Bench-to-bedside review: The role of β-lactamases in antibiotic-resistant Gram-negative infections

Multidrug resistance has been increasing among Gram-negative bacteria and is strongly associated with the production of both chromosomal- and plasmid-encoded β-lactamases, whose number now exceeds 890. Many of the newer enzymes exhibit broad-spectrum hydrolytic activity against most classes of β-lactams. The most important plasmid-encoded β-lactamases include (a) AmpC cephalosporinases produced in high quantities, (b) the expanding families of extended-spectrum β-lactamases such as the CTX-M enzymes that can hydrolyze the advanced-spectrum cephalosporins and monobactams, and (c) carbapenemases from multiple molecular classes that are responsible for resistance to almost all β-lactams, including the carbapenems. Important plasmid-encoded carbapenemases include (a) the KPC β-lactamases originating in Klebsiella pneumoniae isolates and now appearing worldwide in pan-resistant Gram-negative pathogens and (b) metallo-β-lactamases that are produced in organisms with other deleterious β-lactamases, causing resistance to all β-lactams except aztreonam. β-Lactamase genes encoding these enzymes are often carried on plasmids that bear additional resistance determinants for other antibiotic classes. As a result, some infections caused by Gram-negative pathogens can now be treated with only a limited number, if any, antibiotics. Because multidrug resistance in Gram-negative bacteria is observed in both nosocomial and community isolates, eradication of these resistant strains is becoming more difficult.

demonstrate a pharmacodynamic and safety profi le that makes them attractive therapeutic agents.
As shown in Table 1, the most common families of βlactams that are used to treat infections caused by Gramnegative pathogens include extended-spectrum cephalosporins such as ceftriaxone and cefepime, penicillin-βlactamase inhibitor combinations such as amoxicillinclavulanic acid and piperacillin-tazobactam, and the carba penems [4]. Infection types range from uncom plicated community-acquired infections such as otitis media to serious nosocomial infections, including ventilatorassociated pneumonia. Orally administered β-lactams such as amoxicillin-clavulanic acid and the oral cephalosporins cefi xime, cefpodoxime, and cefuroxime axetil are recommended for community infections. For nosocomial infections that are not resistant to cephalosporins, parenteral drugs that may be eff ective include the injectable penicillin-β-lactamase inhibitor combinations and cephalosporins. Carbapenems are often reserved to treat the most serious infections caused by many multidrug-resistant pathogens as they are most able to escape at least some of the common β-lactam resistance mechanisms that aff ect the other β-lactam families. However, the continued use of these antibiotics for important disease states has maintained the pressure on many pathogenic and commensal bacteria to retain enzymatic inactivation mechanisms that render penicillins ineff ective in many disease states. Th is pressure has resulted in a surge in β-lactam resistance due to inactivating enzymes, particularly in Gramnegative pathogens.

Background of β-lactamase resistance
In 1940, an enzyme produced by a strain of Bacillus coli, now known as Escherichia coli, was shown to destroy the ability of penicillin to kill bacterial cells [5]. Th is fi rst report of β-lactamase activity occurred before widespread use of penicillin, demonstrating the presence of βlactam-inactivating enzymes in the natural environment. Th ese enzymes have the ability to hydrolyze the β-lactam chemical bond that distinguishes β-lactam antibiotics from other antibacterial agents, thereby rendering the molecules incapable of killing bacteria.
Today, over 890 unique β-lactamases have been identi fied in naturally occurring bacterial isolates [6,7]. These enzymes have been separated into groups, either accord ing to the amino acid sequences of the enzymes or according to their inactivating properties for different classes of β-lactams. The molecular classification scheme divides β-lactamases into four classes based on the amino acid sequences of the proteins [8][9][10], whereas numeri cally more functional groups have been assigned based on the hydrolysis and inhibition profiles of the enzymes [7,10,11]. The major functional groups of current clinical importance are shown in Table 2, where enzyme groups are commonly named according to the most important class of βlactam that is inactivated. Correlations between the molecular and functional categories are also provided in this compilation.

Cephalosporinases
Group 1/Class C cephalosporinases are among the most abundant β-lactamases on the basis of the number of organisms that produce these enzymes. Th ese cephalosporinases, frequently named as species-specifi c AmpC enzymes, are often found in most Enterobacteriaceae as chromosomal enzymes. At low production levels, they can abolish the antibacterial activity of cephalosporins and can also demonstrate inactivating capabilities toward other β-lactams, especially when produced at high levels.
Th ese enzymes are generally present at a low (basal) level but may be induced to high levels in the presence of selected inducing agents such as amoxicillin or clavulanic acid. Cephalosporinases may also be produced at very high levels in the absence of an inducer, in a 'derepressed' state. Th is has been reported to occur after selection of stable mutants during therapy with broad-spectrum cephalosporins, but not cefepime [12,13]. Th ese mutants often arise as a result of a multistep process, with elevated cephalosporin MICs (in the high susceptible or intermediate range) observed in those organisms with impaired permeability characteristics prior to selection of high-level AmpC mutants that are fully resistant [14]. Treatment of organisms producing an inducible AmpC cephalosporinase has created some controversy. Although a group of investigators has recommended that any AmpC-inducible Enterobacteriaceae be regarded as resistant to all third-generation cephalosporins [13], the clinical data supporting this recommendation are mixed [14]. Studies have indicated that the selection of resistant Enterobacter spp. has ranged as high as 19% in a 1991 study (6/31), with higher rates observed in patients with bacteremia [15], but more recent studies have not corroborated those observations [12]. For example, in an 18-month study (2005)(2006) of 732 patients infected with Enterobacteriaceae capable of producing AmpC βlactamases, treatment with broad-spectrum cephalosporins resulted in resistance in 5% (11/218) of the patients compared with 0% (0/20) of the patients treated with cefepime, with the emergence of resistance being more frequent in Enterobacter spp. [12].
Th e suggestion has been made that therapeutic decisions be made on the basis of specifi c cephalosporin MICs rather than on the presence of an inducible AmpC enzyme [14]. Th is year, the Clinical and Laboratory Standards Institute (Wayne, PA, USA) lowered susceptibility breakpoints for cefotaxime, ceftriaxone, and ceftazidime against the Enterobacteriaceae [16]. Lower breakpoints should help to decrease the number of 'susceptible' isolates that may be likely to emerge as resistant strains following therapy with broad-spectrum cephalosporins but will allow cephalosporins to be used against highly susceptible organisms. Th e new cephalosporin breakpoints will also classify as resistant those ESBL-producing Enterobacteriaceae that will not respond pharmacodynamically to approved therapeutic doses, thereby reducing the need for specifi c ESBL testing other than for epidemiological purposes [16].
Plasmid-encoded AmpC cephalosporinases closely related in sequence to chromosomal AmpC enzymes from Enterobacter cloacae, Citrobacter freundii, or Aero monas spp. also appear in Enterobacteriaceae in organisms that produce at least one other β-lactamase [17]. Although some of the plasmid-encoded AmpC enzymes are inducible, most of the enzymes are produced at much higher levels than seen for basal AmpC cephalo sporinases, similar to isolates with derepressed AmpC enzymes [17].

Penicillinases
Common penicillinases (functional group 2b, molecular class A) include the SHV-1 enzyme in Klebsiella pneumoniae and the TEM-1 β-lactamase found in many strains Table 2 Classifi cation based on Bush, Jacoby, and Medeiros [11] and Bush and Jacoby [7]. b Classifi ed according to primary amino acid sequence [8][9][10]. c β-lactams that are resistant solely as a function of β-lactamase production. d β-lactams that are resistant as a function of β-lactamase production, usually at high levels, in combination with effl ux or porin modifi cations.

. Major groups of β-lactamases in Gram-negative bacteria that threaten the role of β-lactam antibiotics
of Neisseria gonorrheae [18] and Haemophilus infl uenzae [19]. Th ese two enzymes occurred in high prevalence among the Enterobacteriaceae prior to the introduction of the broad-spectrum cephalosporins such as cefotaxime and ceftazidime [20,21]. Th e group 2b enzymes are readily inhibited by clavulanic acid, sulbactam, and tazobactam [11]. As a result, infections caused by organisms producing a single penicillinase can be readily treated with a β-lactamase inhibitor combination such as amoxicillin-clavulanic acid, ampicillin-sulbactam, or piperacillin-tazobactam.

Extended-spectrum β-lactamases
Particularly worrisome β-lactamases are found among the extended-spectrum β-lactamases, or ESBLs (functional group 2be or molecular class A). Th ese enzymes were recognized shortly after the introduction of 'βlactamase-stable' cephalosporins and aztreonam. ESBLs were initially identifi ed as variants of the common SHV-1 or TEM-1 β-lactamase, often diff ering from the parent enzymes by only one or two amino acids. After their recognition in the late 1980s concomitantly in Europe [22] and the US [23,24], they became associated with major outbreaks of cephalosporin-resistant infections caused by ESBL-producing E. coli and K. pneumoniae [22,25]. Th e genes that coded for these enzymes were generally found on plasmids that conferred resistance to multiple antibiotic classes and that were readily transferable among species [26]. Even during the fi rst reported outbreaks, other Enterobacteriaceae such as C. freundii, Enterobacter aerogenese, and Serratia marcescens were identifi ed as ESBL producers [27].
ESBLs are still associated with major outbreaks of βlactam resistance. However, the early SHV and TEM variants have been largely replaced by the CTX-M family of ESBLs. Th e fi rst two CTX-M enzymes were identifi ed at approximately the same time in the early 1990s in western Europe and South America in individual clinical isolates [28,29]. Within a decade, the CTX-M β-lactamases became the predominant ESBL family in many medical centers such that they have largely replaced most of the TEM-and SHV-derived ESBLs throughout the world [30][31][32].
CTX-M enzymes were particularly slow to emerge in the US [33] but have recently begun to dominate the ESBL populations of some US health centers [34]. Th e MYSTIC (Meropenem Yearly Susceptibility Test Information Collection) surveillance study of 2007 identifi ed CTX-M genes in 80% of the US medical centers that reported ESBL-producing isolates in their survey [35]. Th e early CTX-M-producing isolates were fre quently resistant to cefotaxime and ceftriaxone but susceptible to ceftazidime because of a strong preference for hydrolysis of the former cephalosporins. However, some members of the CTX-M family demonstrate high hydrolysis rates for all of the extended-spectrum cephalosporins as a result of single amino acid mutations, resulting in complete cephalosporin resistance in all of the producing pathogens [36,37]. It is possible that regional diff erences in the emergence of the CTX-M family were due to localized preferences for therapeutic regimens employing cefotaxime or ceftriaxone in Europe and South America compared with more frequent use of ceftazidime in the US and Canada.

OXA β-lactamases
Th e 'oxacillinase' family of β-lactamases, the OXA enzymes, was originally named to refl ect the ability to hydrolyze oxacillin (or cloxacillin) [19]. Recently, a highly important subgroup of the OXA enzymes was identifi ed as a causative factor for decreasing carbapenem susceptibility in non-fermentative bacteria such as Acinetobacter spp. and Pseudomonas aeruginosa. Th e OXA enzymes in Acinetobacter spp. are mainly chromosomally located, with little apparent transfer among strains [38]. Although these enzymes are structurally related to the older OXA β-lactamases, these new members of the OXA family have carbapenem-hydrolyzing activity. Because hydrolysis rates may be slow compared with other carbapenemases, full carbapenem resistance may require an additional resistance mechanism such as a porin mutation or upregulated effl ux pump [39]. Importantly, carbapenemases within the OXA family have the ability to confer carbapenem resistance to organisms that may already have intrinsic resistance to multiple antibiotic classes [40]. Th is is especially evident in the Acinetobacter spp. that have been identifi ed from soldiers and civilian populations returning from military duty in the Middle East [40].

Serine carbapenemases
Another emerging family of β-lactamases is the serine carbapenemase group (group 2f/class A). Th is group of enzymes has the ability to hydrolyze most β-lactam antibiotics, including the carbapenems that are generally stable to hydrolysis by all other β-lactamases that have serine at the active site of the enzyme [11]. Early reports of this group of chromosomal enzymes were from single isolates of Enterobacteriaceae in the western US and London, followed by small outbreaks in Boston and Chicago a decade later [41]. Th e fi rst plasmid-encoded serine carbapenemases in the KPC enzyme family were reported from the mid-Atlantic region of the US, again in single Klebsiella isolates (from the late 1990s) that were not immediately transferred within the reporting hospitals [42]. However, KPC-producing K. pneumoniae strains soon began to spread into the New York City metropolitan area, then on to Israel, France, and now to many other areas of the world, including southern Europe, southeast Asia, and South America, with high clonality among and within geographical locations [43]. In addition, the genes encoding the KPC enzymes have spread into other Enterobacteriaceae as well as into P. aeruginosa and Acinetobacter spp. [41,44].

Metallo-β-lactamases
Metallo-β-lactamases, or MBLs (functional group 3/ molecular class B) belonging to another carbapenemase family, were initially recognized as species-specifi c βlactamases with limited contributions to the overall resistance profi le in most medical centers [41]. Th ese zinc-requiring chromosomal enzymes appeared in the same strain with other β-lactamases that generally had higher hydrolysis rates for penicillins and cephalosporins. Although these enzymes have relatively weak β-lactamase activity against all β-lactams except monobactams (aztreonam), their distinctive property is the ability to hydrolyze carbapenems [41]. Th e MBLs were thereby responsible for carbapenem resistance in organisms such as Stenotrophomonas maltophilia, Aeromonas spp., and a small population of Bacteroides fragilis. However, the identifi cation of IMP-1, a plasmid-encoded MBL, in 1990 [45] changed our perspective on the group 3 β-lactamases.
Plasmid-mediated carbapenem resistance has now become a serious clinical issue in parts of the world such as southern Europe, Japan, Brazil, and Asia. Th e initial IMP-1 enzyme now belongs to a family with 26 variants; the VIM family of MBLs has 23 variants [6]. Th ese enzymes were fi rst identifi ed in non-fermentative bacteria such as P. aeruginosa and Acinetobacter baumannii but have now spread to many of the Enterobacteriaceae, including Enterobacter spp., E. coli, C. freundii, Klebsiella spp., and S. marcescens [41]. Th ey are frequently associated with integrons (genetic dissemination systems) containing multiple antibiotic resistance deter minants that are easily transferred among species. As with the species-specifi c MBLs, the plasmidic MBLs appear in organisms that almost always produce at least one other β-lactamase with an overlapping hydrolysis profi le. Th us, it is possible that the MBL gene may be lost in the absence of specifi c antibiotic pressure, especially carbapenem therapy.

Nosocomial issues
When broad-spectrum cephalosporins were introduced into clinical practice, ESBL-producing Enterobacteriaceae began to arise prolifi cally, resulting in the loss of these previously eff ective agents. In the late 20th century, many resistant Gram-negative bacteria were often treated with carbapenems as single therapeutic agents. At that time, it seemed that almost all β-lactamase-mediated resistances could be overcome by imipenem or meropenem. Th ese antibiotics were not hydrolyzed by the AmpC cephalosporinases or by ESBLs. Carbapenem monotherapy was the approach taken by some hospitals in which ceftazidime-resistant, ESBL-producing, K. pneumoniae isolates represented a majority of the Klebsiella isolates in the intensive care unit (ICU) [46]. For example, in a New York hospital [46] with a large outbreak of infections caused by ESBL-producing pathogens in the early 1990s, cephalosporin usage was reduced 80% throughout the hospital, accompanied by a 71% reduction in ceftazidimeresistant Klebsiellae in the ICU. However, the collateral damage was that overuse of carbapenems resulted in a 69% increase in imipenem-resistant P. aeruginosa [46] and in infections caused by imipenem-resistant Acinetobacter that were treatable only with polymyxin B or ampicillin-sulbactam [47]. Interestingly, a recent study from a New York ICU showed that strict infection control combined with routine rectal surveillance cultures was the most important factor associated with a reduction in carbapenem-resistant K. pneumoniae isolates but not with the reduction of carbapenem-resistant P. aeruginosa or Acinetobacter spp.; antibiotic usage was not correlated with this reduction [48].
Today, we are seeing situations that are even more complex as multiple broad-spectrum β-lactamases are spreading throughout Gram-negative pathogens. Global dissemination of new β-lactamases is expected, with epidemiological characteristics easily traced with currently available molecular techniques. Recent studies have demon strated the clonal appearance of a highly homo geneous CTX-M-15-producing E. coli strain that was present in both hospital and community isolates from eight countries in Europe, North America, and Asia [49]. Th e fi rst KPC-2-producing K. pneumoniae isolate reported in France was directly traced to a patient who had previously been treated in a New York City hospital [50]; subsequent analyses of KPC-producing K. pneumoniae isolates from the Centers for Disease Control and Prevention (CDC) identifi ed a major sequence type that accounts for over 70% of the CDC KPC isolates [43]. OXA carbapenemases have been identifi ed in multiple Acinetobacter spp. clones worldwide [40,51,52].
Organisms with multiple β-lactamases responsible for high-level resistance to most β-lactams are also appearing more frequently. A K. pneumoniae clinical isolate has been reported to produce eight diff erent β-lactamases, including at least one ESBL, an AmpC, and a KPC enzyme [53]. Th e MBL VIM-1 was identifi ed in a Greek K. pneumoniae isolate that also produced the KPC-2 carbapenemase [54]. Carbapenem-resistant Acineto bacter isolates have now been reported with both a VIM-2-like MBL as well as an OXA-23 carbapenemase in the same strain [55]. Th ese organisms are resistant not only to all β-lactams but to many other antibiotic classes as well. Multidrug resistance has recently been reported in KPCproducing K. pneumoniae in metropolitan New York City [56] and in MBL-producing isolates [57] because of plasmid-encoded resistance determinants that confer resistance to aminoglycosides, fl uoroquinolones, trimetho prim, sulphonamides, and chloramphenicol. Th e Infectious Diseases Society of America (Arlington, VA, USA) made Gram-negative resistance a major focus in their 2009 list of ESKAPE pathogens that include multidrug-resistant K. pneumoniae, Acinetobacter spp., P. aeruginosa, and Enterobacter spp. [3].
On a more optimistic note, perhaps plasmid-encoded MBLs and their co-production with non-carbapenemases may not be the disaster some have envisioned. In the absence of a carbapenem-selective pressure in a patient, it is possible that the MBL gene may be lost, although that organism would still maintain its resistance profi le to other β-lactams. Th is speculation is based on the experiences of Japan and Italy, two countries where plasmid-encoded MBLs were fi rst identifi ed. In Japan and Italy, MBL genes that have spread countrywide after their initial reports have been responsible primarily for small sporadic outbreaks in these countries and have not been a continuous cause of carbapenem resistance throughout all Gram-negative species at a single medical center [58,59]. However, co-production of two carbapenemases, such as a VIM and KPC or a VIM and an OXA carbapenemase, may lead to much more serious consequences in which β-lactam antibiotics could never be useful.

Community reservoirs
Although the main focus has been on multidrug-resistant infections in the hospital setting, community sources for β-lactamase-mediated resistances are being reported more frequently. Community-acquired ESBLs in locations such as nursing homes or long-term health care resi dences have been described for several years [31,32,[60][61][62]. As one might expect, the even more deleterious plasmid-encoded carbapenemases are also being described in patients outside the hospital. Nine E. coli isolates producing the KPC-2 and KPC-3 carbapenemases, enzymes most frequently found in K. pneumoniae strains, were identifi ed in seven nursing homes in the New York City area [63]. In Greece, a set of 45 patients was identifi ed with community-onset urinary tract infections caused by VIM-2 MBL-producing P. aeruginosa isolates [64]. It is noteworthy that the fi rst CTX-M-producing E. coli isolates in the US were identifi ed from sources that included a number of urinary tract isolates originating from outpatients [34]. Th us, even if infection control procedures are instituted and strictly enforced in hospitals and medical centers, resistant Gram-negative pathogens have already been released into the community, where they can reside in relatively healthy people until challenged by changes in disease status, at which time they may become the causative pathogen for a new infection.

Conclusions
Resistance in Gram-negative pathogens is increasing at an alarming rate. New β-lactamases that are transferred among species on plasmids with multiple resistance factors are also being described as a continuing exercise. Th e implications for therapeutic use of the workhorse β-lactam antibiotics are sobering. However, it is possible that some of these β-lactamases may exact a high price from the producing organism, especially when multiple β-lactamases are produced by a single strain. Small changes in therapeutic approaches may allow the fl ora to be altered within a medical center so that bacterial producers of ESBLs or KPCs or MBLs are only a small portion of the nosocomial fl ora and can be contained in isolated areas. Combination therapy with a β-lactamase inhibitor combination or a carbapenem, and at least one agent from another antibiotic class, may be eff ective against isolates that have decreased susceptibilities. Use of the maximally approved therapeutic doses should always be considered for serious infections when βlactam MICs are close to the susceptibility breakpoints. Finally, β-lactam antibiotics should not be used if their MICs are in the highly resistant category, particularly in a hospital known to have high levels of plasmid-encoded β-lactamases that are being transferred throughout the facility; it should be noted that this epidemiological information may need to be determined with the assistance of a reference laboratory. With judicious use of antibiotics and strict infection control procedures, it may be possible to limit the eff ects of these newer β-lactamases until the time when new antibacterial agents are developed to counteract their eff ects.
She has received compensation from Bristol-Myers Squibb Company and Wyeth, both of which hold patents on β-lactam antibiotics. No organization supported the writing or publishing of this manuscript.