Exploring the use of conformationally locked aminoglycosides as a new strategy to overcome bacterial resistance

Published on Web 12/14/2005

Exploring the Use of Conformationally Locked

Aminoglycosides as a New Strategy to Overcome Bacterial

Resistance

Agatha Bastida,^†Ana Hidalgo,^†Jose Luis Chiara,^†Mario Torrado,^†

Francisco Corzana,^†Jose Manuel Pe´rez-Can˜adillas,^‡Patrick Groves,^§

Eduardo Garcia-Junceda,^†Carlos Gonzalez,^‡Jesu´s Jimenez-Barbero,^§and

Juan Luis Asensio*^,†

Contribution from the Instituto de Qu´ımica Orga´nica General (CSIC), 28006 Madrid, Spain,

Instituto de Qu´ımica F´ısica Rocasolano (CSIC), Madrid 28006, Spain, and Centro de

InVestigaciones Biolo´gicas (CSIC), Madrid 28040, Spain

Received June 30, 2005; E-mail: iqoa110@iqog.csic.es

Abstract: The emergence of bacterial resistance to the major classes of antibiotics has become a serious

problem over recent years. For aminoglycosides, the major biochemical mechanism for bacterial resistance

is the enzymatic modification of the drug. Interestingly, in several cases, the oligosaccharide conformation

recognized by the ribosomic RNA and the enzymes responsible for the antibiotic inactivation is remarkably

different. This observation suggests a possible structure-based chemical strategy to overcome bacterial

resistance; in principle, it should be possible to design a conformationally locked oligosaccharide that still

retains antibiotic activity but that is not susceptible to enzymatic inactivation. To explore the scope and

limitations of this strategy, we have synthesized several aminoglycoside derivatives locked in the ribosome-

bound “bioactive” conformation. The effect of the structural preorganization on RNA binding, together with

its influence on the aminoglycoside inactivation by several enzymes involved in bacterial resistance, has

been studied. Our results indicate that the conformational constraint has a modest effect on their interaction

with ribosomal RNA. In contrast, it may display a large impact on their enzymatic inactivation. Thus, the

work presented herein provides a key example of how the conformational differences exhibited by these

ligands within the binding pockets of the ribosome and of those enzymes involved in bacterial resistance

can, in favorable cases, be exploited for designing new antibiotic derivatives with improved activity in resistant

strains.

Introduction

detailed knowledge about the interactions that stabilize the drug/

RNA complexes.

Aminoglycoside antibiotics are highly potent, wide-spectrum

bactericidals.^1-4These drugs bind to the decoding region

aminoacyl-tRNA site (A-site) inducing codon misreading and

inhibiting translocation, eventually resulting in cell death.^1-4

From a chemical point of view, aminoglycosides are poly-

cationic oligosaccharides carrying several hydroxyl and amino

functional groups. Most include one or more amino sugars

attached to a 2-deoxystreptamine ring (ring II in Scheme 1).

The structures of several aminoglycosides, such as gentamycin,

tobramycin, and paromomycin, bound to the target ribosomal

A-site RNA have been determined recently by both X-ray and

NMR methods.^5-7These structures have provided an extremely

Several years after the introduction of aminoglycosides in

human antibacterial chemotherapy, resistant organisms began

to appear. Indeed, the emergence of bacterial resistance to the

major classes of antibiotics has become a serious problem over

recent years.^8,9Acquired resistance to aminoglycoside antibiotics

can occur via three different mechanisms: mutation of the

ribosomal target, reduced permeability for the antibiotics, and

enzymatic modification of the drugs, thus leading to inactivation.

The most prevalent source of clinically relevant resistance is

conferred by the third mechanism, the enzymatic inactivation

of the drugs.^8,9Enzymes involved in bacterial defense against

aminoglycoside antibiotics can be broadly classified as N-

acetyltransferases (AACs), O-adenyl transferases (ANTs), and

^†Instituto de Qu´ımica Orga´nica General (CSIC), Madrid 28006, Spain.

^‡Instituto de Qu´ımica F´ısica Rocasolano (CSIC), Madrid 28006, Spain.

^§Centro de Investigaciones Biolo´gicas (CSIC), Madrid 28040, Spain.

(1) Walter, F.; Vicens, Q.; Westhof, E. Curr. Opin. Chem. Biol. 1999, 3, 694-

704.

(2) Hermann, T. Angew. Chem., Int. Ed. 2000, 39, 1890-1905.

(3) Schroeder, R.; Waldsich, C.; Wank, H. EMBO J. 2000, 19, 1, 1-9.

(4) Sucheck, J. S.; Wong, C. H. Curr. Opin. Chem. Biol. 2000, 4, 678-686.

(5) Carter, A. P.; Clemons, W. M.; Brodersen, D. E.; Morgan-Warren, R. J.;

Wimberly, B. T.; Ramakrishnan, V. Nature 2000, 407, 340-348.

(6) (a) Vicens, Q.; Westhof, E. Structure 2001, 9, 647-658. (b) Vicens, Q.;

Westhof, E. Chem. Biol. 2002, 9, 747-755.

(7) (a) Yoshizawa, S.; Fourmy, D.; Puglisi, J. D. EMBO J. 1998, 17, 22, 6437-

6448. (b) Fourmy, D.; Recht, M. I.; Blanchard, A. C.; Puglisi, J. D. Science

1996, 274, 1367-1371. (c) Lynch, S. R.; Gonzalez, R. L., Jr.; Puglisi, J.

D. Structure 2003, 11, 43-53.

(8) Magnet, S.; Blanchard, J. Chem. ReV. 2005, 105, 477-497.

(9) Smith, C. A.; Baker, E. N. Curr. Drug Targets 2002, 2, 143-160.

9

100

J. AM. CHEM. SOC. 2006, 128, 100-116

Conformationally Locked Aminoglycosides

A R T I C L E S

Scheme 1. Schematic Representation of the Conformational Differences (Highlighted with a Dotted Box) Exhibited by Aminoglycosides in

the Binding Pockets of RNA and Some of the Enzymes Involved in Bacterial Resistance^a

^aThe glucose, 2-deoxy-streptamine, and ribose units are numbered as I, II, and III, respectively. The values of the glycosidic torsion angles Φ(H1-

C1-Ox-Cx) and Ψ(Hx-Cx-O1-C1) for each glycosidic linkage in the bound state, together with the pdb codes of the complexes, are indicated. The

position of the aminoglycoside modified by the enzyme is marked by an arrow. A schematic representation of the GlcR(1-4)2-deoxy-streptamine fragment

present in neomycin (X ) NH₂, Y ) NH₂, R ) 2,6-diamino-2,6-dideoxy-L-Idoâ(1-3)Rib, R′ ) H), paromomycin (X ) OH, Y ) NH₂, R ) 2,6-diamino-

2,6-dideoxy-^L-Idoâ(1-3)Rib, R′ ) H), ribostamycin (X ) NH₂, Y ) NH₂, R ) Rib, R′ ) H), and kanamycin (X ) NH₂, Y ) OH, R ) H, R′ ) 3-amino-

3-deoxy-R-Glc) is shown in the top left-hand corner.

O-phosphotransferases (APHs). It is evident that a detailed

knowledge of the structures of these enzymes and their

interactions with the drugs should provide a framework that

will facilitate the rational design either of novel aminoglycosides,

not susceptible to modification, or of inhibitors of these

enzymes, which could be administered in tandem with existing

aminoglycosides. Indeed, the search for new derivatives not

susceptible to modification by bacterial defense proteins con-

stitutes an active field of research.^10-13

in the enzyme binding site significantly differs from that

observed in the corresponding aminoglycoside/A-site RNA

complex (see Scheme 1). For example, Staphylococcus aureus

ANT(4′) selects a high energy conformation of the ligand,

kanamycin.¹⁵In particular, the GlcR(1-4)-2-deoxy streptamine

fragment (numbered as units I and II in Scheme 1), present in

all aminoglycosides with different patterns of amination/

hydroxylation at the glucose unit (I), adopts an anti-Ψ geometry

(Φ/Ψ, -21/151). In contrast, for tobramycin and paromomycin,

the analogous glycosidic linkage is defined by a syn-Ψ confor-

mation (Φ/Ψ ≈ -40/-30) in the RNA-bound state. The “syn-

Ψ” geometry also constitutes the most populated minimum for

the free antibiotics, according to NMR data.¹⁶The drug confor-

mation selected by ANT(4′) represents a high-energy minimum

that must be stabilized by specific aminoglycoside-protein

interactions. In a similar way, the X-ray structure of Mycobac-

terium tuberculosis AAC(2′) in complex with ribostamycin

shows that the Ribâ(1-5)-2-deoxy streptamine fragment (num-

The 3D structures of several aminoglycoside-enzyme com-

plexes have been described recently by X-ray methods.^9,14

Interestingly, in some cases, the 3D structure of the antibiotic

(10) (a) Kondo, S.; Iinuma, K.; Yamamoto, H.; Maeda, K.; Umezawa, H. J.

Antibiot. 1975, 26, 412-417. (b) Kondo, S.; Hotta, K. J. Infect. Chemother.

1999, 5, 1.

(11) Inoue, M.; Nonoyama, M.; Okamoto, R.; Ida, T. Drugs Exp. Clin. Res.

1994, 20, 233-239.

(12) Fujimura, S.; Tokue, Y.; Takahashi, H.; Nukiwa, T.; Hisamichi, K.; Mikami,

T.; Watanabe, A. J. Antimicrob. Chemother. 1998, 41, 495-499.

(13) (a) Roestamadji, J.; Graspas, I.; Mobashery, S. J. Am Chem. Soc. 1995,

117, 11060-11069. (b) McKay, G. A.; Roestamadji, J.; Mobashery, S.;

Wright, G. D. Antimicrob. Agents Chemother. 1996, 40, 2648. (c) Haddad,

J.; Vakulenko, S.; Mobashery, S. J. Am. Chem. Soc. 1999, 121, 11922-

11923. (d) Grapsas, I.; Lerner, S. A.; Mobashery, S. Arch. Pharm. 2001,

334, 295.

(15) Pedersen, L. C.; Benning, M. M.; Holden, H. M. Biochemistry 1995, 34,

13305-13311.

(16) Asensio, J. L.; Hidalgo, A.; Cuesta, I.; Gonza´lez, C.; Can˜ada, J.; Vicent,

C.; Chiara, J. L.; Cuevas, G.; Jime´nez-Barbero, J. Chem. Eur. J. 2002, 8,

22, 5228-5240.

(14) Vicens, Q.; Westhof, E. Biopolymers 2003, 70, 42-57.

9

J. AM. CHEM. SOC. VOL. 128, NO. 1, 2006 101

A R T I C L E S

Bastida et al.

Chart 1. Schematic Representation of the Proposed Structure-Based Chemical Strategy to Overcome Bacterial Resistance

bered as units III and II in Scheme 1) also adopts a high energy

“anti-Ψ” conformation (Φ/Ψ ) 28/-173).¹⁷This geometry is

significantly different to the bioactiVe “syn-ψ” orientation

adopted by the same glycosidic bond in the paromomycin/A-

site complex.^5,6An additional key example of conformational

differences of aminoglycosides between the RNA- and protein-

bound states is provided by the recently reported structure of

the Salmonella enteritidis AAC(6′) in complex with ribosta-

mycin.¹⁸In this case, the Ribâ(1-5)-2-deoxy streptamine

fragment (units III and II in Scheme 1) of the oligosaccharide

presents a rather unusual nonexo-anomeric conformation (char-

acterized by a Φ angle of -5°). In addition, the furanose ring

(numbered as III in Scheme 1) is bound in a C2-endo

conformation, in contrast with the C3-endo geometry selected

by the ribosomal RNA. A similar furanose puckering is also

recognized by the Enterococcus faecalis APH(3′) enzyme

according to the X-ray structure of its complex with neomycin-

B.¹⁹

The design of our conformationally constrained aminogly-

cosides is based on the structural information available for the

RNA-paromomycin complex.^5,6The structure of neomycin-B

(1) along with the numbering employed for the different sugar

units along this manuscript (I, II, III, and IV for the glucose,

2-deoxy-streptamine, ribose, and idose units, respectively) is

shown in Scheme 2. It is known that positions O5_IIIand N2_I

are hydrogen bonded within the ribosome bound state. Thus,

in our analogues, this interaction has been replaced by either a

direct covalent bond, as in 2, a two-methylene bridge, as in 3,

or by a salt bridge, as in 4 (see Scheme 2).

It should be pointed out that OH5_IIIis involved in RNA

recognition^5,6and might also have an influence on the enzymatic

inactivation of the antibiotics. Since cyclization of neomycin-B

to give 2 or 3 required the removal of this OH group, the deoxy-

derivatives 5 and 6 were also synthesized as controls, to isolate

the effect of rigidification.

In the first part of the manuscript we present a detailed

description of the synthesis of the proposed neomycin-B

derivatives. The next section deals with the structural NMR

analysis of the obtained aminoglycosides. This study, essential

in order to establish any structure-activity relationship, shows

that, as expected, the obtained antibiotics exhibit a significantly

reduced flexibility with respect to natural neomycin-B and, in

addition, closely resemble its bioactive ribosome-bound con-

formation. In the next two sections the effect of the conforma-

tional restriction on the RNA binding properties and the

biological activity of these analogues is determined. Finally the

susceptibility of the obtained derivatives to enzymatic modifica-

tion has been tested employing three different enzymes involved

in bacterial resistance. This final analysis suggests that confor-

mational restriction might constitute a useful approach to prevent

aminoglycoside inactivation by bacterial defense proteins.

The different 3D-shapes adopted by aminoglycosides in the

RNA- and enzyme-bound states suggest a possible structure-

based chemical strategy to obtain antibiotics with a better

activity against resistant bacteria. Assuming that, in these cases,

some degree of conformational distortion of the substrates is

required for enzymatic activity, it should be possible to design

a conformationally locked oligosaccharide that still retains

antibiotic activity but that is not susceptible to enzymatic

20

inactivation (Chart 1).

To test this hypothesis, we have designed and synthesized

several neomycin-B analogues locked in the RNA-bound

(bioactiVe) structure of the antibiotic.²¹Thus, we have studied

the effect that structural preorganization has on the strength of

the aminoglycoside-RNA interaction. In a second step, we have

determined the susceptibility of the conformationally locked

derivatives to enzymatic inactivation, employing the enzymes

Staphylococcus aureus ANT(4′), Mycobacterium tuberculosis

AAC(2′), and Enterococcus faecalis APH(3′) as model systems.

Results and Discussion

Synthesis of the Proposed Neomycin-B Derivatives.

A. Synthesis of Neomycin-B Conformationally Constrained

Analogue 2. Compound 2 was prepared following the simple

four-step procedure shown in Scheme 3. Thus, the hexa-N-

(benzyloxycarbonyl) derivative (7), readily prepared from

neomycin-B (1),²²was treated with 2,4,6-triisopropylbenzene-

sulfonyl chloride in pyridine²³to yield the primary sulfonate

(17) Vetting, M. W.; Hegde, S. S.; Javid-Majd, F.; Blanchard, J. S.; Roderick,

S. L. Nat. Struct. Biol. 2002, 9, 653-658.

(18) Vetting, M. W.; Magnet, S.; Nieves, E.; Roderick, S. L.; Blanchard, J. S.

Chem. Biol. 2004, 11, 565-573.

(19) Fong, D. H.; Berghuis, A. M. EMBO J. 2002, 21, 2323-2331.

(20) A key example of how the enzymatic activity can be modulated by small

conformational preferences of the substrates was reported recently: Galan,

M. C.; Venot, A. P.; Boons, G. J. Biochemistry 2003, 42, 8522-8529.

(21) A preliminary communication on some of the aspects described above has

been recently published: Asensio, J. L.; Hidalgo, A.; Bastida, A.; Torrado,

M.; Corzana, F.; Garc´ıa-Junceda, E.; Can˜ada, J.; Chiara, J. L.; Jime´nez-

Barbero, J. J. Am. Chem. Soc. 2005, 127, 8278-8279.

(22) Kumar, V.; Remers, W. A. J. Org. Chem. 1981, 46, 4298-4300.

(23) Michael, K.; Wang, H.; Tor, Y. Bioorg. Med. Chem. 1999, 7, 1361-1372.

9

102 J. AM. CHEM. SOC. VOL. 128, NO. 1, 2006

Conformationally Locked Aminoglycosides

A R T I C L E S

Scheme 2. Schematic Representation of the Target Conformationally Locked Aminoglycosides (2-4) and the Deoxy Derivatives (5,6)^a

^aThe numbering employed for the different sugar units is indicated for neomycin-B (compound 1).

(8), regioselectively. Deprotection of the amino groups by

palladium-catalyzed hydrogenolysis in MeOH/trifluoroacetic

acid gave crude sulfonate (9) that smoothly cyclized in a highly

regioselective way by mild heating in water at pH 7.0 to give

2 in 56% yield.

triphenylphosphorane in acetonitrile gave the corresponding R,â-

unsaturated ester 13, which was reduced with NaBH₄in the

presence of catalytic LiCl to give alcohol 14 in 89% yield.

Initially, we tried the selective activation of the primary alcohol

in 14, by treatment with 2,4,6-triisopropylbenzenesulfonyl

chloride in pyridine. However, the low regioselectivity obtained

in the reaction forced us to use a three-step strategy similar to

that described above. Thus, regioselective protection of the

primary hydroxyl group in 14 as a triisopropylsilyl ether, yielded

derivative 15. Acetylation of the secondary hydroxyl groups

followed by desilylation gave derivative 16, which was trans-

formed into the corresponding primary sulfonate by treatment

with p-toluenesulfonyl chloride in pyridine. Finally, clean

deacetylation of this derivative was best performed by mild

KCN-catalyzed methanolysis,²⁴to give 17. The carbamate

groups were removed by palladium-catalyzed hydrogenolysis

B. Synthesis of Neomycin-B Conformationally Con-

strained Analogue 3. The synthesis of compound 3 is shown

in Scheme 4. The key step involves a two-carbon atom

elongation of ring III side chain through a Wittig-type reaction

on aldehyde 12. The following four-step strategy was used to

obtain this aldehyde in good yield. First, regioselective protec-

tion of the primary hydroxyl group of 7, by treatment with

triisopropylsilyl trifluoromethanesulfonate and 2,6-lutidine, fol-

lowed by acetylation of the secondary hydroxyl groups gave

compound 10, which was desilylated by treatment with HF‚

pyridine to give 11. Finally, oxidation of 11 using the Dess-

Martin periodinane gave the required aldehyde 12 in 57% overall

yield from 7. Treatment of 12 with (methoxycarbonylmethylen)-

(24) Mori, K.; Tominaga, M.; Takigawa, T.; Matsui, M. Synthesis 1973, 790-

791.

9

J. AM. CHEM. SOC. VOL. 128, NO. 1, 2006 103

A R T I C L E S

Scheme 3^a

Bastida et al.

^a(a) CbzCl, Na₂CO₃, MeOH/H₂O 3:1, 0 °C, 3 h, 92%; (b) 2,4,6-triisopropylbenzenesulfonyl chloride (TIBSCl), Py, rt, 72 h, 44%; (c) H₂(1 atm), Pd/C,

MeOH, trifluoroacetic acid, rt, 8 h, 90%; (d) H₂O, pH 7.0, 60 °C, 5 d, 56%. In 7 and 8, R₁) 2,6-diamino-2,6-di-N-(benzyloxycarbonyl)-2,6-dideoxy-â-

L-idose.

in MeOH/TFA. The resulting crude sulfonate was cyclized at

40 °C and pH 7.0 in H₂O to give 3 in 62% yield.

by including all the experimental NOE-derived distances and J

data as time-average restraints with the AMBER 5.0²⁶program,

as described.¹⁶

The cyclization proceeded with an astonishing high regio-

selectivity considering the number of amino groups present in

this neomycin-B derivative (see Figure S1). In fact, compound

3 is the only macrocyclic derivative isolated from the reaction

mixture. Close inspection of the corresponding molecular model

shows that the two amino groups present in unit IV could also

participate in the nucleophilic attack to the side chain of III.

However, preliminary molecular mechanics calculations suggest

that the cyclization involving NH₂at position 2 of unit IV would

proceed through a high energy transition state. The preferred

cyclization through the NH₂at position 2 of ring I instead of

that at position 6 of unit IV can be explained considering the

different pK_avalues of these groups (7.4 vs 8.5, respec-

tively). In fact, under the employed experimental conditions (pH

7.0), the later amino group is fundamentally in its protonated

state.

The obtained trajectories were able to quantitatively reproduce

all the NMR experimental data. In all cases, the analyses of the

J couplings for units I and II unambiguously show that the ⁴C₁

conformer is the very major one in water (>98%). In contrast,

for ring IV the ¹C₄conformer with three axial and two equatorial

substituents is the only one experimentally detected. A similar

behavior has been reported for the natural antibiotic,¹⁶and

therefore the cyclization of neomycin-B to give 2-3 does not

produce significant distortions in any of the three pyranose rings.

According to the experimental data, derivative 2 is extremely

rigid. Both J and NOE values (see Table 1 and Figure 1) are

consistent with a pure “gt” (ω_O-C4-C5-O5≈ 60°) orientation

for the side chain of unit III and a unique puckering for its

furanose ring. In fact, only minor fluctuations around the I/II

and III/II glycosidic linkages are compatible with this macro-

cyclic structure, according to the MD-tar simulations.

Compound 6 was readily obtained from 17 by palladium-

catalyzed hydrogenolysis of the Cbz protecting groups and

subsequent reduction of the primary sulfonate under the same

conditions using a longer reaction time.

A summary of the experimental data employed in the NMR

analysis of 3 is shown in Figure 2. In this case, a large number

of lower bound constraints had to be included in the MD-tar

simulations (together with the NOE derived distances and J

values) to properly define its structure in solution. As expected,

the experimental data indicates the presence of a larger degree

of flexibility. Thus, NOEs and J values are consistent with the

Finally derivatives 4 and 5 were easily obtained from 11 as

described in the Supporting Information.

NMR Analysis of the Conformationally Locked Amino-

glycosides 2-4. Selective 1D-NOE experiments were carried

3

out at 313 K and pH 5.0 for analogues 2-4. In addition, J

values were measured for all the sugar units (Table 1). To get

an experimentally derived ensemble, MD-tar (molecular dynam-

ics with time-averaged restraints) simulations²⁵were carried out

(25) Pearlman, D. A. J. Biomol. NMR 1994, 4, 1-16.

(26) Pearlman, D. A.; Case, D. A.; Caldwell, J. W.; Ross, W. S.; Cheatham, T.

E.; DeBolt, S.; Ferguson, D.; Siebal, G.; Kollman, P. Comput. Phys.

Commun. 1995, 91, 1-41.

9

104 J. AM. CHEM. SOC. VOL. 128, NO. 1, 2006

Conformationally Locked Aminoglycosides

A R T I C L E S

Scheme 4^a

^a(a) Triisopropylsilyl trifluoromethanesulfonate (TIPSOTf), 2,6-lutidine, THF, rt, 1 h, 80%; (b) Ac₂O, Py, rt, 5 h, ∼100%; (c) HF‚Py, THF, rt, 5 h, 90%;

(d) Dess-Martin periodinane, CH₂Cl₂, rt, 2h, 79%; (e) Ph₃PdCHCO₂Me, CH₃CN, rt, 3h, 82%; (f) NaBH₄, cat. LiCl, EtOH/THF, rt, 12 h, 89%; (g) TIPSOTf,

2,6-lutidine, THF, 0 °C, 1 h, 60%; (h) Ac₂O, Py, rt, 5h, ∼100%; (i) HF‚Py, THF, rt, 5 h, 79%; (j) p-toluenesulfonyl chloride (TsCl), Py, cat. DMAP, rt, 3

h, 79%; (k) cat. KCN, MeOH/THF 2:1, rt, 10 h, 58%; (l) H₂(1 atm), Pd/C, MeOH, trifluoroacetic acid, rt, 8 h, ∼88%; (m) H₂O, pH 7.0, 40 °C, 10 d, 62%;

(n) H₂(1 atm), Pd/C, MeOH, trifluoroacetic acid, rt, 48 h, ∼60%; In 14, 15, and 17 R₁) 2,6-diamino-2,6-di-N-(benzyloxycarbonyl)-2,6-dideoxy-â-L-idose.

In 10-13 and 16 R₂) 3,4-di-O-acetyl-2,6-diamino-2,6-di-N-(benzyloxycarbonyl)-2,6-dideoxy-â-L-idose.

presence of several orientations for the propylidene bridge that

links units I and III. For the side chain of ring III, a major

(>80%) “gt” (ω_O-C4-C5-C6≈ 60°) and a minor (<20%) “gg”

(ω_O-C4-C5-C6≈ -60°) orientation are detected. Moreover, the

furanose ring presents two different puckerings characterized

by phase angles of 25° (“North”) and 160° (“South”) (for phase

angles the Altoona and Sundaralingan²⁷convention is employed)

and relative populations of 40:60, respectively. According to

the MD-tar trajectories, the macrocyclic structure present in 3

is compatible with relatively large fluctuations around the I/II

and III/II glycosidic linkages (see Figure 3).

Finally, our combined NMR/MD approach indicates that

derivative 4 also exhibits a significant degree of flexibility

(Figure 4). As previously observed for 3, ring III presents two

different puckerings characterized by phase angles around 5°

and 170° and relative populations of 30:70, respectively. The

polar interaction between NH2_Iand the carboxylate present in

III (see Figure 4) seems to be compatible with the presence of

relatively wide fluctuations around the I/II and III/II glycosidic

linkages.

The obtained ensembles of conformers for 2-4 that quanti-

tatively reproduce the experimental data are shown in Figure

5. These ensembles were compared to that obtained for the

natural antibiotic 1 in the free (experimentally derived from the

NMR data)¹⁶and RNA-bound (derived from unrestrained MD

simulations of the neomycin-B/A-site complex) states.¹⁶As

expected, analogues 2-4 are more rigid than 1 in the free state.

The RMSD for heavy atom superimposition for the ensemble

corresponding to 1 in the free state is 2.1 Å changing to 0.4 Å

for the bound ligand. The corresponding values for 3 and 4 are

1.6 Å. The conformational restriction imposed by covalently

bonding C5_IIIto N2_Iin 2 is especially severe with an RMSD

value of 0.9 Å.

Interestingly, the locked aminoglycosides 2-4 closely re-

semble the bioactive conformation of the natural antibiotic. The

most representative minimum energy conformations present in

solution for 2 and 3, superimposed on the X-ray structure of

paromomycin in complex with A-site RNA, are shown in Figure

6. For 2, deviations in Φ/Ψ values with respect to the

paromomycin bound structure are less than 11° and 21° for the

I/II and III/II linkages, respectively. The corresponding devia-

(27) Altona, C.; Sundaralingham, M. J. Am. Chem. Soc. 1972, 94, 8205-8208.

9

J. AM. CHEM. SOC. VOL. 128, NO. 1, 2006 105

A R T I C L E S

Bastida et al.

Table 1. Experimental and Theoretical Distances (above) and

Coupling Constant Values (below) Obtained for Compounds 2, 3,

and 4 at pH 5.0^a

tionally constrained derivative; they are all more rigid than the

natural antibiotic neomycin-B and closely resemble its bioactive,

RNA-bound conformation.

compound 2

compound 3

compound 4

Fluorescence-Based Analysis of the Interaction between

Aminoglycosides 1-6 and the Ribosomal Decoding Site. As

a next step, the binding properties of compounds 1-4 to

ribosomal RNA were analyzed. It is known that, for natural

paromomycin, OH5_IIIis involved in RNA recognition.^5,6Cy-

clization of neomycin-B to give 2 and 3 requires the removal

of this OH group. To isolate the effect of rigidification on the

aminoglycoside-RNA interaction, the binding properties of the

deoxyderivatives 5 and 6 were also tested. The fluorescence-

based approach, developed independently by Pilch²⁸and Her-

mann,²⁹was employed. This methodology makes use of a

modified 27-mer RNA fragment (A-site(2AP)) including the

A-site sequence and the fluorescent base analogue 2-amino-

purine (2AP) replacing A1492 (Figure 7a). It is well established

that major groove binding of aminoglycosides to the A-site

induces a significant conformational change in the RNA so

that the adenine residues at positions 1492 and 1493 are

displaced from the helical stack toward the minor groove.^5-7

As the fluorescence of 2AP is modulated by hydrophobic

stacking interactions, the employed RNA fragment (shown in

Figure 7a) provides an adequate means for monitoring and

quantifying the destacking of A1492 that accompanies complex

formation.

MD-tar

d(Å)

expmt

(5 ns)

expmt

(80 ns)

expmt

(5 ns)

H1_I-H4_II

H1_I-H5_II

H1_I-H1_III

H1_I-H2_III

H1_I-H3_III

H1_I-H5_III

2.4

2.3

3.8

4.5

3.7

3.3

5.1

3.3

2.5

3.3

3.9

3.4

2.5

3.2

4.0

3.2

>3.3

R

S

2.4

2.3

2.4

H1_I-H6_IIIR/S

>3.3/>3.3 3.3/4.6

>3.3/>3.3 3.3/3.6

H1_I-H7_IIIR/S

H2_I-H5_III

H2_I-H7_III

R

<2.4

2.6

H3_I-H7_IIIR/S

H1_III-H5_II

>3.3/>3.3 3.5/3.3

2.6

>3.5

3.5

2.6

4.5

3.6

3.7

2.2

>3.3

3.1

2.3

4.4

3.7

3.0

2.5

H1_III-H4_II

H1_III-H6_II

3.6

3.5

3.3

H1_III-H4_III

H2_III-H7_IIIR/S

H2_III-H6_II

3.6

>3.3/>3.3 3.3/4.9

>3.3

4.2

3.2

3.3

H2_III-H6_IIIR/S

H2_III-H5_III

>3.3/>3.3 4.9/3.4

R

<2.4

2.2

H3_III-H7_IIIR/S

H3_III-H6_IIIR/S

H4_III-H6_IIIR/S

H1_IV-H2_III

>3.3/>3.3 3.3/4.6

>3.3/>3.3 4.3/3.2

>3.3/>3.3 3.2/3.8

>4

2.5

3.5

4

2.6

3.5

H1_IV-H3_III

H1_IV-H4_III

2.5

>3.3

2.3

4.0

2.4

3.8

2.5

3.9

It has been shown that paromomycin binding to the A-site(2AP)

induces a marked increase in the fluorescence intensity at 369

nm. These changes were monitored as a function of the [drug]/

[RNA] ratio in titration experiments, to derive Kd values. Some

typical binding curves are shown in Figure 7b. The affinity

constants measured for compounds 1-6 are shown in Figure

8. According to our titration experiments, natural neomycin-B

binds to the A-site with a Kd value of 0.07 µM in 100 mM

NaCl, 10 mM phosphate and pH 7.6. This result is in agreement

with previously reported values obtained by other methods under

comparable experimental conditions.³⁰

MD-tar

(80 ns)

MD-tar

(5 ns)

J (Hz)

expmt

(5 ns)

expmt

H1_III-H2_III

H2_III-H3_III

H3_III-H4_III

H4_III-H5_III

H7_IIIR-H6_III

H7_IIIS-H6_III

3.9

4.9

<1.0

11.6

4.0

4.5

4.2

1.1

11.0

4.7

3.5

4.1

3.9

4.0

4.4

4.0

4.1

4.3

3.5

<3.5

9.5

3.0

8.5

4.5

4.1

8.8

3.4

7.9

7.2

R

S

R

S

R

7.0

^aMolecular dynamic simulations with time-averaged restraints (MD-

tar) for compounds 2 and 4 were carried out using explicit solvent,

counterions, periodic boundary conditions (PBC), and Ewald sums for the

treatment of electrostatic interactions. For compound 3, in vacuo 80 ns MD-

tar trajectories were collected (see the Materials and Methods). The

stereochemistry of the protons involved in the distance constraints is

indicated when appropriate.

The maximum fluorescence increases observed for the

neomycin-B derivatives (2, 3, 5, and 6) are similar to that

detected for the parent compound (1), suggesting the presence

of similar binding modes. The corresponding Kd values are

shown in Figure 8a. It can be observed that deoxy derivatives

5 and 6 present a 10-fold reduction in affinity with respect to

the natural compound 1. This result is not unexpected as OH5_III

is involved in a hydrogen bond with G1491 in the paromomycin/

A-site complex. Therefore, the ∆G contribution of this hydro-

gen bond can be estimated as ∼1.3 kcal/mol. The observed

affinities for the macrocyclic derivatives 2 and 3 are in the same

range of those observed for the corresponding open deoxy-

analogues 5 and 6, respectively. According to these data, under

the employed experimental conditions, the ∆G contribu-

tion resulting from the conformational restriction is almost

negligible. ^31,32

tions for 3 are larger (15° and 45°). However, the global shape

of the molecule is very similar to that observed for paromomycin

in the RNA-bound state, as shown by the low RMSD values

for heavy atom superimposition (see Figure 6).

Previous studies have shown that neomycin-B in solution is

characterized by a remarkable flexibility, with different con-

formations around the three glycosidic linkages and two different

conformational populations for the unit III furanose ring (with

phase angles ∼20° and ∼160° and populations of 60-70% and

30-40%, respectively) in fast exchange.¹⁶On the other hand,

MD simulations of neomycin-B complexed to ribosomal A-site

RNA suggest that its flexibility is severely reduced upon binding

and only one of the furanose conformations is selected by the

RNA receptor. Interestingly, in analogue 2, the unit III furanose

is locked in a very similar conformation to that present in the

A-site/paromomycin complex, with an almost identical orienta-

tion of the key polar groups, O2 and O3 (Figure 6).

(28) Kaul, M.; Barbieri, C. M.; Pilch, D. S. J. Am. Chem. Soc. 2004, 126, 3447-

3453.

(29) Shandrick, S.; Zhao, Q.; Han, Q.; Ayida, B. K.; Takahashi, M.; Winters,

G. C.; Simonsen, K. B.; Vourloumis, D.; Hermann, T. Angew. Chem., Int.

Ed. 2004, 43, 3177-3182.

(30) (a) Wang, Y.; Hamasaki, K.; Rando, R. R. Biochemistry 1997, 36, 768-

779. (b) Ryu, D. H.; Litovchick, A.; Rando, R. R. Biochemistry 2002, 41,

10499-10509. (c) Fourmy, D.; Recht, M. I.; Puglisi, J. D. J. Mol. Biol.

1998, 277, 347-362.

In conclusion, according to our structural analysis, derivatives

2-4 satisfy the basic requirements for an optimal conforma-

9

106 J. AM. CHEM. SOC. VOL. 128, NO. 1, 2006

Conformationally Locked Aminoglycosides

A R T I C L E S

Figure 1. (a) Selective 1D NOE experiments with the 1D-DPFGSE NOE pulse sequence, corresponding to the inversion of H5_IIIand H1_Iin compound 2.

Key NOEs probing the spatial proximity of units I and III are highlighted in the upper part of the figure. The numbering employed for the different sugar

units is also indicated. (b) MD-tar Φ/Ψ distributions obtained for the I/II (upper panel) and III/II (lower panel) glycosidic linkages. The values corresponding

to the paromomycin-RNA bound form are indicated with a yellow dot.

Conformational freedom has been shown to play a key role

in the recognition of ligands by receptors.³³Several attempts

have been made to exploit the decrease of conformational

entropy of a particular ligand by either locking or biasing

conformations toward the bioactive one. In some cases, the

constrained derivatives presented significantly larger affinities

for the receptor than the natural ligand.³⁴In contrast, reported

results for carbohydrate-protein interactions range from modest

∆G benefits (<0.5 kcal/mol)³⁵to small penalties^35,36(<0.3 kcal/

mol). In these cases, the small effect of the conformational

constraint on the global ∆G has been attributed to an entropy-

enthalpy compensation phenomenona.³⁶Our results indicate that

despite the different chemical nature of both the ligand and the

receptor, aminoglycoside-RNA interactions exhibit a similar

behavior to those previously observed in the recognition of

neutral carbohydrates by proteins. Especially striking is the lack

(31) While this manuscript was in preparation, Tor and co-workers reported

the interaction of both the locked derivative 2 and the corresponding

paromomycin analogue with the same labeled A-site RNA fragment.

Although the buffer conditions employed for the fluorescence measurements

differ from ours, the reported decrease in affinity for the locked derivatives

is 22-14-fold with respect to the parent aminoglycosides, in agreement

with our results; Blount, K. F.; Zhao, F.; Hermann, T.; Tor, Y. J. Am.

Chem. Soc. 2005, 127 (27), 9818-9829.

(32) The 3D structure of derivative 2 in complex with the A-site, reported by

Hermann and col while this manuscript was under revision, shows that

rings III and IV present a slightly different orientation in the RNA major

groove from that observed for natural Neomycin-B. As a result, the number

of ligand-RNA contacts established by these two units is reduced. Zhao,

F.; Zhao, Q.; Blount, K. F.; Han, Q.; Tor, Y.; Hermann, T. Angew. Chem.,

Int. Ed. 2005, 44, 5329-5334.

(33) (a) Williams, D. H.; Searle, M. S.; Mackay, J. P.; Gerhard, U.; Maplestone,

R. A. Proc. Natl. Acad. Sci. U.S.A.. 1993, 90, 1172-1178. (b) Calderone,

C. T.; Williams, D. H. J. Am. Chem. Soc. 2001, 123, 6262-6267.

(34) (a) Choi, Y.; Moon, H. R.; Yoshimura, Y.; Marquez, V. E. Nucleoside

Nucleotide Nucleic Acids 2003, 22, 547-557. (b) Kim, H. S.; Ohno, M.;

Xu, B.; Kim, H. O.; Choi, Y.; Ji, X. D.; Maddileti, S.; Marquez, V. E.;

Harden, T. K.; Jacobson, K. A. J. Med. Chem. 2003, 46, 4974-4987.

(35) Navarre, N.; Amiot, N.; van Oijen, A. H.; Imberty, A.; Poveda, A.; Jime´nez-

Barbero, J.; Nutley, M. A.; Boons, G. J. Chem. Eur. J. 1999, 5, 2281-

2294.

(36) Bundle, D. R.; Alibe´s, R.; Nilar, S.; Otter, A.; Warwas, M.; Zhang, P. J.

Am. Chem. Soc. 1998, 120, 5317-5318.

9

J. AM. CHEM. SOC. VOL. 128, NO. 1, 2006 107

A R T I C L E S

Bastida et al.

Figure 2. (a) Selective 1D NOE experiments with the 1D-DPFGSE NOE pulse sequence, corresponding to the inversion of the anomeric protons in compound

3. (b) Upper panel: Schematic representation of some of the experimental information employed in the structural analysis of 3. The upper left panel shows

the observed NOEs as red arrows and those not observed as blue arrows. The upper right panel shows the structurally relevant experimental J couplings

measured for 3. The ring IV is omitted for simplicity. Lower panel: some representative NOE buildup curves. NOEs corresponding to known fixed distances,

employed as references, are labeled in gray. NOEs corresponding to unknown distances are labeled in red.

of effect that preorganization of the unit III furanose has on the

binding strength. Close inspection of molecular models shows

that the global shape of the antibiotic is extremely dependent

on the precise puckering of this ring. In fact, the specific drug-

RNA contacts observed in the paromomycin-RNA complex,^5,6

involving units III and IV and presumably required for tight

binding, are only feasible when the furanose adopts a puckering

in the “South” region of the pseudorotational path, which is

the minor one in the free state. Nevertheless, the original NMR

studies by Puglisi and co-workers⁷showed that, for paromo-

mycin, these rings are dynamic in the ribosome-bound state,

and therefore the contacts observed in the crystal structure of

the complex might be transient. In addition, the Westhof and

Hanessian groups have recently described, for a paromomycin

derivative, a new binding mode to the A-site, for which the

ribose ring (unit III) is recognized by the RNA receptor in a

C2-endo conformation.³⁷

Finally, for compound 4, the maximum increase in fluores-

cence observed upon complex formation is less than half the

value observed for neomycin-B, suggesting a smaller structural

change in the receptor upon complex formation. It is known

that, in the paromomycin-A-site complex, OH5_IIIis located in

a strongly electronegative region of the RNA binding pocket.

Probably, the substitution of this OH group by a negatively

charged carboxylate precludes a deep penetration of the ligand

in the major groove. However, despite this modification, the

observed affinity for 4 is in the micromolar range.

Effect of the Conformational Constraint on the Antibiotic

Activity. The biological activity of the locked neomycin-B

derivatives 2-4 was tested against E. coli (see Table 2). For

derivative 2, a more extensive testing was performed employing

both Gram+ and Gram- bacteria. Despite the observed decrease

in activity with respect to the natural compound 1 (3-20-fold

depending on the bacteria and the compound), derivatives 2 and

3 are still significantly active (with MIC values in the 2.5-20

µg/mL range). In contrast, derivative 4 showed a poor antibiotic

activity. This suggests that 4 is adversely affected by other

factors such as a different binding mode (as suggested by

fluorescence experiments above), different kinetics for crossing

the bacterial membrane, digestion or modification by bacterial

enzymes, etc. With respect to 2 and 3, the obtained results show

a good correlation with the binding affinities previously

measured for ribosomal RNA and indicate that these locked

derivatives might potentially be employed as antibiotics.

Therefore, the inactivation of these analogues by aminoglycoside

modifying enzymes was tested.

Effect of the Conformational Constraint on Aminoglyco-

side Inactivation by Bacterial Defense Proteins. To analyze

the effect that conformational restriction has on the antibiotic

enzymatic inactivation, three different enzymes were chosen as

model systems: Staphylococcus aureus ANT(4′), Myco-

bacterium tuberculosis AAC(2′), and Enterococcus faecalis

APH(3′). These proteins are representative of the three main

families of enzymes that modify aminoglycosides: adenyltrans-

ferases, acyltransferases, and phosphotransferases. In addition,

there is high-resolution X-ray structural information available

(37) Francois, B.; Szychowski, J.; Adhikari, S. S.; Pachamuthu, K.; Swayze, E.

E.; Griffey, R. H.; Migawa, M. T.; Westhof, E.; Hanessian, S. Angew.

Chem., Int. Ed. 2004, 43, 6735-6738.

9

108 J. AM. CHEM. SOC. VOL. 128, NO. 1, 2006

Conformationally Locked Aminoglycosides

A R T I C L E S

Figure 3. (a) MD-tar Φ/Ψ distributions obtained for the I/II (upper panel) and III/II (lower panel) glycosidic linkages in compound 3. Minimum energy

geometries are indicated with a cyan dot. The Φ/Ψ values observed for the ligand in the paromomycin-RNA complex are indicated with a yellow dot. (b)

Representation of the major conformational populations detected in solution for 3. Unit IV is omitted for simplicity.

for the three enzymes in complex with several antibiotics.

Analysis of these data permits the deduction of either minor

(as in APH(3′)) or major (as in ANT(4′) or AAC(2′)) confor-

mational differences, for the aminoglycoside, in the protein-

bound state with respect to the RNA-bound state (see Scheme

1).^15,17,19

310 K was monitored by 1D NMR, and the final products were

characterized by 2D NMR and MALDI-TOF MS. As shown in

Figure 9, at 100 µM neomycin-B, 100 µM ATP, 2.5 mM MgCl₂,

and 0.2 µM ANT(4′), the adenylation reaction is completed

within 22 min. NMR analysis permitted us to confirm that

ANT(4′) selectively modifies OH4_I. In fact, if only 1 equiv of

ATP is employed, this is the only product detected. Interestingly,

if 2 ATP equiv are used, nonselective secondary adenylations

are detected but at much longer reaction times. For example, at

3 mM ATP, 1.5 mM neomycin-B, 5 mM MgCl₂(pH 7.0, 310

K), and 1 µM ANT(4′), the primary adenylation is completed

in 2 min. In contrast, the enzyme requires more than 12 h to

consume the second ATP equivalent, once the primary adenyl-

ation is completed. The presence of diadenylated neomycin-B

derivatives were confirmed by MALDI-TOF.

For the rigid analogue 2, at 100 µM aminoglycoside, 100

µM ATP, 2.5 mM MgCl₂, and 0.2 µM ANT(4′), no reaction

was detected even 24 h after enzyme addition (see Figure 9).

At larger ANT(4′) concentrations (up to 1 µM ANT(4′)), slow,

nonselective adenylation processes (similar to those described

above for 1) were observed. In contrast, the related deoxy

derivative 5 reacted smoothly to give the monoadenylated

compound within a few minutes, which demonstrates that the

behavior exhibited by 2 has its origin in the extra rigidity of

the antibiotic and that it is not related to the absence of OH5_III.

Therefore, the simple modification leading to conformationally

Staphylococcus aureus ANT(4′) catalyzes the transfer of one

adenyl group from ATP to position O-4 (see Scheme 1) of ring

I. This transfer leads to a sharp decrease in the drug affinity for

its target RNA. The 3D structure of ANT(4′) complexed to

kanamycin has been determined by using X-ray.¹⁵In addition,

the structures of related analogues, such as gentamycin, tobra-

mycin, and paromomycin bound to the target ribosomal RNA

have also been determined recently.^5,6Interestingly, the oligo-

saccharide conformation recognized by the RNA and by the

enzyme is remarkably different, as previously mentioned (see

Scheme 1). If it is assumed that some degree of conformational

distortion at the I /II fragment is in fact required for enzymatic

activity, the conformational restriction introduced in derivatives

2 and 3 might provide a certain degree of protection against

the enzymatic inactivation of the antibiotic.

To test this hypothesis, enzymatic reactions were monitored

in NMR tubes. ATP (100 µM-3 mM) and the aminoglycosides

(100 µM-1.5 mM) were dissolved in phosphate buffer (10 mM

phosphate, 2.5-5.0 mM MgCl₂, pH 7.0). After addition of the

enzyme (0.2-1.0 µM), the evolution of the reaction mixture at

9

J. AM. CHEM. SOC. VOL. 128, NO. 1, 2006 109

A R T I C L E S

Bastida et al.

Figure 4. (a) Upper panel: Schematic representation of some of the observed NOEs employed in the structural analysis of 4. The salt bridge interaction

between units III and I is highlighted with a dotted box. Lower panel: Representative NOE build-up curves. (b) MD-tar Φ/Ψ distributions obtained for the

I/II (upper) and III/II (lower) glycosidic linkages. Minimum energy geometries are indicated with a cyan dot. The Φ/Ψ values corresponding to the

paromomycin-RNA bound form are indicated with a yellow dot.

Figure 5. MD-tar ensembles of conformers for 2-4. The experimental ensemble corresponding to 1 in the free state (left) and that obtained from MD

simulations of the neomycin-RNA complex¹⁶(right) are shown for comparison. RMSDs (Å) for heavy atom superimposition are given below. The numbering

employed for the different sugar units is shown on the right.

restricted 2 provides an effective protection against aminogly-

coside inactivation by S. aureus ANT(4′).

enzymatic activity. For ANT(4′) a strong conformational restric-

tion (as present in 2) prevents the antibiotic enzymatic adenyl-

ation, although the limited flexibility present in 3 restores the

enzymatic activity.

In vivo activities of neomycin-B (1) and mimic 2 were tested

employing the bacteria E. coli DH5R (pBBR1MCS-2), which

expresses the resistance enzyme ANT(4′). As expected, the MIC

value for the natural antibiotic is significantly increased (from

3 to 60 µg/mL). In contrast, the cyclic derivative 2 maintains

the same activity observed for the nonresistant bacteria (20 µg/

mL). In conclusion, this simple modification leading to the

conformationally restricted 2 provides an effective protection

Surprisingly, the locked analogue 3 exhibited a similar

behavior to that observed for the natural antibiotic, neomycin-

B. Indeed, the specific activities of ANT(4′) for aminoglycosides

1 and 3 are almost identical (see Figure 9). According to these

data, it seems that ANT(4′) is able to recognize and inactivate

aminoglycoside conformations different from the anti-Ψ ge-

ometry observed in the enzyme-kanamycin complex¹⁵by X-ray

methods. In fact, this conformation is not accessible for

derivative 3. In any case, these results prove that some degree

of conformational flexibility in the ligand is critical for

9

110 J. AM. CHEM. SOC. VOL. 128, NO. 1, 2006

Conformationally Locked Aminoglycosides

A R T I C L E S

Table 2. Experimental MIC Values (µg/mL) Measured for

Neomycin-B and Its Conformationally Constrained Analogues 2-4

MICs (µg/mL)

1

2

3

4

E. coli (DH5R)

E. coli (BL21)

S. epidermis

Bacillus cereus

Alcaligenes faecalis

3

0.50

2

1

20

10

5

9

2.5

>100

20

concentrations (0.2-20 µM). Therefore, the conformational

constraint introduced in 2 and 3 provides complete protection

against aminoglycoside modification by AAC(2′).

Figure 6. Structures representative of the main conformational families

present in solution for neomycin-B derivatives 2 (left) and 3 (right)

superimposed on the X-ray structure of paromomycin in the complex with

ribosomal RNA, according to X-ray data. The maximum Φ/Ψ deviations

(for each glycosidic linkage) and RMSD (Å) values for heavy atom

superimpositions are shown. Unit IV is omitted for simplicity.

The in vivo activities of neomycin-B (1) and mimic 3 were

tested employing the bacteria E. coli BL21(DE3)(pET23a-

AAC(2′)), which expresses the resistance enzyme AAC(2′). The

MIC values measured for 1 and 3 were 2 and 2.5 µg/mL,

respectively. As expected, 3 is not affected by the expression

of AAC(2′), and this locked aminoglycoside exhibits a good

antibiotic activity. However, under the employed experimental

conditions, the AAC(2′) enzymatic activity seems to have a low

influence on the antibiotic properties of the natural compound

1. Although merely speculative, acetylation at position 2 of ring

I might have a small influence on the stability of the drug/A-

site complex. In fact, according to the structural information

available,^5-7the amino group located at this position of the

antibiotic is not involved in any key contact with the RNA target,

and therefore acetylated 1 might retain a significant activity.

Finally, the effect of neomycin-B flexibility on its enzymatic

inactivation by Enterococcus faecalis APH(3′) was tested. This

enzyme catalyzes the transfer of a phosphate group from ATP

to position 4 (see Scheme 1) of ring I. In this case, the X-ray

structure of the enzyme in complex with neomycin-B shows

that the antibiotic structure recognized by the enzyme and the

target RNA are remarkably similar (see Scheme 1). In fact, they

mainly differ in the conformation adopted by the furanose ring

of III (C3-endo and close to C2-endo for the RNA- and enzyme-

bound conformations, respectively).

against aminoglycoside inactivation by S. aureus ANT(4′), both

in vivo and in vitro, while maintaining a significant antibiotic

activity.

Mycobacterium tuberculosis AAC(2′) catalyzes the transfer

of one acetyl group from Acetyl CoA to position N-2 (see

Scheme 1) of the ring I. The 3D structure of AAC(2′) complexed

to ribostamycin has been recently determined by X-ray.¹⁷As

previously mentioned, the oligosaccharide conformations bound

by the RNA and by the enzyme are remarkably different (see

Scheme 1). Interestingly, the recognition of a high energy

conformation of the antibiotic seems essential for the enzymatic

activity. Thus, by forcing an anti-Ψ geometry around the III/II

linkage, the protein increases the accessibility of the reacting

NH2_Igroup which, otherwise, would be hydrogen-bonded to

the nonvicinal unit III in the low energy syn-Ψ minimum.

According to the conformational analysis described above, for

cyclic derivatives 2 and 3, this conformational change is no

longer possible. Therefore, the geometrical restriction introduced

in these derivatives is expected to fully protect the antibiotics

from the enzymatic acetylation. To test this hypothesis, enzy-

matic reactions were performed in NMR tubes employing

similar conditions as those previously described for ANT(4′)

(see Materials and Methods). As shown in Figure 10, at 100

µM neomycin-B, 100 µM AcCoA, 2.5 mM MgCl₂, and 0.2 µM

AAC(2′), the acetylation reaction is completed within 1 h. A

similar behavior was observed for the deoxy derivatives 5 and

6. In contrast, no reaction was observed for the locked

aminoglycosides 2 and 3, even after 24 h and higher enzyme

As shown in Figure 11, at 100 µM neomycin-B, 100 µM

ATP, 2.5 mM MgCl₂, and 0.2 µM APH(3′), a 50% antibiotic

transformation is achieved in 17 min, and the reaction is almost

complete in 40 min. Under identical conditions, the enzyme

needs 35 min to modify 50% of analogue 3. For the more rigid

derivative 2, the reaction is even slower (around 100 min for a

50% transformation). In conclusion, although the extra rigidity

Figure 7. (a) Schematic representation of the modified RNA fragment employed in our binding studies and the structural change induced by aminoglycoside

binding. (b) Some representative binding curves obtained for derivatives 1 (black), 2 (red) and 3 (cyan) under identical conditions.

9

J. AM. CHEM. SOC. VOL. 128, NO. 1, 2006 111

A R T I C L E S

Bastida et al.

neomycin-B locked derivatives 2 and 3 have been specifically

designed to mimic the bioactive RNA-bound structure of the

drug. In fact, both antibiotics maintain a reasonable affinity for

the target A-site RNA and exhibit significant biological activi-

ties. For compound 2, the conformational constraint provides

full protection against the aminoglycoside modification by

Staphylococcus aureus ANT(4′) and Mycobacterium tubercu-

losis AAC(2′) and partially protects against modification by

Enterococcus faecalis APH(3′). In fact, 2 presents an improved

biological activity with respect to neomycin-B in bacteria

expressing Staphylococcus aureus ANT(4′). In addition, deriva-

tive 3 is not susceptible to modification by Mycobacterium

tuberculosis AAC(2′). However, for derivatives 2 and 3, the

price paid in terms of RNA binding affinity is not negligible

(around 1.6 kcal/mol in ∆G). Clearly, more structure activity

studies are needed in the future to address this problem.

Interestingly, our RNA binding studies show that for these

compounds the observed decrease in affinity has its origin,

mainly, in the loss of the OH at position 5 of ring III. In contrast,

the conformational restriction does not seem to be critical.

According to X-ray studies OH5_IIIis involved in a hydrogen

bond with G1491 in the neomycin-B/ribosome complex. Our

results indicate that this particular interaction plays a significant

role in the stabilization of the complex. Thus by participating

in a direct ligand-RNA interaction the OH5_IIImight help to

maintain an appropriate orientation of units III and IV within

the RNA binding pocket.³²Therefore, efforts should be made

in future to optimize the ligand-RNA contacts established by

rings III and IV while maintaining the conformational restriction.

Figure 8. (a) Kd values obtanied for aminoglycosides 1-6. (b) Schematic

representation of the ∆G values measured for the binding of aminoglycosides

1, 2, 3, 5 and 6 to the A-site.

of the locked derivatives provides a certain protection against

modification by APH(3′), in this case, it is insufficient to prevent

the enzymatic inactivation of the drugs. This result is not totally

unexpected, given the close similarity of the aminoglycoside

structures in the binding pockets of both receptors (the A-site

RNA and the enzyme).^6,7,19

The in vivo activities of neomycin-B (1) and mimic 2 were

tested employing the bacteria E. coli DH5R(pACYC177), which

expresses the resistance enzyme APH(3′). The MIC values for

the natural antibiotic and cyclic derivatives 2 and 3 were in all

cases >150 µg/mL. Therefore, in this case, although the extra

rigidity present in compound 2 provides a certain in vitro

protection against the antibiotic enzymatic inactivation, this is

clearly insufficient to warrant a good biological activity.

In our opinion, this example represents a test case for the

validity of a structure-based approach for the design and

preparation of ligands that specifically interact with a given

receptor and that, in this particular case, might potentially be

used as antibiotics. Finally, we would like to stress that the utility

of conformational restriction to prevent the enzymatic inactiva-

tion of a particular drug while preserving its biological activity

might not be restricted to aminoglycosides and could find further

applications within medicinal chemistry.

Conclusions

We have shown that the conformational differences exhibited

by aminoglycosides within the binding pockets of the ribosome

and of those enzymes involved in bacterial resistance can, in

certain favorable cases, be exploited for designing new antibiotic

derivatives not susceptible to enzymatic inactivation. Thus, the

Figure 9. Time evolution (at pH 7.0 and 310 K) of neomycin-B (1, left), derivative 2 (middle), and derivative 3 (right) at 100 µM concentration in the

presence of ATP (100 µM), phosphate (10 mM), and MgCl₂(2.5 mM) after addition of 0.2 µM ANT(4′) followed by 1-D NMR. For derivative 2, no reaction

was detected. 1-Ad and 3-Ad stand for the adenylated derivatives 1 and 3, respectively. The aminoglycoside position modified by the enzyme is shown

schematically in the inset.

9

112 J. AM. CHEM. SOC. VOL. 128, NO. 1, 2006

Conformationally Locked Aminoglycosides

A R T I C L E S

Figure 10. Time evolution (at pH 7.0 and 310 K) of neomycin-B (left), derivative 2 (middle), and derivative 3 (right) at 100 µM concentration in the

presence of AcCoA (100 µM) and phosphate (10 mM) after addition of 0.2 µM AAC(2′), followed by 1-D NMR. For derivatives 2 and 3, no reaction was

detected even after 24 h, independently of the concentration of the enzyme employed. 1-Ac stands for acetylated neomycin-B. The aminoglycoside position

modified by the enzyme is shown schematically in the inset. The X-ray structure of ribostamycin in complex with AAC(2′) is also represented.

Figure 11. Time evolution (at pH 7.0 and 310 K) of neomycin-B (left), derivative 2 (middle), and derivative 3 (right) at 100 µM concentration in the

presence of ATP (100 µM), phosphate (10 mM), and MgCl₂(2.5 mM) after addition of 0.2 µM APH(3′) followed by NMR. 1-P, 2-P, and 3-P stand for the

modified antibiotics 1, 2 and 3, respectively. The aminoglycoside position modified by the enzyme is shown schematically in the inset.

MHz, CD₃OD ): δ 7.4-7.2 (m, 30H, CH₂Ph), 5.43 (broad s, anomeric

proton), 5.24 (broad s, anomeric proton), 5.08 (broad s, anomeric

proton), 5.2-5.0 (m, 12H, CH₂Ph), 2.91 (m, 3H, CH(CH₃)₂), 1.21 (18H,

CH(CH₃)₂). ¹³C NMR (75.47 MHz, CD₃OD): 130-129 (CH₂Ph), 111.0

(anomeric carbon), 99.8 (anomeric carbon), 99.0 (anomeric carbon),

69-68, (CH₂Ph), 35.5 (CH(CH₃)₂), 35.4 (C2_II), 25.4 (CH(CH₃)₂), 81.8,

81.1, 71.8, 71.2, 69.3, 57.7, 54.2, 52.8, 52.6, 43.3, 42.8. MALDI-TOF

m/z 1708.3 [M + Na⁺].

Materials and Methods

Synthesis of the Aminoglycoside Derivatives. Compound 8.

Compound 7 (500 mg, 0.34 mmol) was dissolved in pyridine (10 mL),

and 2,4,6-triisopropylbenzene sulfonyl chloride (3.2 g, 10.6 mmol) was

added under argon at rt. After stirring for 72 h, the mixture was diluted

with EtOAc (300 mL) and H₂O (100 mL) and then neutralized with 1

N HCl. The aqueous phase was extracted with EtOAc (2 × 100 mL),

and the combined organic layers were washed with brine (50 mL), dried

over Na₂SO₄, and evaporated at reduced pressure; the crude was purified

by column chromatography on silica gel (CH₂Cl₂/MeOH, 20:1) to give

Compound 9. Palladium on charcoal (1:5 catalyst/substrate by

weight) was added to a solution of 8 (50 mg, 0.030 mmol) and

trifluoroacetic acid (20 µL, 0.26 mmol) in MeOH (3 mL). After the

1

8 (250 mg, 44%). R_f) 0.5 (CH₂Cl₂/MeOH 10:1). H NMR (500.13

9

J. AM. CHEM. SOC. VOL. 128, NO. 1, 2006 113

A R T I C L E S

Bastida et al.

suspension was stirred under H₂(1 atm), at rt for 8 h, the catalyst was

removed by filtration, and the solvent was evaporated to give 9 (24

mg, 90%), which was used in the next step without further purification.

R_f) 0.21 (NH₄OH/CHCl₃/n-BuOH/EtOH 3:1:2:2.5) ¹H NMR (500.13

MHz, D₂O ): δ 5.55 (broad s, 1H, H1_I), 5.25 (broad s, 1H, H1_III), 5.01

(broad s, 1H, H1_IV), 1.90 (m, 1H, H2_IIequatorial), 1.15 (m, 1H, H2_II

axial). MALDI-TOF m/z 903.7 [M + Na⁺].

was removed at reduced pressure, and the crude was purified by column

chromatography on silica gel (EtOAc/hexane, 1:1) to give compound

13 (215 mg, 82%). R_f) 0.3 (AcOEt/hexane 2:1). H NMR (500.13

1

MHz, CDCl₃): δ 7.4-7.2 (m, 30H, CH₂Ph), 7.16 (broad m, 1H), 6.33

(broad s, 1H), 5.2-4.9 (12H, CH₂Ph), 3.90 (s, 3H, CO₂CH₃), 2.1-1.8

(18H, CH₃CO₂), 1.98 (m, 1H, H2_IIequatorial), 1,40 (m, 1H, H2_IIaxial).

¹³C NMR (75.47 MHz, CDCl₃): 130-129 (CH₂Ph), 68-67 (CH₂Ph),

34.6 (C2_II), 21-20 (CH₃CO₂). MALDI-TOF m/z 1748.7 [M + Na⁺].

Compound 14. To a solution of compound 13 (200 mg, 0.12 mmol)

in THF/EtOH (1:2, 10 mL) at rt was added NaBH₄(54 mg, 1.46 mmol)

and LiCl (6 mg, 0.15 mmol). After the mixture was stirred at rt for 12

h, EtOAc (25 mL) was added, and the mixture was washed with

aqueous saturated NaHCO₃and brine. The organic phase was dried

over Na₂SO₄, the solvent was removed at reduced pressure, and the

crude was purified by flash chromatography (CH₂Cl₂/MeOH, 20:1) to

give 14 (155 mg, 89%). R_f) 0.4 (CH₂Cl₂/MeOH 10:1). ¹H NMR

(500.13 MHz, CD₃OD): δ 7.4-7.2 (m, 30H, CH₂Ph), 5.16 (broad s,

1H, H1_I), 5.04 (broad s, 1H, H1_III), 5.2-5.0 (12H, CH₂Ph), 4.78 (broad

s, 1H, H1_IV), 1.99 (m, 1H, H2_IIequatorial), 1.62-1.60 (m, 4H, -CH₂-

CH₂-CH₂-OH), 1.39 (m, 1H, H2_IIaxial). ¹³C NMR (75.47 MHz, CD₃-

OD): 130-129 (CH₂Ph), 111.3 (C1_III), 101.2 (C1_I), 100.9 (C1_IV), 69-

68, (CH₂Ph), 35.1 (C2_II), 30.6 (-CH₂-CH₂-CH₂-OH), 29.5 (-CH₂-

CH₂-CH₂-OH), 86.8, 82.8, 76.4, 74.9, 73.6, 73.4, 72.9, 71.6, 69.4,

63.0, 57.7, 54.2, 52.6, 43.1, 42.5. MALDI-TOF m/z 1469.7 [M + Na⁺].

Compound 15. Treatment of compound 14 (150 mg, 0.10 mmol)

at 0 °C following the same procedure described for the preparation of

compound 10 gave the silyl ether 15 (97 mg, 60%). R_f) 0.4 (CH₂Cl₂/

Compound 2. A diluted solution of 9 (50 mg, 0.056 mmol) in H₂O

(100 mL, pH 7.0) was stirred at 60 °C for 5 days. The solvent was

removed at reduced pressure, and the crude was purified by ion

exchange column chromatography (0.5 cm × 7 cm) with Amberlite

CG-50 (0-7.5% NH₄OH) to give, after lyophilization, compound 2

(18 mg, 56%). R_f) 0.16 (NH₄OH/CHCl₃/n-BuOH/EtOH 3:1:2:2.5).

¹H NMR (500.13 MHz, D₂O ): δ 5.75 (d, J ) 3.9 Hz, 1H, H1_III), 5.20

(d, J ) 3.5 Hz, 1H, H1_I), 5.03 (broad s, 1H, H1_IV), 2.72 (1H, H5_III),

2.55 (1H, H2_I), 1.95 (m, 1H, H2_IIequatorial), 1.16 (m, 1H, H2_IIaxial).

¹³C NMR (75.47 MHz, D₂O): δ 100.3 (C1_IV), 109.5 (C1_III), 98.2 (C1_I),

60.7 (C2 ), 48.4 (C5_III). MALDI-TOF m/z 619.1 [M + Na⁺].

I

Compound 10. To a solution of compound 7 (1.0 g, 0.7 mmol) in

dry THF (7 mL) at 0 °C under argon was added 2,6-lutidine (0.24 mL,

2.1 mmol) and TIPSOTf (0.28 mL, 1.05 mmol). The mixture was

allowed to warm to rt and stirred for 1 h. The solvent was evaporated,

and the crude was purified by flash chromatography (CH₂Cl₂/MeOH,

15:1) to give the silyl ether derivative (0.88 g, 80%). R_f) 0.4 (CH₂Cl₂/

MeOH 10:1). To a solution of this derivative (1.0 g, 0.63 mmol) in

pyridine (15 mL) was added at rt Ac₂O (7 mL). After the mixture was

stirred for 5 h at rt, the solvent was removed at reduced pressure, and

the crude was purified by column chromatography on silica gel (EtOAc/

hexane, 1:1) to give 10 (1.02 g, 100%). R_f) 0.6 (EtOAc/hexane 2:1).

¹H NMR (500.13 MHz, CD₃OD): δ 7.4-7.2 (m, 30H, CH₂Ph), 5.2-

5.0 (12H, CH₂Ph), 2.1-1.8 (18H, CH₃CO₂), 1.97 (m, 1H, H2_II

equatorial), 1.36 (m, 1H, H2_IIaxial), 1.32 (m, 3H, SiCH(CH₃)₂), 1.10

(18H, SiCH(CH₃)₂). ¹³C NMR (75.47 MHz, CD₃OD ): 130-129

(CH₂Ph), 69-68, (CH₂Ph), 35.4 (C2_II), 21-20 (CH₃CO₂), 18.5

(SiCH(CH₃)₂). MALDI-TOF m/z 1850.2 [M + Na⁺].

1

MeOH 10:1). H NMR (500.13 MHz, CD₃OD): δ 7.4-7.2 (m, 30H,

CH₂Ph), 5.26 (broad s, 1H, H1_I), 5.07 (broad s, 1H, H1_III), 5.2-5.0

(12H, CH₂Ph), 4.85 (broad s, 1H, H1_IV), 1.96 (m, 1H, H2_IIequatorial),

1.66-1.63 (m, 4H, -CH₂-CH₂-CH₂-OTIPS), 1,38 (m, 1H, H2_II

axial), 1.04 (18H, CH(CH₃)₂). ¹³C NMR (75.47 MHz, CD₃OD): 130-

129 (CH₂Ph), 111.0 (C1_III), 100.7 (C1_IV), 100.5 (C1_I), 69-68, (CH₂Ph),

35.1 (C2_II), 31.1 (-CH₂-CH₂-CH₂-OTIPS), 26.7(-CH₂-CH₂-

CH₂-OTIPS), 18.31 (CH(CH₃)₂), 86.7, 82.7, 82.2, 75.8, 74.7, 73.5,

73.1, 72.0, 69.3, 64.8, 57.9, 54.3, 53.3, 52.5, 43.4, 42.5. MALDI-TOF

m/z 1626.2 [M + Na⁺].

Compound 16. Acetylation of 15 (97 mg, 0.060 mmol) employing

the same methodology described for 10 gave the fully acetylated

derivative in quantitative yield. R_f) 0.3 (CH₂Cl₂/MeOH 20:1). From

this compound (111 mg, 0.060 mmol), derivative 16 (81 mg, 0.048

mmol) was obtained following a procedure identical to that described

for 11; yield: 79%. R_f) 0.25 (CH₂Cl₂/MeOH 20:1). ¹H NMR (500.13

MHz, CD₃OD): δ 7.4-7.2 (m, 30H, CH₂Ph), 5.1-4.9 (12H, CH₂Ph),

2.1-1.9 (18H, CH₃CO₂), 1.94 (m, 1H, H2_IIequatorial), 1.63 (m, 1H,

H2_IIaxial). ¹³C NMR (75.47 MHz, CD₃OD): 130-129 (CH₂Ph), 68-

67, (CH₂Ph), 34.5 (C2_II), 21-20 (CH₃CO₂). MALDI-TOF m/z 1721.8

[M + Na⁺].

Compound 17. To a solution of compound 16 (81 mg, 0.048 mmol)

in pyridine (2 mL) was added DMAP (1 mg, 0.008 mmol) and

p-toluenesulfonyl chloride (14 mg, 0.073 mmol). After stirring the

mixture, at room temperature under argon for 3 h, the solvent was

evaporated and the crude was purified by flash chromatography (EtOAc/

hexane, 1:1) to give the primary sulfonate (70 mg, 79%). R_f) 0.4

(CH₂Cl₂/MeOH 20:1). To a solution of this derivative (70 mg, 0.038

mmol) in MeOH/THF (2:1, 300 µL) was added KCN (2 mg, 0.031

mmol). After stirring at room temperature for 10 h, the solvent was

removed and the crude was purified by column chromatography on

silica gel (CH₂Cl₂/MeOH, 20:1) to give the deacetylated derivative 17

(36 mg, 58%). R_f) 0.44 (CH₂Cl₂/MeOH 10:1). ¹H NMR (500.13 MHz,

CD₃OD): δ 7.75 (d, J ) 8 Hz, 2H, O-Ts), 7.36 (d, J ) 8 Hz, 2H,

O-Ts), 7.4-7.2 (m, 30H, CH₂Ph), 5.24 (broad s, 1H, H1_I), 5.01 (broad

s, 1H, H1_III), 5.2-5.0 (12H, CH₂Ph), 4.81 (broad s, 1H, H1_IV), 1.97

(m, 1H, H2_IIequatorial), 1.65-1.60 (m, 4H, -CH₂-CH₂-CH₂-OTs),

1.36 (m, 1H, H2_IIaxial). ¹³C NMR (75.47 MHz, CD₃OD): 130-129

(CH₂Ph), 111.6 (C1_III), 100.6 (C1_IV), 100.5 (C1_I), 69-68, (CH₂Ph),

Compound 11. To a solution of the silyl ether 10 (0.5 g, 0.28 mmol)

in dry THF (5 mL), in a polyethylene vessel at 0 °C, was added HF/

pyridine (0.3 mL). After stirring at rt for 5 h, the mixture was diluted

with Et₂O and neutralized with aqueous saturated NaHCO₃. The

combined organic layer was dried, evaporated, and purified by flash

chromatography (EtOAc/hexane, 1:1) to give alcohol 11 (0.42 g, 90%).

1

R_f) 0.3 (AcOEt/hexane 2:1). H NMR (500.13 MHz, CD₃OD): δ

7.4-7.2 (m, 30H, CH₂Ph), 5.81 (broad s, anomeric proton), 5.72 (broad

s, anomeric proton), 5.18 (broad s, anomeric proton), 5.2-4.9 (m, 12H,

CH₂Ph), 2.2-1.9 (18H, CH₃CO₂) 1.95 (m, 1H, H2_IIequatorial), 1,64

(m, 1H, H2_IIaxial). ¹³C NMR (75.47 MHz, CD₃OD ): 130-129

(CH₂Ph), 110.5 (anomeric carbon), 98.2 (anomeric carbon), 68-67,

(CH₂Ph), 34.4 (C2_II), 21-20 (CH₃CO₂), 84.1, 83.4, 76.1, 74.1, 72.4,

69.8, 62.4, 54.9, 51.1, 42.1. MALDI-TOF m/z 1694.6 [M + Na⁺].

Compound 12. Dess-Martin periodinane (95 mg, 0.26 mmol) was

added to a solution of alcohol 11 (250 mg, 0.15 mmol) in CH₂Cl₂(5

mL). After the mixture was stirred for 2 h at rt, Et₂O was added (20

mL), and the mixture was poured into saturated aqueous NaHCO₃

solution (25 mL) and Na₂S₂O₅(100 mg). The organic layer was washed

with saturated aqueous NaHCO₃(25 mL) and brine (25 mL) and dried,

and the solvent evaporated at reduced presure to give, after purification

by flash chromatography (EtOAc/hexane, 2:1), aldehyde 12 (199 mg,

79%). R_f) 0.3 (AcOEt/hexane 2:1). ¹H NMR (500.13 MHz, CDCl₃):

δ 9.4 (broad s, 1H, CHO), 7.4-7.2 (m, 30H, CH₂Ph), 5.2-4.9 (12H,

CH₂Ph), 2.1-1.9 (18H, CH₃CO₂), 1.92 (m, 1H, H2_IIequatorial), 1.41

(m, 1H, H2_IIaxial). ¹³C NMR (75.47 MHz, CDCl₃): 130-129

(CH₂Ph), 68-67, (CH₂Ph), 34.6 (C2_II), 21-20 (CH₃CO₂). MALDI-

TOF m/z 1691.58 [M + Na⁺].

Compound 13. To a solution of 12 (250 mg, 0.15 mmol) in dry

acetonitrile (6 mL) was added (methoxycarbonylmethylen)triphenyl-

phosphorane (66 mg, 0.19 mmol). After stirring at rt for 3 h, the solvent

9

114 J. AM. CHEM. SOC. VOL. 128, NO. 1, 2006

Conformationally Locked Aminoglycosides

A R T I C L E S

35.4 (C2_II), 30.8 (-CH₂-CH₂-CH₂-OTs), 26.5(-CH₂-CH₂-CH₂-

OTs), 86.8, 82.5, 81.9, 76.3, 73.7, 73.3, 72.2, 69.5, 68.4, 57.8, 54.6,

52.8, 52.5, 43.5, 42.6. MALDI-TOF m/z 1623.3 [M + Na⁺].

time averaged J coupling constraints. A r^-6^-1/6average was used for

the distances, and a linear average was used for the coupling constants.

For derivatives 2 and 4, final trajectories were carried out using explicit

solvent, counterions, periodic boundary conditions (PBC) and Ewald

sums for the treatment of electrostatic interactions. An exponential decay

constant of 500 ps and simulation lengths of 5 ns were employed.

For compound 3, the experimental data could only be quantitatively

reproduced after (in vacuo ꢀ ) 80) 80 ns MD-tar simulations.

Preliminary tests showed that, for this flexible molecule, the use of

exponential decay constants shorter than 5 ns produced unstable

trajectories and led in some cases to severe distortions of the sugar

rings. This problem could be solved by increasing the exponential decay

constant value to 8-10 ns. It has been estimated that simulation lengths

of ca. 1 order of magnitude larger than the exponential decay constant

should be used to generate reliable estimates of average properties.

Therefore the final simulation length used was 80 ns. We would like

to point out that for MD-tar runs the sensitivity of the results to the

simulation conditions (solvent, counterions, cutoffs, dielectric constant,

etc.) used is extremely dependent on the amount of experimental

information included as time-averaged constraints. For derivative 3 we

employed 23 distance constraints and 7 J values. Under these conditions,

the conformational regions sampled by the trajectory (within the

esterically allowable regions) are mainly determined by the constraints,

and the fine details of the simulation are not critical. In fact similar

results were obtained using ꢀ ) 40 and ꢀ ) 20.

Finally, to further clarify the conformational properties of 3, we

performed additional calculations. Thus, a geometry representative of

each of the five most populated conformational families present for

derivative 3 in solution, according to the in vacuo ꢀ ) 80 MD-tar

calculations, was subjected to 5 ns unconstrained MD simulations in

the presence of explicit water, counterions, periodic boundary condi-

tions, and Ewald sums for the treatment of electrostatic interactions.

The obtained trajectories show that, as deduced from the in vacuo MD-

tar simulations, the macrocyclic structure of compound 3 is compatible

with wide fluctuations around the glycosidic linkages, different

puckerings for ring III and different orientations for the three methylene

bridge. In fact, the conformational regions populated by 3 according

to these simulations are similar those deduced from the in vacuo 80 ns

MD-tar trajectories (Figure S7).

Compound 3. Treatment of 17 (12 mg, 0.0075 mmol) under the

same conditions described for the preparation of compound 9 gave the

fully deprotected tosilate (5 mg, 88%) which was used in the next step

without further purification. R_f) 0.16 (NH₄OH/CHCl₃/ⁿBuOH/EtOH,

3:1:2:2.5). A diluted solution of this compound (5 mg, 0.0066 mmol)

in H₂O (100 mL, pH 7.0) was stirred at 40 °C for 10 days. The solvent

was removed at reduced pressure, and the crude was purified by ion

exchange column chromatography (0.5 cm × 7 cm) with Amberlite

CG-50 (0-7.5% NH₄OH) to give, after lyophilization, compound 3 (3

mg, 62%). R_f) 0.16 (NH₄OH/CHCl₃/n-BuOH/EtOH 3:1:2:2.5). ¹H

NMR (500.13 MHz, D₂O): 5.9 (d, J ) 3.9 Hz, 1H, H1_I), 5.5 (d, J )

3.5 Hz, 1H, H1_III), 5.2 (d, J ) 1.5 Hz, 1H, H1_IV), 4.3 (1H, H2_III), 4.2

(1H, H4_III), 3.6 (1H, H2_IV), 3.6 (1H, H7_III), 3.5 (1H, H2_I), 3.1 (1H,

H7_III), 2.5 (m, 1H, H2_IIequatorial), 2.3 (1H, H5_III), 2.1 (1H, H6_III), 1.8

(m, 1H, H2_IIaxial), 1.7 (1H, H5_III), 1.7 (1H, H6_III). ¹³C NMR (75.47

MHz, D₂O): 108.7 (C1_III), 96.4 (C1_I), 96.1 (C1_IV), 73.2 (C2_III), 59.3

(C2_I), 50.8 (C2_IV), 49.8 (C7_III), 31.3 (C5_III), 28.2 (C2_II), 21.8 (C6_III).

MALDI-TOF m/z 648.1 [M + Na⁺].

Compound 6. Palladium on charcoal (1:5 catalyst/substrate by

weight) was added to a solution of 17 (25 mg, 0.016 mmol) and

trifluoroacetic acid (20 µL, 0.26 mmol) in MeOH (3 mL). After stirring

the suspension under H₂(1 atm) at room temperature for 48 h, the

catalyst was removed by filtration, and the solvent was evaporated at

reduced pressure. Derivative 6 (6.3 mg, 0.010 mmol) was purified by

ion exchange column chromatography (0.5 cm × 7 cm) with Amberlite

CG-50 (0-7.5% NH₄OH); yield: ∼60%. R_f) 0.24 (NH₄OH/CHCl₃/

ⁿBuOH/EtOH, 3:1:2:2.5). ¹H NMR (500.13 MHz, D₂O ): δ 5.9 (broad

s,1H, H1_I), 5.3 (d, J ) 3.5 Hz, 1H, H1_III), 5.2 (broad s, 1H, H1_IV), 4.3

(1H, H2_III), 3.5 (1H, H2_IV), 3.4 (1H, H6_I), 3.4 (1H, H6_IV), 3.4 (1H,

H6_IV), 3.3 (1H, H2_I), 3.2 (1H, H6_I), 2.3 (m, 1H, H2_IIequatorial), 1.7

(m, 1H, H2_IIaxial). ¹³C NMR (75.47 MHz, D₂O): 109.7 (C1_III), 95.7

(C1_IV), 95.1(C1_I), 79.4 (C2_III), 53.5 (C2 ), 50.9 (C2_IV), 40.5 (C6_IV),

I

40.2 (C6_I), 29.7 (C2_II). MALDI-TOF m/z 649.8 [M + Na⁺].

NMR and Molecular Modeling of 2-4. RESP atomic charges for

neomycin-B analogues 2-4 were derived by applying the RESP module

of the AMBER 5.0 package²⁶to the HF/6-31G(d) ESP charges

calculated with Gaussian-94.³⁸All MD simulations were carried

out using the SANDER module within AMBER and the Cornell et al.

force field.³⁹Parameters for the acetalic functions were taken from

GLYCAM.⁴⁰

Binding Experiments. The labeled 27mer A-site model was obtained

in its polyacrylamide gel electrophoresis- (PAGE-) purified sodium salt

form from Dharmacon Research, Inc. (Lafayette, CO). Buffer conditions

for the binding experiments were 100 mM NaCl, 10 mM sodium

phosphate (pH 7.6). Fluorescence experiments were conducted on a

Fluorolog-3 spectrophotometer (Jobin Yvon-Spex Instruments S.A.,

Inc., New Jersey) equipped with a thermostat. A quartz cell with a 1

cm path length in both the excitation and emission directions was used

in all the experiments that were performed at 25 °C. The RNA

concentration was, in all cases, 1 µM in strand. For renaturing the RNA,

it was placed in a water bath at 20 °C, heated to 85 °C for 1 min and

then slowly cooled back to 20 °C over a 2 h period. With the excitation

wavelength set to 310 nm, the change in the fluorescence emission at

369 nm upon complex formation was followed in titration experiments.

Aliquots of the antibiotic derivatives were added sequentially, and 2

min of equilibration time were allowed before each fluorescence

measurement. Fluorescence intensities were corrected for volume

changes according to the following equation: F_i,corr) F_i,obs*V_i/V_owhere

F_i,corris the corrected intensity for point i of the titration, F_i,obsis the

measured intensity at point i, V_iis the volume after the i^thaddition and

V_ois the initial volume (typically 2 mL). The Kd values for simple

binding were determined by plotting F_i,corragainst the ligand to RNA

ratio. In all cases, the data were found to fit well assuming a 1:1

complex.

NMR Structure of Derivatives 2-4. NMR experiments were

recorded on a Varian Unity 500 and Bruker Avance 500 at 313 K.

Selective 1D NOE experiments were recorded employing the 1D-

DPFGSE NOE pulse sequence. NOE intensities were normalized with

respect to the diagonal peak at zero mixing time. Selective T₁

measurements were performed on the anomeric and several other

protons to obtain the above-mentioned values. Experimental NOEs were

fitted to a double exponential function, f(t) ) p0(e^-p1t)(1 - e^-p2t) with

p0, p1, and p2 being adjustable parameters.⁴¹The initial slope was

determined from the first derivative at time t ) 0, f ′(0) ) p0p2. From

the initial slopes, interproton distances were obtained by employing

the isolated spin pair approximation.

To obtain a NMR-derived ensemble, MD-tar²⁵simulations were

performed for compounds 2-4 employing a protocol identical to that

described by Asensio et al.¹⁶NOE-derived distances were included as

time averaged distance constraints, and scalar coupling constants, as

(38) Pople, J. A. et al. Gaussian-94; Gaussian, Inc.: Pittsburgh, PA, 1995.

(39) Cornell, W. D.; Cieplack, P. C.; Bayly, I.; Gould, I. R.; Merz, K.; Ferguson,

D. M.; Spellmeyer, D. C.; Fox, T.; Caldwell, J. W.; Kollman, P. A. J. Am.

Chem. Soc. 1995, 117, 5179-5197.

(40) Woods, R. J.; Dwek, R. A.; Edge, C. J.; Fraser-Reid, D. J. Phys. Chem.

1995, 99, 3832-3839.

Cloning, Overexpression, and Purification of the Aminoglyside

Modifying Enzymes. The plasmid codifying for the AAC(2′)-Ic from

Mycobacterium tuberculosis was kindly provided by J. S. Blanchard

(41) Haselhorst, T.; Weimar, T.; Peters, T. J. Am. Chem. Soc. 2001, 123, 10705-

10714.

9

J. AM. CHEM. SOC. VOL. 128, NO. 1, 2006 115

A R T I C L E S

Bastida et al.

(Albert Einstein College of Medicine, New York). This plasmid was

transformed in the E. coli strain BL21(DE3). The expression and

purification of Mycobacterium tuberculosis AAC(2′) was performed

as described by J. S. Blanchard and co.⁴²

20 µM), the evolution of the reaction mixture at 37 °C was monitored

by 1D ¹H NMR, and the final products were characterized by 2D NMR

and MALDI-TOF MS.

Biological Activity. Minimal inhibitory concentrations (MIC) were

measured according to a published method.⁴³The different bacteria

were grown in 1 mL of Mueller-Hinton broth to an optical density

OD₆₀₀of 0.5 units. The desired concentrations of antibiotic were added

from stock solutions. The samples were incubated at 37 °C for 24 h.

After this time the control culture (with no antibiotic) had an OD₆₀₀of

1.2-1.6. Then the OD₆₀₀of each sample was read, and the MIC was

taken as the lowest antibiotic concentration inhibiting bacterial growth

by greater than 90%.

To determine the effect that expression of AAC(2′) has on the MIC

values the previously described protocol was slightly modified. The

different bacteria were grown in 1 mL of Mueller-Hinton broth to an

optical density OD₆₀₀of 0.3 units. The temperature was switched to

30 °C, and the culture was induced with 0.5 mm IPTG. When the OD₆₀₀

reached 0.5 units, the desired concentrations of antibiotic were added

from stock solutions. The samples were incubated at 30 °C for 24 h,

and the MIC was taken as the lowest antibiotic concentration inhibiting

bacterial growth by greater than 90%.

Enterococcus faecalis APH(3′) and Staphylococcus aureus ANT(4′)

were subcloned from the plasmids pACYC177 and pBBR1MCS-2,

respectively. PCR amplifications were performed in a 100 µL reaction

mixture containing DNA (3 µL) of the plasmids as a template, water

(70.5 µL), buffer (100 mM Tris-HCL, 500 mM KCl, pH 8.0, 10 µL),

MgCl₂(2 mM, 8µL), dNTPs (200 µM), primers (1 µM), and Taq DNA

polymerase (2.5 U). The primers employed for Enterococcus faecalis

APH(3′) were Nt (NdeI) (5’tataacatatgagccatattcaaggg3′) and Ct (EcoR

I)- (5’ttatagaattcttagaaaaactcatc3′), and those used for Staphylococcus

aureus ANT(4′) were Nt(Hind III)(5′gccgcaagattctgaatggaccaataata3′)

and Ct(Xho I) (5′gcctcgagtcaaaatggtatgcgttt3′). The reaction was

subjected to 30 cycles of amplification. The cycle conditions were set

as follows: denaturation at 94 °C for 1 min, annealing at 50 °C/1.0

min for Enterococcus faecalis APH(3′), and 55 °C/1.0 min for

Staphylococcus aureus ANT(4′) and elongation at 72 °C for 1 min.

The genes obtained from the PCR were purified with the Wizard PCR

Preps DNA Purification System (Promega) and ligated in pET-28b(+).

The pET28b-APH3′ and pET28b-ANT4′ expression vectors constructed

in this way were then transformed into E. coli BL21(DE3) competent

cells.

Selected clones for Enterococcus faecalis APH(3′) and Staphylo-

coccus aureus ANT(4′) were grown on 100 mL of LB medium

containing 26 µg/mL kanamycin at 37 °C. When the cell growth reached

an optical density at 600 nm (OD₆₀₀) of 0.5, the temperature was

switched to 30 °C and the culture was induced with 0.5 mm IPTG.

After 24 h of induction, the expression level was analyzed by SDS-

PAGE using gels with 13% polyacrylamide in the separation zone.

Proteins were purified employing a Ni²⁺-IDA-agarose column. The

enzymes after this step were found to be more than 90% pure as

assessed by SDS-polyacrylamide gel electrophoresis.

Acknowledgment. We thank Prof. J. Blanchard (Albert

Einstein College of Medicine, New York) for supplying the

plasmid encoding Mycobacterium tuberculosis AAC(2′). Fi-

nancial support from DGES (Grant CTQ2004-04494) is grate-

fully acknowledged. F.C. thanks Comunidad Auto´noma de la

Rioja for a postdoctoral fellowship. M.T. thanks the Ministerio

de Educacio´n y Ciencia for a predoctoral fellowship. The authors

thank CESGA (Santiago de Compostela) for computer support.

Supporting Information Available: Synthesis of derivatives

4 and 5. Supplementary Figures S1 to S7 showing NMR spectra

of aminoglycosides 2-6 and solvated MD simulations for

derivative 3. This material is available free of charge via the

Internet at http://pubs.acs.org.

Enzymatic Activity. Enzymatic reactions were performed in NMR

tubes. The nucleoside triphosphate (ATP 100 µM-3.0 mM) and the

aminoglycoside (100 µM-1.5 mM) were dissolved in 10 mM phosphate

buffer, pH 7.0, 2.5-5.0 mM MgCl₂. After addition of the enzyme (0.2-

JA0543144

(43) Waterworth, P. M. In Laboratory Methods in Antimicrobial Chemotherapy;

(42) Hegde, S. S.; Javid-Majd, F.; Blanchard, J. S. J. Biol. Chem. 2001, 276,

49, 45876-45881.

Garrod, L., Ed.; Churchill Livingstone Press: Edinburgh, 1978; pp 31-

40.

9

116 J. AM. CHEM. SOC. VOL. 128, NO. 1, 2006

Article Doi

Exploring the use of conformationally locked aminoglycosides as a new strategy to overcome bacterial resistance

DOI: 10.1021/ja0543144

Source and publish data:

Authors:

Article abstract of DOI:10.1021/ja0543144

Full text of DOI:10.1021/ja0543144

Products guided by the article

R&D Labs maybe for 875105-30-9

Relevant to this article

Hot Product