Arnfinn Sundsfjord, Department of Microbiology and Virology, Faculty of Medicine, University of Tromsø, N-9037 Tromsø, Norway. e-mail: firstname.lastname@example.org
Accurate and rapid diagnostic methods are needed to guide antimicrobial therapy and infection control interventions. Advances in real-time PCR have provided a user-friendly, rapid and reproducible testing platform catalysing an increased use of genetic assays as part of a wider strategy to minimize the development and spread of antimicrobial-resistant bacteria. In this review we outline the principal features of genetic assays in the detection of antimicrobial resistance, their advantages and limitations, and discuss specific applications in the detection of methicillin-resistant Staphylococcus aureus, glycopeptide-resistant enterococci, aminoglycoside resistance in staphylococci and enterococci, broad-spectrum resistance to β-lactam antibiotics in gram-negative bacteria, as well as genetic elements involved in the assembly and spread of antimicrobial resistance.
Antimicrobial susceptibility testing of bacterial pathogens is one of the primary functions of a diagnostic microbiology laboratory. Individual results have important therapeutic implications for the patient. Empiric treatment schemes for infectious diseases are based on accumulated susceptibility testing data gathered at the local, regional, or national level. Finally, careful detection of resistant bacteria provides a fundamental basis for infection control measures and antimicrobial surveillance systems.
The global emergence and spread of antimicrobial resistance poses a major risk for human health due to the impact on morbidity, mortality, and health care costs (33, 49, 106, 182). The Nordic countries are considered a low-prevalence area for antimicrobial resistance (40, 47, 116, 153, 160). However, in general the Nordic countries are experiencing a slow increase in the prevalence of several types of resistance and in particular the MRSA-situation is worrisome (50, 51, 110). Moreover, imported cases of multiresistant pathogens illustrate the dynamic situation with respect to epidemic multiresistant bacterial pathogens or resistance genes that do not require visas when crossing country borders (63). Thus, we need accurate and rapid diagnostic methods to guide antimicrobial therapy and infection control interventions. The objectives of this paper are to examine the rationale for using genetic methods in the detection of antimicrobial resistance, their advantages and limitations, to address some of the technical aspects, and then to discuss the application of these techniques for specific purposes.
UNDERSTANDING ANTIMICROBIAL RESISTANCE AT THE MOLECULAR LEVEL
During the past 30 years tremendous progress has been made in our understanding of the genetics and biochemistry of antimicrobial resistance, the origins of resistance determinants, and routes of transmission of resistance determinants between bacteria (reviewed in (3, 36, 136, 140)). Most antibiotics in clinical use are mother nature's own products or derivatives and bacteria have evolved mechanisms to avoid their inhibitory action. Antimicrobial resistance determinants may be exchanged between bacteria sharing common ecological niches through transformation, transduction or conjugation (3). Horizontal gene transfer events may be infrequent and the acquisition of new DNA involves a biological cost for the recipient (90–92). However, antibiotic selection favours these events creating an environment for biological amplification of resistance and genetic compensation for fitness costs that may favour long-term persistence of antimicrobial-resistance bacteria (4, 9, 12, 18, 79).
The genetic basis for antimicrobial resistance includes: (i) The acquisition and expression of new DNA by horisontal gene transfer or (ii) mutations in cellular genes or acquired genes that alter antimicrobial target sites or affect gene expression. The genetic alterations mediate a diversity of biochemical mechanisms of resistance that have been further refined by pathogenic bacteria: (i) Enzymatic inactivation of the antimicrobial agents; (ii) Target substitutions, amplification or modifications bypassing the binding or reducing the affinity for the antimicrobial agent; (iii) Barriers or efflux pumps reducing access to the target. The diversity of resistance mechanisms is illustrated in Fig. 1 and has recently been the subject of a comprehensive review (140).
Genetic and biochemical research into antimicrobial resistance has also provided insight into the molecular basis for cross- and co-resistance. The concept of cross-resistance to various drug families is illustrated by overlapping targets for antibiotics, as shown by the decreased susceptibility to structurally unrelated macrolides, lincosamides and streptogramins (MLS) following the synthesis of ribosomal methylases (141). Drug efflux has been recognised as common resistance mechanisms in both Gram-positive and Gram-negative bacteria (reviewed in (114, 127)). Analysis of bacterial genomes has revealed a number of putative pumps and their contribution to reduced antimicrobial susceptibility is probably not fully understood. The broad substrate transporters may account for reduced susceptibility to several antibiotic families. The increased occurrence of genetically linked and co-expressed resistance determinants illustrates the concept of co-resistance. This is best exemplified by integrons (144), first described in Gram-negative (65) and now also in Gram-positive bacteria (112, 113), as well as mobile plasmids carrying multiple resistance genes (138). Genetic linkage and co-expression implies that the use of any antibiotic that is a substrate for one resistance mechanism will co-select for resistance to the others and thus maintain the entire gene set.
In more practical terms, the elucidation of biochemical mechanisms for antimicrobial resistance and their genetic support has been fundamental for improving antimicrobial susceptibility testing and therapeutic interpretation of resistance phenotypes (35, 97). Interpretive reading of antibiogram data takes advantage of an observed phenotype in combination with an understanding of the underlying resistance mechanism. The interpretation predicts the resistance mechanism(s) from the actual phenotype and acts as a basis for decision making in antimicrobial therapy. For example, the detection of methicillin resistance in staphylococci allows reporting of resistance to other β-lactams that have not been tested because the resistance mechanism predicts treatment failure. Qualified interpretation requires accurate bacterial speciation and careful selection of indicator drugs that are best able to discern certain resistance mechanisms. This notion is illustrated by the use of oxacillin to screen for reduced susceptibility to penicillin in pneumococci. Finally, the concept of detection and characterisation of antimicrobial resistance at the genetic level has evolved as a direct consequence of our increased understanding of antimicrobial resistance at the molecular level (34).
PHENOTYPIC VERSUS GENOTYPIC METHODS: ADVANTAGES AND LIMITATIONS
Conventional antimicrobial susceptibility testing requires bacteria in pure culture after biological amplification from the clinical sample. Hence, it may routinely take at least 24–48 h to obtain an antimicrobial profile. Obviously, the development of rapid genetic assays would be an attractive approach to targeting specific resistance determinants (30, 34, 55).
The advantages of genotypic detection of antibiotic resistance include: (i) A YES (presence) or NO (absence) answer for a defined resistance determinant. (ii) Genotypic detection is not dependent upon phenotypic categories such as susceptibility, intermediate susceptibility and resistance for which breakpoints may vary between countries. (iii) Resistance mechanisms involved in low-level resistance could be difficult to detect using phenotypic methods. (iv) Genetic assays can be performed directly with clinical specimens and bypass phenotypic expression, reducing the detection time. This is particularly important for difficult-to-culture organisms. (v) Easy and early interpretation allows early therapeutic predictions. (vi) Genetic assays may reduce the biohazard risk associated with conventional culture methods.
On the other hand, the genotypic approach contains certain limitations and pitfalls: (i) Genetic detection is based on screening for resistance determinants whereas decision making in antimicrobial therapy is preferably based on the detection of susceptibility. (ii) You can only screen for what you already know and genetic methods do not take into account new resistance mechanisms. (iii) There are silent genes and pseudogenes that may cause false-positive results. (iv) Accordingly, mutations in primer binding sites may preclude PCR amplification, generating false-negative results. (v) There is low clinical sensitivity when performed directly on mixed flora samples due to inhibition of nucleic acid amplification or a limited number of targets. (vi) Finally, regulatory mutations that affect gene expression are not detected unless a quantitative measurement of the specific mRNA is targeted. This is particularly relevant for resistance mechanisms occurring in multiresistant non-fermenting gram-negative bacteria such as Stenotrophomonas maltophilia, Pseudomonas aeruginosa and Acinetobacter species (67, 93, 102).
Hence, the genetic approach based on today's test principles cannot substitute for phenotypic methods in routine antimicrobial susceptibility testing. Novel resistance mechanisms will arise continuously or unknown pre-existing resistance genes will be mobilized from environmental reservoirs and spread under antimicrobial selection (10). The most notable resistance mechanisms that have emerged include extended-spectrum β-lactamases (95) and metallo-β-lactamases (98, 115) in gram-negative bacteria, transferable high-level resistance to glycopeptide resistance in enterococci and staphylococci (23, 87, 169), as well as broad-spectrum high-level aminoglycoside resistance mediated by the 16S rRNA methylase gene in Pseudomonas aeruginosa (184). Thus, the role of traditional susceptibility testing will continue to be important. Rather the rationale for genetic assays is to complement conventional phenotypic analyses: (i) Confirm specific resistance mechanisms. For example methicillin resistance in staphylococci may demonstrate heterogenous low-level expression not readily detected by culture-based methods. Differentiation between MRSA and borderline oxacillin-resistant S. aureus (BORSA) strains may challenge phenotypic tests (99). (ii) Rapid identification or exclusion of resistance determinants in complex clinical samples for early intervention in infection control, e.g. screening samples for MRSA or glycopeptide-resistant enterococci (GRE). (iii) For molecular epidemiological purposes to analyse the spread of specific resistant pathogens and/or resistance determinants. (iv) Detection of resistance mechanisms in slow-growing organisms e.g. Mycobacterium tuberculosis. (v) Detection of genetic elements involved in the accumulation (integrons) and spread of resistance genes (conjugative plasmids and transposons).
The application of genetic assays for detection of antimicrobial resistance is also dependent upon potential costs savings and user-friendly testing formats of the techniques. The recent developments in multiplex and real-time PCR assays have fuelled clinical acceptance of genetic tests and will certainly lead to increased use (31).
GENOTYPIC METHODS AND TECHNICAL ASPECTS: FROM RESEARCH TO ROUTINE DIAGNOSTICS
Genetic methods for the detection of antimicrobial resistance genes and their expression take advantage of the development of nucleic acid hybridisation and amplification techniques. Both approaches depend on available genetic information in large databases that is used to design labelled single-stranded nucleic acids (probes) and amplification oligonucleotides (primers) complementary to the target of interest. The choice of target is of fundamental importance for the analytic sensitivity and specificity of the method and should take into consideration available information on conserved and variable regions within the antimicrobial resistance genes or their genetic support target. Thus, nucleic acid probes and primers may be specific for a defined gene or single nucleotide polymorphism or universal for a group of related resistance determinants. Alignment of multiple nucleotide sequences from the target of interest that are available from public databases or other sources will discern conserved or polymorphic regions that can be used for primer or probe selection. Several commercial programmes as well as free resources on the Internet are available for DNA sequence analysis and use when designing amplification primers (Tables 1 & 2). Selected primers and probes should be carefully checked to exclude potential cross-reacting sequences, especially when used with complex flora samples.
Table 1. Resources for primer design and analyses used in our laboratory
Polymerase chain reaction (PCR) has been the most commonly used nucleic acid amplification technique in the detection of antimicrobial genes and their genetic support (7, 111, 145). Convential PCRs, defined as separate amplification and post-PCR detection assays, have been described for most resistance determinants (55, 134). The laborious post-PCR work and problems with carry-over contamination have been largely removed by the advent of real-time PCR defined as the ability to monitor the amplified product during amplification. Real-time PCR techniques have permitted the development of routine diagnostic applications for the microbiology laboratory (100). Several reports have described the use of these techniques for detection of resistance determinants and surveillance of antimicrobial-resistant bacteria (26, 52, 56, 61, 69, 73, 75, 80, 105, 108, 122, 133, 135, 147, 162, 181). The ability to monitor the accumulating amplicon in real time is based on labelled primers, oligonucleotide probes and/or fluorescing amplicons producing a detectable quantitative signal related to the amount and specificity of the amplicon. Several improvements have been introduced. Reduced amplicon size, shorter cycling times and removal of separate post-PCR detection systems have allowed automation, reduced the detection time, and minimised the risk for carry-over contamination. Technical aspects in the recent developments of real-time PCR methods and their use in diagnostic microbiology have recently been elegantly reviewed (100). Other significant technical developments include multiplex PCR assays using more than one primer set for simultaneous detection of several antimicrobial resistance genes (43, 46, 103, 123, 128, 131, 157, 170). Fluit and colleagues have published an extensive review of genetic techniques and their application in the detection of resistance genes (55). Furthermore, a comprehensive table describing primers for detection, probing and sequencing of genes or mutations associated with antimicrobial resistance has recently been published (134).
The latest developments in nucleic acid sequence techniques have made the detection of mutational resistance easier by rapid DNA sequence analysis (142). These innovative techniques have been used in the detection of linezolid resistance in enterococci (154) as well as rapid bacterial identification (143). Their diagnostic potential in detection of antimicrobial resistance mechanisms associated with single nucleotide polymorphism in housekeeping genes needs to be pursued and they may replace older techniques (19, 29).
DNA microarray technologies offer a promising method for detection of antimicrobial resistance genes and mutational resistance (13, 94, 183). The method is based upon gene-specific probes (oligonucleotides or PCR amplicons) deposited on a solid surface in a lattice pattern. The test DNA is extracted labelled and hybridized to the array. Target-probe duplexes are detected with a reporter system. Detection and identification of multiple tetracycline resistance genes by DNA microarray were recently described (21). Microarray technology enables detection of a large number of resistance genes in a single experiment and has the potential for significant automation in a microchip format. However, a cost-effective and user-friendly format for application in antimicrobial susceptibility testing remains to be developed.
QUALITY ASSURANCE: A MUST
It is critical that genetic assays are validated and quality assured. Amplification methods are more easily adapted in the laboratory compared to DNA probe assays and are the preferred methods for genetic detection of resistance determinants. An internal amplification control for both sample preparation and amplification is recommended to exclude false-negative results using consensus 16S rDNA primers or a more genus- or species-specific target; e.g. the nuc or femA gene for Staphylococcus aureus (17, 74, 171). It is also critical that negative controls without template DNA and positive controls with defined targets be included to check for false- positive and false-negative results, respectively. Physical separation of specimen handling and preparation of amplification reagents is preferred. The use of conventional PCR requires amplicon detection procedures in separate rooms to prevent carry-over contamination. The specificity of the amplicon can be confirmed by various methods: probe assays, electrophoretic mobility, restriction fragment length polymorphism (RFLP) analysis, single-strand conformational polymorphism (SSCP) analysis or DNA sequencing. The recent developments within real-time PCR using in-tube monitoring of specific amplicons have eliminated the need for post-PCR confirmation.
It is obligatory to perform an in-house validation of PCR methods before they can be used for clinical diagnostic purposes. This also applies to published methods as they may not have undergone rigorous testing. The validation of any genetic method should also take into consideration whether its intended use is based on nucleic acid extracts from biologically amplified material (i.e. pure bacterial cultures) or clinical samples containing a complex sample background of diverse microbial and host cell DNA. The availability of commercial kits with integrated amplification and detection, built-in controls, etc. will overcome some of the problems associated with “in house” tests and be more convenient (31).
DETECTION OF METHICILLIN-RESISTANT STAPHYLOCOCCUS AUREUS (MRSA): “DO NOT LET THE DEVIL OVER THE BRIDGE”
Staphylococcus aureus is one of the major human pathogens within hospitals and in the community. The emergence of methicillin resistance in S. aureus expressing cross-resistance to all β-lactams has been associated with higher morbidity and mortality as well as increased hospital costs (33, 49). Moreover, the increased prevalence of MRSA has a profound effect on the overall use of antibiotics in the clinical setting. The increased use of glycopeptide antibiotics has been associated with the emergence of vancomycin-resistant enterococci and the recent transfer of high-level glycopeptide resistance to S. aureus (23). Thus, it is of public health interest to reduce the spread of MRSA. The Nordic countries are in a particularly favourable position with a prevalence rate below 1% (47). However, an increasing number of MRSA-infections have been reported from several Nordic countries, indicating that future work on the containment of MRSA will require additional efforts, including rapid and accurate microbiological methods (50, 51, 116).
Methicillin resistance in S. aureus is caused by the acquisition of the mecA gene encoding a β-lactam low-affinity penicillin-binding protein (PBP), termed PBP 2a or PBP2′. PBP2′ substitutes for the essential functions of the high-affinity PBPs in the presence of β-lactam antibiotics, hence rendering the bacteria resistant to this general and important class of antimicrobials (24). Consequently, rapid and accurate identification of MRSA in clinical samples is of considerable importance for the institution of early correct therapy and to reduce the workload associated with MRSA control and surveillance. Numerous molecular methods have therefore been developed to confirm phenotypically suspected MRSA and to reduce the detection time of MRSA in clinical samples, including blood culture (11, 17, 132, 171). Primers (mecA and nuc) used in our laboratory for multiplex PCR confirmation of MRSA are given in Table 3. Real-time PCR assays have significantly improved the application of PCR for these purposes (52, 56, 66, 73, 75, 80, 135, 147, 162). Genetic methods for the detection of MRSA have traditionally been based on two targets: a species-specific gene target for identification of S. aureus and the mecA gene encoding PBP2′. These methods are together with the immunological detection of PBP2′ used as standard confirmatory tests for MRSA. Fig. 2 illustrates multiplex real-time amplification of mecA and nuc using SYBR® Green and melting point determination (73).
Table 3. PCR primers used in our laboratory for conventional PCR detection of MRSA (mecA, nuc), GRE (vanA, vanB1, vanB consensus, vanC1, vanC2–3), class A β-lactamases (blaTEM, blaSHV, blaCTX-M), class C β-lactamase (blaCMY-2), metallo-β-lactamases (blaVIM, blaIMP, blaSPM), class 1 integron, vanB-operon, and Tn1546
a Annealing temperature in 1×Applied Biosystems PCR buffer with 1.5 mM MgCl2. b Annealing temperature in 1.2×Applied Biosystems XL PCR buffer with 1.1 mM Mg(OAC)2. c Annealing temperature in 1×Applied Biosystems XL PCR buffer with 1.4 mM Mg(OAC)2.
GGG ATC ATA GCG TCA TTA TTC AAC GAT TGT GAC ACG ATA GCC
GCG ATT GAT GGT GAT ACG GTT AGC CAA GCC TTG ACG AAC TAA AGC
GTT GCA ATA CTG TTT GGG GG CCC CTT TAA CGC TAA TAC GAT CAA
Considerable progress has recently been made in the characterisation of the genetic support of the mecA gene. Several molecular markers have been identified that may be useful for epidemiological studies as well as diagnostic purposes (57, 72, 76, 118). The 2.1 kb mecA gene is carried by a new class of genetic elements, designated the mec element or the Staphylococcal Cassette Chromosome mec (SCCmec), inserted near the chromosomal origin of replication (72). Excision of SCCmec has been observed in vivo and was dependent on the recombinase genes (ccrAB) located within the element. Three homologous pairs of ccrAB genes have been described and recently a new ccr-gene homologue called ccrC was described (76A). The mecA gene and the upstream regulatory genes, mecI and mecR, form the mecA gene complex (mecI-mecR1-mecA). Different mec-gene complexes have been described in staphylococci based upon insertions (IS431) 3′ of mecA and partial deletions in the mecI-mecR1-region (72, 76). The SCCmec types I-V (21–67 kb) vary in their overall genetic composition, type of recombinase genes (ccrAB and ccrC) and class A, B, C, or D mec classes (72, 76 and 76A). Community-acquired MRSA (CA-MRSA) is emerging worldwide with no connection to hospitals. CA-MRSA is generally susceptible to other classes of antimicrobial agents, expresses heterogeneous methicillin resistance more frequently and carries the SSCmec type IV (72, 76). Epidemic CA-MRSA clones producing the Panton-Valentine leukocidin have also been described (76). More SCCmec diversity is to be expected. Recently, new variants of SCCmec were identified in Norwegian clinical isolates of S. aureus and coagulase-negative staphylococci (CoNS) (68). The closer genetic relationship among SCCmec-sequences within Norwegian staphylococci than between Norwegian and international MRSA indicates an ongoing transfer of mecA-containing mobile elements within the staphylococcal community. This and other observations challenge the original concept of the horizontal transfer of mecA as a rare event and that the spread of methicillin resistance in S. aureus seems to be predominantly due to the clonal expansion of very few lineages (119). The origins(s) of SCCmec and the mechanism(s) for mecA transfer remain to be elucidated.
Understanding the detailed genetic organization of SCCmec has led to new concepts in MRSA-specific PCRs. Traditional PCR approaches which have been based on the detection of two different targets, one specific for S. aureus and one specific for SCCmec, cannot be applied for the direct detection of MRSA from clinical specimens because they may contain multiple staphylococci, including mecA-positive coagulase-negative staphylococci (CoNS) and S. aureus. On the other hand, the integration of SCCmec at a specific site (attBscc) in an open reading frame (ORF) of unknown function (orfX) the S. aureus chromosome has allowed the development of MRSA-specific PCR-strategies that bridge the SCCmec-chromosomal junction site. Huletsky and co-workers have recently described a real-time multiplex PCR assay for the detection of MRSA directly in clinical specimens containing different staphylococci (75). Their multiple primers target the various SCCmec right extremity sequences as well as the chromosomal orfX gene located to the right of the SCCmec integration site in combination with molecular beacon probes. The validation procedures revealed the detection of 1,636/1,657 (98.7%) MRSA isolates and the misidentification of 26/569 (4.6%) MSSA strains. Moreover, none of the 62 non-staphylococcal bacterial species or 212 methicillin-resistant or 74 methicillin-susceptible CoNS was detected by this method. The assay allowed the fluorescence detection of MRSA < 1 h with an analytic sensitivity of ∼25 CFU per sample using MRSA-negative nasal specimens containing MSSA, MRCoNS and MSCoNS spiked with MRSA. Accordingly, real-time PCR-methods targeting the flanking regions of the integration site for SCCmec seem to offer a powerful approach when developing methods that can specifically detect MRSA directly in clinical samples. Still, the methods need additional validation for use directly with clinical samples and should take into consideration the most prevalent MRSA-types at the local, regional and national level. The concept may also be challenged by new SCCmec types appearing as described in the Nordic environment (68).
In conclusion, the “search-and-destroy” strategy for MRSA in the Nordic countries requires a rapid and reliable microbial diagnosis of MRSA. A multiplex PCR assay targeting the mecA gene and an inherent S. aureus gene are therefore recommended for verification of suspected MRSA isolates. A user-friendly real-time PCR format should be a valuable tool in the routine laboratory. This approach could also be implemented in MRSA-screening to reduce the turnaround time for results and the cost of cultures and patient isolation (reviewed in (45)). The use of real-time PCR directly with clinical samples amplifying a chromosomal target in MRSA that links the SSCmec and the chromosomal insertion site is a promising concept (75). This application is discussed later in this paper.
DETECTION OF GLYCOPEPTIDE RESISTANCE IN ENTEROCOCCI AND STAPHYLOCOCCI: THE VAN ALPHABET
Glycopeptide antibiotics are important alternative therapeutic agents in the treatment of infections caused by multiresistant gram-positive bacteria. Vancomycin and teicoplanin block the cell wall synthesis by binding to the C-terminal D-alanyl-D-alanine (D-Ala-D-Ala) residues and inhibiting the assembly of peptidoglycan precursors (8). The first description of glycopeptide-resistant enterococci (GRE) in the UK (169) and France (87) was characterised by transferable high-level resistance later classified as VanA-type glycopeptide resistance. The genetic basis and biochemical mechanisms for glycopeptide resistance have been extensively described (reviewed in (8)). Glycopeptide resistance is phenotypic and genotypic heterogeneous. Several genes within an operon are responsible for the expression of the resistance phenotype, consistent with the presence of an alternative pathway for peptidoglycan synthesis producing precursors with a low affinity for glycopeptide antibiotics.
Six types of glycopeptide resistance have currently been described in enterococci: the van alphabet A-E and G (1, 8, 42, 54, 129). The different types can be genotyped based on sequence differences in the ligase gene (27, 38, 46, 123, 124, 131). (See Table 3 for description of van-primers used for conventional PCR.) Depardieu and co-workers have recently described a multiplex PCR covering the whole alphabet (43). However, the ever growing subtypes within the vanD-group has already emphasized the need for consensus D-primers (14A). VanA and VanB-phenotypes are the most commonly encountered forms and clinically the most important due to their epidemic appearance in the clinical setting (159, 167). (i) The VanA-type of resistance is characterised by inducible high-level resistance to both vancomycin and teicoplanin. The vanA gene cluster is located on the non-conjugative transposon Tn1546, which can be part of mobile chromosomal or extrachromosomal elements. Typing of Tn1546 elements has been used for molecular epidemiological purposes to track the spread of the vanA gene cluster between different reservoirs, including farm animals (64, 78, 152, 158, 178, 180). Primers used for PCR-restriction fragment length polymorphism (PCR-RFLP) analysis of Tn1546 (64) are described in Table 3. (ii) VanB-type strains express a variable level of inducible vancomycin resistance. The strains are generally susceptible to teicoplanin, but teicoplanin- resistant strains have been detected during antibiotic selection both in vitro and in vivo. The vanB gene cluster is more heterogeneous at the DNA sequence level and can be divided into three subtypes that seem to reflect divergent evolution in different reservoirs (38, 124). Primers for PCR-RFLP typing of vanB1- and vanB2-operons are given in Table 3 (64). The vanB2-operon seems to be linked to the putative conjugative transposon Tn5382/Tn1549 associated with intercellular spread of vanB-resistance (22, 39, 58).
In summary, the inducible character of the glycopeptide resistance genes implies problems with detection of some GRE-strains using conventional phenotypic tests, because of the time lag between induction and expression of resistance to a detectable level (reviewed in (161)). However, several methods have now been improved and validated for phenotypic detection of vancomycin resistance. The vancomycin agar-screening test has been extensively used and is recommended by NCCLS (179). Thus, genotypic methods are mostly used for confirmatory, epidemiological and infection control purposes (32). Recent developments in real-time PCR targeting the vanA and vanB genes support the use of PCR directly on clinical samples in GRE-screening efforts to reduce costs associated with conventional culture screening techniques (122). The potential application in detection of GRE-carriage is discussed later in this paper.
GENOTYPIC DETECTION OF RESISTANCE TO AMINOGLYCOSIDES IN ENTEROCOCCI AND STAPHYLOCOCCI: IMPORTANT DISCREPANCIES BETWEEN GENOTYPIC AND PHENOTYPIC ASSAYS
Aminoglycosides are important antimicrobial agents often used in combination with glycopeptides and β-lactams for the treatment of invasive infections caused by several agents including α-haemolytic streptococci, staphylococci and enterococci. The prevalence of aminoglycoside resistance among staphylococci and enterococci is relatively high in European countries, although geographical variations occur (47, 149). The prevalence of high-level gentamicin in E. faecalis and E. faecium may vary between hospitals in the Nordic countries (153, 175). The presence of aminoglycoside resistance in staphylococci is highly correlated with methicillin resistance due to genetic linkage between resistance determinants (44, 84, 76, 116, 149). Thus, the prevalence of gentamicin resistance in S. aureus is low and in contrast to the high frequency of aminoglycoside resistance in coagulase-negative staphylococci (CoNS) in Nordic hospitals (84).
The main mechanism of aminoglycoside resistance in gram-positive cocci is drug inactivation by aminoglycoside-modifying enzymes (AME) encoded within mobile genetic elements (plasmids and transposons). The following three AMEs are the most prevalent and of clinical significance since they modify and thereby inactivate the traditional aminoglycosides of therapeutic importance (Table 4): the bifunctional enzyme aminoglycoside-6′-N-acetyltransferase/2′′-O-phosphoryltransferase [AAC(6′)/APH (2′′)] encoded by the aac(6′)-Ie- aph(2′′)-Ia gene; aminoglycoside-4′-O-nucleotidyltranseferase I [ANT(4′)-I] encoded by the ant(4′)-Ia gene; aminoglycoside-3'-O-phosphoryltranseferase III [APH(3′)-III] encoded by the aph(3′)-IIIa gene. The genes encoding these AMEs are highly conserved within enterococci and staphylococci (121). Hence, the same primers can be used for their detection in both genera. Table 4 summarises information on primers used for conventional PCR of aminoglycoside resistance determinants in our laboratory. These primers could also be used in a multiplex PCR format (170).
Table 4. PCR primers (170) used in our laboratory for conventional PCR detection of genes encoding aminoglycoside-modifying enzymes (AME) in staphylococci and enterococci, amplicon sizes and corresponding phenotype
GEN=gentamicin, TOB=tobramycin, NET=netilmicin, AMK=amikacin, KAN=kanamycin. S=susceptible, R=resistant, r=reduced zones but likely to remain susceptible at BSAC breakpoints according to Livermore 2001 (97).
Variation in substrate specificity of these enzymes explains differences in antibacterial activity among the aminoglycosides, and may have important clinical consequences with regard to the choice of indicator drug for antimicrobial susceptibility testing. For example, the most frequently encountered AME in staphylococci and enterococci, the bifunctional enzyme, displays AAC(6′) and APH(2′′) activity that modifies to a different degree essentially all the clinically available aminoglycosides, except streptomycin (107, 146, 149, 168). The bifunctional enzyme has higher substrate specificity for gentamicin. Thus, gentamicin is the preferred substrate in testing for aminoglycoside resistance in gram-positive cocci. AAC(6′)/APH(2′′) abolish the bactericidal activity of amikacin and netilmicin, excluding them as therapeutic agents. However, the lower affinity for amikacin and netilmicin compared to gentamicin affects their bacteriostatic activity to a lesser extent. Consequently, aminoglycoside susceptibility testing with those substrates may give misleading results and should not be performed (35, 107, 121).
GENETIC DETECTION OF BROAD-SPECTRUM β-LACTAM RESISTANCE IN GRAM-NEGATIVE BACTERIA: A NEVER-ENDING ARMADA OF GENES
β-lactamases are the most common cause of resistance to β-lactam antibiotics in gram–negative bacteria. B-lactamases disrupt the β-lactam (amide) bond and inactivate the antibiotic. A diversity of different families of β-lactamases have been isolated and characterised. (For the most recent reviews see (14, 15, 95).) The Ambler and the Bush-Jacoby-Medeiros systems are currently used for classification of β-lactamases based on similarities in amino acid sequence and functional activities, respectively. The Ambler scheme separates β-lactamases in 4 classes; A, C and D are serine β-lactamases, whereas B are sink-dependent metallo-β-lactamases. We will here focus on (i) the extended-spectrum β-lactamases (ESBLs) and their favourite host species, E. coli and Klebsiella spp., and (ii) metallo-β-lactamases. We will not discuss the chromosomally encoded clinical important, inducible and potential stably expressed AmpC β-lactamases in Enterobacteriaceae, especially encounter in Enterobacter spp., Citrobacter freundii, Serratia spp., Morganella morganii, Providencia stuartii, and P. rettgeri (reviewed in 95).
ESBLs in E. coli and Klebsiella spp.
The two most commonly encountered class A β-lactamases in E. coli and Klebsiella spp. are TEM-1 (after the patient named Temoniera) and SHV-1 (for sulphydryl variable). TEM-1 is of unknown origin, plasmid mediated, and was first described in 1965 (41). SHV-1 is inherent in K. pneumoniae and usually plasmid mediated in other klebsiellae and E. coli. TEM-1 and SHV-1 are penicillinases with little activity against cephalosporins. However, they are progenitors of the most common extended ESBLs. ESBLs are enzymes with an extended substrate profile due to amino acid changes enabling hydrolysis of most cephalosporins, including broad-spectrum cephalosporins (third and fourth generation cephalosporins).
ESBLs were first described in the mid 1980s (82). The impact of ESBL detection is important both from a therapeutic point of view and for infection control purposes. ESBL-producing strains are considered resistant to all penicillins and cephalosporins. Clinical observations and animal model studies have shown therapeutic failures and increased mortality when treating infections with ESBL-producing bacteria with cephalosporins, even with MIC-values around the breakpoint for the corresponding antibiotic (86, 125, 148, 155). Derivatives of TEM or SHV enzymes and CTX-M (cefotaximase) enzymes are the most common ESBLs in clinical isolates of Enterobacteriaceae (14, 15, 139). Unlike most TEM and SHV ESBLs, CTX-M enzymes hydrolyse cefotaxime better than ceftazidime. A large number of other ESBLs have been described and importantly some of these have been found within plasmid-mediated integrons (15), underlining their broad-spectrum transfer potential. A recent update on the Internet revealed >130 TEM, >50 SHV and >30 CTX-M different ESBLs (www.lahey.orgstudies).
Phenotypic criteria and detection methods for ESBL-production in E. coli and Klebsiella sp. are based on the combined use of a β-lactamase inhibitor, usually clavulanic acid, and one or several oxymino-cephalosporins, usually cefotaxime, ceftazidime, ceftriaxone and/or cefpodoxime (88, 96, 120). According to these criteria ESBLs are placed into the functional group 2be of Bush, Jacoby & Medeiros (20). The reduced level of resistance to the oximino-cephalosporin in combination with clavulanic acid is thus the major diagnostic criterion for ESBL detection by various methods (77, 96, 172). However, the sensitivity and specificity of phenotypic tests varies with the chosen cephalosporin due to the different substrate profiles of various ESBLs. The combined use of two broad-spectrum cephalosporins, e.g. cefotaxime and ceftazidime, or cefpodoxime alone is thus recommended for susceptibility testing in E. coli and K. pneumoniae. ESBL-production is subsequently verified by clavulanic acid synergy tests (reviewed by Livermore (96)). The importance of using ceftazidime and cefotaxime in susceptibility testing was also recently shown for ESBL-producing clinical strains of E. coli and K. pneumoniae in Norway (163). The diversity of β-lactamases with the ability to hydrolyse broad-spectrum cephalosporins has been further expanded and made more complex by the detection of class C and class D β-lactamase genes in transferable plasmids (1, 139, 140). Moreover, class C and D β-lactamases are in general not inhibited by clavulanic acid and are thus not considered ESBLs, which is the topic of this paper. To complicate matters even further some OXA-type enzymes mainly found in P. aeruginosa are considered ESBLs (reviewed in (15)). Updated information on OXA-type enzymes and plasmid-mediated AmpC-β-lactamases has recently been published (2, 71).
Several genetic methods have been developed for detection of ESBL-genes, including DNA probes specific for TEM and SHV, PCR, oligotyping, PCR-single strand conformational polymorphism (PCR-SSCP), PCR-RFLP, and LCR (reviewed in (15)). Recently, real-time PCR detection methods have also been described (133). However, the increasing number of additional subtypes within each ESBL-family has placed strict limitations on these techniques with regard to their ability to cover the whole range of variants within each family. The combined use of type-specific PCRs and restriction fragment length analysis (PCR-RFLP) may, however, cover a number of subtypes for TEM (6), SHV (25, 117) and CTX-M (48). Type-specific primers used in our laboratory for conventional PCR detection of TEM-, SHV, CTX-M, and CMY-1/2 (class C bla-genes) genes and subsequent typing by RFLP or sequence analysis in our laboratory are given in Table 3. The gold standard for typing of ESBLs is nucleic sequencing techniques, and recent developments within rapid sequencing techniques will probably make this approach more readily available and cost-effective in ESBL-typing (143).
In conclusion, genetic methods for detection of broad-spectrum β-lactamases are complex and challenging due to the diversity of genotypes and phenotypic expression. Their use is therefore mainly restricted to reference laboratories and to molecular epidemiological studies. For routine diagnostic purposes we recommend susceptibility testing of E. coli and Klebsiellae spp. with the combined use of ceftazidime and cefotaxime or cefpodoxime alone to detect reduced susceptibility to broad-spectrum cephalosporins. ESBL-production is confirmed with clavulanic synergy. For the reference laboratory receiving difficult-to-type strains we recommend interpretive reading of an extended MIC profile for different β-lactam antibiotics. The antibiogram in combination with accurate species identification will most often lead to a qualified direction of further genetic typing of clinically relevant β-lactamases, e.g. enabling the differentiation between class A and class C β-lactamases with extended hydrolysing capacity (15, 95, 139, 140). In particular, one should be aware of the potential misidentification of K. oxytoca with hyperproduction of the chromosomally encoded class A β-lactamase (K1-enzyme) as an ESBL-producing Klebsiella sp. (130). However, careful bacterial identification in combination with a typical antibiogram with decreased susceptibility to all β-lactams except ceftazidime, cephamycins (cefoxitin) and carbapenems should lead to the correct microbial diagnosis of K1-hyperproducing strains. For molecular epidemiological purposes we prefer a PCR-RFLP approach in combination with nucleic acid sequence analysis for detection and characterization of the most common broad-spectrum β-lactamases.
Detection of metallo-β-lactamases
Acquired β-lactamases that significantly hydrolyse carbapenems have been described within Ambler class A, B, C and D (reviewed in (71, 98, 115)). The class B β-lactamases are considered the most clinically important acquired carbapenemases. As metalloenzymes, their biochemical activities are dependent on zinc or other heavy metal ions and hence inhibited by chelating agents. They confer resistance to virtually all β-lactam compounds except aztreonam and are resistant to inactivation by clinically available β-lactamase inhibitors. Mobile metallo-β-lactamases are usually encoded within class 1 integrons that may be spread by transposons and conjugative plasmids (5, 164–166). The epidemic potential is illustrated by their worldwide distribution and nosocomial outbreaks (115, 150). The first metallocarbapenemase-producing clinical isolate of P. aeruginosa in Scandinavia was recently described (63).
The most commonly encountered mobile metallo-β-lactamases are within the IMP and VIM series associated with mobilized class 1 integrons but new subfamilies are emerging (151, 164–166). Their favourite hosts include non-fermenting gram-negative bacteria such as P. aeruginosa and Acinetobacter spp., as well as members within the Enterobacteriaceae.Bacteroides spp. are another clinical important group of bacteria that may possess mobile metallo-β-lactamases conferring carbapenem resistance associated with clinical failure (37). Expression of these enzymes may vary in clinical isolates (150, 151). The variability of MIC values may explain the observed difficulties in detecting IMP- and VIM-positive isolates in the routine clinical laboratory (115). It is recommended that gram-negative clinical isolates, including Enterobacteriaceae, Acinetobacter baumannii and Pseudomonas species with borderline susceptibility to carbapenems, should be considered producers of mobile metallo-β-lactamases and examined in reference laboratories (115). Phenotypic detection methods for metallo-β-lactamases based on synergy between a β-lactam substrate and a β-lactamase inhibitor (e.g. EDTA) have been developed (151, 174, 185). Discrepancies between phenotypic and genetic susceptibility testing have been discerned, which underlines the need for genetic methods (151).
Hence, methods used in reference laboratories should include genetic methods for detection of metallo-β-lactamase genes because of the variable level of expression and difficulties in phenotypic detection. Primers used for conventional PCR detection of class B β-lactamase genes as well as class 1 integrons used in our laboratory are given in Table 3.
RAPID DETECTION OF ANTIMICROBIAL-RESISTANT BACTERIA CARRIAGE: A LONG-DESIRED AIM
Early detection of antibiotic-resistant bacteria is crucial not only for therapeutic decisions but also for infection control. A critical strategy to minimize spread of antimicrobial-resistant bacteria in health care institutions is detection of colonized patients and health care workers to initiate efficient infection control measures. Hence, the benefits of genetic techniques could also be used for the rapid detection of human and non-human reservoirs for antimicrobial-resistant organisms in hospitals (45, 83).
Prevention and control of MRSA and GRE have received particular attention. New developments in traditional culture techniques using enrichment broth for detection of persons colonized with MRSA or GRE are sensitive but laborious and require 2–3 days or more to finalize (176). High costs are also associated with patient isolation, which itself may have negative consequences for patients (81). These problems could at least partly be overcome by the application of rapid, sensitive and specific genetic detection assays either directly using clinical samples or with enrichment broths.
A screening method for MRSA by the combined use of enrichment broth and real-time PCR could be a valuable tool for rapid identification of MRSA in clinical samples as well as reducing the time to obtain negative results (52). Fang & Hedin showed that nearly 90% of the negative samples could be identified by negative nuc-amplification of selective broth cultures within 18 h. Only nuc-positive samples and nuc-negative samples with PCR inhibitors were cultured on plates and processed further, significantly cutting the cost of processing negative samples. This is especially cost-effective in countries such as the Nordic countries with a low prevalence of MRSA. A similar approach has been used for the detection of vanA and vanB genes in enrichment broth cultures from faecal material, with corresponding results (122).
Several groups have reported the use of these techniques directly on clinical samples for the detection of MRSA or GRE (56, 75, 122, 126). The detection of MRSA in mixed flora samples is complicated by the fact that the SSCmec-element may be harboured by S. aureus and/or CoNS. Thus, the genetic detection of SSCmec DNA must be physically linked to DNA derived from S. aureus for valid detection of MRSA . In contrast, the vanA and vanB glycopeptide resistance determinants are almost exclusively linked to enterococci, although vanB-containing elements have been detected in anaerobic bacteria in the human gastrointestinal tract (156). Therefore, the detection of vanA- and vanB resistance genes itself is considered epidemiologically significant. Francois and co-workers evaluated a one-step anti-protein A, immunomagnetic enrichment technique for S. aureus directly on mixed flora samples followed by a triplex quantitative PCR for detection of the mecA, S. aureus femA, and S. epidermidis femA genes (56). Application on 48 clinical samples revealed a 100% sensitivity but only 64% specificity compared to culture, indicating that the assay needs to be further refined. The real-time PCR assay described by Huletsky et al. (75) targeting the SSCmec right extremity sequence and the linked S. aureus chromosomal orfX gene was evaluated with MRSA-negative mixed flora nasal specimens spiked with MRSA. The detection limit was ≈25 CFU of MRSA per nasal swab. A commercial real-time PCR test based upon the innovative concept described by Huletsky et al. for use on nasal specimens (IDI-MRSA; Infectio Diagnostic, Quebec City, Quebec, Canada) has been cleared by FDA and Health Canada (45). Performance characteristics are available on the Internet (IDI-MRSA package insert; www.idi-mrsa.com). The assay described by Palladino et al. (122) for the detection of vanA and vanB genes is also commercially available through Roche Diagnostics as an analyte-specific reagent kit. The performance on faecal enrichment broth cultures was acceptable, but the application directly from rectal or faecal swab samples was complicated by a high-level of PCR inhibition.
Detection of microbial targets based on nucleic acid amplification consists of three steps: sample preparation, amplification and detection. The advent of real-time PCR technologies has made amplification and detection of defined microbial targets in bacterial cultures convenient for routine diagnostic purposes. The major challenge with direct detection of antimicrobial resistance targets in clinical specimens is associated with clinical sensitivity. The sample preparation step is complicated by the concentration limit of target detection and the presence of inhibitory substances. Hence, improvements in sample preparation and affinity capture methods are needed.
DETECTION OF GENETIC ELEMENTS INVOLVED IN THE BUNDLING AND DISSEMINATION OF ANTIMICROBIAL RESISTANCE DETERMINANTS: PREDICTING LINKAGE OF RESISTANCE DETERMINANTS AND UNDERSTANDING CO-SELECTION
Several genetic elements are involved in the assembly and spread of antimicrobial resistance determinants (28, 36, 137, 138, 144). Hence, the detection of the genes encoding the functions necessary for bringing together, mobilizing and transferring resistance will identify the genetic elements involved in the development of transferable multiresistance as well as the expression of co-resistance to clinically relevant antimicrobial resistance. This notion is best illustrated by integrons in gram-negative bacteria (10, 144) and conjugative transposons in gram-positive bacteria (137, 138).
Integrons are described as natural cloning systems that bring together and express open reading frames (ORFs), including genes encoding antimicrobial resistance. The integrons consist of an integrase gene (intI) and a recombination site (attI). Several classes of integrons have been defined based upon sequence divergence among integrase genes. The integrase mediates recombination between attI and an attC site (or 59-base pair element) in free single ORFs termed gene cassettes. In this context the gene cassettes are antimicrobial resistance genes. PCR analyses of class 1, 2 and 3 integrons in clinical isolates of Enterobacteriaceae have revealed a dominance of gene cassettes encoding resistance to aminoglycosides and trimethoprim (70, 89, 104, 177). Class 1 integrons are also of particular interest in the spread of metallo-β-lactamases and sequencing of associated gene cassettes has demonstrated the presence of multiple co-expressed resistance determinants (164, 166). The observation of resistance integrons in mobile genetic elements such as plasmids and transposons supports their capacity for intra- and interspecies transfer (144).
Conjugative transposons are chromosomally located genetic elements that encode the functions necessary for their own excision and intercellular transfer (28). The Tn916-1545 family of conjugative transposons is a clinically relevant example in gram-positive bacteria linked to the spread of resistance determinants (137). The genetic linkage between erm-genes mediating MLSB-resistance and tetracycline resistance determinants has been demonstrated within these promiscuous elements carried by oral commensals as well as Streptococcus pneumoniae and S. pyogenes (62, 109, 173). Thus, the appearance of transferable co-resistance to tetracyclines and macrolides in pathogenic streptococci is understood at the molecular level.
Several other illustrative examples of genetic elements mediating multidrug resistance have been reported (36, 138). These mechanisms underline the ability of bacteria to withstand antibiotic selection and may be used as a target for molecular epidemiological studies of multiresistant bacteria as well as defined antimicrobial resistance mechanisms (48, 53, 57, 68, 79, 89, 101, 178).
The advent of real-time PCR offers a cost-effective, user-friendly format for genetic methods that will fuel their use for the detection and characterization of antimicrobial resistance determinants in routine diagnostic microbiology. The implementation of these assays to detect resistance in clinically important slow-growing organisms, to rapidly identify clinically important resistance mechanisms and to overcome laborious and time-consuming culture techniques in the control and surveillance for MRSA and GRE-carriage is of particular interest (45, 59). For reference laboratories it is important to have a broad repertoire of genetic assays to confirm defined resistance determinants, to sort out ambiguous phenotypic results, as well as to provide a reliable scientific basis for molecular surveillance of antimicrobial-resistant bacteria and resistance determinants in a global network. In the Nordic countries with their shared epidemiology of antimicrobial-resistant bacteria it would be both interesting and cost-effective for their reference laboratories to expand their network with respect to complementary activities in this field.
Upfront investment and molecular expertise are required for the development and validation of in-house genetic techniques and may hinder application in smaller laboratories. Commercialization will certainly improve the user friendliness of these techniques and increase their use. However, we should keep in mind potential overuse and apply these techniques as part of a larger strategy to minimize the development and spread of resistant bacteria.
Technical developments in molecular diagnostic microbiology focus on fully integrated systems for rapid sample preparation and analysis of small volumes. A portable genetic analysis microsystem, including PCR amplification in 200 nl chambers and capillary electrophoresis for fluorescence detection of S. aureus and mecA within 30 min, has recently been described (85). The present perspective points to the development of rapid portable diagnostic devices used for bedside diagnostics. However, despite impressive developments respecting diagnostic microdevices, they are based upon today's test principles and are searching for what is already known. Thus, for antimicrobial susceptibility testing we still require viable bacterial cultures which – through their fascinating diversity of morphology, colour, odour and new antimicrobial properties – can challenge traditional phenotypic methods.
We thank Patrice Courvalin for valuable information and pre-published material.