Synthesis, antiribosomal and antibacterial activity of 4′-O- glycopyranosyl paromomycin aminoglycoside antibiotics

Article abstract of DOI:10.1039/c4md00119b

A series of 4′-O-glycopyranosyl paromomycin analogs and a 4′-O-(glucosyloxymethyl) analog were synthesized and evaluated for their ribosomal activity to determine the influence of the glycosyl moiety on drug activity and selectivity. Antibacterial activity against clinical strains of Escherichia coli and Staphylococcus aureus was also investigated. While all compounds were less active than paromomycin itself, differences in activity were seen between the gluco-, manno-, and galactopyranosyl series and between individual anomers. These differences in activity, which are discussed in terms of variations in affinity for the ribosomal decoding A site, may prove useful in the design of subsequent generations of improved aminoglycoside antibiotics with reduced toxicity. This journal is the Partner Organisations 2014.

Full text of DOI:10.1039/c4md00119b

MedChemComm
View Article Online
View Journal | View Issue
CONCISE ARTICLE
Synthesis, antiribosomal and antibacterial activity
of 40-O-glycopyranosyl paromomycin
aminoglycoside antibiotics†
Cite this: Med. Chem. Commun., 2014,
5, 1179
Weiwei Chen,a Takahiko Matsushita,a Dimitri Shcherbakov,b Heithem Boukari,b
c
b
a
*
*
*
Andrea Vasella, Erik C. Bottger and David Crich
¨
A series of 40-O-glycopyranosyl paromomycin analogs and a 40-O-(glucosyloxymethyl) analog were
synthesized and evaluated for their ribosomal activity to determine the influence of the glycosyl moiety
on drug activity and selectivity. Antibacterial activity against clinical strains of Escherichia coli and
Staphylococcus aureus was also investigated. While all compounds were less active than paromomycin
itself, differences in activity were seen between the gluco-, manno-, and galactopyranosyl series and
between individual anomers. These differences in activity, which are discussed in terms of variations in
affinity for the ribosomal decoding A site, may prove useful in the design of subsequent generations of
improved aminoglycoside antibiotics with reduced toxicity.
Received 14th March 2014
Accepted 5th April 2014
DOI: 10.1039/c4md00119b
www.rsc.org/medchemcomm
administration in the form of a single large dose daily, which
leaves AGA ototoxicity as the major area of concern in the future
development of novel AGAs.1–3,6 AGAs bind to the ribosomal
Introduction
The aminoglycosides (AGAs) are an established class of broad
spectrum antibiotics with proven efficacy against a number of
clinical pathogens.1–3 The evolution of widespread resistance
mechanisms coupled with concerns arising from nephrotoxicity
and ototoxicity together with the advent of combinatorial
chemistry and high throughput screening methods, however,
combined to end the classical phase of semi-synthetic AGA
discovery and development in the pharmaceutical industry in
the 1980 and 1990's.1–3 The continued emergence of drug-
resistant pathogens coupled with recent advances in the
understanding of the mode of action of the AGAs, and the
paucity of new antibiotic drug classes to emerge from
the screening efforts of the major pharmaceutical companies
suggests that the time is now ripe for the further exploration of
the AGAs.3,4 Indeed, the semi-synthetic 4,6-AGA plazomicin,
currently under investigation for the treatment of drug resistant
bacteria, is the rst demonstration of the renewed potential of
the AGAs.5 The molecular basis of resistance mechanisms to
AGAs is for the most part well-understood such that resistance
can be overcome by design, and AGA nephrotoxicity may be
circumvented in the clinic by careful monitoring and especially
decoding A-site and recent advances in molecular biology, most
notably in the engineering of eukaryotic ribosomal A-site
cassettes into bacterial ribosomes,7 have led to the hypothesis
that ototoxicity relates to limitations in drug target selectivity.
Thus, AGA binding to the decoding A-site of eukaryotic ribo-
somes,8–11 in particular the mitochondrial decoding A site, is the
root cause of ototoxicity.8–10 On this basis a series of functional
ribosomal assays using mutant hybrid ribosomes enables the
prediction of ototoxic activity alongside antibacterial activity in
collections of semisynthetic AGAs and other A site binders.7–10
Using this method we recently predicted that the natural atyp-
ical AGA apramycin 1 would have reduced ototoxicity and
demonstrated that this is indeed the case in the ex vivo murine
cochlear explant model and in the in vivo guinea pig auditory
brainstem response model.12 Working in the same manner we
next discovered that the target selectivity of the potent natural
pseudotetrasaccharide paromomycin 2, a member of the 4,5-
AGA class of molecules, can be increased by substitution at the
40- and 40,60-positions as exemplied by 3 and 4, respectively.13
Recognizing the structural similarity of apramycin 1 with 3 and
4, both in solution and in complex with fully constituted
bacterial ribosomes,13 we were spurred to prepare and investi-
gate the series of 40-O-glycopyranosyl paromomycin derivatives
on which we report here, whose substituent is positioned so as
to mimic the terminal apramycin 4-amino-4-deoxyglucopyr-
anosyl moiety. In previous work Baasov and coworkers have
synthesized a series of 500-O-glycosylated derivatives 5–8 of
neomycin B 9, which were active against various bacterial
strains.14 All of these compounds may be considered
aDepartment of Chemistry, Wayne State University, Detroit, MI 48202, USA. E-mail:
dcrich@chem.wayne.edu
b
¨
¨
¨
¨
Institut fur Medizinische Mikrobiologie, Universitat Zurich, 8006 Zurich, Switzerland.
E-mail: boettger@imm.uzh.ch
c
¨
¨
¨
Laboratorium fur Organische Chemie, ETH Zurich, 8093 Zurich, Switzerland. E-mail:
vasella@org.chem.ethz.ch
† Electronic
supplementary
information
(ESI)
available:
Complete
characterization data for all new compounds. Copies of the 1H and 13C NMR
spectra of 20a,b–23a,b and 32. See DOI: 10.1039/c4md00119b
This journal is © The Royal Society of Chemistry 2014
Med. Chem. Commun., 2014, 5, 1179–1187 | 1179

View Article Online
MedChemComm
Concise Article
regiosiomeric analogs of the pseudopentasaccharide livid- compound 2 for bacterial wild-type ribosomes. Even taking into
omycin A 10, a 30-deoxyparomomycin derivative characterized in account the approximately 20% increase in molecular weight
the early AGA literature as being less ototoxic than paromomy- due to the incorporation of the additional sugar units, these
cin itself.15
IC50 values are predictive of a substantial decrease of antibac-
terial activity. Closer inspection of the data suggests that the
a-anomers, with their axial glycosidic linkage to the newly
introduced hexopyranosyl residue, have lower IC50 values than
the corresponding equatorial (b-) glycosides, although the
difference is small and perhaps only signicant for compounds
26a,b and 27a,b (2.7- and 2.4-fold, respectively). The ability of
27a to inhibit translation in bacterial ribosomes, which
approaches that of paromomycin and apramycin, may be in
part attributed to the incorporation of an additional amino
group and thus to increased electrostatic attraction between the
positively charged AGA and the negatively charged ribo-
some.26–28 However, the difference in IC50 with its anomer 27b
suggests that other, more specic, interactions also play a role.
X-ray crystallographic studies of paromomycin in complex
with the bacterial ribosomal 30S subunit29 and with truncated
oligonucleotide models of it's decoding A-site30 indicate a
hydrogen bond between O40 of the drug and the phosphate
backbone of the ipped out base A1493. However, 40-deoxy-
paromomycin experiences only a minimal loss of antibacterial
activity (MIC values)16,31 suggesting that the 40-OH/A1493 H
bond is only a minor component of the overall binding energy.
Suppression of the 40-OH/A1493 H bond by derivatization of
the 40-OH group is thus not the major cause of the loss of activity
seen in the 40-O-substituted paromomycin derivatives.
Presumably, the substitution pattern at the 40-position affects
the crystallographically revealed29,30 CH–p32,33 interactions of
the b-face of paromomycin ring with G1491 at the bottom of the
binding pocket, and the interplay between the AGA ring 1 and
its hydration shell. Alteration of the electron density at C40 by
removal of O40 or by substitution at O40 may modulate the CH–p
interaction between H40 and G1491 by a combination of elec-
tronic, hydrophobic, and steric effects. For example, alkylation
of O40, as in derivative 3, will have the effect of lowering the
electron density at O40. Inclusion of O40 in either an acetal
linkage (compounds 4, 32) or a glycosidic bond (compounds 24–
27) will further reduce the electron density at that locus beyond
simple alkylation. The various substitutions at O40 may result in
an unfavourable steric interaction with G1491 and so further
reduce the CH–p interaction with H40, indeed, this may be the
origin of the observed differences in antiribosomal activity
between the a- and b-anomers of the various glycosides studied.
X-ray crystallographic studies of apramycin (1),12 of the 40-
phenylpropyl paromomycin ether (3)13 and of the paromomycin
40,60-O-phenylpropylidene acetal (4)13 all reveal a common mode
of binding with paromomycin to the ribosomal RNA decoding A
site. The aminoglucosyl residue of apramycin and the 40(60)-
Results and discussion
Chemical synthesis
Coupling of the known selectively protected paromomycin
derivative16 11 with a series of four known thioglycosides 12–
1517–20 with low temperature activation by the diphenyl sulf-
oxide/triuoromethanesulfonic anhydride (Tf2O) couple21 in
dichloromethane in the presence of the hindered non-nucleo-
philic base tri-tert-butylpyrimidine (TTBP)22 at temperatures
between ꢀ72 ꢁC and room temperature gave the a- and
b-glycosides 16–19 (Scheme 1) with the yields and anomeric
ratios indicated in Table 1. No attempt was made to improve
stereoselectivity in these coupling reactions, but the anomeric
mixtures were separated by standard chromatographic tech-
niques and subject to deprotection by treatment with trime-
thylphosphine and subsequent hydrogenolysis leading to the
isolation of the 40-O-paromomycin derivatives 24a,b–27a,b
(Scheme 1) with the yields indicated in Table 1.
In order to mimic more closely the trehalose-type linkage in
apramycin, the known trichloroacetimidate 2823 was coupled24
with phenylthiomethanol25 to give the phenylthiomethyl a-D-
glucoside 29 (Scheme 2). Subsequent N-iodosuccinimide (NIS)
and triuoromethanesulfonic acid (TfOH) mediated coupling24
of 29 with the acceptor 11 gave the derivative 30, which, on
subjection to the Staudinger protocol followed by hydro-
genolysis in the standard manner afforded the 40-O-a-D-gluco-
pyranosyloxymethyl derivative 32 of paromomycin (Scheme 2).
Antiribosomal activity
The ability of the synthetic AGAs to prevent translation in cell- substituents in 3 and 4 also share a common location extending
free ribosome assays was assessed using the parental drug beyond the main binding pocket into the bulk solvent between
paromomycin (2), apramycin (1), and the 40-paromomycin A1492 and G1491. The glucosyl (24) and mannosyl (25) deriva-
derivatives 3 and 4 as comparators. The data presented in Table tives exhibit very similar antiribosomal activity for a given
2 indicate that, with the single exception of the 4-amino-a-D- anomer, while the galactosyl isomers (26) are marginally less
glucopyranosyl derivative 27a, all compounds display an IC50 active. These results are consistent with the notion that gluco-
value in mg mlꢀ1 ꢂten or more times greater than the parental and mannopyranosides may better t into and disrupt the
1180 | Med. Chem. Commun., 2014, 5, 1179–1187
This journal is © The Royal Society of Chemistry 2014

View Article Online
Concise Article
MedChemComm
Scheme 1 Synthesis of 40-O-glycosyl paromomycin derivatives.
structure of organized water less than galactopyranosides;34–37 (Table 2) is representative of the human mitochondrial A site,
thus, the galactosyl derivatives 26a,b are the most disruptive of while the G1491A mutant (Table 2) mimics one of the two
the solvent shell in the exposed region between A1492 and changes in the human cytosolic decoding A site, with the
G1491.
In order to determine the inuence of individual bases on ined with the aid of the A1408G mutant (Table 2).
the ribosome–drug interaction, the 40-substituted AGA deriva-
The data (Table 2) reveals that all compounds exhibit a
importance of the second change independently being exam-
tives were assayed for their ability to inhibit mutant ribosomes signicant loss of antiribosomal activity on going from the wild
in each of which a single base in the bacterial decoding A site type to the G1491C mutant, as is the case for the parental
was replaced by that found in human mitochondrial rRNA or in compound paromomycin 2, for apramycin 1, and for the paro-
human cytosolic rRNA. In Fig. 1 the decoding A site of the momycin derivatives 3 and 4. With the exception of the 40-O-(4-
bacterial ribosome is aligned with those of human cytosolic and amino-4-deoxyglucosyl) derivatives 27a- and b- which show
human mitochondrial rRNA. Thus, the G1491C mutant comparable affinity, the a-anomers inhibit the G1491C mutant
Table 1 Synthesis of 40-O-glycosyl paromomycin derivatives
Entry
Donor
12
Glycosylation yield, ratio
16, 42%, a : b ¼ 3.2 : 1
17, 51%, a : b ¼ 1 : 3.6
18, 60%, a : b ¼ 6.4 : 1
19, 50%, a : b ¼ 4.5 : 1
Staudinger yield
Hydrogenolysis yield
1
2
3
4
5
6
7
8
20a, 64%
20b, 55%
21a, 91%
21b, 66%
22a, 81%
22b, 78%
23a, 64%
23b, 48%
24a, 39%
24b, 87%
25a, 33%
25b, 25%
26a, 62%
26b, 39%
27a, 64%
27b, 29%
13
14
15
This journal is © The Royal Society of Chemistry 2014
Med. Chem. Commun., 2014, 5, 1179–1187 | 1181

View Article Online
MedChemComm
Concise Article
Scheme 2 Synthesis of a 40-O-glucosyloxymethyl paromomycin derivative.
to a greater extent than the corresponding b-anomers. A very mutants with the noncanonical C1409$C1491 and C1409$A1491
similar pattern is seen for inhibition of the G1491A mutant base pairs, respectively. The greater loss of activity seen for all
ribosomal activity, namely an overall decrease in activity and for compounds on going from the wild type to the G1491C mutant
the mannose and galactose series (25 and 26) preferential than to the G1491A mutant is indicative of greater disruption to
binding of the a-anomer (the two anomers of the glucose and 4- binding on replacement of the purine G by the pyrimidine C
amino-4-deoxyglucose series exhibiting little or no difference). than by the purine A. This may be the result of greater distortion
Overall this pattern suggests a related binding mode for all of the binding pocket geometry, of a larger perturbation of the
compounds studied with the bacterial drug binding pocket, CH–p interaction with the b-face of the AGA ring 1, or of
with the cognate C1409]G1491 base pair at the base of the changing interactions with the varying hydration shells around
ribosomal decoding A site, and in the G1491C and G1491A the different bases.
Table 2 Compound interaction with polymorphic residues in the drug binding pocket (IC50, mg mlꢀ1 a
)
Bacterial A site
Wild type
Cmpd
40-O-Substitution type
G1491C
G1491A
A1408G
0.26 ꢃ 0.04
2
1
3
4
24a
24b
25a
25b
26a
26b
27a
27b
32
None (paromomycin)
Na (apramycin)
0.03 ꢃ 0.02
0.06 ꢃ 0.04
0.20 ꢃ 0.07
0.10 ꢃ 0.03
0.32 ꢃ 0.05
0.37 ꢃ 0.06
0.29 ꢃ 0.03
0.46 ꢃ 0.02
0.55 ꢃ 0.08
1.52 ꢃ 0.30
0.12 ꢃ 0.04
0.29 ꢃ 0.07
0.31 ꢃ 0.01
10.42 ꢃ 2.86
31.21 ꢃ 7.27
0.57 ꢃ 0.09
5.00 ꢃ 1.36
128.89 ꢃ 41.49
14.52 ꢃ 8.60
3.81 ꢃ 2.44
21.84 ꢃ 6.87
9.32 ꢃ 1.36
5.10 ꢃ 0.78
27.91 ꢃ 2.04
12.01ꢃ 1.17
36.46 ꢃ 2.23
7.06 ꢃ 0.52
2.39 ꢃ 0.21
17.16 ꢃ 1.89
Phenylpropyl ether
210.73 ꢃ 39.98
124.07 ꢃ 0.03
109.01 ꢃ 33.61
294.83 ꢃ 21.70
81.32 ꢃ 50.80
297.39 ꢃ 21.16
118.67 ꢃ 8.58
623.16 ꢃ 46.00
32.79 ꢃ 2.05
51.26 ꢃ 15.72
43.01 ꢃ 11.31
27.08 ꢃ 10.41
15.50 ꢃ 2.72
8.79 ꢃ 0.12
40,60-O-Phenylpropylidene acetal
a-D-Glucopyranosyl
b-D-Glucopyranosyl
a-D-Mannopyranosyl
b-D-Mannopyranosyl
a-D-Galactopyranosyl
b-D-Galactopyranosyl
4-Amino-a-D-glucopyranosyl
4-Amino-b-D-glucopyranosyl
a-D-Glucopyranosyloxymethyl
51.48 ꢃ 17.39
18.68 ꢃ 2.37
78.05 ꢃ 2.35
6.07 ꢃ 0.06
32.03 ꢃ 2.21
4.87 ꢃ 0.46
200.30 ꢃ 0.42
33.45 ꢃ 2.21
a
All measurements were made in triplicate.
1182 | Med. Chem. Commun., 2014, 5, 1179–1187
This journal is © The Royal Society of Chemistry 2014

View Article Online
Concise Article
MedChemComm
Fig. 1 Decoding A sites of bacterial and eukaryotic ribosomes. (The AGA binding pocket is boxed. The prokaryotic numbering scheme is
illustrated for the AGA binding pocket. Changes in the binding pocket from the bacterial ribosome are coloured green.)
The A1408G mutation causes a reduction in ribosomal gt conformation.29 A comparable pseudobase pair is formed
activity for all compounds (Table 2). In the wild type A1408 with apramycin,12 albeit the gg conformation of the C50–C60
forms a pseudobase interaction with the ring oxygen and the 60- bond is enforced by the rigid bicyclic nature of its ring 1.
OH of paromomycin ring 1 with its hydroxymethyl group in the Modelling studies38 suggest that an alternative approximately
Fig. 2 Decoding A sites of the bacterial and mutant hybrid ribosomes. (The AGA binding pocket is boxed. Eukaryotic A sites replacing bacterial
h44 are shaded. The prokaryotic numbering scheme is illustrated for the AGA binding pocket. Changes in the binding pocket from the bacterial
ribosome are coloured green. The A1555G mutant conferring hypersusceptibility to aminoglycoside ototoxicity is coloured red.)
This journal is © The Royal Society of Chemistry 2014
Med. Chem. Commun., 2014, 5, 1179–1187 | 1183

View Article Online
MedChemComm
Table 3 Compound interaction with hybrid ribosomes (IC50, mg mlꢀ1 a
Eukaryotic A site
Concise Article
)
Bacterial A site
wild type
Mitochondrial
wild type
Mitochondrial
mutant A1555G
Cmpd
Substitution type
Cyt 14
RRL
2
1
3
4
24a
24b
25a
25b
26a
26b
27a
27b
32
None (paromomycin)
Na (apramycin)
Phenylpropyl
0.03 ꢃ 0.02
0.06 ꢃ 0.04
0.20 ꢃ 0.07
0.10 ꢃ 0.03
0.32 ꢃ 0.05
0.37 ꢃ 0.06
0.29 ꢃ 0.03
0.46 ꢃ 0.02
0.55 ꢃ 0.08
1.52 ꢃ 0.30
0.12 ꢃ 0.04
0.29 ꢃ 0.07
0.31 ꢃ 0.01
50.54 ꢃ 13.04
58.03 ꢃ 18.80
49.85 ꢃ 15.39
330.74 ꢃ 115.30
1.86 ꢃ 0.36
5.83 ꢃ 2.27
26.06 ꢃ 7.08
105.68 ꢃ 39.23
126.88 ꢃ 54.36
13.98 ꢃ 2.56
83.64 ꢃ 24.45
18.05 ꢃ 4.68
85.70 ꢃ 27.23
14.95 ꢃ 0.11
209.22 ꢃ 52.33
3.62 ꢃ 1.22
10.39 ꢃ 2.57
49.17 ꢃ 26.01
268.57 ꢃ 54.97
183.86 ꢃ 84.90
46.26 ꢃ 4.65
9.78 ꢃ 4.47
23.77 ꢃ 7.48
nd
40,60-O-Phenyl-propylidene
a-D-Glucosyl
nd
30.64 ꢃ 1.40
60.75 ꢃ 1.21
36.82 ꢃ 0.9
47.64 ꢃ 7.7
23.75 ꢃ 1.20
104.57 ꢃ 0.54
19.59 ꢃ 4.54
32.41 ꢃ 5.11
43.14 ꢃ 5.40
b-D-Glucosyl
a-D-Mannosyl
b-D-Mannosyl
a-D-Galactosyl
59.12 ꢃ 6.57
2.19 ꢃ 0.07
117.53 ꢃ 29.68
51.72 ꢃ 16.80
90.35 ꢃ 26.69
35.63 ꢃ 2.23
54.60 ꢃ 8.3
2.23 ꢃ 0.33
b-D-Galactosyl
141.35 ꢃ 3.08
0.34 ꢃ 0.01
286.76 ꢃ 32.27
18.95 ꢃ 5.37
4-Amino-a-D-glucosyl
4-Amino-b-D-glucosyl
a-D-Glucosyl-oxymethyl
7.62 ꢃ 3.30
17.38 ꢃ 5.12
96.03 ꢃ 14.50
21.43 ꢃ 8.53
24.11 ꢃ 2.31
130.61 ꢃ 13.31
a
All measurements were made in triplicate.
isosteric pseudobase interaction is possible following the A somewhat greater loss of activity is seen in the case of the
A1408G exchange at least in the case of paromomycin. The mitochondrial mutant A1555G in which the non-Watson–Crick
lower antiribosomal activity of paromomycin 2 and of its C1494$A1555 base pair immediately below the C1493$C1491
40-phenylpropyl ether 3 against the A1408G mutant indicates noncanonical pair at the base of the decoding A site is replaced
that the alternative pseudobase interaction of ring 1 with G1408 by the Watson–Crick C1494]G1555 base pair (Fig. 2). Thus, in
is weaker than that with A1408. The greatly reduced activity of the deafness allele the two sequential noncanonical pairs at the
apramycin toward the A1408G mutant ribosome suggests that base of the mitochondrial A site are replaced by a single non-
the G1408-ring 1 pseudobase interaction is less adaptable than Watson–Crick pair. Presumably, the more exible mitochon-
the A1408-ring 1 interaction and does not accommodate the gg drial A site is better able to accommodate substitution at the 40-
conformation of the C50–C60 bond. In the case of the 40,60-O- position of the AGA than is its A1555G deafness allele. As with
alkylidene paromomycin derivative 4 a single hydrogen bond the bacterial wild type and the G1491C single point mutation
remains possible between A1408 N6 or G1408 N1 and O50, both the mitochondrial hybrid ribosome and its A1555G deaf-
despite the enforced tg conformation of the C50–C60 bond, ness allele were inhibited to a greater extent by the a- than by
without distortion of the overall geometry of the pseudobase the b-anomers of the 40-O-glycosides.
pair interaction.13 Glycosylation at O40 presumably affects the
The Cyt 14 hybrid ribosome carrying the A site of human
A1408-ring 1 pseudobase interaction through the involvement cytosolic rRNA incorporates both of the A1491A and A1408G
of the 60-OH group in inter-residue hydrogen bonds with the single point mutations. Not surprisingly therefore all
40-O-glycosyl ring, altering the conformational preference of the compounds studied have lower antiribosomal activity toward
C50–C60 bond, and reducing the availability of the 60-OH group the Cyt 14 hybrid than toward the bacterial wild type. SAR data
for hydrogen bonding to the ribosomal base. Any such modu- obtained with eukaryotic ribosomes from rabbit reticulocytes
lation of this pseudobase interaction will impact the weaker were comparable to that from the Cyt 14 hybrid ribosome. The
G1408-ring 1 interaction to a greater extent leading to the loss of pattern of greater inhibition by the a-anomers of 40-O-glycosides
activity seen with the A1408G mutant.
than by the b-isomers, which was partially obscured for the
Having examined the inuence of single point mutations in G1491A and A1408G single point mutants, is restored for the
the bacterial A site on the drug ribosome interaction we turned Cyt 14 hybrid and rabbit reticulocyte ribosomes.
to the use of hybrid mutant ribosomes in which complete
The difference in antiribosomal activity between the indi-
eukaryotic A site cassettes are inserted into the bacterial ribo- vidual anomers, while typically favouring the a-anomer, is small
some.7 The three constructs employed in this manner repre- for bacterial, cytosolic, and rabbit reticulocyte ribosomes but
sentative of the mitochondrial wild type, the mitochondrial noticeably larger for the mitochondrial wild type and deafness
mutant A1555G associated with enhanced incidence of drug- alleles (Table 3). The large difference in activity of the anomers
induced deafness (deafness allele),39 and the cytosolic wild type seen with the mitochondrial ribosomes results in the b-anom-
(Fig. 2, Table 3), were complemented by the use of native ers displaying greater bacterial/mitochondrial selectivity. While
eukaryotic cytosolic ribosomes from rabbit reticulocytes.
still signicant the activity difference between pairs of anomers
Compared to bacterial ribosome, all compounds prepared in for the deafness allele is smaller, most likely reecting the less
this study showed reduced antiribosomal activity toward the accommodating nature of the less exible binding site of the
hybrid ribosome containing the mitochondrial A site (Table 3). deafness allele.
1184 | Med. Chem. Commun., 2014, 5, 1179–1187
This journal is © The Royal Society of Chemistry 2014

View Article Online
Concise Article
MedChemComm
Table 4 Minimal inhibitory concentrations (MIC, mg mlꢀ1) of clinical isolates
Strain
Staphylococcus aureus (MRSA)
Escherichia coli
Compound
AG 038
AG 039
AG 042
AG 044
AG 006
AG 001
AG 055
AG 003
2
1
3
4
24a
24b
25a
25b
26a
26b
27a
27b
32
4
8
>256
8
8
8
32
$128
$128
$128
128
$128
8–16
>64
$128
>256
8
4
4–8
32
>128
>128
$128
64
>128
8–16
>64
128
4–8
16
4–8
16
16–32
128
32–64
$128
64
>128
4
32
2–4
4
8
16–32
16–32
32
16
64
32
16–32
16
16
32–64
128
>128
>128
>128
>128
>128
32
8
8
16
32
128
>128
>128
>128
>128
>128
32
8–16
8–16
8–16
32
2–4
4–8
32
128
64
$128
64
$128
4
128
>128
$128
>128
>128
>128
16–32
>64
128
2–4
16
32
64–128
>64
$128
>64
$128
64–128
64
$128
The small differences in antiribosomal activity (Tables 2 and
3) observed between the 40-O-gluco-, manno-, and galacto-
pyranosides and between the individual anomers in each series
recalls the modest inuence of substituents in SAR studies of
40,60-O-alkylidene and 40-O-arylalkyl derivatives of paromomycin
related to compounds 3 and 4.13 Together this suggests the
absence of a specic interaction between the 40-(60-)-O-substit-
uent and the ribosomal decoding A site, but rather the existence
of one or more less specic interactions, none of which is
dominant.
Conclusion
A series of 40-O-D-hexopyranosyl analogues of the AGA paro-
momycin have been prepared by chemical synthesis and their
individual a- and b-anomers assayed for their ability to inhibit
ribosomal protein synthesis by bacterial ribosomes and their
mutants designed to mimic the various eukaryotic drug
binding pockets. All compounds were also assayed for anti-
bacterial activity against clinical isolates of Staphylococcus
aureus and Escherichia coli. With the exception of the 4-amino-
4-deoxy glucosides, all glycosides showed less antibacterial
activity than paromomycin, apramycin, and two 40-O-deriva-
tives of paromomycin. Among the glycosides prepared the a-
anomers were generally more active than their b-isomers and
the glucopyranosides more active than the manno- and
galactopyranosides.
Antibacterial activity
All glycosides prepared were assessed for their antibacterial
activity by the determination of MIC values against clinical
isolates of Staphylococcus aureus and Escherichia coli (Table 4).
The parental drug paromomycin 2, apramycin 1, and the 40-O-
(phenylpropyl) and 40,60-O-alkylidene derivatives 3 and 4 of
paromomycin were used as comparators (Table 4). With the
exception of two strains of S. aureus which were resistant to
paromomycin, all compounds other than the 4-amino-4-deoxy
Experimental part
General procedure for glycosylation
glucoside 27a were less active than the comparators reecting A dichloromethane solution of glycosyl donor (0.47 mmol,
the IC50 data for inhibition of the wild type bacterial ribosome 1.3 eq.), diphenyl sulfoxide (0.55 mmol, 1.5 eq.), 2,4,6-tri-
˚
(Table 2). Consistent with the IC50 data for the wild type tert-butylpyrimidine (1.05 mmol, 2.9 eq.) and activated 3 A
bacterial ribosome (Table 2) the a-anomers of all 40-O-glycosides molecular sieves in dichloromethane (3 ml) was stirred at
are more active than their b-counterparts. No advantage is room temperature for 3 h before cooling to the indicated
gained by the inclusion of a spacer group between the AGA 40- temperature (see ESI†). Triuoromethanesulfonic anhydride
position and the glycoside. 40-O-(4-Amino-4-deoxy)-a-D-gluco- (0.55 mmol, 1.5 eq.) was added slowly to the solution. Aer 1
pyranosyl paromomycin 27a has comparable activity to the h at the same temperature a solution of acceptor 11 (ref. 16)
40,60-O-(2-phenylpropylidene) derivative 4 of paromomycin, (0.36 mmol, 1 eq.) in dichloromethane (1.5 ml) was added
which we attribute to the presence of the additional amino dropwise to the reaction mixture. The resulting solution was
group in the sugar moiety and which is consistent with the wild stirred at the same temperature for 2 h before the reaction
type antiribosomal activity. Comparators 1, 3, and 4, the 40-O-a- was quenched by the addition of saturated aqueous
glucoside 24a, and both anomers of the 4-amino-4-deoxy NaHCO3. The reaction mixture was diluted with EtOAc and
glucoside 27a and 27b, unlike paromomycin itself 2, are active the organic phase was washed with saturated aqueous
against the S. aureus strains AG 039 and AG 042. Presumably, AG NaHCO3 three times. The organic layer was dried (Na2SO4),
039 and AG 042 achieve resistance through the operation of a 40- concentrated and puried by column chromatography over
or a 30-aminoglycoside modifying enzyme.5,40,41
silica gel.
This journal is © The Royal Society of Chemistry 2014
Med. Chem. Commun., 2014, 5, 1179–1187 | 1185

View Article Online
MedChemComm
Concise Article
2 B. Becker and M. A. Cooper, ACS Chem. Biol., 2013, 8, 105–
115.
General procedure for the Staudinger reduction of azides to
amines16
3 J. L. Houghton, K. D. Green, W. Chen and S. Garneau-
Tsodikova, ChemBioChem, 2010, 11, 880–902.
4 S. B. Vakulenko and S. Mobashery, Clin. Microbiol. Rev., 2003,
16, 430–450.
5 J. B. Aggen, E. S. Armstrong, A. A. Goldblum, P. Dozzo,
M. S. Linsell, M. J. Gliedt, D. J. Hildebrandt, L. A. Feeney,
A. Kubo, R. D. Matias, S. Lopez, M. Gomez,
K. B. Wlasichuk, R. Diokno, G. H. Miller and H. E. Moser,
Antimicrob. Agents Chemother., 2010, 54, 4636–4642.
6 S. Magnet and J. S. Blanchard, Chem. Rev., 2005, 205, 477–
497.
The substrate (0.06 mmol, 1 eq.) in THF (5 ml) was treated at
room temperature with 0.1
M aqueous NaOH (0.3 ml,
0.03 mmol, 5 eq.) and 1 M trimethylphosphine in THF
(0.6 mmol, 10 eq.). The reaction mixture was stirred for 2 h at
50 ꢁC, and then cooled to room temperature and neutralized
with 1 M aqueous AcOH to pH 7 before concentration. The
resulting slurry was subjected to silica gel chromatography,
eluting rst with MeOH (100 ml), followed by 0.5% aqueous
NH4OH in MeOH (150 ml) to give the product.
General procedure for hydrogenolysis
7 S. N. Hobbie, S. K. Kalapala, S. Akshay, C. Bruell, S. Schmidt,
The substrate (25 mg) was dissolved in a mixture of methanol
(1 ml), deionized water H2O (2 ml), and glacial acetic acid
(0.1 ml). A catalytic amount of Pd(OH)2 on carbon (20 wt %) was
added, and the reaction mixture was stirred at room tempera-
ture under 1 atm of hydrogen (balloon) overnight. Aer
completion, the reaction mixture was ltered over Celite® and
ltrate was neutralized by the addition of Amberlite-IRA400 to
pH 7 and ltered. The ltrate was concentrated to dryness and
dissolved in 0.002 M aqueous AcOH (pH 4, 2 ml) before it was
charged to a Sephadex column (CM Sephadex C-25). The
Sephadex column was ushed sequentially with deionized
water H2O (50 ml), 0.5% aqueous NH4OH (20 ml), and 1.5%
NH4OH (40 ml). The fractions containing the hydrogenolysis
product were combined and evaporated to give a sticky white
solid, which was dissolved in 0.00002 M acetic acid (pH 5, 1 ml).
The resulting solution was frozen by immersion in a dry ice
acetone bath, and then the water was removed by lyophilization
to give the product in the form of its acetate salt.
¨
S. Dabow, A. Vasella, P. Sander and E. C. Bottger, Nucleic
Acids Res., 2007, 35, 6086–6093.
8 S. N. Hobbie, S. Akshay, S. K. Kalapala, C. Bruell,
D. Shcherbakov and E. C. Bottger, Proc. Natl. Acad. Sci.
¨
U. S. A., 2008, 105, 20888–20893.
9 R. Akbergenov, D. Shcherbakov, T. Matt, S. Duscha,
M. Meyer, D. Perez-Fernandez, R. Pathak, S. Harish,
I. Kudyba, S. R. Dubbaka, S. Silva, M. Ruiz Ruiz, S. Salian,
¨
A. Vasella and E. C. Bottger, in Ribosomes: Structure,
Function, and Dynamics, ed. M. V. Rodnina, W.
Wintermeyer and R. Green, Springer-Verlag, Vienna, 2011,
pp. 249–261.
¨
10 E. C. Bottger and J. Schacht, Hear. Res., 2013, 303, 12–19.
11 S. P. Francis, J. Katz, K. D. Fanning, K. A. Harris,
B. D. Nicholas, M. Lacy, J. Pagana, P. F. Agris and
J.-B. Shin, J. Neurosci., 2013, 33, 3079–3093.
12 T. Matt, C. L. Ng, K. Lang, S.-H. Sha, R. Akbergenov,
D. Shcherbakov, M. Meyer, S. Duscha, J. Xie,
S. R. Dubbaka, D. Perez-Fernandez, A. Vasella,
Bacterial strains. Clinical isolates of E. coli and S. aureus were
obtained from the Diagnostic Department, Institute of Medical
Microbiology, University of Zurich. MIC values were determined
by broth microdilution assays.
Recombinant microorganisms. The construction of these
strains derived from single rRNA allelic M. smegmatis DrrnB, has
been described previously.7,38
¨
V. Ramakrishnan, J. Schacht and E. C. Bottger, Proc. Natl.
Acad. Sci. U. S. A., 2012, 109, 10984–10989.
13 D. Perez-Fernandez, D. Shcherbakov, T. Matt, N. C. Leong,
I. Kudyba, S. Duscha, H. Boukari, R. Patak, S. R. Dubbaka,
K. Lang, M. Meyer, R. Akbergenov, P. Freihofer, S. Vaddi,
P. Thommes, V. Ramakrishnan, A. Vasella and
Cell-free translation assays. Rabbit reticulocyte lysate
(Promega), S-30 extracts and puried ribosomes were used for
cell-free translation assays as described previously.12 Firey
luciferase mRNA was used as reporter to monitor translation
activity. Luminescence was measured using a luminometer
Flx800 (Bio-Tek Instruments).
¨
E. C. Bottger, Nat. Commun., 2014, 5, 3112.
14 M. Fridman, V. Belakhov, S. Yaron and T. Baasov, Org. Lett.,
2003, 5, 3575–3578.
15 H. Tamura, Y. Fukakusa, Y. Ogawa and J. Ouchi, Oyo Yakuri,
1972, 6, 845–855.
16 R. Pathak, D. Perez-Fernandez, R. Nandurdikar,
¨
S. K. Kalapala, E. C. Bottger and A. Vasella, Helv. Chim.
Acta, 2008, 91, 1533–1552.
17 J. Dinkelaar, A. R. de Jong, R. van Meer, M. Somers,
Acknowledgements
We gratefully acknowledge support from the University of Zur-
ich and Wayne State University.
´
G. Lodder, H. S. Overklee, J. D. C. Codee and G. A. van
der Marel, J. Org. Chem., 2009, 74, 4982–4991.
18 D. Crich and S. Sun, Tetrahedron, 1998, 54, 8321–8348.
19 C. M. Taylor, C. A. Weir and C. G. Jørgensen, Aust. J. Chem.,
2002, 55, 135–140.
References
1 E. S. Armstrong, C. F. Kostrub, R. T. Cass, H. E. Moser, 20 W. A. Greenberg, E. S. Priestley, P. S. Sears, P. B. Alper,
A. W. Serio and G. H. Miller, in Antibiotic Discovery and
Development, ed. T. J. Dougherty and M. J. Pucci, Springer
Science+Business Media, 2012, pp. 229–269.
C. Rosenbohm, M. Hendrix, S.-C. Hung and C.-H. Wong,
J. Am. Chem. Soc., 1999, 121, 6527–6541.
1186 | Med. Chem. Commun., 2014, 5, 1179–1187
This journal is © The Royal Society of Chemistry 2014

View Article Online
Concise Article
MedChemComm
´
´
˜
´
21 J. D. C. Codee, L. J. van den Bos, R. E. J. N. Litjens, 32 J. L. Asensio, A. Arda, F. J. Canada and J. Jimenez-Barbero,
H. S. Overklee, C. A. A. van Boeckel, J. H. van Boom and
G. A. van der Marel, Tetrahedron, 2004, 60, 1057–1064.
22 D. Crich, M. Smith, Q. Yao and J. Picione, Synthesis, 2001,
323–326.
Acc. Chem. Res., 2013, 46, 946–954.
´
´
´
33 T. Vacas, F. Corzana, G. Jimenez-Oses, C. Gonzalez,
´
A. M. Gomez, A. Bastida, J. Revuelta and J. L. Asensio,
J. Am. Chem. Soc., 2010, 132, 12074–12090.
¨
¨
23 H. Waldmann, G. Bohm, U. Schmid and H. Rottele, Angew. 34 S. A. Galema, J. B. F. N. Engberts, H. Hoiland and
Chem., Int. Ed., 1994, 33, 1944–1946.
G. M. Ferrland, J. Phys. Chem., 1993, 97, 6885–
24 D. Crich and F. Yang, Angew. Chem., Int. Ed., 2009, 48, 8896–8899.
6889.
25 L. L. Gundersen and T. Benneche, Acta Chem. Scand., 1991, 35 S. A. Galema, M. J. Blandamer and J. B. F. N. Engberts, J. Org.
45, 975–977. Chem., 1992, 57, 1995–2001.
26 R. Benveniste and J. Davies, Antimicrob. Agents Chemother., 36 S. A. Galema and H. Hoiland, J. Phys. Chem., 1991, 95, 5321–
1973, 4, 402–409. 5326.
27 M. Kaul, C. M. Barbieri, J. E. Kerrigan and D. S. Pilch, J. Mol. 37 C. J. Capicciotti, M. Leclere, F. A. Perras, D. L. Bryce,
Biol., 2003, 326, 1373–1387.
28 S. Salian, T. Matt, R. Akbergenov, S. Harish, M. Meyer,
H. Paulin, J. Harden, Y. Liu and R. N. Ben, Chem. Sci.,
2012, 3, 1408–1416.
¨
S. Duscha, D. Shcherbakov, B. B. Bernet, A. Vasella, 38 P. Pster, S. Hobbie, Q. Vicens, E. C. Bottger and E. Westhof,
¨
E. Westhof and E. C. Bottger, Antimicrob. Agents
ChemBioChem, 2003, 4, 1078–1088.
Chemother., 2012, 56, 6104–6108.
39 S. N. Hobbie, C. M. Bruell, S. Akshay, S. K. Kalapala,
¨
29 A. P. Carter, W. M. Clemons, D. E. Brodersen, R. J. Morgan-
Warren, B. T. Wimberly and V. Ramakrishnan, Nature, 2000,
407, 340–348.
D. Shcherbakov and E. C. Bottger, Proc. Natl. Acad. Sci.
U. S. A., 2008, 105, 3244–3249.
40 T. Ida, R. Okamoto, C. Shimauchi, T. Okubo, A. Kuga and
M. Inoue, J. Clin. Microbiol., 2001, 39, 3115–3121.
30 Q. Vicens and E. Westhof, Structure, 2001, 9, 647–658.
31 C. Battistini, G. Cassinelli, G. Franceschi, F. Arcamone and 41 R.-B. Yan, M. Yuan, Y. Wu, X. You and X.-S. Ye, Bioorg. Med.
R. Mazzoleni, J. Antibiot., 1981, 34, 240–242.
Chem., 2011, 19, 30–40.
This journal is © The Royal Society of Chemistry 2014
Med. Chem. Commun., 2014, 5, 1179–1187 | 1187