Outbreak of Pseudomonas aeruginosa Infections with PER-1 Extended-Spectrum β-Lactamase in Warsaw, Poland: Further Evidence for an International Clonal Complex

Abstract

Forty-one Pseudomonas aeruginosa isolates with extended-spectrum β-lactamases (ESBLs) from a hospital in Warsaw, Poland, were analyzed. Thirty-seven isolates from several wards were collected over 9 months in 2003 and 2004. The isolates were recovered from patients with multiple types of infections, mostly respiratory tract and postoperative wound infections. All 41 isolates produced the PER-1 ESBL, originally observed in Turkey but recently also identified in several countries in Europe and the Far East. The bla_PER-1 gene resided within the Tn1213 composite transposon, which was chromosomally located. Pulsed-field gel electrophoresis and multilocus sequence typing (MLST) revealed the presence of three separate clones among the isolates. Two of these, corresponding to sequence types (STs) ST244 and ST235, were responsible for parallel outbreaks. Apart from PER-1, all the isolates produced OXA-2 oxacillinase. ST235 isolates additionally expressed a novel enzyme, OXA-74, differing by one amino acid from the OXA-17 ESBL identified originally in PER-1- and OXA-2-positive P. aeruginosa isolates from Ankara, Turkey, in 1992. These earlier Ankara isolates with PER-1, OXA-2, and OXA-17 were also classified into ST235, which is a single-locus variant of two other STs, ST227 and ST230. ST227, ST230, and ST235 all correspond to the recently described clonal complex BG11, which seems to be internationally distributed, having spread in Turkey, Greece, Italy, Hungary, Poland, Sweden, and much of Russia. It is associated with various β-lactamases, including PER-1 and VIM metalloenzymes. This work further demonstrates the value of MLST of P. aeruginosa.

Pseudomonas aeruginosa is one of the most important nosocomial pathogens, being responsible for various types of infections with more and more limited therapeutic options. This situation is due in large part to the rapid accumulation of antimicrobial resistance mechanisms in P. aeruginosa, including extended-spectrum β-lactamases (ESBLs) that confer resistance to oxyimino-β-lactams. PER (Pseudomonas extended resistant) β-lactamases are one of the rarer ESBL families; however, their prevalence may be increasing. Two PER types have been described previously. PER-1 was identified first, in a P. aeruginosa isolate from 1991 recovered in France from a Turkish patient (36, 37). Later, PER-1 was recognized to be widespread in Turkey among strains of P. aeruginosa, Acinetobacter spp., and Enterobacteriaceae and is the most prevalent ESBL in the first of these species (1, 2, 12, 24, 48, 49, 50, 51). PER-2 production among Enterobacteriaceae from South America, mostly Argentina, is known to occur (4, 5, 44, 52).

Since 1995, PER-1-producing organisms have been disseminating in Italy (15, 30, 38, 39, 40) and, more recently, in Belgium (6, 34), France (13, 14, 43), Spain (33), Romania (35), Hungary and Serbia (27, 46), Korea (23, 55), Japan (54), and China (20). In Europe, the bla_PER-1 gene is found mostly in P. aeruginosa, where it resides within a specific transposon, named Tn1213 by Poirel et al. (42) and Tn4176 by Mantengoli and Rossolini (31). This element is formed by two different, novel insertion sequences of the IS4 family with almost identical inverted repeats. The expression of PER-1 (and that of PER-2) usually confers clear resistance to oxyimino-β-lactams, especially ceftazidime, ceftibuten, and aztreonam (4, 12, 37, 38).

Although frequent in Enterobacteriaceae, ESBLs in P. aeruginosa strains in Poland are very rare, with only a few documented identifications of ESBL-producing P. aeruginosa isolates (26, 53). In this paper, the first identification of PER-1-producing P. aeruginosa isolates in Poland is described and the relationship of the isolates to earlier producer strains from Turkey is explored.

MATERIALS AND METHODS

Clinical isolates.

We examined 41 clinical isolates of P. aeruginosa from one hospital in Warsaw, Poland (Table 1), a large (∼1,000-bed), tertiary-care medical center treating patients from all over Poland, as well as foreign residents and travelers to Warsaw. Thirty-seven isolates were collected during an apparent outbreak of P. aeruginosa infections between September 2003 and May 2004; the remaining four were recovered from patients with sporadic infections between July 2005 and January 2006. Over the study period, all the isolates were identified by the hospital microbiology laboratory as nonreplicate P. aeruginosa putative ESBL producers by using the double-disk synergy test (22). The isolates were from patients from several wards, mostly those in the intensive care unit (ICU; 21 isolates) and the surgery ward (10 isolates), and were mainly from tracheostomy tube swabs (14 isolates), postoperative wound swabs (12 isolates), and urine (6 isolates). Of all the affected patients, 11, mostly those in the ICU, died; however, all these patients had severe underlying diseases (e.g., cancer, diabetes, circulatory failure, and chronic renal failure), and only in one case could the death be unambiguously attributed to bacterial infection. Other patients were successfully treated according to susceptibility data with either piperacillin-tazobactam or imipenem. Infection control procedures included mostly personnel education and strict contact precautions. The microbiological investigation of the ICU environment did not identify any materials contaminated with infecting organisms, and a personnel carriage investigation was not performed.

TABLE 1.

Clinical data, PFGE types, MLST results, and β-lactamases corresponding to the P. aeruginosa isolates studied

PFGE type (no. of isolates)	Date(s) (mo/yr) of isolation	Source(s)	Hospital ward(s) or location	MLSTa		β-Lactamase pIs	β-Lactamase variantsb
				Allelic profile	ST
A, subtypes 1 to 5 (37)	9/2003-1/2006	Tracheostomy tube swabs, wound swabs, urine, blood, peritoneal fluid, drainage fluid, ear swabs	ICU, surgery, neurology, cardiology, internal medicine	17-5-12-3-14-4-7	ST244	5.3, 7.7, 8.2	PER-1, OXA-2, OXA-50f
B, subtypes 1 and 2 (3)	2/2004-3/2004	Urine, peritoneal fluid, ear swab	Surgery, neurology, nephrology	38-11-3-13-1-2-4	ST235	5.3, 6.5, 7.7, 8.2	PER-1, OXA-2, OXA-74, OXA-50g
Cc (1)	7/2005	Tracheostomy tube swab	Neurology	39-6-12-11-3-15-2	ST245	5.3, 7.7, 8.2	PER-1, OXA-2, OXA-50h
D (2)	2/1992-3/1992		Ankara, Turkey	38-11-3-13-1-2-4	ST235	5.3, 6.5, 7.7, 8.2	PER-1,c OXA-2,d OXA-17,d OXA-50g

^aMLST of eight ST244 isolates, three ST235 isolates, one ST245 isolate, and two Turkish isolates was performed.

^bPCR analyses of the bla_PER-1-, bla_OXA-2-, bla_OXA-10-, and bla_OXA-50-like genes of all isolates were performed; full sets of amplified genes from three representative ST244 isolates and all ST235 and ST245 isolates from Warsaw were sequenced. Only the bla_OXA-50-like genes from the Turkish isolates were sequenced in the present study; the other β-lactamase genes of these isolates were characterized previously (11, 12).

^cReference 12.

^dReference 11.

The isolates were sent to the National Medicines Institute in Warsaw for further examinations. They were reidentified by the ATB ID32GN test (bioMérieux, Charbonnieres-les-Bains, France) and confirmed as ESBL producers by the double-disk test (22) using ceftazidime (30 μg), cefotaxime (30 μg), and amoxicillin and clavulanate (20 and 10 μg) disks (Oxoid, Basingstoke, United Kingdom).

P. aeruginosa isolates 871 and 873, producing PER-1, OXA-2, and OXA-17 β-lactamases, were identified in 1992 in a hospital in Ankara, Turkey (11, 12), and were included in this work for comparison. One study isolate, 2622/03, was used previously to partially characterize the genetic context of bla_PER-1 (42).

Antimicrobial susceptibility testing.

MICs were evaluated by the Clinical and Laboratory Standards Institute (CLSI) agar dilution method (7) using aztreonam and cefepime (Bristol-Myers Squibb, New Brunswick, NJ); cefotaxime and gentamicin (Polfa Tarchomin, Warsaw, Poland); ceftazidime, ticarcillin, and clavulanate (GlaxoSmithKline, Stevenage, United Kingdom); ciprofloxacin (Bayer AG, Leverkusen, Germany); imipenem (Merck, Rahway, NJ); meropenem (AstraZeneca, Macclesfield, United Kingdom); piperacillin-tazobactam (Wyeth, Pearl River, NY); and tobramycin (ICN Biomedicals, Eschwege, Germany). Clavulanate and tazobactam concentrations in combination tests were fixed at 2 and 4 μg/ml, respectively. P. aeruginosa ATCC 27853, Escherichia coli ATCC 25922, and E. coli ATCC 35218 were used as reference strains in susceptibility testing.

PFGE.

Pulsed-field gel electrophoresis (PFGE) typing was performed as described by Grundmann et al. (19) or, in the case of isolates with rapidly degrading DNA, as described by Römling and Tümmler (45). DNA was cut using the XbaI restriction enzyme (MBI Fermentas, Vilnius, Lithuania) and electrophoresed in a contour-clamped homogenous electric field DRIII PFGE system (Bio-Rad, Hercules, CA). The results were interpreted according to the criteria of Tenover et al. (47).

MLST.

Multilocus sequence typing (MLST) was performed as proposed by Curran et al. (9). Seven internal fragments of the genes acsA, aroE, guaA, mutL, nuoD, ppsA, and trpE were amplified, and the resulting PCR products were purified with the QIAquick PCR purification kit (QIAGEN, Hilden, Germany). Sequencing was carried out with primers as corrected at the http://pubmlst.org website by using an ABI 310 automatic sequencer (Applied Biosystems, Foster City, CA). Sequences were edited and assembled with the SEQEDV 10.3 software (Applied Biosystems). The database at http://pubmlst.org was used to assign numbers to particular alleles and to identify sequence types (STs).

IEF of β-lactamases.

β-Lactamases were profiled by isoelectric focusing (IEF) as described by Bauernfeind et al. (3) by using a model 111 mini IEF cell (Bio-Rad). Gels were run at pHs over the range of 3 to 10 and were developed with 0.5 mM nitrocefin (Oxoid) or with 0.5 mM nitrocefin plus 0.3 mM cloxacillin (Polfa Tarchomin).

PCR detection and sequencing of β-lactamase (bla) genes.

Entire protein-encoding regions of all bla genes were amplified using the primers shown in Table 2. DNA sequencing was performed as described above, with the appropriate primers presented in Table 2.

TABLE 2.

PCR primers used to analyze β-lactamase genes

Primer designation	Target gene or region	Sequence (5′-3′)	Purpose(s)	Expected size(s) (bp) of amplicon(s) (corresponding primer)a	Reference
PERA	blaPER-1	ATGAATGTCATTATAAAAGC	PCR, sequencing	925 (PERD)	12
PERD	blaPER-1	AATTTGGGCTTAGGGCAGAA	PCR, sequencing		12
S	blaOXA-50 type	AATCCGGCGCTCATCCATC	PCR, sequencing	867 (AS)	16
AS	blaOXA-50 type	GGTCGGCGACTGAGGCGG	PCR, sequencing		16
OXA-2A	blaOXA-2	ATGGCAATCCGAATCTTCG	PCR, sequencing	828 (OXA-2B)	This work
OXA-2B	blaOXA-2	TTATCGCGCAGCGTCCG	PCR, sequencing		This work
OXA-2C	blaOXA-2	CTTGAATGTCGATGCAGGC	Sequencing		This work
OXA-2D	blaOXA-2	GGTCGCAACTGGATACTGC	Sequencing		This work
OXA-10A	blaOXA-10 type (blaOXA-74)	ATGAAAACATTTGCCGCATATG	PCR, sequencing	801 (OXA-10B)	This work
OXA-10B	blaOXA-10 type (blaOXA-74)	TTAGCCACCAATGATGCCC	PCR, sequencing		This work
OXA-10C	blaOXA-10 type (blaOXA-74)	CCTGATGCTCATTCTTTATG	Sequencing		This work
OXA-10D	blaOXA-10 type (blaOXA-74)	CACCTGAATATCTAGTGCA	Sequencing		This work
ISPa12.B		GATCTCGCTTTACATTTACC	PCR	1,678 (PER-1E)	42
ISPa14.B		GCCTAATTCGATGCCTTAT	PCR	1,993 (PER-1E)	42
PER-1E	blaPER-1	GCACTGGAACACTAAACTCG	PCR		This work
PERC	blaPER-1	ACACAGCTGTCTGAAACCTC	PCR	1,883 (ISPa13.A); 2,528 (ISPa14.A)	12
ISPa13.A		TAACCATATGCACTCAACGG	PCR		42
ISPa14.A		AATCAAATGTCCAACCTGCC	PCR		42

^aPCR product sizes are shown in conjunction with the forward primer in each primer pair; the corresponding primer in the pair is identified in parentheses.

Context and location of bla genes.

The presence of bla_PER-1 genes within Tn1213 (42) was checked by separate PCR analyses of two overlapping parts of the element; Table 2 shows the sequences of the primers used. The 5′ part of the element was amplified with Tn1213-specific forward primer ISPa12.B or ISPa14.B and the bla_PER-1-specific reverse primer PER-1E, whereas the 3′ part was amplified with the forward primer PERC and the reverse primer ISPa13.A or ISPa14.A. The location of bla_PER-1 genes in chromosomal or plasmid DNA was studied by the I-CeuI technique (28) with probes specific for rRNA genes and bla_PER-1. The bla_PER-1 probe was a ∼0.9-kb PCR product obtained with primers PERA and PERD using total DNA of P. aeruginosa 2623/03 as a template. The rRNA gene probe was a mixture of PCR amplicons encompassing fragments of 16S and 23S rRNA genes of E. coli, obtained as described previously (18). I-CeuI-digested and PFGE-separated DNA of P. aeruginosa isolates was blotted onto a Hybond-N+ membrane (Amersham Pharmacia Biotech, Little Chalfont, United Kingdom). Probe labeling, hybridization, and signal detection were performed with the ECL random-prime labeling and detection system (Amersham).

Nucleotide sequence accession numbers.

The nucleotide sequences analyzed in this work were deposited in the EMBL database under the following accession numbers: AJ854182 for bla_OXA-74, AM117126 for bla_OXA-50f, AM117127 for bla_OXA-50g, and AM117128 for bla_OXA-50h.

RESULTS

PFGE and MLST.

PFGE revealed three separate clusters of isolates (Table 1). One of these, type A, was predominant, corresponding to 34 isolates from the main collection period and three isolates from patients with sporadic infections; this cluster split into five subtypes. Three other outbreak isolates were classified into PFGE type B, which was differentiated into two subtypes. The remaining sporadic-infection isolate was classified into PFGE type C. No correlation between the PFGE type and the hospital ward or the type of infection was found.

Twelve representative isolates were analyzed by MLST. The results correlated well with the PFGE data (Table 1). Isolates of PFGE type A were classified into ST244, and isolates of PFGE type B corresponded to ST235, whereas the PFGE type C isolate was classified into ST245. All the MLST alleles found had known sequences; however, ST244 and ST245 represented novel combinations of alleles.

β-Lactamase content.

IEF analyses of crude sonication products from all the study isolates were performed; results are shown in Table 1. All the ESBL-positive isolates expressed β-lactamases with pIs of 8.2, 7.7, and 5.3. Additionally, the three ST235 isolates produced pI 6.5 enzymes. The pI 8.2 β-lactamases ceased to be visualized when the nitrocefin was supplemented with cloxacillin, indicating these to be P. aeruginosa-specific AmpC cephalosporinases.

PCR detection and sequencing of bla genes.

The genomic DNA of each isolate was tested by PCR for the presence of bla_PER-1- and bla_OXA-2-type genes, which might encode the pI 5.3 and pI 7.7 β-lactamases, respectively. Amplicons of the expected sizes of ∼0.9 and ∼0.8 kb, respectively, were obtained in all cases. The three ST235 isolates additionally were positive for bla_OXA-10-type genes, possibly coding for the pI 6.5 enzymes and yielding amplicons of the expected size of ∼0.8 kb. Finally, all the isolates were positive for the bla_OXA-50-type (poxB) genes, recently found to be characteristic of P. aeruginosa (16, 25), giving PCR products of the expected size (∼0.85 kb).

Three ST244 isolates and all the ST235 and ST245 isolates were selected for sequencing of the entire coding regions of their amplified bla genes. Sequence analysis confirmed the presence of classical bla_PER-1 and bla_OXA-2 in all these isolates, with sequences identical to those described originally (10, 37). The bla_OXA-10-type genes of the ST235 isolates specified a novel β-lactamase variant, now designated OXA-74 (http://www.lahey.org/studies/webt.htm). bla_OXA-74 differed by two mutations, C197→T and A218→G (numbered from the first ATG codon of the coding region), from bla_OXA-10 (21). These mutations cause two amino acid substitutions in OXA-74, Ala69Val and Asn76Ser (numbered according to the class D β-lactamase scheme [8]), respectively, compared to OXA-10. The Ala69Val substitution differentiates OXA-74 from OXA-17, identified in P. aeruginosa isolates from Turkey that also produced PER-1 and OXA-2 (11, 12).

Three allelic variants of bla_OXA-50 specific to the three STs were found (Table 3). A comparison of the sequences of the bla_OXA-50-type alleles with those available in databases (16, 25) revealed that these sequences were unique. bla_OXA-50f alleles from the ST244 isolates and bla_OXA-50h from the ST245 isolate differed by five and four nucleotides, respectively, from the bla_OXA-50 gene of P. aeruginosa PAO1 (16), and their closest relative was poxB27, previously identified in the P. aeruginosa strain DSM 1253 of unreported origin (25). The bla_OXA-50g allele of the ST235 isolates was much different from the other bla_OXA-50 alleles, with 8, 13, and 12 polymorphisms compared with bla_OXA-50, bla_OXA-50f, and bla_OXA-50h, respectively. It differed by only a single silent mutation from bla_OXA-50b, identified in P. aeruginosa COL-1 from France (16).

TABLE 3.

Nucleotide sequences of identified bla_OXA-50 alleles in the P. aeruginosa isolates studied

Strain or ST/allele	Nucleotide at positiona:
	9	21	36	46	66	74	96	105	126	145	282	289	327	492	500
PAO1/blaOXA-50	T	T	T	A (Thr)	C	A (Gln)	C	G	C	C (Arg)	G	C (Arg)	C (Asp)	A	G (Arg)
DM1253/poxB27b	C									T (Cys)			G (Glu)		A (His)
COL-1/blaOXA-50bc		C	C	G (Ala)		G (Arg)	A	T	T		A			G
ST244/blaOXA-50f	C									T (Cys)		T (Cys)	G (Glu)		A (His)
ST235/blaOXA-50g		C	C	G (Ala)		G (Arg)	A	T	T		A
ST245/blaOXA-50h	C				T								G (Glu)		A (His)

^aNucleotide positions are numbered consecutively, starting with the first ATG codon of the bla_OXA-50 coding sequence. Only nucleotide differences with respect to the sequence of the bla_OXA-50 gene from P. aeruginosa PAO1 (GenBank accession no. AE004964) (16) are shown. In cases in which polymorphisms cause amino acid differences, the amino acid symbols are given in parentheses.

^bGenBank accession no. AY597445; reference 25.

^cGenBank accession no. AY306131; reference 16.

Comparative typing of ST235 isolates and P. aeruginosa isolates from Turkey.

The similar complexities of the β-lactamase contents of the ST235 (PFGE type B) isolates, producing PER-1, OXA-2, and OXA-74, and two P. aeruginosa isolates identified in 1992 in Ankara, Turkey, producing PER-1, OXA-2, and OXA-17 (11, 12), suggested possible relatedness. The two Ankara isolates, therefore, were compared with isolates from Warsaw by PFGE, MLST, and sequencing of bla_OXA-50-type genes. In the PFGE analysis, the two Turkish isolates were classified into a separate type, type D, but their DNA banding patterns showed clear similarity to those of PFGE type B (data not shown). Moreover, the isolates from Ankara corresponded to ST235 by MLST and carried the bla_OXA-50g gene variant, which revealed a fundamental relationship.

Location of bla genes.

A previous study of a PFGE type A isolate showed that its bla_PER-1 gene was chromosomally located, inside a Tn1213 composite transposon (42). PCRs demonstrated that this genetic context of bla_PER-1 was conserved in all of the Warsaw clones, irrespective of ST and PFGE type, and in the isolates from Turkey (Table 1). I-CeuI analysis, carried out with several isolates representing Warsaw ST244 and ST235 isolates and with the two Turkish isolates, confirmed the chromosomal location of bla_PER-1 (data not shown).

Susceptibility testing.

Most of the study isolates were resistant to multiple drugs, including β-lactams, aminoglycosides, and ciprofloxacin (Table 4). All but seven isolates were, however, susceptible to piperacillin by CLSI (though not European) criteria, and all but one were susceptible to piperacillin-tazobactam on the same basis. Seventeen isolates were intermediately susceptible or resistant to imipenem, and 27 isolates were intermediately susceptible or resistant to meropenem.

TABLE 4.

Antimicrobial susceptibilities of clinical isolates^a

Strain or ST (no. of isolates)	MIC (μg/ml) of:
	TIC	TIM	PIP	TZP	CTX	CAZ	CAZ-CLA	FEP	ATM	IPM	MEM	GEN	TOB	CIP
ST244 (37)	256->512	32-256	16->256	8->256	128->256	>256	4-128	32->256	64->256	0.5->32	2->32	8->64	2->64	>32
ST235 (3)	512	256	64	64	128	128->128	16-32	64-128	128->128	8	16	>64	>64	16->16
ST245 (1)	512	32	32	16	>256	>256	32	256	>256	4	4	4	2	1
P. aeruginosa ATCC 27853	16	16	2	2	16	1	2	2	2	2	0.5	2	0.5	0.25
E. coli ATCC 25922	4	8	2	2	0.06	0.25	0.125	0.06	0.06	0.06	0.015	0.5	0.25	0.015

^aAbbreviations: TIC, ticarcillin; TIM, ticarcillin with clavulanate; PIP, piperacillin; TZP, piperacillin-tazobactam; CTX, cefotaxime; CAZ, ceftazidime; CLA, clavulanate; FEP, cefepime; ATM, aztreonam; IPM, imipenem; MEM, meropenem; GEN, gentamicin; TOB, tobramycin; CIP, ciprofloxacin.

DISCUSSION

This study is the first report of PER-1-producing microorganisms in Poland. For years, PER β-lactamases were thought to be significant only in Turkey (PER-1) and Argentina (PER-2) (4, 12, 32, 41, 50, 51). However, the recent identification of PER-1 producers in several European countries and in the Far East suggests their proceeding dissemination (20, 27, 33, 35, 46, 54). In Korea, as in Turkey, PER-1 production by Acinetobacter spp. has been reported often, but in Europe, it has been identified mostly in P. aeruginosa.

With 37 patients in several wards affected between September 2003 and May 2004, the outbreak described in the present study is one of the largest caused by P. aeruginosa with PER-1 to be described so far. Molecular typing revealed two clones among the outbreak isolates, with one of these, ST244 (PFGE type A), being greatly predominant. The clonal dissemination of Acinetobacter baumannii, P. aeruginosa, or Proteus mirabilis strains with PER-1 in hospitals in Korea, Turkey, Italy, and Belgium has been reported previously, but the corresponding outbreaks were limited and restricted to ICUs (6, 23, 30, 39, 51). The concurrent presence of two or more unrelated clones of A. baumannii or P. aeruginosa with the PER-1 enzyme in some Korean, Turkish, and Italian centers has been reported previously (1, 38, 55).

In all of the study isolates, bla_PER-1 resided in the same transposon, Tn1213 (Tn4176), as in many producers from Turkey, Italy, France, and Belgium (31, 42). The constant genetic environment of bla_PER-1 in organisms from multiple countries implies a common origin and horizontal transmission of the gene, even though it often occurs in chromosomal DNA (42). PER-1 expression was inferred to be the major source of oxyimino-β-lactam resistance in the study isolates, although they also produced other β-lactamases. The ST235 isolates expressed a novel oxacillinase, OXA-74, which differed by one amino acid only from the OXA-17 β-lactamase from P. aeruginosa isolates with PER-1 obtained in Ankara in 1992 (12). OXA-17 has ESBL activity conferring cefotaxime resistance (11), and it is possible that OXA-74 does as well. The ST235 isolates did not differ significantly from ST244 isolates (without OXA-74) in resistance levels; however, the β-lactam MICs for the ST235 isolates were more often at the upper end of the MIC ranges.

It is difficult to identify the origin and molecular epidemiology of PER-1-producing P. aeruginosa strains in Warsaw. Genetic similarity in terms of β-lactamase content, ST, and the bla_OXA-50 allele demonstrated the relatedness of the minor outbreak clone, ST235, to two Ankara isolates from 1992 (11, 12). One may therefore propose the import of ST235 from Turkey and the transfer of bla_PER-1 to ST244, responsible for the main outbreak, and later to ST245. However, the first of the Warsaw ST235 isolates was recovered in February 2004, 6 months after the first ST244 isolate. It is entirely possible that the ST235 strain had been present but undetected before the main ST244 infection outbreak; nevertheless, the origins of P. aeruginosa strains with the PER-1 enzyme in the hospital remain unclear.

The recently introduced P. aeruginosa MLST scheme (9) has permitted interesting epidemiological analyses. Apart from the Warsaw (2004) and Ankara (1992) isolates examined here, ST235 isolates have been identified as producers of VIM-2 or VIM-4 metallo-β-lactamases circulating since 2001 in northern Poland (J. Empel and M. Gniadkowski, unpublished results). ST235 also has been found in three distant Russian cities (Moscow, Omsk, and Krasnodar; 2002 to 2004), though the β-lactamase content of these isolates was not reported (http://pubmlst.org). Moreover, two single-locus variants (SLVs) of ST235, namely, ST227 (with VIM-1) and ST230 (with VIM-4), have recently been described in Italy (where such isolates were obtained between 1997 and 2003) and in Greece and Sweden (where such isolates were obtained between 2001 and 2002), respectively. ST227 and ST230 have been classified as clonal complex BG11 (17), and being an SLV of these lineages, ST235 provides a link between them and also belongs to the BG11 complex. Most probably, BG11 is a truly international complex that has been disseminating over a long period and a wide area, acquiring various resistance determinants in particular locales. By contrast, of the two other STs identified here, ST245 is a singleton among the STs so far reported, whereas ST244 has an SLV (ST336) but one on which no clinical data have yet been presented (http://pubmlst.org). ST244 and ST245 have not yet been reported elsewhere. It is noteworthy, finally, that the distribution of STs perfectly matched that of the bla_OXA-50 alleles seen, confirming the validity of the MLST scheme and suggesting that bla_OXA-50 may be another potential clonality marker for P. aeruginosa.

Along with other recent studies, this work illustrates the ongoing dissemination of organisms with PER-type ESBLs. Only time will show whether these ESBLs can mimic the recent and dramatic spread of CTX-M-type enzymes (29). This study is also a good example to demonstrate the value of the wider implementation of MLST in order to identify and track the spread of multidrug-resistant P. aeruginosa clones.