Multiresistant Gram-negative bacteria: the role of high-risk clones in the dissemination of antibiotic resistance


  • Editor: Teresa Coque

Correspondence: Neil Woodford, Microbiology Services - Colindale, Health Protection Agency, ARMRL, 61 Colindale Avenue, London NW9 5EQ, UK. Tel.: +44 20 8327 7255; fax: +44 20 8327 6264; e-mail:


Multilocus sequence typing reveals that many bacterial species have a clonal structure and that some clones are widespread. This underlying phylogeny was not revealed by pulsed-field gel electrophoresis, a method better suited to short-term outbreak investigation. Some global clones are multiresistant and it is easy to assume that these have disseminated from single foci. Such conclusions need caution, however, unless there is a clear epidemiological trail, as with KPC carbapenemase-positive Klebsiella pneumoniae ST258 from Greece to northwest Europe. Elsewhere, established clones may have repeatedly and independently acquired resistance. Thus, the global ST131 Escherichia coli clone most often has CTX-M-15 extended-spectrum β-lactamase (ESBL), but also occurs without ESBLs and as a host of many other ESBL types. We explore this interaction of clone and resistance for E. coli, K. pneumoniae, Acinetobacter baumannii– a species where three global lineages dominate – and Pseudomonas aeruginosa, which shows clonal diversity, but includes the relatively ‘tight’ serotype O12/Burst Group 4 cluster that has proved adept at acquiring resistances – from PSE-1 to VIM-1 β-lactamases – for over 20 years. In summary, ‘high-risk clones’ play a major role in the spread of resistance, with the risk lying in their tenacity – deriving from poorly understood survival traits – and a flexible ability to accumulate and switch resistance, rather than to constant resistance batteries.


Multiresistance is considered to be a key indicator of problematic bacterial strains because (1) it undermines empirical treatment regimens, thereby delaying the administration of appropriate antibiotic therapy, and (2) it reduces the options of treatments that are appropriate. Both factors contribute to increased patient mortality, and limiting the spread of multiresistant strains is considered to be an infection control priority. Despite this, the term ‘multi-resistance’ remains poorly defined and there is conflicting usage in the literature (Schwarz et al., 2010). There are presently moves to define ‘multiresistance’ as resistance to three or more drug classes and ‘extreme’ resistance as susceptibility to no more than two classes, but this remains contentious, not least because the latter definition potentially varies between laboratories that test different panels of agents. In this review of important Gram-negative clones, we will focus on bacteria that have acquired resistance to carbapenems and/or oxyimino-cephalosporins in addition to other drug classes, but will also mention clones associated with other or, in some cases, no problematic resistances. We focus on four species as exemplars, Escherichia coli, Klebsiella pneumoniae, Acinetobacter baumannii and Pseudomonas aeruginosa, although widespread clones of other genera, including, but not limited to Enterobacter, Proteus, Citrobacter and Serratia, have been reported and may be endemic in certain areas.

Multiresistant bacteria serve as hosts for the multiple genetic elements (genes, integrons, transposons and plasmids) that confer their antibiotic resistance phenotypes. These are discussed elsewhere in this issue. A ‘successful’ bacterial strain is an extremely effective vehicle for the dissemination of these elements for at least two reasons: firstly, all of the hosted resistance elements are transmitted vertically (i.e. from mother to daughter cells) by virtue of the strain's spread and its increasing prevalence; secondly, a successful strain has multiple opportunities to act as a donor and to transfer its resistance elements horizontally to other strains, species or genera.

Multiresistance prompts investigation, particularly when associated with an outbreak in a healthcare or a community setting, and this commonly leads to molecular epidemiological analysis of the causative bacteria and their resistance genes. The recognition of a prevalent and successful multiresistant strain or clone should prompt further questions, not least about: (1) whether it is genuinely a newly emerged entity, or a previously minor strain, with resistance a key driver in its rise to success, or (2) whether it was a pre-existing successful, but antibiotic-susceptible strain receiving increased attention owing to the recent acquisition of multiple resistance by gene transfer. These possibilities are often difficult to distinguish because fully antibiotic-susceptible bacteria attract less attention, which may confound attempts to consider resistant strains in the context of the population biology of a particular species. Evidence is emerging to indicate that particular host clones acquire similar resistance genes on multiple occasions, and that evolution fine-tunes these associations to maintain or increase bacterial fitness (Deschamps et al., 2009).

Defining ‘strains’ and ‘clones’

The terms ‘strain’ and ‘clone’ are used widely, but their meaning can be open to interpretation. As far as possible, we will follow the recommendations of recently published guidelines (van Belkum et al., 2007). Thus, ‘strain’ should not be confused with ‘isolate’ because isolates that are indistinguishable by typing can be considered as descendants of the same strain, whereas ‘strains’ should be distinguishable from one another by typing methods. There is no precise definition of how closely related isolates need to be in order to be considered to represent the same strain, and the decision reached may depend on the typing method used.

Determining whether to use the term ‘strain’ or ‘clone’ is even more difficult, especially given that the logical usage of the word ‘clone’ is to describe isolates that are directly descended from an original; the word is commonly used as an alternative to ‘strain’. Nevertheless, it was proposed that ‘clone’ be used ‘to describe isolates that, although they may have been cultured independently from different sources in different locations and perhaps at different times, still have so many identical phenotypic and genotypic similarities that the most likely explanation is a common origin’ (Orskov & Orskov, 1983). This wider explanation commonly applies to multiresistant bacteria found in multiple locations.

Among the most common methods currently used for genotyping of bacteria are multilocus sequence typing (MLST), pulsed-field gel electrophoresis (PFGE), amplified fragment length polymorphism (AFLP) analysis, other PCR fingerprinting methods and multiple-locus variable tandem repeat number analysis (MVLA). MLST, which uses sequence variation in a number of housekeeping genes to define types, is excellent for evolutionary studies, and for readily comparing isolates, but may lack the discrimination required for outbreak analysis (van Belkum et al., 2007; Diancourt et al., 2010). It has led to the definition of major sequence types (STs) and clonal complexes (CCs) of significance in a number of species, and the growing recognition of successful, international clones, such as K. pneumoniae ST258, increasingly associated with KPC carbapenemase, and the community-acquired K. pneumoniae CC23, associated with invasive disease (Baraniak et al., 2009; Brisse et al., 2009; Kitchel et al., 2009a; Samuelsen et al., 2009b). The relationships between the various STs and CCs can conveniently be depicted by e-BURST diagrams (Feil et al., 2004).

PFGE, following restriction of genomic DNA with an appropriate rare-cutting enzyme, provides excellent discrimination and has been used widely for typing of a range of bacterial species (Coenye et al., 2002; Lau et al., 2008a; Long et al., 2010), but is not as ‘portable’ as are methods such as MLST and MVLA, which describe isolates numerically. PCR fingerprinting methods are popular typing methods, too, but, with the exception of AFLP, are often considered not very reproducible between laboratories. An automated rep-PCR method, however, is increasingly being used, and may provide discrimination similar to that of PFGE (Diaz et al., 2010; Ligozzi et al., 2010; Ratkai et al., 2010). MVLA, a PCR-based method that involves determining the number of repeat units at multiple loci with short sequence repeats, is rapidly gaining popularity for epidemiological investigations, and is also portable (Lindstedt, 2005; van Belkum, 2007; Vu-Thien et al., 2007; Kendall et al., 2010; Teh et al., 2010); its discriminatory power varies with the loci chosen, providing the potential to tailor the typing at an appropriate resolution. The ability of this method to provide higher resolution than PFGE has been exploited in a number of studies (Turton et al., 2009; Li et al., 2010).

The typing method used must be appropriate to the question asked. To track individual transmissions, a highly discriminatory method is required and many typing techniques fall short of this. On the whole, SNP and whole-genome approaches are more suitable for this (Zhang et al., 2006; Niemann et al., 2009; Lewis et al., 2010). On the other hand, such techniques may overemphasize minor differences and fail to recognize that isolates are largely similar. Hence, PFGE subdivided isolates of A. baumannii so well that the fact that most isolates belonged to European clones I and II was not recognized for some time (Turton et al., 2007; Higgins et al., 2009, 2010a). MLST- and sequence-based typing have been instrumental in identifying major international clones, with many of those included here having been described by this method, but each level of discrimination is important; in the above example, the finding of distinct PFGE types of the same clonal lineage of A. baumannii, indicating independent selection from a common ancestor, rather than simple spread, is an important observation that is critical to an understanding of the epidemiology.

Extraintestinal E. coli

Escherichia coli is the most common agent of urinary tract infections and the leading Gram-negative cause of bacteraemia. Since the mid 1990s or early 2000s, according to the country, there has been a worldwide increase in the prevalence of isolates that are resistant to oxyimino-cephalosporins (e.g. and produce extended-spectrum β-lactamase (ESBLs), particularly CTX-M type enzymes (Canton & Coque, 2006; Livermore et al., 2007). Considered globally, CTX-M-15 and CTX-M-14 are the most prevalent members of this family of >110 ESBL variants. As a consequence of this increase, research interest in the biology of multiresistant, extraintestinal pathogenic E. coli strains has increased dramatically, and has included studies to elucidate the epidemiology of the current ‘CTX-M ESBL pandemic’.

There are several MLST schemes for E. coli, but those in widest use are the ‘Achtman’ (Wirth et al., 2006) ( and ‘Pasteur’ (Jaureguy et al., 2008) ( schemes. Inconveniently, there has been only a limited attempt to match the ST designations assigned by these schemes (Jaureguy et al., 2008) ( The application of MLST to isolates producing CTX-M-15 ESBL led to the recognition of an internationally disseminated clone, B2,O25:H4-ST131 (with its ST defined according to the ‘Achtman’ scheme) (Nicolas-Chanoine et al., 2008) (Fig. 1). Representatives were originally reported in Canada, France, Lebanon, Portugal, South Korea, Spain and Switzerland, but this clone has subsequently been reported far more widely (Johnson et al., 2010; Peirano & Pitout, 2010; Platell et al., 2010; Rogers et al., 2011). ST131 is a ‘virulent’ phylogroup B2, uropathogenic E. coli lineage, which can be recognized by allele-specific PCR of the pabB gene (Clermont et al., 2009; Dhanji et al., 2010), and may be subdivided into multiple variants by PFGE (Coque et al., 2008; Lau et al., 2008a; Nicolas-Chanoine et al., 2008). These variants have diverse complements of virulence factors, albeit with key common components (Karisik et al., 2008; Nicolas-Chanoine et al., 2008; Bert et al., 2010; Johnson et al., 2010). Perhaps reflecting this diversity, it would appear that not all ST131 variants are created equal, with some better able to spread than others. In the United Kingdom, the most prevalent variant, designated strain A, has been isolated in >50 microbiology laboratories, serving one or more hospitals each, whereas a second variant, strain D, has been restricted to just one hospital (Woodford et al., 2004b). The biological factors that contribute to success and varied epidemiological features of some variants/strains within a clone remain undefined.

Figure 1.

 ‘Population Snapshot’ determined by eBURST analysis ( showing the clusters of linked STs and unlinked STs in the Escherichia coli MLST database (‘Achtman’ scheme; 1899 STs;; last accessed 30 December 2010). ST labels have been removed. Major CCs and STs referred to in the text are indicated.

In the United Kingdom, the five initially most prevalent strains of E. coli with CTX-M enzymes were defined as A–E by PFGE (Woodford et al., 2004b), and the fact that they all belonged to ST131 was not recognized for several years (Lau et al., 2008a). This clearly illustrates the benefits of investigating emergence events at multiple levels, using both a population-focused approach, such as MLST, and a strain typing method such as PFGE.

Most ST131 isolates in the United Kingdom produce CTX-M-15 ESBL (Woodford et al., 2004b; Lau et al., 2008a, b), but some have the CTX-M-3 enzyme (Rooney et al., 2009), while others have additionally acquired a CMY-type AmpC enzyme (Woodford et al., 2007). This illustrates the fact that members of single widespread clones may harbour different resistance genes and so may display different susceptibility profiles. If clones are circulating widely, they are prone to acquire whatever resistance plasmids are locally prevalent and ST131 variants may harbour diverse plasmids (Woodford et al., 2009). In Japanese hospitals, clone O25:H4-ST131 was one of two dominant clones among ESBL-producing E. coli isolates, but variously produced CTX-M-14, CTX-M-2 or -35 ESBLs rather than CTX-M-15. The other dominant clone in this study was O86:H18-ST38 with CTX-M-9 or CTX-M-14 enzymes (Suzuki et al., 2009). ESBL-negative ST131 isolates have also been found for example in stools of 7% of healthy residents in Paris (Leflon-Guibout et al., 2008). Recently, E. coli ST131 isolates in Madrid were reported to produce the inhibitor-resistant enzymes TEM-30, -33 and -37, all encoded by IncFII plasmids (Martin et al., 2010). The current geographic distribution of ST131 and its resistance traits have been reviewed recently (Rogers et al., 2011). ST131 is the predicted founder of a clonal cluster that comprises 13 single locus variants (SLVs) and three double locus variants (DLVs) (Fig. 1).

Escherichia coli lineages of ‘virulent’ phylogroup D that have been associated with multiresistance include ST69 (‘clonal group A’; CgA), ST405 and O15:K52:H1 (Tartof et al., 2005; Cagnacci et al., 2008; Coque et al., 2008; Platell et al., 2010). ‘Clonal group A’ accounts for up to 50% of cotrimoxazole-resistant isolates from urinary tract infections in some United States locales (Manges et al., 2001); many of its resistance genes are located in a 23-kb ‘genomic resistance module’ within a 105-kb chromosomal island that is inserted into a leuX tRNA integration hotspot (Lescat et al., 2009). ST405 is increasingly reported worldwide, associated or not with production of CTX-M-15 ESBL (Jones et al., 2008; Fam et al., 2010; Mihaila et al., 2010; Smet et al., 2010). The phylogroup D clones O1:HNM-D-ST59, O15:H1-D-ST393/ST1394 and O20:H34/HNM-D-ST354 have been reported among producers of CTX-M-14 ESBL in Spain (Mora et al., 2011). The CMY-2 enzyme was found in eight phylogroup D STs in Spain, mainly STs 57, 115, 354, 393 and 420 (Oteo et al., 2010a).

Both ST131 (Clermont et al., 2008) and ST405 (Mihaila et al., 2010) are highly virulent, as judged by PCR screening for virulence genes and using both in vitro and animal models. Indeed, transmission among family members has been documented for each of these clones (Ender et al., 2009; Mihaila et al., 2010).

In addition to lineages of phylogroups B2 and D, some E. coli clones belonging to groups A and B1 are increasingly reported and associated with resistance: STs 10 and 23 (both phylogroup A) have been associated with ESBLs (Oteo et al., 2009; Valverde et al., 2009) and hyperexpressed AmpC enzymes (Guillouzouic et al., 2009; Cremet et al., 2010); STs 155 and 359 (both phylogroup B1) have been isolated from Spanish, Portuguese and Brazilian outpatients (Valverde et al., 2009; Oteo et al., 2010b).

Several multiresistant E. coli clones (e.g. ST131, ST69, ST23) have been isolated from non-human sources, including farm and companion animals, river water and foods, indicating the wide distribution and the potential complexity of transmission routes (Johnson et al., 2009; Ewers et al., 2010; Guenther et al., 2010; Ozawa et al., 2010; Vincent et al., 2010; Dhanji et al., 2011; Rogers et al., 2011).

Although E. coli has emerged as the major producer of CTX-M-type ESBLs, including in the community setting, it is less often associated with carbapenemases than K. pneumoniae. NDM-1 carbapenemase was detected recently in France in an isolate of E. coli belonging to the ST131 lineage (Poirel et al., 2010a). If carbapenemases become widely established in this or other successful clones, it would create a significant public health concern, given the burden of disease caused by E. coli.

Klebsiella pneumoniae

An MLST scheme for K. pneumoniae was established in 2005. The original analysis identified 40 STs among 67 isolates (Diancourt et al., 2005), and found evidence for only two CCs, which comprised STs 14 and 15 and STs 16–22, respectively. Nineteen ceftazidime-resistant isolates producing ESBLs were divided between 11 STs, and the authors concluded that ‘resistance is not restricted to a few genetic backgrounds and that it is a problem of multiple emergence rather than one of interhospital spread of a few clones.’ (Diancourt et al., 2005). The database ( now includes 542 STs and many workers have documented unequivocally the international spread of certain STs, particularly those associated with KPC carbapenemases.

A blaKPC gene is a readily identifiable marker and has sufficient clinical importance to justify a detailed investigation, meaning that we know the epidemiology of the host strains in considerable detail. Producers of KPC enzymes, particularly the dominant KPC-2 and KPC-3 variants, have become endemic in the United States (Yigit et al., 2001; Smith Moland et al., 2003; Bradford et al., 2004; Woodford et al., 2004a; Bratu et al., 2005c; Endimiani et al., 2008, 2009a, b; Kitchel et al., 2009a), Greece (Cuzon et al., 2008b; Tsakris et al., 2008b; Giakoupi et al., 2009; Pournaras et al., 2009) and Israel (Leavitt et al., 2007, 2010; Navon-Venezia et al., 2009), and with increasing numbers of reports with a wide geographic scatter, including from China (Wei et al., 2007; Zhang et al., 2007; Mendes et al., 2008), South America (Villegas et al., 2006; Monteiro et al., 2009; Pavez et al., 2009; Peirano et al., 2009) and many countries in Europe (Naas et al., 2005; Woodford et al., 2008; Samuelsen et al., 2009b; Curiao et al., 2010; Kassis-Chikhani et al., 2010; Mammina et al., 2010; Toth et al., 2010). The European isolates are often, but by no means always associated with an original patient transfer from an endemic area; similar transfer has also been reported to Canada (Goldfarb et al., 2009). Although K. pneumoniae is the major host species, blaKPC genes are associated with transposons and transmissible plasmids and have been reported in other Enterobacteriaceae (Miriagou et al., 2003b; Hossain et al., 2004; Bratu et al., 2005a, 2007; Navon-Venezia et al., 2006; Zhang et al., 2007, 2008; Cai et al., 2008; Petrella et al., 2008; Tibbetts et al., 2008; Bennett et al., 2009; Goren et al., 2010), in Pseudomonas spp. (Villegas et al., 2007; Akpaka et al., 2009; Bennett et al., 2009; Wolter et al., 2009b; Poirel et al., 2010b) and in Acinetobacter spp. (Robledo et al., 2010).

The globally distributed ST of K. pneumoniae associated with KPC enzyme production is ST258 (Kitchel et al., 2009a; Navon-Venezia et al., 2009). This is part of a CC, founded on ST292, that includes 96 STs, (Fig. 2). Isolates belonging to this complex appear to be widespread independent of KPC carbapenemase. For example, ST258 has also been identified as a host of CTX-M-14 ESBL in South Korea (Ko et al., 2010). Despite this, the rapid, multifocal emergence of clonally related isolates with KPC carbapenemase is highly suggestive of direct, human-mediated international spread, rather than the repeated acquisition of similar plasmids by a prevalent clone.

Figure 2.

 ‘Population Snapshot’ determined by eBURST analysis ( showing the clusters of linked STs and unlinked STs in the entire Klebsiella pneumoniae MLST database (542 STs;; last accessed 30 December 2010). ST labels have been removed. There is a large CC comprising 96 STs (‘CC292’; green box) and with ST292 as the predicted founder. This CC includes many internationally prevalent and multiresistant STs, including STs 11, 14 and 258.

The first isolate with KPC carbapenemase in France was cultured from a patient who had been treated 2 months previously in a New York City hospital (Naas et al., 2005), although its ST was not recorded. The first producers in Norway, Sweden (Samuelsen et al., 2009b) and the United Kingdom (Woodford et al., 2008) included ST258 isolates recovered from patients who had been repatriated from Greece and Israel where KPC-positive ST258 isolates are well known to be prevalent. An isolate belonging to ST11, i.e. the same CC as ST258 (Fig. 2), and also producing a KPC enzyme was imported more recently to the United Kingdom from Curaçao in the Caribbean (N. Virgincar et al., unpublished data). This latter ST also seems to be widespread independent of KPC carbapenemase; it was the most prevalent ST among 77 ESBL-producing isolates isolated from urinary tract infections (57%) and blood (70%) in a multicentre survey of 10 university hospitals in South Korea (Ko et al., 2010). These ST11 isolates from Korea had at least eight different ESBL genotypes (none had KPC enzymes), implying multiple acquisition events and the presence of multiple circulating variants of the clone. The authors emphasized that ‘specific PFGE patterns were not related with a specific hospital or a specific ESBL gene.’ ST11 is also associated with the spread of CTX-M-15 ESBL throughout Hungary (Damjanova et al., 2008).

The first isolate with a KPC enzyme in Hungary was also from a patient repatriated from a Greek hospital (Toth et al., 2010). This, and subsequent isolates, belonged to ST258 and harboured blaKPC-2, blaSHV-12, blaTEM-1 and blaSHV-11. The authors also highlighted a wider association between ST11 isolates and SHV-11 ESBL in Hungary. They proposed that CC258 and 340 belong to a hyperepidemic CC – including ST11, ST258, ST265, ST270, ST277, ST340, ST379, ST407, ST418 and ST437 – that occurs worldwide (Toth et al., 2010) ( This view is supported by MLST data, although the CC encompasses far more STs than originally envisaged (currently 96; Fig. 2).

By contrast, the first KPC-positive isolates reported in Madrid were from eight patients and belonged mainly to a novel type, ST384, with a single isolate of ST388. Isolates belonging to this latter type had already persisted in the hospital for over 10 years as hosts of CTX-M-10 ESBL and had probably acquired blaKPC through horizontal plasmid spread (Curiao et al., 2010). Despite the international dominance of ST258, this study provides a clear warning of the potential for the spread of plasmids encoding KPC carbapenemases (and, indeed, any resistance) to other strains, including those that may be locally prevalent and more usually associated with distinct antibiograms. Molecular epidemiological studies of bacteria with a newly emerged resistance pattern should include local strains as comparators wherever possible to put the emergence into the proper local context.

Other K. pneumoniae STs reported as hosts of KPC enzymes include, in the United States, STs 14, 21, 37, 45, 101, 228, 234, 257 and 259, with most found in at least two states (Kitchel et al., 2009a, b), and, in Israel, STs 277, 327, 340 (all members of ‘CC292’) and 376 (Leavitt et al., 2010).

In addition to strains with KPC carbapenemases, K. pneumoniae isolates that produce VIM-type metallo-β-lactamases (MBLs) are also endemic in Greece. The first 17 producer isolates were detected in 2002 among patients in the intensive care units of three hospitals in Athens (Giakkoupi et al., 2003). These isolates had four different PFGE profiles and isolates belonging to the two dominant PFGE ‘types’ had similar plasmids, containing a class 1 integron with blaVIM-1, aac6, dhfrI and aadA cassettes, indicating that horizontal plasmid spread was occurring in addition to the spread of strains. Although most Greek K. pneumoniae with VIM enzymes are nonclonal, Ikonomidis et al. (2005) showed the spread of a PFGE-related isolates with a similar integron to two hospitals in other parts of Greece. As with plasmids encoding the KPC enzyme, those encoding VIM-1 have been found in other Enterobacteriaceae (Miriagou et al., 2003a; Galani et al., 2007; Tsakris et al., 2007a, b). Currently, there are few MLST data for K. pneumoniae with VIM carbapenemases, although isolates belonging to ST383 have been reported in Greece (Papagiannitsis et al., 2010) and to ST11 in Hungary (Kristof et al., 2010). As more MLST data become available, we may learn of the relationships among organisms that appear to be unrelated by PFGE.

Recent reports describe K. pneumoniae isolates in Greece that have both VIM and KPC enzymes (Giakkoupi et al., 2009; Meletis et al., 2010; Pournaras et al., 2010; Zioga et al., 2010), again illustrating the potential for bacteria to carry seemingly superfluous genetic baggage. These ‘double carbapenemase producers’ appear to represent the acquisition of blaKPC genes by strains that already had VIM enzymes (Zioga et al., 2010); the converse – acquisition of blaVIM by isolates of ST258 with KPC enzymes – has not yet been reported. Isolates of any species producing multiple acquired carbapenemases have not yet been described widely elsewhere, although anecdotal reports exist. Strains that produce multiple noncarbapenemase β-lactamases are well known (Moland et al., 2007).

Klebsiella pneumoniae with a distinct class of carbapenemase, the class D enzyme OXA-48, have caused hospital outbreaks in Istanbul, Turkey (Carrer et al., 2008), and there is evidence to suggest that producers are widely scattered in that country (Gulmez et al., 2008; Carrer et al., 2010), although there has been no reported national surveillance. Elsewhere, reports of OXA-48 producers are sporadic, from Belgium, Egypt, France, Lebanon, Tunisia and the United Kingdom (Cuzon et al., 2008a, 2009, 2010; Zhang et al., 2009; Carrer et al., 2010), with spread of the gene to other Enterobacteriaceae (Gulmez et al., 2008; Castanheira et al., 2009; Goren et al., 2009; Carrer et al., 2010). A hospital outbreak, centred on a renal unit, of K. pneumoniae with the OXA-48 enzyme has been described in London, UK; although there was a dominant ‘outbreak’ strain, several minor strains were defined by PFGE, suggesting the spread of an OXA-48-encoding plasmid (Thomas et al., 2009), and all were distinct by PFGE from ‘clones A and B’, which have caused outbreaks in Istanbul (Carrer et al., 2008). The London ‘outbreak’ strain belonged to the novel ST353 (C.P. Thomas and N. Woodford, unpublished data; Recently, a gene predicted to encode a novel OXA-48-like β-lactamase (with four amino acid differences) was detected in carbapenem-resistant isolates of K. pneumoniae from three geographically dispersed centres in India (Bell et al., 2009).

As in other species, the emergence of carbapenemases in K. pneumoniae limits therapeutic options. Polymyxins are often regarded as agents of ‘last hope’ (Fernandez et al., 2010) and many carbapenemase-producing K. pneumoniae isolates remain susceptible to them. However, resistance to colistin and polymyxin B is an emerging problem in the species, both in countries where the resistant isolates predominantly belong to successful clones and in others where they are diverse. For example: (1) colistin-resistant isolates of the usually susceptible ST258 clone with KPC carbapenemase have been reported in the United States (Bratu et al., 2005b) and Greece (Zarkotou et al., 2010); (2) the first ST258 isolate with the KPC enzyme in Hungary was susceptible to colistin [minimal inhibitory concentration (MIC), 0.5 mg L−1], eight further isolates were resistant (MICs, 16–32 mg L−1) (Toth et al., 2010), indicating either that resistance can emerge rapidly or, perhaps, the reintroduction of a polymyxin-resistant variant; (3) 15 colistin-resistant K. pneumoniae isolates, none with KPC carbapenemase, collected from four hospitals in South Korea during 2006–2007 represented 6.8% of the isolates sampled from nine hospitals, but were diverse and belonged to 14 STs (ST11, ST27, ST37, ST218, ST354, ST356, ST358, ST359, ST363, ST364, ST365, ST366, ST367 and ST369) (Suh et al., 2010). Because polymyxin resistance involves mutations that cause changes in lipopolysaccharide composition, controlling the spread of the resistant strains is the key issue, rather than the horizontal spread of their resistance genes.

Recently, K. pneumoniae isolates with the NDM-1 (New Delhi Metallo-carbapenemase) enzyme have been identified, initially in Sweden (Yong et al., 2009) and subsequently in the United Kingdom and throughout the Indian subcontinent (Abdul Gafur, 2010; Deshpande et al., 2010; Kumarasamy et al., 2010). Although there is evidence of strain spread in some cities (specifically Haryana) in the subcontinent (Kumarasamy et al., 2010), there has been no large-scale investigation or comparison of NDM producers to date. Importation of producers to Europe, and more recently to the United States (Anonymous, 2010), is often associated with patients who have had prior contact with the healthcare systems in India or Pakistan, and most of the isolates have diverse PFGE profiles (Kumarasamy et al., 2010), meaning that the European and the United States cases probably represent a near-random ‘sampling’ of strains present in different parts of the subcontinent. It is noteworthy, however, that the first-reported NDM-positive isolate (Yong et al., 2009), two isolates from Oman (Poirel et al., 2011a), and seven from Nairobi, Kenya (Poirel et al., 2011b), all belonged to ST14, a member of the large CC292 complex. Isolates with NDM-1 carbapenemase are extremely multiresistant: they typically harbour ESBLs and acquired AmpC enzymes besides the NDM MBL, thus conferring resistance to all β-lactams; they also produce 16S methylases, often ArmA or RmtC, which confer resistance to all clinically useful aminoglycosides. Most remain susceptible to polymyxins, although here too resistance has been observed in some isolates (Kumarasamy et al., 2010).

The international spread of bacteria, particularly K. pneumoniae, with NDM or KPC carbapenemases represents one of the most pressing resistance-related public health issues. Both have the potential to impact dramatically on healthcare, although it is impossible to predict which will rise to dominance.

Acinetobacter baumannii

Two MLST schemes exist for A. baumannii (Bartual et al., 2005; Diancourt et al., 2010) with only three loci in common ( and The population structure of clinical isolates of A. baumannii is dominated by three international clonal lineages known as European clones I, II and III, corresponding to CC1 (comprising ST1, ST7, ST8, ST19 and ST20), CC2 (comprising ST2, ST45 and ST47) and CC3 (ST3 and ST14) in the MLST scheme of Diancourt et al. (2010), and with most outbreak strains belonging to the first two of these lineages. In the other scheme (Bartual et al., 2005), reference strains of European clones I and II have STs of 12 and 6, respectively; ST22, and SLVs clustered into CC22, belonging to European clone II, have been reported widely (Fu et al., 2010; Mugnier et al., 2010; Park et al., 2010). A sequence-based typing scheme based on three genes (ompA, csuE and blaOXA-51-like) has also been described (Turton et al., 2007) (; multiplex PCRs based on this facilitate the rapid identification of representatives of the main clonal lineages. Sequencing of the blaOXA-51-like gene alone has also been used (Hamouda et al., 2010).

The first description of the international lineages was based on comparisons of cell envelope protein profiling, ribotyping and AFLP genomic fingerprinting of epidemic and nonepidemic A. baumannii strains from geographically distinct European hospitals (Dijkshoorn et al., 1996). Since then, there have been abundant reports of outbreaks due to sublineages of these clones, including the French AYE VEB-1 strain (European clone I) (Naas et al., 2006), the United Kingdom OXA-23 clone 1 and the United Kingdom South East clone (European clone II) (Coelho et al., 2006). These have each affected large numbers of hospitals. There are numerous other examples from many countries (Da Silva et al., 2007; Nemec et al., 2008; Giannouli et al., 2009; Donnarumma et al., 2010; Fu et al., 2010). A few other CCs, distinct from European clones I to III, have been described. They also include genetically similar isolates from diverse geographical locations, clearly fitting the description of the term ‘clone’ as defined by Orskov & Orskov (1983). They include a novel international clone (ST15). These clonal lineages have long been associated with multiple antibiotic resistances, and the selective advantage these confer appears to be driving their expansion (Diancourt et al., 2010). While the core genome of these lineages is stable, the accessory genome is more fluid, and carbapenem resistance among these isolates has become common, usually by acquisition of OXA carbapenemase genes (blaOXA-23-like, blaOXA-58-like, blaOXA-40-like, and blaOXA-143-like), although it may simply involve insertion sequence-mediated upregulation of the intrinsic blaOXA-51-like gene. These genes may be chromosomally or plasmid encoded (Poirel & Nordmann, 2006) (Higgins et al., 2010b) (Mugnier et al., 2010). There is a geographical variation in the distribution of acquired OXA genes, with blaOXA-58-like associated with Greece and Italy (Tsakris et al., 2008a; Di Popolo et al., 2011; Donnarumma et al., 2010), blaOXA-40-like with Spain and Portugal (da Silva et al., 2004; Quinteira et al., 2007; Ruiz et al., 2007) and blaOXA-23-like with Northern European countries, Asia and South America (Coelho et al., 2006; Lee et al., 2009; Mendes et al., 2009; Zarrilli et al., 2009; Fu et al., 2010). Acinetobacter baumannii isolates with VIM, IMP, SIM and NDM MBLs have also been reported (Lee et al., 2005; Poirel & Nordmann, 2006; Tsakris et al., 2006; Chiu et al., 2010; Karthikeyan et al., 2010). The ability of these clonal lineages to adapt to advances in antimicrobial therapy is likely to be a key factor in their success.

Pseudomonas aeruginosa

Pseudomonas aeruginosa is a key host species for genes encoding metallo-carbapenemases, particularly, but not limited to, VIM types. In contrast to the Enterobacteriaceae, carbapenem resistance is not unusual in this species. It is intrinsically resistant to ertapenem and strains with horizontally acquired carbapenemases, such as VIM types, must be distinguished against a larger background of isolates that are resistant to imipenem, meropenem and doripenem through mutations leading to loss of OprD porin (imipenem) or a combination of this mechanism with upregulation of efflux pumps, particularly MexAB-OprM (meropenem and doripenem).

An MLST scheme for P. aeruginosa was published in 2004 (Curran et al., 2004) and there are currently 972 defined STs ( The species has a nonclonal and epidemic structure, and recombination has played an important role in its evolution. Because of this, attempts to infer the relationships between all except the most related groups are essentially meaningless (Curran et al., 2004). There is a huge diversity of STs and, typically, a huge overlap between isolates from clinical and environmental sources (Wiehlmann et al., 2007). One clear exception to this generalization is the international, and frequently multiresistant, serotype O12 lineage (Pitt et al., 1989, 1990). Members of this epidemic lineage cluster very tightly when analysed by a variety of methods (Mifsud et al., 1997; Pirnay et al., 2009). This clone includes only clinical isolates and the evidence indicates that it emerged during the 1980s, perhaps selected by heavy antibiotic use (Pirnay et al., 2009). The O12 lineage falls into P. aeruginosa CC/BURST Group (BG) 4 – in context, a CC or BG comprises those STs that have at least five alleles in common. An eBURST representation ( of the relationships between P. aeruginosa STs reported to harbour MBLs or ESBLs is shown in Fig. 3.

Figure 3.

 eBURST analysis ( to show linked and unlinked STs for Pseudomonas aeruginosa isolates reported in the MLST database ( to produce an MBL or ESBL (SLVs are linked lines and DLVs enclosed within balloons).

The most prevalent serotypes of P. aeruginosa are O11, O12, O6 and O1, with O11 and O12 particularly common among multiresistant isolates (Pirnay et al., 2009). Giske et al. (2006) performed MLST on P. aeruginosa isolates with VIM MBLs from four European countries: Greece, Hungary, Italy and Sweden. All of the O12 isolates belonged to ST229 (BG 4), but included isolates with either VIM-2 or VIM-4 carbapenemase; these VIM MBL variants are phylogenetically distinct and represent separate acquisition events rather than evolution of the enzyme within the strain. In contrast to the relative uniformity shown by O12 isolates, those of serotype O11 show greater genetic diversity. Although the O11 isolates all belonged to BG 11, those from Italy produced the VIM-1 enzyme and belonged to ST227, whereas isolates from Greece, Hungary and Sweden produced the VIM-4 enzyme and belonged to the distinct, but related ST230 (Giske et al., 2006; Samuelsen et al., 2010).

More recently, Samuelsen et al. (2010) found that isolates of the international complexes CC111 (specifically, STs 111 and 229; serotype O12; blaVIM-2; corresponding to CC/BG 4) and CC235 (STs 235 and 230; serotype O11; blaVIM-4; CC/BG 11) accounted for 10 of 12 VIM-type MBL producers isolated in Norway and Sweden between 1999 and 2007. These included importations from Greece, Cyprus and Denmark, although six of the eight O12, CC111 isolates represented onward transmission in Sweden. The remaining two VIM MBL-producing isolates were distinct and were associated with repatriations from Ghana (ST233, serotype O6; blaVIM-2) – the first Norwegian MBL producer – and Tunisia (ST654; serotype O11; blaVIM-2) (Samuelsen et al., 2009a, 2010). The authors concluded that ‘both import of successful international clones and local clonal expansion contribute to the emergence of MBL-producing P. aeruginosa in Scandinavia’. These two CCs also account for many of the VIM MBL-producing isolates from Hungary (Libisch et al., 2008b), although other lineages, such as serotype O4, ST175 and serotype O6, ST395, also have a countrywide distribution (Libisch et al., 2009). Pseudomonas aeruginosa with VIM-2 MBL and belonging to ST175 also caused a recent outbreak of urinary tract infections in 11 patients in Germany (Elias et al., 2009). VIM-type MBLs have been identified in STs 155, 179 and 811 in Spain (

The predicted founder of the BG 11 is ST235 (Fig. 3), which has been found in many other countries, including Austria, Belgium, France, Greece, Hungary, Italy, Poland, Russia, Serbia, Singapore, Sweden and Turkey, in association not just with VIM enzymes but also with BEL-, IMP- OXA-, PER-, PSE- and SPM- β-lactamases (Empel et al., 2007; Shevchenko & Edelstein, 2007; Edalucci et al., 2008; Lepsanovic et al., 2008; Libisch et al., 2008a; Duljasz et al., 2009; Viedma et al., 2009; Glupczynski et al., 2010; Juan et al., 2010). Isolates belonging to a single PFGE-defined strain of ST235 accounted for 50% of multiresistant P. aeruginosa collected in a hospital in Madrid, and were susceptible only to polymyxins (Viedma et al., 2009). This strain harboured an integron with cassettes that encoded both the GES-1 ESBL and the GES-5 carbapenemase, together with blaOXA-2 and three aminoglycoside resistance genes: aac(6)-33, aacA4 and aadA1 (Viedma et al., 2009).

Pseudomonas aeruginosa isolates with IMP-type MBLs are geographically scattered and belong to diverse STs. Those included in the MLST database include STs 235 and 357 (IMP-1, Japan), 260 (IMP-14, Norway), 620 and 621 (IMP-22 and -13, respectively, Austria), 741, 742, 743 (IMP-1, Singapore), 744, 745 (IMP-7, Singapore) and also STs 235, 593 and 654 (unspecified IMP alleles, Brazil and Singapore) (Duljasz et al., 2009; Kouda et al., 2009; Samuelsen et al., 2010) (; Fig. 3).

The first SPM-1 MBL-producing isolate of P. aeruginosa was identified in Sao Paulo, Brazil, in the late 1990s (Toleman et al., 2002; Murphy et al., 2003). Isolates belonging to a single PFGE-defined strain have caused hospital outbreaks throughout Brazil (Gales et al., 2003; Zavascki et al., 2005; Carvalho et al., 2006). This strain typically is resistant to all antibiotics, except polymyxins (Carvalho et al., 2006; Fonseca et al., 2010), although rare isolates may be susceptible even to carbapenems (Pellegrino et al., 2008). It harbours an integron with aacA4, blaOXA-56 and aadA7 cassettes, and a putative transposase gene (Carvalho et al., 2006). The blaSPM-1 gene is not on this integron, and is probably located on large (c. 400 kb) plasmids. These are not transferable in vitro to E. coli or other P. aeruginosa recipients (Toleman et al., 2002; Poirel et al., 2004), but have achieved some horizontal spread to other strains of P. aeruginosa in the clinic (Castanheira et al., 2008), but not (yet) to other genera. Some isolates also produce RmtD, a 16S rRNA methyltransferase that confers broad aminoglycoside resistance (Doi et al., 2007a, b; Castanheira et al., 2008).

Most of the SPM-1 MBL-producing isolates, including the main PFGE-defined strain, belong to ST277 (Fonseca et al., 2010), which was previously identified from Austria and China, suggesting that it is likely to be a widespread international clone. It has three known SLVs, STs 364 (Austria), 659, 758 (China) and one DLV, ST 206 (Canada) ( It is worth noting that SPM-1 MBL has also been detected in ST235 (CC235; Producers of SPM-1 MBL are distinct from other multiresistant Brazilian P. aeruginosa isolates (Fonseca et al., 2010); these comparators, similarly, were susceptible only to polymyxins, but lacked SPM-1 and belonged to the unrelated ST244, which, together with its SLVs and DLVs, represents another widely distributed lineage, having also been identified in China, Poland and the United Kingdom (Empel et al., 2007) (

It is not known whether the non-Brazilian isolates of ST277 and related STs that are present in the MLST database also produced SPM-1 MBL. However, we are aware of only one report outside Brazil of a P. aeruginosa isolate with the SPM-1 enzyme (el Salabi et al., 2010), and this was from a patient who had been repatriated to Switzerland from a Brazilian hospital; the isolate was highly related by PFGE to SPM-1-positive comparators from Brazil. Given the level of international interest in carbapenemase-producing bacteria and the inclusion of primers for blaSPM-1 in MBL screening assays (Ellington et al., 2007; Mendes et al., 2007), it seems improbable that P. aeruginosa isolates with the SPM-1 enzyme are widespread globally, even if ST277 is prevalent. Rather, it seems likely that the acquisition of SPM-1 MBL is a feature limited to ST277 in Brazil, reflecting the escape of the blaSPM-1 gene from an unidentified species with subsequent national dissemination and a single example of international transfer.

Pseudomonas aeruginosa isolates from the lungs of cystic fibrosis (CF) patients are often resistant to multiple antibiotic classes, and in some cases are pan-resistant, including to polymyxins. Long-term exposure of isolates in the CF lung to antibiotics leads to the accumulation of mutations that confer highly complex phenotypes in strains with defective mismatch repair systems (Oliver et al., 2000). These mechanisms usually involve complex combinations of reduced porin expression and upregulation of multiple efflux pumps and intrinsic β-lactamases (Lister et al., 2009; Wolter et al., 2009a; Tomas et al., 2010). Carbapenem resistance is frequent, but there is, to our knowledge, only one report of MBL production by CF isolates with the VIM-2 enzyme in Portugal (Cardoso et al., 2008).

Although Curran et al. (2004) included many sputum isolates in their original MLST study; no CF-specific analyses were presented. The largest subsequent application of MLST specifically to CF isolates was undertaken by Waine et al. (2009), who found that most patients were colonized by isolates belonging to one (50%) or two (37.5%) STs, but up to four STs were present in some patients. Furthermore, these authors found that patients could be colonized simultaneously by unique STs and ‘epidemic’ STs, with the latter including ST146 (the Liverpool strain, widespread in the United Kingdom) and ST148 (the Midlands 1 strain) (Waine et al., 2009).

Pirnay et al. (2009) recently investigated the contentious issue of CF-specific clones and concluded that there was little evidence for them; multiresistant CF isolates from different parts of the world were genotypically diverse and unlikely to be directly related, but clustered into a core lineage, rather than a true clone. The authors could find no evidence for ‘widespread or global transmission of successful P. aeruginosa CF strains’ and concluded that strains belonging to the core lineage were ubiquitous in the environment. The frequency at which particular strains are isolated from CF patients is likely to reflect the prevalence of these strains in the local environment, and most patients will acquire the strains independently (Pirnay et al., 2009); certainly members of the Liverpool strain vary considerably in antibiogram among patients, ranging from fully susceptible to pan-resistant (Scott & Pitt, 2004).

Concluding remarks

Inferring the relatedness of bacterial isolates based on typing data is most straightforward when there is strong supportive epidemiological evidence (e.g. an outbreak situation). It is more difficult to interpret similarity (or even identity) with confidence in the absence of such evidence, and especially when isolates have seemingly disparate origins. It is easy to fall into the trap of assuming that isolates with similar DNA profiles must represent a transmission chain, although from different parts of the globe, but we now recognize that some Gram-negative bacterial species include globally distributed strains or clones and so we must be careful not to over-infer direct relationships where there may be none. Prevalence and/or wide geographic scatter may simply reflect a strain/clone's biological success and repeated acquisition of resistance through convergent evolution (Deschamps et al., 2009) rather than contemporaneous transmission events, although the latter must also occur.

When considering multiresistance, one should be alert to this possibility when representatives of a single clone may carry markedly different resistance genes; this probably indicates an ability to acquire whatever genes are locally prevalent, rather than direct spread with repeated loss and gain of resistance genes. Conversely, ‘recent’ dissemination is likely to have been a significant factor in epidemiological success when representatives of a particular clone are found widely in combination with a specific resistance gene(s), although even here it may not indicate direct transmission.

MLST is a definitive methodology that allows isolates anywhere to be typed and compared. However, it is expensive and many workers seek to introduce cheaper alternatives, such as VNTR/MLVA, to determine information about relatedness. Unless the same next-generation approach is widely adopted, there is potential for contrasting methodologies to make interstudy comparisons (including with existing MLST data) more difficult or impossible. This is a cause for concern and, in our opinion, would be a retrograde step. Clearly, we need to embrace alternative methods if they are faster, cheaper than and as powerful as MLST, but we must establish a comprehensive ‘dictionary’ to ‘translate’ the clonal group designations of different schema. There is a similar requirement where multiple MLST schemes exist for a particular species; this has been at least partially addressed in some cases, for example, for E. coli (Jaureguy et al., 2008).

There are numerous examples in the literature of hospital and community outbreaks, in some instances substantial, caused by MDR Gram-negatives for which there has been extensive molecular investigation, but for which there are no MLST data. Hence, although local purposes are served, much add-on value of potential national or international importance is lost. We would urge laboratories to undertake MLST on representatives of all outbreak strains and to use it in parallel with any other current or newly introduced method. Only by doing this can we generate a more complete picture and fuller understanding of the importance of high-risk clones in the international dissemination of antibiotic resistance.


No external funding was provided for this work. D.M.L. has received conference support from numerous pharmaceutical companies. He also holds shares in AstraZeneca, Merck, Pfizer, Dechra and GlaxoSmithKline and, as Executor, manages further holdings in GlaxoSmithKline and Eco Animal Health. N.W. has received conference support and has spoken at symposia sponsored by many pharmaceutical companies. None of these declared interests poses a conflict with the contents of this review. J.F.T., none to declare.