Anaplasma phagocytophilum—a widespread multi-host pathogen with highly adaptive strategies

The bacterium Anaplasma phagocytophilum has for decades been known to cause the disease tick-borne fever (TBF) in domestic ruminants in Ixodes ricinus-infested areas in northern Europe. In recent years, the bacterium has been found associated with Ixodes-tick species more or less worldwide on the northern hemisphere. A. phagocytophilum has a broad host range and may cause severe disease in several mammalian species, including humans. However, the clinical symptoms vary from subclinical to fatal conditions, and considerable underreporting of clinical incidents is suspected in both human and veterinary medicine. Several variants of A. phagocytophilum have been genetically characterized. Identification and stratification into phylogenetic subfamilies has been based on cell culturing, experimental infections, PCR, and sequencing techniques. However, few genome sequences have been completed so far, thus observations on biological, ecological, and pathological differences between genotypes of the bacterium, have yet to be elucidated by molecular and experimental infection studies. The natural transmission cycles of various A. phagocytophilum variants, the involvement of their respective hosts and vectors involved, in particular the zoonotic potential, have to be unraveled. A. phagocytophilum is able to persist between seasons of tick activity in several mammalian species and movement of hosts and infected ticks on migrating animals or birds may spread the bacterium. In the present review, we focus on the ecology and epidemiology of A. phagocytophilum, especially the role of wildlife in contribution to the spread and sustainability of the infection in domestic livestock and humans.


INTRODUCTION
The bacterium Anaplasma phagocytophilum has been known to cause disease in domestic ruminants (Europe) (Foggie, 1951) and horses (USA) (Gribble, 1969) for decades. More recently, the infection has been detected in several mammalian species, including humans, in areas on the northern hemisphere with endemic occurrence of Ixodes ticks. A. phagocytophilum as a bacterial species appears to be a generalist, infecting a wide range of animals. Multiple genetic variants of the bacterium have been characterized (Scharf et al., 2011) and subpopulations within the species are now being discussed. In this review, we present updated information especially concerning the ecology and epidemiology of A. phagocytophilum.

HISTORY
During an experimental study on louping-ill (LI) in Scotland last century, some sheep contracted an unknown fever reaction on tick-infested pastures. The fever reaction was transmitted to other sheep by blood inoculation, but gave no protection against a later LI-virus infection. The disease was given the provisional name "tick-borne fever" (TBF), and the responsible pathogen was assumed to belong to the class Rickettsia (Gordon et al., 1932(Gordon et al., , 1940. The name TBF is still used for the infection in domestic ruminants in Europe. Anecdotally it could be mentioned that the Norwegian synonym of TBF is "sjodogg," and this name was already used to describe a devastating illness in ruminants as early as year 1780 in a coastal area of western Norway (Stuen, 2003).
The causative agent of TBF was first classified as Rickettsia phagocytophila (Foggie, 1951). However, due to morphological resemblance with Cytoecetes microti, an organism found in the polymorphonuclear cells of the vole Microtus pennsylvanicus (Tyzzer, 1938), it was later suggested to include the TBF agent in the genus Cytoecetes in the tribe Ehrlichia, as C. phagocytophila (Foggie, 1962).
In 1974, the organism was named Ehrlichia phagocytophila in Bergey's manual of determinative bacteriology (Philip, 1974). The discovery of E. chaffeensis in 1986, causative agent of human monocytic ehrlichiosis (Maeda et al., 1987;Anderson et al., 1991), and the agent of human granulocytic ehrlichiosis (HGE) in 1994 Chen et al., 1994), initiated new studies on the host associations, epidemiology and taxonomy of the granulocytic Ehrlichiae (Ogden et al., 1998). Genus Ehrlichia was divided into three genogroups, of which the granulocytic group contained E. phagocytophilum, E. equi [described in horses (Gribble, increased aspartate and alanine aminotransferase activity in sera are reported (Bakken and Dumler, 2008). However, most human infections probably result in minimal or no clinical manifestations. Reports from the US, indicate a hospitalization rate of 36%, of which 7% need intensive care, while the case fatality rate is less than 1% (Dumler, 2012). A recent cohort study from China however, describes a mortality of 26.5% (22/83) in hospitalized patients (Li et al., 2011).

CLINICAL SIGNS
Clinical signs in ruminants may be sudden onset of high fever (>41 • C) and drop in milk yield, while symptoms in horses, dogs, and cats may be more vague and unspecific. In humans, a flulike symptom 2-3 weeks after tick exposure is an indicator of infection. However, laboratory confirmation is required to verify the diagnosis (Woldehiwet, 2010). To our knowledge, chronic infection has not yet been confirmed in any host, although persistent infections have been found to occur in several mammalian species.

DIRECT IDENTIFICATION
Light microscopy of blood smears taken in the initial fever period is normally sufficient to state the diagnosis. Stained with May-Grünwald Giemsa, the organisms appear as blue cytoplasmic inclusions in monocytes and granular leucocytes, especially neutrophils (Foggie, 1951). Electron microscopy may also confirm the diagnosis of acute Anaplasma infection in blood or organs. Single or multiple organisms are then identified in clearly defined cytoplasmic vacuoles (Tuomi and von Bonsdorff, 1966;Rikihisa, 1991). Immuno-histochemistry on tissue samples could also be performed to confirm the diagnosis (Lepidi et al., 2000).

POLYMERASE CHAIN REACTION (PCR) AND CULTIVATION
Several PCR techniques (conventional, nested, and real-time) for the identification of A. phagocytophilum infection in blood and tissue samples have been established primarily on basis of the 16S rRNA, groEL, and p44 genes Courtney et al., 2004;Alberti et al., 2005a). Multiple variants of A. phagocytophilum have been genetically characterized. Identification and stratification into phylogenetic subfamilies have been based on cell culturing, experimental infections, PCR and sequencing techniques (Dumler et al., 2007). Cultivation of A. phagocytophilum in cell cultures has been described for variants isolated from human, dog, horse, roe deer, and sheep Bjöersdorff et al., 2002;Woldehiwet et al., 2002;Silaghi et al., 2011c).

SEROLOGY
The presence of specific antibodies may support the diagnosis. A complement fixation test, counter-current immunoelectrophoresis test and an indirect immunofluorescent antibody (IFA) test can be used (Webster and Mitchell, 1988;Paxton and Scott, 1989). Several ELISA tests have also been developed (Ravyn et al., 1998;Magnarelli et al., 2001;Alleman et al., 2006;Woldehiwet and Yavari, 2012). A SNAP®4Dx® ELISA test is commercially available for rapid in-house identification of A. phagocytophilum antibodies in dog serum, but the kit has also been used successfully on horse and sheep sera (Granquist et al., 2010a;Hansen et al., 2010).

PATHOLOGY
An enlarged spleen, up to 4-5 times the normal size with subcapsular bleedings, has for decades been regarded as indicative of TBF in sheep (Gordon et al., 1932;Øverås et al., 1993). No other typical pathological changes have been described (Munro et al., 1982;Campbell et al., 1994;Lepidi et al., 2000). An enlarged spleen with subcapsular bleedings has also been observed in roe deer and reindeer (Stuen, 2003).
Relative sensitivity of the diagnostic tests used for laboratory diagnostic confirmation of A. phagocytophilum infection in humans is shown in Table 2.

TREATMENT, PREVENTION, AND CONTROL
The drug of choice is tetracycline (Woldehiwet and Scott, 1993;Dumler, 1996). Doxycyclin hyclate, given orally or intravenously, has been effective in treating clinical cases of human granulocytic anaplasmosis, and has led to clinical improvement in 24-48 h. In human patients, treated with doxycycline for 7-10 days, infections have resolved completely and relapses have never been reported. In patients at risk of adverse drug reactions, rifampin therapy should be considered (Bakken and Dumler, 2006). Current disease prevention strategies in domestic animals are based on the reduction of tick infestation by chemical acaricides, for instance at turn out on tick pasture. This is mostly done be dipping or with a variety of pour-on applications (Woldehiwet and Scott, 1993;Stuen, 2003). This treatment has to be repeated during the tick season. In the UK, long-acting tetracycline has also been used as a prophylactic measure given before animals are moved from tick-free environment into tick-infested pasture (Brodie et al., 1986;Woldehiwet, 2007). However, there is a growing concern about the environmental safety and human health, increasing costs of chemical control and the increasing resistance of ticks to pesticides (Samish et al., 2004).
Biological tick control is becoming an attractive approach to tick management. Biological control of tick infestations has been difficult because ticks have few natural enemies. Studies so far have concentrated of bacteria, entomopathogenic fungi, and nematodes (Samish et al., 2004). However, the main challenge is to create a sustainable biological control of ticks in the natural habitat. Vaccines against A. phagocytophilum are not yet available. Several vaccine candidates have been suggested, but the development of an effective vaccine has so far been difficult (Ijdo et al., 1998;Herron et al., 2000;Ge and Rikihisa, 2006). In order to develop a vaccine, one challenge is to choose antigens that are conserved among all variants of A. phagocytophilum.
Vaccines against ticks are also an alternative option. The development of vaccines that target both ticks and pathogen transmission may provide a mean of controlling tick-borne infections through immunization of the human and animal population at risk or by immunization of the mammalian reservoir to minimize pathogen transmission (de la Fuente and Kocan, 2006). Gut-, salivary-, or cement antigen vaccines (recombinant Bm/Ba 86, Bm91, and 64TRP) have been tested, and TickGUARDPLUS and Gavac (both recombinant Bm86) are examples of commercially available vaccines from the early 1990's (Willardsen, 2004;Labuda et al., 2006;de la Fuente et al., 2007;Canales et al., 2009). Other vaccines that inhibit subolesin expression are now being tested. These vaccines cause degeneration of gut, salivary gland, reproductive -and embryonic tissues and causes sterility in male ticks (de la Fuente et al., 2006a,b,c). Tick vaccines are feasible control methods, cost-effective and environmentally friendly compared to chemical control (de la Fuente and Kocan, 2006).

TRANSMISSION AND COLONIZATION
A. phagocytophilum has, as its name implies, a partiality to phagocytic cells and is one of very few bacteria known to survive and replicate within neutrophil granulocytes (Choi et al., 2005). During tick feeding, neutrophil-associated-inflammatoryresponses are modulated by various stimuli deployed by the tick sialome components (Beaufays et al., 2008;Guo et al., 2009;Heinze et al., 2012). Orchestration of vector-and bacterial interactions with the defensive mechanisms of the host animal seem to promote infection and transmission rather than controlling it, resulting in increased availability of infected cells in the circulating blood and at the site of tick bite (Choi et al., , 2004Granquist et al., 2010b;Chen et al., 2012). The low level of circulating organisms, detected between periods of bacteremia (Granquist et al., 2010c), may indicate temporary clearance of infected cells, possible margination of infected granulocytes to endothelial surface or immunologically modified intervals in generations of antigenically different organisms Beninati et al., 2006;Granquist et al., 2008). Because of the short-lived nature of circulating neutrophils, the role of these cells in establishing and maintaining infection has been questioned (Herron et al., 2005), however to date little is known about alternative cellular components involved in the invasion and colonization of A. phagocytophilum in the host organism (Granick et al., 2008).
A. phagocytophilum modulates the distribution of potential host cells and infected neutrophils, by inducing cytokine secretion and their receptors (Akkoyunlu et al., 2001;Scorpio et al., 2004) and promoting the loss of CD162 and CD62L . The bacterium further interacts with host cell ligands Granick et al., 2008), by surface exposed proteins known as adhesins Ojogun et al., 2012) in order to facilitate internalization in the host cell (Wang et al., 2006). The translocation of bacteria to the inside of host cells is receptor mediated and depending on transglutaminase activity (reviewed by Rikihisa, 2003). However, host cell specific differences to receptors and their components as well as their importance in the infection process seem to exist, which may explain why certain bacterial strains, e.g., ruminant Ap Variant 1 strain, are refractory to culture in commercially available cell lines (like the HL-60 cell line) Herron et al., 2005;Reneer et al., 2006Reneer et al., , 2008Massung et al., 2007). Previous reports have shown that various tissues and cells are susceptible to infection by A. phagocytophilum (Klein et al., 1997;Munderloh et al., 2004). It has been shown that intravascular myeloid cells (mature) have a higher infection rate than cells located in the bone marrow which may indicate that precursor stages of myeloid cells express ligands different from mature neutrophils, thus being more refractory to binding and internalization of the organism (Bayard-Mc Neeley et al., 2004). The coincidence that A. phagocytophilum uses CD162 when infecting neutrophils, led to the hypothesis that endothelium may have a function in the pathogenesis of A. phagocytophilum infection in vivo (Herron et al., 2005). However, a field study of skin biopsies in sheep observed A. phagocytophilum in inflammatory cell infiltrates comprised of PMNs and macrophages in the dermis and subcutis, and occasionally restricted to the mid-and peripheral parts of the blood vessel walls during tick attachment, thus questioning the role of endothelium in the pathogenesis of A. phagocytophilum infection in in the earliest phases of tick bite inoculation (Granquist et al., 2010b). Interestingly A. phagocytophilum has the ability to delay host cell apoptosis by activation of an anti-apoptosis cascade (Sarkar et al., 2012). This is critical for intracellular survival and reproduction of A. phagocytophilum in the normally short lived neutrophil granulocytes (Yoshiie et al., 2000;Lee and Goodman, 2006). Unlike other Gram-negative bacteria, A. phagocytophilum lacks lipopolysaccharides and peptidoglycans, but compensates for the loss of membrane integrity by incorporation of cholesterol which allows the escape of Nod Like Receptor and Toll Like Receptor activation pathways to successfully infect vertebrate immune cells (Lin and Rikihisa, 2003a,b;Hotopp et al., 2006;Xiong et al., 2007). However, recent studies in mice have surprisingly shown that alternative pathways involving the Nod 1 and 2 associated receptor interacting protein 2 may be important in control and clearance of A. phagocytophilum infection (Sukumaran et al., 2012).
The ability of A. phagocytophilum to persist in immunecompetent hosts between seasons of tick activity is a complex and coordinated interaction that through evolutionary steps, have left the genomes of A. phagocytophilum and related organisms, heavily reduced to comprise essential genes allowing for nearly infinite numbers of recombined antigens and macromolecular exchange with its host cell (Rikihisa, 2011;Rejmanek et al., 2012).
Cyclic bacteremias display as periodic peaks containing genetically distinct variants of major surface proteins (MSP) (Granquist et al., , 2010a. The capacity to generate novel antigens when other organisms are already present (superinfection) results in persistence and maintenance of the organism in natural transmission cycles and possibly allows spatial spread in nature (Barbet et al., 2003;Rodriguez et al., 2005;Futse et al., 2008;Ladbury et al., 2008;Stuen et al., 2009). Variants of MSPs such as MSp2 (or P44) contain epitopes recognized by antibodies appearing subsequently, but not prior to the respective peaks of rickettsemia in which they are expressed (Barbet et al., 2003;Granquist et al., 2010c), indicating a true process of antigenic variation influenced by the host immune response. Sequence variation may be achieved by segmental gene conversion of a single polycistronic expression site by insertion of total or partial pseudogene sequences (Barbet et al., 2000;Granquist et al., 2008) with the possible formation of mosaics or chimeras . The large repertoire of donor sequences in A. phagocytophilum suggests that this bacterium may however only require simple gene conversion to evade host immune surveillance (Lin et al., 2003). On the other hand, the close proximity of the partial recombinase gene, recA, which is commonly involved in homologous recombinations supports the theory that recombination of pseudogenes by insertion in the expression site occurs (Barbet et al., 2003;Lin et al., 2003).

VECTORS AND COMPETENT VECTORS OF A. phagocytophilum
A. phagocytophilum is transmitted by hard ticks of the I. persulcatus-complex. The main vector in Europe is I. ricinus (commonly known as sheep tick or castor been tick); in the Eastern US I. scapularis (deer tick or black-legged tick); in the Western US I. pacificus (Western black-legged tick), and in Asia I. persulcatus (taiga tick) (Woldehiwet, 2010). Vector competence has been proven for the American tick species I. scapularis (previously I. dammini), I. pacificus, and I. spinipalpis (Telford et al., 1996;Des Vignes et al., 1999;Zeidner et al., 2000;Teglas and Foley, 2006). Transovarial transmission has not been proven in Ixodes species, but in Dermacentor albipictus, which lifecycle involves a single host animal, representing a distinct ecological niche (Baldridge et al., 2009). As to current knowledge, a vertebrate reservoir host is necessary in nature for keeping the endemic cycle.
Prevalence data on molecular detection of A. phagocytophilum in questing ticks, show great variations within countries or continents where such studies have been performed. The infection rate in I. scapularis ranges from <1% up to 50% and in I. pacificus from <1% up to ∼10% in the US. Additionally, A. phagocytophilum has been detected in questing I. dentatus, Amblyomma americanum, Dermacentor variabilis, and D. occidentalis (Table 4; Goethert and Telford, 2003). In Asia, detection rates varied in I. persulcatus between <1% up to 21.6% and questing I. ovatus, I. nipponensis, D. silvarum, Haemaphysalis megaspinosa, H. douglasii, H. longicornis, and H. japonica also contained DNA of A. phagocytophilum ( Table 5). The greatest number of studies has been performed on questing I. ricinus ticks in Europe, where the prevalence rates vary between and also within countries. On average, the A. phagocytophilum-prevalence in I. ricinus in Europe ranges between <1% and ∼20%, in I. persulcatus-endemic areas in Eastern Europe between 1.7 and 16.7%, and additionally DNA of A. phagocytophilum has been detected in questing D. reticulatus, H. concinna, and I. ventalloi (Table 3). Detailed information on worldwide prevalence rates of A. phagocytophilum in unfed ticks from the vegetation can be found in Tables 3-5.
Based on molecular detection in questing ticks, A. phagocytophilum seems to appear in all countries across Europe. In the US, the majority of studies have been performed in Eastern and Western (California) parts. From Northern US such data are lacking for several geographical regions, however serological evidence indicate exposure to A. phagocytophilum in large parts of the continent (Dugan et al., 2006;Bowman et al., 2009;Villeneuve et al., 2011). Two recent studies revealed the presence of A. phagocytophilum in questing ticks also in the Southern US (Florida and Georgia) (Clark, 2012;Roellig and Fang, 2012). Only few studies have been carried out in Asia, namely in Russia, China, Japan, and Korea ( Table 5). It seems likely that other parts of Asia also belong to the endemic area of this pathogen.
Additionally to the ticks mentioned above, molecular detections have been reported from the following tick species (collected engorged from animals): I. hexagonus, I. trianguliceps, I. spinipalpis, I. ochotonae, and D. nutalli (Zeidner et al., 2000;Bown et al., 2003;Yaxue et al., 2011;Silaghi et al., 2012a). However, the vector competence of a lot of the tick species in which A. phagocytophilum has been detected as well as their contribution to the endemic cycle of A. phagocytophilum remain to be investigated.
The tick species I. ricinus, I. persulcatus, I. scapularis, and I. pacificus are found ubiquitously in their distribution range, have an open questing behavior and a broad host range, including many mammalian species (Sonenshine, 1993). These tick species may consequently also transmit the bacterium from animal reservoir hosts to humans. Aside from these aforementioned antropophilic and exophilic ticks, the involvement of nidicolous, and more host-specific endophilic ticks have been discussed in the context of so-called niche cycles, which may additionally keep the infection in nature. Examples for such proposed niche cycles involve cottontail rabbits (Sylvilagus spp.), I. dentatus and I. scapularis in the US (Goethert and Telford, 2003); field voles (Microtus agrestis), I. trianguliceps and I. ricinus in the UK (Bown et al., 2003); and hedgehogs (Erinaceus europaeus), I. hexagonus and I. ricinus in Europe . The mentioned animals harbor two to three developmental stages of both endophilic and exophilic tick species and can thus transmit the agent from the animal host to humans via the anthropophilic tick species. Considering the large number of host specific and/or nidicolous ticks all around the world, it is likely that more potential niche cycles will be uncovered in the future .
Due to the comparatively low prevalence of A. phagocytophilum in I. pacificus in the Western US, I. spinipalpis has been suggested as a bridging vector for HGA (Zeidner et al., 2000). This nidiculous tick species infests, among others, Mexican woodrats (Neotoma mexicana) (in which A. phagocytophilum DNA has also been detected) and also occasionally bites humans and may thus transmit the agent from zooendemic cycles to humans. Infection rates reported in many studies are higher in adult ticks than in nymphs. Due to the transstadial transmission, but lack of transovarial transmission, larvae are considered free of A. phagocytophilum. Adult ticks have had an additional blood meal in comparison to nymphs, and thus twice the chance of acquiring the infection. Variations in prevalence in questing ticks have also been observed with regard to the year of collection and in-between study areas and different geographic locations Wicki et al., 2000;Hildebrandt et al., 2002;Cao et al., 2003;Holman et al., 2004;Ohashi et al., 2005;Grzeszczuk and Stanczak, 2006;Wielinga et al., 2006;Silaghi et al., 2008Silaghi et al., , 2012bSchorn et al., 2011;Overzier et al., 2013b).
When looking at these variations, it has to be taken into account, that variations can be due to local variations, such as habitat structure or host availability, variation in methodology and sampling approach. Most studies shown in Tables 3-5 are single studies providing a spot prevalence, while studies including longitudinal data are scarce.
Variations in the prevalence of A. phagocytophilum in ticks may be attributed to several factors, such as the susceptibility of individual tick species, the susceptibility of certain tick populations, and the vector competence of tick species; the transmissibility of the A. phagocytophilum variant involved, the susceptibility of different host species, the susceptibility of individual hosts or host populations and the reservoir competence of the host. Especially the availability of different reservoir hosts and the adaptation strategy of A. phagocytophilum seem to be crucial factors in this variability. The availability of reservoir hosts depends on factors such as landscape structure and fragmentation (Medlock et al., 2013). In addition, effects exerted by changes in climate, demography, and agriculture may influence the tick distribution and density and their hosts.

HOSTS AND RESERVOIRS
Viable A. phagocytophilum organisms have been isolated from several hosts, such as cattle, sheep, goat, dog, horse, human, red deer (Cervus elaphus), roe deer (Caperolus capreolus), and white-tailed deer (WTD) (Odocoileus virginianus) (Foggie, 1951;Goodman et al., 1996;Munderloh et al., 1996;Woldehiwet et al., 2002;Massung et al., 2007;Stuen et al., 2010;Silaghi et al., 2011c). However, several prerequisites have to be fulfilled for a reservoir to be competent for a transstadially transmitted pathogen. A reservoir host must be fed on by an infected vector tick; it must take up a critical number of the infectious agent; it must allow the pathogen to multiply and survive for a period and it must allow the pathogen to find its way into other feeding ticks (Kahl et al., 2002). Several mammals may serve as hosts and reservoirs.

WILD RUMINANTS
In Europe, Asia, and America, A. phagocytophilum has been detected in local wild ruminant species (Tables 6-8). Wild ruminants such as WTD and roe deer are among the major feeding hosts for ticks in the Eastern US and Europe, respectively, and thus considered to contribute to a rapid increase in the population of ticks (Spielman et al., 1985;Vázquez et al., 2011;Medlock et al., 2013). WTD is considered one of the major reservoir hosts for an apathogenic variant (Ap-V1) of A. phagocytophilum in the Eastern Frontiers in Cellular and Infection Microbiology www.frontiersin.org July 2013 | Volume 3 | Article 31 | 5    US (Massung et al., 2005). Several genetic variants of A. phagocytophilum have been found in roe deer in Europe and there seem to be both potentially pathogenic and apathogenic variants occurring in roe deer (Silaghi et al., 2011b;Overzier et al., 2013a). A high roe deer density is associated with a high tick density (Jensen et al., 2000;Carpi et al., 2008;Rizzoli et al., 2009) and both presence and high density of roe deer seems to have a positive effect on the A. phagocytophilum prevalence (Rosef et al., 2009). Similarly, the density of WTD influences the density of I. scapularis ticks in the north-eastern US (Rand et al., 2003). For example, the elimination of WTD from certain areas lead to a drastic reduction of the occurrence of I. scapularis (Wilson et al., 1988). In a later study, however, there was no direct effect of a deer culling program on the occurrence of I. scapularis developmental stages (Jordan et al., 2007). In the US, WTD has prevalence rates of A. phagocytophilum of up to 46.6% (Table 6), while detection of A. phagocytophilum in wild ruminants other than WTD are scarce so far. In Europe, roe deer show prevalence rates reaching up to 98.9% (Overzier et al., 2013a). Other deer species seem to contribute to the endemic Frontiers in Cellular and Infection Microbiology www.frontiersin.org July 2013 | Volume 3 | Article 31 | 9 cycles in Europe, and may also constitute efficient reservoir hosts, as the pathogen has been detected in red deer with up to 87% prevalence, in fallow deer (Dama dama) with up to 72%, and in sika deer (Cervus nippon) with up to 50% ( Table 7). A. phagocytophilum has also been identified in deer species in Asia, namely sika deer and water deer (Hydropotes inermis) with prevalence rates of up to 46% and of 63.6%, respectively (Jilintai et al., 2009;Kang et al., 2011; Table 8). However, the studies that have been conducted in Asia on wild ruminants are too few as to draw any definite conclusion on the distribution of A. phagocytophilum.

SMALL MAMMALS
The second large group of animals that A. phagocytophilum is found in endemic countries are in small mammals such as rodents and insectivores. These animals also are major feeding hosts for ticks, especially for the developmental stages (Kiffner et al., 2011). DNA of A. phagocytophilum was found in different mouse, vole, other rodent and insectivore species in the US, Europe, and Asia (Tables 6-8).

Rodents
In Europe, yellow-necked mice (Apodemus flavicollis) were infected with ranges from <1 to 15%, wood mice (Apodemus sylvaticus) from <1 to 11% and bank voles (Myodes glareolus) from 5 to 19.2%. In mouse species, detection with higher prevalence rates represents only single studies, whereas detection in bank voles seemed higher and more consistent. This was also the case for other vole species in Europe (Table 6). In the UK, the field vole has been discussed as a potential small mammal reservoir (Bown et al., 2003). However, in several studies on rodents in Europe, no DNA of A. phagocytophilum has been detected or at such low prevalence rates, that a reservoir role of this group of animals in Europe remains unclear (Barandika et al., 2007;Silaghi et al., 2012b; Table 6).
On the contrary, in the Eastern US, the white-footed mouse (Peromyscus leucopus) is considered one of the major reservoir hosts for the human pathogenic variant (Ap-ha) . P. leucopus is found as the predominant small mammal in forested habitats throughout the Eastern and Central US and it is one of the major hosts for the larval stages of I. scapularis (Sonenshine, 1993). The white-footed mouse has reservoir competence for the AP-ha variant, but reservoir competence could not be shown for the apathogenic Ap-V1 variant , as opposed to the aforementioned WTD as a major reservoir hosts for Ap-V1 (Massung et al., 2005). Different lengths of infections with the two strains have also been shown in an experimental WTD study: Ap-V1 from tick cells resulted in lasting parasitemia, whereas infection with Ap-ha was short-lived (Reichard et al., 2009). By contrast, both Ap-V1 and Ap-ha were infectious for goats and goats are reservoir competent to Ap-V1 .
Ap-V1 was isolated from goats and I. scapularis and propagated in the ISE6 tick cell line, but it could not be cultivated in the human HL-60 cell line. This stands in contrast to A. phagocytophilum strains which have been isolated from human cases in the US, which readily grow in HL-60 cell lines (Horowitz et al.,   1998; Massung et al., 2007), suggesting differing host specificity for these two strain types. Apart from the white-footed mouse, A. phagocytophilum DNA has been detected in several rodent species such as voles and chipmunks in the Eastern US, cotton mice and cotton rats in Florida and several mouse-, chipmunk-, and squirrel-species as well as the dusky-footed woodrat (Neotoma fuscipes) in the Western US ( Table 7). Prevalence ranges from 1.8 to 88.4%. The gray squirrel (Sciurus carolinensis) has also been found to be reservoir competent (Levin et al., 2002) and the redwood chipmunk (Tamias ochrogenys) and sciurid rodents are discussed as important reservoir hosts for A. phagocytophilum in the Western US (Nieto et al., 2010;. Similarly to other small mammals that have been suggested to maintain niche cycles, the redwood chipmunk hosts both antropophilic (I. pacificus) and nidicolous (I. angustus) ticks .
In Asia, comparatively high prevalence rates in small mammals also seem to indicate a reservoir function of this group of mammals (Table 8). For example, in China, wood mice showed prevalence rates up to 10.0% (Zhan et al., 2008), Korean field mice (A. peninsulae) up to 25% (Zhan et al., 2010) and blackstriped field mice (A. agrarius) up to 20.8% (Cao et al., 2006). In Korea, prevalence rates in the black-striped field mouse was also up to 23.6% (Kim et al., 2006) and therefore, A. agrarius has been discussed as one of the major reservoir host in Asian countries. In the Asian part of Turkey, however, all captured rodents were serologically negative for A. phagocytophilum (Güner et al., 2005).
Additionally to mice, voles, chipmunks, and squirrels, DNA of A. phagocytophilum has also been detected in rats on all three continents, in hamsters (China) and in a porcupine (Italy) (Tables 6-8).

Insectivores
There are very few published studies on the role of insectivores in the life cycle of A. phagocytophilum. The common shrew (Sorex araneus) has been discussed as a reservoir host for A. phagocytophilum in the UK (Bown et al., 2011). In that study, prevalence reached 18.7%. Other insectivores which have been investigated in Europe were the greater white-toothed shrew (Crocidura russula) and the European hedgehog ( Table 6). DNA of A. phagocytophilum has also been detected in short-tailed shrews (Blarina brevicauda) with 17.2% prevalence in the US and in Asia in whitetoothed shrews with 63.6% prevalence (Tables 6, 8). Detection rates of A. phagocytophilum in insectivores were generally high, with average prevalence rates around 20%, reaching over 80%. However, the role of insectivores in the life cycle of A. phagocytophilum needs further investigation.

OTHER ANIMAL SPECIES
Apart from wild ruminants, rodents and insectivores, there are several other vertebrate species in which DNA from A. phagocytophilum has been isolated. Whether these contribute to the endemic cycle of A. phagocytophilum is currently not clear. Amongst these animals are mammals such as wild boars, foxes, and bears, but also birds and reptiles (Tables 6-8). The prevalence rates in these animal species seem similar to the potential reservoir hosts discussed above, but studies have been very few so a final conclusion is not yet possible. In the US, raccoons (Procyon lotor) have been found to be reservoir competent for A. phagocytophilum (Levin et al., 2002;Yabsley et al., 2008), while wild boar (Sus scrofa) has recently been discussed as a host for human pathogenic variants of A. phagocytophilum in Europe (Michalik et al., 2012).
The questions which remain open are whether many different animal species get infected only temporarily with potentially nonspecies specific strains of A. phaogcytophilum and constitute deadend hosts such as human beings, whether they develop clinical signs of disease or if they contribute in any way to the endemic cycle.

DOMESTIC ANIMALS
Dogs in Europe were positive for DNA of A. phagocytophilum at about 1-6% prevalence, regardless whether they show symptoms of canine granulocytic anaplasmosis or not. By comparison, the Frontiers in Cellular and Infection Microbiology www.frontiersin.org July 2013 | Volume 3 | Article 31 | 13    prevalence rates in cats were much lower, with <0.5%. In horses, prevalence was higher ranging up to 80%, however, several of the studies investigated horses with symptoms of equine granulocytic anaplasmosis. Without any clinical signs, the prevalence in horses was less than 6.7% (Tables 6-8). Furthermore, several case reports and case series have been published on domestic animals in North America (e.g., Cockwill et al., 2009;Granick et al., 2009;Uehlinger et al., 2011), and serological studies have shown a wide evidence of dogs, cats, and horses being in contact with A. phagocytophilum in USA, Canada, and Asia (e.g., Magnarelli et al., 2001;Billeter et al., 2007;Bowman et al., 2009;Villeneuve et al., 2011;Bell et al., 2012;Ybañez et al., 2012). Additionally, serological and molecular evidence have been provided from North Africa (which also is an endemic area for I. ricinus) that horses and dogs become infected with A. phagocytophilum (M'Ghirbi et al., 2009(M'Ghirbi et al., , 2012. This important finding broadens the known geographic range of A. phagocytophilum to Africa as another continent. The role of dogs as reservoir hosts has been discussed (Schorn et al., 2011). Furthermore, a report of granulocytic anaplasmosis has been described in another member of the canine family, a captive timber wolf (Canis lupus) (Leschnik et al., 2012). The question remains open whether dogs can contribute to the natural cycle of A. phagocytophilum: Is the infection persistent enough for subsequent ticks to become infected, and do dogs host enough nymphal stages of ticks to contribute to the spread? Animals which host mainly adult ticks cannot effectively contribute to the life cycle of A. phagocytophilum, as transovarial infection does not seem to occur.

DOMESTIC RUMINANTS
Infection with A. phagocytophilum has also been detected in several domestic ruminant species such as sheep, goats, cattle, and yaks (Tables 6-8). In Europe, domestic ruminants have been found infected with DNA with rates of up to 20% (cattle), 37% (sheep), and 5.6% (goats) ( Table 6). However, larger scale molecular studies on domestic ruminants in Northern America are lacking, but cases of granulocytic anaplasmosis have been described in llama (Lama glama) and alpaca (Vicugna pacos) in California and Massachusetts, respectively (Barlough et al., 1997a,b;Lascola et al., 2009). Furthermore, serological evidence has been provided for A. phagocytophilum antibodies in cattle in Connecticut (Magnarelli et al., 2002).

SPREAD OF INFECTION
A. phagocytophilum may be spread between different geographic regions by both infected ticks and infected hosts. Expansion of existing endemic areas or to new geographic regions occurs when populations of competent vectors and reservoirs or the abundance of susceptible hosts increase both in total number and in geographic range.
Roe deer carry large number of ticks and moves over long distances (Vor et al., 2010) and may therefore add to the spread of the pathogen itself as well as by moving infected ticks to other areas (Overzier et al., 2013a). Factors contributing to a wider occurrence of suitable hosts such as WTD, white-footed mice, roe deer, field mice etc. may be landscape changes leading to an expansion in the distribution range as well as in the density of those hosts.
Landscape changes such as reforestation may also lead to an expansion of the anthropophilic ticks which are spread also when their primary feeding hosts expand (Sonenshine, 1993).
The increase and spread of I. scapularis in the Eastern US has lead to an increase in Lyme Borreliosis cases (Sonenshine, 1993) and may similarly contribute to the expansion of A. phagocytophilum. In Europe, the increasing geographic range of I. ricinus as well as the expansion to higher altitudes has recently been discussed by several authors (Materna et al., 2005;Jore et al., 2011;Jaenson et al., 2012;Medlock et al., 2013).

Frontiers in Cellular and Infection Microbiology
www.frontiersin.org July 2013 | Volume 3 | Article 31 | 17  Domestic animals including pet animals such as the dog and farm animals such as sheep and cattle may be transported to other areas, in-between countries, even continents, and can thus also add to the spread of infection. Ticks may be spread by birds over long distances and with them A. phagocytophilum-infected ticks. Studies from Europe indicate that migrating birds may be important in the dispersal of A. phagocytophilum infected I. ricinus (Alekseev et al., 2001b;Bjöersdorff et al., 2001). However, A. phagocytophium DNA has sometimes been detected in ticks collected from birds at low prevalence, and it was questioned by some authors whether birds may really be involved in the spreading of the pathogen whereas other authors discussed their possible involvement (Daniels et al., 2002;Ogden et al., 2008;Franke et al., 2010;Hildebrandt et al., 2010a;Dubska et al., 2012;Palomar et al., 2012;Hornok et al., 2013;Kang et al., 2013). The involvement of birds and their ticks in the life cycle of A. phagocytophilum has also been tested in a transmission study in the US. For the two bird species [American robin (Turdus migratorius) and Gray catbird (Dumetella carolinensis)] involved, no significant role in the life cycle was found (Johnston et al., 2013). However, the establishment of ticks in a new habitat depends on the density of hosts in that area, the habitat structure, and the character of the local microclimate and its changes (Daniel, 1993). As an example of this complexity, Figure 1 shows a summary of several direct and indirect factors which are influencing the occurrence and the spread of A. phagocytophilum to humans.

GEOGRAPHIC DISTRIBUTION AND GENETIC VARIATION
As already shown in Tables 3-8, A. phagocytophilum has a wide geographical distribution. However, there is a huge lack of knowledge on ecology, epidemiology and source attributions, vector biology and the clinical implication of different pathogenic strains, related to risk posed on animals and humans (Zhang et al., 2013). This intercepts with the development of effective prevention, control, and eradication strategies for A. phagocytophilum. As already mentioned, transovarial transmission does not seem to occur in tick species associated with infection of humans or animals and the dependence on reservoir animals for maintenance of infection in nature seems crucial (Ogden et al., 1998;Liz et al., 2002). Understanding the extent and mechanisms behind bacterial strain diversity, geographical distribution, and host-pathogen fitness on vector and animal level is increasingly important to give accurate estimates to veterinary and public health risks. Former and future developments in methodologies in molecular epidemiology and genetic fingerprinting like multi-locus sequence typing (MLST), pulse field gel electrophoresis (PFGE), high throughput genome sequencing, blood meal genetic analyses, and the study of microbiomes by for instance metagenomic analyses are powerful approaches to delineating bacterial population structures and the evolutionary processes that underlie these Bown et al., 2007;Dark et al., 2012).
A. phagocytophilum is currently viewed as a single bacterial species, seemingly capable of infecting a broad range of hosts based on 16S rRNA gene analyses. The appearance of 16S rRNA gene variants in ticks seems to be dependent on the habitat structure and therefore of the occurrence of specific potential reservoir hosts, which supports the theory of a host association of some variants (Overzier et al., 2013a,b). The situation appears to be even more complex and delicate in its partiality for certain hosts than previously foreseen, when high resolution methods are used to further delineate strains at host level. Strain variation with potential specific host tropism seems to be abundant in A. phagocytophilum and as such, this has to be taken into account when considering the spread of infection, and the contribution of wildlife such as wild ruminant species in infection cycles involving domestic animals and humans.
A. phagocytophilum is sometimes seen to circulate between hosts sharing similar ecological niches (Al-Khedery et al., Michalik et al., 2012). For example, phylogenetic investigations of the groEL gene have revealed a clustering of sequences into those from roe deer and those from others, as well as a clustering according to geographic origin (Alberti et al., 2005a,b;Silaghi et al., 2011c,d).
Investigations on several A. phagocytophilum strains from different hosts in California indicated that multiple unique strains of A. phagocytophilum with distinct host tropisms exist . Furthermore, one study in the Western US showed no overlap in the endemic cycles found with variants from HGA cases and from the suggested wild-life reservoir, the dusky-footed wood rat (Foley et al., 2008a,b).
A. phagocytophilum 16S rRNA gene variants and possibly also msp4, groEL or ankA gene variants, may cycle differently in the blood of infected hosts, however, the epidemiological consequences of cyclic variation during persistent infection in different hosts are still unknown (Granquist et al., 2010c). The MSP4 is believed to be involved in the host-pathogen interaction and therefore may show host specific characteristics due to selective pressures exerted by the host immune systems, thus a high sequence heterogeneity is observed among A. phagocytophilum strains in this particular gene de la Fuente et al., 2005a). Red deer for instance, previously shown to carry strains that show similarities with ovine strains in the 16S rRNA (100%) and ank (99%) gene sequences (Stuen et al., 2001), have recently been shown to carry msp4 genotypes that appear distinct from sheep variants (Stuen et al., 2013). This stands in contrast to earlier assumptions that red deer and occasionally roe deer may contribute to a natural transmission cycle in Europe, also involving livestock and humans (Alberdi et al., 2000;Rymaszewska, 2008). Characterization of variations in the msp4 sequence, have shown similar structures of strains isolated from humans and dogs in the US (de la Fuente et al., 2005a). Homologous isolates from horse and donkey in California and Italy, respectively, and separate clustering in ruminants are additional examples of evolutionary aspects related to host susceptibility and geographical distribution of this organism (de la Fuente et al., 2005a). Similar patterns have been observed when comparing human, dog, and rodent strains with horse and ruminant strains based on components of the type IV secretion system (Al-Khedery et al., 2012). A German roe deer strain is different in the MSP4 by 23 amino acid changes, compared to the HZ-reference strain representing an outlier of the diversity within the species (de la Fuente et al., 2005a;Ladbury et al., 2008). The diversity of partial msp4 gene in Norwegian sheep and Austrian wild ungulates have shown great variation in sequence types (Ladbury et al., 2008;Silaghi et al., 2011b), while little heterogeneity has been shown for this gene among isolates from horses (Silaghi et al., 2011b,d).
Investigations of the variable part of the msp2 (p44) gene have shown a clustering into variants obtained from ruminant species and those from dogs, horses, and humans, as well as a clustering into those from Europe and the US (Silaghi et al., 2011b,d).
The ank gene has also been used to assess the degree of phylogenetic relationship between strains of A. phagocytophilum as this gene is considered less conserved among strains and even more appropriate for high resolution phylogenetic studies von Loewenich et al., 2003). In one study, ankA gene sequences were found to separate into four clearly distinct clusters. Sequences from dogs, humans, horses, and cats were found exclusively in cluster I, whereas samples from sheep, cows, European bison, and red deer were parts of clusters I and IV. Roe deer sequences were almost exclusively contained in clusters II and III. Based on these results, roe deer seems unlikely to be reservoir of human granulocytic anaplasmosis (Scharf et al., 2011), which supports the findings from studies mentioned earlier.

RESEARCH GOALS AND APPROACHES
Thus far, it is not clear if the differences in infection rates in vectors and hosts outlined above truly reflect differences in vector competency of the vector species and reservoir competency of the host species or whether they reflect differences in the opportunities to acquire the infections (i.e., encounter rates). Previous studies have indicated the existence of enzootic cycles of gene variants in relation to species of ticks and hosts. The knowledge about infection cycles are important for infection and disease control in domestic animals and humans. Future studies should therefore investigate the relationship between genetic strains of A. phagocytophilum, ticks and different hosts, by genetic fingerprinting and blood meal analysis in order to unravel the ecology and phylogeographic distribution of A. phagocytophilum in nature for evidence based risk assessment and risk management. Vector competence of different tick species should be studied, especially considering the potential niche cycles and great variety of strains and variations in the different geographic areas. Which hosts and vectors that competently can keep which variants in endemic cycles in nature should be unraveled.
Further studies should investigate pathogenesis and mechanisms of persistence in host infections. The complexity of cellular and humoral immune responses in rickettsial diseases may be important targets of prophylactic and metaphylactic treatment strategies to control and cure infections by A. phagocytophilum in animals and humans. Factors involve in pathogenicity of the different variants should therefore be elucidated.
Cell culturing and novel molecular tools allow for rapid sequencing and annotation of whole genome structure. Several comprehensive contributions on A. phagocytophilum proteomics from experimental studies in culture systems, tick-and mouse models have been provided (Lin et al., 2011;Troese et al., 2011;Mastronunzio et al., 2012;Kahlon et al., 2013). However, tick and ruminant host interactions with highly pathogenic strains of the bacterium, like the Norwegian Sheep variant 1 (Stuen et al., 2002), should be studied by use of proteomic approaches to reveal key elements for future control strategies in management of this intrusive disease in livestock production. Longitudinal studies to investigate antigenic variation on genomic levels during persistent infections may reveal hitherto unknown mechanisms of immune evasion and persistence, useful in development of diagnostic and therapeutic approaches. To achieve prophylaxis by vaccination further studies on mechanisms of immune evasion and infection strategies are required. The whole genome of several variants of the bacterium has to be sequenced in order to do comparative genomics and develop proper recombinant vaccine antigens for future cross-infection studies.