ABSTRACT
Plasmid-mediated mechanisms, comprising TEM hyperproduction, TEM derivative production, and OXA production, lead to amoxicillin-clavulanic acid resistance in enterobacteria. The ability of the single-strand conformation polymorphism (SSCP)-PCR method to differentiate the genes encoding inhibitor-resistant β-lactamases was evaluated with three bla TEM primer pairs. Thebla TEM genes, which were known to be different on the basis of their nucleotide sequences (bla TEM-1A, bla TEM-1B,bla TEM-2, bla TEM-30,bla TEM-32, andbla TEM-35), were identified as different by their electrophoretic mobilities. Thebla TEM-33, bla TEM-34,bla TEM-36, bla TEM-37,bla TEM-38, andbla TEM-39 genes, whose sequence differences have been established by oligotyping, displayed different SSCP profiles for different fragments, suggesting genetic differences in addition to those defined by oligotyping. Confirmed by sequencing, these additional genetic events concerned silent mutations at certain positions and, notably, a G→T transversion at position 1 of the −10 consensus sequence in bla TEM-34,bla TEM-36, bla TEM-37, and bla TEM-39. Applied to eight clinical isolates of Escherichia coli resistant to amoxicillin-clavulanic acid, the SSCP method detected TEM-1 in three strains and TEM-30, TEM-32, and TEM-35 in three other strains, respectively. A novel TEM derivative (TEM-58) was detected in another strain, and the deduced amino acid sequence showed two substitutions: Arg244Ser, which is known to confer amoxicillin-clavulanic acid resistance in TEM-30, and Val261Ile, which has not been described previously. The eighth strain produced an OXA β-lactamase. Given the discriminatory power and the applicability of SSCP-PCR, this method can be proposed as a means of following the evolution of the frequencies of the different inhibitor-resistant β-lactamases.
Resistance to amoxicillin-clavulanic acid appeared first in Escherichia coli isolates, then in other species of enterobacteria, and most recently in Haemophilus influenzae (7, 12, 13, 21, 22, 32). Four enzymatic mechanisms for this resistance have been described in E. coli: hyperproduction of class C chromosomal β-lactamase (cephalosporinase), hyperproduction of plasmid-mediated TEM-1 or TEM-2, production of inhibitor-resistant TEM (IRT), and production of a relatively inhibitor-resistant OXA-type β-lactamase (5, 8, 9,29, 36). Epidemiological studies carried out in Europe have shown that the frequency of each mechanism varies in different countries (17, 33). The frequency of cephalosporinase hyperproduction was similar in France and England, but IRT production seemed to be much more frequent in France, whereas OXA production seemed to be more frequent in England. Such differences may be related to the different use of β-lactam antibiotics in the two countries, and the evolution of the incidence of the different mechanisms may influence the future use of different β-lactams in the hospital as well as in the community. Thus, it is important to define a strategy which allows continued observation of the frequency of the different mechanisms.
However, because only cephalosporinase hyperproduction can be indisputably detected by the classical antibiogram (resistance to both cephalothin and cefoxitin, in addition to amoxicillin-clavulanic acid resistance), other methods must be used to differentiate the mechanisms involving plasmid-encoded amoxicillin-clavulanic acid resistance. Such methods include the determination of the β-lactamase isoelectric point, determination of β-lactamase kinetic parameters, and/or oligotyping, but these methods are fastidious procedures and are too time-consuming for a national epidemiological survey of amoxicillin-clavulanic acid resistance. In the present study, the single-strand conformation polymorphism (SSCP)-PCR technique was evaluated to differentiate the plasmid-encoded enzymatic mechanisms of amoxicillin-clavulanic acid resistance (25). This technique, which is able to detect any genetic modification, i.e., a point mutation, deletion, or insertion, should be able to differentiate wild-type bla TEM genes from IRT-encoding genes derived from bla TEM genes by point mutations. Moreover, the use of specific primers should allow differentiation between bla TEM and bla OXAgenes.
MATERIALS AND METHODS
Reference genes.The genes bla TEM-1A,bla TEM-1B, and bla TEM-2, which were expressed by E. coli C600, were used as wild-type reference genes (15, 34). Previously described IRT-encoding genes were also included as reference genes in the study. These genes have been either sequenced (bla TEM-30 fromE. coli E-GUER, bla TEM-32 fromE. coli 1408, and bla TEM-35 fromE. coli CF0042) or defined by oligotyping from clinicalE. coli isolates (bla TEM-33,bla TEM-34, and frombla TEM-36 to bla TEM-39) (4, 6, 16, 31). The nucleotide differences which have been defined by the gene sequence and those which have been determined only by oligotyping are indicated in Table 1.
Nucleotide substitutions inbla TEM-1B, bla TEM-2, and IRT-encoding genes in comparison with the nucleotides ofbla TEM-1A
Primers.Three pairs of primers were designed from the nucleotide sequence of the bla TEM-1A gene and were used to amplify three overlapping fragments (Table2). The first pair of primers allowed the amplification of a fragment including the whole-gene promoter. The positioning of each fragment in relation to thebla TEM gene is included in Table 1. One pair of primers was selected to amplify an internal 609-bp fragment ofbla OXA-1 (Table 2).
Nucleotide sequences of the oligonucleotides used forbla TEM andbla OXA-1 amplification
SSCP-PCR.Samples were prepared by suspending a freshly grown colony in 0.5 ml of lysis buffer (20 mM Tris HCl [pH 8.3], 50 mM KCl, 0.1% Tween 20) which was heated at 94°C for 10 min. Five microliters of this preparation was submitted to PCR in a final volume of 25 μl. To the PCR master mixture containing 20 mM Tris HCl (pH 8.0), 100 mM KCl, 3 mM MgCl2, 400 μM (each) deoxynucleotide triphosphate, 2.5 U of Taq DNA polymerase (Boehringer Mannheim GmbH, Mannheim, Germany), and 25 pmol of each primer, 0.25 μl (2.5 μCi) of radioactive [α-32P]dCTP was added. The amplification reaction consisted of 36 cycles of 30 s of denaturation at 94°C, 30 s of hybridization at 42°C, and 60 s of extension at 72°C, with a final extension step at 72°C for 10 min. The radioactive PCR product was diluted 1:2 with SSCP dilution buffer (2 mM EDTA, 0.1% sodium dodecyl sulfate), and 5 μl of the diluted product was mixed with 5 μl of loading buffer (95% formamide, 0.05% bromophenol blue, 0.05% xylene cyanole, 50 mM EDTA). Immediately prior to loading of the SSCP gel the samples were denatured for 15 min at 94°C, cooled on ice, and loaded onto a nondenaturating polyacrylamide gel. The nondenaturating polyacrylamide gel was prepared by mixing 20 ml of acrylamide-bisacrylamide (29:1) with 80 ml of 1× TBE (Tris-borate-EDTA). The gel was run for 4 h at 65 W with constant cooling. After termination of the run, the gel was transferred to a filter paper, dried, and exposed for 2 h to an X-ray film at −70°C with an intensifying screen.
Sequencing.The PCR products for sequencing were prepared as indicated above but the radioactive nucleotide was omitted. The PCR products were purified with the QIAquick PCR Purification Kit (QIAGEN, Courtaboeuf, France) following the manufacturer’s recommendations. The nucleotide sequences of the purified PCR fragments were determined with the Sequenase PCR Product Sequencing Kit (Amersham, Les Ulis, France) and by following the manufacturer’s indications exactly. The sequencing reactions were run on a standard denaturating sequencing gel.
Clinical isolates.Eight clinical isolates of E. coli (isolates AP1 to AP8), obtained from Ambroise-ParéHospital between 1993 and 1994, were studied because they were resistant to amoxicillin and amoxicillin-clavulanic acid and susceptible to cefoxitin and broad-spectrum cephalosporins by the disk diffusion test according to the recommendations of the Antibiogram Committee of the French Microbiology Society (1).
Characterization of the amoxicillin-clavulanic acid resistance in the eight clinical isolates.The MICs of amoxicillin (SmithKline Beecham, Nanterre, France), alone and associated with a fixed concentration of 2 μg of clavulanic acid (SmithKline Beecham) per ml, and piperacillin (Léderlé, St. Cloud, France), alone and associated with a fixed concentration of 4 μg of tazobactam (Léderlé) per ml, for the eight clinical isolates were measured by a dilution method on Mueller-Hinton agar with a Steers replicator device and an inoculum of 104 CFU per spot.
For each clinical E. coli isolate, crude extracts were submitted to isoelectric focusing as described previously (3), and their pIs were compared with the pI values of the following enzymes: RP4/TEM-1, pI 5.4; R111/TEM-2, pI 5.6; pUD101/TEM-30, pI 5.2 (4); and RGN238/OXA-1, pI 7.4 (26). The kinetic parameters for the enzymes, the β-lactamase specific activity (in milliunits per milligram of total protein), and Km values (in micromolar) were determined with crude extracts by computerized microacidimetry as described previously (20). All extracts were first studied at pH 7 and 37°C in the presence of NaCl. When an OXA-type β-lactamase was suspected, a complementary set of experiments was performed at pH 7 and 20°C and in the presence of Na2SO4 instead of the NaCl solution, since OXA enzymes are inhibited by chloride ions. After preincubation of the crude extracts for 10 min at 37°C with 100 μg of clavulanic acid per ml, the residual activity of the β-lactamases was determined in order to differentiate TEM enzymes which display a residual activity of ≤10% from IRT enzymes which display a residual activity of >20% (11). Under these experimental conditions, it has been previously observed that OXA enzymes are highly unstable (11).
RESULTS
SSCP-PCR of the wild-type reference genes.In accordance with the nucleotide sequences (Table 1), we obtained three different SSCP profiles for fragment 1 covering 389 bp starting at position −7 (34) for the three wild-type reference genesbla TEM-1A, bla TEM-1B, andbla TEM-2 (Fig.1A). This was also the case for fragment 2, which covers 426 bp starting at position 301 (Fig. 1B). For fragment 3 we obtained, as expected (Table 1), two different SSCP migration profiles, one for both bla TEM-1 genes and one for bla TEM-2 (Fig. 1C).
SSCP-PCR of wild-type bla TEMgenes. The SSCP profiles of amplified fragments 1 (A), 2 (B), and 3 (C) are shown. Lanes 1, bla TEM-1A; lanes 2,bla TEM-1B; lanes 3,bla TEM-2.
SSCP-PCR of sequenced IRT-encoding genes.According to the nucleotide sequences of bla TEM-30,bla TEM-32, andbla TEM-35, we observed two different and specific SSCP profiles of fragments 1 ofbla TEM-30 and bla TEM-35, whereas the profile for fragment 1 of bla TEM-32was identical to that of fragment 1 ofbla TEM-1A. For fragments 2 (Fig.2) and 3, each of the IRT-encoding genes displayed a specific profile that was in accordance with the previously determined nucleotide sequence.
Comparative SSCP-PCR profiles of the amplified fragment 2 of wild-type bla TEM genes and previously sequenced IRT-encoding genes. Lane 1,bla TEM-35; lane 2,bla TEM-30; lane 3,bla TEM-1A; lane 4,bla TEM-32; lane 5,bla TEM-2; lane 6,bla TEM-1B.
SSCP-PCR of oligotyped IRT-encoding genes.No data were available for the nucleotide sequences corresponding to fragment 1 of the oligotyped IRT encoding genes. By the SSCP method, we found that fragment 1 of bla TEM-33 was clearly different from the fragments 1 of all reference genes included in this study. For bla TEM-34,bla TEM-36, bla TEM-37, andbla TEM-39, the SSCP migration profile of fragment 1 differed slightly from that of fragment 1 ofbla TEM-30, whereas the profile of fragment 1 ofbla TEM-38 differed slightly from that of fragment 1 of bla TEM-1B.
According to oligotyping analysis, at least two SSCP profiles different from those of the three bla TEM genes and the three sequenced IRT-encoding genes could be expected for fragment 2: one for bla TEM-34,bla TEM-36, and bla TEM-38and one for bla TEM-39. On the other hand, the SSCP profile of fragment 2 of bla TEM-33 could be expected to be identical to that of fragment 2 ofbla TEM-35, and that of fragment 2 ofbla TEM-37 could be expected to be identical to that of fragment 2 of bla TEM-32. In fact, we obtained specific fragment 2 profiles for five of the six oligotyped IRT-encoding genes because it was shown thatbla TEM-33 differed frombla TEM-35, bla TEM-37differed from bla TEM-32, andbla TEM-38 differed from bothbla TEM-34 andbla TEM-36. The fifth specific profile corresponded to bla TEM-39, as expected (Fig.3A).
SSCP-PCR profiles of fragments 2 and 3 of oligotyped IRT encoding genes. (A) Fragment 2. Lane 1,bla TEM-32; lane 2,bla TEM-35; lane 3,bla TEM-33; lane 4,bla TEM-34; lane 5,bla TEM-36; lane 6,bla TEM-37; lane 7,bla TEM-38; lane 8,bla TEM-39. (B) Fragment 3. Lane 1,bla TEM-33; lane 2,bla TEM-34; lane 3,bla TEM-36; lane 4,bla TEM-38; lane 5,bla TEM-39; lane 6,bla TEM-37.
For fragment 3, the oligotype and sequence comparison could suggest a migration profile identity between fragments 3 ofbla TEM-1, bla TEM-33, andbla TEM-34 on the one hand and between fragments 3 of bla TEM-35,bla TEM-36, and bla TEM-37on the other hand. Inversely, a migration specificity could be expected for bla TEM-38 andbla TEM-39. These two IRT-encoding genes effectively displayed specific fragment 3 SSCP profiles, but such was also the case for bla TEM-34, which differed frombla TEM-33, which displayed the same profile asbla TEM-1 (Fig. 3B), and forbla TEM-36 and bla TEM-37, which differed from each other (Fig. 3B) and which both differed frombla TEM-35.
Characterization of the amoxicillin-clavulanic acid resistance in the eight clinical isolates. (i) Kinetic parameters of the β-lactamases.The amoxicillin-clavulanic acid resistance was evaluated first by the disk diffusion test for the eight clinical isolates and was confirmed, as indicated in Table3, by the MICs (MICs, >16 μg/ml). Compared to the MICs of amoxicillin, those of piperacillin were lower, and in association with 4 μg of tazobactam per ml, piperacillin was active against six of the eight clinical isolates (MIC, <8 μg/ml) (Table 3). The β-lactamases of strains AP1, AP2, AP3, and AP5 comigrated with TEM-1 (pI 5.4) and that of strain AP4 comigrated with OXA-1 (pI 7.4). The three remaining strains, AP6 to AP8, produced a β-lactamase with a pI of 5.2 (Table 3). According to theKm values for penicillin G, the clinical isolates could be separated into three groups:Km of <10 for strains AP4 and AP5,Km of ∼25 for strains AP1 to AP3, andKm of >100 for strains AP6 to AP8. Among the β-lactamases which displayed a pI of 5.4, three (those from strains AP1 to AP3) showed a very low residual activity (<10%) in the presence of clavulanic acid, suggesting the production of TEM enzymes, whereas one (that from strain AP5) kept a relatively high residual activity (56%), similar to those which were measured for the three β-lactamases having pIs of 5.2 (Table 3). These findings suggested that the clinical isolates AP6 to AP8 and AP5 produced IRT β-lactamases. The necessity of using Na2SO4instead of NaCl solution to measure the β-lactamase activity of the enzyme produced by strain AP4 and the fact that this enzyme was highly unstable under the experimental conditions established for measurement of the residual activity suggested the presence of an OXA β-lactamase in this strain.
MICs and enzymatic kinetic parameters for the clinicalE. coli isolates
(ii) SSCP-PCR.Strain AP4, which was expected to produce an OXA-1 β-lactamase, did not yield an amplification product with the TEM-derived primers in repetitive experiments but yielded products with the specific primers of bla OXA-1 (data not shown). As indicated in Table 4, strains AP1 to AP3, whose β-lactamases displayed a very low residual activity in the presence of clavulanic acid, were shown to produce TEM enzymes by the SSCP method. The bla gene of strain AP1 could be identified as bla TEM-1B, and that of strain AP3 could be identified as bla TEM-1A. An unusual situation was observed with the bla gene of strain AP2 because the SSCP profiles of its fragments 1 and 2 corresponded to those of bla TEM-1A andbla TEM-1B, respectively. The four clinical strains which showed an important residual β-lactamase activity in the presence of clavulanic acid could be expected to produce IRT enzymes according to the SSCP results, namely, TEM-32 for strain AP5, TEM-35 for strain AP6, and TEM-30 for strain AP8. The amplification product obtained for fragment 3 of thebla TEM gene expressed by strain AP7 showed a particular migration profile that did not correspond to the profiles of any of the reference genes, while those of fragments 1 and 2 showed the same electrophoretic mobility as fragment 1 ofbla TEM-33 and fragment 2 ofbla TEM-1B, respectively.
SSCP-PCR profiles of the clinical E. coliisolates in comparison with the SSCP-PCR profiles of the referencebla TEM genes
Nucleotide sequencing.Given the great diversity of SSCP profiles of oligotyped IRT-encoding genes, sequencing of amplified fragments 1 and 2 and some of fragment 3 was carried out and showed that different silent point mutations occurred. As indicated in Table5 and in comparison with thebla TEM-1A gene sequence, these mutations consisted of a C→T transition at position 32 forbla TEM-38, a G→T transversion at position 162 for bla TEM-34,bla TEM-36, bla TEM-37, andbla TEM-39, an A→G transition at position 175 for bla TEM-33 andbla TEM-38, a C→T transition at position 226 for bla TEM-38, an A→G transition at position 346 for bla TEM-34,bla TEM-36, bla TEM-37, andbla TEM-39, a C→T transition at position 436 for all the bla TEM-derived genes, a G→T transversion at position 604 forbla TEM-33, a T→C transition at position 682 for bla TEM-34, bla TEM-36, and bla TEM-37, and a G→A transition at position 925 for bla TEM-36. By sequencing thebla TEM gene of clinical isolate AP7, for which the SSCP analysis was not able to define the type of IRT produced, three nucleotide changes were observed (Table 5). The first one consisted of an A→G transition at position 175, as was found inbla TEM-33 and bla TEM-38. The second one concerned a C→A transversion at position 929, leading to the amino acid substitution arginine→serine at position 244, while the third one was a G→A transition at position 980, leading to the amino acid replacement valine→isoleucine at position 261 (numbering of Ambler et al. [2]).
Nucleotide substitutions determined by sequencing of the previously oligotyped IRT-encoding genes and thebla TEM gene of strain AP7 compared with the sequence of bla TEM-1A
DISCUSSION
The amoxicillin-clavulanic acid resistance which appeared for the first time about 10 years ago in E. coli has now been observed in Klebsiella pneumoniae, Proteus mirabilis, Salmonella typhimurium, Shigella flexneri, and H. influenzae (7, 12, 13, 21, 22,32). Contrary to E. coli, the recently involved species do not possess a chromosomal cephalosporinase whose hyperproduction results in amoxicillin-clavulanic acid resistance. Subsequently, strains of these species have acquired plasmid-encoded mechanisms, comprising TEM hyperproduction and IRT production, which were first identified in E. coli (7, 13, 21, 32). OXA production, which is the third plasmid-encoded mechanism, seems to be limited to E. coli. The reasons why bacteria, particularly E. coli, have developed so many mechanisms are unknown. However, we can observe differences in the incidences of the mechanisms in individual countries and differences in the spectrum of β-lactam resistance according to the amoxicillin-clavulanic acid resistance mechanism (17, 28, 30, 33).
Because plasmid-encoded mechanisms involved either completely different genes (bla TEM, bla OXA) orbla TEM-derived genes (IRT-encoding genes), the PCR-based SSCP method seemed suitable to us for the rapid identification of the genes responsible for amoxicillin-clavulanic acid-resistant phenotypes. In fact, using this method we were able to differentiate bla TEM genes which were known to be different on the basis of their nucleotide sequences. SSCP-PCR was able to indicate the occurrence of further genetic events in the genes for which only short stretches were oligotyped, and these were confirmed by sequencing. Thus, a G→T transversion at position 1 of the −10 consensus sequence was identified inbla TEM-34,bla TEM-36, bla TEM-37, and bla TEM-39. Such a transversion, which is known to be at the origin of a higher level of enzyme production, was recently described in the bla TEM-1 gene expressed by an S. flexneri isolate resistant to amoxicillin-clavulanic acid and in IRT-encoding genes, notably,bla TEM-45 (10, 32). Our study shows, as has already been described by Caniça et al. (10), that there is a great molecular diversity in thebla TEM genes encoding IRT but that some mutations are very frequent in such genes, notably, A346G, C436T, and T682C transitions. Although we have observed the G925A transition only in bla TEM-36, this mutation also seems to be frequent in IRT-encoding genes according to the study of Caniça et al. (10). Inversely, the C32T and G604T transitions which we identified in bla TEM-38 andbla TEM-33, respectively, seem to be rare in IRT-encoding genes (10).
The eight amoxicillin-clavulanic acid-resistant clinical isolates were studied by both enzymatic methods and SSCP-PCR. The results obtained by each method were in perfect concordance. Enzymes determined by the SSCP-PCR technique to be TEM enzymes showed by enzymatic analysis the typical Km values and residual activity of TEM enzymes (9, 11). Those enzymes determined by the SSCP-PCR method to be IRT enzymes displayed kinetic parameters which have previously been published for such enzymes (10). Nevertheless, the SSCP-PCR method, which is more rapid and whose performance is less fastidious, allowed us to characterize precisely the type of the IRT enzymes. The presence of TEM-32 in strain AP5, suggested by the Km value, which is known to be particularly low for this enzyme, was definitely proven by SSCP-PCR (11). Moreover, SSCP-PCR allowed us to detect a novelbla TEM gene (bla TEM-58) in clinical isolate AP7. The deduced amino acid sequence showed a previously unknown amino acid substitution, Val261Ile, in addition to the already known amino acid substitution Arg244Ser, which confers amoxicillin-clavulanic acid resistance in TEM-30 (8). By the two crystal structures of TEM-1 β-lactamase which are available (entries 1BTL and 1XPB; Protein Data Bank, Brookhaven National Laboratory, Brookhaven, N.Y. [14, 19]), Val261 was shown to be located on sheet s-5 in a group of four hydrophobic residues: Ile, Val, Val, and Ile, starting at position 260. Moreover, the side chain of Val261 is in the close vicinity of at least Leu221 (helix h10), Leu250, and Leu286 (helix h11), which form a highly hydrophobic pocket. The role of the substitution Val261Ile on the enzymatic activity of this TEM-derived enzyme should be determined in further studies.
We can also note that the bla TEM gene of strain AP7 showed the same migration profile as fragment 1 ofbla TEM-33, for which we demonstrated a guanine at position 175. Such a mutation is extremely rare, because it is apparently present only in bla TEM-1B,bla TEM-5, bla TEM-6b, andbla TEM-38; however, it is present in association with a thymine at position 216 in these four genes (15, 18,27).
Three clinical isolates (isolates AP1 to AP3) were shown by the SSCP-PCR method and kinetic parameters to produce TEM enzymes. Three molecular mechanisms have previously been described to explain the amoxicillin-clavulanic acid-resistant phenotype related to wild-typebla TEM genes: up-mutation in the regulatory region of the gene, the presence of multiple copies of the plasmid, and a gene dose effect (23, 29, 32, 35). Because we did not observe for the TEM enzymes produced by strains AP1 to AP3 specific SSCP-PCR profiles of fragment 1 covering the promoter region, an up-mutation cannot be assumed to be responsible for the amoxicillin-clavulanic acid resistance. However, AP2 has been shown by determination of kinetic parameters and SSCP-PCR to yield a higher level of β-lactamase activity, and its β-lactamase gene has been shown to have an unusual structure.
In conclusion, this study shows that the SSCP-PCR method is a suitable tool for rapidly screening for the different plasmid-encoded mechanisms which are known so far to confer amoxicillin-clavulanic acid resistance in enterobacteria. We have shown that this method can also be used to identify the different IRT enzymes and detect novel enzymes of this type. However, because it is a comparative technique and because IRT-encoding genes have a great molecular diversity, a large number of reference genes must be used. To decrease the number of reference genes the amplified fragments could be shortened and concentrated on mutations leading to relevant amino acid substitutions, as was shown for SHV derivatives (24).
ACKNOWLEDGMENTS
This work received financial support from the Beecham Institute, Paris, France.
We thank Danielle Sirot for providing us with strains producing the oligotyped IRT enzymes and Patrice Courvalin for providing us withbla TEM-1b and bla TEM-2.
FOOTNOTES
- Received 4 August 1997.
- Returned for modification 16 December 1997.
- Accepted 31 January 1998.
- Copyright © 1998 American Society for Microbiology
