Each of the 18 strains with reduced susceptibility to cefoxitin was also resistant to ampicillin, cephalothin, and amoxicillin/clavulanate acid. All were imipenem susceptible. Isoelectric focusing demonstrated chromosomal AmpC in all strains; no other β-lactamases were identified. A summary of promoter and attenuator mutations, as well as alterations in outer membrane profiles, is shown in Table 1. Strains QE1–QE9 were identical or closely related by PFGE. Each strain had a 1-bp insertion in the spacer region between the –35 and –10 boxes. This insertion created a consensus 17-bp spacer. In addition to this mutation, these strains had additional mutations at –73, +6, and +81. Strain QE7 also had a deletion in the loop of the attenuator. None of these strains had changes in their outer membrane protein profiles (Figure 1).

The other nine strains (QE10-QE18) had different PFGE patterns and came from diverse locations (different hospitals and from both outpatients and inpatients). Strain QE10 had a C to T mutation at –42 and a G to A at –18, which created a novel consensus –35. Two strains had the T to A mutation at –32 that is necessary to create the stronger –35 consensus sequence (QE14, 15). Strain QE14 also had the –11 C to T mutation that created the stronger –10 consensus sequence (TACAAT).

Only two strains had no promoter or attenuation loop mutations and no abnormalities of the outer membrane profile (QE16, QE17). Both of these strains had the C to T mutation at position +58 of the attenuator. This mutation would appear not to influence the development of the attenuation loop.

Of the 10 susceptible strains that were serially passaged on gradient plates, 8 developed resistance to cefoxitin (strains LD1–LD5, LD8–LD10). One of these strains had mutations in the AmpC promoter or attenuator regions (Table 2); this strain had two mutations, including a C to T mutation in the left stem of the attenuator, which would result in the transcription of a weak attenuation loop. Both of the mutations were also seen in the initial clinical isolates. This strain also had absent Omp C and Omp F. The remaining cefoxitin-resistant strains either lacked Omp C and Omp F (six strains) or had decreased amounts of OmpC and OmpF (one strain). One strain had no mutations in the promoter or attenuator and normal amounts of Omp C and Omp F.

The emergence of E. coli strains resistant to extended-spectrum cephalosporins and cephamycins should be a cause of concern to clinicians managing infections in both the community and institutional setting. Extended-spectrum cephalosporins and penicillins combined with β- lactamase inhibitors are frequently used for both empirical and definitive treatment of E. coli infections. Strains resistant to cephamycins have emerged in recent years. Some of these strains have become resistant by virtue of their hyperproduction of chromosomally encoded AmpC β-lactamases (5–7). Others have acquired plasmidic β-lactamases; most often those derived from Citrobacter freundii (8) (Figure 2).

In our survey of E. coli strains identified at our center, we found that no strains were resistant as a result of acquired-plasmid–mediated β-lactamases. These strains all produced substantial amounts of AmpC β-lactamase with the isoelectric point characteristic of the E. coli–derived AmpC. The predominant reason for the hyperproduction of AmpC appeared to be promoter mutations, attenuator mutations, or both.

Half of the strains appeared to be clonal in origin in that they had the same promoter mutation and the same attenuator mutations. Several of the strains had additional attenuator mutations; however, all of these strains had the same pulsed-field pattern. None of the strains had alterations in their OmpC or OmpF profiles. These strains came from diverse locations within the Queen Elizabeth II Health Sciences Centre complex and from other hospitals. Since we had limited information on the movement of the infected patients within the healthcare system, we are unable to define a common source.

The other nine strains had different pulsed-field patterns. Two strains each had similar promoter mutations, attenuator mutations, or both. Again, these strains came from different locations within our center, the community, and from referring hospitals. Two of these strains had altered outer membrane profiles; one lacked OmpF, the other OmpC. The promoter and attenuator mutations we detected were very remarkably similar to those that had been previously described in France, South Africa, Sweden, and Toronto, Canada. In particular the –42, –31, –18 mutations and spacer inserts have been consistently reported. The –11 promoter mutation that we observed was the second report of this mutation in the E. coli AmpC promoter (14).

We were able to develop resistance in 8 of 10 strains in vitro. Upon examination, seven of these strains had altered outer membrane protein profiles; six lacked both OmpC and OmpF, and one had decreased production of both of these porin proteins. One strain did not have promoter or attenuator mutations or alterations in the outer membrane protein profile. Sequencing of the promoters and attenuators indicated that only one had mutations that might have affected the amount of β-lactamase produced. However, this mutation was present in the wild strain as well.

We found that promoter abnormalities were common in clinical strains but not in lab-generated cefoxitin-resistant isolates. Increased resistance to antimicrobial agents has been shown to be associated with loss of porins in E. coli and other gram-negative organisms (15–19). The major porin proteins of E. coli, OmpF and OmpC, are differentially expressed at the transcriptional level by the two-component regulatory proteins, EnvZ and OmpR. The sensory protein, EnvZ, phosphorylates the transcriptional factor OmpR in response to environmental stress. The cellular level of the active form of OmpR, OmpR-phosphate (OmpR-P), is responsible for the differential expression of ompF and ompC (20). Therefore, the tension between the kinase and phosphatase activities of EnvZ controls the level of active OmpR-P in the cell. In the resistant strains selected on gradient plates, we found that most lacked both major outer membrane proteins in comparison to the respective parent strain. The lack of outer membrane protein expression in these isolates could occur by several mechanisms, including loss of kinase activity of EnvZ and mutation of the transcription factor OmpR.

Alternatively, resistance may arise because of decreased pore diameter (21,22). The in vitro mutant without quantitative changes in Omp C or OmpF or promoter or attenuator mutations may have had this molecular lesion.

We postulate that, while porin deficient mutants are more readily selected by antimicrobial pressure, they are likely less fit. This finding is reflected by the fact that Omp changes are easily created in the laboratory but not found in clinical samples. As a result of their lower fitness, they are more likely to be replaced by wild E. coli when antimicrobial pressure has been removed. On the other hand, the widespread clonal dissemination of strains that hyperproduce AmpC by virtue of promoter or attenuator mutations suggests that they are much better able and more likely to contribute to the spread of cephamycin resistance.

The ampC β-lactamase produced by E. coli hydrolyzes penicillins, cephalosporins, and cephamycins. In doing so, β-lactamase increases the MICs to third-generation cephalosporins but, as our data suggest, seldom above the breakpoint set by the National Committee for Clinical Laboratory Standards (NCCLS). This situation is analogous to that of many TEM- and SHB-derived β-lactamases. Using third-generation cephalosporins or penicillin/β-lactamase–inhibitor combinations to treat serious infections caused by ampC up-regulated strains may be more imprudent. Just as we would not be inclined to treat an E. coli bearing an extended spectrum β-lactamase with a ceftriaxone MIC of 2 mg/L with ceftriaxone, we would not treat an ampC β-lactamase up-regulated E. coli strain with a third-generation cephalosporin. To the best of our knowledge, no publication has documented treatment failures in such circumstances, and no NCCLS guidelines exist. Nevertheless, our practice is to report all of these strains as resistant to third-generation cephalosporins and to cautioning against their use.