ABSTRACT
We evaluated three molecular methods for identification of Francisella strains: pulsed-field gel electrophoresis (PFGE), amplified fragment length polymorphism (AFLP) analysis, and 16S rRNA gene sequencing. The analysis was performed with 54 Francisella tularensis subsp. holarctica, 5 F. tularensis subsp. tularensis, 2 F. tularensis subsp. novicida, and 1 F. philomiragia strains. On the basis of the combination of results obtained by PFGE with the restriction enzymes XhoI and BamHI, PFGE revealed seven pulsotypes, which allowed us to discriminate the strains to the subspecies level and which even allowed us to discriminate among some isolates of F. tularensis subsp. holarctica. The AFLP analysis technique produced some degree of discrimination among F. tularensis subsp. holarctica strains (one primary cluster with three major subclusters and minor variations within subclusters) when EcoRI-C and MseI-A, EcoRI-T and MseI-T, EcoRI-A and MseI-C, and EcoRI-0 and MseI-CA were used as primers. The degree of similarity among the strains was about 94%. The percent similarities of the AFLP profiles of this subspecies compared to those of F. tularensis subsp. tularensis, F. tularensis subsp. novicida, and F. philomiragia were less than 90%, about 72%, and less than 24%, respectively, thus permitting easy differentiation of this subspecies. 16S rRNA gene sequencing revealed 100% similarity for all F. tularensis subsp. holarctica isolates compared in this study. These results suggest that although limited genetic heterogeneity among F. tularensis subsp. holarctica isolates was observed, PFGE and AFLP analysis appear to be promising tools for the diagnosis of infections caused by different subspecies of F. tularensis and suitable techniques for the differentiation of individual strains.
The family Francisellaceae, which belongs to the γ subclass of the class Proteobacteria, includes closely related organisms within the single genus Francisella. There are two recognized species, Francisella tularensis and F. philomiragia (25). Of these, F. philomiragia is relatively rare, is often associated with water, and is less virulent than F. tularensis (14). The more common species, F. tularensis, has four subspecies, all of which have been related to the acute infectious disease known as tularemia, although they differ in their virulence for humans and rabbits (16).
F. tularensis subsp. tularensis (also called F. tularensis subsp. nearctica or biovar type A), the causal agent of tularemia in North America, has recently been observed for the first time in Europe (Slovakia) (13). This subspecies is a highly virulent pathogen for many mammalian species, including humans, and is mainly associated with tick-borne tularemia in lagomorphs (16). F. tularensis subsp. palaearctica (also called F. tularensis subsp. holarctica or biovar type B) is more widely distributed in nature and is found in Europe, Asia, and to a lesser degree in North America. This subspecies is linked to waterborne disease of rodents and hares, it is considered less pathogenic for mammals than F. tularensis subsp. tularensis, and it usually causes the ulceroglandular form of tularemia in humans (28). F. tularensis subsp. mediaasiatica has been found only in the former Soviet republics of Central Asia (23). The fourth subspecies, F. tularensis subsp. novicida, has been recovered from water samples and has rarely been associated with human illness (14).
The subspecies of F. tularensis have been differentiated by biochemical analysis, and the two main subspecies have been differentiated by 16S rRNA hybridization (8, 23). However, such tests have obvious drawbacks for unambiguous strain classification. In addition, Francisella strains have similar antigenic compositions, and it is not possible to distinguish the different subspecies from each other by serological methods (21). Therefore, the development of new molecular typing methods is of special interest.
F. tularensis has been described as a genetically homogeneous species, despite differences in virulence and geographic origins (18), with a limited diversity among isolates. In this respect, some molecular techniques have been able to discriminate the species within the genus but not the subspecies within the species (9, 15, 17); however, other recent taxonomic work based on either repetitive element sequence-based PCR (rep-PCR), random amplified polymorphic DNA (RAPD) analysis (4, 16), or allele size determinations (6, 18) has successfully demonstrated variability in subspecies and individual strains of F. tularensis. In addition, a capture enzyme-linked immunosorbent assay based on lipopolysaccharide-specific monoclonal antibodies and a handheld immunochromatographic assay have been applied to the detection of F. tularensis subsp. tularensis and F. tularensis subsp. holarctica (12).
Some recently developed molecular methods with strong discriminatory power, such as macrorestriction pulsed-field gel electrophoresis (PFGE) and amplified fragment length polymorphism (AFLP) analysis, have been used to differentiate species and clinical isolates of a variety of gram-positive and gram-negative organisms other than Francisella (7, 11, 30). PFGE provides a highly reproducible restriction profile that typically shows distinct, well-resolved fragments representing the entire bacterial chromosome in a single gel (20). AFLP analysis is a genome fingerprinting technique based on the selective amplification of a subset of DNA fragments generated by restriction enzyme digestion. It was originally applied to the characterization of plant genomes and more recently has been applied to the typing of bacteria (29).
This report describes the genetic characterization by PFGE and AFLP analysis of Spanish F. tularensis isolates recovered from hares, humans, voles, and ticks and compares the genetic profiles with those of other F. tularensis strains recovered from patients and animals with tularemia in the United States and Europe, as well as with those of F. philomiragia isolates. These two molecular techniques are also compared with a more traditional typing method, 16S rRNA gene sequencing. Although the latter method usually shows limited variability between strains of a bacterial species (22), any mutation in the genes of this highly conserved region could be of interest for epidemiological purposes.
MATERIALS AND METHODS
Bacterial strains and culture conditions.Sixty-two Francisella tularensis strains were studied (Table 1). Forty-two F. tularensis subsp. holarctica isolates were recovered from a tularemia epidemic (which involved ulceroglandular, typhoidal, glandular, pneumonic, oculoglandular, and atypical clinical forms diagnosed in humans) that occurred in Castilla y León, which is in northwestern Spain, between 1997 and 1999 (4, 10). Of these isolates, 31 were from hares, 8 were from humans, 1 was from a vole, and 2 were from ticks. The geographical origins of these isolates are shown in Table 1. In addition, three Czech, two French, two Russian, and five U.S. isolates were included in the study. Five F. tularensis subsp. tularensis, two F. tularensis subsp. novicida, and one F. philomiragia strains were also included. All isolates, which had been stored frozen at −80°C, were grown aerobically at 37°C for 2 days on modified Thayer-Martin agar plates containing GC medium base (36 g/liter), hemoglobin (10 g/liter), and Vitox supplement (2 vials/liter; Oxoid Ltd., Basingstoke, England).
Characteristics of the strains of the genus Francisella used in this study
16S rRNA gene sequencing.Sequencing of the 16S rRNA genes was performed with universal primers directed at a 1.4-kb conserved region of the gene (19). The PCR amplification products were purified in CentriSep (Princeton Separations, Inc., Adelphia, N.J.) columns and were sequenced with the Taq Dye Deoxy Terminator Cycle Sequencing kit on a model 310 genetic analyzer (Applied Biosystems, Foster City, Calif.) according to the instructions of the manufacturer. Sequence comparisons were done with the BLAST (version 2.0) system (1).
Analysis of chromosomal DNA restriction pattern by PFGE.The Francisella strains were grown as described above. A loop filled with cells from each strain was suspended in SE buffer (25 mM EDTA, 75 mM NaCl [pH 7.5]). The cells were harvested by centrifugation (3,500 × g, 10 min, 4°C) and washed twice with the same buffer. Agarose plugs were made from a 1:1 mixture of 2% low-melting-point agarose and the cell suspension. The plugs were lysed in buffer (50 mM Tris-HCl, 50 mM EDTA, 1% [vol/vol] lauroyl sarcosine, 500 μg of lysozyme per ml [pH 9.5]) for 24 h at 37°C. The cells were then treated for 24 h at 56°C with the same volume of a solution (50 mM Tris-HCl, 50 mM EDTA, 1% [vol/vol] lauroyl sarcosine [pH 9.5]) containing proteinase K at 500 μg/ml and were washed three times with Tris-EDTA buffer for 1 h at 4°C. XhoI, BamHI, SpeI, XbaI, SmaI, BglII, and SacII (all from MBI Fermentas) and ApaI (Promega Corp.) were used for restriction endonuclease digestion in accordance with the instructions of the manufacturers. The fragments were resolved by PFGE in electrophoresis-grade agarose (1%; Boehringer Mannheim) by using a CHEF-DR III system (Bio-Rad). The following parameters were used: running time, 24 h; temperature, 14°C; voltage gradient, 200 V; included angle, 120°; an initial pulse time of 0.1 s and a final pulse time of 10 s for XhoI, SpeI, XbaI, BglII, and SacII; an initial pulse time of 0.1 s and a final pulse time of 15 s for BamHI; and an initial pulse time of 0.1 s and a final pulse time of 25 s for SmaI and ApaI. The gels were stained with ethidium bromide (0.5 μg/ml) for 15 min, destained in distilled water, and photographed under UV light. A bacteriophage lambda ladder PFGE marker (concatemers of bacteriophage lambda cI857 Sam7; Boehringer Mannheim) was used for molecular size determination. Any nonidentity in terms of the presence, absence, or apparent mobility of a band was considered one difference. Variations in band intensity were not counted as a difference.
AFLP analysis.AFLP analysis of DNA was performed in duplicate with the AFLP Microbial Fingerprinting kit (Applied Biosystems) according to the recommendations of the manufacturer. Briefly, 10 ng of purified DNA was simultaneously restricted with EcoRI and MseI endonucleases (New England Biolabs, Beverly, Mass.) and ligated to the adapter-linker sequences with T4 DNA ligase (New England Biolabs) for 2 h at 37°C in a final volume of 11 μl. After dilution, 4 μl was amplified with the EcoRI and MseI core primer sequences in a final volume of 20 μl. Selective amplification was performed with primers EcoRI-C and MseI-A, EcoRI-T and MseI-T, EcoRI-A and MseI-C, and EcoRI-0 and MseI-CA by using 1.5 μl of the diluted preselective amplification reaction mixture in a final volume of 10 μl. All reactions were performed in an Applied Biosystems model 9600 thermal cycler with 200-μl reaction tubes.
Separation and detection of the AFLP fragments were performed with an Applied Biosystems model 310 genetic analyzer equipped with a 47-cm capillary and by using polymer POP-4. Samples were injected for 8 s at 15 kV and separated for 28 min at 15 kV and 60°C by using denaturing conditions. Each sample contained GeneScan 500[ROX] (Applied Biosystems) as an internal size standard. Determination of the sizes of the fragments was performed automatically with GeneScan software (version 3.1; Applied Biosystems). Additional analysis and construction of dendrograms and trees were done with GelCompar II software (version 2.0; Applied Maths, Sint-Martens-Latem, Belgium). The dendrograms were constructed by binary band matching by using the Dice correlation and the unweighted pairwise group method with mathematical averaging (UPGMA). The unrooted maximum parsimony tree was constructed by using 500 bootstrap simulations.
RESULTS
16S rRNA gene sequencing.Only the 42 Spanish F. tularensis subsp. holarctica isolates were tested by the 16S rRNA gene sequencing method. All strains tested were found to have the same 16S rRNA gene sequence, and that sequence was compared with the 16S rRNA sequences previously reported to GenBank. The sequence detected in this study shared 100% similarity to that of F. tularensis subsp. holarctica strain LVS (Table 2) and 99% similarity to those of strains from another Spanish outbreak of tularemia (2). In the previous study (2), the amplicons were directly amplified from clinical or environmental samples but not from F. tularensis subsp. holarctica isolates. The similarities observed between our sequences and those from the other F. tularensis subspecies listed in Table 2 were also 99%; the sequences of the strains shared lower percent similarities with those of F. philomiragia strains.
16S rRNA gene sequence similarity among the 42 Spanish F. tularensis subsp. holarctica isolates tested in this study and other Francisella strains
PFGE analysis.Only 49 Francisella strains were studied by PFGE (Table 1). Of the eight restriction enzymes tested, SmaI and ApaI provided profiles with only four bands, and these profiles could be used because of the small number of bands. On the other hand, SpeI, XbaI, BglII, and SacII gave patterns which were extremely difficult to interpret because of the large number of bands (more than 30 bands, with all bands having molecular sizes below 40 kb). XhoI was used as the primary enzyme, and five different DNA fragment profiles were generated, with a mean of 20 bands per isolate, all of which were less than 100 kb (Table 1 and Fig. 1A). The 42 Spanish F. tularensis subsp. holarctica clinical isolates exhibited the same XhoI pattern (designated pattern B), irrespective of the host from which they were recovered; this profile was also shared by the three Czech F. tularensis subsp. holarctica isolates. The human F. tularensis subsp. tularensis isolate, the F. tularensis subsp. holarctica strain isolated from a tick in Russia, F. tularensis subsp. novicida ATCC 15482, and F. philomiragia were assigned profiles A, C, D, and E, respectively.
PFGE fingerprint patterns of Francisella strains after digestion with XhoI (A) and BamHI (B). (A) Lanes: 1, F. tularensis subsp. holarctica isolate 26 (pattern B); 2, F. tularensis subsp. tularensis strain Schu (isolate 1 in this study; pattern A); 3, F. tularensis subsp. holarctica strain 503 (isolate 51 in this study; pattern C); 4, F. tularensis subsp. novicida reference strain ATCC 15482 (isolate 54 in this study; pattern D); 5, F. philomiragia reference strain ATCC 25015 (isolate 56 in this study; pattern E); 6, bacteriophage lambda ladder PFGE marker. (B) Lanes: 1 and 6, bacteriophage lambda ladder PFGE marker; 2, F. tularensis subsp. holarctica isolate 8 (type A); 3, F. tularensis subsp. holarctica isolate 11 (type B); 4, F. tularensis subsp. holarctica strain CAPM 5536 (isolate 26 in this study; type C); 5, F. tularensis subsp. holarctica strain 503 (isolate 51 in this study; type D).
BamHI also provided good discrimination and was used with F. tularensis subsp. holarctica isolates to help interpret minor differences in the PFGE patterns obtained with XhoI. Therefore, it was applied only to the Spanish and Czech isolates and Russian strain 51, yielding four distinguishable patterns, with 12 to 16 bands with sizes ranging from about 27 to 291 kb. These profiles differed from each other by three or fewer restriction fragments. Thirty-five of the Spanish isolates (83.3%) as well as the Czech hare isolates were designated type A, four Spanish hare isolates recovered in 1998 were designated type B (9.5%), three other Spanish hare isolates recovered between 1997 and 1998 (7.1%) were designated type C, and finally, the Russian tick isolate was designated type D (Table 1 and Fig. 1B).
When the results obtained with XhoI and BamHI were analyzed together, the 49 isolates studied could be divided into seven pulsotype groups (Table 1). Most strains (77.6%) belonged to pulsotype III, which included the majority of the Spanish F. tularensis subsp. holarctica isolates and Czech isolates. The F. tularensis subsp. tularensis isolate tested by PFGE was assigned to pulsotype I, a limited number of Spanish isolates were referred to as pulsotypes II and IV, the Russian F. tularensis subsp. holarctica strain isolated from a tick was assigned to pulsotype V, and the F. tularensis subsp. novicida and F. philomiragia strains were grouped into pulsotypes VI and VII, respectively.
AFLP analysis.All 62 strains were analyzed by AFLP analysis with four primer pairs: EcoRI-T and MseI-T, EcoRI-0 and MseI-CA, EcoRI-C and MseI-A, and EcoRI-A and MseI-C. These primers produced a total of 192 DNA fragments, of which 92%, or 177, were polymorphic for one or more strains. The patterns seen for F. philomiragia were quite different from those seen for the F. tularensis strains and contributed greatly to the percentage of polymorphic bands (data not shown). If the contributions from F. philomiragia are removed, there remained 68 polymorphic bands among the total of 125 bands, or 54%. Of these 68 polymorphic bands, a little more than one-half (n = 35) were due to F. tularensis subsp. novicida.
Figure 2 shows the dendrogram obtained after clustering by UPGMA when the data obtained with all four primer pairs were analyzed as a composite. A total of four primary clusters were observed among the 62 Francisella strains tested. The five F. tularensis subsp. tularensis strains were assigned to primary cluster A. Two subclusters were seen, one for isolate 62 (subcluster A2) and the other for the remaining F. tularensis subsp. tularensis isolates (subcluster A1). A genetic similarity of 91% was observed between these two subclusters. In addition, a minor variation between isolate 1 and the other three isolates in subcluster A1 was seen. Primary cluster B included all the F. tularensis subsp. holarctica isolates. Three major subclusters were detected within cluster B (with similarities among them of about 94%). Subcluster B1 contained four isolates recovered from monkeys and humans in the United States (greater than 96% similarity), subcluster B2 contained the Czech and Russian isolates (greater than 98% similarity), and subcluster B3 included the Spanish and French isolates, as well as an isolate (isolate 52) of unknown origin (greater than 99% similarity). The isolates in primary cluster B shared less than 90% similarity with the isolates in primary cluster A. The two F. tularensis subsp. novicida isolates, which were slightly different from each other, were assigned to primary cluster C and had about 72% similarity to the other F. tularensis subspecies. Finally, a genetically distant cluster (cluster D) was generated for F. philomiragia (less than 24% similarity to the F. tularensis strains). This species had just 15 AFLP bands in common with the other strains (data not shown).
Dendrogram showing the results of cluster analysis on the basis of the fingerprints of 62 Francisella strains obtained by AFLP analysis. The dendrogram was constructed by UPGMA clustering on a matrix based on the Dice coefficient. The banding patterns obtained with four primer pairs were combined for the dendrogram.
An unrooted maximum parsimony tree is presented in Fig. 3. Three primary branches from a common origin can be seen. One branch contains the F. tularensis subsp. tularensis (subclusters A1 and A2) and F. tularensis subsp. holarctica (subclusters B1, B2, and B3) clusters, the second branch contains the F. tularensis subsp. novicida (cluster C) cluster, and the third branch represents the unique species, F. philomiragia (cluster D). The lengths of the lines between the nodes represent the relative degrees of dissimilarity.
Unrooted maximum parsimony tree of the 62 Francisella strains. Five hundred bootstrap simulations were performed by using simulated annealing to optimize the topology. The bands obtained with all four primer pairs were used in the calculations. The lengths of the lines between nodes are relative degrees of dissimilarity. The labels represent the subclusters indicated in Fig. 2.
Comparative analysis of the results obtained by PFGE and AFLP analysis.A comparative analysis of the results obtained by PFGE and AFLP analysis for the 49 Francisella strains analyzed by both methods is presented in Table 1. The F. tularensis subsp. tularensis isolate belonging to PFGE cluster I had AFLP profile A1. The four strains yielding PFGE cluster II had AFLP profile B3. The 35 strains belonging to PFGE cluster III could also be grouped into AFLP profile B3, and 3 strains belonging to PFGE cluster III were assigned to AFLP profile B2. The three strains belonging to PFGE cluster IV had AFLP profile B3. Finally, the strains that were grouped into PFGE clusters V, VI, and VII had AFLP profiles B2, C, and D, respectively.
DISCUSSION
Although the genus Francisella comprises two species, almost all knowledge originates from studies with F. tularensis and, more concretely, from studies with two of the subspecies of F. tularensis (F. tularensis subsp. tularensis and F. tularensis subsp. holarctica), which are extremely infectious bacteria that cause the zoonotic disease tularemia. In this study, PFGE, AFLP analysis, and 16S rRNA gene sequencing were used as tools to characterize F. tularensis subsp. holarctica isolates of different origins and to compare them with other subspecies and species of the genus Francisella.
No nucleotide differences were detected among the amplicons of the 42 Spanish isolates by 16S rRNA sequencing. This single sequence matched base by base the 16S rRNA sequence of F. tularensis subsp. holarctica strain LVS (GenBank accession no. L26086 ), and therefore, this method confirmed the identities (at the species and subspecies levels) of the Francisella isolates recovered from patients involved in the tularemia outbreak that occurred in northwestern Spain between 1997 and 1999, but it did not allow us to discriminate individual isolates.
It is interesting that the sequence reported in this study was not identical to those recently obtained in Spain from a second outbreak of tularemia; that outbreak was associated with crayfish (Procambarus clarkii) fishing (2) and was unrelated to the first hare-associated outbreak, the isolates from which were tested in our investigation. The published sequences of the isolates involved in the second Spanish outbreak were reported to be 100% identical to that of F. tularensis subsp. holarctica strain LVS (2). However, our analysis showed that all these other sequences differed by one base from the sequence of the LVS reference strain, and, consequently, they also differed at the same base from the sequences of the 42 Spanish isolates listed in Table 1. This slight difference was found at a position equivalent to base 1066 of the published sequence of the LVS reference strain, for which the sequences reported by Anda et al. (2) lacked an A residue. This finding suggests that the isolates recovered from humans, hares, ticks, and voles involved in the first Spanish tularemia outbreak were different from those recovered from humans, crayfish, and water involved in the second outbreak. This means that two variants of F. tularensis subsp. holarctica are present in Spain, and so far they have been involved in completely unrelated outbreaks from an epidemiological point of view. In addition, the lack of variability in the 16S rRNA genes of Francisella species observed in previous studies (9, 15, 23) makes this finding of special interest.
The initial PFGE experiments were done with restriction endonucleases routinely used for gram-negative organisms (XhoI, SpeI, XbaI, SmaI, ApaI, BglII) (27). However, the results obtained were difficult to analyze because of either two few or too many bands, with the exception of the patterns produced by XhoI. Paradoxically, a high discriminatory power (four types among 46 F. tularensis subsp. holarctica isolates) was obtained after digestion with BamHI, an endonuclease normally used with gram-positive organisms. This fact could be explained by the low G+C contents of the DNAs of members of the genus Francisella (33 to 36 mol%), which are more similar to those of some gram-positive organisms (for instance, gram-positive cocci) than to those of other closely related gram-negative organisms.
The banding pattern generated for F. philomiragia isolates by PFGE was completely different from those generated for F. tularensis isolates, and similarly, the pulsotypes observed for the three subspecies within the latter species were easily distinguishable from each other (Fig. 1A and B), demonstrating the usefulness of this technique for the identification of Francisella strains to the subspecies level. Within subspecies, the PFGE profiles were quite homologous among the F. tularensis subsp. holarctica isolates analyzed. Isolates indistinguishable by their PFGE patterns were found in geographically unrelated areas (the Czech Republic and Spain). The existence of an identical genetic pattern for the isolates cited above might suggest that a close epidemiological relationship exists between them. This possible relationship was supported by AFLP analysis, even though they had different profiles (cluster B2 for Czech isolates and cluster B3 for Spanish isolates). These two clusters had 97% similarity, whereas there was only 94% similarity between the European and the U.S. isolates (cluster B1). The fact that Czech and Spanish isolates shared the same pulsotype but differed by AFLP analysis (about 3% diversity) could be explained by the higher discriminatory power of AFLP analysis compared to that of PFGE, a finding already reported for other organisms (5). In addition, the use of RAPD analysis with the universal M13 primer in an earlier study also allowed discrimination between Czech and Spanish isolates (4).
Some evidence for differentiation among isolates was seen after BamHI digestion of F. tularensis subsp. holarctica isolates; the Spanish isolates were grouped into three types. These results demonstrate that isolates characterized by different PFGE profiles may be present in a limited geographic area (Castilla y León, Spain; about 11,000 km2). The differences seen were not more than two or three bands for every pattern, and according to the system for standardization of the interpretation of PFGE patterns (27), these isolates should be regarded as closely related and considered subtypes. However, the original guidelines suggested by Tenover et al. (27) were meant to apply to smaller localized outbreaks, while geographically distributed populations of isolates were examined in our study. Therefore, a limited genomic divergence in F. tularensis subsp. holarctica isolates could be epidemiologically significant. For this reason, we believe that the differences in the BamHI pulsotypes, although slight, might be better interpreted as indicating different types rather than different subtypes, as previously reported for other organisms that tend to appear clonal, such as Escherichia coli O157:H7 (3). The genetic diversity seen by PFGE was also similar to that seen previously when these same isolates were studied by rep-PCR and RAPD analysis (4).
The degree of similarity observed for F. tularensis subsp. tularensis and F. tularensis subsp. holarctica by AFLP analysis was less than 90%, while the profiles seen with the F. tularensis subsp. novicida isolates were quite different (only about 72% homology) from those generated for the 59 strains of the other F. tularensis subspecies studied. According to Savelkoul et al. (24), the similarity results confirm that all these isolates belong to the same species. However, maximum parsimony analysis of the data obtained by AFLP analysis placed F. tularensis subsp. novicida on a branch separate from the branch containing F. tularensis subsp. tularensis and F. tularensis subsp. holarctica. On the other hand, the homology between F. philomiragia and the remaining isolates was very weak, less than 24%, a percentage considerably lower than the 40% that Savelkoul et al. (24) proposed as the lowest limit for isolates of the same genus but of different species. However, F. philomiragia was previously transferred from the genus Yersinia to the genus Francisella on the basis of the results of biochemical studies and DNA relatedness analyses, as well as those of cellular fatty acid analyses (14). This study, however, contained too few isolates from groups other than F. tularensis subsp. holarctica to reach any firm conclusion on the placement of F. tularensis subsp. novicida and F. philomiragia on the basis of the results of AFLP analysis.
Our study makes evident that a certain degree of genetic diversity exists among the five F. tularensis subsp. tularensis isolates studied, especially with regard to strain 6223 (isolate 62), whose sequence showed a 9% divergence from those of the other F. tularensis subsp. tularensis isolates tested. It is interesting that this strain is the only known avirulent strain within F. tularensis subsp. tularensis. The patterns generated by AFLP analysis for the F. tularensis subsp. holarctica isolates showed extensive similarity (about 94%), and they were assigned to a single primary cluster, with three major subclusters and minor variations within them. A slight difference could be observed for Spanish isolate 40 (about 1% divergence); however, the remaining 41 F. tularensis subsp. holarctica isolates recovered in Spain shared 100% identity by AFLP analysis and were closely related to French isolates. One additional strain of unknown origin (isolate 52) also clustered with this group. This strain is known only as Grousse but was provided to two of us (M.E.D. and T.L.H.) along with strains St. Germaine (isolate 47) and Chateneux (isolate 46), which were from villages of those names in France. It is possible, given the AFLP pattern, that strain Grousse is also from France since there exist French villages of that name.
The five F. tularensis subsp. holarctica strains isolated in the Czech Republic or Russia showed a 3% divergence compared to the Western European subcluster, and globally, all the European isolates were more closely related to each other than to the four isolates recovered from humans and monkeys in the western United States. The two European clusters showed a 6% divergence from the U.S. cluster. Although the sample size is small, we believe that these data support the conclusion that AFLP analysis can be used to identify regional variations among F. tularensis subp. holarctica strains.
AFLP analysis has established itself as a broadly applicable genotyping method with high degrees of reproducibility and discriminatory power (24). Its procedures are more labor-intensive than those of other PCR-based typing methods, which have shown equally high differentiation powers (4); however, it is more reproducible and has the advantages of using whole-genome analysis and automated data acquisition, as does PFGE. Finally, results are obtained faster by AFLP analysis than by PFGE (22), and unlike the latter technique, AFLP analysis generates polymorphism patterns based on the mutations not only in the restriction site but also in the sequences adjacent to the restriction sites and complementary to the selective primer extensions, as well as insertions or deletions within the amplified fragments (24).
The hare F. tularensis subsp. holarctica isolates tested in this study could be separated into three clusters (clusters II, III, and IV) by PFGE and into two subclusters by AFLP (subclusters B2 and B3) (Table 1 and Fig. 2). This fact confirms the existence of a certain degree of genetic diversity among the isolates recovered from this animal, a finding also observed when the same isolates were tested by different PCR methods (4). By contrast, although the number of strains was too limited to make epidemiological conclusions, minor diversity was observed among the human F. tularensis subsp. holarctica isolates. All of them belonged to a single cluster (cluster III) by PFGE and to two subclusters (subclusters B1 and B3) by AFLP analysis, unlike the results previously obtained by rep-PCR and RAPD analysis (4).
The two Spanish isolates obtained from ticks and the one isolate from a vole shared pulsotype III and AFLP subcluster B3. The Russian tick isolate yielded a unique pulsotype (pulsotype V) by PFGE, but the AFLP profile was shared with the Czech hare isolates. These isolates were also easily differentiated by the PCR approaches used in a previous study (4). The fact that pulsotype III and AFLP subcluster B3 included vole, tick, hare, and human isolates seems to suggest the transmission of tularemia among humans, wild animals, and their ectoparasites during the Spanish outbreak. In this respect, it must be pointed out that most of the human infections occurred during the hunting season, a finding also reported by Stewart (26).
In conclusion, PFGE and AFLP analysis appear to be excellent tools for the discrimination of Francisella species, subspecies, and, to a certain extent, individual isolates, as well as for the rapid analysis of tularemia outbreaks. Moreover, these methods allowed samples to be treated in such a way that laboratory-acquired disease can be avoided. Finally, the data reported in this study contribute to the knowledge of the epidemiology of tularemia in Spain, which is important for defining future control measures.
ACKNOWLEDGMENTS
We thank Laboratorio Central de Sanidad Animal, Ministerio de Agricultura, Pesca y Alimentación, Algete, Madrid, Spain; Servicio de Sanidad Animal and Laboratorio de Sanidad Animal de León, Consejería de Agricultura y Ganadería, Junta de Castilla y León, León, Spain; Laboratorio de Microbiología, Hospital Virgen de la Concha, Insalud, Zamora, Spain; Departamento de Microbiología Médica, Hospital Princesa Sofía, Insalud, León, Spain; and National Defense Research Establishment, Umea, Sweden, for providing us the Francisella strains. We also thank Jane Wong of the California Department of Health Services for providing two of us (M.E.D. and T.L.H.) with the U.S. isolates used in this study.
This work was supported by grant LE 04/00B from the Junta de Castilla y León, Consejería de Educación y Cultura of Spain.
FOOTNOTES
- Received 12 December 2001.
- Returned for modification 11 February 2002.
- Accepted 18 May 2002.
- Copyright © 2002 American Society for Microbiology
