
dinger reaction conditions gave the desired subset of 4’,5-di-O-
ether nebramine derivatives 4a, 4c, and 4e–j in 79–99% yield
of the isolated products. An additional subset of nebramine-
derived cationic amphiphiles was generated by n-alkylation
(n=5–9 carbon atoms in the chain) of each of the three alco-
hol groups of tetra-azido nebramine 5 (Scheme 1C) to yield
5 f–j, which were deprotected to yield 4’,5,6-tri-O-n-alkyl nebr-
amine derivatives 6 f–j.
Finally, to test the effect of the number of positively charged
amine groups on the biological performance of AG-derived an-
timicrobial cationic amphiphiles, we treated 6h with N-
hydroxysuccinimide (NHS) ester-activated and di-Boc-protected
histidine (Boc-His(Boc)-NHS) to yield 7h (72% yield) with high
chemoselectivity to the 6’-amine position of 6h (Scheme 1D).
A short treatment of 7h with TFA resulted in 8h, the 6’-NH-
histidine analogue of 6h. Compound 6h has four positively
charged amine groups under physiological conditions, whereas
8h has four primary amine groups and a histidine imidazole
ring that are positively charged under physiological conditions.
Hence, biological evaluation of 6h and 8h enabled us to de-
termine the effects of the addition of the histidine unit on the
biological activity of these nebramine-derived cationic amphi-
philes. The synthetic antimicrobial cationic amphiphiles were
treated with TFA to yield the corresponding ammonium TFA
salts. The exception was 2b, which was TFA labile and there-
fore used in its free amine form. All of the compounds were
1
characterized by means of H and 13C NMR spectroscopic and
low-resolution (R) mass-spectrometric analysis. Proton assign-
ments were achieved by 1D TOCSY experiments. The molecular
weights of all the compounds that were subjected to biologi-
cal evaluation were confirmed by high-resolution (HR) mass-
Chem. Eur. J. 2015, 21, 4340 – 4349
www.chemeurj.org
4342
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
spectrometric analysis, and their purity was greater than 95%,
as determined by LCMS (see the Experimental Section and
Supporting Information).
Of the subset of tobramycin derivatives 2a–d containing the
substituted benzyl ether cationic amphiphiles, only the benzyl
ether derivative 2a and the 4-chlorobenzyl derivative 2c dem-
onstrated antimicrobial activity against all the tested Gram-
positive strains; the compounds in this subset were ineffective
against the tested Gram-negative strains (Table 1). Of the
subset of compounds 2e–i containing the tobramycin core
that were substituted with five linear alkyl ether chains, the
penta-O-n-pentyl derivative 2 f demonstrated the most potent
antimicrobial activity against the tested Gram-positive strains
(MIC=2–16 mgmLꢀ1); however, this compound was ineffective
against the Gram-negative strains. Moreover, of this subset,
compounds with alkyl chains longer or shorter than n-pentyl
were less effective antimicrobial agents than those with the n-
pentyl chains. MIC tests of the subset containing the 4’,5-di-O-
ether nebramine derivatives 4a, 4c, and 4e–j revealed that
only those compounds with linear n-alkyl ether chains longer
than six carbon atoms were potent against the tested Gram-
positive strains; again, none were active against the tested
Gram-negative strains. The di-n-nonyl derivative 4j was the
most potent antimicrobial agent of this subset with MIC values
of 4 mgmLꢀ1 against all of the tested Gram-positive bacterial
strains.
Antimicrobial activity
The MIC values of the 23 cationic amphiphiles against five
Gram-positive and five Gram-negative bacterial strains were
determined by using the standard double-dilution method
(Table 1).[21] The chosen bacteria represent a diverse selection
of pathogens with drug resistance to the parent AG tobramy-
cin and/or to the clinically used membrane-disrupting antibiot-
ics that were used as controls in this study, the non-cationic
antimicrobial peptide mixture GRM D and the cationic lipopep-
tide polymyxin B (PMX). Amongst the five Gram-positive
pathogens were the antibiotic-resistant and pathogenic methi-
cillin-resistant Staphylococcus aureus (MRSA, strain A) and
Staphylococcus aureus Cowan (strain B).
The panel of five Gram-negative bacteria included two
strains of Klebsiella pneumonia (strains G and H), which are en-
capsulated by negatively charged uronic acid-rich polysaccha-
rides and are therefore resistant to several CAPs.[10,29]
Finally, of the 4’,5,6-tri-O-n-alkyl ether nebramine subset 6 f–
j, the optimal antimicrobial activity was observed for the tri-O-
n-heptyl nebramine 6h, which had MIC values of 1–4 mgmLꢀ1
against all of the tested Gram-positive strains. Moreover, 6h
was the only compound with potent antimicrobial activity
against all of the tested Gram-negative strains (MIC=4–
8 mgmLꢀ1). Compound 8h, the more hydrophilic and positively
charged analogue of 6h, was ineffective against the tested
Gram-negative bacteria. Compound 8h had slightly reduced
efficacy against the tested Gram-positive bacteria relative to
the parent compound 6h.
Table 1. MIC values of evaluated compounds against Gram-positive (+)
and Gram-negative (ꢀ) strains.[a,b]
Cpd.
A
B
C
D
E
F
G
H
I
J
(+) (+) (+) (+)
(+) (ꢀ) (ꢀ) (ꢀ) (ꢀ) (ꢀ)
PMX
64
64
64
8
16
4
4
4
64
0.5
4
4
8
8
64
2
32
8
2
16 >64
>64 >64 >64 >64 >64
>64
2
4
GRM D >64 >64
1
>64
16
32 >64
8
>64 >64 >64 >64
>64
4
8
8
2
4
1
2a
2b
2c
2d
2e
2 f
2g
2h
2i
4a
4c
4e
4 f
4g
4h
4i
>64 >64 >64 >64 >64
>64 >64 >64 >64 >64
8
8
16 >64 >64 >64 >64 >64
64 >64 >64 >64 >64 >64
>64 >64 >64 >64 >64 >64
64
16
64
32
2
4
16
2
4
4
8
4
8
>64 >64
>64 64
32 >64 >64
64 64 64
Connection between LPS affinity and antimicrobial-activity
spectrum
>64 >64 >64 >64
>64 >64 >64 >64
>64 >64 >64 >64
64 >64 >64 >64 >64 >64
>64 >64 >64 >64 >64 >64
>64 >64 >64 >64 >64 >64
>64 >64 >64 >64 >64 >64
>64 >64 >64 >64 >64 >64
>64 >64 >64 >64 >64 >64
>64 >64 >64 >64 >64 >64
32 >64 >64 >64 >64 >64
Compound 6h was the only one that demonstrated potent an-
timicrobial activity against all five tested Gram-negative patho-
gens (Table 1). Therefore, we examined whether the potency
of 6h against Gram-negative bacteria was associated with an
affinity for LPS, which is a major component of the outer leaf-
let of the outer membrane of Gram-negative strains. Hence,
we initially tested the MIC values of 6h against the Gram-neg-
ative E. coli ATCC 25922 (strain G) and against the Gram-posi-
tive S. pyogenes glossy (strain C) in the presence of LPS in
a range of concentrations. The LPS was purified from E. coli (se-
rotype O111:B4) by means of ion-exchange chromatography
and evaluated for binding to a toll-like receptor (TLR). The anti-
Gram negative cationic lipopeptide antibiotic PMX that acts by
binding to the lipid A core of LPS was tested as a control. An
increase in LPS concentration led to an increase in the MIC
values of PMX against E. coli ATCC 25922 of up to eightfold
(Table 2). A similar effect was observed for 6h against both
tested bacteria. We tested the effect of added LPS on the anti-
microbial activity of 8h, the analogue of 6h that demonstrated
>64 >64
32
32
>64 >64 >64 >64
>64 >64 >64 >64
>64 >64 >64 >64
64 >64
32
4
8
4
4
8
4
4
16
4
16
4
64 >64 >64 >64 >64
>64 64 64 32 64
4j
6 f
6g
6h
6i
>64 >64
32
2
4
16
2
1
64 >64 >64 >64 >64 >64
8
2
4
2
4
2
64 >64
32
8
32 >64
4
8
4
8
8
4
4
4
16 >64 >64
32
32 >64
6j
8h
32
4
32
4
64
2
32
4
32 >64 >64 >64 >64 >64
32 64 32 >64 >64
4
[a] MIC values are presented in units of mgmLꢀ1. [b] Strain A: MRSA;
strain B: S. aureus Cowan ATCC 12598; strain C: S. pyogenes glossy;
strain D: S. pyogenes JRS75; strain E, S. epidermidis RP62A ATCC 35984;
strain F: E. coli ATCC 25922; strain G: K. pneumoniae K36; strain H; K. pneu-
moniae K21; strain I, P. aeruginosa PAO1; and strain J, P. aeruginosa ATCC
33347. All the experiments were performed in triplicate, and the results
were obtained from two independent experiments. MRSA=methicillin-re-
sistant S. aureus.
Chem. Eur. J. 2015, 21, 4340 – 4349
www.chemeurj.org
4343
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
strated that 6h and 8h interacted with LPS; hence, that in ad-
dition to LPS affinity, the anti-Gram negative activity of 6h is
likely to result from interactions with additional Gram-negative
membrane determinants.
Table 2. Effect of LPS on the MIC values of PMX, 6h, and 8h.[a]
Bacterial strain
E. coli (ATCC 25992)
S. pyogenes glossy
6h[b]
E. coli LPS[b]
6h[b]
PMX[b]
8h[b]
0
10
4
8
2
4
4
4
2
4
50
100
16
32
8
16
16
32
8
16
Evaluation of membrane selectivity
The selectivity of the AG-derived cationic amphiphiles for bac-
terial cell membranes relative to mammalian cell membranes
was evaluated by measuring the hemolytic effect of the com-
pounds on isolated rabbit or rat erythrocytes. Erythrocytes are
frequently used as models for the evaluation of mammalian
cell-membrane damage. Of the tobramycin derivatives 2a–d
that were substituted with aromatic ether functionalities,
benzyl ether derivative 2a, and 4-methoxybenzyl ether deriva-
tive 2b caused almost no measurable hemolysis at a concentra-
tion of 128 mgmLꢀ1 (2–4%; Table 4). In contrast, 4-chlorobenzyl
ether derivative 2c caused significant hemolysis in the same
range as the MIC value of this compound against several of
the tested Gram-positive strains (8ꢁ2% at 8 mgmLꢀ1). Of the
n-alkyl ether tobramycin derivatives 2e–i, those that were
potent antimicrobial agents were also highly hemolytic
(Table 4); for example, 2 f (MIC= 2–4 mgmLꢀ1 against four of
the five tested Gram-positive strains) caused 100% hemolysis
at a concentration of 64 mgmLꢀ1. A similar trend was observed
in the case of the 4’,5-di-O-ether nebramine subset 4a, 4c, and
[a] LPS was purified by ion-exchange chromatography from E. coli (sero-
type O111:B4) grown in BHI broth. Test compounds were preincubated
for 1 h at room temperature with LPS and then the bacteria were added.
The results were obtained after 18 h of incubation, and repeated in two
independent experiments. [b] Values are in units of mgmLꢀ1
.
potent antimicrobial activity against all of the tested Gram-
positive strains but not against the tested Gram-negative
strains. As observed for 6h, increased concentrations of LPS
led to an increase in the MIC value of 8h against S. pyogenes
glossy (Table 2).
These findings can be rationalized by interactions between
LPS and either PMX or 6h or 8h that effectively reduce the
concentration of the free test compound available to interact
with the bacterial cell membrane, thus leading to a decrease in
their antimicrobial activity.[30]
The fact that added LPS had a similar deactivating effect on
the antimicrobial activities of PMX and compounds 6h and 8h
suggested that these compounds interact with LPS regardless
of their potency against Gram-negative bacteria. To test this
hypothesis, the interactions between LPS and the ribosome-
targeting tobramycin (negative control), LPS-targeting PMX
(positive control), 6h, and 8h were quantitatively evaluated by
measuring the increase in fluorescence as a result of the com-
petitive displacement of the LPS binding fluorescent dye
BODIPY-cadaverine by the tested molecules.[31] As expected,
the parent ribosome-targeting AG tobramycin demonstrated
a very low LPS displacement effect. Interestingly, the maximal
displacement effects (Ymax) of both 6h, which demonstrated
anti-Gram-negative activity, and 8h, which was effective only
against Gram-positive strains, were significantly higher than
that of PMX (Table 3). There was little difference between the
EC50 values of 6h and 8h, but the EC50 value of PMX was ap-
proximately threefold lower than 6h and 8h (Table 3). The re-
sults of the competitive-displacement assay clearly demon-
Table 4. Hemolysis of rabbit erythrocytes [%] in the presence of the indi-
cated concentrations of the evaluated compounds.[a]
Concentration [mgmLꢀ1
]
Cpd.
128
0
64
32
16
8
4
0
2
0
PMX
GRM D
2a
2b
2c
2d
2e
2 f
2g
2h
2i
4a
4c
4e
4 f
4g
4h
4i
4j
6 f
6g
6h
6i
0
0
0
0
52ꢁ8
39ꢁ6
30ꢁ4
20ꢁ4
10ꢁ2 5ꢁ2 2ꢁ1
4ꢁ1
2ꢁ1
98ꢁ2
26ꢁ2
0
0
0
0
0
0
0
0
0
0
0
0
0
89ꢁ9
16ꢁ1
0
47ꢁ9
9ꢁ1
0
17ꢁ4
3ꢁ1
0
8ꢁ2 3ꢁ1 1ꢁ0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
100
100
67ꢁ13 30ꢁ13
3ꢁ1
99ꢁ1
70ꢁ1
14ꢁ2
1ꢁ0
2ꢁ0
0
99ꢁ1
43ꢁ10
8ꢁ2
1ꢁ1
1ꢁ1
0
89ꢁ4
19ꢁ4
4ꢁ1
1ꢁ0
1ꢁ1
0
57ꢁ3
8ꢁ1
2ꢁ1
1ꢁ0
1ꢁ0
0
0
0
0
3ꢁ1
25ꢁ5
0
4ꢁ1
58ꢁ3
12ꢁ5
6ꢁ1
9ꢁ1
16ꢁ2 2ꢁ1
4ꢁ1 1ꢁ1
1ꢁ1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Table 3. BODIPY-cadaverine fluorescent displacement assay.[a]
37ꢁ4
4ꢁ0
25ꢁ6
100
5ꢁ0
95ꢁ5
0
3ꢁ0 1ꢁ0
Compound
EC50 [mM]
Hill coefficient
Ymax [%]
100
100
7ꢁ1 2ꢁ1 1ꢁ0
tobramycin
PMX
6h
12.3
7.1
25.8
21.3
1.2
3.4
1.4
1.2
8
43
100
71
4ꢁ1
0
100
100
0
0
0
0
0
0
0
0
0
0
100
100
100
61ꢁ2
0
100
7ꢁ1
5ꢁ2
8h
99ꢁ1
62ꢁ9
12ꢁ1
28ꢁ2
6j
8h
69ꢁ7
95ꢁ5
28ꢁ3
82ꢁ4
3ꢁ1 1ꢁ1
[a] LPS of E. coli (serotype O111:B4) was incubated with BODIPY-cadaver-
ine and a range of concentrations of the tested compound. Fluorescence
was measured at l=580 nm. Maximal probe displacement (Ymax) of 6h
was defined as 100%. The Hill coefficients n (in arbitrary units) were cal-
culated by using OriginLab software. The experiments were carried out in
triplicate.
4ꢁ1 2ꢁ1 2ꢁ1
[a] Rabbit erythrocytes were incubated with the test compounds for 1 h
at 378C. All the experiments were performed in triplicate, and the results
are the average of three independent experiments on blood samples
from three different laboratory rabbits.
Chem. Eur. J. 2015, 21, 4340 – 4349
www.chemeurj.org
4344
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
4e–j. For example, 4’,5-di-O-n-nonyl nebramine derivative 4j
was the most potent of this subset against the tested Gram-
positive bacteria and was also the most hemolytic, thus caus-
ing 95ꢁ5% hemolysis at a concentration of 32 mgmLꢀ1. Final-
ly, of the 4’,5,6-tri-O-n-alkyl ether nebramine subset 6 f–j, those
with linear aliphatic chains with 6–8 carbon atoms were highly
hemolytic (Table 4). Compound 6h, the most potent antimicro-
bial in this study, was also the most hemolytic; that is, this
compound caused 58ꢁ3% hemolysis at a low concentration
of 16 mgmLꢀ1. Compound 8h, the more hydrophilic and posi-
tively charged analogue of 6h, caused significantly less hemol-
ysis than 6h. Exposure of rabbit erythrocytes to 32 mgmLꢀ1 of
6h resulted in 100% hemolysis, whereas exposure to 8h at
a similar concentration led to only 28ꢁ2% hemolysis.
Correlation of partition coefficient values with biological
activity
To test whether the biological activities of the tobramycin- and
nebramine-derived antimicrobial cationic amphiphiles were
correlated with their hydrophobicity/hydrophilicity ratios, we
calculated the partition coefficients (logP) for each of the 23
compounds in this study (see Table 1S in the Supporting Infor-
mation). The calculated logP values ranged from ꢀ0.67 to
13.60. All of the most potent antimicrobials in this study fell
into a narrow logP range of 5.36–6.08 (Table 6), thus suggest-
ing that logP may be used as a general molecular-design con-
sideration for the development of this type of antimicrobial
cationic amphiphile.
The standard MIC tests were conducted in a brain–heart in-
fusion (BHI) broth, which contains a yeast extract that is rich in
negatively charged molecules, such as nucleotides and pro-
teins, which may interact with antimicrobial cationic amphi-
philes and reduce their antimicrobial activity. In contrast, the
hemolysis tests were performed in a phosphate-buffered saline
(PBS) solution. Therefore, we evaluated the hemolytic effect of
AG-derived antimicrobial agents in BHI broth to determine
whether the media affected the membrane selectivity. For this
analysis, we chose the tobramycin-derived 2 f and nebramine-
derived 6h, which were potent antimicrobial agents that were
also highly hemolytic. The hemolysis tests, carried out in a BHI
broth identical to that used for the MIC tests, showed that 2 f
and 6h were significantly less hemolytic than in PBS (Table 5).
For example, 2 f caused 89.8ꢁ8.0% hemolysis in PBS at
Table 6. Hemolysis, MIC range, and logP values of selected compounds.
Compound
Hemolysis [%][a]
MIC range [mgmLꢀ1
]
logP[b]
Gram (+)
Gram (ꢀ)
2a
2 f
6h
4j
0
100
100
100
4–16
2–16
1–4
4
>64
32–>64
4–16
5.36
5.58
5.30
6.08
32–>64
[a] Hemolysis was determined after exposure of rabbit erythrocytes to
64 mgmLꢀ1 of the test compound. [b] The logP values were calculated for
the free-base form of the compound by using the MarvinSketch software
(Marvin 6.3.1).
LogP was not predictive, however, of the hemolytic activity
or antimicrobial-activity spectrum. For example, 2a has a logP
value of 5.36 and was the most potent antimicrobial of the
subset of aromatic ether-substituted tobramycin derivatives
against Gram-positive bacteria, but this compound was inac-
tive against all of the tested Gram-negative strains and was
the least hemolytic of all of the cationic amphiphiles in this
study. In contrast, 6h, with a similar logP value of 5.30 and
was the most potent antimicrobial of the 4’,5,6-tri-O-n-alkyl
ether nebramine derivatives, was active against all Gram-posi-
tive and Gram-negative strains and was one of the most hemo-
lytic compounds tested.
Table 5. The effect of the BHI media on the hemolytic activity.[a]
Concentration [mgmLꢀ1
64 32
]
Cpd.
2 f
BHI [%]
128
16
8
4
0
100
0
99ꢁ1
97ꢁ3
52ꢁ8
100
100
90ꢁ8
8ꢁ1
16ꢁ3
1ꢁ1
2ꢁ0
0
4ꢁ0
1ꢁ0
0
0
0
0
93ꢁ7
100
100
95ꢁ1
78ꢁ19
50ꢁ8
16ꢁ1
6h
100
[a] Rat erythrocytes were incubated with the test compounds for 1 hour
at 378C. All the experiments were performed in triplicate, and the results
are the average of three independent experiments on blood samples
from two individual laboratory rats.
Evaluation of bacterial-membrane disruption effect
a concentration of 32 mgmLꢀ1, whereas this compound was
approximately one order of magnitude less hemolytic (8.3ꢁ
0.8%) at the same concentration in BHI. The significance of the
media in which the hemolysis tests were conducted was di-
minished at higher concentrations of the test compounds
(Table 5). The antimicrobial cationic amphiphiles likely interact
with negatively charged molecules found in the rich media.
Saturation of these interactions at higher concentrations of the
antimicrobial cationic amphiphiles reduced this rich-media
effect and left sufficient concentration of the free compounds
to act on the membranes.
The reduced hemolytic activity of the penta-O-benzyl tobramy-
cin derivative 2a relative to all of the other compounds in this
study suggests that this compound may act on bacteria
through a different mode of action than the other compounds
evaluated. To test this possibility, we monitored the lytic effect
of 2a on bacteria that express cytosolic yellow fluorescent pro-
tein (YFP) and compared it to that of penta-O-n-pentyl tobra-
mycin 2 f.[21,32] Similar to 2a, 2 f demonstrated antimicrobial ef-
ficacy against all five tested Gram-positive strains; however,
unlike 2a, 2 f demonstrated poor membrane selectivity when
an exposure of rat erythrocytes at
a concentration of
16 mgmLꢀ1 of 2 f led to 30ꢁ13% hemolysis.
Chem. Eur. J. 2015, 21, 4340 – 4349
www.chemeurj.org
4345
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
A culture of untreated YFP-expressing Bacillus subtilis (Fig-
ure 2A–C) was treated with 2a and 2 f at concentrations that
were twofold the MIC value (2 mgmLꢀ1). Although most of the
bacterial cells in the untreated sample were viable and fluores-
cently labeled, a significant drop in the bacterial cell number
and in cytosolic fluorescence was observed in bacterial culture
containing 2a or 2 f (Figure 2D–I). Similar to the loss of hemo-
globin during the hemolysis tests, the observed loss of cytosol-
ic fluorescent protein content likely resulted from a disruption
in the membrane integrity caused by 2a and 2 f, thus leading
to bacterial cell lysis. We concluded that even though 2a has
relatively low hemolytic activity, it appears to be toxic to bacte-
ria due to membrane disruption and not from a different
mode of antimicrobial action.
whether there is a general parameter that can be used to
design AG-derived antimicrobial cationic amphiphiles by calcu-
lating the partitioning coefficient for each of the cationic am-
phiphiles. As a fourth direction, we studied the mode of action
of the potent antimicrobial agents.
Compound 6h, the most potent antimicrobial cationic am-
phiphile in this study, had a significantly broader spectrum of
antimicrobial activity relative to the antimicrobial activity of
the parent AG tobramycin, the membrane-disrupting antimi-
crobial peptide-mixture gramicidin D, and the cationic lipopep-
tide PMX, all of which are clinically used antimicrobial agents.
Compound 6h was at least 32 times more effective than the ri-
bosome-targeting antibiotic tobramycin and the membrane-
targeting antibiotic gramicidin D against several of the tested
Gram-positive and Gram-negative bacteria and, unlike PMX,
was effective against the tested Gram-positive pathogens. MIC
experiments revealed that compounds that exhibited potent
antimicrobial activity against the tested Gram-positive bacteria
were not necessarily active against the tested Gram-negative
strains. On the other hand, our observations suggest that activ-
ity against Gram-negative strains serves as an indication of po-
tency against Gram-positive bacteria.
In general, potent antimicrobial activity was accompanied
by higher erythrocyte hemolytic activity. Erythrocyte mem-
brane damage caused by the AG-derived antimicrobial cationic
amphiphiles was significantly diminished in rich physiological
environments relative to PBS solutions in which it is usually
evaluated. An exception was observed in the case of penta-O-
benzyl tobramycin 2a because this compound had potent an-
timicrobial activity against the tested Gram-positive strains, yet
caused almost no measurable hemolysis at a concentration
range of 16–32-fold greater than its effective MIC range (4–
8 mgmLꢀ1).
Figure 2. Bright-field and epi-fluorescence microscopy. B. subtilis (PY79) that
constitutively express the YFP gene were imaged after cells had been incu-
bated for 30 min with 2a or with 2 f at 4 mgmLꢀ1 (2ꢁMIC). A) Control (un-
treated cells), bright field; B) control, fluorescent field; C) control, merge,
D) 2a, bright field; E) 2a, fluorescent field; F) 2a, merge; G) 2 f, bright field;
H) 2 f, fluorescent field; I) 2 f, merge.
The calculated logP values of the four most potent antimi-
crobials, which vary in their molecular structures, fell into
a narrow range of 5.36–6.08. This outcome suggests that this
parameter may be used as a consideration for the design of
other AG-derived antimicrobial cationic amphiphiles. However,
logP values did not correlate with the antimicrobial-activity
spectrum or the extent of the hemolytic activity of the com-
pounds in this study and cannot be used to predict these bio-
logical activities.
Conclusion
Four defined subsets of novel cationic amphiphiles derived
from the AGs tobramycin and its pseudo-disaccharide segment
nebramine were synthesized by etherification of some or all of
the alcohol groups on the parent compounds. Antimicrobial
activity tests revealed that the most potent antimicrobial cat-
ionic amphiphiles were compounds 2a, 2 f, 4j, and 6h. Biolog-
ical evaluation of the cationic amphiphiles in this study fo-
cused on four properties: First, we performed antimicrobial
tests to determine the potency and efficacy spectrum of the
compounds. Second, we assessed the selectivity of the cationic
amphiphiles for bacterial membranes by evaluating the
damage that these compounds caused to red blood cells,
which served as a model for the evaluation of mammalian cell-
membrane damage. We also evaluated the significance of per-
forming the hemolysis assay in the same rich BHI broth used
for the antimicrobial activity tests. Third, we investigated
Finally, we investigated whether there was a connection be-
tween the anti-Gram negative activity of 6h and its affinity to
LPS as a possible mode of action. We concluded that the anti-
Gram negative activity of 6h is likely to result from interactions
with LPS and with additional Gram-negative membrane deter-
minants. Second, we investigated if the antimicrobial cationic
amphiphiles in this study (represented by 2a and 2 f) induce
bacterial cell-membrane damage, hence causing the loss of in-
tracellular contents. This investigation was accomplished by
observing the loss of fluorescent cytosolic protein from cyto-
solic YFP-expressing bacteria after incubation with the tested
cationic amphiphiles. From these experiments, we concluded
that 2a disrupts bacterial membrane structure, even though it
did not cause significant red-blood-cell lysis at concentrations
near the MIC values.
Chem. Eur. J. 2015, 21, 4340 – 4349
www.chemeurj.org
4346
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
the TFA salt of 2a. HRMS (ESI): m/z calcd for C53H68N5O9: 918.5017
[M+H]+; found, 918. 5015; 1H NMR (400 MHz, D2O) d=7.45–7.25
(m, 18H; benzyl), 7.21–7.15 (m, 3H; benzyl), 7.11–7.05 (m, 4H;
benzyl), 5.19 (d, J=3.0 Hz, 1H; H-1’’), 5.05 (d, J=2.1 Hz, 1H; H-1’),
4.85 (d, J=11.9 Hz, 1H; CH2 benzyl), 4.77–4.70 (m, 2H; CH2 benzyl),
4.66–4.55 (m, 3H; CH2 benzyl), 4.57–4.48 (m, 1H; CH2 benzyl), 4.40
(bd, J=10.2 Hz, 1H; H-5’), 4.30 (d, J=11.1 Hz, 1H; CH2 benzyl), 4.16
(d, J=11.8 Hz, 1H; CH2 benzyl), 4.11–4.03 (dd, J=8.9, 9.1 Hz 1H; H-
4), 4.00 (d, J=11.8 Hz, 1H; CH2 benzyl), 3.83 (m, 1H; H-5’’), 3.81–
3.75 (m, 2H; H-5, H-6), 3.74–3.54 (m, 4H; H-4’, H-2’’, H-3’’, H-4’’),
3.52–3.36 (m, 2H; H-1, H-3), 3.40 (dd, J=11.6, 2.4 Hz, 1H; H-6’’),
3.21 (m, 1H; H-2’), 3.18–3.05 (m, 2H; H-6’, H-6’’), 2.99 (dd, J=14.0,
4.1 Hz, 1H; H-6’), 2.38 (dt, J=12.6, 4.3 Hz, 1H; H-2eq), 2.18 (m,1H;
H-3eq), 1.89–1.71 ppm (m, 2H; H-2ax, H-3ax); 13C NMR (100 MHz, D2O)
d=163.09 (q, J=35.4 Hz; CF3CO2H), 137.28, 136.88, 136.30, 136.19,
135.44, 129.44, 129.35, 129.08, 129.05, 128.84, 128.79, 128.64,
128.49, 128.40, 128.21, 126.67, 116.34 (q, J=291.8 Hz, CF3CO2H),
98.28, 92.90, 81.80, 81.54, 78.14, 75.35, 75.04, 74.66, 74.38, 73.74,
73.06, 71.59, 71.23, 69.88, 66.56, 52.74, 49.56, 48.17, 46.72, 37.78,
30.22, 27.26, 25.59 ppm.
To conclude, the AG-derived antimicrobial cationic amphi-
philes, such as those presented herein, offer a promising direc-
tion for the development of novel antibiotics for the treatment
of topical infection caused by pathogens with resistance to the
currently available repertoire of clinically used antimicrobial
agents.
Experimental Section
General methods
1H NMR spectra (including 1D TOCSY) were recorded on Bruker
Avance 400 or 500 spectrometers, and chemical shifts (reported in
ppm) were calibrated to CD3OD, D2O, or CDCl3(d=3.33, 4.63, and
7.27 ppm, respectively). 13C NMR spectra were recorded on Bruker
Avance 400 or 500 spectrometers at 100.6 or 125 MHz, respectively.
Multiplicities are reported with the following abbreviations: b=
broad, s=singlet, d=doublet, t=triplet, dt=doublet of triplets,
dd=doublet of doublets, ddd=doublet of doublet of doublets,
m=multiplet, eq=equatorial, ax=axial. Coupling constants (J) are
given in Hertz. Low-resolution electron spray ionization mass spec-
tra were measured on a Waters 3100 mass detector. High-resolu-
tion electron spray ionization mass spectra were measured on
a Waters Synapt instrument. Chemical reactions were monitored
by TLC analysis (Merck, Silica gel60 F254). Visualization was ach-
ieved by using a cerium molybdate stain ((NH4)2Ce(NO3)6 (5 g),
(NH4)6Mo7O24·4H2O (120 g), H2SO4 (80 mL), and H2O (720 mL)). All
the reactions were carried out in an argon atmosphere with anhy-
drous solvents, unless otherwise noted. All chemicals, unless other-
wise stated, were obtained from commercial sources. Compounds
were purified by means of flash chromatography (SiO2, Merck, Kie-
selgel60).
General procedure for the preparation of nebramine deriva-
tives from the corresponding tobramycin derivatives
Concentrated sulfuric acid (1.5 n, 2.5 mL) was added to 1a
(200 mg, 0.19 mmol) was dissolved in MeOH (60 mL). The reaction
mixture was heated to reflux for 48 h. Progress of the reaction was
monitored by TLC analysis (petroleum ether/EtOAc 92:8). Upon
completion, the reaction mixture was quenched with saturated
sodium bicarbonate and extracted with ethyl acetate (3ꢁ100 mL).
The organic layer was dried over anhydrous MgSO4, filtered, and
concentrated under reduced pressure. Purification by flash column
chromatography (SiO2, petroleum ether/EtOAc 99:1!92:8) gave
3a (100 mg, 88%) as a colorless oil. LRMS (ESI): m/z calcd for
C26H30N12O5: 613.24 [M+Na]+; found 613.16; 1H NMR (500 MHz,
CD3OD) d=7.41 (m, 2H; benzyl), 7.38–7.23 (m, 8H; benzyl), 5.54 (d,
J=3.4 Hz, 1H; H-1’), 5.12 (d, J=10.2 Hz, 1H; CH2 benzyl), 4.79 (d,
J=10.2 Hz, 1H; CH2 benzyl), 4.68 (d, J=11.6 Hz, 1H; CH2 benzyl),
4.50 (d, J=11.6 Hz, 1H; CH2 benzyl), 4.22 (m, 1H; H-5’), 3.66–3.42
(m, 8H; H-1, H-3, H-4, H-5, H-6, H-2’, H-4’, H-6’), 3.28 (m, 1H; H-6’),
3.15–3.05 (m, 1H; H-3’eq), 2.43 (dt, J=11.0, 4.3 Hz, 1H; H-2eq), 2.25
(dt, J=12.3, 4.0 Hz, 1H; H-3’ax), 2.01 ppm (m, 1H; H-2ax); 13C NMR
(125 MHz, CD3OD) d=138.38, 138.09, 128.05, 127.91, 127.71,
127.62, 127.47, 127.24, 96.45, 84.73, 77.36, 77.23, 74.60, 72.36,
71.10, 70.45, 60.80, 59.98, 55.99, 51.04, 31.80, 27.75 ppm.
General procedure for etherification of azide-protected ami-
noglycoside scaffolds
Benzyl bromide (1.2 mL, 10.0 mmol, 6.0 equiv), TBAI (617 mg,
1.67 mmol, 1.0 equiv), and NaH (60%, 402 mg, 10.0 mmol,
6.0 equiv) to
a
solution of penta-azidotobramycin[27] (1.0 g,
1.67 mmol) dissolved in dry DMF (10 mL) under argon. The reac-
tion mixture was stirred at ambient temperature overnight. Prog-
ress of the reaction was monitored by TLC analysis (petroleum
ether/EtOAc 75:25). Upon completion, the reaction mixture was di-
luted with EtOAc (100 mL), and the organic layer was washed with
brine, dried over anhydrous MgSO4, filtered, and concentrated
under reduced pressure. Purification by flash column chromatogra-
phy (SiO2, petroleum ether/EtOAc 99:1!75:25) gave 1a (1.7 g,
Synthesis of 7h
1
K2CO3 (25.7 mg, 0.18 mmol, 1.1 equiv) and bis-Boc-l-histidine N-hy-
droxysuccinimide ester (84.15 mg, 0.18 mmol, 1.1equiv) were
added to 6h (110 mg, 0.17 mmol) in MeOH (5 mL). The reaction
mixture was stirred at room temperature for 24 h. Progress of the
reaction was monitored by TLC analysis (2.8% NH4OH solution in
MeOH/dichloromethane 10:90). Upon completion, the reaction
mixture was concentrated under reduced pressure and further pu-
rified by flash chromatography (SiO2, 2.8% NH4OH solution in
MeOH/ dichloromethane 0:100!10:90). The fractions containing
the pure product were concentrated under reduced pressure. The
residue was dissolved in a minimal volume of H2O and freeze-dried
to afford the pure product 7h (112 mg, 75%) as a white powder.
LRMS (ESI): m/z calcd for C49H91N7O10: 838.38 [MꢀCO2C(CH3)3 +2H]+;
97%) as a colorless oil. The H NMR spectra was in full agreement
with the previously reported spectra.[28]
General procedure for reduction of azide protecting groups
Compound 1a (130 mg, 0.12 mmol) was dissolved in THF (4.0 mL)
and H2O (0.5 mL). The mixture was added to NaOH (0.1n, 0.2 mL)
and trimethylphosphine in THF (1m, 1.24 mL, 1.24 mmol,
10.0 equiv). Progress of the reaction was monitored by TLC analysis
(2.8% NH4OH solution in MeOH/dichloromethane 10:90). After 24 h
the reaction mixture was evaporated under reduced pressure. Pu-
rification by flash column chromatography (SiO2, 2.8% NH4OH solu-
tion in MeOH/dichloromethane 0:100!10:90) gave 2a (100 mg,
88%). The pure residue was dissolved in 95% TFA (0.5 mL), which
was removed after 2 min under reduced pressure. The residue was
dissolved in a minimal volume of H2O and freeze-dried to afford
1
found 838.74; H NMR (500 MHz, CD3OD) d=7.64 (s,1H; histidine),
6.89 (s, 1H; histidine), 4.93 (s, 1H; H-1’), 4.32 (dd, J=9.0, 5.2 Hz,
1H; CH-a-histidine), 3.89–3.79 (m, 2H; n-heptyl (2H)), 3.74 (dt, J=
Chem. Eur. J. 2015, 21, 4340 – 4349
www.chemeurj.org
4347
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
9.2, 4.6 Hz, 1H; H-5’), 3.73–3.55 (m, 3H; n-heptyl (3H)), 3.52–3.42
(m, 2H; H-6’ (2H)), 3.37 (m, 1H; n-heptyl (1H)), 3.29 (t, J=9.3 Hz,
1H; H-4), 3.21 (t, J=9.3 Hz, 1H; H-5), 3.12–3.03 (m, 2H; H-4’, CH-b-
histidine (1H)), 2.93 (t, J=9.5 Hz, 1H; H-6), 2.90–2.72 (m, 3H; H-3,
H-2’, CH-b-histidine (1H)), 2.67 (ddd, J=12.2, 9.8, 4.1 Hz, 1H; H-1),
2.17 (dt, J=11.6, 4.2 Hz, 1H; H-3’eq), 1.95 (dt, J=12.9, 4.2 Hz 1H; H-
2eq), 1.72–1.55 (m, 6H; n-heptyl (6H)), 1.50 (m, 1H; H-3’ax), 1.45–
1.28 (m, 42H; Boc-histidine (18H), n-heptyl (24H)), 1.20 (m, 1H; H-
2ax), 0.97–0.81 ppm (m, 6H; n-heptyl (9H)); 13C NMR (125 MHz,
CD3OD) d=173.01, 156.24, 134.80, 99.93, 86.82, 86.14, 83.73, 79.20,
73.91, 73.38, 73.30, 71.06, 68.66, 54.89, 51.40, 50.69, 49.75, 39.88,
35.98, 31.87, 31.76, 30.20, 30.16, 29.76, 29.11, 29.05, 28.88, 27.33,
25.91, 25.85 ppm.
Protocol LPS BODIPY-cadaverine displacement assay
Stock solutions of LPS from E. coli 0111:B4 purified as above
(5 mgmLꢀ1) and BODIPY-TR-cadaverine (500 mm) were prepared in
tris(hydroxymethyl)aminomethane (Tris) buffer (pH 7.4, 50 mm).
The LPS stock (1 mL) and BODIPY-TR-cadaverine stock (1 mL) were
mixed and diluted in Tris buffer to a final volume of 100 mL, thus
yielding final concentrations of 50 mgmLꢀ1 of LPS and 5 mm
BODIPY-TR-cadaverine. BODIPY-TR-cadaverine/LPS mixture (40 mL)
was added to each well of the 96-well plate. The test compound
was added and diluted with Tris buffer to a final volume of 60 mL.
Fluorescence measurements were made at 258C on a TECAN mi-
croplate reader (infinite F200 PRO). The BODIPY-TR-cadaverine exci-
tation wavelength was l=580 nm. Emission spectra were taken at
l=620 nm.
Synthesis of 8h
Compound 7h (50 mg, 0.05 mmol) was treated at room tempera-
ture with 95% TFA (1 mL) for 3 min. The TFA was removed under
reduced pressure, and the residue was dissolved in a minimal
volume of H2O and freeze-dried to afford 8h (39 mg, quantitative
yield) as a white foam. HRMS (ESI): m/z calcd for C39H76N7O6:
Protocol for erythrocyte hemolysis assay
The hemolysis assay was performed by following a previously re-
ported protocol.[23] These experiments were done in triplicate, and
the results reported are the averages of three independent experi-
ments on three different rabbit blood samples. Hemolysis assays
conducted in BHI broth were performed as follows: A sample of
rat red blood cells in BHI medium (2% w/w) were incubated with
each of the tested compounds for 1 h at 378C in 5% CO2 by using
the double-dilution method and starting at concentration of
128 mgmLꢀ1. The negative control was BHI, and the positive con-
trol was 1% v/v solution of Triton X100 in BHI (100% hemolysis).
Following centrifugation (2000 rpm, 10 min, ambient temperature),
the supernatant was removed and absorbance at l=550 nm was
measured by using a microplate reader (SpectraMax-M2). The re-
sults are expressed as percentage of hemoglobin released relative
to the positive control (Triton X100). The experiments were per-
formed in triplicate, and the results are an average of experiments
on blood samples taken from two rats.
1
738.5857 [M+H]+; found, 738.5861; H NMR (500 MHz, CD3OD) d=
8.61 (s, 1H; histidine), 7.38 (s, 1H; histidine), 5.28 (d, J=1.8 Hz, 1H;
H-1’), 4.33 (dd, J=8.3, 5.7 Hz, 1H; CH-a-histidine), 4.23 (t, J=
7.1 Hz, 1H; H-5’), 4.00 (dd, J=10.0, 9.2 Hz, 1H; H-4), 3.85–3.78 (m,
2H; n-heptyl (2H)), 3.74 (m, 1H; n-heptyl (1H)), 3.66–3.52 (m, 5H;
H-2’, H-6’ (2H), n-heptyl (2H)), 3.50–3.34 (m, 6H; H-3, H-5, H-6, H-4’,
n-heptyl (1H), CH-b-histidine (1H)), 3.30–3.20 (m, 2H; H-1, CH-b-
histidine), 2.46 (dt, J=12.3, 4.2 Hz, 1H; H-2eq), 2.40–2.21 (m, 2H; H-
3’eq, H-3’ax), 1.93 (m, 1H; H-2ax), 1.72–1.57 (m, 6H; n-heptyl (6H)),
1.43–1.27 (m, 24H; n-heptyl (24H)), 1.00–0.83 ppm (m, 9H; n-
heptyl (9H)); 13C NMR (100 MHz, CD3OD) d=168.33, 161.83 (q, J=
34.7 Hz, CF3CO2H)), 135.05, 128.45, 118.08, 117.07 (p, J=292.6 Hz;
CF3CO2H)), 93.53, 83.25, 80.42, 78.41, 75.77, 74.06, 73.90, 71.12,
69.80, 52.39, 49.52, 48.63, 38.33, 31.80, 31.75, 31.72, 30.37, 29.79,
29.66, 29.22, 29.16, 29.09, 28.16, 27.40, 26.29, 26.24, 25.85, 25.75,
22.41, 13.13 ppm.
Protocol for epi-fluorescence microscopy
The epi-fluorescence microscopy assay was performed as previous-
ly reported[21] with the following changes: B. subtilis PY79 cells that
constitutively express YFP were used. B. subtilis PY79 cells carrying
YFP taken from a freshly streaked plate were grown (378C, ꢂ14 h)
in lysogeny broth (LB; 5 mL) supplemented with erythromycin
(3 mgmLꢀ1). The overnight culture (0.06 mL) was diluted into fresh
LB broth (6 mL) and grown to OD600 =0.3. The cells were treated
with 2a or 2 f at 8 mgmLꢀ1 (4ꢁMIC) and incubated at 378C. After
30 min in the presence of 2a or 2 f, aliquots (1 mL) were transferred
to agarose slabs made of LB. Snapshot fluorescence images were
taken with a 100ꢁ lens of an inverted epi-fluorescence microscope
(Nikon TiE).
Protocol for minimal inhibitory-concentration tests
MIC values were determined as reported previously.[23] In each ex-
periment, the samples were analyzed in triplicate, and the results
were obtained from two independent experiments. MIC values
(mgmLꢀ1) were determined as the lowest concentration at which
no bacterial growth was observed.
Protocol for MIC experiments with added LPS
Starter cultures of E. coli ATCC 25922 (Strain G) and S. pyogenes
glossy were incubated for 24 h (378C, 5% CO2, aerobic conditions)
and then diluted in fresh BHI medium to obtain an optical density
of 0.004 (OD600). Tested compounds were diluted by using the
double-dilution method starting at 64 mgmLꢀ1 into 96-well plates
(Corning) at a volume of 100 mL. The LPS (50 mL) was prepared
from E. coli 0111:B4, purified by ion-exchange chromatography, and
tested to ensure that it could serve as a TLR ligand. The LPS was
added in final concentrations of 0, 10, 50, and 100 mgmLꢀ1. The
compounds and LPS were incubated for 1 h at room temperature.
Diluted bacteria solution (50 mL) was added to each well to obtain
a final volume of 200 mL. This mixture was incubated for 18 h
(378C, 5% CO2, aerobic conditions) and 3-(4,5-dimethylthiazol-2-yl)-
2,5-diphenyltetrazolium bromide (MTT; 50 mL, 1 mgmLꢀ1 in H2O)
was added to each well, followed by an additional incubation at
378C for 30 min.
Acknowledgements
This work was supported by the Israel Science Foundation
Grant 6-14. We thank Prof. Doron Steinberg (The Hebrew Uni-
versity of Jerusalem) for the gift of bacterial strains.
Keywords: cationic amphiphiles
· antibiotic resistance ·
antibiotics · bacterial membranes · glycosides
[1] J. G. Hurdle, A. J. O’Neill, I. Chopra, R. E. Lee, Nat. Rev. Microbiol. 2011, 9,
62–75.
Chem. Eur. J. 2015, 21, 4340 – 4349
www.chemeurj.org
4348
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
[2] R. E. W. Hancock, Exp. Opin. Invest. Drugs 2000, 9, 1723–1729.
[3] H. Jenssen, P. Hamill, R. E. W. Hancock, Clin. Microbiol. Rev. 2006, 19,
491–511.
man, Angew. Chem. Int. Ed. 2012, 51, 5652–5656; Angew. Chem. 2012,
124, 5750–5754.
[22] I. M. Herzog, M. Feldman, A. Eldar-Boock, R. Satchi-Fainaro, M. Fridman,
MedChemComm 2013, 4, 120–124.
[4] M. Zasloff, Nature 2002, 415, 389–395.
[5] G. F. Gause, M. G. Brazhnikova, Nature 1944, 154, 703–703.
[6] V. Nizet, Curr. Issues Mol. Biol. 2006, 8, 11–26.
[7] R. E. W. Hancock, H.-G. Sahl, Nat. Biotechnol. 2006, 24, 1551–1557.
[8] S. Kristian, V. Datta, J. Bacteriol. 2005, 187, 6719–6725.
[9] J. S. Gunn, K. B. Lim, J. Krueger, K. Kim, L. Guo, M. Hackett, S. I. Miller,
Mol. Microbiol. 1998, 27, 1171–1182.
[10] M. A. Campos, M. A. Vargas, V. Regueiro, C. M. Llompart, S. Albertꢂ, A.
Josꢃ, Infect. Immun. 2004, 72, 7107–7114.
[11] T. Jin, M. Bokarewa, T. Foster, J. Mitchell, J. Higgins, A. Tarkowski, J. Im-
munol. 2004, 172, 1169–1176.
[12] A. E. Shinnar, K. L. Butler, H. J. Park, Bioorg. Chem. 2003, 31, 425–436.
[13] F. Guilhelmelli, N. Vilela, P. Albuquerque, L. D. S. Derengowski, I. Silva-
Pereira, C. M. Kyaw, Front. Microbiol. 2013, 4, 1–12.
[14] X. Lai, Y. Feng, J. Pollard, Acc. Chem. Res. 2008, 41, 1233–1240.
[15] Y. Eun, M. H. Foss, D. Kiekebusch, D. A. Pauw, W. M. Westler, M. Than-
bichler, D. B. Weibel, J. Am. Chem. Soc. 2012, 134, 11322–11325.
[16] N. Srinivas, P. Jetter, B. J. Ueberbacher, M. Werneburg, K. Zerbe, J. Stein-
mann, B. Van der Meijden, F. Bernardini, A. Lederer, R. L. A. Dias, et al.,
Science 2010, 327, 1010–1013.
[23] Y. Berkov-Zrihen, I. M. Herzog, M. Feldman, A. Sonn-Segev, Y. Roichman,
M. Fridman, Bioorg. Med. Chem. 2013, 21, 3624–3631.
[24] Y. Berkov-Zrihen, I. M. Herzog, M. Feldman, M. Fridman, Org. Lett. 2013,
15, 6144–6147.
[25] M. Ouberai, F. El Garch, A. Bussiere, M. Riou, D. Alsteens, L. Lins, I. Baus-
sanne, Y. F. DufrÞne, R. Brasseur, J.-L. Decout, M.-P. Mingeot-Leclercq,
Biochim. Biophys. Acta. 2011, 1808, 1716–1727.
[26] G. Sautrey, L. Zimmermann, M. Deleu, A. Delbar, L. Souza Machado, K.
Jeannot, F. Van Bambeke, J. M. Buyck, J.-L. Decout, M.-P. Mingeot-Le-
clercq, Antimicrob. Agents Chemother. 2014, 58, 4420–4430.
[27] P. T. Nyffeler, C.-H. Liang, K. M. Koeller, C.-H. Wong, J. Am. Chem. Soc.
2002, 124, 10773–10778.
[28] F. Agnelli, S. J. Sucheck, K. A. Marby, D. Rabuka, S.-L. Yao, P. S. Sears, F.-S.
Liang, C.-H. Wong, Angew. Chem. Int. Ed. 2004, 43, 1562–1566; Angew.
Chem. 2004, 116, 1588–1592.
[29] E. Llobet, J. M. Tomꢄs, J. A. Bengoechea, Microbiology 2008, 154, 3877–
3886.
[30] A. Giacometti, O. Cirioni, R. Ghiselli, F. Mocchegiani, F. Orlando, C. Silves-
tri, A. Bozzi, A. Di Giulio, C. Luzi, M. L. Mangoni, et al., Antimicrob. Agents
Chemother. 2006, 50, 2478–2486.
[31] S. Wood, K. Miller, S. David, Comb. Chem. High Throughput Screening
2004, 7, 239–249.
[32] A. Eldar, V. K. Chary, P. Xenopoulos, M. E. Fontes, O. C. Losꢅn, J. Dworkin,
P. J. Piggot, M. B. Elowitz, Nature 2009, 460, 510–514.
[17] S. K. Vooturi, C. M. Cheung, M. J. Rybak, S. M. Firestine, J. Med. Chem.
2009, 52, 5020–5031.
[18] H. Zou, J.-J. Koh, J. Li, S. Qiu, T. T. Aung, H. Lin, R. Lakshminarayanan, X.
Dai, C. Tang, F. H. Lim, et al., J. Med. Chem. 2013, 56, 2359–2373.
[19] I. M. Herzog, M. Fridman, MedChemComm 2014, 5, 1014–1026.
[20] B. Findlay, G. G. Zhanel, F. Schweizer, Antimicrob. Agents Chemother.
2010, 54, 4049–4058.
[21] I. M. Herzog, K. D. Green, Y. Berkov-Zrihen, M. Feldman, R. R. Vidavski, A.
Eldar-Boock, R. Satchi-Fainaro, A. Eldar, S. Garneau-Tsodikova, M. Frid-
Received: December 9, 2014
Published online on February 4, 2015
Chem. Eur. J. 2015, 21, 4340 – 4349
www.chemeurj.org
4349
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim