ABSTRACT
A huge variety of extended-spectrum β-lactamases (ESBLs) have been detected during the last 20 years. The majority of these have been of the TEM or SHV lineage. We have assessed ESBLs occurring among a collection of 455 bloodstream isolates of Klebsiella pneumoniae, collected from 12 hospitals in seven countries. Multiple β-lactamases were produced by isolates with phenotypic evidence of ESBL production (mean of 2.7 β-lactamases per isolate; range, 1 to 5). SHV-type ESBLs were the most common ESBL, occurring in 67.1% (49 of 73) of isolates with phenotypic evidence of ESBL production. In contrast, TEM-type ESBLs (TEM-10 type, -12 type, -26 type, and -63 type) were found in just 16.4% (12 of 73) of isolates. The finding of TEM-10 type and TEM-12 type represents the first detection of a TEM-type ESBL in South America. PER (for Pseudomonas extended resistance)-type β-lactamases were detected in five of the nine isolates from Turkey and were found with SHV-2-type and SHV-5-type ESBLs in two of the isolates. CTX-M-type ESBLs (blaCTX-M-2 type and blaCTX-M-3 type) were found in 23.3% (17 of 73) of isolates and were found in all study countries except for the United States. We also detected CTX-M-type ESBLs in four countries where they have previously not been described—Australia, Belgium, Turkey, and South Africa. The widespread emergence and proliferation of CTX-M-type ESBLs is particularly noteworthy and may have important implications for clinical microbiology laboratories and for physicians treating patients with serious K. pneumoniae infections.
During the last 2 decades, extended-spectrum β-lactamases (ESBLs) found in gram-negative bacilli have emerged as a significant mechanism of resistance to oxyimino-cephalosporin antibiotics (8, 10, 21, 29, 37). The ESBLs mediate resistance to broad-spectrum cephalosporins (e.g., ceftazidime, ceftriaxone, and cefotaxime) and aztreonam. The genes encoding ESBLs are usually found on plasmids, along with genes encoding mechanisms of resistance to aminoglycosides and trimethoprim-sulfamethoxazole. Additionally, Klebsiella pneumoniae isolates harboring ESBLs are significantly more frequently found to be resistant to quinolones than non-ESBL-producing strains (22, 30). Finally, the combined effect of multiple β-lactamases and outer membrane protein (OMP) deficiencies may lead to resistance of ESBL-producing enteric bacteria to β-lactam-β-lactamase inhibitor combinations and, occasionally, even to cephamycins and carbapenems (24).
Most commonly, ESBLs derive from genes for the narrower-spectrum TEM-1, TEM-2 or SHV-1 β-lactamases by mutations that alter the amino acid configuration around the active site of these enzymes (27, 28). More than 100 genetically distinct TEM-type and SHV-type ESBLs have now been characterized (www.lahey.org/studies/webt.asp). Additionally, other types of ESBLs have been documented. These include the CTX-M-type ESBLs, β-lactamases that have less than 40% homology with TEM and SHV types, and PER (for Pseudomonas extended resistance)-type ESBLs (1, 2, 5, 33, 42). As a group, the CTX-M-type β-lactamases are closest in amino acid identity to the chromosomal cephalosporinases of Kluyvera georgiana, Kluyvera cryorescens, and Kluyvera ascorbata (14, 20, 34). PER-type β-lactamases represent a distinct class A cephalosporinase phenotype so far only restricted to South America and Europe (3, 42). Although possessing only 26% identity to the TEM-type ESBLs, PER β-lactamases also confer resistance to oxyimino-β-lactams, such as cefotaxime, ceftazidime, and aztreonam (6, 7, 40, 43, 44).
Curiously, ESBLs are most commonly detected in K. pneumoniae. This organism is a cause of significant community- and hospital-acquired infections. Bloodstream infections with K. pneumoniae may arise from the lungs (community- and ventilator-acquired pneumonia), the urinary tract, intra-abdominal pathologies, and central venous line-related infections. In this study, we have performed molecular characterization of ESBLs from bloodstream isolates of K. pneumoniae collected from 12 hospitals in seven countries. Our goal is to present a partial sequence analysis of the types of β-lactamases found in these isolates. The demographic features, clinical characteristics, and outcomes of the patients harboring these ESBLs have been reported (30, 31).
MATERIALS AND METHODS
Study design.A prospective, observational study of consecutive, sequentially encountered patients with K. pneumoniae bacteremia was performed in 12 hospitals in the United States, Taiwan, Australia, South Africa, Turkey, Belgium, and Argentina (30). The study period was 1 January 1996 to 31 December 1997. Patients older than 16 years of age with blood cultures positive for K. pneumoniae were enrolled, and a 188-item study form was completed. Patients were monitored for 1 month after the onset of bacteremia to assess clinical outcome, including mortality and infectious complications.
Microbiologic methods.The K. pneumoniae isolates potentially harboring ESBLs were those with a positive phenotypic confirmatory test for ESBLs according to current National Committee for Clinical Laboratory Standards (NCCLS) criteria (26). These isolates were initially screen positive in that the MICs of ceftazidime, cefotaxime, ceftriaxone, or aztreonam for these organisms were ≥1 μg/ml according to standard broth dilution techniques. A phenotypic confirmatory test was then performed by testing MICs for ceftazidime, ceftazidime-clavulanic acid, cefotaxime, and cefotaxime-clavulanic acid. A ≥threefold concentration decrease in a MIC of either ceftazidime or cefotaxime tested in combination with clavulanic acid versus its MIC when tested alone was indicative of phenotypic confirmation of ESBL production.
aIEF.We performed initial characterization of the β-lactamases in these clinical isolates by analytical isoelectric focusing (aIEF) as previously described (32). Ten microliters of the crude enzyme extract was loaded onto a precast gel (Ampholine PAGplate; Amersham Pharmacia Biotech, Piscataway, N.J.). We used gels with a pH range of 3.5 to 9.5 as part of the initial screen. Isolates with previously characterized β-lactamases were used as controls. These were obtained as a kind gift from P. Bradford (Wyeth Pharmaceuticals, Pearl River, N.Y.). In addition, RTEM-1 enzyme from Escherichia coli 205 (Sigma Chemical Co., St. Louis, Mo.) and SHV-1 from K. pneumoniae 15571 were loaded onto a gel as control β-lactamases (pI 5.4 and 7.6) (32).
PCR amplification and bla gene sequencing.A 10-μl aliquot of an overnight culture of the test isolate was diluted 1:10 with water and boiled for 10 min. PCR amplification was then performed with 10 μl of this dilution as the DNA template. The primer sets are described in Table 1. The PCR conditions used were 35 cycles of amplification at a denaturation temperature of 94°C for 30 s, an annealing temperature of 60°C for 1 min (70°C for the CTX-M-2 primers, 43°C for PER-1 primers, and 45°C for TEM primers), and an extension temperature of 72°C for 1 min. This step was followed by a final extension at 72°C for 10 min. PCR products were resolved on 1% agarose gels, stained with ethidium bromide, and photographed with UV illumination. ΦX174 replicative-form DNA HaeIII fragments (GIBCO BRL Life Technologies, Rockville, Md.) were used to assess PCR product size.
Primer sets used in characterizing β-lactamasesa
Direct sequencing of amplified products was performed on an ALF Express automated DNA sequencer (Amersham Pharmacia Biotech, Piscataway, N.J.) by using the Thermo Sequenase fluorescent-labeled primer cycle sequencing kit (Amersham Pharmacia Biotech). Under some circumstances, amplicons were cloned into the pCR 2.1-TOPO vector (Invitrogen, Carlsbad, Calif.) and sequenced with Cy5-labeled M13 reverse and M13 forward primers (17). We repeated amplification and sequencing twice for each isolate. Sequencing primers for blaSHV were used as previously described (17).
ELISA.We examined K. pneumoniae isolates for the presence or absence of SHV- or CMY-2-type β-lactamases by using a sensitive and specific enzyme linked immunoabsorbent assay (ELISA) as a screening test (18). This assay detects these two β-lactamases with a sensitivity and specificity of greater than 94%.
RESULTS
Four hundred fifty-five episodes of K. pneumoniae bacteremia occurred in 440 patients during the study period; the isolates were from Argentina (n = 41), South Africa (n = 116), Europe (n = 27), the United States (n = 58), Australia (n = 71), and Taiwan (n = 142). Eighteen percent (85 of 455) of the isolates had phenotypic evidence of ESBL production. Of these, 73 isolates were available for aIEF and gene sequencing. These isolates came from Argentina (n = 18), South Africa (n = 27), Turkey (n = 9), the United States (n = 11), Australia (n = 2), Belgium (n = 3), and Taiwan (n = 3).
aIEF.All 73 isolates possessed at least one β-lactamase (mean, 2.7; range, 1 to 5) (Table 2). The numbers of β-lactamases produced by each isolate were one (1.4%; 1 isolate), two (49.3%; 36 isolates), three (34.2%; 25 isolates), four (12.5%; 9 isolates), and five (2.7%; 2 isolates). In certain isolates that possessed CTX-M-type β-lactamases, it was difficult to assess the precise number of β-lactamases by using aIEF. For certain isolates, it was not possible from aIEF to enumerate all of the β-lactamases detected.
IEF data from 73 K. pneumoniae bloodstream isolates with phenotypic confirmation of ESBL production
Analysis of bla gene sequencing results.We amplified and sequenced each blaTEM, blaSHV, and blaCTX-M at least twice. Among the isolates studied, 67 of 73 (91.8%) that had a positive phenotypic confirmatory test for ESBL production were found to produce a TEM-, SHV-, PER-, or CTX-M-type ESBL by sequencing (exceptions are isolates 456, 470, 438, 442, 104, and 140; see Table 4). The ceftazidime MICs for four of these isolates were ≥256 μg/ml. One of the 73 isolates was positive by ELISA for SHV β-lactamase, but we were not able to amplify blaSHV by using specific primers. These isolates are undergoing further characterization of their plasmids, β-lactamases, and outer membrane protein profiles. From aIEF gels, it is clear that multiple TEM and SHV β-lactamases are present.
Organisms with multiple ESBL types
Summary of bla sequencing results of each isolate grouped by countrya
Although we did not sequence the entire blaTEM and blaSHV genes, we concentrated our analysis in the regions of these enzymes responsible for the ESBL phenotype (Table 1). In our analysis, 87.7% (64 of 73) isolates produced a TEM-type β-lactamase. These included the non-ESBL enzymes TEM-1A-type (8 isolates), TEM-1B-type (36 isolates), a novel TEM variant we designate TEM-1H-type (1 isolate), and TEM-2-type (7 isolates). The TEM-1H sequence differed from that of TEM-1B by a single nucleotide change at the codon encoding amino acid 171 (GAA to GAG). Only 16.4% (12 of 73) isolates produced a TEM-type ESBL (Table 4). Eight isolates possessed TEM-10-type (amino acid changes R164S and E240K; four isolates from the United States, two from South Africa, and two from Argentina), two isolates harbored TEM-12-type (amino acid change R164S; one isolate each from Argentina and South Africa), one isolate had TEM-26-type (amino acid changes E104K and R164S; from the United States), and one isolate contained TEM-63-type (amino acid changes E104K, R164S, and M182T; from South Africa). TEM-63 has been reported from Durban, South Africa (15).
Ninety percent (66 of 73) of the isolates produced an SHV-type β-lactamase. Notably, 49 isolates produced an SHV-type ESBL. We focused our attention upon the dominant amino acid substitutions that result in the ESBL phenotype for SHV-type β-lactamases and compared each sequence to the blaSHV-1 sequence deposited in GenBank (accession no. AF124984). Fourteen isolates had an amino acid mutation at Ambler position 238 but not at site 240 (amino acid change G238S; therefore, ESBL types SHV-2, -2A, -3, -20, or -21 according to www.lahey/org/studies/webt.asp). These isolates were from Argentina (n = 4), South Africa (n = 5), Turkey (n = 1), the United States (n = 2), Australia (n = 1), and Taiwan (n = 1). We designated these ESBLs as SHV-2-type. Thirty-five isolates had mutations at amino acid positions 238 and 240 (amino acid changes G238S and E240K; therefore ESBL types SHV-4, -5, -7, -9, -10, -12, or -15; for simplicity we designated the isolates as SHV-5-type). These isolates were from Argentina (n = 2), South Africa (n = 21), Turkey (n = 3), United States (n = 5), Australia (n = 1), Belgium (n = 2), and Taiwan (n = 1) (Table 4). In our analysis, we observed that there were also blaSHV genes with silent mutations. At amino acid position 240, Glu is normally encoded by GAG. We found GAA in SHV-2-type enzymes (designated SHV-2-type§ in Table 4). In SHV-5-type enzymes, we found both AAA (designated SHV-5-type ∧) and AAG encoding Lys at 240. In sum, 35 isolates possessed SHV-5-type, and 14 isolates produced SHV-2-type enzymes.
The ESBL found in 23.3% (17 of 73) of our isolates was a CTX-M-type ESBL. CTX-M-type ESBL-producing K. pneumoniae isolates were found in all study countries, except the United States. The CTX-M β-lactamases identified were CTX-M-2 type (14 isolates; 11 from Argentina, 1 from South Africa, 1 from Turkey, and 1 from Belgium) and CTX-M-3 type (1 isolate each from Taiwan, Australia, and South Africa).
PER-1-type β-lactamases were detected in five of the nine isolates from Turkey and were found with both SHV and TEM β-lactamases. All five isolates possessed SHV and TEM β-lactamases.
More than one ESBL was found in 19.2% (14 of 73) of isolates (Table 4). Ten of the isolates had TEM- and SHV-type ESBLs; two had TEM-, SHV-, and CTX-M-type ESBLs; and two had SHV- and PER-type ESBLs.
DISCUSSION
To our knowledge, this study is among the first to give a snapshot of the SHV and TEM sequence variability of ESBLs found in bloodstream isolates of K. pneumoniae in different continents at a single point in time (1996 to 1997). This is a necessary first step in our quest to understand how the genotypes of these complex antibiotic resistance phenotypes emerge in the clinical setting. A major finding was that SHV-type ESBLs were by far the most dominant ESBL type. Although, we did not fully sequence all 49 of the ESBL blaSHV genes, we focused our attention upon the amino acid sequence at the crucial 238 and 240 sites. This showed that 35 isolates with SHV-type ESBLs had mutations at both sites 238 and 240 and were therefore of the ESBL phenotype SHV-4, -5, -7, -9, -10, -12, or -15. Why this remains the preferred site of mutation to evolve an ESBL phenotype in SHV remains to be established. It has been demonstrated that the mutation G238S in the SHV β-lactamase (SHV-2) preserves efficient catalytic activity against both penicillins and cephalosporins (17). The mutation G238S coupled with the E240K mutation is also associated with increased steady-state β-lactamase expression relative to the G238S mutant β-lactamase. Other reports show that the molecular heterogeneity of blaSHV associated with ESBLs centers about the mutations at amino acid positions 238 and 240 (4, 15, 29, 36). Data do not yet exist to support the notion that the E240K mutation functions in any way to stabilize the effect of other mutations in SHV, especially G238S. This mutation may simply be enhancing the affinity of broad-spectrum cephalosporins to the active site (19). We also found that in 10 K. pneumoniae isolates, the amplified blaSHV gene differed from wild type SHV-1 by one nucleotide that did not result in an amino acid change (Table 4). This was found in 3 SHV-2-type β-lactamase amplicons. Like the situation that exists in TEM (TEM-1A to -G), there is molecular heterogeneity of blaSHV (23, 35).
It is perplexing to us that in seven of the isolates, our SHV PCR screening primers did not detect blaSHV (isolates 8, 9, 10, 11, 465, 466, and 467 [see summary in Table 4]). blaSHV is reported to be “universal” in K. pneumoniae (G. S. Babini and D. M. Livermore, Letter, Antimicrob. Agents Chemother. 44:2230, 2000). blaSHV is mainly a chromosomally encoded species-specific enzyme (12). Isolates 8, 9, 10, 11, 465, 466, and 467 were positive with the PCR screen for CTX-M-2 by CTX-M-2-specific primers (hence explaining the phenotype) but did not amplify with blaAmpC primers and did not demonstrate a signal that could be detected by the ELISA for AmpC or SHV β-lactamase. Since the aIEF was difficult to interpret due to the presence of a unique pattern of bands due to CTX-M enzymes (data not shown), we used PCR amplification and ELISA to detect these enzymes. We currently have no information that identifies a related β-lactamase in lieu of SHV, nor do we have data to propose that the presence of blaCTX-M excludes blaSHV. It has been previously observed that many isolates of K. pneumoniae possess LEN-1- or LEN-2-type β-lactamases migrating at pI 7.1 (16). We had a number of isolates with β-lactamases in this range (Table 2). In further studies, we plan to expand our analysis to seek LEN-1 and LEN-2.
Although more varieties of TEM-type ESBLs have been described than SHV-type ESBLs, TEM-type ESBLs were found less frequently in our study. TEM-10-type was the most frequently detected TEM-type ESBL and was found in disparate regions (the United States, South Africa, and Argentina). As far as we are aware, this finding of TEM-type β-lactamases in Argentina represents the first report of TEM-type ESBLs from South America. We also noted that TEM-1B-type was the most common TEM variant found, being sequenced in more than half of the isolates studied. TEM-1B-type occurred most often in association with SHV-2-type and SHV-5-type enzymes (23 of 73 isolates). A substantial proportion (19.2%) of organisms produced multiple ESBL types. Two isolates produced a TEM-type ESBL, an SHV-type ESBL, and a CTX-M-type ESBL.
Perhaps most significantly, this report extends the geographical spread of CTX-M-type ESBLs to Australia, Belgium, Turkey, and South Africa (9, 11, 13, 39, 41). There are no other published reports of the discovery of CTX-M-type β-lactamases in these nations. Additionally, in our study, CTX-M-type ESBLs were more numerous than TEM-type ESBLs. Since cefotaxime and ceftriaxone are used worldwide, it is not surprising that CTX-M-type ESBLs are now being found in multiple countries. Although no epidemiologic studies have yet been performed that have linked cefepime use with infection with a CTX-M-type ESBL, it is noteworthy that elevated cefepime MICs are frequent for K. pneumoniae isolates producing CTX-M-type ESBLs. Of increasing importance is the potential effect of the presence of a CTX-M-type ESBL on detection of ESBLs by the clinical microbiology laboratory. Those laboratories, which rely on resistance to ceftazidime as a surrogate marker for ESBL production, will likely not be aware of organisms producing CTX-M-type ESBLs.
As noted above, PER-type enzymes were detected in more than half of the K. pneumoniae isolates from Turkey. PER-type β-lactamases have recently been recovered from E. coli, Proteus mirabilis, Salmonella enterica serovar Typhimurium, K. pneumoniae, Acinetobacter baumannii, and Alcaligenes faecalis (33, 42-44). PER-2 shares 86% homology with PER-1 and has been found predominately in South America (3). The PER-type ESBLs are among the most efficient β-lactamases, able to hydrolyze broad-spectrum cephalosporins (11-fold dilution increase when transformed into E. coli C600) (43). The residues responsible for the ESBL phenotype in PER-1 are distinct from SHV and TEM, and the binding cavity of this β-lactamase is quite different (6, 40). Our finding of this ESBL in K. pneumoniae in Turkey raises significant clinical concern. These PER β-lactamases were detected in K. pneumoniae isolates possessing TEM-2 and either SHV-2 or SHV-5 type (in isolates with three β-lactamases).
In summary, we have documented the dominance of K. pneumoniae SHV ESBL types worldwide and highlighted the emergence of CTX-M-type ESBLs in numerous countries. It is important to consider that the number of reports of novel ESBLs of SHV- and TEM-type have diminished in recent years. The growing number of ESBLs of different varieties challenges us to ponder if K. pneumoniae is one of the main pathogens in which ESBLs evolve. A limitation to our study is that only a select number of hospitals in each country were assessed. There may be peculiarities in prescribing antibiotics or other forces that may have biased the ESBL types seen in the study hospitals. It has not escaped our attention that an isolate may have more than one blaTEM or blaSHV gene present, and amplification and sequencing efforts only detected a single genotype. If multiple blaTEM or blaSHV genes are present, the predominant one will preferentially amplify and produce sequence. Nevertheless, we have been able to gain a unique assessment of ESBL-types occurring in consecutive patients with K. pneumoniae bacteremia at the same point in time. As antibiotic usage changes over time, we speculate that types of β-lactamase produced by K. pneumoniae may progress as well. In turn, the antibiotic susceptibilities of this organism will evolve, and it behooves us to constantly reevaluate both laboratory detection of ESBLs and potential treatment options for organisms producing ESBLs (45). This study complements investigations directed at increasing the awareness of β-lactamases in K. pneumoniae in the United States and other hospitals worldwide (25). Studies are planned to examine the number and type of plasmids present in these isolates.
ACKNOWLEDGMENTS
A grant from AstraZeneca supported the work of Kristine M. Hujer. The Veterans Affairs Department Merit Review Program supported Louis B. Rice and Robert A. Bonomo. The Cottrell Foundation of the Royal Australasian College of Physicians supported some of the work of David L. Paterson.
We thank Marion S. Helfand for critical review of the manuscript.
The International Klebsiella Study Group comprises Victor L. Yu (Veteran Affairs Medical Center, Pittsburgh, Pa.), Herman Goossens (University Hospital, Antwerp, Belgium), Jose Maria Casellas (Sanatorio San Lucas, Buenos Aires, Argentina), Wen Chien Ko (National Cheng Kung University Medical College, Tainan, Taiwan), Keith Klugman (Emory University, Atlanta Ga.), Joseph G. McCormack (University of Queensland, Brisbane, Australia), Anne Von Gottberg (South African Institute of Medical Research, Johannesburg, South Africa), Gordon Trenholme (Rush Presbyterian St. Lukes Medical Center, Chicago, Ill.), and Lutfiye Mulazimoglu (Marmara University, Istanbul, Turkey).
FOOTNOTES
- Received 13 March 2003.
- Returned for modification 28 April 2003.
- Accepted 22 August 2003.
↵† Members of the International Klebsiella Study Group are listed in Acknowledgments.
- American Society for Microbiology
