INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Legionellae are environmental organisms that can cause disease in humans (2). Clinical manifestations of legionella infection range from no symptoms to potentially fatal pneumonia and multisystem disease. There are 42 species in the genus Legionella (5), more than half of which have been implicated in human disease (2).

Transmission of the bacteria from the environment to humans occurs via inhalation or aspiration of Legionella-containing aerosols (6, 8). A suspected cluster or outbreak of cases of legionellosis requires careful epidemiological investigation to identify possible sources of infection. Such investigations also require a sensitive and discriminatory subtyping technique to identify similarities and differences between possibly related strains (3).

Legionella longbeachae is an uncommon pathogen in most parts of the world (17) but causes up to half the cases of legionellosis in many regions in Australia (1, 11). The reason for this is not clear. It has commonly been isolated from soil and decomposing materials, such as bark or sawdust used in potting mixes (33), and has been detected occasionally in water (29). L. longbeachae has caused at least two outbreaks of legionellosis in Australia, one in Western Australia (7) and the other in South Australia (19). Studies of Australian clinical strains of L. longbeachae by multienzyme electrophoresis (19), ribotyping (19), and random amplified polymorphic DNA (RAPD) typing (9) have suggested that L. longbeachae serogroup 1 strains are largely clonal. This similarity between strains has thwarted attempts to develop a discriminatory subtyping method, which would be useful to link environmental isolates to cases of clinical disease.

Legionella pneumophila serogroup 1 causes up to 95% of the cases of legionellosis worldwide and most outbreaks and sporadic cases in Australia (1, 11). For this reason, it has been the focus of most subtyping techniques, including typing with different panels of monoclonal antibodies (20), plasmid analysis (10, 26), multienzyme or alloenzyme electrophoresis (18), restriction fragment length polymorphism (15, 16), ribotyping (18), arbitrary primed PCR (13, 14), RAPD typing (30), and macrorestriction enzyme digestion followed by pulsed-field gel electrophoresis (PFGE) (28, 31). At present, PFGE, following restriction digestion with the enzymes SfiI or NotI, is the most discriminatory method. This technique has also been used to subtype epidemiologically linked strains of L. pneumophila serogroup 6 (24), Legionella bozemanii (22), and Legionella micdadei (21) but was not used previously to type isolates of L. longbeachae.

In this study we used macrorestriction enzyme digestion followed by PFGE to investigate the genetic variability of clinical and environmental isolates of L. longbeachae serogroup 1 from five states in Australia over a period of 10 years.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial isolates. The 68 isolates of L. longbeachae serogroup 1 investigated in this study included 24 clinical isolates from New South Wales, 16 clinical and 5 environmental isolates from Queensland, 1 clinical isolate from Tasmania, 4 clinical isolates from South Australia, and 12 clinical and 6 environmental isolates from Western Australia (Table 1). One environmental isolate which had been identified as L. longbeachae serogroup 2 from Western Australia was also tested. Controls were selected from the American Type Culture Collection (ATCC): L. pneumophila serogroup 1 (Philadelphia 1; ATCC 33152), L. longbeachae serogroup 1 (Long Beach 4; ATCC 33264), and L. longbeachae serogroup 2 (Tucker 1; ATCC 33484).

Identification of isolates. A rapid latex test (Serobact; Disposable Products, Adelaide, South Australia) was used for presumptive identification of the isolates as L. longbeachae serogroup 1. These results were confirmed by direct immunofluorescence with a panel of pooled monovalent Legionella antibodies (MarDx Diagnostics, Scotch Plains, N.J.) and a monoclonal antibody to L. pneumophila groups 1 to 14 (Genetic Systems, Seattle, Wash.) according to the manufacturers' recommendations. Direct immunofluorescence with eight species- or serogroup-specific monovalent antibodies, including L. longbeachae serogroups 1 and 2 (MarDx Diagnostics), was also performed according the manufacturer's protocol. Isolate identification was confirmed as L. longbeachae with a positive reaction to reagents L. species b to j and L. omni species b to p and a negative reaction to both L. pneumophila reagents 1 to 6 and L. pneumophila 1 to 14, in addition to a positive reaction to L. longbeachae-specific monovalent antibody.

Preparation of PFGE plugs. PFGE plugs were prepared according to a modified version of the methods of Smith and Cantor (32) and Gautom (12). Briefly, bacterial cells were harvested into approximately 3 ml of Pett IV buffer (1.0 M NaCl, 10 mM Tris-HCl [pH 7.6]) and the bacterial suspensions were adjusted to exactly 20% transmittance (equivalent to 3 × 10¹⁰ organisms/ml) with a calibrated bacterial nephelometer (Vitek colorimeter; Hach Company, Loveland, Colo.). After centrifugation, a 400-µl aliquot of the bacterial suspension was concentrated to half its volume and mixed with an equal volume of molten 2.4% low-melting-point agarose (Bio-Rad, Hercules, Calif.) in Pett IV buffer and dispensed into a disposable plug mold (Bio-Rad). The final concentration of the bacterial DNA in the plug was 10 µg (1 µg of DNA/plug slice). The usually recommended preliminary RNase and lysozyme digestion step at 37°C (4, 23, 31) was omitted, as it was found that this step did not affect digestion with the enzyme SfiI. The bacterial plugs were incubated in 2-ml Eppendorf tubes containing 1.5 ml of ESP solution (0.5 M EDTA [pH 8.0], 1.0% N-lauryl sarcosine, 2 mg of proteinase K/ml) and incubated overnight at 55°C.

PFGE plug digestion and electrophoresis. Prior to digestion, the plugs were incubated in a solution of 2 ml of 10 mM Tris-0.1 M EDTA and 1.0 mM phenylmethylsulfonyl fluoride (pH 7.5) (Sigma-Aldrich, St. Louis, Mo.) for 1 h at room temperature, washed in 1× TE (10 mM Tris, 0.1 mM EDTA [pH 7.5]), and then stored in 1× TE at 4°C until required. For restriction enzyme digestion, plug slices were digested overnight at 50°C for SfiI or 37°C for NotI in a 50-µl reaction mixture which contained 2.5 U of SfiI or NotI restriction enzyme/ml of buffer (New England Biolabs, Beverly, Mass.). L. longbeachae serogroup 1 (ATCC 33462) and L. pneumophila serogroup 1 (ATCC 33152) were used as internal controls and digested in parallel with the test organisms. These were included as controls in every gel along with at least three lanes of Saccharomyces cerevisiae chromosomes (catalog no. 345; Promega, Madison, Wis.) as fragment size standards. The fragments were electrophoretically separated by PFGE with a contour-clamped homogeneous electric field system (Bio-Rad Chef Mapper) in 1% PFGE grade agarose (Bio-Rad) and 0.5× TBE running buffer (45 mM Tris, 45 mM boric acid, 1.0 mM EDTA [pH 8.0]). The initial pulse time of 3.51 s was increased linearly to a final switch time of 93.56 s over 24 h at 6 V/cm at 14°C. The gels were then stained with 0.5 g of ethidium bromide/ml for 10 min, destained in water, and photographed under UV transillumination.

Evaluation of reproducibility of the PFGE results. Electrophoretic bands for the PFGE restriction fragments were sized and compared with the software program GelCompar version 4.1 (Applied Maths, Kortrijk, Belgium). Computer comparison was based on the algorithm of the unweighted pair group method for arithmetic averages and the Dice coefficient (25) with 3.2% band tolerance. Band tolerance statistics were calculated on the basis of differences in band positions of a list of identical internal control patterns with the GelCompar program. The lowest band tolerance required to have identical isolates typed as identical by the GelCompar program was 3.2%, and this value was applied to the entire band-matching comparison. No other computer-enhanced optimization or smoothing was used.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Analysis of PFGE typing. Preliminary results showed that digestion with NotI produced too few (three to four) restriction fragments to allow discrimination between strains of L. longbeachae (results not shown). Further analyses were confined to PFGE with SfiI. The number of SfiI fragments varied from four to seven, which ranged in size from approximately 400 to 1,500 kb (Fig. 1, 2, and 3). Interpolation of the known chromosome sizes of S. cerevisiae gave a standard reference curve for the comparison of sample fragment sizes. The genome sizes of the L. longbeachae isolates were calculated by adding the sizes of individual fragments for each strain. They ranged from 3,300 to 4,300 kb (Table 1). A fragment of 1,493 kb was common to all L. longbeachae serogroup 1 isolates and both serogroups 1 and 2 ATCC strains but was absent from the one environmental isolate that had been identified as L. longbeachae serogroup 2 and from the L. pneumophila control strain. The next most common fragments of the L. longbeachae isolates were 389 (94% of isolates) and 701 kb (79% of isolates).

View larger version (68K): [in this window] [in a new window]   FIG. 1.   PFGE of SfiI-cleaved DNAs from L. longbeachae serogroup 1 isolates from Australia. Lanes: 2, L. pneumophila serogroup 1 (ATCC 33152); 3, L. longbeachae serogroup 1 (ATCC 33462); 4, 141; 5, 142; 6, 143; 7, 144; 8, 149; 9, 152; 10, 153; 12, 154; 13, 155; 14, 158; 15, 159; 16, 161; 17, 162; 18, 163; 19, 272. Lanes 1, 11, and 20 contained S. cerevisiae chromosomes as a molecular size standard. e, environmental isolate; c, clinical isolate. The letters A to C and the numbers 1 to 5 indicate pulsotypes and subgroups, respectively.

View larger version (80K): [in this window] [in a new window]   FIG. 2.   PFGE of SfiI-cleaved DNAs from L. longbeachae serogroup 1 isolates from Australia. Lanes: 2, L. pneumophila serogroup 1 (ATCC 33152); 3, L. longbeachae serogroup 1 (ATCC 33462); 4, L. longbeachae serogroup 2 (ATCC 33484); 5, 287; 6, 153; 7, 272; 8, 303; 9, 152; 10, 161; 12, 459; 13, 401; 14, 159; 15, 379; 16, 142; 17, 390. Lanes 1, 11, and 18 contained S. cerevisiae chromosomes as a molecular size standard. t, type strains; e, environmental isolate; c, clinical isolate. The letters A to C and the numbers 1 to 5 indicate pulsotypes and subgroups, respectively. The DNA block in lane 9 moved from its original position on the comb prior to the gel being run.

View larger version (68K): [in this window] [in a new window]   FIG. 3.   PFGE of SfiI-cleaved DNAs from L. longbeachae serogroup 1 isolates from Western Australia. Lanes: 2, L. pneumophila serogroup 1 (ATCC 33152); 3, L. longbeachae serogroup 1 (ATCC 33462); 4, 359; 5, 361; 6, 364; 7, 365; 8, 366; 9, 368; 10, 369; 12, 371; 13, 373; 14, 374; 15, 375; 16, 376; 17, 377; 18, 378. Lane 19, isolate 369, was typed by the MIDI system as an atypical L. longbeachae strain. Lanes 1, 11, and 20 contained S. cerevisiae chromosomes as a molecular size standard. e, environmental isolate; c, clinical isolate. The letters A to C and the numbers 1 to 4 indicate pulsotypes and subgroups, respectively.

View larger version (39K): [in this window] [in a new window]   FIG. 4.   Cluster dendrogram of Australian L. longbeachae isolates generated from SfiI restriction fragments separated by PFGE and analyzed with the program GelCompar ver 4.1. Similarity of >65% and more than a four-band difference in the band pattern divide the different types and clades. Subtype divisions have >88% similarity and band pattern differences of two or three bands. The type strain for L. longbeachae serogroup 1 was designated pulsotype B3, and the type strain for L. long-beachae serogroup 2 was designated pulsotype LL2. Isolates LL1 (i to vi) are representative internal controls of ATCC 33462, which were run on separate PFGE gels.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The purpose of our study was to examine the genetic variability of Australian L. longbeachae serogroup 1 isolates and develop a practicable method for subtyping. Using macrorestriction digestion with SfiI followed by PFGE, we demonstrated three distinct patterns (dendrogram clades) that could be separated by four-band differences and <65% similarity with the Dice coefficient (Fig. 1) among 68 clinical and environmental isolates. The three major L. longbeachae serogroup 1 pulsotypes were subdivided into 11 subgroups, most of which were widely distributed geographically throughout Australia and over significant periods.

These results are in contrast to those of previous studies, in which alloenzyme electrophoresis, ribotyping, and RAPD analyses failed to distinguish among Australian strains of L. longbeachae serogroup 1 and showed only minor differences among strains of L. longbeachae serogroup 2 (9, 19). The results of previous studies have been interpreted as indicating widespread distribution of a single clone of L. longbeachae in Australia (9, 19).

However, by adding the sizes of fragments obtained after digestion (Table 1), we estimated that the genome sizes of different strains varied from 3,300 to 4,300 kb. This is similar to the degree of variation in genome size among strains of L. pneumophila serogroup 1, which has been reported to range from 2,600 to 3,900 kb (31). A previous study had also reported the genome size of L. pneumophila Philadelphia 1 as 3,900 kb (4). This suggests that L. longbeachae is more variable than was previously believed.

This apparent variability is unlikely to be due to incomplete lysis of bacteria or digestion of DNA, since consistent results were obtained on repeat testing up to five times. The largest (1,493-kb) fragment was present in all L. longbeachae serogroup 1 strains and both L. longbeachae ATCC control strains, and the number of sizes of fragments did not vary when lower concentrations of DNA were used (data not shown). Moreover, the L. pneumophila PFGE internal control strain, which was processed in the same way as L. longbeachae isolates, gave the same number and size of fragments as previously described (31).

The appearance of the same pulsotypes and subgroups in different parts of Australia over a period of 10 years may be explained by a low mutation rate of L. longbeachae strains (19). Alternatively, it could be due to the widespread distribution of a common vehicle, such as potting mix, with persistence of L. longbeachae in the environment or in unused potting mix. It has been shown that L. longbeachae is able to survive in potting mix for up to 7 months (33).

Ideally, for PFGE, a restriction enzyme should be chosen that will generate at least 10 fragments per isolate (34). The enzyme NotI generated too few fragments to produce a useful profile, and SfiI generated only four to seven fragments from L. longbeachae DNA. However, in defining their criteria for the use of PFGE for bacterial subtyping, Tenover et al. (34) conceded that modification of the criteria may be necessary for defining pulsotypes among large numbers of isolates over extended periods. We suggest that modification of the criteria is justified to extend the use of SfiI for PFGE to L. longbeachae as well as L. pneumophila (31), L. micdadei (21), and L. bozemanii (22), for which its use has been described. PFGE typing of L. bozemanii with SfiI produced a similar number of fragments (four to eight), and pulsotypes were defined by criteria similar to those used in our study, namely, four or more band differences and <65% band similarity by the Dice coefficient (22).

The routine use of PFGE as a typing method is often limited by the fact that it is time-consuming and labor intensive (12). The standard procedure (23) was reduced to 3 days by using a turbidity standard rather than an optical density reading to estimate bacterial numbers for agarose plugs, and the omission of the lysozyme digestion step eliminated an overnight incubation. Although interpretation of subtyping data is most useful when multiple techniques are used (31), no other sufficiently discriminatory method for subtyping L. longbeachae has been described. However, it is likely that the use of other restriction enzymes, which cleave at different sites within the genome, could provide complementary patterns for subtyping.

SfiI digestion followed by PFGE showed that the Australian L. longbeachae strains are not a single clonal population. The results of this study contribute to an understanding of the distribution of L. longbeachae serogroup 1 strains in Australia. A computer database with a number of mainly unrelated environmental and clinical isolates from five Australian states was established and can now be used in, and supplemented by, investigations of future cases and outbreaks. The PFGE method developed is discriminatory, could be applied to other Legionella species, and is sufficiently rapid to allow a timely investigation of a potential outbreak of legionellosis.