ABSTRACT
Martha's Vineyard, Mass., has been the site of two outbreaks of tularemia (1978 and 2000). Although most patients from both outbreaks presented with pneumonic disease and although aerosol transmission has been suggested, the bite of a dog tick and exposure to rabbits remain the only proven modes of transmission. The factors that precipitated the tularemia outbreaks or the proximal determinants of human risk remain undescribed. We sought to test the hypothesis that the ongoing outbreak is due to a recent introduction event as opposed to amplification of a cryptic enzootic cycle. From 2001-2003, we collected 4,246 dog ticks and tested them in pools for evidence of tularemia by PCR. We then measured the genetic diversity of Francisella tularensis by using multiple-locus variable-number tandem repeat analysis. The prevalence of F. tularensis in dog ticks averaged 0.7%. From 29 positive pools, we identified 10 unique genotypes, which was an unexpectedly large degree of diversity (Simpson's index = 0.86). This degree of genetic diversity is inconsistent with a recent introduction event. We conclude that there has been long-standing enzootic transmission of tularemia on the island.
Martha's Vineyard, Mass., represents a unique site in the epidemiology of tularemia in North America. Although isolated cases of primary pneumonic disease have been reported elsewhere, no other site in the United States has reported two such outbreaks. In 1978, 15 tularemia cases were diagnosed there, including a cluster of 7 from a single household (35). Of these patients, 12 presented with pneumonia, but the mode of exposure remained undescribed. A second outbreak comprising mainly pneumonic patients occurred 23 years later. In 2000, 15 tularemia cases were diagnosed, 12 of which were considered to be primary pneumonic disease, including one fatality. A case control study suggested that increased risk was associated with lawn-mowing and brush-cutting (9, 10), but the nature of the fomites remains elusive. For both outbreaks, the only confirmed mode of exposure, for two patients per outbreak, was ulceroglandular disease due to the bites of dog ticks (Dermacentor variabilis). Also, a few patients (two and one) from each outbreak had exposure to a dead rabbit. The 2000 outbreak has continued to the present, with three to six cases reported each subsequent year (2001-2004).
Although the biology of the agent of tularemia (Francisella tularensis) has been intensively studied, little is known about how it perpetuates in nature, particularly for the main North American subspecies (known as type A or F. tularensis subsp. tularensis). The prevailing scenario is that the distribution of the disease is associated with that of cottontail rabbits (Sylvilagus floridanus) and related species and that zoonotic risk depends upon spillover from epizootics among these hosts and their ectoparasites (20). This dogma has been supported by the fact that until recently, nearly 90% of all reported U.S. cases were associated with rabbit hunting or exposure to rabbits (6). During the ongoing Martha's Vineyard outbreak, rabbit exposure does not appear to be a risk factor (9). Rabbit carcasses have not been frequently noted there during the outbreak, and other evidence of an active epizootic (other dead animals or a scarcity of rabbits) has not been documented by long-term tick-borne disease surveillance. The agent is known to infect diverse vertebrate hosts, however, including small rodents, birds, and carnivores, many of which, like rabbits, rapidly die as a result of infection (5). In addition, a variety of arthropods, including multiple genera of ticks (17, 18, 30) as well as deer flies, fleas, mites, and lice, have been found to be naturally infected or experimentally competent as vectors (11-13). Thus, it is possible that the perpetuation of F. tularensis type A is locally determined and includes cycles that do not depend on rabbits and that such cycles may serve as the basis for the current outbreak.
Identifying the mode of perpetuation of F. tularensis on Martha's Vineyard may provide clues to the proximal determinants of human risk there and, in particular, may help to identify the fomites that served as sources of the apparently aerogenic infections. Martha's Vineyard is an island, and thus it is likely that the agent of tularemia becomes locally extinct because of its propensity to rapidly kill vertebrate reservoir hosts. Episodic outbreaks would quickly follow the random introduction of F. tularensis-infected rabbits and bird-feeding ticks (Haemaphysalis leporispalustris or Ixodes dentatus) by migratory birds. Under such a hypothesis, molecular epidemiologic studies would tend to demonstrate a point-source pattern, with limited genetic diversity within F. tularensis isolates. Accordingly, we sought to test this hypothesis by describing the genetic diversity of F. tularensis on Martha's Vineyard. We sampled dog ticks, a demonstrated vector for F. tularensis, from multiple sites on Martha's Vineyard and analyzed them for evidence of infection by PCR. We then measured the genetic diversity of F. tularensis species infecting dog ticks by conducting multiple-locus variable-number tandem repeat analysis (MLVA).
MATERIALS AND METHODS
Tick collection.Host-seeking adult dog ticks (D. variabilis) were collected from nine locations on Martha's Vineyard, Mass., during the spring and early summer of 2001 to 2003 (Fig. 1). These ticks were collected by dragging a cloth over the vegetation. In some locations, dog ticks were also removed from anesthetized skunks and raccoons that were trapped for other purposes during 2001-2003. During 2003, adult deer ticks (Ixodes dammini) were collected from hunter-killed deer brought to the check station at the Massachusetts State Forest Headquarters.
Collection sites on Martha's Vineyard.
DNA extraction.Hemolymph from individual dog ticks was obtained by cutting a foreleg with scissors. Forceps and scissors were cleaned and flame sterilized between ticks to prevent cross-contamination between samples. A drop of hemolymph from each of six ticks was combined into a microcentrifuge tube with 50 μl of phosphate-buffered saline. Because of their small size, deer ticks were placed individually into wells of 96-well plates containing 50 μl of phosphate-buffered saline and were macerated by use of a glass rod. The rod was also sterilized between wells. Pools of six were created by using 10 μl of each sample. All tick pools were mixed with an equal volume of 4 M guanidium isothiocynate and were extracted by a traditional phenol-chloroform method. DNAs were resuspended in 50 μl of sterile water.
PCR analysis.Samples were screened for evidence of the tularemia agent by nested PCRs targeting the fopA gene, as described previously (14). Amplification was performed by the use of Qiagen Taq polymerase and buffers as recommended by the manufacturer. Ten microliters of the 409-bp product was visualized in a 1.5% agarose gel. Negative controls, physically separated lab areas, and dedicated pipettes and other tools were used to prevent and monitor for possible PCR contamination. Subspecies confirmation was obtained by amplifying the intergenic region between the peptidyl-prolyl trans-isomerase and putative RNA helicase genes (herein referred to as PPI-helicase) by using primers C6 (AGGCGGAGATCTAGGAACCTTT) and C8 (AGCCCAAGCTGACTAAAATCTTT) (A. Johansson, unpublished data). This yielded a variable-length amplicon due to a 30-bp deletion in F. tularensis subsp. holarctica: for F. tularensis subsp. holarctica, the amplicon was 220 bp, and for F. tularensis subsp. tularensis, the amplicon was 250 bp (GenBank accession no. AF524865) (23). Random samples of amplicons from all PCR assays were sequenced to verify the validity of the PCR. For sequencing, amplicons were excised from the gels, purified through spin columns (Qiagen), and sequenced at the University of Maine Sequencing Facility.
MLVA.Two short-sequence tandem repeat (SSTR) loci, SSTR9 (9-bp repeats) and SSTR16 (16-bp repeats), were amplified as described previously (22), with the modification that amplifications were performed directly from dog tick DNA extracts. Amplicons obtained from tularemia-infected cottontail rabbits, S. floridanus, found dead at one of our field sites were included in this analysis. Amplicons were run for 5 h at 90 V in Tris-borate-EDTA buffer next to a 20-bp ladder (Sigma) in a 2% gel made of super-fine-resolution agarose (Amresco). Because SSTR9 was previously shown to have some sequence diversity and since its amplicons were extremely close in size, all of these amplicons were sequenced. SSTR16 has not been shown to have sequence diversity, and the size differences of the amplicons were much larger. Therefore, only a representative sample of each size variant was confirmed by sequencing. The sequences for each gene were aligned by eye by the use of GeneDoc software (27), and the number of repeated units for each sample was counted. The SSTR type was simply the number of repeats for SSTR9 followed by that for SSTR16. Simpson's index of diversity was calculated as described previously (19).
RESULTS
Prevalence of F. tularensis in ticks.Over a period of 3 years, 4,246 dog ticks were assayed for evidence of infection with F. tularensis; 29 (0.7% minimum infection rate [MIR]) tested positive in the fopA PCR assay. All positive samples amplified the 250-bp piece of the PPI-helicase gene that identifies F. tularensis subsp. tularensis (data not shown). Furthermore, all samples sequenced, for both genes, were 100% identical to the sequence of strain Schu (AF240631) (the SchuS4 genome can be found at http://artedi.ebc.uu.se/Projects/Francisella/). Thus, dog ticks and rabbits on Martha's Vineyard are infected with F. tularensis type A.
Because the Francisella-like endosymbiont of D. variabilis (28, 29) is common in dog ticks on Martha's Vineyard (H. K. Goethert, unpublished data), all primers used for this study were tested for their capacity to amplify DNA from this agent. None of our assays amplified the endosymbiont of D. variabilis, and therefore our results were not confounded by this closely related bacterium.
None of the 110 pools of I. dammini, from 660 individual ticks, were infected with F. tularensis. A power calculation suggested that our sample size would be sufficient to detect an absolute prevalence of 0.6% at a power of 80% with an alpha value of 0.05, suggesting that if I. dammini contributes to the force of tularemia transmission on Martha's Vineyard, it does so no more than does D. variabilis.
We were able to detect F. tularensis in host-seeking dog ticks from virtually every field site surveyed at some point during the study (Table 1). However, the overall prevalence of infection was low. The average MIRs were 0.2, 0.3, and 1.1% in 2001, 2002, and 2003, respectively. An unusually large number of infected ticks (4% MIR) was detected from the field site near Squibnocket during the summer of 2003. This prevalence was significantly higher than that detected for the same site in previous years (χ2 test, P = 0.005).
MIRs in questing D. variabilis ticks
To determine whether skunks and raccoons contribute to the force of tularemia transmission, we compared the infection rates of ticks collected from animals with those of ticks flagged from the vegetation at the same site at the same time (Table 2). Although the MIRs did not differ among most of the field sites that we examined (P > 0.1), there was a significantly higher prevalence in host-seeking ticks than in those collected from animals for the field site near Squibnocket during 2003.
Comparison of MIRs of F. tularensis isolates from D. variabilis ticks collected from animals (skunks and raccoons) and from vegetation (by flagging)
MLVA.Two samples that tested positive for both the fopA and PPI-helicase PCRs failed to be amplified with the SSTR primers (SF80-2002 and SQ24-2003). A sequence analysis confirmed that they were indeed infected with F. tularensis, and we can only conclude that the bacterial copy number within these ticks was below the amplification efficiency of the SSTR PCR.
Ten unique SSTR types were identified for 27 positive tick pools that were identified during the course of the study (Table 3). The nucleotide sequences of SSTR9 and SSTR16 were invariant on the island. All polymorphisms detected were derived solely from the numbers of repeated units, that is, all SSTR9 loci had the same structure of repeats, A2B1En (A = AACAAAGAC, B = AATAAAGAC, and E = AATAAGGAT, as defined by Johansson et al. [23]). The sequence of the 16-bp repeating unit of SSTR16 was invariant, as described previously. Simpson's index of diversity was 0.86 for both loci together, 0.71 for SSTR16, and 0.83 for SSTR9.
SSTR genotypes identified on Martha's Vineyard
In total, 11 unique SSTR types were identified on the island. Two SSTR types, 22 10 and 9 9, were identified from more than one field site. These sites were not close to each other (Fig. 1). Multiple SSTR types were often identified from a single site. Furthermore, SSTR types 9 9, 9 13, 10 7, and 11 7 were identified for multiple years. The field site near Squibnocket yielded not only the highest prevalence of tularemia in ticks but also the most diversity. This one site yielded four (36% of the total) unique SSTR types (Tables 3 and 4). Two SSTR types were identified from rabbits that were found dead at this site, one of which, type 8 9, was unique. SSTR type 9 13 seems to be associated only with animals. It was only detected in rabbit spleens and in ticks collected from skunks and raccoons. In contrast, type 10 7 was detected solely in questing ticks. The significance of these SSTR-host associations remains unclear.
SSTR genotypes collected from the field site near Squibnocket
DISCUSSION
During our study, we were able to detect F. tularensis in dog ticks from multiple sites across Martha's Vineyard. The prevalence in ticks was low, generally with a <0.5% MIR at any given site. Despite this, 11 unique genotypes were identified from 30 positive samples by the use of only two SSTR loci. This amount of diversity, with a Simpson's index of 0.86, argues against a point source or recent introduction and is more consistent with the alternative hypothesis, that there has been long-standing enzootic transmission of tularemia on the island. In the only other analyses for which Simpson's indexes have been estimated (7, 22), SSTR9 yielded an index of 0.93 for 56 isolates from diverse sites across the western United States and of 0.95 for 39 isolates from across the world. Thus, the degree of F. tularensis genetic diversity that we detected on Martha's Vineyard was similar to that detected with random samples across continents. Furthermore, if F. tularensis was introduced just prior to the outbreak, it is unlikely that infected ticks would be distributed across the entire island, including Chappaquiddick, which is only connected to the main island by a sandbar during low tide (Fig. 1, see Poucha Pond and Wasque). Although we failed to always detect infected ticks from a given site each year, this may be because the sample size often lacked the power to detect infections with such a low prevalence.
It is not clear whether our findings are typical of a natural focus of tularemia, as molecular epidemiologic analyses of the agent are rare and have not examined infected ticks. However, nine tularemia isolates from Oklahoma (GenBank accession no. AF393400 to -6, AY37287, and AY393399, presumably all type A) were analyzed by use of the SSTR9 loci, although the strain designations for the sequences in GenBank do not match those within a published report (7), making it difficult to ascertain their origins. The diversity among these nine strains was much higher than that documented for the 27 infected ticks from Martha's Vineyard. All of our samples had the same motif of 9-bp repeat units, A2 B1 En. In contrast, the number of every repeated unit varied in the samples from Oklahoma, with the motif A1-4 B2-5 En. Thus, it may be that the genetic diversity of F. tularensis isolates on Martha's Vineyard is relatively limited or that even more diversity is related to areas where transmission may be even more long-standing, such as the American Midwest or locations where new variants may be introduced more easily than in an insular site. The question of what constitutes “typical” type A diversity will remain unanswered until ticks or animals from other North American sites are sampled and analyzed.
Strains of F. tularensis subsp. holarctica (type B) isolated from several Swedish outbreaks have been analyzed by MLVA (22). In sharp contrast to the diversity seen for type A, type B isolates appear to be much more homogeneous. When the same two SSTR loci as those used for this study were employed, only three SSTR types were described for 18 strains, despite sampling from a much larger geographic area and over a 17-year time span. It is unknown whether this difference in diversity is due to fundamental differences in the ecology of the two subspecies or to sampling bias, i.e., the Swedish samples were derived mostly from human cases and our samples were from ticks. Perhaps human infection tends to select for dominant genotypes, thereby causing the appearance of strain homogeneity.
Further support for the hypothesis that type B isolates are less genotypically diverse than type A isolates comes from an analysis of seven type B isolates from a mixture of wild and captive animals collected during an outbreak in Maricopa County, Ariz., in 2000 (7). MLVA using six loci found all isolates to be identical. Although type B appears to be common in Arizona and other states west of the Mississippi River, we noted that the PCR assay that was used for that report to assign the isolates to type B may not have been definitive. The C1 and C4 primers, which target PPI-helicase, are said to have produced a 300-bp amplicon that was consistent with type B isolates (7). However, when we aligned these primers with the deposited sequences in GenBank (accession no. AF524865), it was apparent that a 300-bp amplification product is impossible. Indeed, a 149-bp product would be produced from DNAs of type B organisms, and a 179-bp product would be produced from type A DNAs (21).
Although tandem repeat regions are known to be highly mutable, they are stable enough to be successfully used for strain typing for many other bacterial species (16, 24, 33, 36). Preliminary experiments with laboratory strains of F. tularensis have demonstrated that SSTR loci remain stable after numerous passages on plates and through laboratory animals (22). Our data are consistent with the stability of these markers: similar SSTR types were identified from multiple samples over a period of 3 years and from various sites.
We have not documented type B or Francisella philomiragia from our samples, and we note that MLVA should be used with caution for assigning F. tularensis subspecies. Type A and type B were consistently distinguished by the concurrent use of six SSTR loci (7), but none of the loci could be used on its own to discriminate the two subspecies. Complementary methods must be used for specific identification. We have deliberately not isolated viable bacteria from our tick samples to date so as to maintain a minimal inventory of these select agents. An isolate from the Martha's Vineyard fatality in 2000 was said to ferment glycerol (9). Thus, although we cannot exclude the possibility that type B isolates may be present on Martha's Vineyard, to date all available evidence suggests that the outbreak comprises only type A isolates.
The apparent genetic diversity that we detected among our Martha's Vineyard samples raises interesting questions about the population genetics of F. tularensis. Tularemia appears to be a relatively recently introduced disease in Massachusetts. Prior to 1937, a sole case had been recognized, which was related to the handling of contaminated rabbit meat imported from the Midwest (2, 26, 37). However, beginning in 1937, >20,000 cottontail rabbits from Missouri and Kansas were introduced into Massachusetts (including Martha's Vineyard) for sporting purposes (2). This act prompted a prescient editorial in the New England Journal of Medicine (25) in which the authors predicted the emergence of tularemia in Massachusetts as a result of the introduction “folly.” The first case appeared within the same year that the rabbits were released (1). Subsequently, tularemia was considered endemic to the area, but rare: from 1953 to 1977, only 20 cases were reported to the Department of Public Health (35). Thus, at least 11 genotypes have arisen within or been maintained for 70 to 80 years on a 100-square-mile island. If we model perpetuation on Martha's Vineyard as episodes of island-wide epizootics, followed by low-level enzootic transmission or extinction within most island sites, then genetic bottlenecking during interepizootics would be expected to reduce genotypic diversity. Our results would then require that diversity be generated quickly during epizootic amplification (within the span of a year or two). Alternatively, the natural nidality model (31) suggests that F. tularensis is continuously maintained in cryptic enzootic cycles in very small foci, perhaps on the order of a few dozen square meters; epizootics or human risk becomes apparent as amplification occurs within foci, leading to the coalescence of foci and spillover. Diversity is maintained through a metapopulation of natural foci. Although further work is required, our observations favor the second hypothesis.
Dog ticks, not deer ticks, appear to perpetuate F. tularensis on Martha's Vineyard. We were unable to identify any infected I. dammini ticks, despite testing >600 ticks. This is paradoxical inasmuch as immature stages of both ticks often share the same white-footed mouse host on Martha's Vineyard. To date, we have found no evidence that white-footed mice are involved in the enzootic cycle. Because we were able to detect infection in host-seeking (i.e., nonfed) adult dog ticks, these ticks must have acquired infection either during their previous blood meals (transstadial transmission) or by inheritance (transovarial transmission [TOT]). Both modes of transmission have been experimentally demonstrated with D. variabilis (32). Unfortunately, we were unable to assess the efficiency of TOT in our ticks, as none of the infected females collected from animals had fed long enough to lay eggs.
We have determined that skunks and raccoons are frequently seropositive for tularemia (S. R. Telford and Z. L. Berrada, unpublished data), which demonstrates their exposure and perhaps their capacity to serve as reservoirs. If skunks and raccoons are competent reservoirs, their peridomestic habits may provide urine or fecal fomites that may serve as the basis for inhalational exposure for landscapers. To determine whether they may serve as a source of infection for adult ticks and thereby facilitate TOT, we compared the infection rate for adult ticks attached to skunks and raccoons with that for nonfed host-seeking adults from the same sites. We found no significant difference between the prevalence of infection in ticks collected from animals and the prevalence in those collected from the vegetation (Table 2), and in fact, for the field site near Squibnocket, host-seeking ticks were more frequently infected than those collected from skunks and raccoons. Thus, assuming that the hemolymph assay did not fail to detect newly acquired (tick midgut) infections, D. variabilis ticks did not acquire infection from these animals. It may be the case that circulating antibodies to F. tularensis in seropositive animals can clear the ticks of their infection and that younger seronegative animals may be competent reservoirs. This phenomenon was described by Bell (3), who allowed infected D. variabilis ticks to feed on vaccinated animals and showed that the ticks lost their infection. A similar phenomenon occurs with Lyme disease spirochetes, wherein anti-OspA antibodies circulating within a mouse host reduce that host's competence as a reservoir (4). Whether skunks and raccoons can be incompetent tularemia reservoirs for ticks but can serve as the source for fomites (via their excreta) remains unknown. Skunks and raccoons, however, at the very least help to maintain F. tularensis on Martha's Vineyard because dog tick egg production critically depends upon their blood.
The prevailing dogma is that the distribution and transmission of type A tularemia mainly depend on cottontail rabbits. We were unable to effectively sample rabbits on Martha's Vineyard because their population densities are not as high as in other coastal New England sites, such as nearby Nantucket, where we can easily attain a 10% capture rate with live traps (15). Although there is no direct evidence implicating rabbits as the main human risk factor during the ongoing outbreak, clearly these hosts have been locally involved in an epizootic. We collected six rabbits from our site near Squibnocket; three of them were dead or dying and yielded evidence of infection by F. tularensis. MLVA suggested that there is an interchange between a putative dog tick-driven enzootic cycle and activity in rabbits because the same genotype, SSTR type 9 13, that was identified in rabbits in 2001 was subsequently identified in dog ticks attached to a raccoon in 2002 and to skunks in 2003. Whether the human outbreak initially depended upon a rabbit epizootic or proceeded as a result of amplification within dog ticks and their hosts remains to be described.
Tularemia on Martha's Vineyard is unique because a large proportion of patients present with pneumonic disease. It is the only place in the United States where clusters of pneumonic tularemia cases have been identified. There is significant debate as to whether the pneumonia is primary, due to inhalation of the bacterium from an unknown source, or secondary, due to dissemination of the bacterium after exposure by an arthropod bite or some other route (8, 34). In either case, a hypothesis that must be critically tested is whether F. tularensis variants that are more prone to disseminate to the lungs after an arthropod bite or are more environmentally resistant and capable of aerogenic infection have evolved on Martha's Vineyard. Isolation tends to produce insular endemic forms of animals. Martha's Vineyard is no exception: the white-footed mouse there is a subspecies (P. leucopus fusus Bangs 1905) that is distinct from that found elsewhere in New England (P. leucopus noveboracensis). Whether the human cases from Martha's Vineyard are associated with certain genotypes or represent the spectrum of diversity detected in nature is not known. Unfortunately, only two isolates have been obtained from patients on the island, and these were not examined in our study. Our MLVA of F. tularensis isolates from Martha's Vineyard, however, suggests that there may be some host-associated strain specificity: type 9 13 was only found to be associated with animals, and type 10 7 was only found in questing ticks. These and others of our molecularly defined strains can now be experimentally compared to determine whether there are differences in various virulence-associated parameters or in host ranges.
ACKNOWLEDGMENTS
This work was supported by NIH grants R21 AI 053411 and AI 39002 and by CDC grant U50/CCU119560.
Anders Johansson kindly provided unpublished primer sequences. This work was facilitated by the support of the residents of Martha's Vineyard, particularly John Varkonda and William Searle of the Massachusetts Departments of Environmental Management and Environmental Protection, who provided logistical support. We are grateful for their assistance.
FOOTNOTES
- Received 9 April 2004.
- Accepted 10 July 2004.
- American Society for Microbiology
