Antimicrobial Chemotherapy, 5th Edition
Part 2 - Resistance to Antimicrobial Agents
Mechanisms of Acquired Resistance
Three conditions must be met in order that a particular antimicrobial agent can inhibit susceptible bacteria:
- the antibiotic must be able to reach the target in sufficient concentration and be metabolically active (e.g. optimal pH, redox potential);
- the antibiotic must not be inactivated before binding to the target;
- a vital target susceptible to the action of the antibiotic must exist in the bacterial cell.
The targets of individual antibiotics are often enzymes or other essential proteins. Most antimicrobial agents have to pass through the cell wall and outer membrane to reach their target, and many are carried into the cell by active transport mechanisms that usually transport sugars and other beneficial substances. Some of the differences in susceptibility of bacterial species are therefore related to differences in cell wall structure. For example, the cell envelope of Gram-negative bacteria is a more complex structure than the Gram-positive cell wall (see Chapter 1) and offers a relatively greater barrier to many antibiotics, including penicillins, glycopeptides, and macrolides. Polymyxins exert their effects at the cell surface by disrupting the Gram-negative cell membranes from the outside in a way that resembles the action of some detergents.
The mechanisms by which resistance may occur can be divided into the following major groups:
- destruction or inactivation of the antibiotic (Fig. 9.1, I);
- alteration or protection of the target site to reduce or eliminate binding of the antibiotic to the target (Fig. 9.1, II);
- reduction in cell surface permeability or blockage of the mechanism by which the antibiotic enters cell (Fig. 9.1, III) or removal from the cell (efflux) (Fig. 9.1, IV);
- acquisition of a replacement for the metabolic step inhibited by the antibiotic (Fig. 9.1, V).
Fig. 9.1 Mechanisms of antimicrobial drug resistance.
It is worth emphasizing that certain resistance mechanisms overlap within these groups. Furthermore, bacteria can become resistant to an antibiotic by several different mechanisms; e.g. drug efflux, target protection, target alteration, and enzymatic inactivation may each afford resistance to tetracyclines.
Some of the known mechanisms of antibiotic resistance are summarized in Table 9.1.
Inactivation or modification mechanisms
These are probably the most important resistance mechanisms in clinical practice since they include the common modes of resistance to penicillins and cephalosporins, the therapeutic agents of widest use.
There are many different agents in this group (see Chapter 1) and a correspondingly large number of β-lactamases that catalyse hydrolysis of the β-lactam ring to form an inactive product (Fig. 9.2). In addition, varying levels of β-lactamase production, variable properties of the enzymes, notably the breadth of activity, and differences in the permeability of the
Gram-negative cell envelope, determine the differential susceptibilities of bacteria to these antibiotics.
Table 9.1 Important known resistance mechanisms for the major groups of antibiotics
Fig. 9.2 β-Lactamase hydrolysis of penicillins to form the corresponding penicilloic acid, which is antibacterially inactive. Cephalosporins may be attacked in a similar fashion, but the resultant cephalosporoic acid is usually unstable and disintegrates into smaller fragments.
All bacteria appear to contain enzymes capable of hydrolysing β-lactam antibiotics. Indeed, it has been suggested that the normal function and evolutionary origin of β-lactamases is to break a β-lactam structure that is a transitory intermediate in cell wall synthesis. These constitutive enzymes are encoded by the bacterial chromosome, and are normally bound closely to the cell membrane. In general, they are produced only in small amounts, they attack cephalosporins more readily than penicillins, and they act relatively slowly. In certain organisms, notably Enterobacterspp., Acinetobacter spp., Citrobacter spp., and Pseudomonas aeruginosa, gross overproduction of these chromosomal enzymes has been associated with treatment failure, even with so-called ‘β-lactamase-stable’ cephalosporins and carbapenems. Resistance seems to result from a combination of the slow enzyme-mediated hydrolysis, and cell wall mutations that partially impede antibiotic entry into the bacterial cell.
The classification of β-lactamases has become extremely complex as more and more examples of these enzymes (now numbering more than 200), some differing from one another by only one or a few amino acids, have been described. Various characteristics are used to distinguish the different enzymes, including substrate profile, and the action of enzyme inhibitors such as clavulanic acid and the ion-chelator, ethylenediaminetetraacetic acid (EDTA). Substrate profile refers to the hydrolytic activity of a β-lactamase preparation against a number of β-lactam substrates, often expressed as the ratio to a value for a reference substrate such as benzylpenicillin. Methods based on DNA-DNA hybridization or the polymerase chain reaction (PCR) to identify specific genes, have been used to distinguish newly recognized enzymes. A simplified classification of the common bacterial β-lactamases is shown in Table 9.2.
Gram-negative bacteria produce a greater variety of β-lactamases than Gram-positive bacteria. From a clinical point of view, most interest centres on the large number of plasmid-encoded enzymes, particularly given their potential for widespread dissemination. Plasmid-encoded enzymes are the major cause of bacterial resistance to penicillins and cephalosporins in clinical isolates. Some are located on transposons (described more fully in Chapter 10), which allow movement of genes between plasmids and the chromosome. This means that the distinction between plasmid-encoded and chromosome-encoded enzymes sometimes is blurred.
Among Gram-positive cocci, plasmid-encoded β-lactamases of clinical significance are found almost exclusively in staphylococci. These enzymes rapidly hydrolyse benzylpenicillin, ampicillin, and most other penicillins, but are less active against ‘antistaphylococcal penicillins’ (p. 19) and cephalosporins. Staphylococcal β-lactamases are inducible exo-enzymes that are usually related closely. In streptococci, β-lactamases are usually absent, and these bacteria have consequently remained, with few exceptions, susceptible to benzylpenicillin.
Table 9.2 Simplified categorization of the most common bacterial β-lactamases
The most widely distributed of the plasmid-mediated enzymes is TEM-1, which is encoded by many different plasmids (p. 147) and transposons (p. 149). The resultant genetic promiscuity, coupled with sustained selective pressure from antibiotic prescribing, probably explains the widespread distribution of this and closely related enzymes. Following the first recognition of TEM-1 in Esch. coli in 1965, it was detected in Haemophilus influenzae and Neisseria gonorrhoeae in the mid-1970s, and in Neisseria meningitidis in 1989.
There are more than 100 genetic TEM variants, some of which have an altered substrate spectrum. Some produce β-lactamases that can hydrolyse a wide variety of penicillins and cephalosporins (extended spectrum β-lactamases). Extended-spectrum enzymes unrelated to TEM, have also been described, notably the plasmid-encoded cefotaxime-hydrolysing (CTX-M) class of β-lactamases that has been detected inKlebsiella spp., Esch. coli, and salmonellae. They have become much more prevalent in the UK and other European countries, in particular causing urinary tract infections or septicaemia. The laboratory detection of these enzymes is not straightforward, and relies on a combination of clues obtained from the results of routine susceptibility testing and additional tests based on cephalosporin-induced β-lactamase production in vitro.
Other types of β-lactamases that are encountered in Gram-negative bacilli include: SHV-1 and its many variants—of which there are now more than 50—that are common in Klebsiella spp.; the OXA group of enzymes, which can hydrolyse methicillin and isoxazolylpenicillins; and the PSE group that hydrolyse carbenicillin at least as fast as benzylpenicillin, and which were thought originally to be confined to Ps. aeruginosa.
Control of resistance caused by TEM-1 and some other β-lactamases produced by Gram-negative organisms is afforded by ‘β-lactamase-stable’ cephalosporins, and the use of β-lactamase inhibitors, such as clavulanic acid. However, some plasmid-encoded β-lactamases inactivate even the newer ‘β-lactamase-stable’ β-lactam agents. Many of these novel enzymes seem to be derived by mutation from the widely distributed TEM-1 and SHV-1 β-lactamases.
Most β-lactamases have serine at the active site and are often referred to as serine β-lactamases. However, some β-lactamases require zinc, and thus are known as metallo-β-lactamases; these enzymes are inhibited by the chelating agent EDTA. Metallo-β-lactamases are found in diverse organisms including Acinetobacter spp., Ps. aeruginosa, and the Bacteroides group. They hydrolyse virtually all β-lactam compounds, including the carbapenems (p. 26) and enzyme inhibitors such as clavulanic acid do not offer protection. Serine β-lactamases that inactivate carbapenems have also been described. Fortunately, strains that elaborate such enzymes are still relatively uncommon, although they are now being reported with increasing frequency from many different countries. Several outbreaks have been described involving multiresistantKlebsiella strains that have also acquired a carbapenemase, leaving very few therapeutic options.
Resistance to aminoglycosides results largely from interference with the drug transport mechanism following modification of the antibiotic by one or more of a series of enzymes produced by the resistant bacteria. Such aminoglycoside-modifying enzymes are often plasmid-encoded, but have been associated increasingly with the presence of transposons (p. 149) and integrons (p. 150). They are classified according to the precise type of modification performed, and by the site of modification on the aminoglycoside molecule. Over 30 such modifying enzymes and their variants have been identified by biochemical or nucleic acid-based methods, and these can be divided into three main groups: aminoglycoside acetylating enzymes; nucleotidyltransferase enzymes; and phosphorylating enzymes. Examples of the most widely distributed enzymes are listed in Table 9.3.
Figure 9.3 shows the structure of kanamycin A, a typical aminoglycoside, and indicates the various sites at which modification can take place. The presence or absence of available amino or hydroxyl groupings affects the susceptibility to various enzymes; this is the basis of variability within the aminoglycoside group. The steric configuration of the groupings is also important; thus, the semisynthetic aminoglycoside amikacin is, structurally, related closely to kanamycin A, but is much less susceptible to enzymic modification because of a hydroxyaminobutyric acid side chain that alters the steric configuration of the molecule. The acetylating enzymes, of which there are at least 16 types, catalyse the transfer of acetate from acetyl coenzyme A to an amino group on the aminoglycoside molecule. These enzymes modify only deoxystreptamine-containing aminoglycosides (p. 36) and are, therefore, without effect on streptomycin or spectinomycin. By contrast, aminoglycoside nucleotidyltransferases use adenosine triphosphate or other nucleotides as substrates and attach the nucleotide to exposed hydroxyl groups, while phosphotransferases also modify hydroxyl groups, but by attachment of a phosphate molecule.
Table 9.3 Examples of some of the most common aminoglycoside-modifying enzymes and their characteristic substrates
Fig. 9.3 Structure of kanamycin A, showing the sites at which enzymic modification can occur.
Various patterns of cross-resistance can be shown by bacteria that produce different enzymes (Table 9.3), but these are complicated further because many clinical isolates produce more than one enzyme at any one time. Susceptibility or resistance to any one agent cannot be predicted reliably from results obtained for another; thus, susceptibility tests must be performed with the agent that is to be used therapeutically.
Although aminoglycosides exhibit poor activity against enterococci, they interact synergically with β-lactam antibiotics to achieve a more rapid and complete bactericidal action. Unfortunately plasmid-mediated high-level aminoglycoside resistance, which abolishes synergy, has become much more prevalent in enterococci, so removing the possibility of synergistic β-lactam-aminoglycoside therapy in serious enterococcal infection caused by such strains.
When the aminoglycoside-modifying enzymes were first described, they were considered to be examples of drug-inactivating enzymes analogous to those responsible for resistance to β-lactam antibiotics and chloramphenicol. However, aminoglycoside-modifying enzymes mediate resistance by modifying only small amounts of antibiotic. They are strategically placed near the inner cytoplasmic membrane where they are accessible to acetyl coenzyme A and adenosine triphosphate. As soon as a few molecules of drug are modified, all further transport of drug into the cell becomes blocked.
Resistance to chloramphenicol in both Gram-positive and Gram-negative bacteria is normally associated with production of an enzyme, chloramphenicol acetyltransferase, which converts the drug to either the monoacetate or diacetate. The acetylated drug will not bind to the bacterial ribosome, and so cannot block protein synthesis. Several different acetyltransferases have been described. Some appear to be genus- and species-specific; others, usually plasmid or transposon-associated, are more widespread. This variety is a little surprising since chloramphenicol has been used less widely than many other antibiotics because of its rare, but serious, toxic side effects.
Alteration or protection of the target site
Resistance arising from the selection of rare, pre-existent mutants from within an otherwise susceptible bacterial population has been described for many antibiotics. The mutations usually affect the drug target and often confer high-level resistance in a single step. The emergence of this type of resistance during therapy is an important cause of treatment failure with certain drugs, including rifampicin, the older quinolones, fusidic acid, and various antituberculosis drugs. Use of combinations of antibiotics can prevent the emergence of such resistance during therapy, since the likelihood of independent mutations conferring resistance to two or more unrelated antibiotics appearing simultaneously in the same cell is very small. This strategy has been crucial in antituberculosis therapy. Similarly, monotherapy of staphylococcal infection with rifampicin or fusidic acid should not normally be used.
Variants exhibiting low levels of resistance to almost any antibiotic can be isolated readily from most bacteria. In contrast to single-step mutants they usually develop in a stepwise fashion, and may be accompanied by other phenotypic changes (e.g. slower growth rate, colonial variation on solid media, reduced virulence). These so-called ‘fitness costs’ become more marked as the degree of resistance increases. It is likely that the shifts in penicillin susceptibility of gonococci and pneumococci that have occurred over the years result from such cumulative changes.
The major mechanism of resistance to β-lactam antibiotics is enzymic inactivation (see above), but mechanisms involving target site modification also occur. Streptococcus pneumoniae strains with reduced susceptibility to penicillin exhibit alterations in the target penicillin-binding proteins (PBPs; p. 29) that result in reduced ability to bind penicillin. Similarly, methicillin resistance in staphylococci is associated with the synthesis of a modified PBP (PBP 2′), which exhibits decreased affinity for methicillin and other β-lactam antibiotics. Low-level resistance to penicillin in N. gonorrhoeae has also been associated with alterations to PBPs. Such resistance appears to have developed by rare mutational events and to have become disseminated as a result of considerable antibiotic selection pressure.
Resistance to vancomycin and teicoplanin first emerged in enterococci only after the antibiotic had been available for 30 years. Expression of resistance depends on the presence or absence of several genes and two enzymes (a ligase and a dehydrogenase), which probably originated in non-human pathogens and were transferred to enterococci. The net effect of these is target site alteration. Glycopeptides bind to the D-alanyl-D-alanine terminus of the muramyl pentapeptide of peptidoglycan (p. 31). Enterococci that exhibit high-level resistance to glycopeptides produce a new dipeptide terminus, either D-alanyl-D-lactate or D-alanyl-D-serine. Such substitutions allow cell wall synthesis to continue in the presence of one or both of the currently available glycopeptides, vancomycin and teicoplanin. Several different glycopeptide resistance phenotypes have been described (Table 9.4). Glycopeptide resistance is usually inducible in those strains that posses the necessary genes and enzymes, but some essentially non-pathogenic enterococci (e.g. Enterococcus gallinarum, E. casseliflavus,and E. flavescens) are constitutively resistant to low-moderate levels of vancomycin.
The mechanism of the low-level resistance to glycopeptide antibiotics that has emerged in some Staph. aureus and coagulase-negative staphylococci has not been completely elucidated, but appears to be associated with overproduction of peptidoglycan precursors that require increased amounts of drug to saturate them. Very rare strains of Staph. aureus that are highly resistant (MIC >128 mg/l) to vancomycin and teicoplanin have been described. So far these have all possessed the Van A gene that confers the vanA glycopeptide resistance phenotype in enterococci. Some years before these clinical isolates were first seen, the Van A gene was successfully transferred from an enterococcal strain into Staph. aureus in vitro. It now appears that enterococci can also transfer the genes coding for high-level glycopeptide resistance to Staph. aureus in vivo. The hope is that such occurrences remain rare, and that highly glycopeptide resistant strains do not spread widely.
Table 9.4 Types of glycopeptide resistance in enterococci
Streptomycin binds to a protein (S12) in the smaller (30 S) ribosomal subunit in bacteria. A single amino acid change in the structure of this protein can prevent the binding of streptomycin entirely, so rendering bacteria resistant to very high concentrations of the drug. The alteration is so specific that other aminoglycosides, such as gentamicin, are unaffected by the change.
Erythromycin and chloramphenicol
Changes in the proteins of the larger (50 S) ribosomal subunit have been implicated in resistance to chloramphenicol and macrolides such as erythromycin. However, erythromycin resistance in staphylococci and streptococci results more usually from methylation of the 23 S ribosomal RNA subunit by an inducible plasmid-encoded enzyme. Methylation of the ribosomal RNA also renders the bacteria resistant to other macrolides, lincosamides (p. 49), and streptogramins (p. 50) by reducing ribosomal binding of these drugs. Since erythromycin is a specific inducer of the methylating enzyme, bacteria carrying a plasmid encoding this property are resistant to the other drugs only in the presence of erythromycin. This phenomenon is known as dissociated resistance. Because of a similarity in the sites of action of macrolides and lincosamides, lincosamide resistance is more likely to emerge during treatment of infection caused by strains that are initially erythromycin resistant.
Fusidic acid inhibits translocation of the growing polypeptide chain. Point mutations in the fus A gene lead to altered structure of the protein that regulates this process (elongation factor G), resulting in the cell becoming resistant.
Intermediate levels of resistance to quinolones in Gram-negative rods are usually caused by chromosomal-encoded structural alterations to a subunit of the DNA gyrase target (p. 57). High levels of resistance are associated with additional mutations in the secondary target, topoisomerase IV. In Gram-positive cocci the situation is reversed, since topoisomerase IV is the primary target. In general, mutations that confer reduced susceptibility to older quinolones (‘first stage mutations’) may not reduce the effectiveness of newer more active versions, unless additional (‘second or third stage’) mutations occur; an example of stepwise resistance. There is some evidence that quinolones differ in their propensity to select for resistance mutations. This observation has led to the concept of a ‘mutant protection concentration’; i.e. the concentration that prevents the growth of the least susceptible single-step mutant present in a bacterial population. It has been suggested that these differences between quinolones, related to their achievable tissue concentrations, may influence the likelihood of resistant mutant selection during therapy.
Quinolone resistance can be transferred by plasmids that may additionally code for resistance to other antibiotic classes. This type of resistance is mediated by the qnr gene, which encodes a protein that protects DNA gyrase and DNA topoisomerase from the action of quinolones. Early data suggest that the prevalence of this type of resistance varies markedly.
Resistance to rifampicin is invariably the result of a structural alteration in the rpo gene that encodes the β-subunit of RNA polymerase, which is involved in the transcription of DNA to messenger RNA; this reduces its binding affinity for rifampicin. The location of the rpo gene mutations is usually within a well-defined small area, and molecular tests (p. 174) are now available to detect the responsible alteration.
This oxazolidinone antibiotic inhibits protein synthesis at the stage of ribosomal assembly (p. 51). A ribosomal mutation leads to linezolid resistance in both staphylococci and enterococci. Linezolid resistance is currently very rare in Staph. aureus and is only occasionally seen in enterococci, usually associated with prolonged therapy and failure to remove or drain a focus of infection.
Interference with drug transport and accumulation
In addition to reduced drug accumulation resulting from enzymic modification of aminoglycosides (see above), interference with transport of drugs into the bacterial cell is of proven clinical importance as a cause of resistance to tetracyclines, β-lactam antibiotics, and quinolones.
Uptake of tetracyclines into cells normally involves an active transport mechanism that uses energy and results in accumulation of drug inside the cell. Plasmid or transposon-mediated resistance to tetracyclines is common in both Gram-positive and Gram-negative bacteria. Generally, there is complete cross-resistance, so a strain that is resistant to one tetracycline is resistant to all the others; exceptions to this rule may be found with minocycline. Tigecycline, which is related to minocycline, retains activity against bacteria that are resistant to other tetracyclines, possibly relating to higher affinity binding to the ribosome.
Tetracycline resistance is often associated with the synthesis of a membrane protein that mediates rapid efflux of antibiotic by an active mechanism; thus, drug entering the cell is removed almost simultaneously and so fails to reach an inhibitory level. Such resistance is normally inducible, and full expression of resistance is obtained only after cells have been exposed to subinhibitory concentrations of the drug.
Some bacteria produce a cytoplasmic protein that appears to have the function of protecting ribosomes from tetracycline attack. Also, tetracycline resistance in Helicobacter pylori is mediated by a modification to the ribosomal target.
The outer membranes of Gram-negative bacilli vary greatly in permeability to various penicillins and cephalosporins. Most β-lactam agents reach their targets in Gram-negative bacilli by passing through the water-filled pores (porins) that extend across the outer membrane bilayer. The rate of permeation is governed largely by the physical size of a particular β-lactam molecule in comparison with the size of the porin, but ionic charge also plays a part. In some bacteria, resistance can result from changes to the size or function of the porins, so that passage of the antibiotic is prevented. In a few instances, genes carried on plasmids encode non-specific changes in cell permeability to β-lactam antibiotics; these changes seem to affect the overall outer membrane structure of the cell. Resistance to carbapenems can be caused by loss of porins, sometimes exacerbated by β-lactamase production; e.g. imipenem resistance in Ps. aeruginosa.
A few strains of chloramphenicol-resistant Gram-negative bacilli possess a plasmid that appears to confer the property of impermeability to chloramphenicol upon the host cell.
Gram-negative bacteria have been described in which resistance to quinolones is caused by impermeability associated with a decrease in the amount of the OmpF outer membrane porin protein. Such strains may simultaneously acquire resistance to β-lactam antibiotics and some other agents that gain access through the OmpF porin. Resistance caused by active efflux also occurs in some Gram-negative bacilli and staphylococci. This can be mediated by efflux pumps that are specific for quinolones or by non-specific transporter pumps.
A mechanism of resistance to aminoglycosides, unrelated to enzymic modification, is associated with alterations in membrane proteins that affect active transport of the antibiotic into the cell.
Most common resistance mechanisms can be accommodated in one or other of the three major groups described already. However, there are two known examples in which a plasmid or transposon provides the cell with an entirely new and drug-resistant enzyme that can bypass the susceptible chromosomal enzyme that is also present unaltered in the cell.
Sulphonamides exert their bacteristatic effect by competitive inhibition of dihydropteroate synthetase. Sulphonamide-resistant strains of Gram-negative bacilli synthesize an additional dihydropteroate synthetase that is unaffected by sulphonamides. The additional enzyme allows continued functioning of the threatened metabolic pathway in the presence of the drug. At least two such enzymes are widespread in Gram-negative bacilli throughout the world.
Trimethoprim blocks a later step in the same metabolic pathway by inhibiting the dihydrofolate reductase enzymes in susceptible bacteria. Resistant strains synthesize a new, trimethoprim-insensitive, dihydrofolate reductase as well as the normal drug-susceptible chromosomal enzyme. At least 14 groups of trimethoprim-insusceptible dihydrofolate reductases have been described in Gram-negative bacilli, and a further example is found in multiresistant isolates of Staph. aureus.
The mechanisms discussed above illustrate the diversity of ways in which microbes can become resistant to the drugs deployed against them. However, any attempt to limit the spread of drug resistance requires not only knowledge of the mechanisms themselves, but also an understanding of the genetic factors that control their emergence and continued evolution. These factors are the subject of the next chapter.