AmpC β-Lactamases

The first bacterial enzyme reported to destroy penicillin was the AmpC β-lactamase of Escherichia coli, although it had not been so named in 1940 (1). Swedish investigators began a systematic study of the genetics of penicillin resistance in E. coli in 1965. Mutations with stepwise-enhanced resistance were termed ampA and ampB (84, 85). A mutation in an ampA strain that resulted in reduced resistance was then designated ampC. ampA strains overproduced β-lactamase, suggesting a regulatory role for the ampA gene (180). ampB turned out not to be a single locus, and such strains were found to have an altered cell envelope (236). ampC strains made little if any β-lactamase, suggesting that ampC was the structural gene for the enzyme (46). Most of the amp nomenclature has changed over the years, but the designation ampC has persisted. The sequence of the ampC gene from E. coli was reported in 1981 (144). It differed from the sequence of penicillinase-type β-lactamases such as TEM-1 but, like them, had serine at its active site (161). In the Ambler structural classification of β-lactamases (7), AmpC enzymes belong to class C, while in the functional classification scheme of Bush et al. (47), they were assigned to group 1.

When the functional classification scheme was published in 1995, chromosomally determined AmpC β-lactamases in Enterobacteriaceae and also in a few other families were known (47). Since then, the number of sequenced bacterial genes and genomes has grown enormously. In GenBank, ampC genes are included in COG 1680, where COG stands for cluster of orthologous groups. COG 1680 comprises other penicillin binding proteins as well as class C β-lactamases and includes proteins from archaea as well as bacteria, gram-positive as well as gram-negative organisms, strict anaerobes along with facultative ones, and soil and water denizens as well as human pathogens, such as species of Legionella and Mycobacterium. Sequence alone is insufficient to differentiate an AmpC β-lactamase from ubiquitous low-molecular-weight penicillin binding proteins involved in cell wall biosynthesis, such as d-peptidase (d-alanyl-d-alanine carboxypeptidase/transpeptidase). Both have the same general structure and share conserved sequence motifs near an active-site serine (149, 162). E. coli even produces a β-lactam binding protein, AmpH, which is related to AmpC structurally but lacks β-lactamase activity (121). The AmpC name is not trustworthy since several enzymes so labeled in the literature actually belong to class A (177, 337). Cephalosporinase activity is not reliable either, since some β-lactamases with predominant activity on cephalosporins belong to class A (97, 205, 278, 298). Accordingly, the conservative listing of AmpC β-lactamases in Table 1 includes proteins with the requisite structure from organisms that have been demonstrated to possess appropriate AmpC-type β-lactamase activity. It is undoubtedly incomplete. For example, organisms not yet shown to produce a functional AmpC-type enzyme but with identified ampC genes include such diverse bacteria as Agrobacterium tumefaciens (110), Coxiella burnetii (GenBank accession number YP_001424134), Legionella pneumophila (56), Rickettsia felis (239), and Sinorhizobium meliloti (127). For other organisms, supportive MIC or enzymatic but not structural data are available for the presence of AmpC β-lactamase, including Enterobacter sakazakii (258), Ewingella americana (311), Providencia rettgeri (207), and several species of Serratia (306, 307) and Yersinia (215, 288, 313). The phylum Proteobacteria contains the largest number, but at least one acid-fast actinobacterium also produces AmpC β-lactamase. Sequence variation occurs within each type. For example, more than 25 varieties of AmpC β-lactamase that share ≥94% protein sequence identity have been described for Acinetobacter spp. (137; G. Bou et al., personal communication), and GenBank contains similar multiple listings for E. coli, Enterobacter cloacae, Pseudomonas aeruginosa, and other organisms. Some frequently encountered Enterobacteriaceae are conspicuous by their absence. Klebsiella pneumoniae, Klebsiella oxytoca, Proteus mirabilis, and Salmonella spp. (31) lack a chromosomal bla_AmpC gene, as do Citrobacter amalonaticus (328), Citrobacter farmeri, Citrobacter gillenii (224), Citrobacter koseri (formerly Citrobacter diversus and Levinea malonatica), Citrobacter rodentium, Citrobacter sedlakii (252), Edwardsiella hoshinae, Edwardsiella ictaluri (312), Kluyvera ascorbata (138, 305), Kluyvera cryocrescens (72), Plesiomonas shigelloides (9), Proteus penneri (175), Proteus vulgaris (60), Rahnella aquatilis (30, 308), Yersinia pestis, and Yersinia pseudotuberculosis (313) as well as, probably, Escherichia hermannii (91), Francisella tularensis (27), Shewanella algae (123), and Stenotrophomonas maltophilia (111). However, since bla_AmpC genes occur on transmissible plasmids, the clinical microbiologist needs to consider this resistance mechanism whatever the identification of an organism.

AmpC enzymes typically have molecular masses of 34 to 40 kDa and isoelectric points of >8.0, although the isoelectric points of plasmid-mediated FOX enzymes are lower (6.7 to 7.2) (254), and an AmpC enzyme from Morganella morganii has an isoelectric point of 6.6 (264). The enzymes are located in the bacterial periplasm, with the exception of the AmpC β-lactamase of Psychrobacter immobilis, which is secreted mainly into the external medium (88). They are active on penicillins but even more active on cephalosporins and can hydrolyze cephamycins such as cefoxitin and cefotetan; oxyiminocephalosporins such as ceftazidime, cefotaxime, and ceftriaxone; and monobactams such as aztreonam but at a rate <1% of that of benzylpenicillin (Table 2, which also shows data for class A and D β-lactamases for comparison). Although the hydrolysis rate for such substrates is low due to slow deacylation (99), the enzyme affinity, as reflected by a low K_m, is high (Table 3), a factor that becomes important at low substrate concentrations. The hydrolysis rates for cefepime, cefpirome, and carbapenems are also very low, and the estimated K_m values for cefepime and cefpirome are high, reflecting lower enzyme affinity (283).

With preferred cephalosporin substrates, the turnover rate of the E. cloacae P99 β-lactamase is diffusion limited rather than catalysis limited, implying that AmpC enzymes have evolved to maximal efficiency (45). Such data also suggest that AmpC β-lactamase evolved to deal with cephalosporins rather than for some other cellular function, although there is some evidence to suggest that these enzymes play a morphological role (121).

Inhibitors of class A enzymes such as clavulanic acid, sulbactam, and tazobactam have much less effect on AmpC β-lactamases, although some are inhibited by tazobactam or sulbactam (48, 157, 218). AmpC β-lactamases are poorly inhibited by p-chloromercuribenzoate and not at all by EDTA. Cloxacillin, oxacillin, and aztreonam, however, are good inhibitors (47).

The known three-dimensional structures of AmpC enzymes are very similar (Fig. 1). There is an α-helical domain on one side of the molecule (Fig. 1, left) and an α/β domain on the other (Fig 1, right). The active site lies in the center of the enzyme at the left edge of the five-stranded β-sheet with the reactive serine residue at the amino terminus of the central α-helix (162, 190). The active site can be further subdivided into an R1 site, accommodating the R1 side chain of the β-lactam nucleus, and an R2 site for the R2 side chain (Fig. 2). The R1 site is bounded by the Ω-loop, while the R2 site is enclosed by the R2 loop containing the H-10 and H-11 helices. Overall, the AmpC structure is similar to that of class A β-lactamases (and dd-peptidase) except that the binding site is more open in class C enzymes, reflecting their greater ability to accommodate the bulkier side chains of cephalosporins. Key catalytic residues in addition to Ser64 for AmpC enzymes include Lys67, Tyr150, Asn152, Lys315, and Ala318, with substitutions at these sites lowering enzymatic activity dramatically (54). In the folded protein, most of these essential residues are found at the active site, with Lys67 hydrogen bonded to Ser64 and Tyr150 acting as a transient catalytic base (79).

In many Enterobacteriaceae, AmpC expression is low but inducible in response to β-lactam exposure. The induction mechanism is complex (118, 139, 140). The disruption of murein biosynthesis by a β-lactam agent leads to an accumulation of N-acetylglucosamine-1,6-anhydro-N-acetylmuramic acid oligopeptides. The N-acetylglucosamine moiety is removed to produce a series of 1,6-anhydro-N-acetylmuramic acid tri-, tetra-, and pentapeptides. These oligopeptides compete with oligopeptides of UDP-N-acetylmuramic acid for a binding site on AmpR, a member of the LysR transcriptional regulator family. Displacement of the UDP-N-acetylmuramic acid peptides signals a conformational change in AmpR, which activates the transcription of ampC. In addition, the cell has an enzyme, AmpD, a cytoplasmic N-acetyl-muramyl-l-alanine amidase, that removes stem peptides from the 1,6-anhydro-N-acetylmuramic acid and N-acetylglucosamine-1,6-anhydro-N-acetylmuramic acid oligopeptide derivatives, thus reducing their concentrations and preventing the overexpression of AmpC.

The most common cause of AmpC overexpression in clinical isolates is a mutation in ampD leading to AmpC hyperinducibility or constitutive hyperproduction (289). AmpR mutations are less common but can also result in high-constitutive or hyperinducible phenotypes (118, 153, 165). Least common are mutations in AmpG, which result in constitutive low-level expression. AmpG is an inner membrane permease that transports the oligopeptides involved in cell wall recycling and AmpC regulation into the cytosol (179).

Different organisms add additional features to AmpC regulation. E. coli lacks an ampR gene (129). Consequently, AmpC in E. coli is noninducible but is regulated by promoter and attenuator mechanisms (145), as is AmpC production in Shigella (33). Acinetobacter baumannii also lacks an ampR gene so that its AmpC β-lactamase is noninducible (39). AmpC in Serratia marcescens is regulated by ampR, but the ampC transcript has an unusual untranslated region of 126 bases forming a stem-loop structure that influences the transcript half-life (191). P. aeruginosa PAO1 has three ampD genes, explaining the stepwise upregulation of AmpC production seen in this organism with the successive inactivation of each ampD gene (151). The multiple ampD loci contribute to virulence since a P. aeruginosa strain partially derepressed by the inactivation of one ampD allele remains fully virulent, while double or triple ampD mutants lose the ability to compete in a mouse model of systemic infection (219). Other aspects of AmpC regulation in P. aeruginosa are also more complex than that in the Enterobacteriaceae. AmpR is involved in the regulation of other genes besides AmpC (164), an ampE gene encoding a cytoplasmic membrane protein acting as a sensory transducer has a role in ampC expression as part of an ampDE operon (150), and the CreBCD system as well as dacB, encoding a nonessential penicillin binding protein, are involved in AmpC hyperproduction as well (219a).

β-Lactams differ in their inducing abilities (184, 189, 285, 302). Benzylpenicillin, ampicillin, amoxicillin, and cephalosporins such as cefazolin and cephalothin are strong inducers and good substrates for AmpC β-lactamase. Cefoxitin and imipenem are also strong inducers but are much more stable for hydrolysis (Table 4). Cefotaxime, ceftriaxone, ceftazidime, cefepime, cefuroxime, piperacillin, and aztreonam are weak inducers and weak substrates but can be hydrolyzed if enough enzyme is made. Consequently, MICs of weakly inducing oxyimino-β-lactams are dramatically increased with AmpC hyperproduction. Conversely, MICs of agents that are strong inducers show little change with regulatory mutations because the level of induced ampC expression is already high (Table 4). β-Lactamase inhibitors are also inducers, especially clavulanate, which has little inhibitory effect on AmpC β-lactamase activity (336) but can paradoxically appear to increase AmpC-mediated resistance in an inducible organism (160). The inducing effect of clavulanate is especially important for P. aeruginosa, where clinically achieved concentrations of clavulanate by inducing AmpC expression have been shown to antagonize the antibacterial activity of ticarcillin (181).

The AmpC enzyme in Aeromonas spp. is controlled, along with two other chromosomally encoded β-lactamases, not by an AmpR-type system but by a two-component regulator, termed brlAB in Aeromonas hydrophila (5, 234). BrlB is a histidine sensor kinase, the regulated β-lactamase genes are preceded by a short sequence tag (TTCAC), and an inner membrane protein is also involved in regulation, but the chemical signal for induction is not yet known (10). E. coli has a homologous regulatory system, and there is some evidence that two-component regulators also play a role in the expression of E. coli ampC (128).

In addition to the amount and intrinsic activity of β-lactamase, the rate at which the substrate is delivered to the enzyme is an important determinant of the resistance spectrum. The concentration of β-lactam substrate in the periplasm is a function of the permeability of the cell's outer membrane, in particular the presence of porin channels through which β-lactams penetrate and of efflux pumps, which transport them out of the cell. At one time, the binding of substrate to AmpC β-lactamase was entertained as a mechanism to explain resistance to β-lactams that appeared to be poorly hydrolyzed (316). Vu and Nikaido pointed out, however, that at the concentration of β-lactams in the periplasm needed to inhibit target penicillin binding proteins, AmpC β-lactamases can hydrolyze cephalosporins despite a low V_max if the substrate also has a low K_m (330). Decreasing the number of porin entry channels or increasing efflux pump expression can lower influx and further augment enzyme efficiency. Thus, carbapenem resistance in clinical isolates of P. aeruginosa involves various combinations of overproduction of AmpC β-lactamase, decreased production of the OprD porin channel for imipenem entry, and activation of MexAB-OprM and other efflux systems (114, 163, 185, 268). Also, cephalosporins with both positive and negative charges (i.e., zwitterionic molecules) such as cefepime and cefpirome have the advantage of penetrating the outer bacterial membrane more rapidly than those with a net positive charge, such as cefotaxime and ceftriaxone, thus more easily reaching their lethal targets without β-lactamase inactivation (233).

The serine β-lactamases are ancient enzymes estimated to have originated more than 2 billion years ago. A structure-based phylogeny indicates that the divergence of AmpC-type enzymes predated the divergence of class A and class D β-lactamases from a common ancestor (116). Figure 3 provides an overview of the phylogenetic relationship between the enzymes listed in Table 1. As would be expected, AmpC enzymes from organisms belonging to the same genus cluster together, while the AmpC β-lactamases of Enterobacteriaceae, Pseudomonas, and Acinetobacter are more distantly related.

Plasmid-encoded AmpC genes have been known since 1989 (Table 5) (254, 335). They have been found around the world in nosocomial and nonnosocomial isolates, having been most easily detected in those enterobacteria not expected to produce an AmpC β-lactamase. Minor differences in amino acid sequence have given rise to families. Forty-three CMY alleles are currently known (http://www.lahey.org/Studies/ ), and in GenBank, sequence data can be found (some of it unpublished) for seven varieties of FOX; four varieties of ACC, LAT, and MIR; three varieties of ACT and MOX; and two varieties of DHA. Some of these varieties are determined by chromosomal genes and represent possible progenitors for the plasmid-determined enzymes.

As indicated in Table 5, the plasmid-determined enzymes are related, sometimes very closely, to chromosomally determined AmpC β-lactamases. CMY is represented twice since it has two quite different origins. Six current varieties (CMY-1, -8, -9, -10, -11, and -19) are related to chromosomally determined AmpC enzymes in Aeromonas spp., while the remainder (including CMY-2, the most common plasmid-mediated AmpC β-lactamase worldwide) are related to AmpC β-lactamases of Citrobacter freundii. The LAT enzymes have a similar origin, but of the four original LAT enzymes, improved sequencing disclosed that LAT-2 was identical to CMY-2, LAT-3 was identical to CMY-6, and LAT-4 was identical to LAT-1, which is the only one remaining unique (15).

Like the chromosomally determined AmpC β-lactamases, the plasmid-mediated enzymes confer resistance to a broad spectrum of β-lactams (Table 6) including penicillins, oxyimino-β-cephalosporins, cephamycins, and (variably) aztreonam. Susceptibility to cefepime, cefpirome, and carbapenems is little, if at all, affected. Note that ACC-1 is exceptional in not conferring resistance to cephamycins and is actually cefoxitin inhibited (21, 106).

The genes for ACT-1, DHA-1, DHA-2, and CMY-13 are linked to ampR genes and are inducible (16, 93, 214, 274), while other plasmid-mediated AmpC genes are not, including other CMY alleles and apparently CFE-1 despite its linkage to an ampR gene (142, 229). Nonetheless, the level of expression of both inducible ACT-1 and noninducible MIR-1 is 33- to 95-fold higher than the level of expression of the chromosomally determined AmpC gene of E. cloacae thanks to a higher gene copy number for the plasmid-determined enzymes (2 copies for bla_ACT-1 and 12 copies for bla_MIR-1) and greater promoter strength for the plasmid genes (8-fold increased from the hybrid MIR-1 promoter and 17-fold increased because of a single base change relative to the wild type in the ACT-1 promoter) (275, 276). AmpC plasmids lack ampD genes, but the level of ACT-1 expression is increased with the loss of chromosomal AmpD function (276).

An AmpD-deficient E. coli strain producing ACT-1 remains susceptible to imipenem (MIC, 2 μg/ml) (276), but imipenem MICs of ≥16 μg/ml have been found in clinical isolates of K. pneumoniae carrying ACT-1 plasmids associated with a loss of outer membrane porins (41). In a porin-deficient K. pneumoniae isolate, other plasmid-mediated AmpC enzymes also provide imipenem, ertapenem, and meropenem resistance (141). Such strains generally remain susceptible to cefepime but are otherwise also resistant to oxyimino-β-cephalosporins.

Plasmids carrying genes for AmpC β-lactamases often carry multiple other resistances including genes for resistance to aminoglycosides, chloramphenicol, quinolones, sulfonamide, tetracycline, and trimethoprim as well as genes for other β-lactamases such as TEM-1, PSE-1 (6), CTX-M-3 (55), SHV varieties (119), and VIM-1 (214). The AmpC gene is usually part of an integron but is not incorporated into a gene cassette with an affiliated 59-base element (273). Note that the same bla_AmpC gene can be incorporated into different backbones on different plasmids (50).

A variety of genetic elements have been implicated in the mobilization of AmpC genes onto plasmids (Fig. 4). The insertion sequence ISEcp1 (or truncated versions thereof) is associated with many CMY alleles including CMY-2 (105, 115, 155), CMY-4 (228), CMY-5 (343), CMY-7 (135), CMY-12 (182), CMY-14 (182), CMY-15 (182), CMY-16 (69), CMY-21 (133), CMY-31 (GenBank accession number EU331425), and CMY-36 (GenBank accession number EU331426) as well as the β-lactamases ACC-1 (78, 249) and ACC-4 (247). ISEcp1 plays a dual role. It is involved in the transposition of adjacent genes (261) and has been shown able to mobilize a chromosomal bla gene onto a plasmid (166), and it also can supply an efficient promoter for the high-level expression of neighboring genes. The transcription of at least CMY-7 has been shown to start within the ISEcp1 element and takes place at a much higher level than the expression of the corresponding AmpC gene in C. freundii (135).

Other bla_AmpC genes are found adjacent to an insertion sequence common region (ISCR1) involved in gene mobilization into (typically) complex class 1 integrons (322). Genes for several CMY varieties (CMY-1, -8, -9, -10, -11, and -19), DHA-1, and MOX-1 are so linked (322, 332). On the other hand, the gene for CMY-13 and its attendant ampR gene are bounded by directly repeated IS26 elements made up of a transposase gene (tnpA) with flanking inverted terminal repeat segments (214). Other elements are associated with and may have been involved in capturing the genes for FOX-5 (269), MIR-1 (142), and MOX-2 (271).

Just as amino acid alterations in TEM and SHV β-lactamase have given rise to extended-spectrum enzymes with broader substrate specificities, amino acid insertions, deletions, and substitutions have been described for AmpC β-lactamases that enhance catalytic efficiency toward oxyimino-β-lactam substrates (235). Such changes in both plasmid-determined and chromosomally mediated AmpC enzymes have been described. Their properties are shown in Table 7. The alterations occur either in the Ω-loop, making the enzyme more accessible for substrates with bulky R1 side chains, or at or near the R2 loop, widening the R2 binding site. At both locations, the amino acid alterations can have opposite effects on enzyme kinetics. Generally, the catalytic constant for ceftazidime increased along with the K_m, or the K_m decreased (reflecting greater affinity), but the k_cat decreased as well. In either case, the k_cat/K_m ratio or catalytic efficiency for ceftazidime and related substrates increased compared to that of the wild-type enzyme with the result that the ceftazidime MICs for a strain carrying such enzymes were in the resistance range (MIC ≥ 32 μg/ml), while the MICs for cefotaxime and cefepime usually reflected only reduced susceptibility, such as a cefepime MIC of 8 μg/ml for E. coli with the AmpC enzymes from E. cloacae CHE or Enterobacter aerogenes Ear2. The enzyme from S. marcescens HD, however, when expressed in E. coli, conferred a cefepime MIC of 512 μg/ml (196), and those from E. coli strains EC14, EC18, and BER were associated with cefepime MICs of 16 μg/ml (197, 199). MICs for aztreonam and imipenem were usually little affected except that an aztreonam MIC of 128 μg/ml was produced by CMY-10 (172). Structural gene mutations were often accompanied by promoter mutations that increased the level of expression of the mutant gene (193). Modifications at additional enzyme sites in laboratory mutant have been described (14). Interestingly, the AmpC variant from E. coli HKY28 became more susceptible to inhibition by clavulanic acid, sulbactam, and tazobactam, a curious phenotype previously described for a few other AmpC variants (13, 341).

There are presently no CLSI or other approved criteria for AmpC detection (76). Organisms producing enough AmpC β-lactamase will typically give a positive ESBL screening test but fail the confirmatory test involving increased sensitivity with clavulanic acid (29, 304). This phenotype is not, however, specific for an AmpC producer, since it can occur with certain complex TEM mutants (277), OXA-type ESBLs, and carbapenemases and in strains with high levels of TEM-1 β-lactamase. Except for non-lactose-fermenting gram-negative organisms intrinsically resistant to cephamycins, resistance to cefoxitin as well as oxyimino-β-lactams is suggestive of an AmpC enzyme, but it is not specific since cefoxitin resistance can also be produced by certain carbapenemases (262) and a few class A β-lactamases (331) and by decreased levels of production of outer membrane porins in both K. pneumoniae and E. coli (124, 125, 202, 203). Furthermore, some plasmid-mediated AmpC strains test susceptible to ceftriaxone, cefotaxime, and ceftazidime by current CLSI criteria and could easily be overlooked (315). Other confirmatory tests are needed (Table 9).

The three-dimensional test was designed to detect both AmpC and ESBL production. In the “indirect” form used for AmpC detection, a conventional disk diffusion susceptibility assay is carried out with a susceptible strain, such as E. coli ATCC 25922, as the lawn and a suspension of the test organism, which is added to a circular slit in the agar 3 mm from a disk containing cefoxitin or some other agent. Distortion of the zone of inhibition indicates a positive test, as cefoxitin is hydrolyzed by the presence of an AmpC enzyme (319). In subsequent modifications, a radial slit was employed, and rather than using intact cells, the test organisms were concentrated by centrifugation, and the pellet was freeze-thawed five to seven times to release β-lactamase (66, 200). Direct spot inoculation of the test organism 7 to 8 mm from the cefoxitin disk has also been used successfully (295), as has a heavy inoculum streaked radially from the cefoxitin disk on the agar surface without using a slit (169), although the latter procedure missed some CMY-2- and DHA-1-producing strains. In a further modification, the test organism has been applied to a filter paper disk containing Tris-EDTA to enhance membrane permeability, with the disk then placed onto a lawn of E. coli ATCC 25922 adjacent to a cefoxitin disk (35). In every case, the presence of an AmpC β-lactamase is indicated by a distortion of the inhibition zone around the cefoxitin disk. Organisms producing a carbapenemase can mimic an AmpC β-lactamase in cefoxitin inactivation, so reduced carbapenem susceptibility is important to exclude since otherwise, a carbapenem might be selected for therapy (35).

A variation on the three-dimensional test is to plate the sensitive indicator strain on agar containing 4 μg/ml cefoxitin and add the freeze-thawed cell extract to a well in the plate. After incubation, growth around the well indicates the presence of a cefoxitin-hydrolyzing enzyme (230). This method is reported to be just as sensitive and specific as the three-dimensional test for AmpC detection, is easier to perform, and allows multiple samples per plate to be tested.

Another approach for AmpC detection is the use of an inhibitor for this β-lactamase class analogous to the use of clavulanic acid in a confirmatory test for class A ESBLs. The β-lactams LN-2-128, Ro 48-1220, and Syn 2190 have been evaluated for this purpose, with the best results from the combination of Syn 2190 and cefotetan, which was 100% specific and 91% sensitive in AmpC β-lactamase detection (36, 37). Unfortunately, these inhibitors are not commercially available.

A double-disk test with a 500-μg cloxacillin disk placed between disks containing ceftazidime and cefotaxime on a lawn of the test organism has been explored using 15 AmpC-producing strains. All showed synergy. A central cefoxitin disk produced synergy with ceftazidime and cefotaxime only with ACC-1 β-lactamase and also revealed the inducibility of enzymes such as DHA-1 (280).

Etest strips with a gradient of cefotetan or cefoxitin on one half and the same combined with a constant concentration of cloxacillin on the other half have been evaluated for AmpC detection (38). Either a reduction in cephamycin MIC of at least three dilutions, deformation of the ellipse of inhibition, or a “phantom zone” was interpreted as a positive test. With almost 500 test strains, the overall sensitivity and specificity were 88 to 93% (83).

Boronic acids have long been known as AmpC inhibitors (28). Various boronic acid derivatives have been either added to a blank disk placed near a β-lactam disk or added to the β-lactam disk for comparison with an unmodified β-lactam disk. For example, Yagi et al. found that a disk potentiation test utilizing a ≥5-mm enhancement of the zone of inhibition around a ceftazidime or cefotaxime disk when 300 μg 3-aminophenylboronic acid was added reliably detected all AmpC varieties tested but was negative with strains producing ESBLs and carbapenemases (344), findings that have been confirmed with a different set of strains (143). Strains producing both a plasmid-mediated AmpC β-lactamase and an ESBL have been reliably detected (301), but such a test cannot differentiate between an AmpC enzyme encoded on a plasmid or on the chromosome. Specificity is also a concern since boronic acids also enhance the sensitivity of strains making a non-AmpC enzyme, class A KPC β-lactamase (250, 324).

Phenotypic tests cannot distinguish among the various families of plasmid-mediated AmpC enzymes and may also overlook chromosomally determined AmpC β-lactamases with an extended spectrum (193). For these purposes, and as the current “gold standard” for plasmid-mediated AmpC β-lactamase detection, multiplex PCR has been developed by utilizing six primer pairs (251) to which a seventh pair for CFE-1 β-lactamase (229) could be added. Chromosomal bla_AmpC did not interfere in testing strains of K. pneumoniae, E. coli, P. mirabilis, or S. enterica but could be a problem with bla_AmpC genes in one of the genera from which the plasmid-mediated enzymes are derived. (Table 5). A multiplex asymmetric PCR-based microarray method for detecting genes for both plasmid-mediated AmpC β-lactamases and mutations responsible for the ESBL phenotype in bla_SHV has been described (351). Perfection of a PCR array technology may ultimately allow the automation of AmpC β-lactamase detection for a suitably equipped clinical laboratory.

Is the recognition of plasmid-mediated AmpC enzymes necessary for the average laboratory? Therapeutic and infection control considerations argue that it is. AmpC-producing isolates may appear to be susceptible in vitro to some cephalosporins and aztreonam yet fail to respond if those agents are used so that a specific test for their presence is necessary (318). Compared to ESBL producers, isolates producing AmpC β-lactamase are resistant to additional β-lactams and insusceptible to currently available β-lactamase inhibitors and have the potential for developing resistance to carbapenems. Furthermore, plasmid mediation of AmpC carries the threat of spread to other organisms within a hospital or geographic region. Time will tell whether these considerations will still apply if cephalosporin breakpoints are significantly lowered so that decisions about therapy become based only on low-MIC susceptibility.

Strains with ampC genes are often resistant to multiple agents, making the selection of an effective antibiotic difficult. β-Lactam/β-lactamase inhibitor combinations and most cephalosporins and penicillins should be avoided because of in vitro resistance, the potential for AmpC induction or selection of high-enzyme-level mutants, and documented poor clinical outcomes with ceftazidime, cefotaxime (244), and, in an animal model, piperacillin-tazobactam (329). Whether cefepime can be used is unsettled. Cefepime is a poor inducer of AmpC β-lactamase, rapidly penetrates through the outer cell membrane, and is little hydrolyzed by the enzyme (232, 283), so many AmpC-producing organisms test cefepime susceptible with a conventional inoculum (see Table 6 for examples). If a 100-fold-higher inoculum is used, however, cefepime MICs increase dramatically for some AmpC producers, suggesting caution in its use (154, 256), and some strains are frankly resistant (238). In a pneumonia model using guinea pigs, cefepime, imipenem, and meropenem were equally effective against a porin-deficient K. pneumoniae strain producing FOX-5 β-lactamase (255). Also, in a rat pneumonia model with a K. pneumoniae strain producing ACT-1, β-lactam therapy with imipenem, meropenem, ertapenem, or cefepime gave equivalent results, even if the test strain was porin deficient (243). However, in a mouse pneumonia model with a porin-deficient strain of K. pneumoniae producing CMY-2 β-lactamase, survival with cefepime therapy was no better than that without antibiotic and significantly inferior to that with imipenem treatment (256). Nonetheless, cefepime has cured infections due to multiply resistant Enterobacter spp. including those with reduced susceptibility to ceftazidime (286), and in a prospective, randomized study of ICU patients with nosocomial pneumonia having P. aeruginosa as the most common isolate, cefepime proved to be just as effective as imipenem (350). The jury is still out, but cefepime seems to be an exception to the recommendation to avoid all cephalosporin therapy even if an AmpC-producing isolate tests susceptible to an individual agent.

Temocillin, a 6-α-methoxy derivative of ticarcillin, is active in vitro against many AmpC-producing Enterobacteriaceae whether the enzyme is determined by chromosomal or plasmid genes and is also active against ESBL producers (108, 187), but clinical experience is limited, and it is available in only a few countries. Amdinocillin is also effective in vitro against AmpC-producing E. coli strains but shows a marked inoculum effect unless clavulanic acid is present (43) and is also not available in the United States.

Carbapenem therapy has usually been successful (244) but has also been followed by the emergence of carbapenem-resistant K. pneumoniae associated with ACT-1 β-lactamase production and outer membrane porin loss (3, 41, 152). Reduced imipenem susceptibility (MIC 8 to 32 μg/ml) has also been reported in porin-deficient clinical isolates of K. pneumoniae making AmpC enzymes ACC-1 (34), CMY-2 (171), CMY-4 (49), DHA-1 (171), or an uncharacterized AmpC-type enzyme (246). The same scenario has been described for clinical isolates of E. aerogenes (59, 71, 317, 325, 349), E. cloacae (168), and C. freundii (192) as well as laboratory mutants (131, 270, 327). In E. coli, reduced carbapenem susceptibility or frank resistance (imipenem MIC of 8 to 128 μg/ml) in porin-deficient clinical isolates producing CMY-2 (183) or CMY-4 (303) has been described, while a Salmonella enterica strain lacking a porin and making CMY-4 reached an imipenem MIC of 32 μg/ml (8).

If the isolate is susceptible, fluoroquinolone therapy is an option especially for non-life-threatening infections such as urinary tract infection. Tigecycline is another option. It had good activity in vitro against 88% of AmpC-hyperproducing isolates of E. coli, Enterobacter spp., Klebsiella spp., and Citrobacter spp. from the United Kingdom (130), but few P. aeruginosa isolates (282) and, in some centers, only 22% of nosocomial Acinetobacter isolates (231) were tigecycline susceptible.

AmpC β-lactamases are clinically important cephalosporinases encoded on the chromosome of many Enterobacteriaceae and a few other organisms where they mediate resistance to cephalothin, cefazolin, cefoxitin, most penicillins, and β-lactamase inhibitor/β-lactam combinations. In many bacteria, AmpC enzymes are inducible and can be expressed at high levels by mutation. Overexpression confers resistance to broad-spectrum cephalosporins including cefotaxime, ceftazidime, and ceftriaxone and is a problem especially in infections due to E. aerogenes and E. cloacae, where an isolate initially susceptible to these agents may become resistant upon therapy. Transmissible plasmids have acquired genes for AmpC enzymes, which consequently can now appear in bacteria lacking or poorly expressing a chromosomal bla_AmpC gene, such as E. coli, K. pneumoniae, and P. mirabilis. Resistance due to plasmid-mediated AmpC enzymes is less common than ESBL production in most parts of the world but may be both harder to detect and broader in spectrum. AmpC enzymes encoded by both chromosomal and plasmid genes are also evolving to hydrolyze broad-spectrum cephalosporins more efficiently. Techniques to identify AmpC β-lactamase-producing isolates are available but are still evolving and are not yet optimized for the clinical laboratory, which probably now underestimates this resistance mechanism. Carbapenems can usually be used to treat infections due to AmpC-producing bacteria, but carbapenem resistance can arise in some organisms by mutations that reduce influx (outer membrane porin loss) or enhance efflux (efflux pump activation).