Mechanisms of Resistance to Antibacterial Agents

Emergence and Spread of Antibiotic Resistance

The emergence of antimicrobial resistance phenotypes is inevitably linked to the clinical (or

other) use of the antimicrobial agent against which resistance is directed. One reason for this

association is trivial—we do not generally test for resistance to antibiotics that are not in

clinical use. The second reason is that nature abhors a vacuum, so when an effective

antibiotic eliminates a susceptible biota, resistant varieties soon fill the niche. Once a

resistance phenotype has emerged within a previously susceptible species, the rapidity and

efficiency with which it spreads are affected by a host of different factors, including the

degree of resistance expressed, the ability of the organism to tolerate the resistance

mechanism, linkage to other genes, site of primary colonization, and others. The rapidity and

completeness of resistance gene spread are often unpredictable. For example, the

staphylococcal β-lactamase gene (conferring resistance to penicillin) was first described

shortly after the introduction of penicillin into clinical use and is now almost universally

present within staphylococci in the hospital and the community. It was not until the early

1980s that this gene was described to occur in Enterococcus, and it has never spread widely

in this genus. The reverse appears to be true with the vancomycin resistance genes, which

are found widely in E. faecium but remain exceedingly rare inStaphylococcus aureus.

An important cause of the spread of antimicrobial resistance is the failure to adhere to

appropriate infection control techniques, both within and outside the hospital. It is well

established that strains of methicillin-resist ant S. aureus (MRSA) within individual hospitals,

and even within entire cities, are often clonally related, as determined by genetic techniques

such as pulsed-field gel electrophoresis, multilocus sequence typing, and staphylococcal

protein A typing (241). The spread of these problematic pathogens has been attributed to

transmission from patient to patient, presumably by transiently or persistently colonized

health care workers (262). The primary site of S. aureus colonization is the anterior nares.

Colonization of the nares facilitates aerosol transmission of the resistant bacteria, particularly

during periods of viral upper respiratory infection in the colonized worker. It also facilitates

direct transmission, given the frequent contact between hands and nose in many people and

the frequently poor hand washing practices of health care workers. The clinical consequences

of patient colonization can be significant. Studies have shown a correlation between patient

colonization with MRSA and subsequent infection during periods of high risk, such as the

postoperative period (210).

Although antibiotic resistance is predominantly a nosocomial problem, resistant bacteria are

also spread in the community setting. Sites in which resistant bacteria have been known to

spread include day care centers and nursing homes (2, 310). Penicillin-resistant

pneumococci have been found to colonize as many as 25% of a day care center’s population.

Transmission probably reaches its peak in the winter months, when viral upper respiratory

infections are prevalent. The prevalence of viral upper respiratory infections works in two

ways to increase transmission: (i) it probably increases the inoculum of resistant organisms

being spread by those already colonized (see above references for S. aureus colonization),

and (ii) it makes those who are not colonized more likely to become colonized because of the

increased likelihood that they will receive antimicrobial therapy. Nursing homes are

predisposed to resistance for a variety of reasons, including the debilitated state of much of

their population, frequent movement back and forth to tertiary care hospitals, and frequent

use of antimicrobial agents in an effort to ward off infections that necessitate hospital

admissions.

A final important source of the emergence and spread of antibiotic-resistant bacteria is

nonhuman niches in which antibiotics are used. It is now well established that antimicrobial

use in food animals is associated with both resistance in bacterial species that contaminate

food and infect humans, primarily Salmonella andCampylobacter, and the transfer of

resistance determinants to their human counterparts, such as Enterococcus(80). Compelling

evidence also exists that high rates of ciprofloxacin resistance in E. coli can be associated

with the use of fluoroquinolones in poultry (75). Finally, the European outbreak of

vancomycin-resistant enterococci with the vanA determinant in the 1990s was almost

certainly fueled by the use of avoparcin (a glycopeptide antibiotic) as a growth promoter in

food animals (302). Data are also emerging that the use of antibiotics to promote growth of

animals is often expensive and unnecessary, which has prompted many stakeholders in this

issue to outline specific instances in which such antimicrobial use will be permitted.

Genetic Bases of Resistance

Acquired antimicrobial resistance results from biochemical processes that are encoded by

bacterial genes. A general list of mechanisms of resistance is presented in Table 1. In order

to understand the biochemical processes, it is useful to first discuss the genetic

underpinnings of resistance and its evolution. Antimicrobial resistance arises by (i) mutation

of cellular genes, (ii) acquisition of exogenous resistance genes, or (iii) mutation of acquired

genes.

Mutation of Cellular Genes

All antibiotics have targets, which are often (but not always) proteins with important

functional responsibilities for cell growth or maintenance. Cellular genes encode these

proteins. Interactions between antibiotics and target proteins are often quite specific, and

changing a single amino acid, frequently as a result of a single base change in the gene, can

sometimes alter these interactions. Perhaps the most familiar example of this mechanism is

resistance to rifampin. Rifampin targets the cellular RNA polymerase (encoded by rpoB), and

a single point mutation in this gene may confer complete resistance. These mutations occur

in most bacterial species at a relatively high frequency (ca. 10−8/CFU). Incubating enough

cells with inhibitory concentrations of rifampin eliminates susceptible cells and allows the

resistant mutants to proliferate. The rifampin in the medium does not actually cause

resistance but, rather, selects mutants that occur naturally but which have no selective

advantage for survival in the absence of rifampin in the environment. Other examples of

mutational resistance include resistance to streptomycin by ribosomal mutation (74),

resistance to fluoroquinolones through mutations of cellular topoisomerases (167), and

resistance to linezolid by mutations in the rRNA (224), among others.

Resistance mutations may also be found in genes that regulate cellular processes. Perhaps

the most completely studied example of regulatory mutation resulting in resistance is the

derepression of the chromosomal β-lactamase of Enterobacter spp. (127). Mutations in a

cellular amidase gene (designated ampD) result in buildup of a cell wall breakdown product

that has the effect of dramatically increasing expression of a chromosomal β-lactamase

gene (ampC). Other examples of regulatory changes include the downregulation of

expression of the porin OMPD2 in Pseudomonas aeruginosa associated with resistance to

imipenem (157), or the insertion of an insertion sequence (IS) element upstream of a

chromosomal carbapenemase conferring imipenem resistance on Bacteroides fragilis (214).

Whether mutational resistance is likely to persist depends in some measure on whether the

resistance mutation is tolerable to the cell. For example, although decreased expression of

OMPD2 appears to be readily achievable for P. aeruginosa, the fact that these resistant

strains have not spread widely in the nearly 20 years of carbapenem use probably reflects

the fact that this porin has functions that are beneficial to the bacterium, favoring

reexpression of the porin once the imipenem threat has been dissipated. Similarly,

intermediate levels of susceptibility to vancomycin in S. aureus have thus far been attributed

to marked changes in the composition of the cell wall (265). These changes are unlikely to

be favored in an environment free of vancomycin, since S. aureus likely “decided” a long

time ago the optimal size and composition of its cell wall. The deleterious effects of acquiring

resistance are often referred to as fitness cost.

Disadvantageous resistance mutations do not always disappear. Although initial point

mutations in the rpoBgene that confer rifampin resistance on Salmonella enterica serotype

Typhimurium appear to decrease the fitness of the organism for survival in vivo, persistence

in a live host is frequently associated with compensatory mutations that at least partially

restore fitness to the strain while retaining the resistance (rather than mutating back to

susceptibility) (24). Similarly, transfer of mutated pbp5 into E. faecium strains is often

associated with decreases in the expression of ampicillin resistance, but growth on increased

concentrations of ampicillin easily yields colonies that grow well at higher concentrations

(237). Similar findings have been reported for S. aureus strains transformed with

the mecA gene, encoding methicillin resistance (181). In summary, while mutational

resistance often confers a fitness cost, subsequent adaptations may make expression of

resistance less costly.

Acquisition of Resistance Genes

If resistance is not achievable through mutation, resistance determinants can be acquired.

Most antimicrobial agents are natural products or derivatives of natural products. Therefore,

resistance genes for most antibiotics must exist in the microbial world, either in the species

that produce the antibiotic or within species that live in the same ecological niche as the

antibiotic producers (62). The challenge for susceptible human pathogens is to find and

acquire these resistance determinants. To assist in this acquisition, bacteria have evolved a

range of mechanisms that promote gene exchange. Perhaps the simplest of these techniques

is natural transformation, referring to the ability of some bacterial species to absorb naked

DNA molecules from the environment under the appropriate circumstances (108). Once

taken up by the susceptible bacterium, these foreign pieces of DNA enter the bacterial

chromosome by recombining across regions of sufficient homology. In some cases, functional

genes result from this recombination. If the acquired gene encodes a protein that is less

susceptible to inhibition than the native protein, a reduction in susceptibility may result.

Perhaps the best-studied example of the formation of these “mosaic” genes to confer

resistance is penicillin and cephalosporin resistance in Streptococcus pneumoniae (108). A

variety of mosaic pbp genes have been described to occur in resistant strains, with the level

and degree of resistance determined by the number and nature of gene recombinations.

Mosaic topoisomerase genes have also been described to occur in fluoroquinolone-resistant

transformable bacteria (63).

Most bacteria are incapable of natural transformation and so have developed other

mechanisms for acquiring useful genetic determinants. A commonly employed mechanism

for genetic exchange is the transfer of conjugative plasmids. These extrachromosomal

replicative DNA forms may bear a variety of important genes. Some plasmids are relatively

narrow in their host range, while others transfer into and replicate within several different

species. Transfer frequencies can be very high, as in the F factor of E. coli (virtual complete

transfer in 1 h) or the pheromone-responsive plasmids found in E. faecalis (ca.

10−1 transconjugant/recipient CFU in 24 h), or more modest, as observed with the broadhost-

range enterococcal plasmids such as pAMβ1 (10−7 to 10−6 transconjugant/recipient CFU

in 24 h) (35). Having entered into a new genus on broad-host-range plasmids, resistance

determinants can readily transfer onto more frequently transferable plasmids to increase

their movement through the new genus. In the first case of true vancomycin-resistant S.

aureus described in the literature, it appears that vancomycin resistance transposon

Tn1546 entered into S. aureus fromEnterococcus faecalis on a broad-host-range plasmid and

then transposed to a conjugative plasmid native to the staphylococcus (89, 304). Plasmids

may also integrate into the chromosome of the recipient strains, potentially increasing the

stability of the genetic information they carry.

Bacteria also take advantage of bacterial viruses (bacteriophages) for genetic exchange.

These discreet packages deliver to uninfected cells a quantity of DNA approximating the size

of their genome (in most cases roughly 40 kb). Designed to incorporate their own genome

into the manufactured phage head, they sometimes incorporate bits of chromosomal DNA

adjacent to the phage integration site (specialized transduction) and other times incorporate

an appropriate-size plasmid or chromosomal DNA segment unrelated to the integrated phage

genome (generalized transduction). Since the staphylococcal β-lactamase gene is frequently

identified on nonconjugative plasmids of approximately 35 to 40 kb, and since

bacteriophages have been well described for staphylococci for decades, it has been

speculated that the high prevalence of β-lactamase production in staphylococci has resulted

from bacteriophage-mediated transfer of these plasmids. Bacteriophages have also been

implicated in the transfer of virulence determinants (82).

Nonreplicative mobile elements known as transposons have also been implicated in the

transfer of resistance genes (232). Transposons encode their own ability to transfer between

replicons (autonomously replicating DNA segments). In some cases, the transposons

themselves encode conjugation functions, which allow them to transfer from bacterial

chromosome to bacterial chromosome. The best characterized of these conjugative

transposons is Tn916, an 18-kb element originally described for E. faecalis but which has a

very broad host range (233). Tn916 encodes resistance to tetracycline and minocycline

through the tet(M) resistance gene. Many different Tn916-like transposons have now been

described for enterococci and other organisms, and some of them, such as Tn1545 from S.

pneumoniae, possess additional resistance genes (conferring resistance to erythromycin and

kanamycin) (233). Some investigators have suggested that the conjugation events

associated with Tn916-like transposons are akin to cell fusion events, in which portions of

the genome distinct from that adjacent to the inserted transposon can exchange via

homologous recombination (284). Transposons with structural similarity to Tn916 have been

implicated in the transfer of vancomycin resistance between E. faecium strains (40).

Transposons lacking conjugative functions may also transfer between strains. The most

common mechanism by which this transfer is presumed to occur is either transient or more

permanent integration into transferable plasmids. Among the more common classes of

nonconjugative transposons are the Tn3 family elements (including Tn917, conferring

erythromycin resistance, and Tn1546, conferring VanA-type vancomycin resistance)

(14, 257) and the composite elements formed by mobile IS elements flanking resistance

genes (including Tn4001, conferring high-level gentamicin resistance on many gram-positive

species) (94).

In recent years the importance of “common regions” has been increasingly recognized (282).

These regions, which are often found near the 3′ conserved regions of integrons, have been

implicated in the movement of a range of different antimicrobial resistance determinants.

These regions represent an atypical type of IS element (IS91-like). They lack typical inverted

repeats and transpose by a rolling-circle mechanism that allows them to transpose adjacent

DNA without a flanking second copy of the element.

The precise origin of resistance genes is often difficult to discern, but in some cases it is at

least possible to determine that acquired resistance determinants originated in other genera.

The VanB-type vancomycin resistance gene in enterococci, for example, has a G+C content

of nearly 50% (78). The enterococcal genome, in contrast, has a G+C content of

approximately 35 to 38%. These differences virtually confirm the origin of

thevanB determinant in a genus other than Enterococcus. The likely origin appears to be

streptomycetes, probably species that manufacture glycopeptide antibiotics, with entry into

the enterococcus facilitated by the incorporation of these resistance operons into

transposons.

It is worth noting at this point that any concept of the bacterial genome as a fixed entity is

untenable. Comparisons of E. coli genomes reveal striking differences between

enteropathogenic and uropathogenic strains, with the different regions constituting

pathogenicity islands that confer specific virulence traits that give the strains their clinical

profiles (111, 306). Data emanating from a comparative study of 36 S. aureusgenomes

indicate that 22% of the genome is dispensable, with many of these variable regions

constituting presumed pathogenicity islands and regions of antimicrobial resistance (88).

Finally, it is estimated that 25% of the genome of E. faecalis strain V583 was acquired from

outside the genus (208).

Mutation of Acquired Genes

As bacteria have responded to the challenge of antimicrobial agents, so have we responded

to the challenge of antibiotic resistance. Our typical response to the appearance of

antimicrobial resistance has been a concerted effort to develop novel antimicrobial agents

that are active against resistant strains. The emergence of β-lactamase-mediated resistance

to antibiotics is an instructive example of this interplay. Ampicillin was developed as the first

penicillin with clinically significant activity against gram-negative rods, primarily E.

coli. Within a few years of the clinical introduction of ampicillin, strains of E. coli were

described that were resistant to this antibiotic by virtue of production of a plasmid-mediated

β-lactamase designated TEM (named after the patient from whom the resistant strain was

isolated). S. aureus expressed a similar β-lactamase, prompting a concerted effort on the

part of the pharmaceutical industry to develop β-lactam antibiotics resistant to hydrolysis.

Among the more successful compounds that were developed were methicillin (with activity

against β-lactamase-producing S. aureus), the cephalosporins and carbapenems (with

widespread activity against many β-lactamase-producing species), and the β-lactamase

inhibitors, which restored the activity of β-lactams susceptible to hydrolysis.

The most successful and widely developed class of β-lactamase-resistant β-lactam antibiotics

is the cephalosporins. So many of these agents have been developed for clinical use that

they are frequently lumped into “generations” to facilitate remembering their spectra of

activity. The third-generation, or extended-spectrum, cephalosporins cefotaxime,

ceftizoxime, ceftriaxone, and ceftazidime are particularly potent antibiotics that are resistant

to hydrolysis by the original TEM enzyme. Unfortunately, increasing clinical use of these

agents, particularly ceftazidime, was associated with the emergence of resistant gramnegative

rods, particularly K. pneumoniae (131). Molecular analysis of these resistant strains

revealed that the resistance was mediated by β-lactamase and that many of these β-

lactamases were derived from the native TEM enzyme through one or more point mutations

in the bla TEM gene.

Biochemical Mechanisms of Resistance

Modification of the Antibiotic

Many antibiotic-modifying enzymes have been described, including the β-lactamases, the

AMEs, and chloramphenicol acetyltransferases (CATs). Although these enzymes are in many

cases acquired, some are intrinsic to certain genera. For example, chromosomal β-

lactamases are intrinsic to almost all gram-negative rods. Expression of these enzymes is

often only at a very low level, conferring resistance to only very susceptible β-lactams, as

with K. pneumoniae resistance to ampicillin through expression of the chromosomal SHV-1

enzyme (236), or to no β-lactams at all, as with wild-type E. coli strains. In some bacterial

genera (notably Enterobacter and Pseudomonas), chromosomal β-lactamases are under

regulatory control, with derangements in these regulatory mechanisms resulting in highlevel,

broad-spectrum β-lactam resistance (128). In some instances AMEs are intrinsic to

bacterial species as well, as with the chromosomal acetyltransferases of Providencia

stuartii and Serratia marcescens (231, 258).

Modifying enzymes in general confer high levels of resistance to the antibiotics against which

they have activity. Expression of the TEM-1 β-lactamase by E. coli, for example, can increase

the ampicillin MIC from 8 μg/ml to >10,000 μg/ml. Similarly, expression of the bifunctional

aminoglycoside resistance enzyme in E. faecalis raises the gentamicin MIC from 32 to 64

μg/ml to >2,000 μg/ml. As effective as these mechanisms are, however, some antibiotics

appear to be immune to inactivating enzymes. Vancomycin has been in clinical use since

1958, yet there are still no examples of vancomycin-modifying enzymes in bacteria.

Modification of the Target Molecule

Since antibiotic interaction with target molecules is generally quite specific, minor alterations

of the target molecule can have important effects on antibiotic binding. Numerous examples

exist of antibiotic target modification as a mechanism of resistance, including the many

erythromycin ribosomal methylases that confer resistance to the macrolide- lincosamidestreptogramin

B (MLSB) class of antibiotics (305). Modifications of PBPs can affect the

affinities of these molecules for β-lactam antibiotics, as noted above for S. pneumoniae,and

especially for ampicillin-resistant E. faecium through mutations in PBP 5 (108, 237).

Modifications of PBPs seem to be a favored mechanism of β-lactam resistance in grampositive

bacteria, whereas β-lactamase production is favored in gram-negative rods.

Although the reason for this difference is unknown, it is interesting that β-lactamases

produced by gram-positive bacteria diffuse into the external medium once produced,

whereas those produced by gram-negative rods are kept within the periplasmic space by the

outer membrane. The ability to concentrate β-lactamases enhances their efficacy and may

help explain the preference for this mechanism among gram-negative rods.

Other important examples of target modifications include the altered cell wall precursors that

confer resistance to glycopeptide antibiotics, mutated DNA gyrase and topoisomerase IV

conferring resistance to fluoroquinolone antimicrobial agents, ribosomal protection

mechanisms conferring resistance to tetracyclines, and RNA polymerase mutations conferring

resistance to rifampin. The degree of resistance conferred by target modifications is variable

and may be dependent upon the ability of the mutated target to perform its normal function.

Mutations in PBPs of S. pneumoniae, for example, confer a relatively low level of resistance

(although one that is significant in the treatment of meningitis) (108), whereas VanA-type

vancomycin resistance confers a very high level of resistance to vancomycin in enterococci

(15).

Restricted Access to the Target

It is axiomatic that an antibiotic must reach its target in order to be effective. Therefore, for

targets for which barriers must be crossed by the antibiotic, strengthening these barriers can

be a highly effective mechanism of resistance. All gram-negative bacteria have an outer

membrane that must be traversed before the cytoplasmic membrane can be reached.

Reductions in the quantities of known or presumed porins (channels for movement of

materials across the outer membrane) have been documented as important contributors to

resistance to imipenem in P. aeruginosa, cefepime in Enterobacter cloacae, and cefoxitin or

ceftazidime in K. pneumoniae (151,157, 168). In most instances, this restricted entry must

be in combination with production of an at least moderately active β-lactamase to confer

high-level resistance. Barriers to entry can also exist in the cytoplasmic membrane.

Movement of aminoglycosides across the cytoplasmic membrane is an oxygen-dependent

process, so these antibiotics are inactive in anaerobic environments (and hence against

strictly anaerobic species) (149).

Efflux Pumps

Among the most active areas of research in antimicrobial resistance is the identification and

characterization of pumps that extrude one or more antibiotic classes from the bacterial cell.

Several classes of pumps have been described for gram-positive and/or gram-negative

bacteria. They may be quite selective, or they may have a broad substrate specificity. The

majority of these pumps are located in the cytoplasmic membrane and use proton motive

force to drive drug efflux. The major families of efflux transporters are (i) the major

facilitator superfamily (MFS), which includes QacA and NorA/Bmr of gram-positive bacteria

and EmrB of E. coli;(ii) the small multidrug resistance family, including Smr of S. aureus and

EmrE of E. coli; and (iii) the resistance- nodulation-cell division (RND) family, including

AcrAB-TolC of E. coli and MexAB-OprM of P. aeruginosa. The structure of the AcrAB-TolC

RND-type efflux pump is shown in Fig. 2. Deciphering the crystal structure of this pump was

a major achievement (182). Among other things, it revealed that there was a periplasmic

opening in the pump that could allow passage of molecules, explaining the previously

confusing observation that RND pumps included β-lactam antibiotics (which do not enter the

cytoplasm) among their substrates. In some instances, combinations of different types of

pumps can result in higher levels of resistance than are achieved by the activity of a single

pump alone (150).

β-Lactamase-Mediated Resistance

Classification of β-Lactamases

Two major schemes are currently used to classify β-lactamases: the Ambler classification

system and the Bush-Jacoby classification system (35a). The Ambler classification separates

β-lactamases into four distinct classes (A to D) based on similarities in amino acid sequence.

Classes A, C, and D are serine β-lactamases, whereas class B enzymes are metallo-β-

lactamases that require one or two zinc atoms for activity (see above). The Bush-Jacoby

classification system (formerly referred to as Bush-Medeiros-Jacoby scheme [7, 36,37])

classifies β-lactamases according to functional similarities (substrate and inhibitor profiles).

There are four categories (groups) and multiple subgroups in the updated Bush-Jacoby

system (groups 1, 2 [2a, 2b, 2br, 2d, 2be, 2c, and 2f ], etc.) (36). A comparison of the two

classification systems is summarized in Table 2. In the discussions that ensue, we refer to

both classification systems.

ESBLs of the TEM family, 36 inhibitor-resistant TEMs (IRTs), 9 complex mutants of TEM

(CMTs), 126 SHVs, more than 160 OXA enzymes, and 91 CTX-M ESBLs. The number of class

C β-lactamases has also increased to 44, and the numbers of IMP and VIM metallo- β-

lactamases have increased to 25 and 23, respectively (these families are discussed below).

Most worrisome is the number of β-lactamases that are able to hydrolyze carbapenems. As

these carbapenemases (e.g., Klebsiella pneumoniaecarbapenemase [KPC]) are being

recovered from clinical isolates, the threat to our “last line” therapy is ever increasing.

Presently, there are nine variants of KPC β-lactamases. The reader is referred to the

following website for updates: http://www.lahey.org/studies.

β-Lactamase Mechanism

β-Lactamases are members of a superfamily of active-site serine proteases or D,Dpeptidases

(170). The mechanism of hydrolysis of β-lactams by β-lactamases is best studied

for TEM-1. TEM-1 β-lactamase disrupts the amide bond of a β-lactam in a two-step reaction.

First, the negatively charged carboxylate group of the β-lactam antibiotic is attracted to the

active site by the enzyme’s positively charged residues. There, the β-lactam is properly

positioned, making key hydrogen bonding interactions with the enzyme (142). The residues

in the active site that facilitate this attraction in the serine β-lactamases are often called the

oxyanion hole or electrophilic center. Next, the β-lactam is acylated (Fig. 1). A conserved

serine, Ser70, in the active site of TEM-1 serves as the reactive nucleophile in this acylation

reaction. Recent ultrahigh-resolution x-ray crystallography studies of TEM-1 (0.85 A) indicate

that Glu166, acting through the catalytic water, activates Ser70 for nucleophilic attack of the

β-lactam ring. Then, a strategically positioned water molecule is activated by a general base

(e.g., again Glu166 and the same water molecule). This water molecule deacylates the β-

lactam and regenerates the active β-lactamase. This symmetric mechanism is also supported

by the ultrahigh-resolution structure (0.90 A) of another common class A β-lactamase, SHV-

2, found in K. pneumoniae (196). There is general consensus on the mechanics of

deacylation by the class A TEM-1 β-lactamase (attention to Glu166), but the details of

acylation still remain contentious. Debate centers on which residue in the active site (Lys73

or Glu166) deprotonates the reactive Ser70 or whether either pathway is in competition with

the other (174).

β-Lactamase Processing

Generally speaking, β-lactamases are secreted into the periplasmic space in gram-negative

bacteria or into the surrounding medium by gram-positive organisms. Membrane-associated

enzymes have been rarely reported (Bacillus licheniformis, Bacillus cereus, and Bacteroides

vulgatus). β-Lactamases are synthesized as precursor proteins. As in other bacteria, the

export of proteins into the periplasmic space is mediated by an amino-terminal signal

peptide. After transport, the signal peptide of the β-lactamase is cleaved by a processing

enzyme, signal peptidase I.

Genetic Environment of β-Lactamases

β-Lactamases can be chromosome-, plasmid-, or transposon-encoded enzymes that are

produced in a constitutive or inducible manner. An increasing number of β-lactamases have

been found that are encoded on integrons.

Integrons are genetic elements of variable length that contain a 5′ conserved integrase

gene (int), mobile antibiotic resistance genes (called cassettes), and an integration site for

the gene cassette, attI (att, attachment). To date, five distinct integron classes have been

found to be associated with cassettes that contain antibiotic resistance genes. Three main

classes of integrons (classes 1 to 3) have been described for gram-negative bacteria.

Integrons capture antibiotic resistance gene cassettes by using a site-specific recombination

mechanism. In the class 1 integrons, the 3′ conserved segment includes three open reading

frames, qacEΔ1, a deletion derivative of the antiseptic resistance gene qacE, and sul1, a

sulfonamide resistance gene. As integrons carry multiple resistance determinants and can be

readily mobilized, their impact on antibiotic resistance is significant. In the words of Hall and

Collis, “integrons thus act both as natural cloning systems and as expression vectors” (109).

The capture and spread of antibiotic resistance determinants by integrons underlie the rapid

evolution of multiple antibiotic resistance among diverse gram-negative clinical isolates

(244). As reviewed by Stokes and Hall (273), integrons were originally found on mobile

elements from pathogenic bacteria and were found to be a major reservoir of antibiotic

resistance genes. Analysis of their gene content suggests that integrons are phylogenetically

diverse and have been with bacteria for a long time. Interestingly, integrons have been

found in approximately 9% of sequenced bacterial genomes. It is maintained that the

integron/gene cassette system has a function in evolution rather than just conferring

resistance to antibiotics. Consequently, integrons may be the agents of change that drive

bacterial evolution and adaptation. It should be noted that gene cassettes are mobile and

can also exist in free circular form. However, these cassettes do not include all functions

required for their mobility. Cassettes are formally part of the integron only when they are

integrated at the integron receptor site. The genes within the cassette do not have

promoters.

Integrons are an important source for the spread of bla genes. Integrons containing β-

lactamases have been found in A. baumannii, P. aeruginosa, and other species of gramnegative

bacteria encompassing Ambler class A, B, and D β-lactamases (307). The β-

lactamase enzymes/families found on integrons are VEB-1, VEB-2, GES-1, GES-2, IBI-1, IBI-

1, CTX-M-2, CTX-M-9, PSE-1, and numerous OXA β-lactamases (189, 216, 217). OXA and

metallo-β-lactamases that confer resistance to carbapenems (IMP-1 to IMP-4, IMP-6 to IMP-

8, IMP-12, VIM-1, VIM-2, and GIM-1) are also included in integron-encoded β-lactamases

(45, 307).

Class A β-Lactamase (Bush Group 2b Penicillinases)

Class A β-lactamases possess four important structural motifs that create a complex

hydrogen-bonding network to fix the β-lactam in the substrate-binding pocket. Residues

Ser70-Xaa-Xaa-Lys73, Ser130-Asp131-Asn132 (SDN loop), and Lys/Arg234-Thr/Ser235-

Gly236 define the conserved residues critical for β-lactam binding and hydrolysis (142). the

Ω loop (amino acids Arg164 to Asn179) is unique in class A β-lactamases. A highly conserved

Glu166 that functions as a general base (electron donor) in the catalytic process is located in

the Ω loop (see above). The salt bridge formed between Arg164 and Asn179 define the

limits, or “neck,” of the Ω loop.

The two commonly encountered class A β-lactamases found in Enterobacteriaceae are

designated TEM-1 and SHV-1. TEM-1 and SHV-1 β-lactamases are primarily penicillinases

with diminished activity against cephalosporin substrates. These two families of β-lactamases

have received considerable attention over the past three decades since they are the

progenitors of the ESBLs and IRT β-lactamases, now common in many hospitals.

Bush group 2be. ESBLs are generally class A β-lactamases that have “expanded” or

changed their substrate profile as a result of amino acid substitutions. Normally, extendedspectrum

cephalosporins are very poor substrates for hydrolysis by Bush group 2be enzymes

(high Km and low kcat). Extended-spectrum cephalosporins are extremely potent β-lactams.

Mutations at critical amino acids expand the spectrum of these enzymes and allow the

hydrolysis of extended-spectrum cephalosporins (211). In most cases, ESBL mutations

render the enzymes more susceptible to inhibition by mechanism-based inactivators

(clavulanic acid, sulbactam, and tazobactam). Until recently, an explanation for this

increased susceptibility was not apparent. Studying reactions using Raman spectroscopy and

stopped-flow kinetics, it is now maintained that increased amounts of certain key

intermediates of SHV-2 and SHV-5 are formed (i.e., the enamine intermediate). As a result,

the Ki values of the mechanism-based inhibitors are reduced (increased susceptibility to

inhibitors) up to 50-fold in SHV-2 and SHV-5. The impact of enteric gram-negative rods

possessing ESBLs on the choice of empirical and definitive antimicrobial therapy has been

substantial (143, 158, 203, 205, 206).

Among the TEM family enzymes, five amino acid residues appear to be most important for

conferring the ESBL phenotype: Gly238 and Ala237 (located on the b3 β-pleated sheet),

Arg164 and Asp179 (located on the neck of the Ω loop), and Asp104 (located directly across

from G238 and A237 at the opening of the active-site cavity) (142, 228). Of note, the

substitution of Gly to Ser, Ala, or Asp at Ambler position (ABL) 238 is a common mutation in

both TEM and SHV ESBLs (http://www.lahey.org/studies/webt.html).

Non-TEM, non-SHV ESBLs. Among the non-TEM, non-SHV ESBLs, the CTX-M β-lactamases

are the most prevalent. They can be divided into distinct clusters

(see www.lahey.org/studies/webt.html). Unlike most (but not all) TEM- and SHV-derived

ESBLs, CTX-M β-lactamases hydrolyze cefotaxime and ceftriaxone better than they do

ceftazidime. Current data show that CTX-M enzymes are more readily inhibited by

tazobactam than they are by clavulanic acid. The first CTX-M-type β-lactamase (MEN-1) was

described nearly a decade ago. There are now nearly 60 members of this family.

CTX-M β-lactamases are commonly found in K. pneumoniae, E. coli, typhoidal and

nontyphoidal Salmonella,Shigella spp., Citrobacter freundii, Enterobacter spp., and Serratia

marcescens (141, 146, 204). The chromosome-encoded β-lactamase of Kluyvera

ascorbata is a probable progenitor for some plasmid-encoded CTX-M enzymes (141). Other

CTX-M β-lactamases may derive from Kluyvera georgiana (199). A recent comprehensive

review delineates the lineage of many CTX-M enzymes resulting from Kluyvera spp. (27). Of

note, different genetic elements may be involved in the mobilization of bla CTX-M genes.

Many other clinically important non-TEM, non-SHV ESBLs have been described (K1, GES-1,

PER-1, PER-2, VEB-1, BES-1, IBI-1, IBI-2, and OXA-type) (30, 31, 219).

Structural biology of ESBLs. Important insights have emerged from the study of a number

of atomic structures of class A ESBLs. The common theme that emerges is that the active

site is selectively remodeled and expanded to accommodate the bulky R1 side chain of

extended-spectrum cephalosporin. Although the details of this modification are different for

many of the ESBLs, the remodeling comes at a price. Many of these ESBLs are not as

catalytically efficient (e.g., redcuced kcat/Km) as are the wild-type progenitors against certain

substrates. With the expanded substrate spectrum, one uniformly observes decreases in

penicillin MICs and inkcat/Km ratios. In addition, these enzymes are less stable to proteolysis

and heat. The structures of Toho-1, TEM-52, TEM-64, the Gly238Ala ESBL in TEM, SHV-2,

and K1 β-lactamases all reveal insights into why expanded-spectrum cephalosporins fit in the

active site (49, 124, 125, 196, 200, 263, 287, 300).

Serine carbapenemases of Bush group 2f or class A type. In the past, β-lactamases

able to hydrolyze carbapenems were rare. It is regrettable that this is no longer the case.

Representatives of these carbapenemases are Sme-1, Sme-2, NMC-A, IMI-1, GES-2, and

KPC-2 (217, 317). Usually, class A carbapenemases hydrolyze imipenem but are not

resistant to clavulanic acid inhibition (the exception may be KPC enzymes). blaNMCA

and blaIMI-1 are chromosomally located genes in Enterobacter cloacae and are induced by

cefoxitin and imipenem (194, 229). bla NMC-A is regulated by a LysR-type regulatory protein

(186). It is felt that blaIMI-1 is also regulated in the same manner (229). blaSme-1 is a

chromosomally located gene encoding serine carbapenemase of class A in E.

cloacae and Serratia marcescens.

A notable increase in bacteria expressing class A carbapenemases has occurred in the United

States. Numerous studies are revealing that KPC β-lactamases are becoming endemic in East

Coast cities. First found in K. pneumoniae, these β-lactamases have been detected

in Salmonella enterica serotype Cubana, Klebsiella oxytoca, and Enterobacter sp. strain MS

412 (121, 177, 317, 318), among others. bla KPC-2 is located on a nonconjugative plasmid

(317). In Salmonella and K. oxytoca,blaKPC-2 was isolated from conjugative plasmids

(177, 318). The plasmid bearing blaKPC-3 from New York was also transferable by conjugation

(314). Currently,blaKPC has been localized to two genetic elements, Tn4401 and the KQ

element (186, 238). K. pneumoniaeorganisms bearing KPCs are now found in North and

South America, the Caribbean, Europe, Israel, and Asia. These isolates are highly resistant to

penicillins, cephalosporins, and commercially available β-lactam–β-lactamase inhibitor

combinations and show reduced susceptibility to carbapenems.

Detection of KPC β-lactamases may be a problem for clinical laboratories because of the

positive ESBL confirmation tests (clavulanate-potentiated activities of ceftriaxone,

ceftazidime, cefepime, and aztreonam). To improve detection of KPC-expressing K.

pneumoniae, care in inoculum preparation for broth-based susceptibility methods must be

taken. Investigators in this area have found that resistant isolates were readily detected by

agar-based methods but not by broth methods (33). Bratu et al. first found that using

ertapenem or meropenem improves detection (33).

We stress that the clinical detection systems (i.e., automated systems like Vitek 1 and 2,

Phoenix, and MicroScan) that are currently used by most hospital laboratories may not be

completely accurate for detection of KPC in all strains of K. pneumoniae. Thus, the true

prevalence of KPC- producing isolates may be underappreciated, and therapeutic decisions

based upon susceptibility testing can be adversely affected. We now generally recognize that

screening with ertapenem increases the sensitivity for detecting KPCs by Etest, disk

diffusion, and automated methods, while testing with meropenem and imipenem increases

specificity. Currently, the major phenotypic test that is used is the modified Hodge test

(MHT). In this test, a lawn of susceptible E. coli (e.g., strain ATCC 25922), a carbapenem

disk (ertapenem or meropenem), and a streak of the suspected KPC producer are placed

upon an agar plate and incubated overnight. Growth up to the disk is checked the next day

and, if present, suggests carbapenemase production. Unfortunately, there may be false

positivities. Chromogenic agar tests are also now employed in selected circumstances, but

clinical experience is limited. Chromogenic agar for KPC detection (CHROMagar KPC)

demonstrated a sensitivity and specificity of 100 and 98.4% compared with PCR for blaKPC for

detection of KPC-producing Enterobacteriaceae directly from rectal swab specimens. The

rectal swabs were obtained from 122 unique patients (41 KPC-producing K.

pneumoniae isolates were detected). The study had certain limitations (the sample size was

small, and the study was performed at one institution and involved predominantly a single

clone of carbapenem-resistant K. pneumoniae). In practical terms, the strains turn red

when E. coli producing a carbapenemase is detected and metallic blue when K.

pneumoniae produces a carbapenemase. Currently, there are two different tests available:

the CHROMagar KPC and the CHROM ID ESBL (42, 250). The latter CHROM method can

detect producers of IMP-, VIM-, and KPC-type carbapenemases with high levels of resistance

to cephalosporins and to carbapenems.

Boronic acid disks have also been used to detect KPC producers. In this assay, the boronic

acid disks (e.g., 3-aminophenylboronic acid) when combined with a carbapenem have a

senstitivity and specificity that approaches 100% (67, 288). These assays can be performed

more readily than others and can add to the current need to define the mechanism of

resistance and detect KPCs. In our experience, PCR still remains the “gold standard” and is

readily performed in many reference laboratories throughout the United States.

So far, the crystal structure of only KPC-2 has been determined. KPC-2 has an overall

structure similar to that of other class A enzymes, and, interestingly, this β-lactamase has

only 50% protein sequence conservation compared to CTX-M-1, 39% to SHV-1, and 35% to

TEM-1 (140). The KPC-2 β-lactamase possesses a large and shallow active site, allowing it to

accommodate “bulkier” β-lactams. As a result of these structural changes, KPC-2 is regarded

as a versatile β-lactamase (226). Microbiologists and clinicians observed that many blaKPC-2-

containing strains are resistant to β-lactam–β-lactamase inhibitor combinations. At present,

β-lactam therapeutic options do seem to work against these highly resistant strains.

Among these serine carbapenemases exists a notable curiosity. bla GES-2 is a plasmid-borne

β-lactamase gene found in P. aeruginosa (221). blaGES-2 is a point mutation mutant of bla GES-

1, which encodes a clavulanic acid-inhibited ESBL (a non-TEM, non-SHV ESBL). It is curious

that GES-1 is an ESBL and that a single point mutation (Gly→ Asp at position 170 in the Ω

loop) can add an imipenemase activity to GES-2 (93). The OXA carbapenemases are

discussed below.

Inhibitor-resistant class A β-lactamase: Bush group 2br. Amino acid mutations within

TEM and SHV that confer resistance to inhibition by β-lactamase inhibitors have also been

characterized at the following amino acid positions: Met69, Ser130, Arg244, Arg275, and

Asn276 (28, 64, 79, 223, 254, 276, 281, 293, 296). Most inhibitor-resistant β-lactamases

are variants of the TEM-1 enzyme, with only two descriptions of clinical isolates expressing

SHV β-lactamases that are resistant to inhibitors at the time of this writing (SHV-10 and

SHV-49) (70, 223). It is possible that the frequency of inhibitor-resistant β-lactamases in

TEM and SHV family enzymes is underestimated, since many laboratories in the United

States do not identify these strains routinely (139). Cephalosporins or high-dose piperacillintazobactam

may be effective for the treatment of E. coli expressing some inhibitor-resistant

β-lactamases.

Mutants of TEM β-lactamases are being recovered that maintain the ESBL phenotype but also

demonstrate inhibitor resistance. These are referred to as CMTs. There are now four CMTs

(84, 190, 218, 266). It is still unknown how these enzymes will affect empirical treatment. At

this time, clinical microbiology laboratories do not have the resources to detect these

complex phenotypes.

Class B β-Lactamase (Bush Group 3 Enzymes)

In contrast to the serine-dependent β-lactamases (classes A, C, and D), class B β-lactamases

are metalloenzymes. These enzymes contain an αββα motif with a central β sandwich and

two α helices on each side. Consequently, class B β-lactamases require zinc or another heavy

metal for catalysis, and their activities are inhibited by chelating agents (EDTA). This zinc

atom is held in place by three histidines and a water molecule. Some metallo-β-lactamases

contain a second Zn binding site. These two sites function separately, with the primary Zn

binding site assisted by the secondary site. A coordinated water molecule also plays a critical

role in catalysis (299).

With few exceptions (see below), class B β-lactamases confer resistance to a wide range of

β-lactam compounds, including cephamycins and carbapenems. The class B β-lactamases are

resistant to inactivation by clavulanate, sulbactam, and tazobactam. Aztreonam, a

monobactam, may act as an inhibitor, but there are metallo-β-lactamases that hydrolyze

aztreonam. Of note, isolates expressing VIM-2 can be susceptible to aztreonam (215).

Class B β-lactamases can be grouped into three different subclasses (B1, B2, and B3)

depending on their requirements for zinc. B1 enzymes (IMP-1, VIM-2, and CcrA) are fully

active with one or two zinc ions, B3 enzymes (e.g., L1) require two zinc ions, and B2

enzymes (e.g., CphA) are inhibited by the addition of a second zinc ion. Amino acid positions

can be assigned according to a specific numbering system, designated BBL, that is based

upon structural alignment (97). Although the genes encoding metallo-β-lactamases show

very little primary structure sequence identity (17 to 37%), the three-dimensional structures

of the known metallo-β-lactamases appear to be similar (39, 54, 271, 292).

Because of the metal ion, the catalytic pathway of metallo-β-lactamases does not involve an

acyl enzyme intermediate as it does in classes A and C. The catalytic pathway in class B also

incorporates a hydrolytic water molecule (the “bridging” water molecule) that possesses

enhanced nucleophilicity due to the proximity to the metal ion. The zinc ions coordinate two

water molecules. The addition of the hydroxide to the carbonyl carbon of the β-lactam leads

to the formation of a transient, noncovalent reaction intermediate. The mechanistic details of

this pathway have been recently summarized by Crowder et al. (58) and are beyond the

scope of this chapter.

The majority of metallo-β-lactamases are chromosomally encoded, and their expression may

be constitutive or inducible. The metallo-β-lactamases of B. cereus, Stenotrophomonas

maltophilia, Aeromonas hydrophila, andAeromonas jandaei are inducible. In A.

jandaei, regulation of the metallo-β-lactamase appears to involve two-component signal

transduction systems (5).

The metallo-β-lactamases of the VIM and IMP types are now established as important threats

to our antimicrobial armamentarium. These metallo-β-lactamases are broad-spectrum

enzymes and are active against most β-lactams, including carbapenems, and have been

found in various gram-negative clinical isolates mostly in the Far East and the Mediterranean

regions. bla VIM is an integron-borne metallo-β-lactamase that is usually found in P.

aeruginosa isolates. Unfortunately, the VIM metallo-β-lactamase has spread to other enteric

bacilli (E. coli,Enterobacter aerogenes, E.cloacae, and Klebsiella spp.). bla VIM-2 has now

spread to more than 20 countries (243). The majority of metallo-β-lactamases are mobilized

on integrons, transposons, and mobile common regions (299). A curious finding is the

coisolation of VIM-1 with KPC-2 in Greece (102). Similarly, IMP metallo-β-lactamases are

very widespread. IMP metallo-β-lactamases have been found as part of integrons in the

following bacteria: P. aeruginosa, Pseudomonas putida, Serratia marcescens, Pseudomonas

stutzeri, Acinetobacter baumannii, Pseudomonas fluorescens, K. pneumoniae, K. oxytoca, E.

aerogenes, Achromobacter xylosoxidans, and Escherichia coli (299). In an unusual parallel to

what has been found with VIM and KPC-2 (see above), IMP-1 has also been discovered in

combination with OXA carbapenemases (275).

Recently, a novel class B enzyme that has raised significant concern has been described.

NDM-1 (New Delhi metallo-β-lactamase) is a class B β-lactamase encoded by a very mobile

genetic element, and the pattern of spread is providing to be more complex and apparently

more unpredictable than that of the gene encoding KPC. Moreover, the number of patients

possessing bacteria containing bla NDM-1 is growing. The gene has moved from India and

Pakistan to the United Kingdom, the United States, Ken ya, Japan, Canada, Belgium, The

Netherlands, Taiwan, Oman, and Australia. bla NDM-1 has been found on plasmids of different

sizes and has been located near a pathogenicity island. The resistance determinants flanking

this bla gene seem to be numerous, including chromosomal β-lactamases (bla CMY and blaDHA)

and chloramphenicol and aminoglycoside resistance genes. The global impact of this gene

and its spread must be carefully monitored.

As stated above, the atomic structures of a number of class B β-lactamases have been

solved (39, 54, 97,98). The structures are being used to design novel inhibitors of class B β-

lactamases.

Class C β-Lactamase

Ambler class C (Bush-Jacoby-Medeiros group 1) chromosomal β-lactamases are produced to

a greater or lesser degree by almost all gram-negative bacteria

(Salmonella, Klebsiella, Proteus mirabilis, Proteus vulgaris,and Stenotrophomonas

maltophilia being the major exceptions) (129). Chromosomally encoded (and inducible)

enzymes are particularly important in clinical isolates of C. freundii, E.

aerogenes, E. cloacae, Morganella morganii,P. aeruginosa, and S. marcescens. Although

class C β-lactamases hydrolyze cephalosporins (including extended-spectrum

cephalosporins) more effectively than they do penicillins, it should be kept in mind that these

enzymes have a great efficiency for hydrolysis of penicillins (the Km is very low). Most class C

enzymes are resistant to inhibition by clavulanate, sulbactam, and tazobactam (in the case

of tazobactam, the resistance to inhibition is generally less).

Class C β-lactamases have larger active-site cavities than do class A enzymes, which may

allow them to bind the bulky extended-spectrum cephalosporins (oxyimino-β-lactams)

(57, 159). It is claimed that this conformational expansion and flexibility facilitate hydrolysis

of oxyimino-β-lactams by making the acyl enzyme intermediate more open to attack by

water (57, 159, 175).

The important structural elements described for class A enzymes are also present in class C

β-lactamases. The active-site serine (Ser64) is located near the N terminus of a long helix

and is followed on the next helix turn by a lysine (Ser64-Xaa-Xaa-Lys67). The second

element contains a Tyr-Xaa-Asn (Tyr150) or Ser-Xaa-Asn pattern corresponding to the

Ser130-Asp131-Asn132 loop of class A β-lactamases. The opposite side of the active site is

marked by Lys315/Arg/His-Thr/Ser-Gly corresponding to the KTG motif of class A enzymes.

Class C cephalosporinases acylate β-lactams in the same manner as class A enzymes do. In

many ways the reaction mechanism of class C β-lactamases remains enigmatic (51, 100).

First, the catalytic Ser64 attacks the β-lactam carbonyl carbon and the acyl enzyme forms.

The approach of the activated water molecule is different between AmpC and class A

enzymes (class C from the β-face). This implies that the deacylation mechanism is distinct.

Unlike with class A enzymes, the ring amine of the acyl enzyme of class C facilitates

deacylation. This “substrate-assisted catalysis” generally distinguishes class C from class A

(although there are reports of substrate-assisted catalysis occurring in class A TEM as well)

(34). Examining the catalytic mechanism, a “Glu166 equivalent” in class C β-lactamases is

not readily apparent; this role may be filled by the tyrosine (Tyr150) in the Tyr-Xaa-Asn

motif. The current thinking is that the conserved residue, Tyr150, acts as a general base in

the acylation mechanism by increasing the nucleophilicity of Ser64. However, Tyr150 may

stay protonated during the reaction and is thus not able to serve as an anionic base during

hydrolysis. The proton on Tyr150 helps stabilize the water’s developing negative charge.

However, mutagenesis studies do not rule out a role for Lys67 in the coordinate base

mechanism as well (50).

The first class C β-lactamase structure determined was for the AmpC cephalosporinase of C.

freundii,determined by Oefner et al. (197). The structures of P99 β-lactamase of E.

cloacae, AmpC β-lactamase from E. coli, and E. cloacae GC1 and E. cloacae 908R β-

lactamases have ensued (57, 160) (http://www.rcsb.org). The GC1 β-lactamase of E.

cloacae has improved hydrolytic activity for oxyimino-β-lactam antibiotics because of a

tripepetide insertion in the Ω loop (a tandem repeat Ala211-Val212-Arg213). In a strict

sense, this is a class C ESBL. As a result of this addition, the width of the opening of the

active-site binding cavity is larger and the substrate spectrum has expanded (57). As stated

in the paragraph above, a number of other structures of the E. coli class C β-lactamase have

been solved in an attempt to decipher the mechanism of catalysis. In the clinically important

class C enzyme-producing gram-negative rods, β-lactamase production is normally

repressed. The details of the repression have been mostly elucidated for Enterobacter spp.

(127). Repression and activation are closely linked to the processes of cell wall synthesis and

breakdown. The molecule that serves as both the repressor and the activator

of ampC transcription is AmpR, a transcriptional regulator of the LysR family. AmpR is

present as a repressor by virtue of its interaction with UDP-MurNAc pentapeptide, a

peptidoglycan precursor molecule. In this form AmpR is incapable of activating ampC and in

fact serves as a repressor of ampR expression. In the setting of high concentrations of cell

wall breakdown product anhydro-MurNAc tripeptide (or anhydro-MurNAc pentapeptide),

however, UDP-MurNAc pentapeptide is displaced from its site in AmpR, resulting in the

conversion of AmpR to an activator of ampC transcription.

Increases in ampC expression may result from the action of β-lactam antibiotics, certain of

which cause the release of significant quantities of anhydro-MurNAc tri peptide and/or

pentapeptide from the peptidoglycan. This anhydro-UDP-MurNAc tripeptide enters the cell

through a channel (AmpG) and overwhelms the recycling ability of the cytosolic amidase

(AmpD) specific for recycling of muropeptides. Under these circumstances (induction), β-

lactamase is produced only as long as the antibiotic is present in the medium.

Constitutive high-level production of AmpC β-lactamase most commonly results from a

mutation in the ampDgene, reducing the quantity of (or eliminating) AmpD from the

cytoplasm. Under these circumstances, a constant high level of anhydro-MurNAc tripeptide is

present in the cytoplasm, and AmpR serves as a constitutive activator of ampC transcription.

Constitutive production can also result from deletion of ampR, but in this circumstance β-

lactamase production is generally at a low level.

The widespread dissemination of ampC-type β-lactamase genes on transferable plasmids is a

continuing challenge. These plasmid-encoded AmpC cephalosporinases are separated into

four general groups. Group 1 plasmid-encoded AmpC cephalosporinases consist of those

which originated from the chromosomal AmpC of C. freundii (BIL-1, CMY-2, LAT-1, and LAT-

2). Members of group 2 are related to the chromosomal cephalosporinase of E. cloacae (MIR-

1 and ACT-1), group 3 β-lactamases belong to the AmpC of P. aeruginosa(CMY-1, FOX-1,

and MOX-1), and group 4 enzymes belong to the CMY-1 β-lactamase (CMY-1 cluster).

Plasmid-mediated class C β-lactamases have been described to occur in many gram-negative

organisms from all parts of the world. Host strains harboring these enzymes include K.

pneumoniae, E. aerogenes, Salmonella entericaserotype Senftenberg, E. coli, Proteus

mirabilis, M. morganii, and K. oxytoca. The loss of porin proteins in clinical isolates with

plasmid-encoded AmpC enzymes may result in resistance to carbapenems (32, 273).

Class D β-Lactamases

The OXA-type (oxacillin-hydrolyzing) β-lactamases have been most commonly described

for Enterobacteriaceae, Acinetobacter spp., and P. aeruginosa (187). In terms of their

genetic background in these gram-negative organisms, many class D β-lactamase genes are

associated with class 1 integrons or with ISs (220). OXA enzymes confer resistance to a wide

variety of penicillins. They are only weakly inhibited by clavulanic acid. There are some

interesting exceptions to this rule in that OXA-2 and OXA-32 are inhibited by tazobactam but

not sulbactam or clavulanate. In an unusual manner, OXA β-lactamases are inhibited by

sodium chloride (50 to 75 mM NaCl). Mutagenesis studies suggest that susceptibility to

inhibition by NaCl is related to the presence of a Tyr residue at position 144. Overall, the

amino acid identities between class D and class A or class C β-lactamases are less than 20%.

Their frequent location on mobile genetic elements (plasmids or integrons) facilitates spread

(188, 216, 297).

Several OXA β-lactamases (OXA-11 and OXA-14 to OXA-20) are associated with an ESBL

phenotype. These OXA types have been found exclusively in P. aeruginosa. A comparison of

the crystal structure of the OXA-10 β-lactamase with those of the class C enzyme from E.

cloacae P99 and of the class A TEM-1 enzyme from E. colishows that the class D and class A

enzymes share an common fold (α helices and β-pleated sheets), although the distribution of

secondary structure elements is different. A remarkable feature is the nearly perfect

symmetry of all atoms that constitute the catalytic machinery for acylation (Ser67, Lys70,

Ser115, and Lys205 in OXA-10 and Ser70, Lys73, Ser130, and Lys234 in TEM-1). There

seems to be an extension of the substrate-binding site (tripeptide strand) in OXA-10. The

role of the peptide extension is not known. The “oxyanion hole” is provided by the mainchain

nitrogen atoms of Ser67 and Phe208. In class D, the same residue (Lys70) is involved

in acylation and deacylation (172, 201).

OXA enzymes are assuming greater importance due to the ability of members of this class to

hydrolyze carbapenems. The first description of a serine carbapenemase in A. baumannii was

ARI-1 (OXA-23) in 1985 (207). Although OXA carbapenemases hydrolyze imipenem

inefficiently, their presence in an organism with an active efflux pump or a porin mutation

may confer clinically significant levels of resistance (117). It is notable that A.

baumannii isolates possess a chromosomally encoded oxacillinase, OXA-69, that confers

very-low-level imipenem resistance. This gene is ubiquitous in A. baumannii and is referred

to as a housekeeping gene (116).

Our understanding of the hydrolytic mechanism of class D β-lactamases is based on the

careful study of OXA-10, -13, and -1. OXA β-lactamases are unique because of the direct

role of carboxylation of the active-site Lys70. The carbamic acid on Lys70 can ionize to yield

a carbamate that hydrogen bonds with the nucleophilic Ser67 residue. In this manner, the

carboxylated Lys70 may serve as the general base by activating both Ser67 for acylation and

the hydrolytic water for deacylation.

More relevant to the issue of carbapenem resistance are the crystal structures of OXA-24/40

and OXA-48. OXA-24/40 is one of the most widespread carbapenemases found in A.

baumannii (29). This enzyme was originally part of a clinical epidemic in Spain that involved

a 10-month-long outbreak affecting 29 patients, 23 of them hospitalized in five intensive

care units. The work by Santillana et al. showed that OXA-24/40 has a hydrophobic barrier

formed by Tyr112 and Met223 side chains, which define a tunnel-like entrance to the active

site (253). OXA-48 was initially isolated from K. pneumoniae. Docquier et al. (66) found that

OXA-48 is similar to OXA-10 in structure and not like OXA-24/40. Molecular dynamics

simulation showed that meropenem may position itself between Leu158 and Thr213, with

the C-6 ethoxy group approaching Val120. In this manner, H2O gets near Lys73 and can

attack C=O at the amide bond.

Resistance to Chloramphenicol

Acetyltransferases

Chloramphenicol is a broad-spectrum antimicrobial agent whose use has waned in recent

years due to well- characterized hematologic toxicity and a wealth of less toxic therapeutic

options. The most common mechanism of resistance to chloramphenicol is the elaboration of

CATs. A large number of CAT genes have been reported, and these determinants generally

confer extremely high levels of resistance on the organisms expressing them. Substantial

structural similarities exist among the different CAT variants, although their nucleotide

sequences may be quite divergent (184). Relationships among the different CATs have been

described in detail in a review by Schwarz and colleagues (256). Chloramphenicol contains

two hydroxyl groups that are acetylated in a reaction catalyzed by CAT in which acetyl

coenzyme A serves as the acyl donor. Initial acetylation occurs at the C-3 hydroxyl group to

yield 3-acetoxy-chloramphenicol (260). Following nonenzymatic rearrangement to 1-

acetoxy-chloramphenicol and reacetylation, the 1,3-diacetoxy-chloramphenicol product is

formed. Neither the mono- nor the di-acetoxy derivatives are able to bind to the 50S

ribosomal subunit and inhibit prokaryotic peptidyltransferase (260).

CATs are generally divided into two types: type A (classical) CATs and type B (xenobiotic)

CATs (256). For S. aureus, five structurally similar type A CATs (A, B, C, D, and that encoded

by the prototypic plasmid pC194) have been described (87). The cat genes encoding these

enzymes are commonly located on small, multicopy plasmids, and expression is inducible by

a translational attenuation mechanism (256). E. faecalis and S. pneumoniae also express

inducible CAT genes that are similar to the type D gene of S. aureus. Two cat genes encoding

constitutive CAT expression have been described to occur in Clostridium perfringens. catP is

generally found within transposon Tn4451, whereas catQ (nearly identical

to catD of Clostridium difficile) is chromosomal (256).

Three types of type A CATs (I, II, and III) have been identified in gram-negative bacteria.

The widely prevalent type I enzymes are distinguished by their ability to bind and inhibit

(without acetylation) the activity of fusidic acid. These enzymes are frequently found to be

associated with transposon Tn9 or related elements. Type II CATs are notable for their

sensitivity to inhibition by thiol-reactive agents and by their association with H.

influenzae (183). Most knowledge of the structural features of the type A CAT enzymes

comes from the study of the type III enzyme, for which the tertiary structure is known at

high resolution (184). The structural determinants of binding for each substrate are also

known for this enzyme.

Type B (xenobiotic) acetyltransferases (184) are structurally unrelated to classic CATs, and

those that have been demonstrated to acetylate chloramphenicol confer only low levels of

chloramphenicol resistance even when present in high copy number. Their natural substrate

is likely something other than chloramphenicol, explaining their limited ability to acetylate

this antibiotic. First described to occur in Agrobacterium tumefaciens,they have now been

identified in a wide range of species (256). Included among this class of agents are the

virginiamycin acetyltransferases found in S. aureus and E. faecium (see section on

macrolides below). In fact, although they are members of this class, the vat genes do not

confer resistance to chloramphenicol, nor have they been demonstrated to be able to

acetylate chloramphenicol in vitro (184). They are, however, quite adept at acetylating

streptogramins. The crystal structures of two trimeric type B CATs have been determined

(20, 225).

Decreased Accumulation of Chloramphenicol

It is now well recognized that chloramphenicol serves as a substrate for many of the MDR

efflux pumps that exist in gram-positive and gram-negative bacteria, including those found

in E. coli, P. aeruginosa, Bacillus subtilis, and S. aureus (191). In addition, there are efflux

systems that are specific for chloramphenicol. The first chloramphenicol-specific efflux gene

that was described was cmlA within the In4 integron of Tn1696 (23).cmlA encodes an efflux

mechanism that uses chloramphenicol but not florfenicol (a chloramphenicol derivative

licensed for use in animals in 1996 in the United States for the treatment of bovine

respiratory pathogens) as a substrate. Gram-negative bacteria also express efflux genes

specific for both chloramphenicol and florfenicol(floPp and floSt). These resistance genes are

being reported with increasing frequency for animal-derived E. coliand Salmonella isolates

(26, 308). In fact, the chloramphenicol resistance expressed by MDR Salmonella

enterica serovar Typhimurium DT104 is most commonly encoded by floSt (26), emphasizing

again the potential negative impact of using similar antimicrobial agents in humans and

animals. Very recently, chloramphenicol resistance in Acinetobacter baumannii has been

attributed to the activity of an MFS-type pump designated CraA (242).

Resistance to Daptomycin

Daptomycin is a cyclic lipopeptide antibiotic with activity exclusively against gram-positive

bacteria. It has bactericidal activity against most strains and acts by a cooperative

interaction in the presence of physiological concentrations of calcium that result in the

formation of pores in the cytoplasmic membranes of target bacterial cells. The end result is

leakage of ions from the cell and cell death. Resistance to daptomycin has been found in

both enterococci and staphylococci. Resistance to daptomycin is very rare in surveys, but in

the clinical setting resistance may arise associated with prolonged therapy. The precise

mechanisms of resistance have not been defined, but overexpression of genes associated

with increasing the positive charge of the cytoplasmic membrane has been implicated,

although actual changes in membrane charge have been difficult to demonstrate (178).

Exposure to daptomycin strongly induces the autoregulatory hVISA-associated (see below)

VraRS two-component regulatory system in S. aureus (185) and a similar system in Bacillus

subtilis(105). Perhaps as a result of these common pathways, hVISA strains frequently

express reduced susceptibility to daptomycin (59) and strains expressing reduced

susceptibility to daptomycin exhibit the hVISA (see below) phenotype (37). Unfortunately, S.

aureus strains with reduced susceptibility have been isolated with some frequency during

prolonged treatment of deep-seated infections (91).

Resistance to Glycopeptides

Glycopeptide antibiotics (vancomycin and teicoplanin) inhibit cell wall synthesis by binding to

the pentapeptide peptidoglycan precursor molecule as it exits the cytoplasmic membrane.

This binding prevents the cross-linking (transpeptidation) of peptidoglycan precursors

necessary for the formation of normal, stable cell walls. The large size of the glycopeptide

molecules also appears to inhibit the other major peptidoglycan linkage reaction

(transglycosylation) by steric hindrance. The specific moiety bound by vancomycin is the

terminal D-alanyl–D-alanine of the pentapeptide. The vast majority of bacteria that have

been studied have peptidoglycan precursors that are pentapeptides terminating in D-Ala –DAla,

and therefore are theoretically susceptible to vancomycin. However, the large size of

vancomycin exceeds the exclusion limits of the porins in the outer membranes of gramnegative

bacteria, so vancomycin cannot access the target in these species. Hence,

vancomycin is active only against bacteria lacking outer membranes, which are

predominantly gram positive.

Acquired resistance to vancomycin in gram-positive bacteria comes in three varieties largely

defined by the species within which they have been described: (i) altered precursor

formation in enterococci, (ii) mutational cell wall changes in staphylococci, and (iii) tolerance

in pneumococci. The importance of the first type of resistance is characterized by both its

prevalence and the importance of the species as a cause of infection, whereas the other two

are defined more by the importance of the species than by their prevalence.

To date, six varieties of enterococcal glycopeptide resistance have been described (VanA

through VanE and VanG). Of these, the most clinically important are VanA and VanB (15).

VanA and VanB are encoded by similar operons in which three

genes (vanH, vanA, and vanX or vanHB, vanB, and vanX B) are required for expression of

resistance (15). Two other genes (vanY and vanZ or vanYB and vanW) serve to amplify

resistance but are not required for its expression (11, 12), and two more

genes (vanS and vanR or vanS B and vanRB) regulate the transcription of the three essential

genes (13, 77). The ultimate purpose of these genes is to alter the structure of the

pentapeptide precursor from terminating in D-alanyl–D-alanine to D- alanine–D-lactate, in so

doing reducing the binding affinity of vancomycin to its target roughly 1,000-fold. The

sequence of reactions resulting in this structure is outlined in reference 122. Since the

terminal amino acid is cleaved off of the pentapeptide in the transpeptidation reaction, the

final composition of the cell wall is indistinguishable from that of strains lacking the

resistance determinant. Apparently the enterococcal PBPs, which facilitate transpeptidation,

have no trouble processing the altered precursors.

VanA enterococci are phenotypically resistant to vancomycin and teicoplanin, whereas VanB

strains are resistant to vancomycin but appear to be susceptible to teicoplanin. This

susceptibility results from the fact that teicoplanin does not induce expression of resistance

(77). Once the VanB operon is expressed, however, resistance to teicoplanin results.

Consequently, teicoplanin has been disappointing as a therapy for infections caused by VanB

enterococci, since mutations in the VanB regulatory apparatus resulting in either inducibility

by teicoplanin or constitutive expression occur readily during therapy (16, 112, 135).

Both VanA and VanB operons are carried by transposons. VanA is found exclusively within

transposon Tn1546, a 10.4-kb Tn3 family element that is presumed to disseminate among

enterococci by integrating into conjugative plasmids (14). The genes of the VanA operon are

found to be highly conserved in their sequence when different strains are compared, but the

restriction maps of the operons and of Tn1546 often differ markedly among clinical strains

(61). These differences result from insertions of a variety of IS elements with or without

subsequent deletions of parts of the mobile element and have been used by some

investigators to establish lineages of strains within defined clinical settings. The VanB operon

is most commonly encoded on highly similar transposons designated Tn5382 or

Tn1549 (40, 99). These transposons exhibit significant homology to prototype enterococcal

conjugative transposon Tn916. In contrast to the vanA gene, three allelic variants

of vanB (vanB1, vanB2, and vanB3) have been described. The vanB2 gene is associated with

Tn5382(60).

The overwhelming majority of clinical vancomycin-resistant enterococcal strains are E.

faecium, a predilection that remains unexplained (249). The vast majority of vancomycinresistant

E. faecium strains that cause clinical infection are also resistant to ampicillin, owing

to the expansion of a group of hospital-adapted clones (referred to as clonal complex 17)

that have emerged around the world. Clonal complex 17 strains are found worldwide and are

characterized by their resistance to ampicillin and by the fact that they frequently harbor

putative virulence determinants such as espEfm and hylEfm (147, 148).

Despite in vitro transfer of the VanA determinant to S. aureus (193), and at least 11

instances in which VanA-expressing S. aureus have been described for clinical samples

(209), vancomycin-resistant S. aureus remains exceedingly rare. In all cases resistance has

been conferred by the VanA operon, and in one well-characterized case transfer appears to

have been facilitated by the presence of Tn1546 on a broad-host-range plasmid in E.

faecalis, with transposition of Tn1546 to a staphylococcal plasmid once entry into the

staphylococcus occurred (89, 304).

The VanC operon is intrinsic to the cell wall synthesis machinery of the minor enterococcal

speciesEnterococcus casseliflavus (including the biotype formerly classified as E. flavescens)

and Enterococcus gallinarum (71, 298). The peptidoglycan precursor in VanC strains

terminates in D-alanine–D-serine, reducing vancomycin affinity about sevenfold and resulting

in low levels of resistance. Precursors terminate in D-alanine–D-serine strains of E.

faecalis with VanE (86), whereas VanD E. faecium terminates in D-alanine–D-lactate. The

failure to observe dissemination of VanD may be explained in part by the fact that the VanX

equivalent enzyme in E. faecium BM4339 appears to be ineffective in an enterococcal

background. Resistance was expressed in BM4339 because that strain lacked a functional

cellular ligase (ddl) gene, eliminating the need for VanX activity to express resistance (44).

Mutational resistance to glycopeptides in S. aureus commonly takes the form of reduced

susceptibility, rather than frank resistance. These strains, alternately called hVISA (heterovancomycin-

intermediate S. aureus) or hGISA (hetero-glycopeptide-intermediate S. aureus),

express vancomycin MICs in the 4- to 8-μg/ml range (156). However, within these cultures

are smaller populations of cells that express higher levels of resistance. The resistance

phenotype is characterized by thickened cell wall, which may decrease glycopeptide

susceptibility by providing an excess of false targets for glycopeptide binding. In many, but

not all, cases, conversion to the hVISA phenotype is associated with the vraRS twocomponent

regulatory circuit, which responds to cell wall damage in S. aureus and regulates

more than 40 genes, some of which are associated with the biosynthesis of peptidoglycan

(21). Interestingly, recent work suggests that the hVISA phenotype can be selected by

exposure to β-lactam antibiotics as well as by exposure to glycopeptides (137). Animal

studies suggest that the level of resistance expressed by hVISA strains reduces the

effectiveness of vancomycin therapy (53).

Glycopeptide resistance has also been reported for coagulase-negative species of

staphylococci. In contrast to S. aureus, resistance in Staphylococcus haemolyticus has been

associated with changes in the composition of the cross-links of the peptidoglycan (22). The

mechanism by which this would lead to vancomycin resistance is incompletely understood.

Vancomycin is, in general, a less bactericidal antibiotic than are the β-lactams. Evidence for

the importance of this observation can be found in several species. The bacteremia

associated with S. aureus endocarditis, for example, takes roughly twice as much time to

clear with vancomycin treatment than with treatment by β-lactam oxacillin (154). Recent

clinical data also suggest that vancomycin treatment was associated with higher rates of

failure and relapse than was nafcillin treatment for bacteremia due to methicillinsusceptible

S. aureus(48). Vancomycin efficacy appears to be particularly poor against some

strains of MRSA. In a recent clinical trial, vancomycin successfully treated left-sided

endocarditis in only 2 of 9 cases, a success rate that was similar to that of daptomycin (2 of

10) (91). Reports of vancomycin tolerance in Streptococcus pneumoniae first appeared in

1990 (289). S. pneumoniae is the most common cause of bacterial meningitis in most

patient populations, and bactericidal therapy is optimal for treatment of this condition. At

least one case of presumed recrudescence of meningitis after treatment of a case of

vancomycin-tolerant pneumococcal meningitis has been reported (114). Tolerance appears

to involve mutations within an operon (vex123) encoding an ABC transporter, but the

mechanism by which this occurs remains undefined (104). Further work will be required

before we understand the true importance of pneumococcal tolerance for the treatment of

clinical infections.

Resistance to Linezolid

The oxazolidinone antibiotic linezolid inhibits bacterial protein synthesis by interacting with

the N- formylmethionyl-tRNA–ribosome–mRNA ternary complex commonly referred to as the

initiation complex (264). Linezolid exerts excellent bacteriostatic activity against a wide

range of gram-positive pathogens, including methicillin-resistant staphylococci and MDR

enterococci. Clinical use of this agent has been associated with the emergence of resistant

strains, most commonly after prolonged therapy of difficult-to-eradicate bacteria. Resistance

has now been described for both enterococci and staphylococci, but overall rates several

years after clinical introduction of this agent remain very low (134). Resistance is most often

associated with a G2576U (Escherichia coli numbering scheme) point mutation in the 23S

rRNA, although mutations at other positions may also contribute to resistance (224).

Resistance to linezolid, macrolides, and chloramphenicol has been attributed to a 6-bp

deletion in the gene encoding riboprotein L4 in S. pneumoniae (313). Since the 23S subunit

genes exist in multiple copies in different bacteria (four in E. faecalis and S. pneumoniae, six

in E. faecium, and five or six in S. aureus), more than one copy of the genes must be

mutated to confer resistance, with strains having a higher percentage of mutated 23S genes

expressing greater levels of resistance (166). Gene conversion, or recombination between

mutated genes and wild-type genes, can rapidly increase the levels of resistance once the

first gene mutation has occurred (161). In this fashion, persistent selective pressure exerted

by linezolid can lead to rapid development of high-level (MIC > 128 μg/ml) resistance.

Plasmid-mediated resistance to linezolid through the expression of the cfr rRNA methylase

gene has been reported for staphylococci (10), but the prevalence of this type of linezolid

resistance remains very low (134).

Resistance to Macrolides

Erythromycin (the first macrolide) was initially isolated from Streptomyces erythraeus, a soil

organism found in the Philippines. There are currently four macrolides in common use:

erythromycin, clarithromycin, azithromycin, and roxithromycin. Macrolides inhibit protein

synthesis in susceptible organisms by binding reversibly to the peptidyl-tRNA binding region

of the 50S ribosomal subunit, inhibiting translocation of a newly synthesized peptidyl-tRNA

molecule from the acceptor site on the ribosome to the peptidyl (or donor site).

Erythromycin does not bind to mammalian ribosomes. Most gram-negative organisms are

resistant to erythromycin because entry of erythromycin into the cell is restricted.

Resistance to macrolides occurs by several mechanisms. Among the more important of these

mechanisms is methylation of the ribosome, preventing erythromycin binding (305). This

methylation is most commonly accomplished by different erm (erythromycin ribosomal

methylase) genes. Methylated ribosomes confer resistance to macrolides, the related

lincosamides (clindamycin and lincomycin), and streptogramins B (MLSBresistance).

Many erm genes have been described—erm(A) and the related erm(TR), plus erm(B) and the

related erm(AM)—and resistance is frequently inducible by macrolides but not by clindamycin

(iMLSB). In some strains, erm-type resistance is expressed constitutively (cMLSB), resulting

in resistance to clindamycin as well.

The second major mechanism of resistance to macrolides is expression of efflux pumps

encoded by mef genes (Mef in gram-positive bacteria and Acr-AB-TolC in H.

influenzae and E. coli) (321). The efflux pumps confer resistance to the macrolides but not to

clindamycin, hence the phenotypic description of this resistance as “M” type. mef genes have

been studied most extensively for Streptococcus pneumoniae [mef(E)] and Streptococcus

pyogenes [mef(A)], but similar genes have been described for a variety of gram-positive

genera. The prevalences of mef-mediated resistance versus that mediated by MLSB-type

mechanisms in S. pneumoniaevary in different parts of the world. Minor mechanisms of

resistance to macrolides include esterases that hydrolyze the antibiotics and point mutations

within the 50S rRNA gene.

Resistance to Ketolides

Ketolides belong to a new class of semisynthetic 14-membered-ring macrolides, which differ

from erythromycin by having a 3-keto group instead of the neutral sugar L-cladinose.

Ketolides bind to an additional site on the bacterial ribosome, increasing their binding affinity

relative to that of other macrolides (69). Telithromycin, a ketolide, is uniformly and highly

active against pneumococci (regardless of their susceptibility or resistance to erythromycin

and/or penicillin), erythromycin-susceptible S. pyogenes and erythromycin-resistant S.

pyogenesstrains of the M phenotype or iMLS-B or cMLSB phenotype [in which resistance is

mediated by a methylase encoded by the erm(TR) gene] (18). Ketolides are less active

against erythromycin-resistant S. pyogenesstrains with the cMLSB phenotype or the

iMLSA subtype (in which resistance is mediated by a methylase encoded by the ermB gene),

these strains ranging in phenotype from the upper limits of susceptibility to resistant.

Methicillin-resistant staphylococci, which commonly express a cMLSB phenotype, are not

susceptible to telithromycin (18).

Resistance to Quinupristin-Dalfopristin

Quinupristin-dalfopristin is a mixture of semisynthetic streptogramins A and B licensed in

Europe and the United States. A related streptogramin A and B combination, virginiamycin,

has been used for years as a growth promoter in animal feed. Resistance to these mixtures

can result from resistance to streptogramin A alone and was first described for staphylococci

conferred by genes encoding streptogramin A acetyltransferases [vat(A), vat(B), and vat(C)]

or ATP-binding efflux genes [vga(A) and vga(B)]. Quinupristin- dalfopristin’s excellent

activities against E. faecium and MRSA make it an alternative for the treatment of MDRE.

faecium and health care-associated MRSA infections, especially since the combination retains

in vitro activity against streptogramin B-resistant strains. Two acetyltransferase-encoding

resistance genes have now been described that confer resistance to quinupristin-dalfopristin

in E. faecium: vat(D) [previously sat(A)] and vat(E) [previously sat(G)]. In most cases,

these resistance genes are found along with an erm gene (270), suggesting that resistance

to both streptogramins A and B may be necessary to confer clinically significant levels of

resistance to quinupristin-dalfopristin in E. faecium. These resistance genes are frequently

present on transferable plasmids. Although quinupristin-dalfopristin remains active against

the majority of human E. faecium strains, the use of virginiamycin in animal feeds has been

associated with high percentages of resistance in isolates derived from animals (113). In

many cases, the known mechanisms of resistance to quinupristin-dalfopristin are not present

in these isolates (113), indicating that there is still much to be learned about resistance to

quinupristin-dalfopristin in E. faecium.

Resistance to Metronidazole

Metronidazole is a member of the nitroimidazole family of bactericidal antimicrobials. The 5-

nitroimidazole molecule is a prodrug whose activation depends upon reduction of the nitro

group in the absence of oxygen. An exception to this rule occurs in Helicobacter pylori, with

which the RdxA protein reduces metronidazole in a microaerophilic environment (295). The

nitro group of metronidazole accepts a single electron from electron transport proteins

(ferredoxins) in bacteria, yielding a toxic radical anion. Metronidazole’s activity appears to

result in DNA damage and cell death (73). Resistance to metronidazole is rare. Decreased

uptake and/or a reduced rate of reduction is believed to be responsible for metronidazole

resistance in some cases (72). FiveBacteroides genes, nimA to nimE, have been implicated in

resistance to 5-nitroimidazole antibiotics. Analysis of the NimA susceptible and

resistant Bacteroides strains and recent crystal structure analysis suggest that the enzyme

utilizes pyruvate for a two-electron reaction resulting in an amine that prevents the

formation of the toxic anion radical (41, 152). Expression of nim genes varies depending on

the positioning of a variety of IS elements that supply active promoters (269). Recent data

indicate that the enzyme thioredoxin reductase is responsible for reduction of metronidazole

in Trichomonas vaginalis (153).

Resistance to Nitrofurantoin

The antibiotics nitrofurazone and nitrofurantoin are used in the treatment of genitourinary

infections and as topical antibacterial agents. Nitrofurazone is primarily used as a topical

antiseptic (103). Nitrofurantoin, 1-[(5-nitrofurfurylidene)amino]hydantoin, is a synthetic

antibacterial agent used primarily in the treatment of urinary tract infections. The

mechanism of action of nitrofurazone and nitrofurantoin has not been fully elucidated.

Investigators have reported that the ability of nitrofurantoin to kill bacteria correlates with

the presence of bacterial nitroreductases which convert nitrofurantoin to highly reactive

electrophilic intermediates (173). These intermediates are believed to attack bacterial

ribosomal proteins nonspecifically, causing complete inhibition of protein synthesis. In E.

coli, nitroreductases are type 1 oxygen-insensitive enzymes, encoded by

the nfnA (nfsA) and nfnB (nfsB) genes. Strains of bacteria that are resistant to nitrofurantoin

have been shown to possess diminished nitroreductase activity (227), which may seriously

compromise their fitness (252). Resistance to nitrofurantoin from reduced nitroreductase

activity seems to be present in other genera as well.

Resistance to Polymyxin B and Polymyxin E (Colistin)

Clinical and scientific interest in the cationic polypeptides is increasing. Although they were

first used in the early 1960s, colistin (polymyxin E) and polymyxin B are now often used as

first-line therapy of infections caused by MDR gram-negative bacterial infections. The

polymyxins are polycationic peptide antibiotics isolated from Bacillus polymyxa (133). They

exert their bactericidal activity by binding to the cell membrane of gram-negative bacteria

and disrupting its permeability, resulting in leakage of intracellular components. They also

disrupt bacterial biofilm formation. In mechanistic terms, polymyxin binds to phosphorylated

head groups of lipid A. Hence, by disrupting cell membranes, these agents become rapidly

bactericidal against certain gram-negative bacteria (278, 290).

Not all gram-negative bacteria are susceptible to polymyxins. Organisms that are resistant to

polymyxins have cell walls that prevent access of the drug to the cell membrane. In general,

polymyxins are bactericidal againstP. aeruginosa, Acinetobacter spp., some Proteus

mirabilis strains, and some strains of Serratia

marcescens.Proteus spp., Providencia spp., Neisseria spp., and gram-positive bacteria are

resistant to polymyxins (278,290).

Polymyxin-resistant mutants and bacteria exhibit a modified LPS. In E. coli,

Salmonella serovar Typhimurium, and many other pathogenic gram-negative bacteria,

modification of the phosphate groups of lipid A confers resistance to polymyxin and cationic

antimicrobial peptides. LPS modifications that include alteration of the fatty acid content of

lipid A, phosphoethanolamine addition to the core and lipid A head groups, and 4-amino-4-

deoxy-L-arabinose addition to the core and lipid A regions have been well studied (290).

Recent evidence also implicates the presence of the MtrC-MtrD-MtrE efflux pump and lipid A

modification as well as the type IV pilin secretion system to modulate levels of polymyxin

resistance in Neisseria meningitidis (290) and the PmrAB two-component system in

resistance to colistin in Acinetobacter baumannii (1).

Resistance to Quinolones

The fluoroquinolones are among the most widely used antimicrobial agents in both the

hospital and community settings. Quinolone antibiotics all act by directly inhibiting DNA

synthesis. Their targets include two type 2 topoisomerases: DNA gyrase and topoisomerase

IV. These two enzymes are structurally related in that both exist as tetramers composed of

two different subunits (GyrA and GyrB of DNA gyrase and ParC and ParE of topoisomerase

IV). DNA gyrase acts to maintain negative supercoiling of DNA, whereas topoisomerase IV

separates interlocked daughter DNA strands formed during replication, facilitating

segregation into daughter cells. Fluoroquinolones bind to the topoisomerase-DNA complexes

and disrupt various cellular processes involving DNA (replication fork, transcription of RNA,

and DNA helicase) (118, 261, 312). The end result is cellular death by unclear mechanisms.

The affinities of fluoroquinolones for the two targets vary, explaining to some degree the

differing potencies of the various agents against different bacterial species. The enzyme for

which a particular fluoroquinolone exerts the greatest affinity is referred to as the primary

target (6, 25, 202). In general, the primary target of fluoroquinolones in gram-negative

bacteria is DNA gyrase, whereas in gram-positive bacteria it is topoisomerase IV.

Alterations in Target Enzymes

The most common mechanism of clinically significant levels of fluoroquinolone resistance is

through alterations of the topoisomerase enzymes. These alterations are created by

spontaneous mutations that occur within the respective genes. In GyrA and ParC, resistanceassociated

mutations are often localized to a region in the amino terminus of the enzyme

containing the active-site tyrosine that is covalently linked to the broken DNA strand. This

130-bp region of gyrA has been referred to as the quinolone resistance-determining region.

X-ray crystallographic studies of a fragment of the GyrA enzyme suggest that these

mutations are clustered in three dimensions, lending support to the hypothesis that the

region constitutes a part of the quinolone binding site (180). Particularly frequent sites for

resistance-associated mutations are serine 83 and aspartate 87 of GyrA and serine 79 and

aspartate 83 of ParC (212).

Experimental data suggest that point mutations occur singly in roughly 1 in 106 to 109 cells.

The level of resistance conferred by a single point mutation in the primary target enzyme

depends upon the reduction of enzyme affinity created by the mutation, as well as the

affinity of the fluoroquinolone for the secondary target. In this scenario, it is expected that

fluoroquinolones exhibiting strong affinity for both target enzymes would be less likely to be

associated with the emergence of resistant strains, since the retained activity against the

secondary target would be enough to inhibit the bacterium even in the presence of a primary

target mutation. Fluoroquinolone-species combinations for which single mutations result in

significantly higher MICs (such as ciprofloxacin and S. aureus or P. aeruginosa) would be

expected to readily select out (and have readily selected out [55]) resistant mutants in the

clinical setting.

Most highly resistant strains exhibit more than one mutation in both the GyrA and ParC

enzymes, a phenomenon that can be reproduced in the laboratory by serial passage of

strains on progressively higher concentrations of fluoroquinolones. It is noteworthy in this

context that fluoroquinolone resistance conferred by enzyme mutations is essentially class

resistance. In other words, the activity of all fluoroquinolones is affected by mutations that

result in resistance. Therefore, while single point mutations that confer resistance to one

fluoroquinolone may not result in MICs conferring clinical resistance to another, the MICs of

the second fluoroquinolone will inevitably be increased. In the setting of such preexisting

mutations, the second fluoroquinolone could then select for an additional mutation that

would result in clinically significant levels of resistance. This reasoning has led to the

recommendation that the most potent and broadly active fluoroquinolone always be used

first, to prevent the emergence of resistance. The wisdom of this recommendation remains

to be tested.

Mutations in GyrB and ParE are far less common than in their companion subunits and tend

to cluster in the mid-portion of the subunit (120). A clear understanding of the impact these

mutations have on enzyme structure or function awaits detailed crystallographic studies of

enzyme-fluoroquinolone complexes.

Resistance Due to Decreased Intracellular Accumulation

Fluoroquinolones penetrate the outer membrane of gram-negative bacteria through porins,

so the absence of specific porins can affect the level of susceptibility. However, their ability

to diffuse through outer and cytoplasmic membranes is sufficient to retain activity against

strains solely lacking porins (192). More important in reducing intracellular accumulation of

fluoroquinolones is the expression of MDR pumps (212). The intrinsic efflux pump complexes

in gram-negative bacteria extend from the cytoplasmic membrane through the outer

membrane, whereas gram-positive pumps need only traverse the cytoplasmic membrane.

These pumps move compounds across the bacterial membranes by proton motive force and

are presumed to represent systems by which bacteria rid themselves of toxic materials.

Resistance results when expression of pumps is increased due to mutations within their

regulatory genes (322). By themselves, pumps generally confer only a low level of resistance

to fluoroquinolones. However, their expression may amplify the level of resistance conferred

by point mutations within the topoisomerase genes. By so doing, they may increase the risk

that use of a given fluoroquinolone will select for resistant mutants through single point

mutations. In recent years, a plasmid-mediated efflux pump (QepA) has been recognized

among strains of Enterobacteriaceae (46). This pump extrudes the hydrophilic

fluoroquinolones (ciprofloxacin, enrofloxacin, and norfloxacin).

The major type of plasmid-mediated fluoroquinolone resistance present in gram-negative

bacteria is conferred by the Qnr proteins (130, 195), which protect DNA from quinolone

binding (285, 286). In general, only low levels of resistance are conferred by this

mechanism, but as with other accessory mechanisms, the presence of Qnr can facilitate the

clinical emergence of strains resistant by virtue of point mutations in the topoisomerase

genes. Five variants of Qnr have now been described (A, B, C, D, and S). There are several

alleles within the A, B, and S variants (http://www.lahey.org/qnrStudies/). The prevalence of

this mechanism is increasing, which may be partly explained by the frequent presence

of qnr within complex sulI-type integrons (195) often associated with insertion elements

(282).

Plasmid-mediated fluoroquinolone resistance can also be conferred by the AAC(6′)-Ib-cr

protein, which is a mutant of the AAC(6′)-Ib AME (46). This confers low levels of resistance

to ciprofloxacin and norfloxacin.

Resistance to Rifampin

Rifampin is particularly active against gram-positive bacteria and mycobacteria. It acts by

inhibiting bacterial DNA-dependent RNA polymerase. Point mutations in the

chromosomal rpoB gene confer resistance to rifampin (303). The frequency with which these

point mutations occur precludes using rifampin as a single agent for the treatment of

bacterial infections.

Resistance to Tetracyclines

The tetracyclines are a group of bacteriostatic antibiotics that act by inhibiting attachment of

aminoacyl-tRNA to the ribosome acceptor site, thereby preventing elongation of the peptide

chains of nascent proteins (255). In order to gain access to the bacterial ribosome,

tetracyclines need to enter the cell. In E. coli and presumably other gram- negative bacteria,

they enter the periplasmic space through outer membrane porins OmpC and OmpF, probably

chelated to magnesium ions (255). Once in the periplasmic space, the weakly lipophilic

tetracycline molecule dissociates from the magnesium ion and crosses into the cell by

diffusing though the lipid bilayer in an energy-dependent process. Once inside the cell,

tetracycline-ion complexes bind to the ribosome at a single, high-affinity binding site on the

30S subunit, blocking access of the aminoacyl-tRNA to the ribosome acceptor site. Although

of high affinity, binding of tetracycline to the ribosome is reversible (52).

Tetracyclines are broad-spectrum and effective antimicrobial agents. Unfortunately,

widespread use of tetracyclines to treat clinical infections and for promotion of growth in

livestock has been associated with the emergence and dissemination of a variety of

resistance determinants. As a consequence, the number of infections for which tetracyclines

are recommended as first-line therapy has been limited for many years (248). The vast

majority of tetracycline resistance determinants fall into one of two classes: (i) efflux or (ii)

ribosomal protection. The designations of the different resistance determinants and their

classes can be found in detail in an excellent review of tetracyclines by Chopra and Roberts

(52). Initial designations of tetracycline resistance determinants used the

prefix tet or otr with letters (A, for example) designating the different determinants. Since

the number of resistance determinants now exceeds the number of letters in the alphabet, a

system using numbers has been devised (155).

Tetracycline efflux proteins are all membrane associated and members of the MFS proteins.

They expel tetracycline from the cell by exchanging a proton for a tetracycline-cation

complex. In general, the efflux proteins confer resistance to tetracyclines but tend to spare

minocycline (52). The single exception to this rule is the Tet(B) protein of gram-negative

organisms, which confers resistance to both tetracycline and minocycline. The efflux proteins

have been divided into six groups based on amino acid identity. Group 1 consists of Tet

efflux proteins that are found primarily in gram-negative species [with the exception of

Tet(Z)], whereas group 2 [consisting only of Tet(K) and Tet(L)] is found primarily in grampositive

species. Groups 3 through 6 are small groups consisting of one or two efflux proteins

each.

Ribosome protection proteins comprise the other major mechanism of tetracycline

resistance. These proteins exhibit homology to elongation factors EF-Tu and EF-G and exhibit

ribosome-dependent GTPase activity (251). They act by binding to the ribosome, thereby

changing its conformation and inhibiting binding of tetracycline. Tet(M) and Tet(O) are the

best characterized of these proteins. Ribosome protection genes are widespread in bacteria,

in many cases as a result of their incorporation into broad-host-range conjugative

transposons.

Both efflux proteins and ribosomal protection proteins are regulated in ways that their

expression is increased in the presence of tetracyclines. The efflux proteins of gram-negative

organisms are regulated by repressors that are divergently transcribed relative to the efflux

proteins (119). Binding of the repressors to tetracycline changes the conformation of the

repressor so that it can no longer bind to the operator region, resulting in increased

transcription of both the efflux protein and the repressor genes. The gram-positive efflux

genes are not associated with specific protein repressors; sequence analysis suggests that

these determinants may be regulated by mechanisms similar to translational attenuation, but

study of this area has been limited (274). Transcription of ribosomal protection genes is

augmented by growth in the presence of tetracycline.

Intrinsic mechanisms of tetracycline resistance exist in many, if not all, gram-negative

bacteria. Among the best characterized of these systems is the mar (multiple antibiotic

resistance) operon (4). This locus consists of a repressor (MarR) that represses transcription

of marA, which encodes a transcriptional activator of a variety of genes. Overexpression of

MarA results in decreased expression of OmpF, a porin through which tetracycline enters the

periplasmic space, and increased expression of multidrug efflux pump AcrAB, a member of

the RND family of efflux proteins, which includes tetracyclines among its substrates. Several

similar pump systems have been described for P. aeruginosa and other gram-negative

bacteria (222). As our knowledge of the genomes of different bacterial species becomes

more complete, we will no doubt discover several other pump systems that affect levels of

susceptibility to tetracyclines and other antibiotics.

The remarkable diversity of species within which tetracycline resistance determinants are

found owes much to the inclusion of these resistance genes within broad-host-range

transferable genetic elements. These include transferable plasmids in gram-negative species,

where tet genes may be found included within integrons, and conjugative transposons.

Among the best studied of the conjugative transposons is the Tn916 family, originally

described for E. faecalis (233). The complete sequence of Tn916 has been determined and is

remarkable for its dearth of restriction enzyme digestion sites [except in the region of

the tet(M) gene, which appears to be a late arrival to the element] (198). This lack of

restriction sites likely facilitates its entry into a variety of different bacterial species. Transfer

of Tn916-like elements from enterococci into many other species has been demonstrated in

vitro and in animal models, and the remnants of Tn916-like sequences in N. gonorrhoeae are

impressive testimony to its ability to travel widely (233). Transfer of Tn916-like elements,

which is increased after exposure to tetracycline, has also been suggested to facilitate

transfer of unlinked genes, further amplifying the risks of overexposure to tetracycline in the

environment.

Resistance to Tigecycline

The recent licensing of the glycylcycline tigecycline offers a broad-spectrum antimicrobial

alternative for treating infections due to resistant pathogens, including MRSA and ESBLproducing

K. pneumoniae. Tigecycline’s broad spectrum of antimicrobial activity is due to its

resistance to the common efflux or ribosomal protection mechanisms that confer resistance

to older tetracyclines. Some bacterial species, notably P. aeruginosa and Proteus spp., are

intrinsically resistant to tigecycline because they express RND-type efflux pumps that

effectively extrude the antibiotic (245). Resistance to tigecycline in other gram-negative

species has also been reported, generally resulting from activation of normally repressed

AcrAB-type RND efflux pumps (246). The ultimate importance of these pump activations for

clinical resistance to tigecycline awaits more extensive clinical use.

Resistance to Trimethoprim-Sulfamethoxazole

Biosynthesis of several amino acids and purines depends upon the availability of

tetrahydrofolate. With few exceptions, bacteria are unable to absorb preformed folic acid,

and hence rely upon their ability to synthesize it. Sulfameth oxazole and trimethoprim are

inhibitors of two enzymes (dihydropteroic acid synthase [DHPS] and dihydrofolate reductase

[DHFR]) that act sequentially in the manufacture of tetrahydrofolate. It is thought that the

two inhibitors act synergistically to inhibit folate synthesis, although the mechanism for

possible synergism (since sequential blockage of a fully inhibited pathway should not

augment resistance) is not clear.

Intrinsic Resistance

Trimethoprim-sulfamethoxazole is a remarkably broad- spectrum antimicrobial agent.

Intrinsic resistance is relatively rare and may occur by decreased access to the target

enzymes (P. aeruginosa) (279) or low-affinity DHFR enzymes

(Neisseria spp., Clostridium spp., Brucella spp., Bacteroides spp., Moraxella

catarrhalis, andNocardia spp.) (280) or by the ability to absorb exogenous folate

(Enterococcus spp. and Lactobacillus spp.) (319) or thymine (Enterococcus spp.) (110). The

decreased access to the target enzyme in P. aeruginosaappears to be due to both a

permeability barrier and active efflux from the cell (144, 169). The percentage contribution

of each of these mechanisms to resistance remains unclear.

Acquired Resistance to Trimethoprim

Mutational resistance to trimethoprim has been described for several species and involves

promoter mutations leading to overproduction of DHFR (in E. coli), point mutations within

the dhfr gene leading to resistance (in S. aureus and S. pneumoniae), or both mechanisms

(in H. influenzae) (123). More common is the acquisition of low-affinity dhfr genes, of which

approximately 20 have been described (123). Expression of thedhfrI and variants

of dhfrII genes, which are most commonly found on plasmids in gram- negative bacteria,

increases resistance to levels greatly exceeding clinically achievable concentrations.

Acquired Resistance to Sulfonamides

Point mutations or small insertions of DNA segments within chromosomal dhps genes

conferring resistance to sulfonamides have been reported for many different species

(76, 123). More extensive changes within dhpsgenes resulting in resistance have been

reported for N. meningitidis and S. pyogenes. In these instances, the extensive changes

have suggested acquisition of at least some parts of the dhps genes from other species via

transformation and recombination (267, 277). Plasmid-mediated, transferable resistance to

sulfonamides has been reported for gram-negative bacteria (123). In contrast to the

diversity in dhfr genes, only two acquired low-affinity dhps genes (sulI and sulII) have been

described. Genes conferring resistance to sulfonamides are frequently incorporated into MDR

integrons, which are themselves frequently integrated into transferable plasmids. The

transferability of these resistance plasmids and the frequent association with other resistance

genes explain in part the widespread nature and persistence of resistance to this

antimicrobial combination. One trimethoprim-sulfamethoxazole-resistant E. coli strain was

reported to have spread widely in the United States, causing urinary tract infections in young

women in at least two states (163), although more recent data suggest that this widespread

prevalence may owe more to parallel emergence of related strains than to direct spread of

an outbreak isolate (92).