HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via CrossRef Citing Articles via Google Scholar Google Scholar Articles by Winstanley, C. Search for Related Content PubMed PubMed Citation Articles by Winstanley, C. Agricola Articles by Winstanley, C.

Genotypic and phenotypic characteristics of Pseudomonas aeruginosa isolates associated with ulcerative keratitis

Division of Medical Microbiology and Genitourinary Medicine, School of Clinical Laboratory Sciences, University of Liverpool, Liverpool L69 3GA, UK

Correspondence Craig Winstanley C.Winstanley{at}liv.ac.uk

Received January 4, 2005
Accepted February 18, 2005

A collection of 63 isolates of Pseudomonas aeruginosa associated with ulcerative keratitis, collected from six centres in England, were typed using serotyping and random amplified polymorphic DNA-PCR, and screened for several variable virulence-related genotypes and phenotypes. Sixty-one percent of the isolates were of either serotype O1 or serotype O11, but there was no evidence for a common clone. The majority of isolates (59 %) were PCR-positive for exoU rather than for exoS (38 %), and carried a-type fliC genes (76 %) rather than b-type (24 %). Isolates were PCR-positive for pyoverdine-receptor types at a prevalence of 38 % for type I, 46 % for type II and 8 % for type III. All but one of the isolates exhibited twitching activity. There was a correlation between the presence of exoS and twitching activity (P = 0.04), suggesting that a combination of exoS genotype and good twitching activity may have a role to play in ExoU-independent corneal virulence.

	METHODS

TOP METHODS RESULTS DISCUSSION ACKNOWLEDGEMENTS REFERENCES

 
Keratitis isolates.

The Microbiology Ophthalmic Group is a recently established group of microbiologists and ophthalmologists from six UK centres: London, Birmingham, Newcastle, Bristol, Manchester and Liverpool. Between April 2003 and March 2004, isolates from the corneas of patients with ulcerative keratitis were sent to the reference laboratory in Liverpool for storage. Sixty-eight isolates were presumptively identified as P. aeruginosa or Pseudomonas species. Of these, 63 were confirmed as P. aeruginosa by PCR amplification of the oprL gene as described elsewhere (De Vos et al., 1997). The remaining five isolates were the only isolates that tested negative using a previously published PCR assay for the exoA gene (Panagea et al., 2003), and were also negative in other tests targeting P. aeruginosa genes. These five isolates were excluded from further study. All but four of the remaining 63 isolates exhibited haemolysis on blood agar, and all produced siderophores on chrome azurol S agar (Schwyn & Neilands, 1987). Two common laboratory reference strains were included in the phenotypic assays: PAO1 and PA14 (gift of Laurence Rahme, Harvard Medical School).

Isolate typing.

The 63 P. aeruginosa isolates were subjected to O-antigen serotyping, using a commercially available kit (Bio-Rad), and to random amplified polymorphic DNA (RAPD) typing using primers 272 and 208 as described elsewhere (Mahenthiralingam et al., 1996).

Genotyping.

Multiplex-PCR assays were used to determine the distribution of the TTS genes exoT, exoY, exoS and exoU (Ajayi et al., 2003) and the pyoverdine-receptor type (De Chial et al., 2003). The flagellin gene type (a or b) was determined using PCR assays either with the published primers CW45 and CW46 (Winstanley et al., 1996) or the alternative primers CW50 and CW51. The oligonucleotide primers and annealing temperatures used in this study are listed in Table 1.

Twitching motility.

Twitching motility was assessed using a method similar to that described by Fonseca et al. (2004). Cells from fresh culture plates were stab-inoculated with a toothpick through a thin (approximately 3 mm) Luria agar plate onto the bottom of the Petri dish. The twitching zone between the agar and the Petri dish surface was measured after incubation for 24 h at 37 °C. Attached cells were stained with crystal violet (1 % w/v). Each assay was performed in triplicate and repeated three times. The detection limit used was 5 mm.

	RESULTS

TOP METHODS RESULTS DISCUSSION ACKNOWLEDGEMENTS REFERENCES

 
Isolate typing using serotyping and RAPDs

Using the Bio-Rad slide agglutination tests for the designation of serotypes O1–O16, it was possible to assign 59 of the 63 isolates to a serotype. Each of the remaining four isolates gave agglutination with the polyvalent PME serum (O2, O5, O15 and O16) but failed to agglutinate with monovalent sera. The distribution of serotypes is shown in Table 2. Thirty-nine of the isolates (61 %) were assigned to either serotype O1 or serotype O11.

All 63 isolates were analysed using RAPD typing. Particular attention was paid to comparing strains with common serotypes. Only three pairs of isolates were indistinguishable using RAPD typing. One pair shared the O11 serotype but differed in fpvA type and were isolated in different centres. A second pair shared serotype O10 and were both isolated from Moorfields Eye Hospital, London, but differed in PCR assays for exoY and fliC. Both carried b-type flagellins, but one was PCR-positive with only one of the primer sets for fliC whereas the other isolate was PCR-positive using both sets. Only one of these isolates was PCR-positive for exoY. A third pair, both isolated from Moorfields Eye Hospital, shared the same profile using PCR assays and serotyping. These isolates both gave agglutination only with the polyvalent PME serum and had similar low twitching activity. These two isolates were the only isolates indistinguishable by all the methods used. There was no evidence for any widespread clones.

Distribution of TTS genes

Of the 63 isolates screened, 37 (59 %) were PCR-positive for exoU, 24 (38 %) were PCR-positive for exoS, 51 (81 %) were PCR-positive for exoY and 63 (100 %) were PCR-positive for exoT. In accordance with previous studies, exoS and exoU were mutually exclusive, with the exception of one isolate that was PCR-positive for both and three isolates that were PCR-negative for both. One exoS-positive isolate was only identified when alternative PCR primers were used (Lee et al., 2003).

The distribution of exoS and exoU amongst the various serotypes is shown in Fig. 1(a). There was a significant association (P < 0.001) between serotype O11 and the presence of exoU. In contrast, exoS and exoU were far more evenly distributed amongst serotype O1 isolates. There were only small numbers representing the other serotypes found in this study. In some cases they exhibited an exclusive genotype for one of exoU or exoS (O2, O6, O8 and O10). The three serotype O4 isolates consisted of one exoS-positive, one exoU-positive and one exoS- and exoU-positive isolate.

View larger version (14K): [in this window] [in a new window]   Fig. 1. Distribution of variable genotypes amongst serotypes. The figure shows the number of isolates (y axis) within a serotype (x axis) that were PCR-positive for (a) exoS or exoU, (b) fliCa or fliCb, and (c) pyoverdine-receptor type I, II or III. NT indicates isolates that were non-serotypable using monovalent sera. N indicates isolates that were not assigned to a pyoverdine-receptor type.

Other variable genotypes

The 63 P. aeruginosa isolates could all be assigned to either fliC type a (48 isolates, 76 %) or fliC type b (15 isolates, 24 %). The distribution of fliC types amongst the serotype groupings is shown in Fig. 1(b). There was a clear association between some serotypes and flagellin types. Notably, the two dominant serotypes (O1 and O11) both exclusively carried a-type flagellin genes.

Pyoverdine-receptor typing of the 63 isolates revealed the most common alleles to be fpvAII (29 isolates, 46 %) and fpvAI (24 isolates, 38 %). Five isolates (8 %) carried fpvAIII and the remaining five isolates were PCR-negative. Four of the fpvAII-positive isolates were only identified after using alternative primers (P. Cornelis, personal communication). The distribution of fpvA types amongst the serotype groupings is shown in Fig. 1(c). There was no significant association between serotype and pyoverdine-receptor type.

Twitching motility

The twitching motility of the 63 P. aeruginosa isolates varied from one isolate below the limit of detection to 35 mm. The overall mean value was 26 mm with a standard deviation of 1.46 mm. The distribution of isolates according to twitching zones is shown in Table 3. To facilitate comparison with previous studies we included reference strains PAO1 and PA14 and determined their twitching activity to be 26 mm and 14 mm, respectively. Head & Yu (2004) reported a comparison of twitching activities for a number of clinical and environmental isolates of P. aeruginosa. They reported that strain PA14 had good twitching activity but that in strain PAO1 twitching motility was undetectable. In our study the vast majority (57/63; 90 %) of isolates exhibited better twitching motility than strain PA14. There was no correlation between twitching activity and O-serotype.

	DISCUSSION

TOP METHODS RESULTS DISCUSSION ACKNOWLEDGEMENTS REFERENCES

 
The TTS system of P. aeruginosa facilitates the secretion and translocation of exotoxins into target host cells and is triggered by contact with host cells (Vallis et al., 1999). The substrate effectors, ExoU, ExoS, ExoT and ExoY, have been ascribed various functions. The role of ExoY (an adenylate cyclase) remains unclear and its absence from some strains has been reported elsewhere ( Finck-Barbancon et al., 1997; Feltman et al., 2001; Ajayi et al., 2003). ExoT and ExoS are both ADP-ribosylating enzymes but whilst ExoT appears to be virtually ubiquitous amongst P. aeruginosa, several authors have noted the mutual exclusivity between ExoS and the cytolytic factor ExoU (Fleiszig et al., 1997; Lomholt et al., 2001; Berthelot et al., 2003). General surveys of the proportion of strains carrying each of these TTS effector genes suggest that exoS is more common. Feltman et al. (2001) reported prevalences of 72 % and 28 % for exoS and exoU, respectively, amongst a panel of 115 clinical and environmental isolates that did not include any from eye infections. Even lower prevalences for exoU have been reported for P. aeruginosa isolates from blood (13 %) and lung (9 %) (Hirakata et al., 2000).

Berthelot et al. (2003) separated 92 bacteraemia isolates into four groups on the basis of their cytotoxicity against macrophages and analysed secretion of the ExoU and ExoS proteins. Twenty-six (28.3 %) of the strains were categorized as type I, exhibiting the highest cytotoxicity, and secreted ExoU. Forty-eight (52.2 %) of the strains exhibited slower rates of cytotoxicity (type II) and secreted ExoS. Eighteen (19.6 %) of the strains had either poor or no cytotoxicity and secreted neither ExoU nor ExoS. Overall gene prevalence levels for P. aeruginosa isolates for exoU and exoS were 29 (31.5 %) and 65 (70.7 %), respectively, including two isolates that possessed both genes. The authors reported a statistically significant correlation between TTS protein secretion and the presence or absence of the exoU and exoS genes (Berthelot et al., 2003). Thus, although PCR screening for TTS genes does not correspond directly to secretion phenotype, it can be used as an indicator of probable phenotype.

In a survey of 62 genetically distinct cystic fibrosis (CF) isolates from England, Scotland and Ireland, we found 48 (77.4 %) to be exoS-positive (unpublished data), suggesting that in CF the invasion phenotype may be more useful to P. aeruginosa. In our study of isolates associated with keratitis we found exoU-positive isolates to be in the majority (59 %:38 %, exoU:exoS). This contrasts with previous studies using isolates associated with different infections and suggests that ExoU-mediated cytotoxic activity may be an advantage for P. aeruginosa with respect to corneal infections. However, a significant minority of isolates lack exoU, suggesting that the gene is not essential in P. aeruginosa-mediated keratitis. Lomholt et al. (2001) reported a more even distribution between exoU and exoS genotypes amongst 61 keratitis isolates, but found exoS to be in a majority overall amongst the 145 P. aeruginosa isolates screened (34 %:66 %, exoU:exoS). In a study of 13 isolates associated with corneal infections in the USA, Cowell et al. (2003) reported invasive and cytotoxic phenotypes in equal proportions and suggested that since invasive and cytotoxic strains have different effects on corneal cells, they may require different treatment strategies.

One isolate was PCR-negative for exoS using the published multiplex-PCR assay (Ajayi et al., 2003) but PCR-positive using an alternative primer set (Lomholt et al., 2001). In addition, a further three isolates were PCR-negative for both exoS and exoU. At least one of these isolates was positive for exoU using dot-blot hybridization (unpublished data). These observations suggest that the multiplex-PCR assay, although a useful tool for rapid TTS genotyping, may not be effective for all strains of P. aeruginosa.

In contrast to studies of CF isolates, serotyping is a useful tool for the study of P. aeruginosa eye infection isolates. This supports the evidence that smooth LPS with intact O-antigen is required for corneal infection (Priebe et al., 2004). There have been a number of surveys investigating serotype distributions amongst P. aeruginosa isolates. In a study of 73 P. aeruginosa strains from various clinical and environmental sources, Pirnay et al. (2002) reported the predominant serotypes to be O11 (15.1 %), O1 (12.3 %), O6 (10.9 %) and O12 (9.6 %). Amongst 48 AFLP (amplified fragment length polymorphism) types isolated from burns patients, 58.3 % were reported as serotypes O1, O6, O11 or O12 (Pirnay et al., 2003). In a survey of 92 genetically distinct bacteraemia isolates, O6 (25 %) and O11 (18 %) were reported to be the most common serotypes (Berthelot et al., 2003). These studies contained few or no eye infection isolates. However, in a study of 23 isolates from contact lens wearers, Thuruthyil et al. (2001) reported O1 (30 %), O6 (17 %) and O11 (17 %) as the most common serotypes. Broadly speaking, the serotype distributions found in our study are consistent with these previous surveys but with greater predominance of the O1 and O11 serotypes.

Faure et al. (2003) analysed 99 non-ocular clinical isolates from both chronic and acute infections for association between serotype and secretion of ExoS, ExoT and ExoU. All nine serotype O11 isolates were from patients with acute infections. Seven of these isolates secreted ExoU, one secreted ExoS and one secreted neither. Of 13 serotype O1 isolates (12 from acute infections), seven secreted ExoS and six didn't secrete any of the three exotoxins. No serotype O1 isolates secreted ExoU. Berthelot et al. (2003) reported an association amongst bacteraemia isolates between serotype O1, O10 and O11 strains and exoU, and serotypes O3, O4, O6, O12 and O16 with exoS. Whereas four of five serotype O1 isolates were exoU-positive in this French study, we found exoU in only six of 15 serotype O1 keratitis isolates. However, both studies support an association of serotypes O10 and O11 with exoU. Berthelot et al. (2003) also reported two strains carrying both exoS and exoU genes. One of these isolates was serotypable and shared the same serotype (O4) as the only keratitis isolate in our study carrying both genes.

We found evidence of a correlation between serotypes O1 and O11 and flagellin type a. There is little in the literature with which to compare this novel finding. In 1996 we surveyed 64 clinical isolates and compared O-serotype with fliC type (Winstanley et al., 1996). Three of four serotype O1 and all three O11 isolates were flagellin type a. The overall type a:type b ratio (58 %:41 %) suggested a predominance of flagellin type a, but not to the same degree as the keratitis isolates in this study (76 %:24 %). In another survey of 51 genetically distinct CF isolates from England, Scotland and Ireland, we found the ratio of fliC type a:type b to be 59 %:41 % (unpublished data), a result showing remarkable concordance with the earlier study and confirming that the high proportion of type a flagella amongst keratitis isolates may be unusual. In the USA, Arora et al. (2001) surveyed a collection of 437 isolates and reported dominance of the type a flagellins to different extents in isolates from blood (65 %), CF (80 %) or the environment (69 %). The predominance of type a flagellin was much lower in our study restricted to CF isolates from the British Isles, suggesting that there may be geographical variations in flagellin type.

There is evidence not only that P. aeruginosa flagella play a role in mucin binding, but that such binding varies depending on flagellar type (Scharfman et al., 2001). It is possible, therefore, that the predominance of a flagellin, and hence a flagellar, type associated with a particular site of infection may be influenced by mucin-binding specificities. The apparent linkage between O1/O11 serotype and the type a fliC gene might suggest that selection for lineages carrying these gene combinations has occurred. However, it is clear that recombination/genetic variation has also played a role in the evolution of the O1/O11 serotype keratitis isolates since no such correlation was observed when the fpvA-receptor types were analysed. O1 serotype isolates carried fpvAI and fpvAII in nearly equal numbers and although fpvAII was the most common pyoverdine-receptor type amongst O11 serotype isolates, 11 of the 24 O11 isolates either carried a different fpvA allele or were non-typable.

The distribution of pyoverdine types amongst keratitis isolates was 46 % : 38 % : 8 % for fpvAII : fpvAI : fpvAIII. This compares with a distribution of 60 % : 22 % : 15 % in a study of 62 genetically distinct CF isolates from England, Scotland and Ireland (unpublished data). De Vos et al. (2001) also reported high levels of type II pyoverdine amongst CF isolates (70 %) but did not find any type III pyoverdine producers. In a study of 73 P. aeruginosa strains from various clinical and environmental sources, Pirnay et al. (2002) reported pyoverdine production to be 51 % : 21 % : 19 % for type II : type I : type III pyoverdines.

In our study the vast majority (57/63; 90 %) of isolates exhibited better twitching motility than strain PA14. In a previous study of 30 mainly clinical P. aeruginosa isolates only four had greater twitching activity than strain PA14 (Head & Yu, 2004). However, a comparison between our strain PA14 results and those published previously (Head & Yu, 2004) must be treated with caution in the light of the contradictory results obtained for strain PAO1, suggesting that this very common laboratory strain may exhibit varying phenotypes in different laboratories, possibly due to mutation or recombination events. In the study by Head & Yu (2004), twitching motility was undetectable for 12 of the 30 (40 %) isolates, all from CF patients. Kus et al. (2004) reported that over 71 % (113/159) of non-CF rectal and clinical isolates and 95.8 % (23 of 24) environmental isolates exhibited twitching activity. The authors also contrasted twitching activity rates between paediatric (81.4 %) and adult (54.5 %) CF isolates (Kus et al., 2004). This compares with 68 % lacking twitching activity in a study of 60 genetically distinct CF isolates from England, Scotland and Ireland, with those isolates exhibiting twitching activity only averaging 11.5 mm (unpublished data). In comparison, only one keratitis isolate (1.6 %) in our study appeared to lack twitching activity. It is clear that CF isolates can lose their twitching ability during the course of a chronic infection (Kus et al., 2004). However, the high prevalence of twitching activity observed amongst keratitis isolates in comparison to other clinical isolates (Kus et al., 2004) supports the notion that such activity is a requirement for successful P. aeruginosa corneal pathogens.

Interestingly there was some evidence in our study for a correlation between the presence of exoS and twitching activity (P = 0.04). An association between the expression of TTS exoenzymes, including ExoS, and twitching motility has already been reported (Ahn et al., 2004), and it may be that a combination of exoS genotype and good twitching activity has a role to play in ExoU-independent corneal virulence.

In conclusion, our survey of bacterial keratitis isolates in the UK revealed no evidence for a common clone but identified dominant serotypes (O1 and O11). The isolates were disproportionately exoU-positive when compared to non-keratitis isolates and all but one exhibited twitching motility. There is evidence for correlations between some serotypes and flagellin type, but not pyoverdine-receptor type. This study provides a useful reference with which to compare P. aeruginosa isolates from other clinical and non-clinical sources in order to identify variable genetic factors contributing to particular P. aeruginosa infections.

	ACKNOWLEDGEMENTS

TOP METHODS RESULTS DISCUSSION ACKNOWLEDGEMENTS REFERENCES

 
The other members of the Microbiology Ophthalmic Group are Stephen Tuft (Moorfields Eye Hospital), Mark Batterbury (Royal Liverpool University Hospital), Derek Tole, Stuart Cook and John Leeming (Bristol Eye Hospital), Peter McDonnell and Timothy Weller (Birmingham and Midlands Eye Hospital), Francisco Figueiredo and Steven Pedler (Royal Victoria Infirmary, Newcastle) and Andrew Tullo and Malcolm Armstrong (Manchester Royal Eye Hospital).

	REFERENCES

TOP METHODS RESULTS DISCUSSION ACKNOWLEDGEMENTS REFERENCES