Antibiotic Resistance Mechanisms, with an Emphasis on Those Related to the Ribosome

KATHERINE S. LONG¹ AND BIRTE VESTER²*

[SECTION EDITOR: MICHAEL O’CONNOR]

Posted August 26, 2008

Department of Biology, University of Copenhagen, Copenhagen Biocenter, 3-1-31, Ole Maaløes Vej 5, DK-2200 Copenhagen N,¹ andDepartment of Biochemistry and Molecular Biology, University of Southern Denmark, Campusvej 55, DK-5230 Odense M,² Denmark

*Corresponding author. Mailing address: Department of Biochemistry and Molecular Biology, University of Southern Denmark, Campusvej 55, DK-5230 Odense M, Denmark. Phone: 45 6550 2377, Fax: 45 65502467, E-mail: This e-mail address is being protected from spambots. You need JavaScript enabled to view it .

Antibiotic resistance is a fundamental aspect of microbiology and a subject of basic research, but it is also a phenomenon of vital importance in the treatment of diseases caused by pathogenic microorganisms. An increase in the appearance of bacterial resistance to many antibiotics underscores the importance of elucidating and understanding the mechanisms by which antibiotic resistance occurs. A resistance mechanism can involve an inherent trait or the acquisition of a new characteristic through either mutation or horizontal gene transfer. Antibiotic resistance is a relative concept, and the natural susceptibilities of bacteria to a certain drug vary significantly from one species of bacteria to another and even from one strain to another. In general, the gram-negative bacteria are less susceptible to antibiotics than gram-positive bacteria due to differences in their membranes. Once inside the cell, most antibiotics affect all bacteria similarly.

Some antibiotic resistance mechanisms originate from antibiotic producer organisms that need to protect themselves, whereas others are due to natural random mutations. The latter can result in the perturbation of an antibiotic binding site or the modulation of the expression of a resistance determinant. Moreover, mutations may alter binding specificities and, in this manner, confer a degree of antibiotic resistance. Antibiotic resistance is an ancient phenomenon, and the conditions under which it spreads from antibiotic producers to other organisms are not clearly established. However, it is evident that the use and misuse of antibiotics in medicine and agriculture have had a significant impact on the incidence, selection, and spread of antibiotic resistance in pathogenic bacteria.

Although some Escherichia coli strains are pathogenic, the importance of E. coli in relation to the study of antibiotic resistance is as a model system with well-established genetic and biochemical tools for the investigation and elucidation of antibiotic resistance mechanisms. The field of antibiotic resistance mechanisms is large, and the literature is extensive. This chapter will briefly cover the current state of the whole field and highlight new insights on selected subjects, in particular, ribosomal resistance mechanisms. We refer to other reviews for more details on some subjects in order to cover the field broadly and to recent papers to cover more specific subjects.

MECHANISMS OF ANTIBIOTIC RESISTANCE

Bacteria have evolved numerous strategies to evade the cytotoxic effects of antibiotics, of which the vast majority can be grouped into the following three general mechanisms (Fig. 1): (i) increased antibiotic efflux out of the cell or reduced antibiotic influx into the cell, (ii) enzymatic inactivation of antibiotics through drug modification or cleavage, and (iii) the alteration of the antibiotic binding site. The third mechanism is dependent on the antibiotic target in the cell, whereas the other two are target site independent. The alteration of either the antibiotic or its binding site results in reduced affinity between the antibiotic and its cellular target. Increased efflux and reduced influx lead to a lower intracellular antibiotic concentration without drug alteration.

Fig. 1Cartoon illustrating the principles of the three main resistance mechanisms in bacteria. The cell membrane is depicted in grey, ribosomes are depicted in green, mRNA is depicted in red, and a fictitious antibiotic that targets ribosomes is shown in blue. (i) The antibiotic is pumped out of the cell, lowering the intracellular antibiotic concentration and thereby antibiotic binding to ribosomes. (ii) The antibiotic is inactivated so that it cannot bind to its target and inhibit protein synthesis. (iii) The drug binding sites on the ribosomes (illustrated with white diamonds) are changed so that the drug does not bind and therefore does not inhibit protein synthesis.

A bacterial cell contains many different target sites for antibiotics, as reviewed in reference 94. In general, antibiotic binding to a target interferes with or inhibits an essential cellular pathway or process. The most common antibiotic targets (with examples of the targeting antibiotics given in parentheses) include bacterial cell wall synthesis (β-lactams, glycopeptides, and fosfomycin); the bacterial membrane (daptomycin); the DNA replication machinery (fluoroquinolones and novobiocin); RNA polymerase (rifampin); the protein synthesis machinery, including ribosomes, translation factors, and tRNAs (tetracyclines, aminoglycosides, macrolides, lincosamides, chloramphenicol, oxazolidinones, fusidic acid, and mupirocin); and the folate biosynthesis pathway (trimethoprim and sulfonamides). As the ribosome is the cellular target for numerous antibiotics, it will be treated separately in a later section.

Efflux is the active export of a solute out of a cell. Efflux pumps are found in all organisms and are believed to have the physiological role of pumping toxic substances outside of the cell. In bacteria, genes encoding efflux pumps are found both on the chromosome and on mobile elements such as plasmids. Efflux pumps can be specific for a particular drug or drug class or, alternatively, be capable of transporting multiple chemically dissimilar compounds out of the cell. The substrates of efflux pumps are not limited to antibiotics but also include bile salts, biocides (disinfectants), detergents, dyes, fatty acids, heavy metals, and organic solvents. Antibiotic resistance in an efflux mutant is usually conferred by mutations that either increase the expression of efflux pump proteins or make efflux pumps more efficient.

Reduced drug influx is due to decreases in cell permeability. In gram-negative bacteria, antibiotic resistance has been associated with alterations in the outer membrane (45). One group of alterations involves mutation or modification of the lipopolysaccharide that renders the outer membrane impermeable to hydrophobic molecules and antibiotics. Other mechanisms have been associated with changes in the expression, structures, and selectivity patterns of the outer membrane porin proteins (17). These span the outer membrane and form water-filled channels through which hydrophilic molecules can cross via diffusion.

Antibiotics can be covalently modified or cleaved so that their binding to cellular target sites is either eliminated or markedly reduced, as reviewed in reference 116. These reactions are catalyzed by a diverse group of enzymes and involve a number of chemical strategies to inactivate antibiotics, including hydrolysis and group transfers. Antibiotics inactivated by hydrolysis typically contain a labile ester or amide bond that is susceptible to attack by a water nucleophile. An example is the β-lactamases (amidases) that cleave the β-lactam rings of the penicillin and cephalosporin classes of antibiotics. The largest and most diverse group of enzymes catalyze group transfers that covalently modify antibiotics through acetylation, phosphorylation, nucleotidylation, ribosylation, glycosylation, or thiol transfer reactions. These reactions require donor substrates such as acetylcoenzyme A, ATP, NAD, UDP-glucose, and glutathione. The aminoglycosides are an illustrative example of a class of antibiotics that can be inactivated via acetylation, phosphorylation, and nucleotidylation in reactions catalyzed by aminoglycoside acetyltransferases, kinases, and nucleotidyltransferases, respectively. All these reactions modify amino or hydroxyl groups on the aminoglycoside structure and inactivate the drugs by blocking functional groups important for interaction with 16S rRNA at the ribosomal decoding center. The enzymes are classified according to the position of group transfer on the aminoglycoside structure and are most often encoded on extrachromosomal elements such as plasmids, transposons, and integrons.

Although the majority of characterized enzymes catalyze hydrolysis or group transfer reactions to inactivate the antibiotic substrates, a few enzymes use alternative strategies, such as redox mechanisms and other carbon bond cleavage reactions. An example of a redox enzyme is TetX, which catalyzes the oxidation of the tetracycline antibiotics. TetX is a flavin-dependent monooxygenase that catalyzes the hydroxylation of tetracycline in the presence of oxygen and NADPH (119). The hydroxylation disrupts coordination to a magnesium ion bound to 16S rRNA, which is a critical feature of the tetracycline binding site on the ribosome.

The strategies evolved by bacteria to evade a particular antibiotic can include one, two, or more types of the general resistance mechanisms described above. These resistance mechanisms can work together to produce higher levels of antibiotic resistance. An example is provided by tetracycline, to which resistance can be conferred through enzymatic inactivation by TetX, efflux, or ribosomal protection. Tetracycline efflux genes have been found in both gram-positive and gram-negative bacteria (reviewed in reference 13). On the order of 20 genes that encode membrane-bound efflux proteins belonging to the major facilitator superfamily have been described. Ribosomal protection proteins are soluble cytoplasmic proteins that mediate tetracycline resistance, as reviewed in reference 14. They have sequence similarity to ribosomal elongation factors and are part of the translation factor superfamily of GTPases. The best-studied ribosomal protection proteins, Tet(O) and Tet(M), have ribosome-dependent GTPase activity and can release tetracycline from its binding site on the ribosome.

Another example of an antibiotic for which a variety of resistance mechanisms exists is erythromycin, to which resistance can be conferred through the alteration of the drug binding site, drug efflux, or enzymatic inactivation. The drug binding site can be altered through the mutation of rRNA and ribosomal proteins L4 and L22, as reviewed in reference 105, and by the methylation of rRNA by Erm methyltransferases (83). Erythromycin can be pumped out of the cell through the expression of mef and msr genes, encoding proteins of the major facilitator superfamily and the ATP binding cassette family of efflux pumps, respectively (http://faculty.washington.edu/marilynr/). Finally, erythromycin can be inactivated either by esterase enzymes encoded by the ere genes or, alternatively, by macrolide kinases encoded by the mph genes that phosphorylate the desosamine sugar that interacts with 23S rRNA (http://faculty.washington.edu/marilynr/).

EFFLUX

Bacterial efflux systems are composed of either single or multiple components. Single-component pumps are membrane-bound efflux proteins that transport substrates across the bacterial cytoplasmic membrane. Efflux pumps with multiple components are found in gram-negative bacteria and transport substrates across the entire cell envelope. They are composed of an inner membrane efflux protein, a periplasmic accessory (membrane fusion) protein, and an outer membrane protein (Fig. 2). Efflux proteins can be classified into the following five families: the major facilitator superfamily, the ATP binding cassette family, the small multidrug resistance family, the resistance-nodulation-division (RND) family, and the multidrug and toxic compound extrusion family. The efflux of substrates from the cell is an energy-dependent process that is either coupled to ion transport or, in the case of the ATP binding cassette family, driven by ATP hydrolysis. Intrinsic resistance to some antimicrobials can be conferred by basal levels of efflux, particularly in gram-negative bacteria (73). However, the overexpression of efflux proteins can cause resistance in bacteria that are normally susceptible to a particular antimicrobial agent. Efflux mutants often have multiple drug resistance phenotypes, with decreased susceptibilities to at least three classes of antimicrobial compounds. MICs of efflux pump substrates for mutants that overexpress an efflux pump typically increase two- to eightfold compared with the MICs for an isogenic parent strain (74). Although these increases are modest compared with those associated with other resistance mechanisms, they are nevertheless sufficient to render a strain resistant relative to breakpoint concentrations and can thus have clinical significance. As the expression of some efflux pumps can be induced by their substrates and many efflux pumps typically have broad substrate ranges, bacteria overexpressing these pumps can be selected by a variety of agents. In addition, the overexpression of efflux pumps can increase the survival times of bacteria in the host and lead to the emergence of highly resistant mutants. Indeed, accumulating evidence suggests that efflux pumps play a role in bacterial pathogenicity, as reviewed in reference 74.

Fig. 2Schematic drawing of a multiple-component efflux system found in gram-negative bacteria that pumps antibiotics and other substances out of the cell. The efflux system consists of an outer membrane protein (TolC or OprM), an efflux protein (AcrB or MexB), and membrane fusion proteins (AcrA and MexA).

In gram-negative bacteria, the efflux pumps that are most often associated with clinically relevant antibiotic resistance are multiple-component efflux systems containing RND family efflux proteins, as reviewed in references 45 and 73. The most well-characterized of these are the AcrAB-TolC (E. coli) and MexAB-OprM (Pseudomonas aeruginosa) systems, in which AcrB and MexB are efflux proteins, AcrA and MexA are membrane fusion proteins, and TolC and OprM are outer membrane proteins. The detailed architecture of these efflux systems has emerged from high-resolution structures of the individual components resolved by X-ray crystallography, as reviewed in reference 52. The outer membrane proteins TolC and OprM are structurally conserved, despite having little sequence similarity (1, 44). The proteins form trimers organized into two barrel structures, a 40-Å-long β-barrel that inserts into the outer membrane and a connecting 100-Å-long α-barrel that extends into the periplasmic space and docks with the corresponding efflux proteins (AcrB and MexB). The AcrB protein is also a trimer that contains a transmembrane domain with 12 transmembrane helices bound in the inner membrane and a 70-Å periplasmic domain that may be further subdivided into porter and TolC docking domains. The adjacent porter domains in the trimer form openings that lead into a large central cavity. The accessory proteins AcrA and MexA adopt elongated sickle-shaped structures that allow them to interact with both the efflux and outer membrane proteins (2, 34, 62).

Recent high-resolution crystal structures of efflux protein forms including AcrB and AcrB-drug complexes have provided important insights into the mechanism of drug efflux and evidence to suggest that RND efflux pumps may capture substrates in the periplasm (52, 63, 92). A notable aspect of the structures with and without bound substrates is that AcrB adopts an asymmetric trimer structure, in which the porter domains of each monomer in the trimer adopt distinct conformations. The different conformations are believed to represent consecutive states in the drug efflux cycle and have been named tight, open, and loose, corresponding to drug binding, extrusion, and access, respectively. An ordered binding change mechanism of the trimer has been proposed, analogous to that of the F₁F₀-ATPase of oxidative phosphorylation, in which each monomer cycles sequentially through the three conformations (63). Complexes with doxorubicin and minocycline show that the drugs interact with a site in the porter domain of the monomer that is in the tight conformation. The site is exceptionally rich in aromatic amino acid residues, and the drugs interact with different sets of amino acid residues, consistent with the concept of a multispecific binding site (63).

RIBOSOMAL ANTIBIOTIC RESISTANCE

The ribosome is a major site of antibiotic action in the bacterial cell and is targeted by a large and chemically diverse group of antibiotics, as reviewed in reference 76. A number of these antibiotics have important applications in human and veterinary medicine in the treatment of bacterial infections. The antibiotic binding sites are clustered at functional centers of the ribosome, such as the decoding center on the 30S subunit; the peptidyl transferase center (PTC), the GTPase center, and the peptide exit tunnel on the 50S subunit; and the subunit interface spanning both subunits on the 70S ribosome (Fig. 3). Upon binding, the drugs interfere with the positioning and movement of substrates, products, and ribosomal components that are essential for protein synthesis. Ribosomal antibiotic resistance is due to the alteration of the antibiotic binding sites through either mutation or methylation.

Fig. 3Models of the ribosomal subunits and the assembled ribosome. In the panels depicting the individual subunits and the assembled ribosome, proteins are shown in yellow and cobalt blue, and RNA is shown in orange and light blue for the 30S and 50S subunits, respectively. The antibiotic binding sites mentioned in the text in relation to resistance determinants are indicated with arrows. The coordinates for the whole subunits (E.coli) are from reference 90; the coordinates for the cut view (T.thermophilus) are from reference 93. The models were kindly provided by Jacob Poehlsgaard using VMD (37).

As in other areas of molecular biology, E. coli has been the experimental system of choice in ribosome studies for decades. For no other organism is our knowledge of the ribosome more complete. The large amount of collected genetic and biochemical information on E. coli ribosomes, particularly in the forms of antibiotic footprinting and mutational and posttranscriptional modification data, has facilitated the study of ribosome-related resistance mechanisms. Since 2000, high-resolution structures of the Thermus thermophilus, Deinococcus radiodurans, and Haloarcula marismortui ribosomal subunits bound to antibiotics have made significant contributions to the identification of drug binding sites and the understanding of antibiotic resistance, as reviewed in reference 76. These developments were followed in 2005 by the 3.5-Å resolution structure of the E. coli 70S ribosome (90), making direct correlations between functional data and a high-resolution model possible for the first time. The rest of this chapter will be centered on ribosome-based mechanisms of antibiotic resistance and will focus especially on new developments and aspects that have not been extensively reviewed previously.

THE DECODING CENTER ON THE 30S RIBOSOMAL SUBUNIT

A primary function of the 30S subunit is decoding, the process in which cognate tRNAs are selected through the monitoring of codon-anticodon base-pairing interactions in the A site, as reviewed in reference 82. The selection of tRNAs by the ribosome requires a conformational switch from an open (hyperaccurate) form to a closed (error-prone) form (69). The antibiotics that target the 30S subunit bind almost exclusively to 16S rRNA nucleotides that either are adjacent to mRNA and tRNA binding sites or undergo critical structural rearrangements during decoding. A subset of the antibiotics induce miscoding by stabilizing the closed conformation and thereby perturbing the equilibrium between the open and closed forms of the 30S subunit (69). The mechanisms of resistance against the antibiotics involve RNA and protein mutations, as well as RNA methylations.

The RNA mutations associated with antibiotic resistance are located primarily in the 530 loop and helices 31, 34, and 44 of 16S rRNA. The crystal structures of spectinomycin, streptomycin, paromomycin, tetracycline, pactamycin, hygromycin B, tobramycin, geneticin, and kasugamycin bound to either 30S subunits (9, 12, 75, 87), decoding A-site oligonucleotide fragments (106), or the 70S ribosome (91) have provided a wealth of detail on antibiotic-RNA interactions. They have greatly facilitated the understanding of ribosomal resistance mechanisms, and there is very good agreement between the mutational and structural data, as the mutated nucleotides are generally positioned at or abutting the drug binding cavity.

The typical aminoglycosides include the 4,5- and 4,6-disubstituted deoxystreptamine antibiotics (the former group includes paromomycin, whereas the latter includes tobramycin and geneticin), and they target helix 44 of 16S rRNA. Thus, RNA mutations conferring resistance to these drugs are clustered here and are localized at nucleotide positions 1406, 1408 and 1409, 1491, 1495 and 1496, and 1498 of 16S rRNA (reference 35 and references therein). These resistance mutations are clinically significant in pathogenic Mycobacterium spp. strains that contain a single rRNA operon. A genetic system using a single rRNA allelic derivative of the nonpathogenic species Mycobacterium smegmatis has been used in mutagenesis studies to investigate strains with homogeneous populations of mutant ribosomes, as reviewed in reference 35. In particular, this approach has been used in an exhaustive mutagenesis study of nucleotides in helix 44 to probe individual drug-nucleotide contacts (35). Other aminoglycoside antibiotics that target the 30S subunit but lack a neamine core include streptomycin, spectinomycin, kasugamycin, and hygromycin B. Mutations at 16S rRNA positions 13, 505, 507, 522 to 526, and 912 to 915 confer resistance to streptomycin (references 66 and 98 and references therein). Spectinomycin resistance is conferred by 16S rRNA mutations in helix 34 at positions 1064, 1066, and 1191 to 1193 (5, 8, 55, 68). Kasugamycin resistance is conferred by 16S rRNA mutations at positions 794, 926, and 1519 (108). Although the vast majority of tetracycline resistance mechanisms are mediated by efflux or ribosomal protection proteins, mutations at positions 965 to 967 and 1058 are known to confer tetracycline resistance in Helicobacter pylori (reference 67 and references therein) and Propionibacterium acnes (85). Resistance to the universal antibiotics pactamycin and hygromycin B can occur through the mutation of 16S rRNA positions 694, 795, and 796 (56) and position 1495 (96), respectively.

The 30S ribosomal proteins associated with antibiotic resistance are primarily S4, S5, and S12. They have been genetically linked with 30S subunit function, and mutations in them are known to affect translational accuracy (66). Mutations in two loops of S12 confer streptomycin resistance and a hyperaccurate phenotype. The crystal structure of the streptomycin-30S subunit complex shows that streptomycin interacts directly with one of these loops. Mutations in S4 conferring streptomycin-resistant and hyperaccurate phenotypes on Salmonella enterica serovar Typhimurium have also been described (6). The phenotypes can be explained by a model in which the mutations stabilize the open conformation of the 30S subunit. Mutations in S5 are known to confer spectinomycin resistance and an error-prone ribosomal ambiguity (Ram) phenotype. These mutations occur at the S4-S5 interface and are thought to disrupt the S4-S5 interaction and thereby facilitate the formation of the closed conformation. However, a recently described E. coli S5 mutant contains a mutation at a residue 5 Å from the spectinomycin binding site, and this mutation may thus interfere directly with drug binding (42). There are two reports of mutations of S10 causing resistance to tetracycline, one in Bacillus subtilis (111) and the other in Neisseria gonorrhoeae (36).

Aminoglycoside resistance methyltransferases from a number of aminoglycoside producer organisms have been described. The founding members of this group are the methyltransferases encoded by the kamA and gmrA genes of Streptomyces tenjimariensis and Micromonospora purpurea (reviewed in reference 15), which methylate N-1 of A1408 and N-7 of G1405, respectively, in 16S rRNA (Table 1). The N-1 methylation at A1408 confers resistance to neamine and kanamycin by blocking interactions of aminoglycoside ring I with A1408. The N-7 methylation at G1405 confers resistance to 4,6-disubstituted deoxystreptamine aminoglycosides, including kanamycin and gentamicin, by sterically interfering with the positioning of ring III.

Table 1rRNA methyltransferases associated with antibiotic resistance in bacteria

Until 2003, it was believed that the genes encoding aminoglycoside resistance methyltransferases were confined to antibiotic producers. However, a number of methyltransferase genes, including armA, rmtA, rmtB, rmtC, and rmtD, in human gram-negative pathogens from different parts of the world have since been identified, suggesting their global dissemination (19, 18, 24, 110, 120). The ArmA methyltransferase methylates N-7 of G1405 (50), whereas the methylation position of the Rmt enzymes has not yet been determined. In addition, a recent report describes the isolation of a clinical multiple-aminoglycoside-resistant E. coli strain that harbors the nmpA methyltransferase gene. The NmpA methyltransferase methylates N-1 of A1408 (109).

A mystery surrounding streptomycin resistance has been unraveled recently with the identification of rsmG (gidB) as the gene encoding a methyltransferase that methylates N-7 of 16S rRNA nucleotide G527 (Table 1) (70). Streptomycin is known to interact with the phosphodiester backbone of four different parts of 16S rRNA, including nucleotides 526 and 527 in the 530 loop (12). A low-level streptomycin resistance phenotype is conferred by the lack of methylation (Table 1). Although the rsmG gene is highly conserved and found in all bacterial genomes sequenced to date, the deletion of the gene does not affect the growth rate of E. coli (70). Mutations in the rsmG gene have been shown to confer resistance in E. coli, M. smegmatis, Mycobacterium tuberculosis, Staphylococcus aureus, and Streptomyces coelicolor (64, 70). The clinical importance of rsmG mutations is underscored by the fact that they occur at high frequencies in streptomycin-resistant isolates of M. tuberculosis. Moreover, rsmG mutations are important for the development of high-level streptomycin resistance, as mutants with this phenotype were shown to be more likely to emerge from cells containing rsmG mutations than from wild-type cells (70). This effect is especially dramatic for M. tuberculosis; mutants of this species with high-level streptomycin resistance emerge from rsmG mutants at frequencies >2,000-fold higher than those at which such mutants emerge from wild-type cells.

The ksgA (rsmA) gene of E. coli encodes a methyltransferase that dimethylates N-6 of nucleotides A1518 and A1519 (33, 97). Although these methylations are part of a group of highly conserved posttranscriptional housekeeping modifications that are necessary for optimal rRNA function, the inactivation of ksgA confers a modest kasugamycin resistance phenotype (Table 1). Biochemical and genetic data reveal that the lack of methylation at A1519 is sufficient to confer kasugamycin resistance but that the identity and methylation status of A1518 are not resistance determinants (91, 108). Crystal structures of kasugamycin-bound E. coli 70S ribosomes (91) and T. thermophilus 30S subunits (87) show that the dimethylated A stem-loop does not come in contact with kasugamycin directly. Therefore, the lack of methylation at A1519 is likely to cause kasugamycin resistance indirectly by disrupting interactions between the dimethylated A stem-loop, helix 44 (107), and the 790 loop that contains two nucleotides that interact directly with the drug.

A resistance mechanism involving RNA methylation has also been identified for the universal antibiotic pactamycin. The pct gene of the pactamycin producer Streptomyces pactum encodes a methyltransferase that methylates N-1 of A964 (Table 1) (4). The crystal structure of pactamycin bound to the 30S subunit shows that N-6 of A964 is within hydrogen bonding distance of pactamycin, suggesting that methylation leads to a distortion of the drug binding pocket (9).

THE PTC ON THE 50S SUBUNIT

The PTC on the large ribosomal subunit is composed mainly of the central loop of domain V of 23S rRNA and contains binding sites for various antibiotics of clinical and veterinary importance. It is the site where amino acids are joined together to produce nascent peptides and, therefore, an efficient place for drugs to block protein synthesis. For the same reason, the PTC is the target for alterations that prevent drug binding and thereby cause antibiotic resistance. The PTC is located in an RNA-lined cleft where there are no ribosomal proteins directly at the surface (65). Therefore, the resistance mechanisms established so far at this site are mostly RNA mutations. Exceptions to this trend are a newly found methylation site at A2503 in 23S rRNA (41), causing phenicol, lincosamide, oxazolidinone, pleuromutilin, and streptogramin A (PhLOPS_A) resistance, and the lack of a natural RNA modification at U2584 in 23S rRNA, causing reduced sensitivity to sparsomycin (46). Furthermore, mutations in a tip of ribosomal protein L3 close to the PTC can confer resistance to tiamulin on bacteria (7, 27, 81) and to anisomycin and trichodermin in Saccharomyces cerevisiae (references cited in reference 81).

Much data on antibiotic resistance from mutations in domain V of 23S rRNA had accumulated before the first X-ray structure of the 50S subunit revealed the three-dimensional structure of the PTC (65). The majority of these mutations were either selected in E. coli or found in other organisms and then introduced into E. coli for confirmation and testing of the generality of their resistance effects. Since then, there have been additional publications of X-ray structures of the 50S subunit with chloramphenicol and clindamycin (30, 89, 104), streptogramin A (32), and pleuromutilins (16, 88) bound in the PTC. Superimposing these X-ray structures clearly shows that the drugs bind at overlapping, but not identical, sites (Fig. 4). Each of these drugs has its own specific set of interactions with RNA nucleotides in the PTC. In general, there is excellent agreement between the positions of the RNA mutations and the drug binding site, indicating that the mutations directly change or interfere with contacts between the drug and the RNA cavity.

Fig. 4Example of binding of four classes of antimicrobials to overlapping sites at the ribosomal PTC. A bacterial cell is depicted with a translating ribosome (top left panel). The antimicrobial drugs inhibit bacterial growth by binding to ribosomes and blocking protein synthesis. The structure of the bacterial 50S ribosomal subunit (90) and an expanded view with the structures of the four drugs (depicted in stick form according to the color scheme shown in the bottom panels) bound at the PTC (top right panel) are shown. The target of the Cfr methyltransferase causing resistance, nucleotide A2503, is shown in red. The surrounding RNA is shown in light grey. In the bottom panels, the names and chemical structures of the four antimicrobial agents are shown with background colors that correspond to those of their bound structures depicted in the expanded view. References for the antibiotic-50S subunit complexes are cited in the text.

Tiamulin is a pleuromutilin derivative that is important in veterinary medicine, and its binding site was localized to the PTC region by antibiotic footprinting in 2001 (80). A study of tiamulin resistance in E. coli showed that an L3 mutation conferring resistance can be selected (7). Another study of tiamulin resistance in Brachyspira spp. indicated that multiple RNA mutations and L3 mutations present together in the PTC region cause resistance, although the exact involvement of each mutation could not be elucidated (81). All mutated positions, including the L3 mutations, are close to position 2504 in 23S RNA, which interacts directly with tiamulin according to an X-ray structure of tiamulin bound to the D. radiodurans 50S subunit (88). It was concluded that small alterations close to position 2504 probably disturb the structure in a minor way that is sufficient to prevent the binding of tiamulin and its interaction with the nucleotide at position 2504.

Resistance to well-known antibiotics such as chloramphenicol, lincosamides, and streptogramin A compounds can be caused by 23S RNA mutations in the PTC area. Reduced susceptibility to chloramphenicol can be conferred by mutations at nucleotide positions 2032, 2058, 2061, 2062, 2447, 2451, 2452, 2503, and 2504, as reviewed in reference 54. Resistance to lincosamides (which include clindamycin) can be caused by RNA mutations at nucleotides 2032, 2058, and 2059, as reviewed in reference 25. Streptogramin A resistance mutations have been identified at positions 2059 and 2503 (79).

Additional antibiotics that bind to the PTC that are not currently getting attention but to which 23S RNA mutations have been found to confer resistance are anisomycin (mutations at positions 2447, 2452, and 2453), amicetin (a mutation at position 2438), and celesticetin (a mutation at position 2452), all reviewed in reference 46. More data on sparsomycin resistance have been obtained, including the identification of a resistance mutation at 2499 and "low-resistance" mutations at nucleotides 2452 and 2500 (99) and the finding of reduced sensitivity to sparsomycin caused by the lack of methylation of U2584 (31, 46).

One of the very few novel antibiotics to be introduced commercially in recent years that is not merely a derivative of a previous drug is linezolid, a PTC binding antibiotic belonging to the oxazolidinone class. Oxazolidinones are not active against gram-negative organisms, and this property is likely due to drug efflux. The oxazolidinone binding site has been localized to the A-site of the PTC via modelling studies based on crosslinking data with S. aureus ribosomes (47). This is in agreement with a very recent crystal structure of linezolid bound to the archaeal Haloarcula marismortui 50S subunit (37a). Resistance to oxazolidinones conferred by mutations of 23S rRNA has been observed at at positions 2192, 2447, 2500, 2505, 2512, 2513, 2576, 2608, and 2610 (61, 117, 118), all of which except 2192 are in or close to the PTC region.

Antibiotic resistance from RNA methylation directly in the PTC area has only recently been discovered (41). However, it is possible that other examples will be revealed in the future, as several cases of this type of resistance mechanism have emerged in recent years (Table 1). The cfr gene encodes a methyltransferase that methylates nucleotide A2503. The gene was originally discovered in a florfenicol-resistant Staphylococcus sp. but has also been expressed in an E. coli strain with altered membrane permeability (53). These studies showed that the Cfr methyltransferase confers resistance to five different classes of relevant antimicrobial agents, yielding the PhLOPS_A resistance phenotype. These drugs bind to overlapping sites at the PTC (Fig. 4). The resistance is significant and is likely the result of steric hindrance from a minor perturbation of the structure at or around A2503 caused by the methylation. In the case of E. coli, there is natural methylation at A2503, so the Cfr methyltransferase apparently adds an additional methyl group at A2503 and also prevents the natural ribose methylation at C2498 (41).

As very few new antimicrobial agents appear on the market, the finding that a single methylation confers resistance to five different classes of relevant antimicrobial agents is a matter of concern. This mechanism of resistance works in both gram-positive and gram-negative organisms (53), and it has been found on transposons and in different geographical locations (39, 40). A very recent paper (102) describes an S. aureus hospital strain from Colombia with linezolid resistance caused by a chromosomal cfr gene. The cfr gene in this clinical strain is linked to erm(B), which encodes a methyltransferase that dimethylates A2058 (described in the next section). Thus, the cfr gene has an alarming dissemination potential, as has been observed with the Erm methyltransferases described below.

THE PEPTIDE EXIT TUNNEL ON THE 50S SUBUNIT

The nascent peptides synthesized in the PTC progress into a tunnel that runs through the 50S ribosomal subunit. The upper part of the tunnel adjoins the PTC region and contains additional antibiotic binding sites, as reviewed in reference 76, where the drugs block the passage of the nascent peptide and thereby inhibit protein synthesis. In fact, some of the antibiotics mentioned in the previous section bind directly at the tunnel entrance, and some of the RNA positions at which mutations cause resistance are also part of the upper exit tunnel. To a large extent, the macrolides and the streptogramin B antibiotics, have a common set of resistance mechanisms. The macrolide erythromycin has been used in therapy since the 1950s, and the history of resistance to erythromycin is almost as long (reviewed in reference 48). Later it was discovered that inducible resistance to erythromycin also caused cross-resistance to two other groups of antibiotics, namely, lincosamides (mentioned in the previous section) and streptogramin B drugs, yielding the macrolide-lincosamide-streptogramin B (MLS_B) resistance phenotypes. The resistance was subsequently traced to the expression of a methyltransferase (ErmC) that dimethylates A2058 of 23S rRNA (95).

Since then, many erm genes similar to the gene encoding ErmC have been found, and the Erm family of methyltransferases, the members of which mediate the mono- or dimethylation of A2058, now consists of approximately 40 different classes of methylases (83, 84) (http://faculty.washington.edu/marilynr/). This type of resistance appears worldwide in a number of different bacteria. The erm methylase genes have been identified in a wide range of gram-positive and gram-negative bacteria, with the transposon-borne erm(B), erm(F), and erm(A) genes, as well as the plasmid-borne erm(C) gene, having the broadest host ranges (84). The high incidence of resistance is probably caused by the extensive use of macrolides for the treatment of bacterial infections in humans and animals and by the nontherapeutic use of these drugs as growth promoters (in the form of feed additives given to farm animals to promote animal growth) in agriculture. The acquired resistance is not especially detrimental to bacteria and can persist for a long time, which, in turn, promotes its spread. In addition, the erm resistance is often inducible, so it is expressed only when needed.

While Erm dimethylases cause high-level resistance to all MLS_B antibiotics, the Erm monomethylases acting at A2058 confer high-level resistance to lincosamides but low to medium levels of resistance to macrolides and streptogramin B drugs (72, 112). The MLS_B resistance phenotypes are sometimes divided into class I and II groups, depending on whether they originate from a mono- or a dimethyltransferase (72). Resistance methyltransferases can also work together, as illustrated by the A2058 monomethylase ErmN and the G748 methylase RlmA_II that together confer resistance to tylosin but that alone do not confer this resistance (51). G748 is in domain II of 23S RNA in the loop of helix 35 that is situated farther down the peptide exit tunnel than the erythromycin binding site. The large macrolide tylosin contacts both the G748 and the A2058 nucleotides. The interaction at G748 may enable tylosin to establish contact with monomethylated A2058 by induced fitting, and methylation at G748 prevents this weak but important interaction (reviewed in reference 20).

Other resistance determinants located in the peptide exit tunnel are mutations in ribosomal proteins L22 and L4. Such mutations were found decades ago to confer resistance to erythromycin in E. coli (115). Erythromycin binding to ribosomes containing an L4 mutant protein with one amino acid change is reduced, but binding to ribosomes with an L22 mutant protein with a 3-amino-acid deletion is unchanged relative to binding to wild type ribosomes. Loops of these proteins are components of the tunnel wall at its narrowest place, just below the erythromycin binding site (104). The X-ray structure of an H. marismortui 50S subunit with a mutated L22 has been determined, but the resistance mechanism has not been fully revealed, largely because the antibiotic effect is not fully understood (104). The L22 mutation seems to allow the binding of erythromycin by creating extra space in the tunnel so that the nascent peptide can more easily pass around the drug and through the narrow site. Several mutations in L4 and L22 conferring macrolide resistance have now been found in various pathogenic strains (references 22 and 83 and references therein) and recently also in Campylobacter spp. (10) and Ureaplasma parvum (71).

Mutations in rRNA can also cause resistance to the MLS_B antibiotics, as reviewed in reference 105, and although these mutations are not transferable by horizontal gene transfer, their ease of appearance in bacteria with one or a few rRNA operons has a significant clinical impact. These mutations appear most frequently in bacteria with a small number of rRNA operons (105), in which the effect from a mutation is potentially more prominent. Numerous papers have addressed such resistance in various organisms, all by the mutation of one of the following positions: 2057, 2058, 2059, 2062, 2452, and 2611. The effects of many of these mutations have been verified by the expression of plasmid-borne mutated rRNA operons in E. coli. Examples of most of these mutations have been known for some time (105), except for the position 2062 mutation (related to macrolide resistance) (23) and the position 2609 mutation (related to ketolide resistance) (26). Not all mutations cause a full MLS_B phenotype, and often the effects of the mutations have been investigated with only one or a few antibiotics, so it is unknown if resistance to all MLS_B drugs is conferred. Moreover, different bacteria show different resistance effects from the various mutations, and some mutations are apparently lethal or deleterious in some organisms but not in others.

OTHER SITES ON THE 50S SUBUNIT

The GTPase Center

The GTPase center on the 50S subunit (Fig. 3) is implicated in GTP hydrolysis in connection with the function of translation factors. A few antibiotics inhibit protein synthesis by binding to this functional region. The thiopeptide antibiotics thiostrepton and micrococcin bind to a highly conserved ribonucleoprotein domain composed of ribosomal protein L11 and part of domain II of 23S rRNA (positions 1051 to 1108) (114), and a recent paper has documented the interaction between L11 and thiostrepton (49). The thiopeptide antibiotics are not currently of therapeutic interest, although they have applications in the food industry. Nonetheless, important studies of the site and of drug binding to the site have been performed, and resistance mechanisms have been revealed. Mutations at positions 1067 and 1095 in 23S RNA and at several positions in ribosomal protein L11 cause resistance to thiostrepton (references in reference 11). Thiostrepton is believed to stabilize the local RNA-L11 structure at the binding site and interfere with conformational changes associated with GTP hydrolysis, thereby inhibiting GTP hydrolysis (reviewed in reference 78). Micrococcin binds in the same region as thiostrepton, although the drug enhances elongation factor G-dependent GTP hydrolysis, and mutations in ribosomal protein L11 have also been shown to confer resistance to micrococcin (77). Resistance to thiostrepton by the producer organism is mediated by 2'-O-ribose methylation of A1067, which also confers resistance to micrococcin (Table 1) (101). It is believed that each of the resistance mechanisms functions by increasing the flexibility of the local RNA structure to counteract the stabilizing effects of the drugs (11).

The Orthosomycin Site

The orthosomycin antibiotics target a site on the 50S ribosomal subunit that includes helices 89 and 91 from 23S RNA and protein L16 and is close to the elbow of tRNA positioned in the A site (43) (Fig. 3). The best-known orthosomycin compounds are evernimicin, which was developed for use in humans, and avilamycin, which is used as a growth promoter. Orthosomycin antibiotics act only on gram-positive bacteria, but E. coli ribosomes have nevertheless been used extensively to study their binding and resistance mechanisms. Very little was published about the association between these drugs and ribosomes until 2000, when biochemical details on the subject accumulated rapidly (references in references 43 and 103). Mutations in ribosomal protein L16 were first shown to confer resistance to evernimicin and cross-resistance to evernimicin and avilamicin (references in reference 103). This finding was followed by the identification of mutations in domain V of 23S rRNA that confer resistance to orthosomycins, namely, those at positions 2469 to 2472, 2479 to 2480, and 2535 and 2536 (summarized in reference 43). Antibiotic footprint analyses confirmed the location of the orthosomycin binding site on the 50S subunit. High-resolution structures of 50S subunit-orthosomycin complexes have not been published, but all resistance determinants and positions identified by footprinting cluster in the same region, at a site overlapping with or close to the initiation factor 2 binding site (58). This region is thus considered to be the orthosomycin binding site, and as for the previous mentioned target sites, both methylations and mutations at the binding site have been identified as resistance determinants.

The EmtA methyltransferase acting at 23S rRNA G2470 was found to confer resistance to evernimicin (Table 1) (57). Soon thereafter, two additional RNA methyltransferases providing resistance were described (113). Four genes in the natural producer of avilamycin, the actinomycete Streptomyces viridochromogenes Tü57, were found to be resistance determinants (113). Two genes are similar to those encoding an ATP binding cassette transporter system that probably exports avilamycin across the cell membrane, whereas the other two genes, aviRa and aviRb , encode rRNA methyltransferases. The modifications were later identified as 2'-O-ribose methylation at U2479 (by AviRb) and base methylation at G2535 (by AviRa) (Table 1) (103). Both positions 2479 and 2535 are in the same RNA region on the 50S subunit as all of the resistance mutations mentioned above. AviRb confers a higher level of resistance than AviRa, but both are necessary for self-protection in the avilamycin producer, and they are thought to act together like the tylosin resistance methyltransferases described above.

THE RIBOSOMAL SUBUNIT INTERFACE

Ribosomes are composed of a small ribosomal subunit and a large ribosomal subunit held together through a network of intermolecular contacts called intersubunit bridges. The intersubunit cavity is spanned by tRNAs, with anticodons base paired with mRNA codons bound on the small subunit and 3'-CCA ends carrying amino acids positioned in the peptidyl transferase cavity on the large subunit. Intersubunit bridge B2a has been conserved throughout evolution and is necessary for the association of ribosomal subunits and the binding of tRNAs (3). It is located at the center of the subunit interface between the ribosomal A and P sites and is composed entirely of rRNA nucleotides from helix 44 of 16S rRNA and helix 69 of 23S rRNA (90, 121). Intersubunit bridge B2a is targeted by the cyclic peptide antibiotics capreomycin and viomycin, which are unique in that they recognize the assembled ribosome by interacting with nucleotides on both subunits (38). The antibiotics inhibit translation by trapping the subunits at an intermediate state in ribosomal subunit movement and thereby impeding tRNA movement (21, 38).

Capreomycin is an important drug in the treatment of multidrug-resistant tuberculosis that is resistant to the first-line antibiotics isoniazid and rifampin. The isolation and characterization of several capreomycin-resistant M. tuberculosis and M. smegmatis mutants revealed that these strains contained mutations that inactivated the tlyA gene (60). The tlyA gene has recently been shown to encode a 2'-O-methyltransferase that modifies nucleotides C1409 in helix 44 of 16S rRNA and C1920 in helix 69 of 23S rRNA (Table 1) (38). The inactivation of the TlyA methyltransferase and the resulting lack of these methylations confers capreomycin resistance in M. tuberculosis isolates (38, 60). Thus, TlyA belongs to the group of methyltransferases for which the lack rather than the presence of methylation confers an antibiotic resistance phenotype (Table 1). The lack of natural homologs of tlyA in many bacteria is correlated with decreased susceptibilities to capreomycin and viomycin relative to those of mycobacteria. In support of this observation, the expression of plasmid-borne tlyA in E. coli results in 2'-O-methylation at the above-mentioned nucleotides and increased susceptibilities to capreomycin and viomycin (38).

Mutations in 16S rRNA nucleotides involved in intersubunit bridge B2a have also been associated with resistance to capreomycin and viomycin. Genetic studies with M. smegmatis have shown that the mutation of nucleotide G1491 that is base paired with C1409 confers resistance to viomycin (100). In addition, mutations of nucleotides 1409 and 1491 have been associated with capreomycin and viomycin resistance in M. tuberculosis (59). Moreover, the mutation of nucleotide 1408 has been associated with capreomycin resistance in T. thermophilus (28) and clinical M. tuberculosis isolates (60).

The E. coli rrmA gene encodes an m¹G methyltransferase that methylates nucleotide 745 of 23S rRNA (29). An rrmA mutation resulting in a lack of methylation has been reported to be associated with viomycin resistance (29). However, as nucleotide 745 is buried within the 50S subunit and not near the subunit interface, it is difficult to reconcile the role of this methylation in the binding of viomycin to intersubunit bridge B2a (38).

CONCLUDING REMARKS

Our knowledge of antibiotic resistance mechanisms has increased in recent years, in particular due to the elucidation of the detailed structures of antibiotic-ribosome complexes and the components of the efflux systems. A number of additional mutations and methyltransferases conferring antibiotic resistance have been characterized. These developments are important for understanding and approaching the problems associated with multidrug resistance. Multidrug resistance and cross-resistance are persisting and increasing clinical problems, as some common infections are becoming refractory to antibiotic therapy. An understanding of the mechanisms of drug resistance is important for designing new antimicrobials that are impervious to the known bacterial resistance mechanisms and for preventing the spread of resistance. There is a need for new drugs that are not affected by known mutations or methylations at the target site. For example, drug companies have tried to circumvent the Erm-mediated resistance by developing semisynthetic macrolide antibiotics, such as the ketolide telithromycin, that produce additional interactions with the ribosome (104) and thereby possess improved resistance properties (22). Another approach is the development of efflux inhibitors that may extend the use of existing antibiotics approved for clinical use (52). A similar approach has been used for some time with the combination of amoxicillin, a β-lactam antibiotic, and clavulanic acid, a β-lactamase inhibitor, which has an increased spectrum of action compared to that of the former alone and restored efficacy against β-lactamase-producing amoxicillin-resistant bacteria.

An important problem is the spread of antibiotic resistance determinants. In the case of the erm methyltransferase genes, their proliferation has likely been reinforced by the use of growth promoters. We still lack knowledge on the mechanisms through which resistance determinants are transferred in both agricultural settings and the human community (86). Therefore, the identification of resistance determinants on plasmids, transposons, or gene cassettes is important for estimating the potential for their dissemination via horizontal gene transfer. An example of genetic mobility via transposition has recently been documented, for the cfr methyltransferase gene described above, in staphylococci isolated from pigs in Germany and Denmark (39).