Structure-toxicity relationship of aminoglycosides: Correlation of 2′-amine basicity with acute toxicity in pseudo-disaccharide scaffolds

Article abstract of DOI:10.1016/j.bmc.2008.08.053

A new pseudo-disaccharide NB23 with a 3′,4′-methylidene protection was designed and its properties were evaluated in comparison to other two structurally related pseudo-disaccharides. The basicity of the 2′-amine was found to be well correlated to acute toxicity data in mice: the increase in the basicity is associated with the toxicity increase. Based on these data, a new pseudo-trisaccharide NB45 was constructed. NB45 exhibited significant antibacterial activity while at the same time retained low acute toxicity.

Full text of DOI:10.1016/j.bmc.2008.08.053

Bioorganic & Medicinal Chemistry 16 (2008) 8940–8951
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry
journal homepage: www.elsevier.com/locate/bmc
Structure–toxicity relationship of aminoglycosides: Correlation of 20-amine
basicity with acute toxicity in pseudo-disaccharide scaffolds
Lilach Chen a, Mariana Hainrichson a, Dmitry Bourdetsky b, Amram Mor b, Sima Yaron b, Timor Baasov a,
*
a The Edith and Joseph Fischer Enzyme Inhibitors Laboratory, Schulich Faculty of Chemistry, Technion—Israel Institute of Technology, Haifa 32000, Israel
b Department of Biotechnology and Food Engineering, Technion—Israel Institute of Technology, Haifa 32000, Israel
a r t i c l e i n f o
a b s t r a c t
A new pseudo-disaccharide NB23 with a 30,40-methylidene protection was designed and its properties
were evaluated in comparison to other two structurally related pseudo-disaccharides. The basicity of
the 20-amine was found to be well correlated to acute toxicity data in mice: the increase in the basicity
is associated with the toxicity increase. Based on these data, a new pseudo-trisaccharide NB45 was con-
structed. NB45 exhibited significant antibacterial activity while at the same time retained low acute
toxicity.
Article history:
Received 13 July 2008
Revised 17 August 2008
Accepted 22 August 2008
Available online 27 August 2008
Keywords:
Ó 2008 Elsevier Ltd. All rights reserved.
Aminoglycosides basicity
Protein translation inhibition
Resistance to aminoglycosides
Aminoglycosides toxicity
Acute toxicity
1. Introduction
studies suggested that the interference between aminoglycosides
and some steps of calcium-mediated acetylcholine release at the
2-Deoxystreptamine (2-DOS) aminoglycosides (Fig. 1) are
highly potent, broad-spectrum antibiotics that selectively target
the prokaryotic ribosome and bind to the decoding A-site of the
16S ribosomal RNA leading to protein translation inhibition and
interference with the translational fidelity.1–4 These antibiotics
have been used in the clinic for almost six decades and because
of such prolonged clinical and veterinary use, bacterial resistance
to these drugs has become a serious public health problem. The
primary resistance mechanism to aminoglycosides is the bacterial
acquisition of enzymes which modify the antibiotics by acetyl-
transferase (AAC), adenyltransferase (ANT), or phosphotransferase
(APH) activities.2,5 Each class of these enzymes performs a regio-
specific reaction either on the amine group (AAC) or hydroxyl
group (ANT and APH) and the turnover products of these reactions
lack antibacterial activity.
In addition to emerged resistance, one of the major limitations
in using aminoglycosides as drugs is their high toxicity to mam-
mals through kidney (nephrotoxicity) and ear-associated (ototox-
icity) illnesses. The origin of this toxicity is still controversial and
probably results from a combination of different factors/mecha-
nisms such as interactions with phospholipids, inhibition of phos-
pholipases and formation of free radicals.6,7 Furthermore,
neuromuscular blockage and related respiratory depression are
well known side effects of aminoglycosides therapy.8 Numerous
level of presynaptic structures is the main cause of the neuromus-
cular blockage induced by aminoglycosides.9,10 Indeed, neuromus-
cular blocking effects were found to be well correlated to acute
toxicity data of aminoglycosides11,12 and that co-administration
of Ca2+ antagonized acute toxicity and neuromuscular effects in-
duced by aminoglycosides.12,13 In light of these observations, it
was suggested that both respiratory failure and experimental tox-
icity follow the neuromuscular blocking effects caused by
aminoglycosides.11
Much work has been done to understand the mechanisms by
which side effects of aminoglycosides develop in patients, but little
effort was devoted to understand their structure–toxicity relation-
ship. During the last decades, many examples of synthetic variants
of naturally occurring aminoglycosides were reported.14,15 The
majority of these variants, however, were designed with the aim
to tackle the problem of resistance to aminoglycosides and it is
very rare that the toxicity was either considered in the design or
it was determined for the resulting compounds. It is highly note-
worthy that, although a new synthetic variant may exhibits favor-
able antibacterial activity, high toxicity can prevent its clinical
application. Therefore, the design of novel variants of aminoglyco-
sides which in addition to resisting to aminoglycoside-modifying
enzymes and targeting ribosomal RNA, will also have high proba-
bility of reduced toxicity is of high urgency.
For this propose, we sought that it may be possible to separate
elements of the aminoglycoside structure that induce toxicity from
those that are required for an antibiotic effect. Indeed, earlier study
* Corresponding author. Tel.: +972 4 829 2590; fax: +972 4 829 5703.
E-mail address: chtimor@tx.technion.ac.il (T. Baasov).
0968-0896/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmc.2008.08.053

L. Chen et al. / Bioorg. Med. Chem. 16 (2008) 8940–8951
8941
crease its antibacterial potency.19,20 However, bactericidal activity
of aminoglycosides was found to be poorly correlated with their A
site affinity.21,22 It is generally accepted that the bactericidal activ-
ity of aminoglycosides is determined by their antitranslational im-
pact on prokaryotic protein synthesis.21
The purposes of the current study were: (1) to employ the im-
pact of the 20-NH2 basicity on the acute toxicity and to design a
new pseudo-disaccharide scaffold with potentially low acute toxic-
ity; (2) to compare the basicity of the 20-NH2 between the new and
structurally related pseudo-disaccharide scaffolds and verify the
correlation between the basicity and acute toxicity; (3) to use
the newly designed scaffold for the construction of novel amino-
glycoside derivatives with potent antibacterial activity and low
toxicity.
NH2
O
R2
O
3' NH2
6'
R1
4'
I
I
4'
R1
R2
II
II
NH2
3' NH2
4
NH2
NH2
1 NH2
4
O
O
HO
HO
6
6
O
O
OH
O
O
HO
III
HO
OH
III
NH
H2N
OH
R1 R2
R1
R2
Kanamycin B
Tobramycin
Dibekacin
GentamicinC1 CH3 NHCH3
GentamicinC2 CH3 NH2
GentamicinC1A
OH OH
OH
H
H
H
NH2
H
Towards these ends, here we report the design, synthesis, and
comparative analysis of a series of three pseudo-disaccharides 1–
3 (Fig. 1). The data presented herein verify a relationship between
the basicity of the 20-amine group and the estimated LD50 values in
mice: the increase in the basicity of the 20-amino functionality is
associated with the acute toxicity increase of an aminoglycoside.
Based on the observed data, a new pseudo-trisaccharide 4 (also
called NB45) was constructed, which showed significant antibacte-
rial activity and low acute toxicity.
NH2
NH2
4'
4'
O
O
O
HO
HO
O
3' NH
3'NH2
HO
2: Neamine
NH2
OH
NH2
2 O
O
NH2
NH2
HO
OH
1: NB23
NH2
O
NH2
4'
4'
O
O
O
3'NH
3'NH2
2. Results and discussion
NH2
NH2
2 O
HO
O
NH2
OH
Gentamine C1A
NH2
O
HO
6
2.1. Design hypotheses and synthesis
O
3:
4: NB45
HO
Based on the above-discussed structure–toxicity relationship
analysis, especially the impact of the 20-NH2 basicity on the acute
toxicity of the several aminoglycoside antibiotics, we designed
the pseudo-disaccharide 1 (also named NB23, Fig. 1) as a minimal
structural motif with the following expectations. First, the struc-
ture 1 consists of a 30,40-methylidene protection as a potential
0defense0 against toxicity; the preservation of 30,40-oxygens in 1
should keep the lower basicity of the 20-NH2 group and subse-
quently the lower toxicity than those in the parallel 30,40-dideoxy
analog 3. Second, the methylidene group should also protect 1
from the action of various APH(30) and ANT(40) resistance enzymes.
Third, the 30,40-methylidene protection is supposed to be substan-
tially stable under both the acid and base conditions usually used
in carbohydrate chemistry, and should easily be constructed from
the corresponding 30,40-diol (see the synthesis bellow).
To test the validity of our design, we initially synthesized three
pseudo-disaccharides, compounds 1–3 (Fig. 1), and determined
their acute toxicity along with the pKa values of the individual ami-
no groups in each compound. As a starting material for the synthe-
sis of 1 we used the compound 2 (also named neamine, Fig. 1),
which was readily accessible in a large quantity by direct acid
hydrolysis of neomycin B according to the previously reported pro-
cedure23 (Scheme 1A). Simultaneous conversion of all the amino
groups of 2 into the corresponding azides (TfN3, CuSO4) was fol-
lowed by selective protection of the 5,6-hydroxyl groups by cyclo-
hexylidene ketal to afford 30,40-diol 5. Treatment of 5 with two
equivalents of CH2Br2 under base conditions of NaOH, followed
by hydrolysis of cyclohexylidene ketal with aqueous acetic acid,
afforded the 30,40-methylidene derivative 6. The Staudinger reac-
tion then generated the desired product 1.
NH
OH
Figure 1. Chemical structures of a series of natural and synthetic 2-deoxystrepta-
mine-derived (ring II) aminoglycosides, including compounds 1–4 that were
investigated in this study.
on aminoglycosides structure–toxicity relationship16 revealed that
one of the factors that significantly affect the acute toxicity of ami-
noglycosides is the deletion of the ring hydroxyl group(s) (deoxy-
genation) that is adjacent to the amino group on the same ring.
For example, deletion of 30-OH in kanamycin
B (Fig. 1,
LD50 = 132) gives the significantly more toxic tobramycin
(LD50 = 79). Further deletion of 40-OH in tobramycin results in
dibekacin (30,40-dideoxykanamycin B, LD50 = 71) with only a mar-
ginal increase in toxicity. The observed increase in the toxicity of
the deoxy-analogs could be explained by an increase in the basicity
of the 20-amino group due to the removal of the neighboring elec-
tron-withdrawing oxygen atom (30-OH). Furthermore, the removal
of the 30-OH which is adjacent to the 20-NH2, has much more influ-
ence than removal of the 40-OH, which is more distant from this
amine. Similar results have been obtained by replacement of the
5-OH with 5-fluorine in kanamycin B and its several clinical deriv-
atives.17,18 The toxicities of the resulting fluoro analogs were sig-
nificantly lower than the parent compounds and this
phenomenon again was attributed to basicity reduction of the 3-
NH2 group induced by the strongly electron-withdrawing 5-fluo-
rine. Thus, significantly high acute toxicity of the clinical drugs
such as tobramycin (30-deoxy), gentamicin (30,40-dideoxy), dibeka-
cin (30,40-dideoxy) and arbekacin (30,40-dideoxy) could be ascribed
to the increased basicity of 20-NH2 group (ring I) in these drugs
caused mainly because of the lack of 30-hydroxyl or 30,40-hydroxyl
groups, respectively.
The third pseudo-disaccharide derivative, compound 3 (also
named gentamine C1A, Fig. 1) was most readily accessible from
the mixture of commercial gentamicin in three steps (Scheme
1B). Treatment of gentamicin (a mixture of at least five com-
pounds: gentamicin C1A, gentamicin C1 and gentamicin C2, while
the latter two are a mixture of the corresponding C60-diastereo-
mers, Scheme 1) with anhydrous HCl (AcCl in MeOH) gave a highly
On the other hand, the increase in basicity of a particular amine
group in the aminoglycoside structure may increase its binding
affinity to the target ribosomal RNA A site and subsequently in-

8942
L. Chen et al. / Bioorg. Med. Chem. 16 (2008) 8940–8951
1-N 3-N 2’-N
N3
O
A
A
pH
6’-N
4'
HO
HO
3.40
4.83
5.73
6.53
a
b
3'N3
N3
2
O
N3
O
c
5
6
O
1-N
3-N
2’-N
6’-N
5
N3
O
4'
7.23
8.33
O
1-N
3-N
2’-N
6’-N
1
O
3' N3
N3
3-N 1-N
2’-N
6’-N
6’-N
O
N3
9.26
HO
OH
6
10.10
11.71
3-N
1-N
2’-N
B
R1
O
R2NH
6'
OMe
O
40
36
32 28
24 20
ppm
16
12
+
d
HO
Gentamicin
NH2
H2N
NH
OH
O
NH2
HO
4-H
6-H 6’-H
7
OH
B
3’-H 2’-H 6’-H
8a: R1=R2=CH3
N3
6'
8b: R1=CH3, R2=H
5’-H
2-H-eq
c
e
O
2'
8c:
R1=R2=H
3
ppm
N3
N3
O
N3
HO
15
20
25
30
OH
9
Scheme 1. Reagents and conditions: (a) i—TfN3, Et3N, CuSO4, in CH2Cl2/MeOH/H2O
3:10:3, 90%; (ii) CSA, DMF, cyclohexanonedimethylketal, 65%; (b) i—CH2Br2 (2
equiv), THF/aqueous NaOH (10 M) 1:1, nBu4NBr, 85%; ii—H2O/1,4-dioxane/HOAc
1:1:2, 85%; (c) PMe3 (1 M in THF), THF/NaOH (0.1 M) 3:1, 95–99%; (d) AcCl, MeOH,
86%; (e) i—TfN3, Et3N, CuSO4, in CH2Cl2/MeOH/H20 3:10:3, 93%; ii—flash column
chromatography. CSA, camphor sulfonic acid.
4.0
3.5
3.0
ppm
2.5
2.0
regioselective hydrolysis between the rings II and III to afford
garoseamine derivative 7 (ring III of gentamicin) and a mixture
of the corresponding pseudo-disaccharides 8. Treatment of this
mixture (8) with TfN3 converted all the primary amines to the cor-
responding azides from which the desired component 9 could be
easily separated by flash chromatography. The Staudinger reaction
on 9 furnished the desired product 3 with excellent purity and iso-
lated yield. The structures of 1–3 were confirmed by a combination
of various 1D and 2D NMR techniques, including 2D 1H–13C HMQC
and HMBC, 2D COSY, and 1D selective TOCSY experiments, along
with mass spectral analysis.
Figure 2. (A) Natural abundance 1 D 15N NMR spectra of compound 1 as a function
of pH. The assignments of each nitrogen atom at particular pH are indicated. (B) 2D
1H–15N HMBC spectrum of compound 1 at pH 13.05. The assignments of the
relevant hydrogen and nitrogen atoms are indicated. For the detailed information
about the acquisition of both 1D and 2D spectra see Section 4.
As a representative example, Figure 2A illustrates 15N NMR
complete titration spectra for the compound 1. As can be clearly
seen from these spectra, the complex dependence of the 15N chem-
ical shifts on pH is observed, especially at intermediate pH values,
which made difficult to differentiate between 1-N from 3-N, and
20-N from 60-N. Similar problem was noted earlier by Botto and
Coxon for neomycin B.25 To solve this problem, the latter group
used 15N spin-relaxation experiments in the presence of
Gd[2.2.1] cryptate as a spin-labeling reagent to provide complete
2.2. Determination of the pKa values of the individual amine
groups in compounds 1–3 by using natural abundance 15N NMR
To evaluate the basicity of individual amine groups, natural-
abundance 15N NMR spectra of the free base forms of 1–3 as a func-
tion of pH were recorded. Assignments for individual 15N signals in
each structure was done by chemical shift comparisons with those
previously reported for the neamine part (rings I and II) of neomy-
cin B24–26 and of other aminoglycosides.19,27,28 To further secure
our assignments, we performed complete set of standard NMR
experiments (including 1H, 13C, 1H–13C HMQC and HMBC, 2D COSY,
and 1D selective TOCSY) along with the 2D 1H–15N HMBC spectra
at the low- and high-pH extremes as well as at several intermedi-
ate pHs. While the combination of standard NMR experiments al-
lowed unambiguous assignment of all the hydrogen atoms
assignment of all nitrogen atoms resonances. We used 2D 1H–15
N
HMBC technique as the method of choice to confirm the assign-
ments of all 15N resonances at these problematic pH values as well
as at low- and high-pH extremes. As a representative example,
Figure 2B illustrates the 1H–15N HMBC spectrum of compound 1
at pH 13. The observed correlations: 60-N to 50-H and two 60-Hs;
20-N to 30-H and 20-H; 1-N to 6-H and 2-Heq; 3-N to 4-H and 2-
Heq; unequivocally secured the assignment of all nitrogen atoms
of the compound 1 at this particular pH. Plotting of 15N chemical
shifts as a function of pH (Fig. 3) gave smooth sigmoid titration
curves through the points for each individual nitrogen atoms,
which further confirmed the validity of our assignments over the
entire pH range.
present in each structure, the 2D 1H–15
N HMBC technique
unequivocally secured the assignment of individual 15N signals in
each structure (1–3) by showing correlations between 1H and
15N (Fig. 2).

L. Chen et al. / Bioorg. Med. Chem. 16 (2008) 8940–8951
8943
Table 1
15N NMR-derived pKa values and associated 15N chemical shifts for the individual
amine groups of compounds 1–3 at 25 °Ca
Aminoglycoside
pKa
dNH3 (ppm)b
dNH2 (ppm)b
Compound 1
1-N
8.13 0.30
6.79 0.05
6.02 0.04
8.30 0.05
38.4 1.1
36.9 0.2
34.2 0.3
25.1 0.3
29.8 0.9
30.5 0.1
21.1 0.1
12.4 0.2
3-N
20-N
60-N
Compound 2
1-N
8.64 0.23
7.07 0.08
7.20 0.05
9.30 0.07
38.1 1.0
37.0 0.3
32.6 0.3
25.7 0.5
29.4 0.8
30.7 0.1
20.8 0.2
13.2 0.5
3-N
20-N
60-N
Compound 3
1-N
8.04 0.13
6.67 0.10
8.40 0.04
9.04 0.06
37.9 0.5
36.8 0.3
38.2 0.2
26.0 0.4
29.0 0.4
30.3 0.2
29.5 0.1
15.0 0.3
3-N
20-N
60-N
The pKa values and chemical shifts (d) of protonated (NHþ3 ) and deprotonated
(NH2) amine groups were derived from fits of the pH-dependent 15N chemical shift
data shown in Figure 2 using Eq. 1. The indicated uncertainties in the values of pKa
and d reflect the standard deviations of the experimental data from the fitted
curves.
a
b
15N chemical shifts were calibrated according to the IUPAC unified scale39 using
1H chemical shift referenced to internal DSS = 0.0 ppm.
The influence of this effect is further manifested in a significant
downfield shift of the 20-amine resonance in compound 3 than that
in 2, with
DdNH2 being approximately 8 ppm and DdNH3 being
about 5 ppm (see Fig. 3). Similar influence of the presence/absence
of the 30-OH group on the estimated pKa values and 15N chemical
shifts of the 20-amine of several natural aminoglycosides have pre-
viously reported.19,29
Interestingly, the pKa value of the 20-amine in 1 (pKa 6.02) is
even lower then that of the corresponding amine in 2 (pKa 7.2),
and is almost 2.4 pKa units below the 20-amine in 3 (pKa 8.4). To
our knowledge, a pKa value of 6.02 for 20-amine of the 30,40-meth-
ylidene derivative 1 is the lowest pKa value ever estimated to this
amine group in any natural or synthetic aminoglycoside reported
to date. Two different factors, electronic and steric, are likely oper-
ating synergistically to make certain such a low basicity of this
amine. In view of the electronic effect, previous measurements of
pKa values of amino sugar derivatives indicate that 3-amino-3-
deoxyglycopyranosides are slightly more basic than their 2-OMe
derivatives30 due to a larger inductive effect (-I) of the 2-OMe
group than that of the 2-OH group. In addition, the higher basicity
of the 1-amine to that of the 3-amine of the 2-DOS ring (ring II) in
neomycin B has ascribed by a larger inductive effect of the pyrano-
syloxy group substituent at position 4 than that of the hydroxyl
group at position 6.25 Based on these observations, the methylid-
eneoxy group at 30-position of 1 may be expected to have a larger
inductive effect than the 30-OH group in 2. Consequently, 20-amine
in 2 is expected to be more basic than this amine in 1, as indeed
were observed (Table 1). In view of the steric effect, the lower ba-
sicity of the 20-amine in 1 may be further ascribed due to its prox-
imity to the 30,40-methylidene five-member ring that hinders
proper solvation of the corresponding ammonium anion.24 This
steric effect is likely operational for the 60-amine too whose pKa va-
lue in compound 1 was estimated to be 1 pKa unit lower (pKa 8.3)
than that in compound 2 (pKa 9.3).
Figure 3. The plots of 15N chemical shifts for individual amino groups versus pH of
compound 1 (A), compound 2 (B), and compound 3 (C). The data points reflect the
chemical shifts of the indicated amino groups. These chemical shifts are reported
according to the IUPAC unified scale39 using 1H chemical shift referenced to
DSS = 0.0 ppm. The continuous lines reflect the calculated fits of the data using
Eq. 1.
The pKa values of individual amine groups in each compound
(1–3) were determined from nonlinear least-squares fits of the
pH profiles using Eq. 1.19
ðdNH ꢀ dNH Þð10pH-pK
Þ
a
3
2
d ¼
þ dNH
ð1Þ
3
1 þ ð10pH-pK
Þ
a
In this relationship, dNH and dNH are the 15N chemical shifts of the
3
2
amino nitrogen in their NH3 and NH2 states, respectively. The re-
sulted pKa values are given in Table 1.
As expected, the major difference between the compounds 1–3
emerges with respect to the pKa values of the 20-amine group. The
pKa value of this amine in compound 3 (pKa 8.4) is 1.2 pKa units
greater than the corresponding pKa value in compound 2 (pKa
7.2). The observed increase in the basicity of 20-amine is likely
due to the absence of the 30-OH group in compound 3, while this
is present in 2. The electron withdrawing inductive effect (-I) of
the neighboring 30-oxygen causes significant decrease in the elec-
tron density on the 20-amine and subsequently lowers its basicity.
2.3. Acute lethal toxicity values (LD50) of 1–3 and structure–
toxicity relationship
Having established the impact of the 30,40-methylidene protec-
tion on the basicity of the 20-amine in compound 1, we then at-
tempted to test if the basicity of 20-amine can correlate with

8944
L. Chen et al. / Bioorg. Med. Chem. 16 (2008) 8940–8951
Table 2
(MIC = 192 lg/mL), compound 1 (MIC = >384 lg/mL) was lack of
Acute toxicity data in mice, steady-state kinetic parameters for the interaction with
APH(30)-IIb, and antibacterial activity of compounds 1–3.
antibacterial potency (with the concentration of 384 mg/mL as
the maximum starting concentration of each compound tested).
Thus, even thogh compound 1 exhibited the lowest acute toxicty
in comparison to 2 and 3, its lack of antibacterial activty was highly
concerned. Therefore, we asked the following question: despite of
its lack of significant antibacterial potency, whether or not com-
pound 1 can serve as a potential scaffold for the construction of
new generation of aminoglycosides with diminished toxicity?
To answer this question, initially we focused our design strategy
on the clinically used antibiotic gentamicin. This drug is used as a
mixture of several related structures (Fig. 1) and shows potent
antibacterial activity against various pathogens. Unfortunately,
however, the use of gentamicin in clinic is largely limited because
of its high toxicity (LD50 values for gentamicin C1, gentamicin C2
and gentamicin C1A were reported as 88, 70, and 70 mg/kg, respec-
tively).16 Our strategy was to mimic the structural features of gen-
tamicin in order to preserve its antibacterial activity and at the
same time to reduce its toxicity. For this purpose, since the com-
mercial gentamicin is a mixture of gentamicin C1, gentamicin C2
and gentamicin C1A (Fig. 1), initially we constructed the new pseu-
do-trisaccharide 4 (NB45, Fig. 1) which is a close structural analog
of gentamicin C1A. In the following section, the synthesis and com-
parative antibacterial activities of NB45 and gentamicin C1A are
outlined.
Compound
LD50 (mg/kg)a
303 10
Kinetic datab
Substrate activity Ki (
Antibacterial
activity (l
g/mL)c
l
M)
1
2
—
642 53
—
384<
96
161
5
Km = 10. 3 lM
kcat = 1.7 sꢀ1
3
103
8
–
650 61
192
a
LD50 refers to a median lethal dose. The indicated uncertainties in the values of
LD50 reflect the standard deviations of the experimental data from the fitted dos-
age-mortality curves using Grafit 5 software.41
b
The substrate (Km and kcat values) and inhibitory (Ki values) activities were
determined with the recombinant APH(30)-IIb enzyme as outlined in Section 4.
c
The MIC values were determined against Escherichia coli (R477-100) strain by
using the double-microdilution method, with 384 lg/mL as starting concentration
of each tested compound. All the experiments were performed in duplicates and
analogous results were obtained in three different experiments.
acute toxicity of the aminoglycoside. For this purpose the compar-
ative acute intravenous toxicity values of 1–3 were determined in
mice under the same experimental conditions and the observed
data are shown in Table 2. These data suggest a relationship be-
tween the 20-amine basicity and the estimated acute LD50 values;
the increase in the basicity of 20-amine is associated with the tox-
icity increase of the aminoglycoside. With a pKa value of 6.02 for 20-
amine, compound 1 is considerably less toxic (LD50 303 mg/kg)
than neamine (2, pKa 7.2, LD50 160 mg/kg) and has about three
times lower toxicity to that of the corresponding 30,40-dideoxy
derivative (3, pKa 8.4, LD50 103 mg/kg). This result is consistent
with the earlier reported structure–toxicity relationship of various
fluorinated analogs of kanamycin B.17,18 Furthermore, since the
acute toxicity data of aminoglycosides were found to be well cor-
related to their neuromuscular blocking effects11,12, we suggest
that the reduction in the pKa of 20-amine is linked to the acute tox-
icity attenuation of aminoglycoside due to the subsequent reduced
effects on neuromuscular blocking. This suggestion is supported by
earlier observations which demonstrated that total charge of an
aminoglycoside correlates with the blocking potency31,32, and that
co-administration of Ca2+ antagonizes both acute toxicity and neu-
romuscular blocking effects induced by aminoglycosides.12,13
2.5. Synthesis and comparative antibacterial activity of NB45
and gentamicin C1A
NB45 was assembled by using compound 6 as an acceptor and
the garosamine fragment 7 (Scheme 1) as a precursor of the donor
(Scheme 2A). First, the anomeric mixture of 7 was converted to the
thioglycoside 10 in three steps; selective protection of the second-
ary amine with trichloroethyloxycarbonyl (Troc) group was fol-
lowed by selective benzoylation of the secondary alcohol and by
replacement of the anomeric methoxyde with
Note that the garosamine ring (ring III) in gentamicin is an unusual
-sugar with the b- -arabino configuration [3-deoxy-4-C-methyl-3-
(methylamino)-b- -arabinose] (Fig. 1 and Scheme 2). Therefore, in
order to guarantee the formation of the desired b- -glycosidic link-
a-thioglycoside.
L
L
L
L
age between the garosamine ring (ring III) and the 2-DOS ring (ring
II) in NB45, the benzoate protection in 10 was replaced with the p-
methoxybenzyl ether protection. Treatment of 10 with methyl-
amine (33% MeNH2 in EtOH) resulted in simultaneous removal of
the benzoate and the formation of the oxazolidinone ring as con-
firmed by subsequent structural analysis of the product. Treatment
of this product with NaH and p-methoxybenzylcloride then fur-
nished the oxazolidinone donor 11. NIS-AgOTf-promoted glycosi-
dation of acceptor 6 with the donor 11 gave the corresponding
2.4. Interaction with the resistance enzyme APH(30)-IIb and
antibacterial activity of 1–3
In order to verify the contribution of the 30,40-methylidene
group in protection of 1 against action of resistance enzymes, com-
pound 1 along with the compounds 2 and 3 were examined for
their substrate/inhibitory activities against APH(30)-IIb33 enzyme
(Table 2). We found that, as per design, while the compound 2
functioned as
a substrate (Km = 10.3 0.5 lM and kcat = 1.7
pseudo-trisaccharide 12 as a mixture of anomers (b/a 4:1) in 90%
0.1 sꢀ1), the compounds 1 and 3 exhibited no substrate activity
isolated yield. Oxidative removal of the p-methoxybenzyl ether
protection by treatment with CAN was followed by treatment with
strong base (aqueous NaOH in EtOH) to open the oxazolidinone
ring. Finally, the Staudinger reaction furnished the desired product
and demonstrated only poor inhibitory activities with the esti-
mated apparent Ki values of 642 52 and 650 62 lM
respectively.
From the earlier studies21, it is well known that because of its
smaller size and the global charge in comparison to those of known
aminoglycoside antibiotics, compound 2 exhibits very poor anti-
bacterial activity. Since compound 1 display significantly lower
20-amine pKa than that of the compounds 2 and 3, its binding affin-
ity to the ribosomal RNA and the subsequent antibacterial activity
was expected to be even poorer than that of the compounds 2 and
3. Indeed, from the estimated minimal inhibitory concentration
(MIC) values against Escherichia coli R477-100 (Table 2) it can be
NB45 as a mixture of anomers (b/a 4:1).
Our attempts to separate the desired gentamicin C1A from the
mixture of commercial gentamicin, by using direct column chro-
matography procedures, were unsuccessful. We solved this prob-
lem by subjecting the commercial mixture to two-steps
procedure (Scheme 2B). Treatment of gentamicin mixture with
TfN3 converted all the primary amines to the corresponding azides,
from which the desired component 13 could be easily separated by
flash chromatography. The Staudinger reaction on 13 furnished the
desired product gentamicin C1A with excellent purity and isolated
yield.
seen that while compound 2 (MIC = 96
lg/mL) exhibited slightly
better antibacterial activity than that of compound
3

L. Chen et al. / Bioorg. Med. Chem. 16 (2008) 8940–8951
8945
APH(30)-IIb enzyme33,34, and the multidrug efflux system MexXY.35
P. aeruginosa is a pathogenic bacterium which is a major cause of
mortality among cystic fibrosis patients.36 In recent years, in-
creased numbers of P. aeruginosa clinical isolates were found to
be resistant to several antibiotics which prevent its effective treat-
ment. Therefore, in the current study, in addition to the standard P.
aeruginosa ATCC 27853 strain, we also included clinical isolates of
this bacterium, P. aeruginosa (584/5) and P. aeruginosa (590) (ob-
tained from Rambam Medical Center Hospital, Haifa, Israel).
As seen in Table 3, in general, the observed spectrum of antibac-
terial activities of NB45 is very similar to that of gentamicin C1A. It
is of particular importance to note that the two compounds show
good antibacterial activities against the APH(30) expressing strains
(E. coli XL1(pET9d), and strains of P. aeruginosa) indicating that
APH(30) enzymes are incapable of modifying these compounds be-
cause the lack of free 30-OH group in these structures. We have re-
cently cloned the chromosomal gene aph(30)-IIb that encodes an
aminoglycoside 30-phosphotransferase in P. aeruginosa, and over-
expressed it in E. coli.33 In the latter study we also have demon-
strated that the aminoglycosides tobramicin and gentamicin
which lack 30-OH group, were not substrates to the purified
APH(30)-IIb and that both P. aeruginosa and E. coli strains harboring
APH(30)-IIb were sensitive to these antibiotics. To further verify
that the observed sensitivity of the P. aeruginosa strains to NB45
is due to the lack of the free 30-OH group in this compound, we per-
formed kinetic analysis of NB45 with the purified APH(30)-IIb en-
zyme. We found that NB45 has no substrate activity and
exhibited only poor inhibitory activity with the estimated apparent
α-anomer
A
TolS
BzO
TolS
PMBO
O
b
O
a
7
N
OH
N
O
O
O
CCl3
10
11
O
N3
O
O
c
d
O
N3
O
4
N3
O
N3
HO
(β/α 4:1)
O
PMBO
N
O
12
O
N3
B
O
e
N3
O
Gentamicin
N3
O
N3
HO
β-anomer
13
O
HO
NH
OH
Ki value of 420 38 lM.
f
Gentamicin C1A
In general, antibacterial activity (MIC values) of aminoglyco-
sides correlates with their in vitro prokaryotic antitranslational po-
tency (IC50 values).21 Since both NB45 and gentamicin C1A
exhibited similar antibicaterial activity against a series of both
Gram-negative and Gram-positive bacteria, it was therefore ex-
pected that they would also exhibit similar prokaryotic antitrans-
lational potency. To verify this expectation, we used an in vitro
luciferase assay21 and tested NB45 and gentamicin C1A for the inhi-
bition of protein translation in prokaryotic system, and the ob-
served data are shown in Fig. 4 and Table 3. The measured half-
Scheme 2. Reagents and conditions: (a) i—2,2,2-Trichloroethyl chloroformate,
CHCl3/H2O 5:1, NaHCO3, 93%; ii—BzCl, pyridine, 4-DMAP, 80%; iii—4-methyl
benzenethiol, BF3ꢁEt2O, C2H4Cl2, 40 °C, 53%; (b) i—33% MeNH2 (33% soln in EtOH),
24 h, 90%; ii—PMBCl, NaH, DMF,nBu4NBr, HMPA, 82%; (c) i—6, NIS, AgOTf, Et2O/
CH2Cl2:1, ꢀ10 °C, 24 h, 90% (b/
NaOH (2 M) 1:1, reflux, 10 h; iii—PMe3 (1 M in THF), THF/NaOH (0.1 M) 3:1, 75% [for
2 steps (b/ 4:1)]; (e) i—TfN3, Et3N, CuSO4, in CH2Cl2/MeOH/H2O 3:10:3, 55%; (ii)
a 4:1); (d) i—CAN, CH3CN, 85%; ii—EtOH/ aqueous
a
flash column chromatography; (f) PMe3 (1 M in THF), THF/NaOH (0.1 M) 3:1, 93%.
4-DMAP, 4-dimethyl aminopyridine; HMPA, hexamethylphosphoramide, PMBCl, p-
methoxybenzyl chloride; NIS, N-iodosuccinimide, CAN, cerium ammonium nitrate.
maximal inhibitory concentration (IC50
) values indicate that
NB45 (IC50 = 38 nM) is only slightly better inhibitor than gentami-
cin C1A (IC50 = 59 nM). These data are consistent with the antibac-
terial data obtained for NB45 and gentamicin C1A. Taken together,
the observed data suggest that the 30,40-methylidene protection in
NB45 not just protects it from the deactivation by APH(30) en-
zymes, but it also does not perturb its interaction with the ribo-
somal target or its penetration through the bacterial membrane
that could influence its antibacterial activity.
The comparative antibacterial activities of NB45 and gentamicin
1A were determined by measuring the minimal inhibitory concen-
C
trations (MICs) against both Gram-negative and Gram-positive
bacteria, including resistant and pathogenic strains, using the mic-
rodilution assay (Table 3). Resistant strains included E. coli
XL1(pET9d), and Pseudomonas aeruginosa. E. coli XL1(pET9d) is an
antibiotic-sensitive laboratory strain that harbors plasmid pET9d
with the cloned aminoglycoside kinase APH(30)-Ia (Novagen Inc.).
P. aeruginosa strains possess several different resistance mecha-
nisms to aminoglycosides, including the chromosomal encoded
Encouraged by the impacts of the 30,40-methylidene scaffold in
NB45 on its antibacterial activity and interaction with the APH(30)
resistance enzymes, we were interested whether this scaffold
Table 3
a
Translation inhibition, antibacterial activity and acute toxicity data of the NB45 and gentamicin C1A
Aminoglycoside
Translation inhibition IC50
(l
M)b
Antibacterial activity MIC (
l
g/mL)c
Acute toxicity LD50 (mg/kg)d
Ec-1
Ec-2
Ec-3
Se
Bs
Pa
Pa-1
Pa-2
NB45
Gentamicin C1A
0.038 0.003
0.059 0.01
48
48
48
96
12
12
24
6
12
6
24
12
24
48
48
384
226
1
7016
a
In all biological tests, both NB45 and gentamicin C1A were in their sulfate salt forms.
The half-maximal inhibitory concentration (IC50) values were obtained from concentration response curves (Fig. 3) using Grafit 5 software.41
The MIC values were determined by using the double-microdilution method, with 384 lg/mL as starting concentration of the tested compound. All the experiments were
b
c
performed in duplicates and analogous results were obtained in three different experiments. The bacterial strains used: Ec-1, Escherichia coli (R477-100); Ec-2, Escherichia coli
(ATCC 25922); Ec-3, E. coli XL1 blue/ pET9d expressing the APH(30)-Ia resistance enzyme; Se, Staphylococcus epidermis (ATCC 12228); Bs, Bacilus subtilis; Pa, Pseudomonas
aeruginosa ATCC 27853; clinical isolates of P. aeruginosa: Pa-1, P. aeruginosa (584/5); Pa-2, P. aeruginosa (590).
d
LD50 refers to a median lethal dose. The indicated uncertainty in the LD50 value reflects the standard deviation of the experimental data from the fitted dosage-mortality
curve using Grafit 5 software.41

8946
L. Chen et al. / Bioorg. Med. Chem. 16 (2008) 8940–8951
properties were compared to the structurally related gentamicin
1A. NB45 exhibited similar antibacterial activity to that of genta-
C
micin C1A against various bacterial strains, including pathogenic
and resistant strains, while at the same time retained significantly
low toxicity to that of gentamicin C1A. An in vitro prokaryotic anti-
translational activity, along with the substrate activity tests with
the purified APH(30)-IIb resistance enzyme, revealed that both
NB45 and gentamicn C1A inhibit protein translation process with
the similar extent and both lack substrate activity for the
APH(30)-IIb enzyme with NB45 exhibiting poor inhibitory activity.
Based on these data it is suggested that the presence of 30,40-meth-
0
ylidene protection in compound NB45 provides a defense0 against
the acute toxicity as well as against the action of APH(30) resistance
enzymes while at the same time it retains significant antibacterial
potency.
4. Experimental
Figure 4. Semilogarithmic plots of in vitro translation inhibition in prokaryotic
system of NB45 (d) and gentamicin C1A (s). The translation inhibition was
measured in E. coli cellular extract and quantified in coupled transcription/
translation assays by using active luciferase detection. The percentages of active
luciferase are plotted as functions of drug concentration. Each data point represents
the average of 2–3 independent experimental results.
4.1. General techniques
1H NMR, 13C NMR, DEPT, COSY, 1D TOCSY, HMQC, HMBC, spec-
tra were recorded on a Bruker AvanceTM 500 spectrometer. Chem-
ical shifts reported (in ppm) are relative to internal Me4Si (d = 0.0)
with CDCl3 as the solvent, and to HOD (d = 4.63) with D2O as the
solvent. Mass spectral analyses were performed on a Bruker Dalto-
nix Apex 3 mass spectrometer under electron spray ionization
(ESI), TSQ-70B mass spectrometer (Finnigan Mat) or under MAL-
could also provoke the low toxicity of NB45. For this purpose, the
acute intravenous toxicity value of NB45 was determined and the
16
observed value was compared to that of the gentamicin C1A
which was determined earlier under the same experimental condi-
tions (Table 3). The data in Table 3 show that NB45 is about three
times less toxic than gentamicin C1A (LD50 values of 226 and
70 mg/kg, respectively). Surprisingly, the similar three-fold differ-
ence in LD50 values was observed between their parent pseudo-
disaccharide scaffolds 1 and 3 (LD50 values of 303 and 103 mg/
kg, respectively; Table 2), suggesting that the presence of 30,40-
methylidene moiety should be responsible in acute toxicity atten-
uation of NB45 relative to that of the gentamicin C1A. The observed
somewhat higher toxicity of NB45 (LD50 226) to that of its parent
pseudo-disaccharide scaffold 1 (LD50 303) can be explained by
the difference in the number of amine groups present in these mol-
ecules (five in NB45 versus four in compound 1) and in their size
(NB45 consists of three rings versus two rings in compound 1). In-
deed, several earlier studies demonstrated that aminoglycosides
with different size and charge interfere differently the steps of cal-
cium-mediated acetylcholine release at the level of presynaptic
structures, and that this interaction is the main cause of the neuro-
muscular blockage induced by aminoglycosides and their subse-
quent toxicity.9,11,37
DI-TOF on a
a-cyano-4-hydroxycinnamic acid matrix on a MALDI
Micromass spectrometer. Reactions were monitored by TLC on Sil-
ica gel Gel 60 F254 (0.25 mm, Merck), and spots were visualized by
charring with a yellow solution containing (NH4)6Mo7O24ꢁ4H2O
(120gr) and (NH4)2Ce(NO3)6 (5gr) in 10% H2SO4 (800 mL). Flash
column chromatography was performed on Silica gel Gel 60 (70–
230 mesh). All reactions were carried out under an argon atmo-
sphere with anhydrous solvents, unless otherwise noted. All chem-
icals, unless otherwise stated, were obtained from commercial
sources. DSS [3-(trimethylsilyl)-1-propanesulfonic acid], neomycin
B, and gentamicin were obtained from Sigma. Compound 2 (nea-
mine) was prepared in multigram quantities from neomycin B as
previously reported.23
4.2. Synthetic part
4.2.1. Compound 5
Neamine (compound 2) was converted to its perazido deriva-
tive, 1,3,20,60-tetraazido neamine by using earlier reported38 azi-
do-transfer reaction. The isolated product (90% isolated yield),
after flash chromatography (EtOAc/hexane), has exactly same
spectroscopic properties as it was originally reported.38 The
1,3,20,60-tetraazido neamine (2.0 g, 4.70 mmol) was dissolved in
DMF (70 mL) and to the resulted mixture were added cyclohaxa-
none dimethyl ketal (8.0 mL, 51.60 mmol) and catalytic amount
of camphor sulfonic acid (CSA). The reaction was heated to 60 °C
under argon and the reaction progress was monitored by TLC
(EtOAc/hexane 1:1). After 6 h, the mixture was diluted with EtOAc
and washed with NaHCO3 and brine. The combined organic layer
was dried over MgSO4, evaporated, and purified by flash chroma-
tography (EtOAc/hexane) to give compound 5 (1.5 g, 65%). 1H
NMR (500 MHz, CDCl3) d: 1.61 (bs, 10H, cyclohexane protons), ring
I: 3.25 (dd, 1H, J1 = 3.5, J2 = 10.5 Hz, H-20), 3.50–3.64 (m, 3H, H-40,
H-60, H-60), 3.94 (t, 1H, J = 9.5 Hz, H-30), 3.99 (m, 1H, H-50), 5.55
(d, 1H, J = 3.5 Hz, H-10); ring II: 1.46 (ddd, 1H, J1 = J2 = J3 = 12.5 Hz,
H-2ax), 2.31 (bdt, 1H, H-2eq), 3.42 (t, 1H, J = 9.5 Hz, H-6), 3.50–3.64
(m, 3H, H-1, H-3, H-4), 3.81 (t, 1H, J = 9.0 Hz, H-5). 13C NMR
(125 MHz, CDCl3) d: 23.7 (cyclohexane), 23.8 (cyclohexane), 24.8
(cyclohexane), 33.8 (C-2), 36.0 (cyclohexane), 36.2 (cyclohexane),
3. Summary and conclusions
In the current study, the correlation between acute toxicity and
the 20-NH2 basicity of some aminoglycosides was investigated. For
this purpose, we have designed a new pseudo-disaccharide 1 with
0
a 30,40-methylidene protection as a potential defense0 against tox-
icity, and its properties were evaluated in comparison to the re-
lated pseudo-disaccharides, compounds 2 and 3. By determining
the pKa values of the individual amino groups in each structure
and the acute toxicities of each structure we have demonstrated
a relationship between the basicity of the 20-amine group and
the estimated LD50 values in mice: the increase in the basicity of
the 20-amino functionality is associated with the toxicity increase
of an aminoglycoside with the compound 1 exhibiting the lowest
pKa value for the 20-amine and being substantially least toxic than
the other two derivatives (compounds 2 and 3).
The pseudo-disaccharide 1 was then used as a scaffold on which
the new pseudo-trisaccharide NB45 was constructed, and its

L. Chen et al. / Bioorg. Med. Chem. 16 (2008) 8940–8951
8947
51.2 (C-60), 57.2, 57.3, 60.8, 62.6, 71.2, 71.3, 71.4, 79.3, 79.3, 96.2
(C-10), 113.8 (cyclohexane). ESI m/z 529.2 (M+Na+, C18N12O6H26 re-
quires 529.2).
and the resulted solid was re-dissolved in D2O. 1H NMR (500 MHz,
D2O) d: ring I 3.07–3.15 (m, 2H, H-40, H-60), 3.33 (t, 1H, J = 3.0 Hz,
H-20), 3.30–3.31 (m, 1H, H-60), 3.66 (dd, 1H, J1 = 9.0, J2 = 2.0 Hz, H-
30), 4.24 (m, 1H, H-50), 5.64 (d, 1H, J = 4.0 Hz, H-10); ring II 1.62
(ddd, 1H, J1 = J2 = J3 = 12.5 Hz, H-2ax), 2.22 (dt, 1H, J1 = 12.5,
J2 = 4.0 Hz, H-2eq), 3.07–3.15 (m, 2H, H-1, H-3), 3.43 (t, 1H,
J = 6.0 Hz, H-6), 3.53 (t, 1H, J = 9.0 Hz, H-5), 3.59 (t, 1H, J = 9.5 Hz,
H-4), 5.02 (s, 1H, CH2), 5.08 (s, 1H, CH02)13C NMR (125 MHz, CDCl3)
d: 32.0 (C-2), 42.4 (C-60), 50.2, 52.0, 55.5 (C-20), 70.8 (C-50), 74.5 (C-
4), 77.0, 77.1, 78.9 (C-30), 82.3 (C-6), 98.1 (CH2), 100.3 (C-10). ESI m/
z 357.2 (M+Na+, C13N4O6H26 requires 357.2).
4.2.2. Compound 6
To a stirred solution of compound 5 (2.0 g, 3.95 mmol) in THF
(80 mL) was added an aqueous solution of 10M NaOH (70 mL) at
room temperature. After being stirred for 30 min, nBu4NBr (3.8 g,
11.8 mmol) and CH2Br2 (3.0 mL, 7.9 mmol) were added and the
reaction was heated to 60 °C. The reaction progress was monitored
by TLC (EtOAc/hexane 1:3). After 40 h, the mixture was diluted
with CH2Cl2 and washed with NaHCO3 and brine. The combined or-
ganic layer was dried over MgSO4, evaporated, and purified by
flash chromatography (EtOAc/hexane) to give the corresponding
30,40-methylidene product (1.8 g, 85%). 1H NMR (500 MHz, CDCl3)
d: 1.66–1.70 (m, 10H, cyclohexane), ring I: 3.19 (t, 1H, J1 = 9.5 Hz,
H-40), 3.14–3.55 (m, 2H, H-20, H-60), 3.60 (dd, 1H, J1 = 2.5,
J2 = 13.5 Hz, H-60), 3.86 (t, 1H, J = 9.0 Hz, H-30), 4.29–4.32 (m, 1H,
H-50), 5.66 (d, 1H, J = 3.5 Hz, H-10); ring II: 1.52 (ddd, 1H,
J1 = J2 = J3 = 12.5 Hz, H-2ax), 2.33 (dt, 1H, J1 = 5, J2 = 13.5 Hz, H-
2eq), 3.14–3.55 (m, 3H, H-6, H-1, H-3), 3.66 (m, 1H, H-4), 3.70
(dd, 1H, J1 = 9.0, J2 = 11.0 Hz, H-5), 5.15 (s, 1H, CH2), 5.18 (s, 1H,
CH02). 13C NMR (125 MHz, CDCl3) d: 25.5 (C-20), 25.5 (cyclohexane),
26.7 (cyclohexane), 35.6 (cyclohexane), 37.8 (cyclohexane), 38.0
(cyclohexane), 53.5 (C-60), 59.0, 62.6, 62.9, 73.6, 78.8, 79.0, 79.1,
81.1, 81.1, 98.4 (CH2), 98.6 (C-10) 115.7 (cyclohexane). ESI m/z
541.2 (M+Na+, C19N12O6H26 requires 541.2).
4.2.4. Compound 9
To a stirred solution of the commercial gentamicin (5.0 g,
0.01 mol) in MeOH (10 mL) was added AcCl (5 mL) by a dropwise
addition at 0 °C. The mixture was heated at reflux for 6 h, cooled
and evaporated to dryness. The residue was purified by flash chro-
matography [CH2Cl2/MeNH2 solution (33% solution in EtOH) 15:1];
first eluted the garosamine 7 (2.6 g, 90%) having the Rf value of
0.93, followed by the mixture of 8a–c (2.5 g, 86%) with Rf value
in the range of 0.20–0.46 [CH2Cl2/MeNH2 solution (33% solution
in EtOH) 3:2]. Data for compound 7: 1H NMR (500 MHz, CDCl3)
a
anomer: d 1.06 (s, 3H, CH3), 2.38 (s, 3H, N-Me), 2.20 (d, J = 10 Hz,
1H, H-3), (s, 3H, OMe), 3.28 (dd, J1 = 7.5, J2 = 10.5 Hz, 1H, H-2),
(3.40 (d, J = 12 Hz, 1H, H-5), 3.50 (d, J = 12 Hz, 1H, H-50), 4.16 (d,
J = 9.5 Hz, 1H, H-1). 13C NMR (125 MHz, CDCl3) d: 20.3 (CH3),
36.1(N-Me), 54.7 (OMe), 62.6 (C-3), 66.3 (C-4), 66.4 (C-5), 68.3
(C-2), 104.2 (C-1). MALDI-TOFMS calculated m/z 192.1 (M+H+,
C8H17O4N requires 192.1). b anomer: d 1.04 (s,3H, CH3), 2.36 (s,
3H, N-Me), 2.38 (d, J = 11 Hz, 1H, H-3), 3.21 (s, 3H, OMe), 3.30 (d,
J = 12 Hz, 1H, H-5), 3.61 (dd, J1 = 3, J2 = 11 Hz, 1H, H-2), 3.65 (d,
J = 12 Hz, 1H, H-50), 4.63 (d, J = 4 Hz, 1H, H-1). 13C NMR
(125 MHz, CDCl3) d 20.7 (CH3), 36.1(N-Me), 54.7 (OMe), 62.6 (C-
3), 66.3 (C-4), 66.4 (C-5), 68.3 (C-2), 98.9 (C-1). MALDI-TOFMS cal-
culated m/z 191.1 (M+H+, C8H17O4N requires 192.1).
The purified mixture 8a–c (2.5 g, 8.33 mmol) was dissolved in
H2O (80 mL) to which CuSO4 (0.20 mg, 1.31 mmol), triethylamine
(30 mL, 215 mmol), triflic azide solution (0.5 M in CH2Cl2 150 mL,
75 mmol) and MeOH (200 mL) were added. The resulted mixture
was stirred for 18 h at room temperature and concentrated under
reduced pressure. The residue was purified by flash chromatogra-
phy (EtOAc/hexane) to yield 9 (750 mg, 1.52 mmol) 93% (calcu-
lated from the 32% gentamicin C1A present in the commercial
gentamicin according to the manufacturer0s protocol). 1H NMR
(500 MHz, CDCl3) d: ring I: 1.58 (dd, 1H, J1 = 4.0, J2 = 12.0 Hz, H-
40), 1.76 (dd, 1H, J1 = 3.0, J2 = 13.0 Hz, H-40), 1.97–2.05 (m, 2H, H-
30, H-30), 3.26–3.27 (m, 2H, H-60, H-60), 3.39–3.41 (m, 1H, H-20),
4.18–4.21 (m, 1H, H-50), 5.25 (d, 1H, J = 3.0 Hz, H-10); ring 2:
1.22 (ddd, 1H, J1 = J2 = J3 = 12.5 Hz, H-2ax), 1.48 (dt, 1H, J1 = 12.5,
J2 = 4.0 Hz, H-2eq), 3.3–3.45 (m, 4H, H-1, H-3, H-5, H-6), 3.51 (t,
1H, J = 9.0 Hz, H-4). 13C NMR (125 MHz, CDCl3) d: 21.9 (C-2), 27.0
(C-30), 32.1 (C-40), 54.5 (C-60), 58.6, 59.2, 59.7, 67.9 (C-50), 75.4,
76.3, 81.6, 98.3 (C-10). MALDI-TOFMS m/z 417.2 (M+Na+
C12H1804N12 requires 417.2).
The product from the previous step (600 mg, 1.15 mmol), was
dissolved in 1,4-dioxane (7 mL) and water (7 mL). To the resulted
mixture was added glacial acetic acid (15 mL), heated at reflux
and the reaction progress was monitored by TLC (EtOAc/hexane
7:13). After about 14 h, the mixture was diluted with EtOAc and
washed with NaHCO3 and brine. The combined organic layer was
dried over MgSO4, evaporated, and purified by flash chromatogra-
phy (EtOAc/hexane) to yield 6 (450 mg, 85%). 1H NMR (500 MHz,
CDCl3) d: ring I: 3.20 (t, 1H, J = 9.0 Hz, H-40), 3.49 (dd, 1H,
J1 = 5.5, J2 = 13.5 Hz, H-60), 3.61 (dd, 1H, J1 = 2.5, J2 = 13.5 Hz, H -
60), 3.69 (dd, 1H, J1 = 4.0, J2 = 11.0 Hz, H-20), 3.84 (dd, 1H, J1 = 9.0,
J2 = 11.0 Hz, H-30), 4.36–4.40 (m, 1H, H-50), 5.47 (d, 1H, J = 4.0 Hz,
H-10); ring II: 1.53 (ddd, 1H, J1 = J2 = J3 = 12.5 Hz, H-2ax), 2.34
(bdt, 1H, H-2eq), 3.35–3.42 (m, 4H, H-4, H-1, H-3, H-6), 3.53 (t,
1H, J1 = 9.0, J2 = 13.5 Hz, H-5), 5.16 (s, 1H, CH2), 5.17 (s, 1H, CH0 ),
2
13C NMR (125 MHz, CDCl3) d: 33.8 (C-2), 53.5 (C-60), 60.7, 61.6,
64.3, 73.6, 78.1, 78.6, 78.8, 79.1, 84.1, 98.5 (CH2), 100.7 (C-10). ESI
m/z 461.1 (M+Na+, C13N12O6H18 requires 461.1).
4.2.3. Compound 1
To a stirred solution of compound 6 (200 mg, 0.45 mmol) in THF
(3.0 mL) was added aqueous solution of NaOH 0.1 M (1.5 mL), fol-
lowed by addition of PMe3 (1 M solution in THF, 3.6 mL,
3.60 mmol) at room temperature. The reaction progress was mon-
itored by TLC [CH2Cl2/MeOH/H2O/MeNH2 (33% solution in EtOH)
10:15:6:15], which indicated completion after 2 h. The reaction
mixture was directly poured on a short column of silica gel and
purified by flash chromatography. The column was washed
sequentially with three column volumes of THF, three column vol-
umes of dichloromethane, and three column volumes of MeOH.
The product was then eluted with the mixture of 20% MeNH2 solu-
tion (33% solution in EtOH) in 80% MeOH. Fractions containing the
product were combined and evaporated under vacuum. The resi-
due was re-dissolved in small volume of water and evaporated
again (2–3 repeats) to afford the free amine form of 1 (150 mg,
99%). This product was dissolved in water, the pH was adjusted
to 6.8 with H2SO4 (0.1 N) and lyophilized to afford the sulfate salt
of 1 as a white foamy solid. For spectral analysis, the pH of the final
product was adjusted to about 3.0 with H2SO4 (0.1 N), lyophilized
4.2.5. Compound 3
Compound 9 was subjected to a Staudinger reaction as de-
scribed above for compound 1 [compound 9 (600 mg, 1.52 mmol),
THF (7 mL), NaOH (0.1 M, 3.0 mL), PMe3 (1 M solution in THF,
12 mL, 12.0 mmol)] to yield 3 as a free amine form (420 mg,
95%). The resulted free amine was dissolved in water, the pH was
adjusted to 6.9 with H2SO4 (0.1 N) and lyophilized to give the sul-
fate salt as a white foamy solid. For the spectral analysis, the pH of
the final product was adjusted to about 3.0 with H2SO4 (0.1 N),
lyophilized and the resulted solid was re-dissolved in D2O. 1H
NMR (500 MHz, D2O) d: ring I: 1.25 (dd, 1H, J1 = 3.5, J2 = 12.0 Hz,

8948
L. Chen et al. / Bioorg. Med. Chem. 16 (2008) 8940–8951
H-40), 1.61–1.64 (m, 1H, H-40), 1.51 (dd, 1H, J1 = 3.5, J2 = 12.5 Hz, H-
30), 1.55–1.58 (m, 1H, H-30), 2.50-2.55 (m, 2H, H-60, H-60), 2.71–2.73
(m, 1H, H-20), 3.70-3.74 (m, 1H, H-50), 4.99 (d, 1H, J = 3.5 Hz, H-10);
ring 2: 1.04 (ddd, 1H, J1 = J2 = J3 = 12.5 Hz, H-2ax), 1.82 (dt, 1H,
J1 = 12.5, J2 = 4.0 Hz, H-2eq), 2.53–3.57 (m, 1H, H-1), 2.68–2.70
(m, 1H, H-3), 2.98 (t, 1H, J = 9.0 Hz, H-6), 3.12 (t, 1H, J = 9.0 Hz,
H-4), 3.31 (t, 1H, J = 9.0 Hz, H-5). 13C NMR (125 MHz, D2O) d:
27.4 (C-30), 29 .0 (C-40), 37.3 (C-2), 47.2 (C-60), 51.2, 51.4, 52.0 (C-
1), 71.7 (C-50), 77.8 (C-5), 79.3 (C-6), 89.3 (C-4), 103.2 (C-10). MAL-
DI-TOFMS m/z 329.2 (M+K+ C12H2604N4 requires 329.2).
To a mixture of anomers from the previous step (400 mg,
0.85 mmol) in dichloroethane (6 mL) were added 4-methyl ben-
zenethiol (200 mg, 1.7 mmol) and boron trifluoride diethyl ether-
ate (0.2 mL, 1.7 mmol). The reaction was heated to 40 °C and the
reaction progress was monitored by TLC (EtOAc/hexane 3:7, two
runs). After 1 h the mixture was diluted with EtOAc and washed
with NaHCO3 and brine. The combined organic layer was dried
over MgSO4, evaporated, and purified by flash chromatography
(EtOAc/hexane) to give compound 10 as a single anomer, anomer
a
(250 mg, 53%). 1H NMR (500 MHz, CDCl3) d: 1.20 (s, 3H, CH3),
2.31 (s, 3H, CH3-STol), 3.08 (s, 3H, N-Me), 3.63 (d, J = 12 Hz, 1H,
CH2), 3.77 (d, J = 12 Hz, 1H, CH2), 4.52 (d, J = 11 Hz, 1H, H-3), 4.57
(d, J = 12 Hz, 1H, H-5), 4.73 (d, J = 12 Hz, 1H, H-50), 4.85 (d,
J = 9.5 Hz, 1H, H-1), 5.57 (t, J1 = 9.5, J2 = 11 Hz, 1H, H-2), 7.10 (d,
J = 6.5 Hz, 2H, aromatic), 7.35–7.46 (m, 4H, atomatic), 7.59 (t,
J = 7 Hz, 1H, p-OBz), 8.04 (d, J = 11 Hz, 2H, aromatic). 13C NMR
(125 MHz, CDCl3) d: 19.4 (CH3), 20.7 (N-Me), 30.3 (Me-STol), 60.4
(C-3), 66.6 (C-2), 73.3 (C-5), 77.3 (CH2), 88.7 (C-1), 128.4, 128.5,
129.2, 129.7, 129.8, 129.9, 130, 133.3, 133.4 (aromatic), 156.1 (car-
bonyl), 165 (carbonyl) MALDI-TOFMS calculated m/z 584.1
(M+Na+, C24H26Cl306NS requires 584.1).
4.2.6. Compound 10
The garosamine fragment 7 from the previous steps (100 mg,
0.52 mmol) was dissolved in CHCl3 (2.5 mL) and H2O (0.5 mL)
2,2,2-trichloroethyl chloroformate (0.09 mL, 0.68 mmol) and NaH-
CO3 (80 mg, 1.04 mmol) were added. The reaction was monitored
by TLC (EtOAc/hexane 7:1). After 30 min, the mixture was diluted
with EtOAc and washed with brine. The combined organic layer
was dried over MgSO4, evaporated, and purified by flash chroma-
tography (EtOAc/hexane) to give the selective N-protected product,
a mixture of
a
and b anomers with Troc protection (180 mg, 93%).
anomer: d 1.17 (s, 3H, CH3), 3.10 (s, 3H,
1H NMR (500 MHz, CDCl3)
a
N-Me), 3.45 (s, 3H, OMe), 3.63 (d, J = 12.5 Hz, 1H, CH2), 3.73 (d,
J = 12.5 Hz, 1H, CH2), 3.81 (t, J1 = 5.5, J2 = 7.5 Hz, 1H, H-2), 4.49 (d,
J = 11 Hz, 1H, H-3), 4.53 (d, J = 7.5 Hz, 1H, H-1), 4.60 (d, J = 12 Hz,
1H, H-5), 4.75 (d, J = 12 Hz, 1H, H-50), 13C NMR (125 MHz, CDCl3)
d: 20.7 (CH3), 31.5 (N-Me), 56.5 (OMe), 60.4 (C-3), 67.5 (C-2),
73.3 (C-5), 75.3 (CH2), 105.4 (C-1), 155.1 (carbonyl), 156.7 (car-
bonyl). MALDI-TOFMS calculated m/z 388.0 (M+Na+, C11H18Cl3O6N
requires 388.1). b anomer: d 1.22 (s, 3H, CH3), 3.05 (s, 3H, N-Me),
3.40 (s, 3H, OMe), 3.36 (d, J = 12.5 Hz, 1H, CH2), 3.71 (dd, J1 = 4,
J2 = 5.5 Hz, 1H, H-2), 3.92 (d, J = 12.5 Hz, 1H, CH2), 4.68 (d,
J = 12 Hz, 1H, H-5), 4.79–4.86 (m, 2H, H-3, H-50), 5.16 (d,
J = 3.5 Hz, 1H, H-1). 13C NMR (125 MHz, CDCl3) d: 21.7 (CH3),
30.9 (N-Me), 55.5 (OMe), 56.4 (C-3), 67.5 (C-2), 74.3 (C-5), 75.0
(CH2), 99.2 (C-1), 156.2 (carbonyl), 165.6 (carbonyl). MALDI-TOF-
MS calculated m/z 388.0 (M+Na+, C11H18Cl3O6N requires 388.1).
The mixture of anomers from the previous step (1 g, 2.73 mmol)
was dissolved in pyridine (15 mL) and BzCl (0.9 mL, 6.82 mmol)
and 4DMAP (catalytic amount) were added at room temperature.
The reaction was monitored by TLC (EtOAc/hexane 2:3). After 1 h
the mixture was diluted with EtOAc and washed with aqueous
HCl (2%) and brine. The combined organic layer was dried over
MgSO4, evaporated, and purified by flash chromatography
(EtOAc/hexane) to give two separate anomers with benzoate pro-
4.2.7. Compound 11
Compound 10 (1 g, 1.78 mmol) was dissolved in 33% methyl-
amine solution in EtOH and the resulted mixture was stirred over-
night at room temperature. The reaction progress was monitored
by TLC (EtOAc/hexane 1:1). After about 24 h, the reagent and the
solvent were removed by evaporation and the crude product was
purified by flash chromatography (EtOAc/hexane) to yield the cor-
responding oxazolidinone product (500 mg, 90%). 1H NMR
(500 MHz, CDCl3) d: 1.38 (s, 3 H, CH3), 2.29 (s, 3H, CH3-STol),
2.95 (s, 3H, N-Me), 3.33 (d, J = 6 Hz, 1H, H-3), 3.53 (d, J = 12.5 Hz,
1H, H-5), 3.73 (t, J1 = 6.5, J2 = 13.5 Hz, 1H, H-2), 4.06 (d,
J = 12.5 Hz, 1H, H-50), 4.74 (d, J = 7.5 Hz, 1H, 1-H), 7.09 (d,
J = 7.5 Hz, 2H, aromatic), 7.36 (d, J = 7.5 Hz, 2H, aromatic). 13C
NMR (125 MHz, CDCl3) d 21.0 (CH3- STol), 23.7 (CH3-C-5), 30.4
(N-Me), 65.8 (C-3), 68.2 (C-5), 70.4 (C-2), 88.3 (C-1), aromatic car-
bons: (128.4, 129.8, 132.8, 132.8), 157.4 (carbonyl).
The product from the previous step (70 mg, 0.22 mmol) was
dissolved in DMF (3 mL), and to the resulted mixture were added
p-methoxybenzylchloride (0.04 mL, 0.28 mmol), tetra-n-butyl-
ammonium iodide (104 mg, 0.28 mmol) and hexamethylphospho-
ramide (0.5 mL) at room temperature. The reaction was then
cooled to ꢀ10 °C (ice-water–salt bath) and NaH (8.1 mg,
0.35 mmol) was added. The reaction was brought to room temper-
ature and the reaction progress was monitored by TLC (EtOAc/hex-
ane 2:3). After about 1 h the reaction was diluted with EtOAc,
washed with brine, dried over MgSO4 and concentrated. The crude
material was purified by flash chromatography (EtOAc/hexane) to
yield 11 (80 mg, 82%): 1H NMR (500 MHz, CDCl3) d: 1.50 (s, 3H,
CH3), 2.34 (s, 3H, CH3-STol), 2.86 (s, 3H, N-Me), 3.41 (d, J = 4 Hz,
1H, H-3), 3.61 (d, J = 12 Hz, 1H, H-5), 3.80 (t, J1 = 6, J2 = 1.5 Hz,
1H, H-2), 3.80 (s, 3H, OMe), 4.15 (d, J = 12 Hz, 1H, H-50), 4.49 (d,
J = 11.5 Hz, 1H, CH2-PMB), 4.77 (d, J = 11.5 Hz, 1H, CH2-PMB),
5.24 (d, J = 3 Hz, 1H, 1-H), 6.88 (d, J = 9 Hz, 2H, OPMB-aromatic),
7.14 (d, J = 7.5 Hz, 2H, STol-aromatic), 7.25 (d, J = 9 Hz, 2H,
OPMB-aromatic), 7.37 (d, J = 7.5 Hz, 2H, STol-aromatic). 13C NMR
(125 MHz, CDCl3) d: 21.0 (CH3-STol), 23.9 (CH3-C-5), 30.2 (N-Me),
55.2 (OMe), 64.0 (C-3), 66.2 (C-5), 72.5 (CH2-PMB), 75.0 (C-2),
85.9 (C-1), aromatic carbons: (113.9, 113.9, 128.8, 129.4, 129.8,
130.2, 132.0, 137.8), carbonyl carbon (157.4). MALDI-TOFMS calcu-
lated m/z 452.2 (M+Na+, C23H2705NS observed 452.1).
tection (1 g, 80%). 1H NMR (500 MHz, CDCl3)
a anomer: d 1.20 (s,
3H, CH3), 3.10 (s, 3H, N-Me), 3.52 (s, 3H, OMe), 3.63 (d,
J = 12.5 Hz, 1H, CH2), 3.73 (d, J = 12.5 Hz, 1H, CH2), 4.49 (d,
J = 11 Hz, 1H, H-3), 4.53 (d, J = 7.5 Hz, 1H, H-1), 4.60 (d, J = 12 Hz,
1H, H-5), 4.75 (d, J = 12 Hz, 1H, H-50), 5.54 (t, J1 = 5.5, J2 = 7.5 Hz,
1H, H-2), 7.42 (t, J = 7 Hz, 2H, m-OBz), 7.55–7.58 (m, 1H, p-OBz),
7.99 (d, J = 7 Hz, 1H, o-OBz). 13C NMR (125 MHz, CDCl3) d: 20.7
(CH3), 31.5 (N-Me), 56.5 (OMe), 60.4 (C-3), 67.5 (C-2), 73.3 (C-5),
75.3 (CH2), 104.1 (C-1), 128.4, 129.3, 129.9, 130.1, 133.4 (aro-
matic), 156.1 (carbonyl), 165 (carbonyl). MALDI-TOFMS calculated
m/z 493.0 (M+Na+, C18H22Cl307N requires 493.0). b anomer: d 1.22
(s, 3H, CH3), 3.05 (s, 3H, N-Me), 3.40 (s, 3H, OMe), 3.36 (d,
J = 12.5 Hz, 1H, CH2), 3.92 (d, J812.5 Hz, 1H, CH2), 4.68 (d,
J=12 Hz, 1H, H-5), 4.79–4.86 (m, 2H, H-3, H-50), 5.16 (d, J=3.5 Hz,
1H, H-1), 5.51 (dd, J1=4, J2=5.5 Hz, 1H, H-2), 7.42 (t, J=7 Hz, 2H,
m-OBz), 7.54–7.57 (m, 1H, p-OBz), 7.99 (d, J=7 Hz, 1H, o-OBz). 13C
NMR (125 MHz, CDCl3) d: 21.7 (CH3), 30.9 (N-Me), 55.5 (OMe),
56.4 (C-3), 67.5 (C-2), 74.3 (C-5), 75.0 (CH2), 97.4 (C-1), 128.4,
129.2, 129.9, 130.1, 133.4 (aromatic), 156.2 (carbonyl), 165 (car-
bonyl). MALDI-TOFMS calculated m/z 493.0 (M+Na+, C18H22Cl3O7N
observed 493.0).
4.2.8. Compound 12
To powdered, flame dried 4 Å molecular sieves (300 mg) was
added a mixture of diethyl ether (2.5 mL) and dichloromethane

L. Chen et al. / Bioorg. Med. Chem. 16 (2008) 8940–8951
8949
(0.5 mL), followed by the addition of acceptor
6
(100 mg,
amine form of compound 4 (NB45) as a mixture of anomers (b/
a
4:1) (35 mg, 75% for two steps). The free amine was dissolved in
water, the pH was adjusted to 6.8 with H2SO4 (0.01 M) and lyoph-
ilized to give the sulfate salt of the product. For the spectral anal-
ysis of the product, the pH of the free amine was adjusted to about
3.0, lyophilized, and re-dissolved in D2O. Data for b-4: 1H NMR
(500 MHz, D2O) d 1.42 (s, 3H, CH3), 2.80 (s, 3H, N-Me). Ring 1:
3.25 (t, J = 10 Hz, 1H, H-40) 3.40–3.41 (m, 2H, H-60, H-60), 3.79–
3.81 (m, 1H, H-20), 3.96 (t, J = 9 Hz, 1H, H-30), 4.37–4.40 (m, 1H,
0.22 mmol) and donor 11 (127 mg, 0.29 mmol). After being stirred
10 min at room temperature, the mixture was treated with NIS
(72.4 mg, 0.29 mmol). After additional 5 min at room temperature,
the mixture was cooled to ꢀ40 °C and AgOTf (76 mg, 0.29 mmol)
was added. The reaction was brought to ꢀ10 °C and was left at this
temperature until completion (about 14 h) while the reaction pro-
gress was monitored by TLC (EtOAc/hexane 3:7, three runs). The
reaction was diluted with EtOAc, filtered through celite, and the
celite was washed thoroughly with EtOAc. The combined EtOAc
fraction was extracted with saturated NaHCO3, brine, dried over
MgSO4 and concentrated. The crude material was purified by flash
chromatography (EtOAc/hexane) to yield the corresponding pseu-
do-trisaccharide product 12 (150 mg, 90%) as a mixture of anomers
H-50), 6.03 (d, J = 4 Hz, 1H, H-10). Ring
2: 1.95 (ddd,
J1 = J2 = J3 = 12.5 Hz, 1H, H-2ax), 2.43 (bdt, 1H, H-2eq), 3.46–3.48
(m, 2H, H-1, H-3), 3.77–3.81 (m, 2H, H-4, H-6), 4.05–4.07 (m, 1H,
H-5). Ring 3: 3.40–3.42 (m, 1H, H-300), 3.70–3.77 (m, 1H, H-500),
3.94–3.96 (m, 1H, H-500), 4.11–4.14 (m, 1H, H-2000), 5.02 (d,
J = 3.5 Hz, 1H, H-100), 5.08 (s, 1H, CH2), 5.09 (s, 1H, CH2). 13C NMR
(125 MHz, D2O) d: 20.2 (CH3), 23.8 (N-Me), 27.0 (C-2), 39.9 (C-
60), 47.7, 49.0, 51.9, 62.6, 65.6, 65.7, 67.1 (C-500), 68.9, 73.0, 74.4,
75.7, 76.1, 82.7, 85.2, 97.3 (C-10), 98.8 (CH2), 102.9 (C-100). MAL-
DI-TOFMS m/z 532.3 (M+K+, C20H39O9N5 requires 532.0).
(b/a 4:1). Following is given the spectral data for the b-12 anomer.
1H NMR (500 MHz, CDCl3) d: 1.38 (s, 3 H, CH3), 2.80 (s, 3H, N-Me),
3.79 (s, 3H, OMe). Ring 1: 3.19 (t, J = 9 Hz, 1H, H-40) 3.47–3.57 (m,
3H, H-2, H-60, H-60), 3.85 (t, J = 9 Hz, 1H, H-30), 4.40-4.44 (m, 1H, H-
50), 5.84 (d, J = 3.5 Hz, 1H, H-10). Ring
2: 1.15 (ddd,
J1 = J2 = J3 = 12.5 Hz, 1H, 2-ax), 2.36 (bdt, 1H, H-2eq), 3.33–3.45
(m, 2H, H-1, H-3,), 3.52–3.58 (m, 2H, H-4, H-6,), 3.72 (t, J = 9.5,
1H, H-5). Ring 3: 3.42–3.42 (m, 1H, H-300), 3.97–3.99 (m, 1H, H-
200), 4.51 (d, J = 11.5 Hz, 1H, H-500), 4.82 (d, J = 11.5 Hz, 1H, H-500),
4.96 (d, J = 3.5 Hz, 1H, H-100). 3.81 (d, J = 11 Hz, 1H, CH2OBn), 4.14
Data for a
-4: 1H NMR (500 MHz, D2O) d 1.24 (s, 3H, CH3), 3.10
(s, 3H, N-Me) Ring 1: 3.21 (dd, J1 = 10, J2 = 8.5 Hz, 1H, H-40), 3.50
(dd, J1 = 14, J2 = 5 Hz, 1H, H-60), 3.59 (dd, J1 = 14, J2 = 5 Hz, 1H, H-
60), 3.64–3.69 (m, 1H, H-20), 3.82 (t, J = 10.5 Hz, 1H, H-30), 4.35–
4.40 (m, 1H, H-50), 5.49 (d, J = 3.5 Hz, 1H, H-10). Ring 2: 1.48
(ddd, J1 = J2 = J3 = 12.5 Hz, 1H, 2-ax), 2.27 (dt, J1 = 4, J2 = 13 Hz, 1H,
H-2eq), 3.31–3.48 (m, 4H, H-1,H-3, H-4, H-6), 3.70-3.68 (m, 1H,
H-5). Ring 3: 3.28 (d, J = 8 Hz, 1H, H-300), 3.42 (d, J = 12.5 Hz, 1H,
H-500), 3.84–3.89 (m, 1H, H-200), 4.01 (d, J = 12.5 Hz, 1H, H-5000),
4.72 (d, J = 7 Hz, 1H, H-100), 5.11 (s, 1H, CH2), 5.13 (s, 1H, CH2).
13C NMR (125 MHz, D2O) d: 14.1 (CH3), 20.7 (N-Me), 30.7 (C-2),
51.5 (C-60), 58.7, 58.9, 62.2, 65.3, 67.0, 71.7 (C-500), 73.5, 75.2,
76.1, 77.3, 77.8, 81.6, 83.5, 96.6 (CH2), 98.9 (C-10), 104.1 (C-100),
157.4 (cabonil). MALDI-TOFMS m/z 532.3 (M+K+, C20H39O9N5 re-
quires 532.0).
(d, J = 11 Hz, 1H, CH2OBn), 5.17 (s, 1H, CH2), 5.24 (s, 1H, CH0 ),
2
6.90 (d, J = 6.5 Hz, 2H, aromatic), 7.25 (d, J = 6.5 Hz, 2H, aromatic).
13C NMR (125 MHz, CDCl3) d: 22.9 (CH3), 30.0 (N-Me), 31.4 (C-2),
51.1 (C-60), 55.2, 58.8, 59.4, 61.4, 62.2, 62.8, 67.1 (C-500), 71.1,
72.1 (CH2-benzyl), 74.2, 75.1, 76.6, 76.8, 79.3, 83.9, 92.6, 96.0
(CH2-methylidene), 96.7 (C-10), 97.6 (C-100), aromatic carbons:
113.3, 113.4, 128.9, 129.2. MALDI-TOFMS m/z 766.3 (M+Na+,
C29H37O11N13 requires 766.0).
4.2.9. Compound 4 (NB45)
To a stirred solution of 12 (50 mg, 0.06 mmol) in CH3CN (5 mL)
was added a solution of CAN (110 mg, 0.20 mmol) in minimum
volume of water at ꢀ10 °C. The reaction was then brought to room
temperature and the reaction progress was monitored by TLC
(EtOAc/hexane 1:1). After 3 h, the reaction was diluted with EtOAc,
washed with NaHCO3 and brine. The combined organic layer was
dried over MgSO4, evaporated, and purified by flash chromatogra-
phy (EtOAc/hexane) to give the desired de-benzylated product
(36 mg, 85%). 1H NMR (500 MHz, CDCl3) d: 1.43 (s, 3H, CH3), 2.91
(s, 3H, N-Me). Ring 1: 3.20 (dd, J1 = 8.5, J2 = 10 Hz, 1H, H-40)
3.48–3.58 (m, 3H, H-20, H-60, H-60), 3.84 (dd, J1 = 9, J2 = 10.5 Hz,
1H, H-30), 4.40–4.44 (m, 1H, H-50), 5.82 (d, J = 4 Hz, 1H, H-10). Ring
2: 1.15 (ddd, J1 = J2 = J3 = 12.5 Hz, 1H, H-2ax), 2.36 (dt,, J1 = 4,
J2 = 13 Hz, 1H, H-2eq), 3.30-3.59 (m, 4H, H-1, H-3, H-4, H-6), 3.65
(t, J = 9, 1H, H-5). Ring 3: 3.68 (d, J = 3.5 Hz, 1H, H-300), 3.76 (d,
J = 11.5 Hz, 1H, H-500), 4.10 (d, J = 11.5 Hz, 1H, H-500), 4.26 (t,
J = 3.5 Hz, 1H, H-200), 4.90 (d, J = 3 Hz, 1H, H-100). 5.13 (s, 1H, CH2),
5.15 (s, 1H, CH02). 13C NMR (125 MHz, CDCl3) d 21.0 (CH3), 30.2
(N-Me), 31.8 (C-2), 51.7 (C-60), 58.8, 59.1, 60.4, 61.5, 62.9, 65.0,
67.0 (C-500), 71.8, 73.8, 75.1, 75.5, 76.0, 83.5, 95.6, 96.5 (CH2),
96.6 (C-10), 99.2 (C-100) 157.3 (carbonyl). MALDI-TOFMS m/z 646.2
(M+Na+, C21H29O10N13 requires 646.0).
The purified product from the previous step (60 mg, 0.09 mmol)
was dissolved in EtOH (12 mL) and aqueous solution of NaOH (2 M,
12 mL) was added. The reaction was heated to 90 °C and the reac-
tion progress was monitored by TLC [CH2Cl2/MeNH2 solution (33%
solution in EtOH) 20:1]. After being stirred 10 h at 90 °C, the reac-
tion was diluted with EtOAc and washed with brine. The combined
organic layer was dried over MgSO4 and evaporated. The residue
(60 mg, 0.09 mmol) was then subjected to a Staudinger reaction
as described above for the compound 6 with the following quanti-
ties: THF (2 mL), NaOH 0.1 M (1.5 mL), PMe3 (0.77 mL) to yield free
4.2.10. Compound 13
Commercial gentamicin (15 g, 0.05 mol) was dissolved in H2O
(25 mL) to which CuSO4 (0.58 g, 9 mmol), triethylamine (40 mL,
280 mmol), TfN3 solution (0.5 M, in 30 mL CH2Cl2, 50 mmol) and
MeOH (100 mL) were added. The reaction was monitored by TLC
(EtOAc/hexane 1:1). The resulted mixture was stirred for 18 h at
room temperature, after which concentrated under reduced pres-
sure. The residue was purified by flash chromatography (silica,
EtOAc/hexane) to yield 13 (2.5 g, 55%), (calculated from the 32%
gentamicin C1A present in the commercial gentamicin according
to the manufacturer0s protocol). 1H NMR (500 MHz, CDCl3) d Ring
1: 1.56 (dd, 1H, J1 = 4, J2 = 12 Hz, H-40), 1.75 (dd, 1H, J1 = 3,
J2 = 13 Hz, H-40), 1.90–1.93 (m, 1H, H-30), 2.07 (dd, 1H, J1 = 3.5,
J2 = 12.5 Hz, H-30), 3.22–3.26 (m, 3H, H-60, H-60, H-20), 4.20–4.24
(m, 1H, H-50), 5.52 (d, 1H, J = 3 Hz, H-10). Ring 2: 1.22 (ddd, 1H,
J1 = J2 = J3 = 12.5 Hz, H-2ax), 1.48 (dt, 1H, J1 = J2 = 4 Hz, H-2eq),
3.3–3.45 (m, 4H, H-1, H-3, H-5, H-6), 3.51 (t, 1H, J = 9 Hz, H-4). Ring
3: 2.18 (d, 1H, J = 6 Hz, H-300), 3.72 (d, 1H, J = 12.5 Hz, H-500), 3.94 (d,
1H, J = 12.5 Hz, H-500), 4.12(d, 1H, J = 2 Hz, H-200), 5.05 (d, 1H,
J = 3.5 Hz, H-100). 13C NMR (125 MHz, CDCl3) d 21.9 (C-2), 27.0 (C-
30), 31.7 (C-40), 43.7, 54.5 (C-60), 54.7, 58.6, 58.9, 59.7, 65.0, 66.6,
67.9 (C-50), 75.4, 76.3, 79.6, 81.6, 84.9, 97.8 (C-10), 100.8 (C-100).
MALDI-TOFMS m/z 594.0 (M+K+ C19H31O7N13 requires 594.4).
4.2.11. Gentamicin C1A
Compound 13 was subjected to a Staudinger reaction as de-
scribed above for compound 1 under following conditions: com-
pound 13 (600 mg, 1.08 mmol), THF (7 mL), NaOH (0.1 M, 3.0 mL),
PMe3 (1 M solution in THF, 12 mL, 12.0 mmol) to yield gentamicin
C1A as a free amine form (290 mg, 93%). The resulted free amine
was dissolved in water, the pH was adjusted to 6.9 with H2SO4

8950
L. Chen et al. / Bioorg. Med. Chem. 16 (2008) 8940–8951
(0.1 N) and lyophilized to give the sulfate salt a white foamy solid.
For the spectral analysis, the pH of the free amine form was adjusted
to about 3.0, lyophilized and re-dissolved in D2O. 1H NMR (500 MHz,
D2O) d Ring 1: 1.45–1.49(m, 1H, H-40), 1.75–1.78 (m, 1H, H-40), 1.88–
1.94 (m, 2H, H-30, H-30), 2.95–2.98 (m, 1H, H-60), 3.14 (dd, 1H, J1 = 3,
J2 = 13.5 Hz, H-60), 3.46–3.49 (m, 1H, H-20), 4.0–4.04 (m, 1H, H-50),
LD50 of the aminoglycosides. Mice were maintained on a 12 h
light/dark cycle (light from 07:00 to 19:00) at 20ꢀ22 °C, with rel-
ative humidity of 50ꢀ55%, and were given food and water ad libi-
tum in an animal care facility. Eight animals were housed in each
cage. The aminoglycosides tested were administered in single
intravenous (i.v.) doses of 0, 50, 100, 150, 200, 250, 300, 350, 400
and 500 mg/kg, each concentration was applied to 6–8 animals.
5.76 (d, 1H, J = 3.0 Hz, H-10); Ring
2: 1.21 (ddd, 1H,
J1 = J2 = J3 = 12.5 Hz, H-2ax), 2.43 (dt, 1H, J1 = J2 = 4 Hz, H-2eq),
2.35–3.46 (m, 3H, H-1, H-3, H-4), 3.72 (t, 1H, J = 9 Hz, H-6), 3.96 (t,
1H, J = 9 Hz, H-5). Ring 3: 3.35 (d, 1H, J = 12.5 Hz, H-500), 3.66 (d,
1H, J = 6 Hz, H-300), 3.89 (d, 1H, J = 12.5 Hz, H-500), 4.43 (dd, 1H,
J1 = 3.5, J2 = 11 Hz, H-200), 5.07 (d, 1H, J = 3.5 Hz, H-100). 13C NMR
(125 MHz, CDCl3) d: 22.6 (C-2), 27.4 (C-30), 29 (C-40), 37.3, 47.2 (C-
60), 50.3, 51.2, 51.4, 52 (C-1), 65.1, 67.8, 69.6, 71.7 (C-50), 77.8 (C-5),
79.3 (C-6), 85.4, 89.3 (C-4), 102.1 (C-100), 103.6 (C-10). MALDI-TOFMS
m/z 488.3 (M+K+ C19H39O7N5 requires 488.0).
The aminoglycosides were dissolved in PBS, and 150
solution (pH 7) were injected into the tail vein at the injection
speed of 150 L/7 s. Changes in the appearance and behavior,
and mortality in mice were monitored for 4 days after the admin-
istration. The LD50 values were determined according to the Reed
and Muench method.40
lL of final
l
4.5. Biochemical assays
Prokaryotic in vitro translation inhibition by the different ami-
noglycosides was quantified in coupled transcription/translation
assays21 by using E. coli S30 extract for circular DNA with the pBES-
Tluc plasmid (Promega), according to the manufacturer0s protocol.
4.3. 15N NMR spectroscopy and pKa measurements
15N NMR data were acquired in a 11.7 T field on a Bruker Avance
500 spectrometer using a dual channel broad-band probe with z-
gradients and 2H lock at 25 °C. Aqueous solutions (85% H2O/ 15%
2H2O) of an aminglycoside free base form (at concentration of
about 270–300 mg/mL) at natural abundance 15N were placed into
5.0 mm o.d. glass NMR tubes (Wilmad). 1D 1H NMR spectra
(500.13 MHz resonance frequency) were typically acquired in a
Translation reactions (25 lL) that contained variable concentra-
tions of the tested aminoglycoside were incubated at 37 °C for
60 min, cooled on ice for 5 min, and diluted with a dilution reagent
(tris-phosphate buffer (25 mM, pH 7.8), DTT (2 mM), 1,2-diam-
inocyclohexanetetraacetate (2 mM), glycerol (10 %), Triton X-100
(1 %) and BSA (1 mg mLꢀ1)) into 96-well plates. The luminescence
was measured immediately after the addition of the Luciferase As-
single shot using a 41.6 ls dwell time and 64k points then Fourier
transformed. Assignments of the proton spectra were straightfor-
ward based on chemical shifts, J-couplings and 1D selective TOCSY
experiments. For the 2-DOS ring, standard 2D 1H–13C HMQC and
HMBC experiments ascertained the assignment.
say Reagent (50 lL; Promega), and the light emission was recorded
with a Victor3TM Plate Reader (Perkin-Elmer). The concentration of
half-maximal inhibition (IC50) was obtained from fitting concen-
tration-response curves to the data of at least two independent
experiments by using Grafit 5 software41
1D 15N NMR spectra (50.7 MHz resonance frequency, 18.9 kHz
B1 field) were recorded with 1H-decoupling (waltz-16) during the
acquisition period (inverse gated decoupling). The FIDs were ac-
4.6. Antibacterial activity
quired with a dwell time of 66 ls with 16k digitization points.
The optimal repetition delay was experimentally determined to
be 6s. For good signal-to-noise ratios 1200–2600 transients were
averaged with two dummy scans. Data were processed with zero
filling to 32k using an exponential window function of 2 Hz line
The MIC values were determined using the double-microdilu-
tion method according to the National Committee for Clinical Lab-
oratory Standards (NCCLS)42 with starting concentration of 384
lg/
mL of the tested compound. All the experiments were performed in
triplicates and analogous results were obtained in three different
experiments. The clinical isolates of P. aeruginosa used in this study
were obtained from Rambam Medical Center, Haifa, Israel. The
existence of the chromosomal aph(30)-IIb gene in these strains
was confirmed by PCR analysis, using the same oligonucleotide
primers as we used earlier for the cloning of this gene in E. coli33
(T. Baasov, unpublished data).
broadening. The spectra were calibrated according to the ratio
N
of the resonance frequencies between 15N and 1H nuclei (
N
equal
to 0.10136767)39, where the 1H signal was referenced to internal
DSS = 0 ppm. Assignment of the 15N spectra was based on the as-
signed proton NMR spectra via 2D 1H–15N HMBC. Spectra were ac-
quired with pulsed field gradients and a low-pass J-filter with no
decoupling during acquisition. Inverse detection of the 15N in the
2D experiment averaged at least 32 transients with 16 dummy
scans. At least 2048 ꢂ 32 acquisition points were typically ac-
quired with spectral widths of 3 kHz (1H) and 20–30 kHz (15N).
2D data sets were Fourier transformed in magnitude mode with
zero filling to 1024 ꢂ 512 points with a sine window function ap-
plied in both dimensions.
4.7. Kinetic analysis with the purified APH(30)-IIb
The APH(30)-IIb enzyme was overexpressed and purified using
neomycin B affinity column as previously described.33 The phos-
phoryl transfer activity was monitored by the pyruvate kinase/lac-
tate dehydrogenase coupled assay. The assay measures the rate of
NADH absorbance decrease at 340 nm, which is proportional to
the rate of steady-state ATP consumption. The oxidation of NADH
was followed by continuously monitoring the absorbance at
340 nm using Ultrospec 2100 pro UV/visible spectrophotometer
(Amersham Biosciences). A typical assay mixture contained Tris–
HCl buffer (50 mM, pH 7.5), KCl (40 mM), MgCl2 (10 mM), NADH
(0.11 mg/mL), PEP (2.5 mM), ATP (1 mM), pyruvate kinase (20 u/
mL), lactate dehydrogenase (20 u/mL), varied concentrations of the
The pH of the NMR samples was adjusted by addition of either
HCl or KOH in 85% H2O/15% 2H2O. All pH measurements of NMR
samples were acquired using a sensoreX semi micro electrode pH
meter (SensoreX). The pKa values for compounds 1–3 amino group
were measured by monitoring the 15NMR spectra of each com-
pound at different pH values. Plots of the observed 15N chemical
shifts as a function of pH resulted in well-defined titration curves
(Fig. 2) from which the pKa values were estimated by fitting the ob-
served data to the Eq. 119,27
:
substrate neamine (comp. 2) (1–100
of the inhibitor(0–1000 M). The mixture was preincubatedat 37 °C
for 5 min and the reaction was initiated by addition of purified
APH(30)-IIb (10
L of 0.05 U/mL stock solution). Values of Ki were
determined by nonlinear regression using the program GraFit 5.0.41
lM) and varied concentrations
4.4. Acute toxicity tests
l
Male ICR mice 22 days of age (18ꢀ20 g) purchased from Harlan
l
Laboratories Limited (Jerusalem, Israel) were used to evaluate the

L. Chen et al. / Bioorg. Med. Chem. 16 (2008) 8940–8951
8951
19. Kaul, M.; Barbieri, C. M.; Kerrigan, J. E.; Pilch, D. S. J. Mol. Biol. 2003, 326,
1373.
Acknowledgments
20. Barbieri, C. M.; Kaul, M.; Bozza-Hingos, M.; Zhao, F.; Tor, Y.; Hermann, T.; Pilch,
D. S. Antimicrob. Agents Chemother. 2007, 51, 1760.
We thank Dr. Yael Balazs for her guidance and help in recording
15N NMR spectra. This research was partly supported by the Israel
Science Foundation founded by the Israel Academy of Sciences and
Humanities (Grant No. 515/07), by the US–Israel Binatinal Science
Foundation (Grant No. 2006/301), and by the Technion V.P.R. fund.
21. Greenberg, W. A.; Priestley, E. S.; Sears, P. S.; Alper, P. B.; Rosenbohm, C.;
Hendrix, M.; Hung, S. C.; Wong, C. H. J. Am. Chem. Soc. 1999, 121, 6527.
22. Kaul, M.; Barbieri, C. M.; Pilch, D. S. J. Am. Chem. Soc. 2006, 128, 1261.
23. Park, W. K. C.; Auer, M.; Jaksche, H.; Wong, C. H. J. Am. Chem. Soc 1996, 118,
10150.
24. Blount, K. F.; Zhao, F.; Hermann, T.; Tor, Y. J. Am. Chem. Soc. 2005, 127, 9818.
25. Botto, R. E.; Coxon, B. J. Am. Chem. Soc. 1983, 105, 7.
26. Kaul, M.; Barbieri, C. M.; Srinivasan, A. R.; Pilch, D. S. J. Mol. Biol. 2007, 369,
142.
References and notes
1. Davis, B. D. Microbiol. Rev. 1987, 51, 341.
27. Barbieri, C. M.; Pilch, D. S. Biophys. J. 2006, 90, 1338.
28. Ozen, C.; Serpersu, E. H. Biochemistry 2004, 43, 14667.
29. Dorman, D. E.; Paschal, J. W.; Merkel, K. E. J. Am. Chem. Soc. 1976, 98, 6885.
30. Barlow, C. B.; Bishop, E. O.; Carey, P. R.; Guthrie, R. D. Carbohydr. Res. 1969, 9,
99.
2. Magnet, S.; Blanchard, J. S. Chem. Rev. 2005, 105, 477.
3. Ogle, J. M.; Brodersen, D. E.; Clemons, W. M., Jr.; Tarry, M. J.; Carter, A. P.;
Ramakrishnan, V. Science 2001, 292, 897.
4. Vicens, Q.; Westhof, E. Biopolymers 2003, 70, 42.
5. Jana, S.; Deb, J. K. Appl. Microbiol. Biotechnol. 2006, 70, 140.
6. Forge, A.; Schacht, J. Audiol. Neurootol. 2000, 5, 3.
7. Nagai, J.; Takano, M. Drug Metab. Pharmacokinet. 2004, 19, 159.
8. Antimicrobial Agent: The Aminoglycosides; Chambers, H. F., Sande, M. A., Eds.;
McGraw-Hill: New York, 1996.
9. Vital Brazil, O.; Prado-Franceschi, J. Arch. Int. Pharmacodyn. 1969, 179, 65.
10. Wright, J. M.; Collier, B. J. Pharmacol. Exp. Ther. 1977, 200, 576.
11. Albiero, L.; Bamonte, F.; Ongini, E.; Parravicini, L. Arch. Int. Pharmacodyn. 1978,
233, 343.
31. Rothlin, C. V.; Katz, E.; Verbitsky, M.; Vetter, D. E.; Heinemann, S. F.; Elgoyhen,
A. B. Neuropharmacology 2000, 39, 2525.
32. Talbot, P. A. J. Pharmacol. exp. Ther. 1987, 241, 686.
33. Hainrichson, M.; Yaniv, O.; Cherniavsky, M.; Nudelman, I.; Shallom-Shezifi, D.;
Yaron, S.; baasov, T. Antimicrob. Agents Chemother. 2007, 51, 3.
34. Hachler, H.; Santanam, P.; Kayser, F. H. Antimicrob. Agents Chemother. 1996, 40,
1254.
35. Masuda, N.; Sakagawa, E.; Ohya, S.; Gotoh, N.; Tsujimoto, H.; Nishino, T.
Antimicrob. Agents Chemother. 2000, 44, 2242.
12. Barnett, A.; Ackermann, E. Arch. Int. Pharmacodyn. 1969, 181, 109.
13. Paradelis, A. G.; Triantaphyllidis, C.; Markomichelakis, J. M.; Logaras, G.
Arzneimittel-Forschung 1977, 27, 141.
14. Li, J.; Chang, C. W. Anti-Infect. Agents Med. Chem. 2006, 5, 255.
15. Silva, J. G.; Carvalho, I. Curr. Med. Chem. 2007, 14, 1101.
16. Fujisawa, K.; Hoshiya, T.; Kawaguchi, H. J. Antibiot. (Tokyo) 1974, 27, 677.
17. Shitara, T.; Kobayashi, Y.; Tsuchiya, T.; Umezawa, S. Carbohydr. Res. 1992, 232,
273.
36. Emerson, J.; Rosenfeld, M.; McNamara, S.; Ramsey, B.; Gibson, R. L. Pediatr.
Pulmonol. 2002, 34, 91.
37. Amici, M.; Eusebi, F.; Miledi, R. Neuropharmacology 2005, 49, 627.
38. Alper, P. B.; Hung, S. C.; Wong, C. H. Tetrahedron Lett. 1996, 37, 6029.
39. Kahovec, J.; Harris, R. K. Chemicke Listy 2001, 95, 111.
40. Reed, L. J.; Muench, H. Am. J. Hyg. 1938, 27, 493.
41. Leatherbarrow, R. J. GraFit 5; Erithacus Software Ltd.: Horley, U.K., 2001.
42. NCCLS. National Committee for Clinical Laboratory Standards, Performance
standards for antimicrobial susceptibility testing. Fifth information supplement:
Approved Standard M100-S5; NCCLS: Villanova, PA, 1994.
18. Shitara, T.; Umemura, E.; Tsuchiya, T.; Matsuno, T. Carbohydr. Res. 1995, 276,
75.