Ecology of Francisella tularensis

Abstract

Tularemia is a Holarctic zoonosis caused by the gamma proteobacterium Francisella tularensis, and is considered a vector-borne disease. In many regions, human risk is associated with the bites of flies, mosquitoes, or ticks. But, the biology of the agent is such that risk may be fomite related and large outbreaks can occur due to inhalation or ingestion of contaminated materials. Such well documented human risk factors suggest their role in the enzootic cycle as well. Many arthropods support the growth or survival of the agent, but whether arthropods (ticks in particular) are obligately required for the perpetuation of F. tularensis remains to be demonstrated. As with most zoonoses, our knowledge of the ecology of F. tularensis has been driven with the objective of understanding human risk. In this review, we focus on the role of the arthropod in maintaining F. tularensis, particularly with respect to longterm enzootic persistence.

“Tularemia is a specific infectious disease due to Bacterium tularense and is transmitted from rodents to man by the bite of an infected bloodsucking insect or by the handling and dissecting of infected rodents by market men or laboratory workers.” [42]

1. Introduction.

During plague surveillance in 1909, an epizootic in ground squirrels from Tulare County, California yielded bacteria that did not have the characteristic “safety pin” morphology of plague bacilli [96]. The bacterium was quickly cultivated and named “Bacterium tularense” [97]. Edward Francis of the U.S. Public Health Service, investigating an outbreak with human cases presenting with ulceration, lymphadenopathy, and fever in Utah residents bitten by deer flies incriminated “B. tularense” as the etiologic agent and proposed the name “tularemia” for the disease [42]. His comprehensive investigations included isolating the agent from flybitten humans, from jackrabbits, and from ground squirrels. Francis also provided experimental evidence for transmission of tularemia by the bites of deer flies, mouse lice and bed bugs [42]. Parker and colleagues isolated “B. tularense” from the Rocky Mountain spotted fever vector, Dermacentor andersoni [111], and the related American dog tick, D. variabilis, was incriminated as the source of a human infection in Minnesota [57], thereby formally incriminating ticks as a risk to humans other than the known deer flies. Francis requested that each state’s department of health report any case of tularemia and determined that >90% of >6000 tularemia case reports from 1924–1935 were associated with exposure to cottontail rabbits or hares [48], mainly as a result of market production of rabbit meat but also due to rabbit hunting. His specific mention of laboratory workers in his pithy summary of tularemia is because all 6 of the USPHS staff working on the newly recognized infection became infected [83], including Francis himself. The ease with which laboratory workers became infected by manipulating cultured bacteria or infected animals (the agent was easily propagated by serially transferring infected tissue homogenates to uninfected rodents) became notorious. British bacteriologists, after receiving reference cultures from Francis, became infected and decided to cease work with “B. tularense” because of its great hazard [138].

Very soon after Francis’ seminal work, Hachiro Ohara in Japan described 10 cases of an acute febrile disease that had been acquired during the skinning of wild rabbits [47]. The disease had been noted in the Fukushima area for 20 years but had not drawn attention by clinicians. Rabbit die-offs had been frequently noted by villagers. Definitive proof of rabbits as a source of infection was provided when Ohara’s wife volunteered to be exposed: she acquired tularemia after tissues from a dead rabbit were rubbed on the back of her hand. The clinical details provided by Ohara agreed “even in minor details with the details of tularemia” [47], sera sent by Ohara to Francis in the U.S. agglutinated American “B. tularense” strains, and fresh tissues from a Japanese case sent to Francis produced typical tularemic lesions in laboratory rodents.

Tularemia has had a long history in the former Soviet Union and much knowledge remains in their literature, sadly in Russian and thus less accessible to most Western workers. Pollitzer [125] provides a comprehensive review of the historical Soviet epidemiological literature and the reader is directed to that extraordinary book. To summarize from that authority, tularemia was first definitively identified in the Soviet Union when in 1929 Zarkhi, a clinical researcher in Sverdlosk, sent his own serum to McCoy at the U.S. Public Health Service. Zarkhi had suffered what he suspected was tularemia after necropsying guinea pigs inoculated with bubo contents from patients of a water rat-associated outbreak of a mystery disease. Pollitzer pointed out that the Soviet scientists retrospectively made the connection between outbreaks of “Siberian ulcer” in the 18^th and 19^th centuries (which was originally considered to represent cutaneous anthrax) and tularemia given the low (2%) mortality associated with some outbreaks of Siberian ulcer; the case fatality rate for anthrax would be much greater. Interpretation of historical outbreaks suggested that tularemia had long afflicted people throughout the vast Soviet Union from west to east and north to south, and thus the infection was very old and not recently introduced, for example, with the liberation by the Soviet fur industry of up to 80,000 muskrat from 1928–1945 [27]. Muskrats have been associated with tularemia outbreaks across the Holarctic.

2. Characteristics of the causative agent.

The genus Francisella comprises gram negative coccobacillary, non-spore forming, aerobic/microaerophilic bacteria in the Order Thiotrichales. Isolates with growth or biochemical characteristics similar to those from humans suffering from tularemia have been made from marine and freshwater fish, brackish water and other environmental sources, and ticks. Non-tularemia patients, often immunocompromised, have also yielded novel isolates that classically group with Francisella spp. Whole genome analyses demonstrate that there are at least a dozen distinct species [24]. Three subspecies of F. tularensis are currently recognized: tularensis, holarctica, and mediasiatica. Differences in distribution, ecology, biochemistry, and virulence led to the seminal classification of tularemia into distinct types [107]. Type A organisms (now known as F. tularensis tularensis) are prevalent in North America but not in Eurasia, are frequently transmitted by ticks, and may cause severe disease. Type B (F. tularensis holarctica; F. tularensis palaearctica of some older literature) causes episodic outbreaks (epizootics) in beavers, muskrats, and arvicoline rodents in either North America or Eurasia, may be isolated from water or soil, and may cause a milder disease [74,114]. F. tularensis subsp. mediasiatica was isolated from rodents in Central Asia but there are no reports of human cases. F. tularensis mediasiatica has apparently been isolated from diverse ticks from most metastriate genera as well as Ixodes persulcatus [149]. F. novicida has been considered a subspecies of F. tularensis but this bacterium is a distinct species [79]. F. novicida is an opportunistic pathogen, mainly of immunocompromised or elderly individuals exposed to environmental sources; isolates have only been made from salt or brackish water, never from animals or arthropods. However, it is highly virulent for experimental rodents, causing typical tularemia and can be acquired by feeding, survive transstadially, replicate within and be transmitted by the bites of D. andersoni [132].

Genetic typing methods confirmed the phenotypic distinctions and identified distinct lineages within Type A and Type B. Multilocus variable number tandem repeat analysis split the former into A.I. and A.II. Virtually all isolates from the eastern and central U.S. were A.I. and those from west of the 100^th meridian were A.II. Whole genome sequencing demonstrates that there are 3 lineages of A.I.[17], with A.I.12 being the most widely distributed, suggesting an adaptive advantage over the others. Four major lineages are apparent across the Holarctic Type B isolates [144] with considerable diversity of subpopulations/genotypes even within endemic counties in Sweden [76]. Newer typing methods using single nucleotide polymorphisms are consistent with MLVA but lineages are designated slightly differently (e.g. A.I. is A1).

3. Evolution of Francisella.

The genus Wolbachia was erected by Marshall Hertig to describe bacteria that he and Burt Wolbach (who described the rickettsial etiology and pathology of Rocky Mountain spotted fever) found in the reproductive tract of Culex pipiens in 1924 [60]. Subsequently, a variety of endosymbiotic intracellular bacteria, most identified only by microscopy of stained smears of arthropod tissues, were considered “wolbachia”. Wolbachia persica was isolated from Argas arboreus ticks collected from an Egyptian heron rookery [140] based on its morphological similarity and presence in Malpighian tubules and reproductive tissues. W. persica was pathogenic for guinea pigs, mice and chicks, but not rats or rabbits. Molecular phylogenetic analyses clearly demonstrated that W. persica was not a rickettsial agent but was closely related to F. tularensis [102,103] and not to W. pipientis, the bacterium first identified by Hertig and Wolbach. W. persica has been formally reclassified in the genus Francisella [85]. Francisella-like endosymbionts (FLE) have been identified from most tick genera [32, 53, 71,103, 133, 137, 141, 142]. Early analyses suggested that tick-FLE associations were ancient [103] and thus F. tularensis may have evolved from FLEs. However, co-speciation between Dermacentor spp. and their FLEs was not detected [137] and critical virulence genes for F. tularensis such as type IV pili, LPS, and the type VI secretion system are present but have become pseudogenes in the FLEs [51]. Accordingly, FLEs were acquired by ticks from a virulent F. tularensis-like ancestor. If F. tularensis did not originate with ticks, the presumed central role of ticks for maintaining this infection requires critical analysis.

4. Epidemiology of tularemia.

Tularemia is endemic throughout most of Europe, northern and central Asia and North America. Type B is found in most of Europe, Asia and North America. Type A is found exclusively in North America; there is one report of its identification in Europe but this finding remains enigmatic. F. tularensis mediasiatica has been identified in limited portions of Central Asia. Although tularemia was not previously thought to be present in the southern hemisphere, in 2011 3 human cases were identified in Tasmania, and isolates made from ringtail possums (Pseudocheirus peregrinus) were determined by sequencing to group with Type B strains from Japan (sometimes referred to as biovar japonica)[33].

The clinical presentation of tularemia is varied, with 6 classical forms having been described: ulceroglandular, glandular, oculoglandular, oropharyngeal, typhoidal, and pneumonic. Such presentations generally relate to mode of acquisition; oculoglandular and oropharyngeal, for example, are the result of conjunctival contamination and ingestion, respectively. Ulceroglandular presentation, with ulceration at the site of an arthropod bite or other portal of infectious entry as well as proximal lymphadenopathy, or glandular presentation with regional lymphadenopathy, are the most common, comprising 50–65% of all American cases [40, 154]. The other common form, typhoidal, has an acute onset with sore throat, high fevers, chills, and enteric symptoms. No ulcer or portal of entry is evident, nor is there lymphadenopathy. Pneumonic tularemia may comprise primary inhalational tularemia with a respiratory portal of entry [129], exemplified by laboratory accidents [109] in the years before universal laboratory adoption of biosafety cabinets; and by large agriculturally-related outbreaks [28]. However, pleuropneumonia is a common finding in advanced typhoidal tularemia [8, 21,41], and in 10–20% of ulceroglandular cases. Pneumonic tularemia, presumably primary inhalational, is becoming increasingly reported in the U.S. and in some sites may be more common than ulceroglandular [117]. In Europe, ulceroglandular and glandular presentations comprise 50–70% of reported cases (e.g., [37, 92].

In Europe and across Russia and the Federation of Independent States (due to the seminal contributions of Soviet scientists, “Soviet Union” will be used herein), the epidemiology of tularemia is diverse, reflecting the diversity of the land, fauna, and people [61]. Mosquitoes are strongly associated with tularemia in Sweden [136]. Oropharyngeal tularemia is the most commonly reported form in some countries (Bulgaria, Kosovo, Norway, Serbia and Turkey), thought to be due to contamination of water or food by rodents. Hunting, mainly of hares, is the source of ulceroglandular or glandular infections in the Czech Republic, France, Germany, Slovakia, and Spain. Ticks are not considered as a common source of infection in Europe [61].

In the Soviet Union, 6 epidemiological scenarios were recognized [93]: (1) Meadow-field type, with the vole Microtus arvalis as amplifying host and D. reticulatus the vector and interepizootic reservoir. Risk to people was associated with agriculture and contamination of drinking water. (2) Steppe type, with diverse rodents and hares and the vector D. marginatus; agricultural activity and contaminated water. (3) Forest type, with the vector I. ricinus and its main subadult hosts (Myodes spp., Apodemus spp.); people were infected during hunting hare as well as by tick bites. (4) Floodland-swamp type. Arvicola terrestris, the water rat, was the critical amplifying host, with Dermacentor, Rhipicephalus and Ixodes ticks as vectors. Mosquitoes may also infect water rats. Humans become infected by hunting water rats. (5) Foothill-brook type. Water rats are also the main host, with Ixodes apronophorus as the vector. (I. apronophoros also plays an important role in maintaining Omsk hemorrhagic fever virus in the same sites; [80]). People get infected by the contamination of running water during the summer by water rats. (6) Desert river type. Hares and gerbils and R. pumilio are thought to maintain infection, with risk exclusively to hunters. These Eurasian epidemiologically-based scenarios demonstrate the possible great diversity of enzootic cycles that may characterize F. tularensis elsewhere.

5. Rabbit fever.

The perception that Type A tularemia was due to lagomorphs was largely the influence of Francis himself and also of William Jellison of the Rocky Mountain Laboratories, who compiled and interpreted the existing literature on tularemia biology in a seminal monograph [73]. Jellison argued that human risk and geographic distribution of North American tularemia was strongly associated with cottontail rabbits [75, 113]. Cottontail rabbits are very susceptible to infection by Type A, dying within 7 days, and are large enough animals to attract attention when there is an epizootic, making them good sentinels for transmission activity. Furthermore, because of their value as food, their populations were a focus of attention by local residents and by state game management divisions: 25,000,000 rabbits were killed annually with a value of $5,000,000 during the 1920s [59]. Then too, tularemia cases were common in the northcentral states where there was a tradition of rabbit hunting [155]. Tularemia incidence in the U.S. started to diminish in the 1960s [18], perhaps as a result of a loss in popularity of rabbits as food and of hunting in general.

In Japan, tularemia risk is mainly associated with rabbits [104]. 93% of 1358 cases analyzed from 1924–1994 were considered to have been due to rabbit exposure. In addition, a large spike in the incidence of tularemia from 1945–1955 was attributed to soldiers resettling in their homes eating rabbits as a source of protein in economically challenging postwar times; the incidence started falling dramatically with robust growth in the Japanese economy.

Rabbits or hares clearly have an important role in the epidemiology of tularemia, either as the direct (meat or hunting) or indirect (contamination of the environment by carcasses) source of human infection, or draw attention to epizootics by easily noted die-offs.

6. Ecology of tularemia: general comments.

Tularemia usually comes to our attention as a result of epizootics (spillover from local amplification with clusters of human cases) but endemic tularemia is underappreciated. The Midwestern U.S. reports a fairly constant (limited interannual variation, no more than 2 fold) number of cases each year (largely tick-transmitted), as does the island of Martha’s Vineyard, Massachusetts (about equal numbers of tick transmission and inhalation) [94]. Sweden also has constant annual numbers of mosquito-transmitted infection [36] and that country and Finland report more tularemia cases than any other [61]. In other countries, clinical reporting might be less of a priority and thus it is possible that other countries than Sweden sustain a similar degree of tularemia endemism. Reported cases, however, reflect only an effective epidemiological bridge from an enzootic cycle (endemic transmission) or epizootic (spillover from massive time-limited amplification of the enzootic cycle).

The classical theory of natural nidality [116] posits that most zoonotic agents exist in longstanding foci that comprise optimal physical (weather, geology) and biological (fauna, flora) associations and that humans only become aware of their existence when they intrude. A scenario for F. tularensis perpetuation may comprise a system (metapopulation) of small “natural foci”, each with a prevalent variant and mode of transmission. This fundamental tenet of the ecology of vector-borne agents asserts that such pathogens are perpetuated within specific sites, often on the order of tens of square meters in area, and that transmission may be continuously detected there for decades or longer. Humans become exposed by stumbling into such a natural focus or when transmission risk becomes more homogeneously distributed over a wide area due to many such foci coalescing as a result of local amplification and spillover. We identified one such natural focus on Martha’s Vineyard and demonstrated that infected American dog ticks were found mainly in a 260m diameter site along a longterm transect and that the most genetic variability of F. tularensis along the transect was to be found in ticks in that microfocus [54]. Great genetic diversity was apparent on Martha’s Vineyard as a whole, as much at the time as that described for all existing global isolates [52]; F. tularensis exists there as a metapopulation of isolated microfoci [55]. How such microfoci remain genetically discrete on an island of about 1000 square kilometers is unclear.

Mathematical models suggest that “The maintenance of indirectly transmitted infections do not require the very large host populations that are needed for directly transmitted microparasite infections” [95]. Given the virulence of F. tularensis for rodents or lagomorphs, with the majority of infected animals dying rapidly, some indirect factor (vector, fomite) is required for the bacterium to persist over ecological time. It seems unlikely that long term persistence is due to direct contact between infected and uninfected vertebrate hosts, although this mechanism certainly has the capacity to dramatically amplify transmission during an epizootic given that excreta can be infectious and cannibalism of animals dying from tularemia may result in new infections. Without a persistent reservoir in a vector or fomite, or chronic infection of a longer lived vertebrate, generation of susceptible hosts by immigration or recruitment would need to at least equal that of the removal of infected hosts through death.

Is there a main theme for perpetuation (ticks and rabbits, for example) with regional ecological differences being variations on that theme? Even with the limited literature, it is not a good assumption that the ecology of Type A is similar to that of Type B. Soviet scientists described at least 6 different kinds of natural foci: floodplain/swamp; meadow/field; forest; steppe; water/river; and desert floodplain [106]. F. tularensis Clades A.I. and A.II. are genetically and generally geographically distinct, suggesting that they occupy different ecological niches or were isolated during glaciation [78]. It may be that the diverse sublineages of Type A and Type B [81, 151] might each have differing requirements for maintenance. On the other hand, both A.I. and A.II. were isolated from lagomorphs in one discrete site in a Utah fly-transmitted outbreak, demonstrating co-circulation [118]. This finding does not preclude more than one kind of natural focus within a geographically discrete area.

6.1. Associations with arthropods.

Although tularemia is considered a vector borne disease, an obligate role for arthropods, hematophagous species in particular, in maintaining F. tularensis in nature is not axiomatic. Ticks are considered as main vectors and likely reservoirs for F. tularensis in North America and much of Eurasia [73, 116], but ticks have not been incriminated as relevant to the enzootic cycle in Scandinavia, which reports more tularemia cases than any other country in Europe (although this is due in part to the likely high proportion of cases that are reported). It is possible that many kinds of arthropods, perhaps even nonhematophagous species, may contribute the perpetuation of F. tularensis over the long term in natural foci.

6.2. Associations with vertebrates.

F. tularensis has been isolated or detected from a very large number of animals, including amphibia, birds, rodents, lagomorphs, carnivores, and ruminants [22, 67]. Virtually all of these are considered incidental hosts, although if they die of sepsis due to F. tularensis and there is an appreciable bacterial burden in their tissues, they may contribute to short term maintenance and perhaps over the long term if decomposing tissues can remain infectious under certain conditions. The carcass of a mouse dying of most of the main tularemia lineages may have 9–11 log colony forming units (cfu) within the spleen or liver [100]. Interestingly, rats (Rattus norvegicus) can recover from infection and viable bacteria may be recovered from their tissues for at least 2 weeks after recovery [30, 31] and their role as possible enzootic reservoirs should be critically examined given their global distribution and abundance. Other animals, such as the carnivores or raptors, are typically poorly susceptible to disease (seroconverting with infection) and are excellent sentinels given their scavenging and predator roles [15]. An exception is the domestic cat, which is commonly infected (as many as 12% are seropositive in some sites [91]), suffers disease, not infrequently fatal, and in fact serves as an occupational risk to their owners and to veterinarians [84]. Given the abundance of feral cats [87], at least in the U.S., their potential role in maintaining natural foci should be analyzed further.

Some evidence suggests the possibility that some strains (lineages) of F. tularensis are specifically associated with certain mammals. The prevalence of human cases in the U.S. overlaps with the distribution of cottontail rabbits [113], and Type A strains were most likely to have been isolated from these hosts [81]; A.I. strains came from eastern cottontails (Sylvilagus floridanus) and A.II. from desert cottontails (S. audubonii). However, there does not seem to be any specific relationship between a vertebrate species and lineages of Type B [124]. Of course, existing F. tularensis strains do not represent a random sampling and reflect convenience samples from clinical cases or those from outbreak investigations. Efforts should be made to isolate additional strains from a greater selection of vertebrate species before any conclusions can be made.

6.2.1. Type B ecology: driven by rodents?

Tularemia in Eurasia and non-rabbit or tick associated infection in North America seem to have a strong environmental basis, acquired from agricultural activities such as hay threshing; from water contaminated by muskrats or water voles; hunting hares; or during the trapping of furbearers [3, 61,116, 134, 145]. A large typhoidal (=pneumonic) tularemia outbreak in Castilla y Leon, Spain was associated with farm and harvest activities, as well as contact with voles [3]. A prior large outbreak in the same region had been due to hunting hares, and subsequent genetic analyses of strains isolated from both outbreaks demonstrated strong similarities [7]. The outbreak of 2007–2018 and a lesser one in 2014 coincided with irruptions of common voles, Microtus arvalis [89] in Castila y Leon. American, Soviet, and Swedish workers have long recognized harvest activities and rodent irruptions as major risks for tularemia [68, 77, 114]. Experimental studies with American voles suggested the possibility that some infected animals developed a chronic nephritis and bacteruria that could serve as a protracted source of environmental contamination [12], a finding that was confirmed for voles that were orally infected [108].

7. General comments on tularemia vector studies.

Rodents are exquisitely sensitive to infection by F. tularensis, dying within the week of a sepsis that is characteristically identified by gross lesions on necropsy [30]. Tularemic pathology includes prominent, often caseous lymphadenopathy and readily visible “pinpoint white spots” on the surface of the liver and spleen, representing necrotic foci with masses of bacteria. Thus, early transmission studies either allowed potential vectors to feed on a rodent, or homogenized vectors and inoculated the homogenates into rodents, then waited for a characteristic rapid death and easily scored gross pathology. Good confidence can be placed in these assays even today in the times of ultrasensitive molecular diagnostic methods. Definitive assays rely on recovery of the agent by cultivation, or its safer surrogate (necropsy of tularemic animals is hazardous and propagation in vitro is even more so), evidence of bacterial DNA by PCR. PCR, however, must be done with stringent contamination control to prevent false positives, and usually fails to discriminate viable bacteria from DNA remnants. Strong evidence for viability is thus presented with results from animal inoculation or cultivation but PCR or immunofluorescence methods may have other interpretations. Accordingly, studies from the older literature can be considered to have used sound diagnostic assays and should not be discounted due to their “age”. However, older experimental studies used challenge strains that were uncharacterized and even whether they comprised Type A or B is not clear; exceptions are [29, 64] who used strain “Sm” which is the current Type A standard, Schu.

There is much literature on field surveys for F. tularensis in vector arthropods but other than definitively establishing the presence of a transmission cycle, sometimes very little can be concluded from such data. Removal of hematophagous arthropods from a host rarely allows for any conclusive evidence of vector competence because infection may be present in freshly ingested blood as opposed to having been retained from a previous bloodmeal. Nest parasites such as mites or fleas or lice, of course, must frequently feed on a host and many do not survive long without a bloodmeal. The critical question for the enzootic cycle is whether bacteria remain viable within such arthropods, live or dead, in the absence of the host, implying that a new host acquiring them (or eating them, or becoming exposed to their products, such as “flea dirt”) could become infected. PCR is almost exclusively used today to detect microbial agents during vector surveys, but the specificity of the primer sets used for detecting F. tularensis varies. The great diversity of Francisella spp. that is now known may confound interpretation of some of the previously reported prevalences reported from tick surveys, particularly those that used PCR primers that have not been directly tested against newly recognized Francisella spp. [24].

7.1. Fleas.

During the first investigations of the biology of F. tularensis by McCoy and Chapin [97], fleas were implicated as maintenance vectors. Both “Ceratophyllus acutus” (Diamanus montanus) and C. fasciatus were experimentally infected by feeding on tularemic guinea pigs and ground squirrels, but infectivity for more than a day or two after feeding was not tested. In addition, although “100 fleas” removed from a guinea pig that had died of tularemia were placed in a clean cage with a healthy squirrel, resulting in death of the squirrel, it was not clear whether that squirrel acquired infection by flea bite or by ingesting groomed fleas. Of course, either mode may be effective in enzootic maintenance. Experimental evidence suggests that fleas (Xenopsylla cheopis and Diamanus montanus) may acquire bacteria from infected mice and retain viable infection for more than a month [128], but did not transmit by feeding. Earlier studies [115] with 3 additional fleas, including Pulex irritans, demonstrated survival for only a day and no transmission. Larval fleas fed cultivated F. tularensis could retain infection for no more than 3 days and did not become infected by feeding on dried blood that had been spiked with culture [62]. Surveys of diverse flea species demonstrated natural infection [58, 152 and others summarized in 135] but on the other hand, early surveys for plague in the western U.S., using the technique of inoculating flea homogenates into rodents, rarely found F. tularensis [5], suggesting that fleas were a very minor contributor to the maintenance of this infection, at least in the American West. Hence, the published information on the role of fleas as enzootic vectors remains inconclusive.

7.2. Lice.

The rabbit louse (Haemodipsus ventricosus) transmitted to experimental rabbits but transfer of lice from an infected animal to an uninfected animal needed to occur within 3 hours otherwise infection appeared to have been lost [45], although a few rabbits became infected with lice that had been held for 2–3 days. Francis was very careful to exclude the possibility that infection was due to contamination with secretions or excreta of dying rabbits by placing hair with lice from the dying donor onto naïve rabbits and placing the rabbit within newly cleansed trash cans. Briefer but similar experiments by Francis’ team using Polyplax serratus also demonstrated transmission even a week after lice had been removed from the infected hosts. Francis was careful to not state that infection was due to bites by the lice (recognizing that mice will groom and eat lice), but simply pointed out that mouse infestation led to transmission under conditions that excluded a healthy mouse’s contact with the excreta of infected mice. Price [126, 127] experimentally infected human body lice (Pediculus humanus corporis) by feeding them on rabbits that had intravenously received a large dose of cultivated F. tularensis just before serving as host, as well as a cohort infected by introcoelomic inoculation. Serial sections of infected lice were examined to determine the course of infection over time. Interestingly, there was relatively little multiplication of bacteria in those infected by feeding (ingestion of 6 log cfu and measurements of not much more than 6 log cfu days thereafter). Those inoculated with 3 logs cfu demonstrated 3 log multiplication within 4 days, with quick progression to mortality. Price suggested that when bacteria remained confined to the gut, lice were more likely to survive but that nutritional factors in the hemolymph allowed for rapid multiplication and toxicosis quickly leading to louse death. Given the host specificity of most lice, the fact that new hosts become infested only by very close contact (lice do not persist in fomites), and the very short life of lice in the absence of a host, at most lice can help to amplify infection during an epizootic but would not maintain the agent once all the hosts for that louse species died. Of course, the lice themselves would soon become locally extinct if their hosts were not present.

7.3. Bedbugs.

Bedbugs (Cimex lectularius) were fed on infected mice and guinea pigs, and 3 modes of transmission were confirmed by Francis and Lake [44]: (1) by interrupted feeding (bug removed from infected animal before repletion and allowed to reinfest a naïve animal); (2) by bite after as many as 71 days (infection by bite ensured by allowing bugs to feed on mouse tails; and (3) by allowing mice to eat bugs infected as many as 100 days previously. The possible role of cimicids or triatomines as potential interepizootic hosts has not been explored.

7.4. Flies.

Tularemia was first described as a nosological entity by Francis because of a deer fly transmitted outbreak in Utah. Using field collected Chrysops discalis, Francis successfully transmitted infection by the bite of flies that had fed for short durations (“interrupted feeding”) on an infected guinea pig or rabbit and hours to days later were fed on uninfected guinea pigs, which died of typical tularemia [46]. His team subsequently assayed flies that had fed on infected guinea pigs on a daily basis, injecting fly homogenates into uninfected guinea pigs. “Up to 5 days the flies remained constantly infected” but Francis argued that infection tended to become lost by day 10 after infection and thus there was likely no replication. Hence, flies were considered to be mechanical vectors. Pavlovsky [116] clearly recognized the vector role of tabanid flies and in fact equated their role in tularemia maintenance with their role in that for anthrax: that flies would aggregate on moribund animals and spread the infection by contaminated mouthparts and interrupted feeding on new hosts. Soviet workers incriminated numerous species in the genera Tabanus, Chrysops, Stomoxys, and Haematopota [135, cited in 49]. Interestingly, reports of fly transmitted tularemia are very rare other than in the American west, and indeed Jellison [72] went so far as to state that the only species of any importance as a vector was C. discalis. From our own work, F. tularensis DNA may rarely be found in C. vittatus in our Martha’s Vineyard site and we know of an ulceroglandular case acquired from a deer fly bite on Nantucket (unpublished). It is likely that fly-transmitted infection may be found in most endemic sites but that human ulceroglandular cases are automatically considered to be a result of tick bite in the absence of dermal exposure to contaminated materials.

7.5. Mosquitoes.

Philip and colleagues [122] first examined the possible transmission of F. tularensis by bite, using local species of Aedes as well as A. aegypti. Mosquitoes were infected by feeding on tularemic guinea pigs. No mosquito transmitted by bite during a second bloodmeal, but a small proportion did so after interrupted feeding (this was interpreted as mechanical transmission). Swatting infected mosquitoes onto the skin of guinea pigs also transmitted infection. The excreta deposited after an infectious bloodmeal was infectious when inoculated into guinea pigs. They also noted that F. tularensis survived in dead mosquitoes for at least 4 days, and speculated that this could serve as a means to contaminate bodies of water, reminiscent of Manson’s classic findings with filariasis [25]. Soviet workers regarded mosquitoes as tularemia vectors but of minor importance to the maintenance of natural foci compared to that of ticks and tabanids [116].

Mosquitoes are strongly implicated as vectors in Sweden, given that tularemic ulcers are most frequently found on the upper back, neck, and ears of case-patients [36], where mosquitoes are more likely to feed. In addition, the agent has been isolated from mosquitoes there [105], and mosquito larvae collected from endemic sites that were allowed to develop and emerge as adults in the laboratory contained F tularensis DNA [88]. Host seeking mosquitoes collected from endemic Swedish sites yielded 18 of 791 positive for F. tularensis holarctica DNA [147]; infection was detected in Culex, Aedes, Anopheles, and Coquilletidia spp. In a definitive experiment, second instar A. aegypti larvae were exposed to 7 log cfu of Type B, then washed and allowed to develop to adults. A quarter of the adult mosquitoes contained F. tularensis DNA, and of those that did, the homogenates from a third of them infected mice when intraperitoneally inoculated, demonstrating that virulent bacteria were transstadially passed [9]. Larval Culex quinquefasciatus readily fed on biofilms of F. tularensis holarctica, but their pupation was delayed and emergent adult mosquitoes were smaller than those feeding on the control dog biscuit slurry [90]. Additional studies are needed on the vectorial capacity of mosquitoes for F. tularensis, particularly to clarify the mode of transmission; to date studies have generally failed to transmit by bite and the leading hypothesis is that people become infected by swatting a feeding mosquito and contaminating their skin. Short lived antibody to mosquito salivary [86] or gut proteins might be sought in tularemia patients to further confirm mosquito transmission. Although experimental mosquito related transmission rates appear to be small, the sheer abundance of mosquitoes in nature would compensate for low probability events. Given the evidence that humans acquire infection from mosquitoes, it is a certainty that other animals do as well in enzootic sites.

The role of mosquitoes may be specific to some natural foci. Over 9000 mosquitoes were screened with negative results from a site in the Czech Republic where as many of 2% of host seeking ticks that were concurrently collected yielded F. tularensis isolates [69].

7.6. Mites.

Mesostigmatid mites (Hirstionyssus, Laelaps, Eulaelaps, Haemolaelaps) were found by Francis and Lake [45] and Soviet workers cited in [49] to be naturally infected by F. tularensis. Painstaking studies done by Cluff Hopla for his doctoral dissertation [63] demonstrated the likely role of hematophagous mites in maintaining F. tularensis. Using Ornithonyssus bacoti, the tropical rat mite, Hopla determined that protonymphs acquired infection from septic mice, bacteria were transstadially passed into the nonfeeding deutonymph, and could be transmitted by the adult mite to clean mice. Clear evidence of transovarial transmission was also provided, although by only about a fifth of the adult mites feeding on infected mice. Transmission was not by bite but rather required grooming and eating the mites by the mice. Assays for infection comprised both mouse inoculation as well as cultivation on glucose blood agar. Cultivation allowed for estimation of bacterial burdens, which on average were greater than 6 log cfu per mite even after “prolonged fasting”. Given the great diversity of blood feeding mites, some effort should be made to better describe the vector-pathogen associations in the laboratory and in nature.

7.7. Soft ticks.

Davis [29] determined that Ornithodoros turicata and O. parkeri could remain infected by F. tularensis for over 600 days, but failed to transmit infection by bite. Detailed studies of infected O. moubata, O. parkeri, O. hermsi, and O. turicata, including infection with a known Type A strain (Schu) found viable bacteria for 450 days and demonstrated viable F. tularensis in rectal secretions, coxal fluid, as well as transmission by bite [20]. Contamination by coxal fluid appeared to be the most likely means of transmission inasmuch as O. hermsi, which does not secrete liquid coxal fluid during feeding, was the least competent vector. No evidence of infection related mortality was reported, despite numerous bacteria colonizing virtually every tick tissue. The fact that soft ticks can maintain viable infection for such extended durations suggests that they may help maintain a tularemia natural focus where they are endemic (southern, southcentral, and western U.S.; central Asia).

7.8. Hard ticks

There are many American reports of the vector competence of the ixodid ticks Dermacentor variabilis, D. andersoni, D. parumapertus, Amblyomma americanum, and Ixodes scapularis were all competent for F. tularensis [2, 11, 64, 65, 66, 111, 123]. Soviet workers, particularly Petrov and Dunaeva [119] and Petrov [120, 121], cited in [10], demonstrated intense multiplication and survival of the agent as long as 700 days within D. reticulatus. Balashov [10] stated that ticks are “the most effective natural vector and reservoir” given the many demonstrations of transmission by feeding, intensive multiplication of the bacterium, long survival within ticks with no loss of bacterial viability or virulence, and frequent detection of infection in surveys of host seeking ticks. The multiyear life cycles of most ixodid ticks make them a logical candidate for longterm persistence of F. tularensis natural foci.

The competence of D. variabilis for F. tularensis was measured in two exemplary modern studies [130, 131] which set the standard for any future work. Realistic doses (100cfu) were used to infect mice. Two Type A (A1b, A2) and one Type B strains were compared. Experimental ticks were from a longstanding laboratory colony with precise provenance, and known to be specific pathogen free. Uninfected cohorts were generated from the same tick batches for comparison with the infected ones. Sample sizes were such that statistical comparisons were adequately powered. Assays for infection were definitive: cultivation of the agent from individual ticks. The methods were presented in such detail that the studies could be replicated by other laboratories. The experiments demonstrated that A2 strains caused nymphs to die quickly, but not adult ticks. Infection affected nymphs infected by all 3 strains in some way, either by reducing body size, attachment or feeding success. The A1b strain was not transmitted to mice by infected nymphs, and transmission was poor for A2 and B strains (8%−12% success). Mice could become infected by eating the infesting infected nymphs. Although adult D. variabilis never feed on rodents, this host restriction was overcome by confining ticks to capsules on mice and of those female ticks that fed and were shown to have contained infection, transmission occurred 58%−89% of the time. Accordingly, the competence of one characterized strain of D. variabilis differed depending on the infecting F. tularensis type, but all 3 types were transmitted by bite of adult ticks and 2 by that of nymphs. F. tularensis was inferred to have been transmitted within one day of attachment by infected ticks based on the observation that some mice were diseased within 4 days of repletion; this was calculated using a known prepatent period of 3 days, and a feeding duration of 7 days for female dog ticks.

Many surveys of host-seeking ticks have been published. As with many other tick borne infections, there is no standardization of assays; with PCR assays, specificity should be demonstrated to prevent detection of FLEs or environmental Francisella spp. Sample sizes are often such that confidence intervals around the estimated prevalence are very wide. Hence, it can be difficult to compare published prevalence estimates. Two very detailed surveys for F. tularensis infection in host-seeking ticks provide examples of what would be most informative. Nearly 8000 D. reticulatus were sampled from a natural focus in the Czech Republic during 1995–2013 [70]; 64 F. tularensis isolates were recovered by mouse inoculation for a minimum infection rate of 0.83%. The bacteriological gold standard is isolation of the agent. Our published [52,55] and unpublished observations of the Martha’s Vineyard natural focus during 2001–2011 comprise PCR detection in ticks (median of 1572 host seeking adult D. variabilis tested each year) with a median annual prevalence of about 3.1% (range 0–5.2). Two gene targets were used (fopA for the initial screen and tul4 for a confirmatory assay) and a large proportion of those samples with specific DNA were genotyped [52,55]. Detection of genetic material, however, does not establish viability.

7.8.1. Transovarial transmission.

The literature conflicts with respect to the inheritance of F. tularensis by ticks. Three reports [64, 65, 112] demonstrated inherited infection and that this infection would pass transstadially to the adult when larvae and nymphs were fed on uninfected hosts. Parker and Spencer [112] definitively demonstrated inheritance of the progeny of 2 of 15 female D. andersoni feeding on infected rabbits; the evidence comprised transmission by bite of larvae or nymphs. Another 6 of the 15 were suggestive of inheritance but typical infection (death of the host) was not demonstrable; infection was inferred only by transfer of splenic material from the rabbits fed on by the progeny to uninfected animals. Calhoun and Alford [23] found host seeking A. americanum larvae to be infected (by inoculating animals with homogenate pools). In contrast, Soviet workers suggest that any demonstration of transovarial infection was due to contamination of eclosed larvae by secretions of an infected female tick [121] and dismiss the possible contribution of inheritance by the tick as a means of F. tularensis perpetuation [116]. Bell [11] failed to demonstrate inheritance in D. variabilis. More recent analyses [50] did not find transovarial transmission from female D. reticulatus and I. ricinus, with bacteria appearing to die within previtellogenic oocytes. They noted that 17%−30% of all engorging female ticks died or did not lay eggs. It may be that the efficiency of the process may vary among strains of F. tularensis and could even be associated with tick genetic background (co-adaptation of vector and pathogen).

7.8.2. Paradox of fitness effects due to infection.

Philip and Jellison [123] first reported that ticks infected by the agent of tularemia were likely to die or fail to oviposit. A recent thorough analysis [130] clearly documents that infection of D. variabilis by Type A diminishes nymphal survival; infected nymphs were generally smaller and took twice as long to feed to repletion as uninfected nymphs. Interestingly, these negative fitness effects were not seen with adult dog ticks [131]. Type B-infected nymphal D. marginatus or D. variabilis die more rapidly than do uninfected ticks [130] and in fact only 2% of all D. marginatus feeding as larvae on infected animals developed to the adult stage [120, 121]. Host seeking adult D. variabilis from our Martha’s Vineyard site that died faster in captivity were more likely to contain F. tularensis DNA than those living longer [56]. Diminished longevity of ticks would impact reproduction, which implies that vector competence for Type A should be selected against, yet, at least in our Martha’s Vineyard natural focus, dog ticks containing F. tularensis DNA have been found every year for more than 15 years (unpublished). Like Rickettsia rickettsii, which is also generally a lethal infection for D. andersoni [102], the continued demonstration of infected tick vectors in the field suggests that despite F. tularensis negatively affecting fitness, there is some as yet unidentified compensatory effect that maintains the enzootic vector-pathogen relationship. For example, F. tularensis genotypes or lineages might be specifically coadapted to local tick populations and the elegant experiments of Reese et al. [130, 131] should be repeated using diverse D. variabilis colonies and with A. americanum, the two main zoonotic vectors for Type A.

7.8.3. “Allergic klendusity”.

During vector competence studies of D. variabilis, Bell [11] noted that a rabbit fed upon by F. tularensis infected ticks failed to become infected. That rabbit had previously been fed on by uninfected ticks, and he speculated that sensitization interfered with transmission. In a subsequent experiment, it was found that rabbits previously fed on by D. andersoni (usually larval infestation) were half as likely to be infected by F. tularensis infected ticks [13]. Bell designated this “allergic klendusity” (“disease escaping ability”) and that it was due to anti-tick immunity, referring to Trager’s classic experiments [150]. This interference was manifested at the portal of entry, working for the two possible modes of transmission, viz., by bite of infectious tick or by contamination of the bite site with tick feces containing F. tularensis as the tick is feeding. He also suggested that such a mechanism might serve to limit natural epizootics. This paper serves as the seminal observation that is the basis for current efforts in anti-tick vaccines and their promise for protecting against tick-borne infection.

7.9. Diverse experimental arthropods.

Advances in our understanding of and development of arthropod models for innate immunity have led to some interesting infection models of F. tularensis. Drosophila [101, 153], dubia roaches [34], and caterpillars (Galleria mellonella, [6,148]; Bombyx mori [143]) have been experimentally infected by inoculation of cultivated F. tularensis, typically with dose dependent mortality that was more pronounced at mammalian temperatures than arthropod (lower) temperatures, although B. mori survived for at least a week. Antimicrobial peptides such as those in the imd/relish pathway appear to be activated and prevent overwhelming sepsis; flies with that pathway knocked out succumb rapidly to infection [101]. Melanization was inhibited in B. mori inoculated with live F. tularensis but not with dead bacteria, suggesting some active bacterial response or secretion inhibits innate immunity. F. tularensis mutants with critical virulence genes (as determined for mammalian infection) also survived longer, suggesting that at least some known virulence factors apply to both mammals and arthropods. The new experimental hosts confirm what has been known since Francis’ first investigations: that F. tularensis has an extremely wide arthropod host range to include acarines, dipterans, lepidopterans, hemipterans, anoplurans, and siphonapterans and thus the infection should not be considered to be limited to solely vector species.

7.10. Aquatic invertebrates.

Shrimp or snails could retain viable organisms for 20 days [99], and indeed, invertebrates were first described as contributing to F. tularensis survival within water by Soviet scientists: Pavlovsky [116] states that the “tularemia microbe…may be found in the bodies of… mollusks, crabs and crayfish, water bug larvae…” Crayfish have been associated with human infection and F. tularensis apparently infects them [4], suggesting that additional surveys of aquatic invertebrates using modern methods are warranted in known natural foci.

8. Are vectors really required for maintenance?

Although 10–100 colony forming units of Type A are sufficient to produce a lethal infection in most mice with most F. tularensis strains when delivered by aerosol, and 1000 cfu is a typical LD50 for parenteral inoculation [146], LD50 is 6 log cfu for oral infection (gavage). Mice survived oral infection with 4 log cfu [82], suggesting that eating materials contaminated with F. tularensis (via excreta on food, grooming ectoparasites, cannibalizing moribund animals) could maintain transmission during epizootics and perhaps even over the long term within natural foci (e..g, by eating infected starved bed bugs or soft ticks, vide supra). Cannibalism of moribund cagemates is well known as a mode of transmission for Type A in the laboratory [110] as well as Type B [108]. Of particular interest is the possibility that voles became partially immune due to low level infection resulting from cannibalism of tularemic carcasses and that this immunity allows survival of the vole during subsequent infection, increasing the probability for shedding in the excreta [12]. Such immunity suggests a means of regulating the duration of epizootics and for the development of a new endemic focus.

Water-borne tularemia was first described by Karpoff and Antonoff [77], in their description of an outbreak related to drinking unboiled brook water; 43 hay-harvesters had evidence for hyperaemic oral mucosa, tonsils, or conjunctiva. Brook water readily yielded isolates of F. tularensis when inoculated into guinea pigs. In the U.S., cold waters contaminated by muskrat and beaver repeatedly yielded isolates of F. tularensis [74, 114]. Infected carcasses contaminated water and stored in the cold provided tissues that infected animals after two weeks; naturally contaminated mud remained infectious as long as 10 weeks [114]. Surface water and sediment yielded indisputable DNA sequence evidence of contamination with Type B in Swedish endemic sites, even in years with little epidemiological activity [19]. On the other hand, Type B held in lake water for 120 days failed to kill mice at a dose ten times the typical LD50 for that Type B strain [146], implying a loss of virulence that was not protected by the presence of high nutrient levels or freeliving protozoa. About half of rodents immersed in contaminated water became infected with exposure to as few as 100–1000 cfu/mL[116], which seems like heavy contamination but the spleen alone of a mouse dying of tularemia may have 10 log cfu [100]; Pavlovsky [116] estimates that a single dead rodent could infect 500,000 liters of water. Indeed, many have speculated that environmental persistence depends on continual contamination of the environment by dead animals. The persistence of F. tularensis in non-aquatic environments has received little attention; the bacterium has been lyophilized in a protein matrix and remained viable for 4 years in ampules stored on a desktop [98] suggesting the possibility for longterm persistence in natural foci. We speculated that a key factor in the endemic pneumonic tularemia focus of Martha’s Vineyard, in which landscapers comprise a major risk group [38], is the exposure of soils there to oceanic salt sprays [16] that would promote the viability of contaminating bacteria.

Bacterial endobiosis with free-living amoebae or other cyst-forming protozoa could serve to contaminate the environment with viable bacteria for a longer duration than if the bacteria were present in an extracellular or naked form. Although F. tularensis is said to be environmentally resistant, the naked bacteria are fragile and do not survive in the laboratory for more than 2–3 weeks in spring water or saline [16]; in another study, though, Type B survived 70 days in tap water at 8°C [39]. F. tularensis infects amoebas and ciliates in the laboratory [1, 14, 35] and can enter a viable but noncultivatable state [39]. Viable but nonculturable states might imply the possibility of longterm persistence in the environment with reversion to replicating, infectious bacteria.

9. Future directions for research on tularemia ecology.

Some questions will be very difficult to answer, given the difficulty in finding longstanding natural foci and the inability to experimentally manipulate conditions in the field. How long does infection persist in carcasses or remnants thereof when placed within suitable substrate? Does survival vary according to site physiography or even more specific microhabitat requirements (salinity, soil acidity or other chemical attribute)? The classification of F. tularensis as a U.S. federal Select Agent (of interest to biodefense and hence any possession or manipulation is highly regulated by the government)(https://www.selectagents.gov/regulations.html) makes even the most basic of deliberate field experiments impossible to undertake in the U.S. It would be virtually impossible to bury fresh infected mouse carcasses or infected ticks in a field site (a site that was known to be endemic and using a F. tularensis strain derived from that site) and sample them periodically to determine the duration of bacterial viability because of fears that they might be hijacked for nefarious purposes.

In the U.S., much attention has been focused on ticks; can Type A be enzootic where there are few ticks, as demonstrated for Type B in Sweden? Can a natural focus disappear with continued application of acaricides to reservoirs or to the environment? Long years of antitularemia campaigns by the Soviets rarely attacked just the ticks but relied on human vaccination, rodent elimination and general sanitation [125] but failed to eliminate natural foci.

In this review, we argued that our knowledge of the ecology of tularemia is incomplete mainly because past studies have focused on the zoonotic condition as opposed to identifying the requirements for maintenance. Of course, human exposure (the subject of epidemiology) may provide clues to the mode of perpetuation (ecology) but this is not axiomatic. Zoonotic infections may exist in sites with no implied human risk in the absence of an effective epidemiological “bridge”. Even the very concept that F. tularensis is an obligate vector-borne infection remains to be proven: its ecology may be more like that of Coxiella burnetii, the agent of Q fever (with diverse modes of perpetuation), than of Rickettsia spp.