Phylogenetic sequence variations in the 23S rRNA peptidyltransferase region.
Most rRNA nucleotides in the inner peptidyltransferase region of the large ribosomal subunit are phylogenetically conserved (8
). The macrolide/ketolide family of antibiotics binds to a hydrophobic cleft formed by residues 2058, 2059, and 2611 (E. coli
numbering is used throughout) in domain V of 23S rRNA, with some drugs making additional contacts in domain II (9
) (). The adenines at 23S rRNA positions 2058 and 2059 are phylogenetically conserved in bacteria and play an important role in compound binding, emergence of resistance, and drug selectivity (1
Fig. 1. (a) Structures of erythromycin (blue) (PDB accession code 3OHJ) and telithromycin (pink) (PDB accession code 3OI3) bound to the Thermus thermophilus ribosome. Nucleotides investigated in this study are indicated in black. (b) Secondary structure of domain (more ...)
One key interaction appears to be the hydrogen bond between N-1 of A2058 and the 2′-hydroxyl group of the macrolides' desosamine sugar () (30
). Mutation of A2058 to a guanine (A2058G) greatly impairs the binding of macrolides to ribosomes by both chemical and steric alteration of the binding site (9
). Ribosomal susceptibility to macrolides and ketolides is also affected by proper Watson-Crick base pairing between nucleotides at positions 2057 and 2611 (6
), which are typically G · C (e.g., in Proteobacteria
) and A · U (e.g., in Mycobacteria
) (; for a phylogenetic comparison of the 2057 · 2611 interaction in clinically relevant phyla, see Table S2 in the supplemental material). The composition of the base pair between nucleotides 2057 and 2611 has been shown to affect the resistance phenotype of the A2058G mutation toward ketolides (19
). However, the effect of an A2057G-U2611C substitution in the context of a wild-type A2058 has remained elusive. Changing the A2057 · U2611 base pair in M. smegmatis
to G2057 · C2611, as it is typically found in Proteobacteria
, had no effect on susceptibility to any of the macrolides or ketolides tested (). Similarly, the composition of this base pair had no effect on resistance to erythromycin, clarithromycin, azithromycin, spiramycin, tylosin, and josamycin as conferred by the A2058G mutation. However, we confirmed that the ketolide resistance phenotype of A2058G is indeed dependent on the nature of the 2057 · 2611 base pair, as the A2058G mutant is 16-fold more susceptible to telithromycin in the context of a proteobacterial G2057 · C2611 sequence than in the context of a mycobacterial A2057 · U2611 ().
MICs of various macrolides/ketolides in M. smegmatis 23S rRNA variants
Phylogenetic sequence variations in 16S rRNA helix 44.
Nucleotides of 16S rRNA helix 44 are part of the aminoacyl-tRNA acceptor site (A site) and are highly conserved () (7
). Aminoglycoside antibiotics bind to the A site by direct contacts to helix 44 (2
). While aminoglycosides form a number of hydrogen bonds with different nucleotides in helix 44, their interactions with rRNA residues 1408, 1409, and 1491 (E. coli
numbering) appear to be most critical for drug binding (11
Fig. 2. (a) Secondary structure of 16S rRNA helix 44 decoding site and sequence conservation in eubacteria. rRNA residues are numbered according to their homologous positions in E. coli 16S rRNA. Phylogenetic sequence variations analyzed in this study are highlighted (more ...)
16S rRNA residue 1408 is an adenine in all wild-type bacteria. Among all A-site mutations that confer aminoglycoside resistance, the 1408 adenine-to-guanine mutation (A1408G) is the predominant alteration in clinical drug-resistant strains (24
). This transition mutation alone is sufficient to confer high-level resistance to 6′-NH2
aminoglycosides by disrupting the interaction between A1408 and the compound's ring 1 amino sugar (21
). It is also thought to function as the main specificity determinant of aminoglycosides, because the cytoplasmic ribosomes of higher eukaryotes carry a guanine at this position (1
Bacterial A-site sequence variations within the aminoglycoside binding pocket involve base pair interactions 1409 · 1491, 1410 · 1490, and 1411 · 1489 (). The 1410 · 1490 pair is always a purine/pyrimidine interaction, either G · C (e.g., in Actinomycetales
) or A · U (e.g., in Proteobacteria
). The 1411 · 1489 interaction involves a pyrimidine/purine interaction, either U · A (e.g., in Actinomycetales
) or C · G (e.g., in Proteobacteria
). The 1409 · 1491 interaction involves a purine-pyrimidine switch: C · G (a pyrimidine/purine interaction) or A · U (a purine/pyrimidine interaction) (7
). The majority of eubacteria are characterized by a 1409 pyrimidine · 1491 purine (C · G) interaction, while Propionibacteria
carry a 1409 purine · 1491 pyrimidine (A · U) base pair (; for a phylogenetic comparison with clinically relevant phyla, see Table S2 in the supplemental material).
Bacterial sequence polymorphism of residues 1410 · 1490 and 1411 · 1489 was found not to affect the susceptibility of the wild-type drug binding pocket to aminoglycoside antibiotics, as the corresponding drug binding sites are highly susceptible to these compounds (, compare, e.g., Mycobacteria with Proteobacteria). Likewise, these sequence polymorphisms do not affect the resistance phenotype associated with the predominant mutational resistance alteration A1408G. Independent of the bacterial sequence polymorphism involving residues 1410 · 1490 and 1411 · 1489, an A1408G mutation results in high-level resistance to aminoglycosides with a 6′-NH2 group, such as gentamicin and neomycin, but only limited resistance to aminoglycosides with a 6′-OH group, such as paromomycin ().
MICs of various aminoglycosides in M. smegmatis 16S rRNA variants
The sequence polymorphism at residues 1409 · 1491, which involves a purine/pyrimidine switch, was found to significantly affect the aminoglycoside susceptibility of the wild-type drug binding pocket. Replacing the C · G base pair in M. smegmatis
with a propionibacterial A · U considerably reduced susceptibility to all 4,5- and 4,6-aminoglycosides tested (). This observation is in agreement with the lower aminoglycoside susceptibility of Propionibacteria
reported previously (16
). Amikacin is the least affected among the aminoglycosides tested, presumably because its l
-haba group interacts with additional nucleotides within helix 44 and stabilizes drug binding (17
To study the contribution of a 1409 · 1491 purine/pyrimidine base pair switch to aminoglycoside susceptibility in more detail, we investigated base pair interactions U · A and G · C. Base pair U1409 · A1491 represents a pyrimidine/purine interaction similar to the C · G found in the majority of eubacteria. Introduction of the U · A base pair decreased susceptibility to paromomycin, a 4,5-substituted aminoglycoside with a 6′-OH group, while it had little if any effect on aminoglycosides with a 6′-NH2
group (). Introduction of a G1409 · C1491 purine/pyrimidine interaction similar to the A · U found in Propionibacteria
resulted in minor but significant changes in susceptibility to both 4,5- and 4,6-aminoglycosides (with the exception of amikacin), with the 4,5-compounds being more affected (). From these data we infer that in the presence of an A1408 there is a gradient of drug susceptibility for the 1409 · 1491 interaction, indicating that both the purine/pyrimidine interaction and the specific nucleotide are relevant. In line with previous investigations and the different orientation of the aminoglycosides' sugars linked to position 5 or 6 of the neamine core (20
) (), our results suggest that, in general, the 4,5-substituted compounds and in particular the 6′-OH paromomycin are more dependent on a proper 1409 · 1491 interaction than the 4,6-substituted compounds.
Combining the propionibacterial A1409 · U1491 pair further with an A1408G alteration resulted in high-level resistance to all aminoglycosides, including paromomycin (). Typically, binding of paromomycin, which carries a hydroxyl group at the 6′ position of ring I, is only moderately affected by the A1408G mutation, since it can accept a hydrogen bond from the N-1 and N-2 of G1408 (21
). Apparently, the high-level resistance to paromomycin is the result of a combined effect of perturbing contacts to both G1491 and A1408, which would be in agreement with previous data demonstrating that alteration of residue 1491 primarily increased resistance toward 6′-OH aminoglycosides such as paromomycin (20
Given that stacking of aminoglycoside ring I on G1491 is important for binding (), we wished to study whether the bacterial 1410 · 1490 and 1411 · 1489 sequence variations affect the drug susceptibility pattern associated with C1409 · G1491 base pair disruptions. Accordingly, we replaced G1491 with A, C, or U in isogenic hybrid strains carrying the mycobacterial and proteobacterial A-site sequence. Drug susceptibility testing of the recombinant mutants revealed that the context of a mycobacterial or proteobacterial A site does not affect the specific drug resistance pattern associated with distinct alterations of residue 1491 () (see reference 11
for a structural discussion of resistance patterns conferred by mutational alteration of G1491).
MICs of various aminoglycosides in M. smegmatis 16S rRNA variants with disruption of 1409 · 1491 base pairing
Previously, limitations in genetic manipulation did not allow study of the effect of bacterial A-site polymorphism on aminoglycoside susceptibility in isogenic mutants. Rather, investigations were limited to testing different bacterial species representative of the corresponding sequence polymorphism. In these early studies it was concluded that the C1409 · G1491/A1409 · U1491 polymorphism is not associated with resistance (21
). Using more recently developed genetic techniques, we have now been able to refine this statement and to define the role of the 1409 · 1491 base-pairing polymorphism in aminoglycoside susceptibility more precisely. In addition, we have recently observed that the 16S rRNA 1410 · 1490 interaction has a subtle influence on aminoglycoside susceptibility in ribosomes with a non-Watson-Crick 1409 · 1491 interaction, e.g., C1409 · C1491 (12
). We extend these findings in our current study, which reveals that in the presence of a 1409 · 1491 base pair interaction, bacterial 1410 · 1490 sequence polymorphisms do not measurably affect aminoglycoside susceptibility.