Binding of Neomycin-Class Aminoglycoside Antibiotics to Mutant Ribosomes with Alterations in the A Site of 16S rRNA

Aminoglycoside antibiotics that bind to the aminoacyl-tRNA site(A site) of the ribosome are composed of a common neamine corein which a glycopyranosyl ring is attached to position 4 ofa 2-deoxystreptamine moiety. The core is further substitutedby one (ribostamycin), two (neomycin and paromomycin), or three(lividomycin A) additional sugars attached to position 5 ofthe 2-deoxystreptamine. To study the role of rings III, IV,and V in aminoglycoside binding, we used isogenic Mycobacteriumsmegmatis ${Delta}$ rrnB mutants carrying homogeneous populations of mutantribosomes with alterations in the 16S rRNA A site. MICs weredetermined to investigate drug-ribosome interactions, and theresults were compared with that of the previously publishedcrystal structure of paromomycin bound to the ribosomal A site.Our analysis demonstrates that the stacking interaction betweenring I and G1491 is largely sequence independent, that ringsIII and IV each increase the strength of drug binding to theribosome, that ring IV of the 6'-NH₃⁺ aminoglycosides compensatesfor loss of interactions between ring II and U1495 and betweenring III and G1491, that the aminoglycosides rely on pseudo-basepairing between ring I and A1408 for binding independently ofthe number of sugar rings attached to the neamine core, thataddition of ring V to the 6'-OH 4,5-aminoglycoside paromomycindoes not alter the mode of binding, and that alteration of theU1406 · U1495 wobble base pair to the Watson-Crick interactionpair 1406C-1495G yields ribosomal drug susceptibilities to 4,5-aminoglycosidescomparable to those seen with the wild-type A site.

INTRODUCTION

Top

Introduction

Aminoglycosides form a large family of water-soluble, polycationicamino sugars which are used as broad-spectrum antibacterialagents (14, 15). Common to all aminoglycosides is the neaminecore. The neamine core is composed of a six-membered cyclitol(2-deoxystreptamine; ring II) glycosidically linked to a glucopyranosyl(ring I). Aminoglycosides target the ribosome by direct interactionwith rRNA and they affect protein synthesis by inducing codonmisreading and by inhibiting translocation of the tRNA-mRNAcomplex from the A to the P site (4, 15, 18, 28).

Helix 44 is located in the 1400 to 1500 region of 16S rRNA (Escherichiacoli numbering is used throughout this article) and has beenidentified as the site of codon-anticodon interaction (A site)by a number of studies (3, 17, 19, 32). Although A-site-boundtRNA and aminoglycosides protect the same nucleotides in thedecoding region, namely A1408, A1492, A1493, and G1494, aminoglycosidesdo not act by sterically hindering the binding of the tRNA tothe ribosome. Instead, they interfere with the decoding processby inducing conformational alterations of the ribosomal A site(2, 19). Upon drug binding, nucleotides A1492 and A1493 flipout from helix 44; this change in conformation is normally onlyobserved after the formation of a cognate tRNA-mRNA complex(2, 19, 20). Bulging out of A1492 and A1493 is most likely anessential step in aminoglycoside action, as it explains themiscoding properties of these drugs (4).

The neamine core of the aminoglycosides is essential for drugactivity (4, 6). Nuclear magnetic resonance studies have revealedthat rings I and II of the 4,5-aminoglycosides are sufficientto confer specific binding to a model A site RNA (6). A1408and base pairs C1409-G1491 and U1406 · U1495 representimportant elements of the drug binding pocket (Fig. 1). RingI forms two direct hydrogen bonds to the Watson-Crick side ofA1408 and is involved in a stacking interaction with G1491.Ring II participates in a network of hydrogen bond contactsof which the interactions between the 1-amino group and O-4of U1495 and between the 3-amino group and N-7 of G1494 aresequence specific.

View larger version (29K):
[in this window]
[in a new window]
FIG. 1. A: Superposition of neamine (gray ball and sticks) and ribostamycin (orange) molecules docked in the minimal A site of the ribostamycin crystal structure (orange). B: Superposition of the lividomycin A (magenta ball and sticks) and paromomycin (cyan ball and sticks) molecules docked in the minimal A site of the paromomycin crystal structure (cyan). C: Superposition of the neomycin (yellow ball and sticks) and paromomycin (cyan ball and sticks) molecules docked in the minimal A site of the paromomycin crystal structure (cyan). Data on crystal structures are from references 8 and 29.

The neamine core makes additional direct or water-bridged hydrogenbond contacts to the phosphate backbone of A1492 and A1493 (2,29, 30). Additional sugars (rings III, IV, and V) are attachedto position 5 of the 2-deoxystreptamine moiety. The contactsof ring III are characterized by two direct hydrogen bonds,with N-7 of G1491 and with N-4 of C1407, linking the two RNAstrands of helix 44 together. Ring IV is mainly involved innon-sequence-specific electrostatic interactions with the phosphatebackbone, but the exact positioning of this ring remains tobe established (2, 27). Recently, the structures of complexesbetween oligonucleotides containing the A site and neamine,neomycin, and lividomycin A (8) have been solved (Fig. 1).

RNA sequence determinants for aminoglycoside binding to theribosomal A site have mostly been defined using model RNA oligonucleotides(6, 7, 9, 13, 16, 26). Data on in vivo mutagenesis of the ribosomalA site are scarce and compromised by limitations in Escherichiacoli, which is frequently used as the model in ribosomal genetics.In particular, several rRNA nucleotides involved in formingthe drug binding site could not be subjected to in vivo mutagenesisdue to nonviability of the mutations in E. coli, e.g., U1495A,U1495C, G1491A, and C1409G (5, 25). These limitations were overcomewith the development of the Mycobacterium smegmatis ${Delta}$ rrnB system(21-24, 27). To analyze the contributions of rings III, IV,and V to the binding of 4,5-aminoglycosides, we extended ourexperiments using M. smegmatis ${Delta}$ rrnB strains carrying a largeset of mutations within the decoding site.

MATERIALS AND METHODS

Top

Materials and Methods

Bacterial strains and DNA techniques. The conditions for selective plating and RecA-mediated homologousrecombination to obtain mutant strains of M. smegmatis ${Delta}$ rrnB(for a list of mutants, see Table 1 and Fig. 2) were describedpreviously (23). After mutant selection the strains were culturedon LB agar plates. Nucleic acid sequencing was used to verifythat the point mutations had been introduced into the singlefunctional rRNA operon of M. smegmatis ${Delta}$ rrnB. Taq cycle sequencingwas done with fluorescently labeled nucleotides according tothe manufacturer's instructions (Applied Biosystems).

View this table:
[in this window]
[in a new window]
TABLE 1. Strains used in this study^a

View larger version (20K):
[in this window]
[in a new window]
FIG. 2. Secondary structure of the bacterial A site (E. coli numbering). The introduced mutations are indicated by the arrows, and the highly conserved A site is boxed. Reprinted from reference 12.

Determination of the MICs. Strains of M. smegmatis ${Delta}$ rrnB were cultured on LB agar plates.Single-colony cultures, grown in LB medium supplemented with0.05% Tween 80, were used for MIC tests in a microtiter plateformat; at least three independent clones were analyzed permutation. To 200 µl of starting cultures with an opticaldensity at 600 nm of 0.025, the antibiotics to be tested wereadded in twofold series of dilutions ranging from 0.25 to 1,024µg/ml. Paromomycin (P-9297), lividomycin (L-4518), neomycin(N-6386), and ribostamycin (R-2255) were obtained from Sigma;neamine (158966) was from ICN Biomedicals. The MIC is definedas the drug concentration at which the growth of the culturesis completely inhibited after incubation at 37°C. Cellswere incubated for a period corresponding to 24 generations;the wild type has a generation time of 3 h, and the generationtime of the mutants investigated was in the range of 3 to 6h (U1495A, 8 h).

RESULTS

Top

Results

Experimental system. The 4,5-disubstituted 2-deoxystreptamines investigated shareclose structural similarities. The common core of the 4,5-aminoglycosidesis composed of a 2-deoxystreptamine and a glycopyranosyl ring.The chemical group linked to the 6' position of ring I separatesthe 4,5-aminoglycosides into two subclasses: the 6'-amino subclass(neamine, ribostamycin, and neomycin) and the 6'-hydroxy subclass(paromomycin and lividomycin). The diversity within these twosubclasses is based on the number of sugar moieties attachedto position 5 of ring II: ribostamycin consists of neamine withone additional sugar attached to ring II; addition of a fourthring to ring III of ribostamycin gives rise to neomycin. Exceptfor the 6' position, paromomycin is identical to neomycin; anadditional fifth ring attached to paromomycin gives rise tolividomycin (for a list of drug structures, see Fig. 3 and 4).The number of charges and the nature of the substituent at position6' are identical for neamine and ribostamycin and for paromomycinand lividomycin; in both cases, however, the number of ringsvaries.

View larger version (29K):
[in this window]
[in a new window]
FIG. 3. Resistance profile of 4,5-disubstituted 2-deoxystreptamines with a 6'-NH₃⁺ group (neomycin subclass) represented as comparison graphs. Each axis of the comparison graph represents the influence of the corresponding introduced mutation (highlighted in bold). The relative resistances conferred by each of the mutations are plotted in log scale, with the smallest value in the innermost circle. Points lying on the outermost circle indicate that the corresponding mutations result in a relative resistance outside the measurable range. Neamine and ribostamycin bind weakly to wild-type ribosomes, and thus, quantitative analysis is limited. To reflect this limitation the values outside the measurable range are depicted by a hatched area.

View larger version (31K):
[in this window]
[in a new window]
FIG. 4. Resistance profile of 4,5-disubstituted 2-deoxystreptamines with a 6'-OH group (paromomycin subclass) represented as comparison graphs. Each axis of the comparison graph represents the influence of the corresponding introduced mutation (highlighted in bold). The relative resistances conferred by each of the mutations are plotted in log scale, with the smallest value in the innermost circle. Points lying on the outermost circle indicate that the corresponding mutations result in a relative resistance outside the measurable range.

Mutagenesis in M. smegmatis ${Delta}$ rrnB results in isogenic strainscarrying homogeneous populations of mutant ribosomes (for alist of mutant strains, see Table 1 and Fig. 2). MICs were determinedto investigate ribosomal drug susceptibility. Data interpretationfor neamine, ribostamycin, neomycin, paromomycin, and lividomycinwas based on available X-ray structural information (2, 8, 29).

The drug susceptibility measurements given in Table 2 (see alsocomparison graphs in Fig. 3 and 4) reveal that in wild-typeM. smegmatis ${Delta}$ rrnB the activity of the 4,5-aminoglycosides correlateswith the number of sugar rings attached to the neamine core.Addition of a third or fourth ring increases the activity ofthe drug by approximately eightfold each. However, the presenceof an additional fifth ring, as in lividomycin, does not markedlyincrease drug activity.

View this table:
[in this window]
[in a new window]
TABLE 2. Drug susceptibilities of rRNA mutants^a

The pseudo-base pair between ring I and A1408 was recently definedas a critical interaction for aminoglycoside binding (22, 29,31). Independent of the number of rings attached to the neaminecore, disruption of the pseudo-base pair by the A1408G alterationprevents the binding of all 4,5-aminoglycosides with a 6'-NH₃⁺group (see Table 2). In contrast, for the 6'-OH aminoglycosides(paromomycin and lividomycin) the hydroxyl group at position6 of ring I is likely to become an H-bond acceptor for N-1 orN-2 of A1408G, as discussed previously (22, 29, 31). Thus, thesedrugs can bind to the target site in presence of A1408G (seeTable 2).

Interactions with base pair C1409-G1491. The Watson-Crick base pair C1409-G1491 produces a seat in thepocket at the lower end of the drug binding site, enabling ringI to stack on G1491. In addition, N-7 of G1491 forms a hydrogenbond with the O-5" of ring III (2, 29, 31). The MIC data andthe calculation of relative resistances given in Table 2 demonstratethat, within the 6'-NH₃⁺ subclass, ribostamycin is always affectedthe most by alterations of C1409-G1491. While ribosomal susceptibilityto neamine and neomycin decreased only moderately upon mutationof the C1409-G1491 base pair (relative resistances between 1and 16 µg/ml), significant resistance is conferred toribostamycin (relative resistances between 8 and 128 µg/ml).In particular, transversion mutations (G1491 to C or U and C1409to G) have a profound effect on ribostamycin susceptibility.Compared to neomycin and as reported previously (21), the 6'-OHparomomycin is much more dependent on proper interaction withthe C1409-G1491 base pair. The importance of this interactionis not alleviated by addition of a further amino sugar, as inlividomycin (Table 2).

The double mutations A1408G/C1409G and A1408G/G1491U conferhigh-level drug resistance to all 4,5-aminoglycosides tested.This is to be expected for the 6'-NH₃⁺ subclass, due to disruptionof the pseudo-base pair interaction of ring I with A1408 (21,22). For the 6'-OH subclass this finding indicates that theeffects of different mutations are more than additive, in particularfor the combined A1408G/C1409G alteration, since the relativeresistance of the combined A1408G/C1409G alteration is significantlyhigher than that of the single mutations A1408G and C1409G alone(Table 2).

Interactions with base pair U1406 · U1495. The wobble base pair U1406 · U1495 forms the bindingpocket at the upper stem. One water-mediated contact betweenO-6 of ring II and O-4 of U1406 is observed in the 4,5-disubstituted2-deoxystreptamines (8, 29). Mutation of U1406 to adenine doesnot affect drug susceptibility; alteration of U1406 to cytosinemarginally influences drug susceptibility (relative resistancesof 4 to 16 µg/ml). The U1406C/A1408G double mutation resultsin high-level resistance to all 4,5-aminoglycosides studied,indicating that for the 6'-OH subclass the effects of the individualmutations are more than additive.

N-1 of ring II makes an additional direct hydrogen bond contactwith O-4 of U1495. The transition mutation U1495C exchangesthe O-4 with an ammonium group, rendering the interaction ofcytosine with N-1 of the 4,5-aminoglycosides unfavorable (electrostaticrepulsion). However, the correct insertion of ring I into thebinding pocket and the pseudo-base pair interaction betweenO-5' and N-6' of ring I with A1408 still allow binding of the6'-NH₃⁺ subclass to the ribosome. In contrast, the 6'-OH aminoglycosidesare more affected by the U1495C alteration. The U1495A alterationis expected to prevent contacts between N-1 of ring II and nucleotide1495 even more than a U-to-C mutation owing to the size of theadenine ring.

With the exception of neomycin, the U1495A alteration as wellas the double mutation U1406C/U1495A confer high-level drugresistance to all 4,5-aminoglycosides tested, including neamine,ribostamycin, paromomycin, and lividomycin. Apparently, theinteraction of the 6'-NH₃⁺ ring I with both A1408 and ring IVis required to partly compensate for this major conformationalchange in the ribosomal A site. Unexpectedly, introducing theU1495G alteration into the U1406C mutant, thus replacing theU · U wobble base pair with a 1406C-1495G Watson-Crickinteraction, nearly completely restores ribosomal susceptibilityto the 4,5-aminoglycoside antibiotics.

DISCUSSION

Top

Discussion

Using a well-defined in vivo system, we investigated the roleof rings III, IV, and V in the binding of 4,5-aminoglycosidesto their ribosomal target. Site-directed mutagenesis of theribosomal A site in a genetic model where manipulation of rRNAresults in cells carrying homogeneous populations of mutantribosomes (21-24) allowed a detailed investigation of drug-targetinteractions. The critical role of the 6' group of ring I inthe aminoglycoside-ribosome interaction has previously beeninvestigated by mutagenesis studies and will not be addressedhere (22). In this study we focused on the role of rings III,IV, and V in drug binding.

In line with previous measurements of dissociation constants(7) ring III increases drug potency by a factor of 8, as determinedby MIC measurements (Table 2). Ring III makes sequence-specificcontacts with RNA, i.e., hydrogen bonding between 5"-OH andN-7 of G1491 and between 2"-OH and N-4 of C1407 (2, 29). Additionof ring IV increases the positive net charge, and it contactsthe ribosomal A site at the backbone of helix 44 (2, 29). Thestrength of these electrostatic contacts raises drug activityeightfold as determined by MIC measurements of neomycin andribostamycin. Overall, in our in vivo system neomycin is 64-foldmore active than neamine. This finding is in good agreementwith the dissociation constants reported for neomycin and neamine(0.5 µM versus 100 µM) (7). Addition of ring V affectsthe antibacterial activity and susceptibility of mutant ribosomeslittle, if at all (compare lividomycin to paromomycin), indicatingthat ring V makes only weak contacts with the A site in a non-sequence-specificmanner, most likely with the RNA backbone, as seen in the crystalstructure (8).

For each pair of aminoglycosides with similar properties ofoverall charge and substituent at the 6' position, one can dividethe effect of mutations into two categories: mutations for whichthe ratio of relative resistance is identical and those forwhich this ratio is different. In the neamine-ribostamycin pair,for instance, the C1409G mutation (and to a lesser extent G1491Aor G1491C) confers higher relative resistance to ribostamycinthan to neamine.

To investigate the ring I-G1491 interaction isolated from otherinteractions by additional rings, e.g., between 5"-OH of ringIII and N-7 of G1491, we chose a reductionist approach, i.e.,we determined MICs to neamine in a series of mutants with alterationsin the C1409-G1491 base pair. These studies revealed that transitionsaffect drug binding weakly; transversions have a small albeitsignificant effect. We interpret these findings as evidencethat the ring I-G1491 stacking interaction is largely sequenceindependent. Within the 6'-NH₃⁺ subclass, ribostamycin is mostaffected by alterations of the C1409-G1491 base pair. This findingis rationalized by the X-ray structural data available for paromomycin(2, 27). In contrast to ribostamycin, with its ring III-G1491interaction, neamine interacts with G1491 merely by non-sequence-specificstacking interaction with its ring I. Compared to that of ribostamycin,ring IV of neomycin apparently is able to compensate for theloss of sequence-specific interaction between ring III and G1491and/or introduction of steric clashes as mediated by mutagenesisof C1409-G1491.

A thorough interpretation of the MIC measurements obtained uponmutagenesis of the U1406 · U1495 wobble base pair isin part limited by the low susceptibility of wild-type ribosomesto neamine. Most of the mutations, i.e., U1406C, U1406A, andU1495C, only moderately affect ribosomal susceptibility to the6'-NH₃⁺ aminoglycosides, if at all. The U1495A alteration conferssignificant resistance to ribostamycin (relative resistance,>128 µg/ml) but only moderately affects susceptibilityto neomycin (relative resistance, 16 µg/ml).

The loss of ribosomal susceptibility to neamine and ribostamycinin the U1495A mutant can be rationalized by the loss of interactionbetween the 1-amino group of ring II and the O-4 of U1495 inthe U1495A mutant and by less favorable binding of ring II.Apparently, ring IV of neomycin is able to partly compensatefor this interaction. The U1406C · U1495A alterationconfers high-level drug resistance to all 4,5-aminoglycosidesexcept neomycin, which is only moderately affected. As discussedpreviously (22), the U1406C · U1495A alteration willresult in a major change in base pair geometry precluding anappropriate positioning of ring II near the 1406-1495 pair andconsequently poor insertion of ring I in the drug binding pocket.

Alteration of U1495A to G in the U1406C · U1495A doublemutant replaces the natural U1406 · U1495 wobble basepair with a U1406C-U1495G Watson-Crick base pair and restoresribosomal susceptibility to all 4,5-aminoglycosides of the 6'-NH₃⁺and the 6'-OH subclasses investigated. This observation contrastswith the U1406-U1495A alteration, which confers significantresistance to most 4,5-aminoglycosides. Presumably, the U1406C-U1495Galteration has a conformation and geometry similar to that ofthe U1406-U1495A alteration. How then to explain this differencein drug susceptibility?

One rationale would refer to the different net charge providedby the functional group at position C-6 of U1495A (amino group,positive charge) and U1495G (keto group, negative charge). Theketo group at C-6 of 1495G, together with the N-7 of 1495G,could possibly interact with the ring II N-1 ammonium groupin a fashion similar to that of O-4 of the pyrimidine U1495or, more generally, be involved in a network of electrostaticinteractions between the positively charged aminoglycosidesand their target RNAs (10, 11). The Hoogsteen sites of guaninesare often involved in binding positive ions such as potassiumions (1).

Figure 5 shows the superposition of a 1406C-1495G base pairon the U1406 · U1495 pair. The distances between a typicalaminoglycoside (N-1 and O-6 of ring II) and the Hoogsteen sitesof 1495G are 3.0 Å (O-6) and 4.2 Å (N-7), respectively.It can also be seen how the amino group N-4 of 1406C may replacethe observed water molecule between U1406 and the aminoglycosidesuch that a contact between N-4 of 1406C and O-6 of ring IIcan be made. Interestingly, mutants in which a single uracilis replaced with an adenine display disparate effects. Thus,the U1406-U1495A mutation leads to resistance to aminoglycosides,while the U1406A-U1495 mutant shows susceptibility to the neomycinclass of aminoglycosides but less susceptibility to the kanamycinclass (20).

View larger version (36K):
[in this window]
[in a new window]
FIG. 5. Manual superposition of a 1495G-1406C interaction on a U1495 · U1406 base pair in the crystal structure of paromomycin (29). The N-1 and O-6 atoms of ring II which contact positions 1495 and 1406 are indicated, as well as the N-7 and O-6 of G1495 and N-4 of C1406. Shown in gray is U1495 · U1406 with water molecules bound.

Our data indicate that binding of 4,5-disubstituted aminoglycosidesto the ribosomal A site depends on proper interaction of ringI with nucleotides A1408 and G1491 and is further stabilizedby additional contacts, e.g., with U1495. Taken together, thesedata on site-directed mutagenesis of the drug target site providea detailed analysis of the importance of the different bindingcontacts in the drug-ribosome interaction.

ACKNOWLEDGMENTS

This study was supported by grants from the Swiss National ScienceFoundation to E.C.B. (3200-BO-100780, NRP 49). B.F. is supportedby a grant from the European Commission (QLRT-2001-000892);E.W. acknowledges support from the Institut Universitaire deFrance.

We thank T. Janu $s$ i $c$ for expert technical assistance, A. Vasellafor helpful discussions, and A. Makovec for typing the manuscript.

FOOTNOTES

* Corresponding author. Mailing address: Institut für Medizinische Mikrobiologie, Universität Zürich, Gloriastr. 30/32, CH-8006 Zürich, Switzerland. Phone: 41-44-634 26 64. Fax: 41-44-634 49 06. E-mail: shobbie{at}immv.unizh.ch.

REFERENCES

Top