|Home | About | Journals | Submit | Contact Us | Français|
The Liverpool epidemic strain (LES) of Pseudomonas aeruginosa is widespread among cystic fibrosis (CF) patients in the United Kingdom and has emerged recently in North America. In this study, we report the analysis of 24 “anomalous” CF isolates of P. aeruginosa that produced inconsistent results with regard to either pulsed-field gel electrophoresis (PFGE) or PCR tests for the LES. We used a new typing method, the ArrayTube genotyping system, to determine that of the 24 anomalous isolates tested, 13 were confirmed as the LES. LES isolates could not be clearly distinguished from non-LES isolates by two other commonly used genetic fingerprinting tests, randomly amplified polymorphic DNA (RAPD) analysis and BOX-PCR, and varied considerably in their carriage of LES genomic islands and prophages. The genomic instability of the LES suggests that identification of this emerging transmissible strain could be a challenging task, and it questions whether discrimination is always a desirable feature of bacterial typing methods in the context of chronic CF infections.
Chronic Pseudomonas aeruginosa infections in cystic fibrosis (CF) patients are associated with a decline in lung function and increased mortality (11). Infection usually occurs during adolescence and, once established, the pathogen is impossible to eradicate (17). In recent years, the view of P. aeruginosa infection in CF patients has been altered because of the emergence of transmissible strains. Originally, it was thought that patients acquired their own unique strains from the environment and only in specific circumstances, such as siblings with CF, were patients found to share the same strain (35). However, in 1996 Cheng et al. (6) described the use of molecular typing to demonstrate the spread of a drug-resistant strain of P. aeruginosa (named the Liverpool epidemic strain [LES]) among patients in a children's CF center in Liverpool, United Kingdom (6). Seven years later, an analysis of patient samples taken after the year 2000 identified the LES in 79% of 80 P. aeruginosa-colonized CF patients in the Liverpool adult CF center, confirming the spread and longevity of LES infections (29). In a survey of 31 CF centers in England and Wales in 2004, involving more than 1,200 isolates of P. aeruginosa, the LES was identified as the most common clone (33). Furthermore, the LES has also been identified in Scotland (10). Recently, cases of CF patients infected with the LES have also emerged in Canada (1).
In addition to its transmissibility, the LES appears to be more aggressive than other P. aeruginosa strains. It has been shown to replace previously established strains of P. aeruginosa (superinfection) (24), has infected both non-CF parents of a CF patient (25), and has infected a pet cat, causing significant morbidity (26). Furthermore, the LES is associated with greater morbidity (2) and increased renal failure (3) and appears to have enhanced survival in air (30).
Patient segregation based on LES status requires simple but effective strain typing methods with adequate discriminatory powers to achieve both the initial identification of the LES and subsequent epidemiological surveillance. The genome of P. aeruginosa is a mosaic structure consisting of a core genome and a variable accessory genome (23). The accessory genome includes large insertions, such as prophages and genomic islands, contributing to genome size variations between 5.2 and 7 Mb (38). The earliest archived isolate of the LES (LESB58, from 1988) has recently been genome sequenced (40), revealing the presence of six prophages and five genomic islands.
Specific PCR assays have been developed for detection of the LES (31, 34). Following reports of false positives for the original marker PS21 (20), a second marker (LES-F9) was identified (34). Neither PS21 nor LES-F9, which map to separate genomic islands (LESGI-3 and LESGI-1, respectively), is 100% specific to the LES, but the combination of the two has not been reported previously in any non-LES strain. The use of PCR assays in routine diagnostic tests in Liverpool and elsewhere in the United Kingdom has led to the implementation of successful segregation measures. However, we have a number of putative LES CF isolates that do not give concordant results with respect to the presence of PS21, the presence of LES-F9, and the molecular typing method pulsed-field gel electrophoresis (PFGE). The Clondiag ArrayTube (AT) system is a portable method for interrogating both the conserved and accessory genomes of P. aeruginosa isolates. The process is rapid, relatively inexpensive, and robust and has been used to type P. aeruginosa isolates from diverse habitats (39).
Here, we report the use of the AT system to resolve the strain status of a collection of suspected LES isolates of P. aeruginosa from CF patients. We provide evidence that the genomic instability exhibited by the LES could impact the validity of routine typing schemes, causing both false-positive and false-negative identification, and suggest that normal paradigms for bacterial typing may be limited in the context of chronic infections in CF.
The P. aeruginosa isolates used in this study were obtained from patients with CF attending three United Kingdom hospitals, two in the Northwest (Liverpool Heart and Chest Hospital and Clatterbridge Hospital) (n = 16) and also Papworth Hospital in Cambridge (n = 4). Additional isolates were obtained from the Health Protection Agency (HPA), Colindale, London (n = 4) (Table (Table1).1). All of the isolates had produced at least one negative result with respect to three tests carried out on CF P. aeruginosa isolates for the identification of the LES: PFGE and assays for the markers PS21 (31) and LES-F9 (34). Isolate LESB58 was used as a positive control for the LES throughout.
For each amplification, 1.25 U GoTaq polymerase (Promega), 300 nM each oligonucleotide primer (Sigma-Genosys) (Table (Table2),2), 1× Taq buffer, 2.5 mM MgCl2, and 100 μM nucleotides (dATP, dCTP, dGTP, dTTP) were used with 1 μl DNA from boiled suspensions. PCR amplification (using an Eppendorf Mastercycler) was performed for 30 cycles at 95°C (1 min), at the chosen annealing temperature (2 min), and at 72°C (2 min). A final extension step of 72°C for 10 min was carried out. In the case of LES prophages 2, 3, and 4 and LESGI-5, additional PCR assays were performed using primers designed for the flanking regions of the insertions, and therefore only if the insertion was not present would the target band be observed.
PFGE using SpeI was performed on both LES and non-LES isolates as described previously (19), using isolate LESB58 (40) as the LES control isolate. Electrophoresis was performed (120° angle and 6 V/cm with a pulse ramp of 5 to 33 s for 20 h) alongside a size marker (Sigma pulse marker, 50 to 1,000 kb). The banding pattern produced from the PFGE was analyzed using GelCompar II software and a dendrogram (Dice/unweighted pair-group method using average linkages [UPGMA]) produced to visualize the relatedness of the isolates. For three isolates (079314, WM1, and 444), we were unable to obtain adequate PFGE profiles despite using 50 μM thiourea, increasing proteinase K incubation from 18 h to 36 h, and increasing SpeI incubation time from 18 h to 24 h (32).
Randomly amplified polymorphic DNA (RAPD) PCR fingerprinting was adapted from the method of Mahenthiralingam et al. (22). Briefly, PCR analyses consisted of the following (with all reagents supplied by Qiagen): 1× PCR buffer, 1× Q-solution, 3 mM MgCl2, 200 μM deoxynucleoside triphosphates (dNTPs), 1.6 μM primer 270 (5′-TGCGCGCGGG-3′), 1 U Taq DNA polymerase, and 2 μl template DNA. Thermal cycling was carried out on a Flexigene thermal cycler as follows: 5 min at 94°C, followed by 4 cycles of 5 min at 36°C, 5 min at 72°C, and 5 min at 94°C; and a further 30 cycles of 1 min at 94°C, 1 min at 36°C, and 2 min at 72°C, followed by a 10-min hold at 72°C. BOX-PCR strain typing was adapted from the method of Coenye et al. (7) and carried out in a Flexigene thermal cycler with BOX-A1R primer (5′-CTACGGCAAGGCGACGCTGACG-3′). Each reaction mixture contained 1× PCR buffer, 1× Q-solution, 1.75 μM primer, 240 μM each dNTP, 8 μg bovine serum albumin, 2.5 μl dimethyl sulfoxide (DMSO), 2 U Taq polymerase, and 2 μl template DNA. PCR cycling conditions were as follows: 95°C for 2 min, 35 cycles of 94°C for 30 s, 50°C for 1 min, and 65°C for 8 min, followed by a final step of 65°C for 8 min. PCR products were separated using an Agilent 2100 Bioanalyzer and a DNA 7500 chip. For both RAPD and BOX-PCR analysis, 1 μl PCR product was analyzed on an Agilent 2100 Bioanalyzer using a DNA 7500 chip, following the manufacturer's protocol. The fingerprints were clustered using the software GelCompar II to produce a dendrogram (Dice/UPGMA).
The Clondiag (Jena, Germany) AT genotyping system (39) was used according to the protocol provided by the manufacturer. The AT chip detects two different types of sequences, 13 single nucleotide polymorphisms (SNPs) for analysis of the conserved genome and 38 variable genetic markers for analysis of the accessory genome. The latter includes markers for virulence factors and previously reported genomic islands. The data from the 13 SNPs, flagellin type (a or b), and presence of the mutually exclusive type III secretion exotoxins (S or U), can be converted into a “hexadecimal code” represented by four digits, allowing the published database to be searched (39). This method has been used previously to identify the LES (26).
Twenty-four isolates, chosen on the basis of inconsistency between three LES identification methods (assays for two separate PCR markers and PFGE), were subjected to genotyping using the AT system, which enabled us to identify diagnostic SNPs at various sites in the core genome. A total of 17/24 isolates were PCR negative for the PS21 marker, whereas 2/24 were PCR negative for the LES-F9 marker (Table (Table1).1). Isolates were classified as either LES or non-LES based on their SNP profiles and hexadecimal code (Table (Table1).1). Using the AT method, 13/24 isolates were identified as the LES and will be referred to as “anomalous LES isolates.” The remaining 11 isolates had a different SNP profile than the LES and were therefore classified as non-LES anomalous isolates (Table (Table1).1). The LES isolates have the strain PAO1 “wild-type” version of SNPs at eight loci (oriC, alkB2, citS1, ampC1, ampC3, ampC4, ampC5, and ampC6) and the mutant variant at four loci (oprL, citS2, oprI, and ampC7).
For the anomalous non-LES isolates, five different SNP types were identified (Table (Table1).1). Five of the isolates shared an identical SNP profile (isolates 0666, 079320, 079390, 079391, and 07444) and matched clone A3 from the published P. aeruginosa AT database (Table (Table1)1) (39). These isolates came from two different geographical locations: four isolates (from three different patients) originated from Liverpool and one originated from the HPA. All five of these isolates were PCR positive for LES-F9 but PCR negative for PS21. Isolates 079390 and 079391 were isolated at the same time from the same patient.
A second shared SNP profile was identified among three isolates, two from Clatterbridge Hospital (444 and 438) and one from the HPA (WM1). Each of these isolates was PCR positive for both of the LES markers but was not identified as the LES by PFGE. These isolates matched clone P from the P. aeruginosa AT database. The remaining three non-LES isolates had SNP patterns that did not match any strains in the P. aeruginosa AT database (Table (Table1)1) (39).
The collection of anomalous isolates was compared using PFGE, RAPD, and BOX-PCR genetic fingerprinting. Dendrograms showing the relationships between the strains according to banding patterns generated by PFGE, RAPD, and BOX-PCR analyses are shown in Fig. Fig.1.1. None of the three methods provided a clear distinction between the isolates as to which could be classified as the LES and which as non-LES. In particular, the LES isolates were highly variable according to PFGE and no clustering of LES isolates was seen (Fig. (Fig.1).1). For the RAPD and BOX-PCR analyses, 14 genuine LES isolates, including LESB58, are included in Fig. Fig.1.1. Both methods demonstrated some clustering of LES isolates; however, the LES isolate 0341 was markedly distinct from all other LES samples (<32% similarity to all other isolates) (Fig. (Fig.1).1). Using RAPD typing, the remaining LES isolates clustered in two different groups, A and B, with similarity values of >72% and >78%, respectively (Fig. (Fig.1),1), but considerable fingerprint discrimination was still evident even within these groups. Using BOX-PCR, eight of the genuine LES isolates were indistinguishable (100% similarity; cluster C) (Fig. (Fig.11).
Using the AT system we analyzed the distribution of various genomic islands and variable genes among the 13 isolates identified as the LES (see Table SA1 in the supplemental material). Six of the AT chip accessory genome loci present in the genome-sequenced LES isolate LESB58 were also found in all 13 of the anomalous LES isolates. Each had the fliCb flagellin gene type and was positive for the rarest pyoverdine receptor gene, fpvAIII (9). All LES isolates were also positive for PA2185 (P. aeruginosa PAO1 gene encoding the non-heme catalase KatN), TBC47-2 (from plasmid pKLC102 in clone C), and 47D7-1 (genomic island 47D7). Because of difficulties with interpretation with the exoS locus, due to unclear distinctions between positive hybridization and background, the presence of this locus was confirmed by PCR assay in all of the anomalous LES isolates. The AT chip data indicated considerable variation among the anomalous LES isolates with respect to the presence of other accessory genome loci (see Table SA1 in the supplemental material).
The genome of the sequenced LES isolate (LESB58) contains one defective and five complete prophages (17), including a Pf1-like prophage (LES prophage 6), a D3112-like prophage (LES prophage 4), a D3-like prophage (LES prophage 5) and two novel prophages (LES prophage 2 and LES prophage 3). PCR assays were used to amplify regions within the prophages and flanking regions that only produce a product when the prophages are absent (Table (Table3).3). Although the AT chip includes a locus from the Pf1 prophage (PA0722), the LES equivalent of this gene (PLES_41241) encodes a predicted protein sharing only 48% identity over its 83-amino-acid length with PA0722. Using the PCR assays, the LES version of this prophage locus was identified in 9/13 of the anomalous LES isolates and 5/11 of the non-LES anomalous isolates (Table (Table3).3). PCR assays were also used to screen for the presence of representative open reading frames (ORFs) from each of the four other complete LES prophages. The unstable D3-like prophage (LES prophage 5) was not found in any of the isolates tested. Previously, it has been reported that the D3112-like prophage (LES prophage 4) was found in all LES isolates tested (40). However, only five of the anomalous LES isolates and two of the non-LES anomalous isolates were PCR positive for LES prophage 4. For the anomalous LES isolates, 5/13 isolates produced a positive result for LES prophage 2, and the same five isolates were also positive for LES prophage 3. For the non-LES anomalous isolates, a LES prophage 2 sequence was identified in two isolates (444 and 079444), and a LES prophage 3 sequence was identified in three isolates (444, 438, and 43513). However, the PCR assay for the flanking regions yielded no product, suggesting that any prophages might be inserted in a different genomic location than LES isolates.
The LES has a genomic island (LESGI-3) containing many sequences matching the PAGI-2 island present in P. aeruginosa clone C. This island has a variable cargo region, which in LESB58 includes the PS21 marker. We used an additional PCR test for the cargo region of LESGI-3. Six of 13 anomalous LES isolates were positive for the LESGI-3 marker, using PCR assays, and this correlated with the six isolates that had at least one positive signal for a locus within the PAGI-2/3 group on the AT chip. The remaining seven anomalous LES isolates had a negative result for LESGI-3 and were also lacking the PS21 marker. In the non-LES anomalous isolates, nine isolates were positive for the PAGI2/3 group using the AT method but only five isolates were PCR positive for LESGI-3 (Table (Table3).3). LESGI-4 is similar to the genomic island PAGI-1. PCR amplification using primers targeting sequences within this island revealed that it may be at least partially present in the majority of both the LES and non-LES anomalous isolates. PCR-negative results were obtained from only three isolates (0342, WM1, and 43513) (Table (Table3).3). The AT chip also indicates a high prevalence of this island, and negative results were identified in only three isolates (079345, 444, and 438). However, these negative AT results did not correlate with the PCR results. The differences in these results could be due to small nucleotide changes or deletions resulting in the failure of PCR amplification. Such results could suggest a degree of variability within the island. For the novel genomic island (LESGI-5), the majority of LES anomalous isolates were PCR positive. However, three anomalous LES isolates from different patients were PCR negative (0342, 0491, and 44366) (Table (Table3).3). LESGI-5 was not identified in any of the non-LES anomalous isolates.
Overall, none of the LES anomalous isolates were positive for all of the genomic islands and prophages found in the genome-sequenced isolate LESB58.
In this study, we report analysis of anomalous CF isolates of P. aeruginosa where the results of PCR assays and PFGE for the identification of an important epidemic strain conflict. We have demonstrated that analysis using the genetic fingerprinting methods PFGE, RAPD, and BOX-PCR cannot unequivocally resolve the identification of genuine LES isolates.
Our recommendation has been that PCR assays should be used as a first screen, with any unclear results or possible first-time LES-positive patients being subjected to further analysis (13, 34). The PS21 PCR marker was originally tested against a large panel of isolates (13, 29, 31) and showed good specificity for the LES. However, the existence of false positives has since been recognized (20, 34). Hence, the LES diagnostic PCR test was expanded to include a second marker, LES-F9 (13, 34). In this study, a further four false positives for the PS21 marker were identified. More worryingly, for the first time 11 false negatives were identified for the PS21 marker. For the LES-F9 marker, 10 false positives and one false negative were identified. There were also three false positives for both markers. Thus, further confirmatory tests would be needed for any isolates testing positive for either PS21 or LES-F9 (we have shown that these can be LES or non-LES) or for any isolates testing positive for both markers from patients previously not infected with the LES (we have shown that occasional false positives for both markers can occur). Although we have not sought extensive characterization of the genetic events leading to these anomalies, the nature of the deletion resulting in a false-negative result for PS21 was determined for one isolate (069237). A deletion of approximately 1.5 kb had occurred, resulting in the loss of the PS21 marker (data not shown).
Although macrorestriction coupled with PFGE is considered the “gold standard” for strain identification in many bacterial species, the application of this method to bacteria in chronic infections has been questioned. Tenover et al. devised guidelines to interpret PFGE restriction patterns (37). These guidelines, which enable inferred relationships based on the number of band differences, are often quoted in published work. However, this method was suggested for use in “analyzing discrete sets of isolates obtained during epidemiologic studies of potential outbreaks in hospitals or communities spanning relatively short periods (1 to 3 months)”; therefore, this method of interpretation may not be suitable in the context of the prolonged chronic infections typical in CF patients (28). PFGE is a time-consuming, labor-intensive method, and point mutations, insertions, and deletions of DNA can dramatically change banding patterns, resulting in misleading results. This has been reported for a number of bacteria, including Escherichia coli (5, 28), Shigella sonnei (36), and Campylobacter jejuni (4). It is clear that bacteriophages are major contributors to genomic variation (4, 5, 8, 15, 16, 18, 21, 28, 36). Indeed, Goerke et al. showed that phage conversion in Staphylococcus aureus was able to change the PFGE pattern (15).
Our study shows that, in the case of these anomalous isolates, PFGE failed to distinguish between LES and non-LES isolates. In addition, although demonstrating some clustering of LES isolates, RAPD and BOX-PCR typing methods could also not be used to identify all LES isolates. It is likely that the variations we found in the carriage of bacteriophages and genomic islands in the genomes of anomalous LES isolates contribute to the inability of these commonly used molecular typing methods to distinguish between LES and non-LES isolates.
Selective pressures in the CF patient's lungs during chronic infection, such as host responses and repeated antibiotic treatment, may act as a driver for microevolution. It is thought that genome plasticity is an important feature which contributes to adaptations to the host environment. The genomic instability exhibited by the LES may be an indicator of such adaptability.
It has been reported that the SNP typing method is 99.97% accurate in discriminating between P. aeruginosa strains (27). Wiehlmann et al. (39) also found that the results from the SNP genotyping method did not always correspond with PFGE due to the presence or absence of genomic islands (39). We found the AT system to be a very useful tool for resolving anomalies. A further advantage of this system is the opportunity to compare with a database of genotypes and to therefore monitor the emergence of new clones, including, potentially, new CF epidemic strains.
Effective segregation requires accurate, quick, and ideally cheap strain identification. Our results show that there are major limitations to the current identification methods used to identify the LES, and there may be implications for P. aeruginosa strains in CF generally. In some cases, the genetic variation evident within populations of the LES is too extensive for many current typing methods to cope with. In this context, with the occurrence of extensive instability in the accessory genome, discrimination may not always be a good feature of bacterial typing. Certainly, the observations presented in this and other studies question the efficacy of the gold standard typing method, PFGE, in enabling the correct identification of important strains in the single-strain-but-divergent populations that are typical of CF chronic infections. Since the AT system provides reproducible typing based on core genome SNPs, we have adopted it for identification of the LES.
Many of the anomalous LES isolates were cultured from the same sputum sample as isolates giving clear, unambiguous results and with PS21-negative and PS21-positive LES coexisting. The natural inclination of the medical microbiologist is to isolate and study cultures derived from single bacterial cells. However, in CF and other chronic infections this paradigm may be flawed, not only in the context of typing but also with respect to meaningful antimicrobial susceptibility testing (12, 14). Certainly for identification of the LES, PCR assays targeting bacterial “sweeps” or carried out directly from sputum samples may be the more reliable indicator of presence.
Some of this work was undertaken under the framework of the UK CF Microbiology Consortium, an initiative funded by the Big Lottery Fund in association with the Cystic Fibrosis Trust. We thank the National Institute of Health Research and the Northwest Development Agency (NWDA) for infrastructural and project support in funding the Biomedical Research Centre in Microbial Disease in Liverpool, United Kingdom. J.W. is funded by a Ph.D. studentship from the Natural Environment Research Council.
We also thank Tyrone Pitt, Laboratory of Healthcare-Associated Infection, Health Protection Agency, for providing strains and advice and John E. Corkill for advice on PFGE.
Published ahead of print on 21 April 2010.
†Supplemental material for this article may be found at http://jcm.asm.org/.