INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Throughout the world, microbiologists are encouraged to survey the antibiotic resistances of major pathogens in order to provide the people in charge of public health with epidemiological data helpful in making recommendations on the best use of antibiotics.

We surveyed amoxicillin-clavulanate resistance in Escherichia coli over a 3-year period (1996 to 1998) because amoxicillin-clavulanate is one of the most used antibiotics in France and also because E. coli is the major enterobacterial pathogen. In addition to the evolution of amoxicillin-clavulanate-resistant E. coli isolates, which was studied both globally and according to different parameters (patient status, wards, and biological specimens), we also analyzed the evolution of the principal mechanisms involved in amoxicillin-clavulanate resistance in E. coli. These mechanisms include hyperproduction of the chromosomal class C -lactamase of E. coli, hyperproduction of plasmid-mediated TEM enzymes, production of oxacillinases, and the production of TEM-derived enzymes whose -lactamase activities are no longer inhibited by clavulanate (inhibitor-resistant TEM [IRT] enzymes) (11-13, 25, 26, 28, 33). However, all these mechanisms which are able to generate amoxicillin-clavulanate resistance in E. coli do not generate the same -lactam cross-resistance. Thus, a knowledge of the evolution of these different mechanisms might lead us to give different recommendations for the choice of antibiotics to treat an infection presumed to be caused by E. coli.

(Part of this study was presented at the 39th Interscience Conference on Antimicrobial Agents and Chemotherapy, San Francisco, Calif., 1999.)

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Clinical isolates. During the same 3 months of the years 1996, 1997, and 1998, the microbiological laboratories of 14 (13 in 1997) French hospitals distributed around the country (seven teaching and seven nonteaching hospitals) collected consecutive and nonrepetitive E. coli isolates (only one isolate with a given antibiogram per patient) and filled out a form for each isolate in which the patient status (out-or inpatient), the ward (pediatrics, geriatrics, medicine, surgery, or intensive-care unit) in which each inpatient was hospitalized, and the biological specimen from which each isolate was obtained were specified. The isolates considered resistant to amoxicillin-clavulanate on the basis of the agar disk diffusion method and in accordance with the recommendations of the French Antibiogram Committee (inhibition zone diameter, <14 mm) (1) were sent with their corresponding antibiograms to our laboratory, where we confirmed the resistance and analyzed the underlying mechanisms by molecular methods.

Susceptibility testing. Amoxicillin-clavulanate resistance was confirmed for each isolate by the E-test method (AB Biodisk, Dammartin sur Tigeaux, France) performed according to the manufacturer's recommendations. The amoxicillin-clavulanate-resistant isolates were classified in two groups on the basis of the standard antibiogram performed by each laboratory: group I, comprising the isolates displaying an additional resistance to cefpodoxime (inhibition zone diameter, <21 mm) and a reduced susceptibility to cefoxitin (inhibition zone diameter, <22 mm) (1), and group II, comprising the isolates fully susceptible to these two antibiotics. The isolates of group I were considered to be hyperproducing their chromosomal class C -lactamase and not to be producing plasmid-mediated AmpC-like enzymes on the basis of the conserved susceptibility to broad-spectrum cephalosporins or aztreonam (3, 6, 15), and no further molecular analysis was made in this study to characterize this mechanism of resistance.

Molecular detection of TEM, IRT, and OXA-1 enzymes. As previously described, two specific primers of bla_TEM genes were used to amplify the entire gene and its promoter region and two other primers specific for the gene encoding OXA-1 were used to amplify an internal 609-bp fragment of this gene. When the PCR was positive for the bla_TEM gene, pairs of primers designed to obtain three overlapping fragments were used to carry out the single-strand conformational polymorphism (SSCP)-PCR, as we have previously demonstrated (34). Briefly, each ³²P-labeled amplified fragment was submitted to electrophoresis on a native polyacrylamide gel after heat and formamide denaturation. The electrophoretic mobility of each fragment was compared to that of the corresponding fragment of the reference bla_TEM genes: bla_TEM-1A, bla_TEM-1B, and bla_TEM-2. The bla_TEM genes whose three fragments comigrated with the corresponding fragments of reference bla_TEM genes were assessed as TEM-encoding genes, whereas those whose fragment 1, 2, or 3 did not comigrate with the corresponding fragments of the reference genes were assessed as bla_TEM gene derivatives and subsequently sequenced.

Nucleotide sequence. PCR products for sequencing were prepared under the standard conditions described above. The PCR products were purified using the QIAquick PCR purification kit (QIAGEN, Courtaboeuf, France) following the manufacturer's recommendations. The nucleotide sequences of purified PCR fragments were determined with an automated cycle-sequencing system on a Perkin-Elmer R377 sequencer and the same primers used for PCR (34).

Statistical analysis. The differences between proportions were evaluated by a chi-square test at the 5% level of significance.

Nucleotide sequence accession numbers. The nucleotide sequence data reported here for bla_TEM-76, bla_TEM-77, bla_TEM-78, and bla_TEM-79 have been submitted to GenBank and have been assigned accession no. AF 190694, AF 190695, AF 190693, and AF 190692, respectively.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Distribution of E. coli clinical isolates. One thousand six hundred twenty and 1,606 E. coli isolates were collected in 1996 and 1998, respectively; only 1,483 were collected in 1997 because one of the 14 laboratories was not able to participate in the study that year. There were no statistical differences in the distributions of these isolates according to the different parameters except for the deep-pus specimens, whose frequency was significantly (P < 0.001) higher in 1998 (10.3%) than in 1996 (6.7%) and 1997 (6.8%). Each year, the teaching hospitals provided approximately one-half of the isolates and the nonteaching hospitals provided the other half. Each year, 10% of the isolates came from children and 90% came from adults, and 18 and 82% were from outpatients and inpatients, respectively. About 8% of the isolates were obtained from adult patients hospitalized in intensive-care units, 8% were from patients in geriatrics, 16% were from patients in surgery, and 38% were from patients in medical wards. The great majority of isolates (80% in 1996 and 1997 and 75.5% in 1998) were urinary tract isolates, while 9 and 3% were isolated from blood and lower respiratory tract secretions, respectively.

Epidemiology of E. coli isolates resistant to amoxicillin-clavulanate. Amoxicillin-clavulanate resistance, first determined by the agar disk diffusion method (inhibition zone diameter, <14 mm) for 79 isolates in 1996, 78 in 1997, and 68 in 1998, was subsequently confirmed by the E-test method (MIC >16 µg/ml) (1). Thus, 4.8% of E. coli isolates collected in 1996, 5.2% of those collected in 1997, and 4.2% of those collected in 1998 were resistant, but these frequency differences were not statistically significant.

Mechanisms of resistance to amoxicillin-clavulanate. Among the 225 amoxicillin-clavulanate-resistant E. coli isolates collected over the 3-year period, 98 isolates which displayed a coresistance to amoxicillin-clavulanate and cefpodoxime and a reduced susceptibility to cefoxitin were considered to be hyperproducing their chromosomal class C -lactamase. However, for 47 (48%) of them, the bla_TEM PCR was positive, and TEM enzyme production was demonstrated in 41 of the cases on the basis of SSCP-PCR experiments, as the three amplified bla_TEM fragments of the corresponding genes comigrated with those of reference bla_TEM genes. For six isolates which displayed SSCP-PCR profiles different from those of the reference bla_TEM genes, the nucleotide sequences of the corresponding bla_TEM genes showed the presence of either bla_TEM-like genes (n = 3) or bla_TEM genes encoding IRT enzymes (n = 3). The bla_TEM gene of strain 602042 differed from bla_TEM-1B by the silent mutation G162T, whereas that of strain 710020 resembled bla_TEM-1A except for the replacement of cytosines 436 and 913 by thymidines (Table 4). The third bla_TEM-like gene, found in strain 606160, was more closely related to bla_TEM-2, although the nucleotide mutation leading to the amino acid substitution Glu39Lys was absent and the promoter region was different, showing thymidine 32 replaced by cytosine and guanine 162 replaced by thymidine (Table 4). The amino acid sequences deduced from the nucleotide sequences of the bla_TEM genes of strains 610087, 710085, and 609046 showed that the two first strains produced IRT-2 (TEM-30) whereas the third produced a new TEM derivative in which there was only one amino acid substitution, namely, Asn276Asp (Table 4).

View larger version (79K): [in this window] [in a new window]   FIG. 1.   SSCP profiles of fragment 3 of blaTEM genes expressed by four clinical E. coli isolates and of reference genes blaTEM-1 and blaTEM-2. As the nucleotide sequences of blaTEM-1A and blaTEM-1B are identical for fragment 3, there is only one SSCP profile for the two genes, which is shown in lane 3. The SSCP profile of fragment 3 of the reference blaTEM-2 gene is shown in lane 4. Two clinical isolates (lanes 1 and 5) expressed a blaTEM gene whose fragment 3 comigrates with that of blaTEM-2. The migration profile of fragment 3 of the blaTEM gene expressed by a third clinical isolate (lane 2) differs slightly from that of blaTEM-2 (lane 4). In lane 6 is shown a migration profile consisting of four bands. As two of these bands comigrate with those of blaTEM-1 and the other two do not comigrate with those of any reference genes, this profile is thought to result from the coamplification of blaTEM-1 and a blaTEM derivative.

Evolution of the mechanisms involved in amoxicillin-clavulanate resistance. As indicated in Table 5, the frequency of E. coli isolates whose amoxicillin-clavulanate resistances were considered to be related to the hyperproduction of their chromosomal class C -lactamase decreased over the 3-year period (48% in 1996 and 39.7% in 1998), whereas the frequency of those whose amoxicillin-clavulanate resistances were presumed to be related to the production of IRT enzymes increased (30.4% in 1996 and 41.2% in 1998). However, this opposite evolution was not significant as was also the case for the decrease in TEM-producing isolates (15% in 1996 and 10.3% in 1998) and the increase in OXA-1-producing isolates (3.8% in 1996 and 7.3% in 1998). The association of two mechanisms responsible for amoxicillin-clavulanate resistance, such as OXA-1 and IRT enzymes or IRT enzymes and presumed hyperproduced class C -lactamase, remained rare (Table 5).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Amoxicillin-clavulanate, which was first produced commercially almost 20 years ago, continues to be a contributing antibiotic to overcome the problem of class A -lactamase-producing bacteria, especially fastidious bacteria, such as Haemophilus influenzae and Moraxella catarrhalis. Such a feature can certainly explain why amoxicillin-clavulanate is so intensively prescribed, and sometimes abused, to empirically treat otitis, sinusitis, and respiratory tract infections (19). In the case of the normally amoxicillin-susceptible enterobacteria, particularly E. coli, the emergence of isolates less susceptible to amoxicillin-clavulanate was noted as early as 1987 among the amoxicillin-resistant isolates (25). Since then, several studies which evaluated the antibiotic susceptibilities of enterobacteria in different European countries have shown that the frequency of E. coli with reduced susceptibility to amoxicillin-clavulanate has not exceeded 25 to 30% in hospitals (8, 17, 24, 27, 29, 31; M. H. Nicolas-Chanoine, J. Sirot, and H. Chardon, Abstr. 39th Intersci. Conf. Antimicrob. Agents Chemother., abstr. 2253, 1999) and 10% in the community (16) for the last 10 years. Nevertheless, some studies reported frequencies of nonsusceptible E. coli isolates of 40% (21, 27), whereas others reported frequencies as low as 5% (35). As recently pointed out by Simpson et al., differences in amoxicillin-clavulanate susceptibility among the E. coli isolates of different countries can result from real localized resistance problems but also from methodological differences in susceptibility testing and breakpoint criteria (32). Moreover, the group of nonsusceptible isolates includes intermediate as well as resistant isolates, and the classification of isolates into these two subcategories also depends on methodological criteria. Our study focused on E. coli strains for which the amoxicillin-clavulanate MIC was higher than 16 µg/ml, i.e., on E. coli considered to be resistant to amoxicillin-clavulanate irrespective of the methodology used (32).

To our knowledge, this is the first study which has observed the evolution of the amoxicillin-clavulanate susceptibility of E. coli over a 3-year period in a large number of hospitals by taking into account different criteria, such as patient characteristics, hospital status, medical specialties, and sample origins.

During the 3-year period of our study (1996 to 1998), the overall frequency of amoxicillin-clavulanate-resistant E. coli remained stable at about 5%. A similar observance of a stable frequency was also made by Stapleton, who conducted a survey from 1990 to 1994 but described much lower resistance rates (35). The percentage of 5% resistant isolates was found, in our study, for urinary tract isolates of E. coli, and this is quite a bit lower than that of Henquell et al., who reported about 25% resistant isolates obtained from a single hospital and 10% obtained from private laboratories in Clermont Ferrand, France, in 1993 (21). As their study included only one hospital, this difference could be attributed to the local spread of E. coli clones.

The highest frequencies of amoxicillin-clavulanate-resistant E. coli isolates were observed in samples from the lower respiratory tract. These frequencies, which increased from 9.4% in 1996 to 14.6% in 1997 and to 15.2% in 1998, were significantly higher than those in all other samples. Such an observation should be taken into account by clinicians in charge of patients with chronic lung diseases who are hospitalized for acute lower respiratory tract infections because amoxicillin-clavulanate is one of the primary drugs recommended for the treatment of such patients (2, 22). With regard to the different medical specialties, it is interesting to note that E. coli isolates obtained from patients in surgery wards, independent of the sample origin, were significantly more often resistant to amoxicillin-clavulanate than those from patients in medical wards in 1998. Those surgeons who are in charge of patients with abdominal problems should be aware of this result, because amoxicillin-clavulanate, which is indicated in France for the treatment of intra-abdominal infections, is one of the most prescribed antibiotics for its concomitant activity against enterobacteria and anaerobes.

In addition to the clinical epidemiology of amoxicillin-clavulanate-resistant E. coli, we also studied the corresponding mechanisms of resistance and their evolution. Although relevant for assessing the adaptation of bacteria to antibiotic pressure, studies of the molecular epidemiology of mechanisms of resistance remain rare. Henquell et al., who studied the mechanisms of resistance to amoxicillin-clavulanate of urinary tract E. coli isolates exclusively, found that TEM production (48%) was the first mechanism and IRT production (27.5%) the second mechanism for E. coli isolated from hospitalized patients and the converse (34% TEM and 45% IRT production) for E. coli isolates collected by private laboratories in Clermont Ferrand, France (21). In our study, we found quite a different distribution of mechanisms, showing presumed hyperproduction of the class C -lactamase as the predominant mechanism (48%) in 1996, followed by IRT production (30.4%), while both mechanisms were found to be equivalent in 1997 (38.4 and 37.2%) and 1998 (39.7 and 41.2%). The fact that IRT production tended to increase could appear paradoxical, because this mechanism is linked to a narrower spectrum of -lactam hydrolysis than that of class C -lactamase production. However, it might reflect the antibiotic selective pressure by penicillin molecules, which have been widely used over the last 10 years. Amoxicillin-clavulanate resistance due to the pure production of TEM enzymes was, irrespective of the study year, always at a lower rate than that found by Henquell et al. (21). However, in both studies the frequencies of OXA-1-producing E. coli remained low. We confirm in this study the previously described relevance of the SSCP-PCR method for screening and differentiating genes encoding TEM- and TEM-derived -lactamases. By using this method, we were able to detect new bla_TEM-like genes and new IRT-encoding bla_TEM genes. Regarding the new bla_TEM-like genes, we found bla_TEM-1B genes with either the strong promoters Pa and Pb (C32T), described by Chen (14) and initially present in bla_TEM-2, or the strong promoter (G162T) recently designated P4 by Goussard et al. (18). Given that bla_TEM-1B-derived genes, such as bla_TEM-30 (10), bla_TEM-33b (18), and bla_TEM-38 (34), possess promoters Pa and Pb, we consider that we have discovered the ancestral bla_TEM-1B gene of these derivatives. A third bla_TEM1B-like gene also displays modifications in the promoter, namely, the mutation C141G located at position 3 in the 35 consensus of the Pribnow box, which is a new genetic feature in the promoters of bla_TEM genes. A fourth bla_TEM gene is similar to bla_TEM-1C (18) except for the mutation C913T, whereas a fifth is similar to bla_TEM-1B except for the mutation A346G, previously described in the bla_TEM-2 gene. We suggest designating these two genes bla_TEM-1D and bla_TEM-1E, respectively. The last bla_TEM-like gene we described is completely new. In fact, it possesses all the silent mutations observed in the bla_TEM-2 gene in comparison with the bla_TEM-1A gene but not the amino acid substitution Gln39Lys, and it displays the strong promoter P4 instead of Pa and Pb as in the bla_TEM-2 gene. As these silent mutations and this promoter are shared by several bla_TEM genes encoding IRT enzymes (TEM-33c, -34, -35, -36, -37, -39, and -45) (9, 10, 18, 34), we think we have identified the ancestor of these derivatives, and we therefore suggest calling this gene bla_TEM-1F.

Our study has also contributed to the discovery of eight new IRT enzymes. One has a mutation only at position 276 which has been previously described only in an in vitro mutant (30). A second resembles TEM-59, as it possesses the amino acid substitution Ser130Gly, but derived from TEM-1 and not from TEM-2 (5, 23). A third differs from TEM-30, -31, -41, and -51 by a glycine at position 244 (7).

Three new enzymes show an association of two amino acid substitutions. The first combines two substitutions already described individually in TEM-33 and TEM-30 (4, 20). The second includes a known substitution and a new substitution in terms of its position. The third resembles TEM-45 (9).

The two remaining new IRT enzymes display three amino acid substitutions. One of them corresponds to a new combination of 3-amino-acid substitutions, whereas the other resembles TEM-39 (20).

In conclusion, this study, which permitted a large-scale evaluation of the frequencies of amoxicillin-clavulanate-resistant E. coli isolates in hospitals, also permitted the discovery of new bla_TEM genes as well as new bla_TEM derivatives.