Molecular epidemiology of the emerging zoonosis agent Anaplasma phagocytophilum (Foggie, 1949) in dogs and ixodid ticks in Brazil

Background Anaplasma phagocytophilum is an emerging pathogen of humans, dogs and other animals, and it is transmitted by ixodid ticks. The objective of the current study was a) detect A. phagocytophilum in dogs and ixodid ticks using real-time Polymerase Chain Reaction (qPCR); and b) Determine important variables associated to host, environment and potential tick vectors that are related to the presence of A. phagocytophilum in dogs domiciled in Rio de Janeiro, Brazil. Methods We tested blood samples from 398 dogs and samples from 235 ticks, including 194 Rhipicephalus sanguineus sensu lato, 15 Amblyomma cajennense, 8 Amblyomma ovale and 18 pools of Amblyomma sp. nymphs. A semi-structured questionnaire was applied by interviewing each dog owner. Deoxyribonucleic acid obtained from ticks and dog buffy coat samples were amplified by qPCR (msp2 gene). The sequencing of 16S rRNA and groESL heat shock operon genes and a phylogenetic analysis was performed. The multiple logistic regression model was created as a function of testing positive dogs for A. phagocytophilum. Results Among the 398 blood samples from dogs, 6.03% were positive for A. phagocytophilum. Anaplasma phagocytophilum was detected in one A. cajennense female tick and in five R. sanguineus sensu lato ticks (four males and one female). The partial sequences of the 16S rRNA, and groESL genes obtained were highly similar to strains of A. phagocytophilum isolated from wild birds from Brazil and human pathogenic strains. The tick species collected in positive dogs were R. sanguineus sensu lato and A. cajennense, with A.cajennense being predominant. Tick infestation history (OR = 2.86, CI = 1.98-14.87), dog size (OR = 2.41, IC: 1.51-12.67), the access to forest areas (OR = 3:51, CI: 1.52-16.32), hygiene conditions of the environment in which the dogs lived (OR = 4.35, CI: 1.86-18.63) and Amblyomma sp. infestation (OR = 6.12; CI: 2.11-28.15) were associated with A. phagocytophilum infection in dogs. Conclusions This is the first report of A. phagocytophilum in ixodid ticks from Brazil. The detection of A. phagocitophylum in A. cajennense, an aggressive feeder on a wide variety of hosts, including humans, is considered a public health concern.

Although A. phagocytophilum has been reclassified, investigations based on the molecular bases of the groEL and 16S rRNA genes [7], have shown that A. phagocytophilum exhibits some genetic heterogeneity. Genetic variations were found in 16S rRNA, groEL, msp2, msp4 and Anka genes in A. phagocytophilum isolated from ticks and mammals [8].
Five genetic variants of A. phagocytophilum with 1-2 nucleotide differences in the 16S rRNA sequence were identified in dogs [9]. Organisms detected in dogs from Switzerland had an identical 16S rRNA gene sequence found in humans with granulocytic anaplasmosis [3], demonstrating that these dogs might serve as hosts of genetic variants of A. phagocytophilum able to infect humans.
A number of aspects related to the host, the environment and the agent have been incriminated as risk factors for infection of A. phagocytophilum in dogs. These factors include season, co-infections, clinical signs and the genetic variants of the parasite [9]. Anaplasma phagocytophilum infections are diagnosed most frequently in months with peaks of nymph and adult stages of the tick vectors [10,11]. Recent studies have shown a greater tendency toward infection in adult dogs, reflecting a higher probability of tick infestation with time [12]. Furthermore, older dogs may be more susceptible.
In Brazil, there are reports of A. phagocytophilum infection in dogs from November to May in Rio de Janeiro state [5]. The authors found a strong association between A. phagocytophilum-positive dogs and the presence of Amblyomma ticks. These data suggest that this tick genus might be involved with A. phagocytophilum transmission in this region. Also in Brazil, molecular detection of A. phagocytophilum was observed in wild birds and sheep [13,14]. Seroprevalence studies have found evidence of A. phagocytophilum infection in small ruminants [15], equines [16] and dogs [17], in Pernambuco state, Notheastern Region of the country, in São Paulo and Paraná state, respectively. Despite this evidence, it is not possible to consider granulocytic anaplasmosis as an endemic disease in Brazil, since studies are isolated and samples are not representative. The aim of this study was to investigate the presence of A. phagocytophilum DNA in blood samples and in ixodid ticks collected from dogs using Real-Time PCR and study epidemiological aspects related to the presence of A. phagocytophilum DNA in dogs associated with variables related to the host, the environment and the possible tick vectors in the state of Rio de Janeiro, Brazil.

Description of the studied area
This study was carried out in the municipalities of Itaguaí and Seropédica (43°10' 376" W and 22°5 7'133 "S), respectively, located in the metropolitan mesoregion of the state of Rio de Janeiro, from November 2009 to November 2010.

Sample size of dogs and ticks
The sampling of dogs in these municipalities was calculated according to a previously described equation [18] with a confidence interval of 95% and an error margin of 3%, assuming an expected prevalence of 10%, based on the data regarding the frequency of A. phagocytophilum in Brazil [5]. Three hundred and ninety-eight dogs were selected (Itaguaí, n = 136; Seropédica, n = 262), based on convenience sampling.
To determine the occurrence of A. phagocytophilum in a tick population, the sample size was determined considering a prevalence of 10% of infected ticks in an infinite population, based on the value of maximum prevalence reported for Ehrlichia chaffeensis in the Amblyomma americanum tick [19,20]. Thus, the minimum number of ticks to be tested was 138, assuming an absolute precision of 5% and a confidence interval of 95% were assumed. Nevertheless, 235 ticks were evaluated in this study.

Epidemiologic questionnaire
A semi-structured questionnaire was applied by interviewing each dog owner to collect information related to the dog and the environment. This information included the following items: sex (female or male); age (<2 years old or ≥ 2 years old); racial definition (defined breed dogs or mixed-breed dogs); dog size as large (50 cm or more) or small/medium (less than 50 cm); history of tick infestation (yes or no); local access of the dog (urban environment/ yard/grazing/forest); contact with other animal species (yes or no); acaricide treatment (yes or no); dog habitat as urban or rural (based on the urban perimeter delimited by municipal governments); condition of environment hygiene as unsatisfactory or satisfactory (based on the environment cleanliness and the presence of feces in the location that the dog lives); veterinary assistance (yes or no); tick infestation (presence or absence of ticks at the time of collection of the blood sample); presence of Amblyomma spp. (presence or absence of ticks of the genus Amblyomma at the time of sample collection); presence of Rhipicephalus sanguineus sensu lato (presence or absence of this tick species at the time of sample collection) and presence of ectoparasites (presence or absence of ticks, fleas, lice and mites at the time of collection).

Buffy coat and tick collection in the dogs
After obtaining the owner's consent, the animals were restrained for clinical examination and blood collection. A sample of 5 mL of peripheral blood was drawn from each animal by cephalic venipuncture and placed under a vacuum in sterile tubes containing anticoagulant. Buffy coat samples were obtained after centrifugation at 2,500xg for 5 minutes, placed in 1.5 mL micro tubes and stored at−80°C until DNA extraction.
The entire body of each animal was examined for ticks, with particular attention paid to the auricular pavilion region, the head, the neck, the chest, the armpits, the inguinal region and the region under the tail. In infested animals, the level of tick infestation was categorized as follows: low infestation (< five ticks), moderate infestation (≥ five <15 ticks), high infestation (≥15 <30 ticks) and very high infestations (≥30 ticks). The ticks were preserved in tubes containing isopropyl alcohol. The adult tick species were identified using dichotomous keys [21,22]. Nymphal stage ticks were classified to genus level.
Ticks were separated according to the species, sexual dimorphism, developmental stage, animal source, collection date and locality of origin.

Search for morulae
All dogs sampled were examined for the presence of A. phagocytophilum morulae in granulocytes using the blood smear method. Briefly, thin blood smears were prepared; the smears were air-dried and stained using Giemsa and then examined by light microscopy at 1,000 × .

DNA extraction
The deoxyribonucleic acid (DNA) was extracted from 70 μL of buffy coat using the DNeasy Blood and Tissue kit (Qiagen, Valencia, CA, USA), according to the manufacturer's recommendations.
Genomic material of adult ticks was extracted from a single specimen, whereas the DNA of nymphs was extracted from a pool of five specimens. DNA was extracted from tick samples according to the protocol described previously [23]. The concentration and purity of the DNA from all blood samples and ticks was determined using spectrophotometry (Nanodrop ND-2000®, Thermo Scientific, Wilmington, DE, USA). DNA samples were diluted to obtain a final concentration of 30 ng/μL.
The positive control of A. phagocytophilum was obtained from antigenic substrates in commercial slides prepared for immunofluorescence (Fuller Laboratories, Fullerton, CA, USA). Genomic material in the slides was purified using the DNeasy Blood and Tissue kit (Qiagen, Valencia, CA, USA) according to manufacturer's recommendations.

Real-time PCR assay (qPCR)
DNA samples obtained from the dog buffy coat and ticks were analyzed by qPCR with targets in the msp2 gene of A. phagocytophilum [24]. The reactions were performed in triplicate using the Real-Time PCR System StepOnePlus® instrument (Applied Biosystems). The quantification cycle (Cq) was standardized between plates and Cq was manually allocated three cycles after the fluorescent base. Samples with Cq values of less than or equal to 40 cycles were considered as positives. Samples were considered positive for A. phagocytophilum if any 1 of 3 replicate samples showed amplified DNA for A. phagocytophilum relative to negative controls.
The amplification of 16S rRNA (546 bp) and groESL (1715 bp) heat shock operon genes was performed from positive samples in the qPCR [25,26] for diagnosis confirmation.

Cloning, sequencing and phylogenetic analysis
The nucleotide sequences of the 16S rRNA and groESL heat shock operon genes were determined for A. phagocytophilum samples from 3 ticks and 12 dogs. PCR products were purified, cloned and sequenced [27]. The sequencing was performed on the equipment ABI 3730 DNA Analyzer (Applied Biosystems / Perkin Elmer, CA, USA). The identity of the fragments was analyzed using multiple alignments by the Basic Local Alignment Search Tool (BLAST).
The phylogenetic position of A. phagocytophilum isolated from ticks and Brazilian dogs was inferred using the Neighbor-Joining method. The combination of phylogenetic clusters was assessed by bootstrap test with 1000 replicates. The evolutionary distances were calculated by Kimura 2-parameter method. There were a total of 420 nucleotides for the 16S rRNA gene and 1167 for the groESL heat shock operon in the final data set. The analyzes were conducted in MEGA 5.0 [28].

Analytical sensibility of qPCR to the Anaplasma phagocytophilum diagnostic
The analytical sensitivity of the assay was determined by evaluating serial decimal dilutions of the amplicon cloned into the pGEM-T plasmid [29]. The sensitivity of the real-time PCR assay was evaluated with and without the addition of 1 μL of DNA extracted from ticks (Amblyomma cajennense) and a dog buffy coat. The concentration and purity of the plasmid DNA were measured using a spectrophotometer (Nanodrop ND-2000®, Thermo Scientific, Wilmington, DE, USA). The concentration of the plasmid DNA was used to calculate the number of plasmids. A standard curve was constructed with six points representing six serial decimal dilutions ranging from 1 to 100,000 plasmid copies containing a 122-bp fragment of the msp2 gene of A. phagocytophilum. The standard curves were performed with and without the addition of 1 μL of A. cajennense DNA (laboratory colony) and the whole blood of an uninfected dog.

Statistical analysis
The chi-square test or the G test at the 20% level of significance was used to assess whether the presence of A. phagocytophilum DNA in the dogs was associated with independent variables collected through the epidemiological questionnaire. The Spearman test was used for those variables exhibiting p <0.20 in the chi-square test or G test to remove the highly correlated variables of the multiple logistic regression analysis. The most biologically important of two highly correlated variables was maintained in the multiple logistic regression analysis.
Independent variables with p <0.20 and ρ <0.7 were included in the multiple logistic regression model as a function of the dogs testing positive for A. phagocytophilum using qPCR. An error of 5% was assumed in the final model.
To verify the association level between the positive dogs in qPCR test for A. phagocytophilum and the tick infestation level, a frequency ratio was calculated at a 5% significance level using uninfested dogs as a reference variable. This analysis was performed using BioEstat, version 5.0 [30].
A bivariate analysis was performed using BioEstat, version 5.0 [30], and the multiple logistic regression model was conducted using the R statistical software, version 2.11.1 [31].

Results
Inclusions in neutrophils suggestive of A. phagocytophilum were not observed in 398 blood samples analyzed by blood smear. Of all DNA samples tested by qPCR, 24 (6.03%) tested positive for A. phagocytophilum. Quantitative PCR was conducted on 194 samples of R. sanguineus sensu lato (100 females, 82 males and 12 pools of five nymphs), on 15 samples of A. cajennense (nine females and six males), on eight samples of Amblyomma ovale (five females and three males) and on 18 pools of Amblyomma sp. nymphs. The frequency of A. phagocytophilum positive ticks found was 2.55% (n = 6/235). A higher frequency of positive results was observed in A. cajennense ticks [one female, 6.67% (n = 1/15)] than in R. sanguineus sensu lato ticks [n = 5/194; 7.76% in males (n = 4/194) and 0.51% in females (n = 1/194)]. All positive ticks for A. phagocytophilum were collected from negative dogs.
Fifteen positive samples (12 samples obtained from dogs and three from ticks) were cloned and sequenced. All of them showed 100% identity with each other. A single partial sequence of each gene (msp2, 16S rRNA and groESL) was submitted to GenBank under the access number [HQ670750, KF836093 and KF836094, respectively].
Of the 24 blood samples positive in qPCR, only 12 were positive for the 16S rRNA and groESL heat shock operon genes. The nucleotide sequences of the 16S rRNA gene were identical to each other and the same was observed for the groESL operon. Regarding the six tick samples, only three were positive for both genes with an identical nucleotide sequence for the 16S rRNA and groESL heat shock operon genes. The 16S rRNA and groESL sequenced genes of A. phagocytophilum showed 100% identity with other sequences isolated from dogs (Figures 1 and 2).
The analytical sensitivity of the qPCR technique was evaluated. The detection limit of the technique was one plasmid copy containing the A. phagocytophilum msp2 gene (Figure 3). No significant difference was found in the analytical sensitivity (p > 0.05) when 1 μL of an equimolar mixture of DNA from buffy coat of an uninfected dog and DNA from an uninfected A. cajennense tick were added to the qPCR. High amplification efficiency was demonstrated by identical slopes between amplification curves (Figure 4).
No association was found (p > 0.05) between the variables such as sex, breed, age and animal size, and the positive result in the real-time PCR (Table 1). Tick infestation history was the variable related to the host that was associated with a positive qPCR result (p < 0.05). Dogs with a tick infestation history were 2.86 times more likely to yield positive results in qPCR for A. phagocytophilum.
Variables including acaricide treatment, contact with other species, the dog's habitat [rural or urban] and veterinary care were not associated (p > 0.05) to infection with A. phagocytophilum in the dogs. Despite absence of significant association among these variables, the frequency of positive dogs in close contact with other animal species (7.94%, n = 17/214) was higher than dogs that were not exposed to other species (3.80%, n = 7/184). The variable "kind of environment accessed" was associated with A. phagocytophilum infection. In this study, 18.75% (n = 9/48) of the dogs that had access to forest areas were infected with A. phagocytophilum, whereas 4.48% (n = 15/350) of the dogs that had access to pasture areas/ urban environment/backyard were positive when analyzed by qPCR. The hygiene conditions of the environment in which the dogs lived were also associated (OR = 4.35, CI = 1.86-18.63) with A. phagocytophilum infection.
The variable "ectoparasite infestation" was removed from the final logistic regression model because of the high correlation (ρ > 0.7) with the variable "tick infestation". Ectoparasite infestation was observed in 79.16% (n = 19/24) of the positive dogs diagnosed by qPCR. The presence of ticks was verified in 66.67% (n = 16/24) of the positive dogs; the presence of ticks parasiting dogs had no association (p = 0.19) with A. phagocytophilum infection. Dogs in which the presence of ticks of the single genus Amblyomma was evaluated showed an odds ratio of 6.21 times, demonstrating a strong association with A. phagocytophilum infection. The species of ticks found most commonly parasiting positive dogs were A. cajennense (50%, n = 12/24) and R. sanguineus sensu lato (37.50%, n = 9/24). Co-infestation with A. cajennense and R. sanguineus sensu lato were   The frequency ratio between the infestation levels by ticks and A. phagocytophilum infection in dogs ranged from 1.7 to 3.0 with increasing infestation level ( Table 2).
The level of infestation was evaluated without considering the tick species found on the dogs.

Discussion
The negative result for A. phagocytophilum in the analysis of blood smears can be explained by the short period of bacteremia, less than 28 days [32]. In experimental infections, the morulae of A. phagocytophilum could be seen in neutrophil cytoplasm from 4 to 14 days post-infection and could be observed for a period of 4 to 8 days [33]. Infected neutrophil percent in the acute phase of the disease varied from 1 to 42% [9,11,33], and probably affected blood smear diagnosis, since in the present study, none of the dogs showed clinical symptoms. During blood smears analyses, one aspect that needs attention is the A. phagocytophilum misdiagnosis based only on intracytoplasmic inclusions in neutrophils, since Ehrlichia ewingii species, present in dogs from Brazil [34], have similar morphological characteristics when compared with A. phagocytophilum [35]. Accordingly, PCR analysis is the best choice for a diagnostic test, and it is the most reliable and specific evaluation for granulocytic ehrlichiosis diagnosis [36].
The 16S rRNA gene fragment [KF836093] amplified from DNA of blood from dogs and ticks samples showed 100% identity with sequences isolated from wild birds [JN217096, JN217094 and JN217095] in Brazil [13]. These results showed that A. phagocytophilum circulates in Brazil and has wild birds and dogs as hosts. 16S rRNA gene sequences isolated from Brazilian dogs and ticks showed 100% identity with the prototype strain isolated    [37]. A sequencing artifact cannot be excluded in this case, but it seems to be unlikely, as the PCR was performed with high fidelity DNA polymerase and sequencing was performed three times in both directions from cloned PCR products.

Number of Cycles
The current study analyzed characteristics of dog blood samples infected with A. phagocytophilum. The results demonstrated no statistical association with age, although a significant number of positive dogs (83.33%, n = 20/24) belonged to the group with two years or more. In Sweden, the percentage of seropositive dogs increased with age [39], reflecting an increased likelihood of exposure to infected tick vectors over time. The average age of dogs infected with A. phagocytophilum reported in the literature is approximately six to eight years old [9,[40][41][42].
At the final logistic regression model, the variable dog size was associated with the presence of A. phagocytophilum DNA in dogs (OR = 2.41, IC: 1.51-12.67). Most likely, the size of the dog did not influence the susceptibility of infection with A. phagocytophilum, but rather the habits or purpose of the breed; frequently, the dogs in this study were kept in the vicinity of the home as guard animals or as hunting dogs. Hunting exposes dogs to forest areas and several tick species, which increases the risk of infection by vector-borne pathogens, including A. phagocytophilum.
Anaplasma phagocytophilum is retained in the natural cycle between rodents and ticks of the genus Ixodes [43,44]. In the studied area, the species Ixodes amarali has been detected parasitizing opossums of the genus Didelphis [45]; there are no reports of dog parasitism by the Ixodes species in the state of Rio de Janeiro, despite the fact that several surveys have been conducted during the last decade. It was found that A. phagocytophilum infection in dogs was associated with the tick infestation history reported by the owners; an identical finding was observed in Taipei, Taiwan [46]. This association was not observed in relation to the presence of ticks on dogs at the time of collection, although a tendency for A. phagocytophilum infection was verified in dogs infested with ticks at the time of collection (66.67%). In the multiple logistic regression model, tick species that belonged to the genus Amblyomma were strongly associated (OR = 6.12; CI: 2.11-28.15) with A. phagocytophilum infection in dogs. This association did not occur for dogs infested with R. sanguineus sensu lato (p = 0.75). A number of studies have shown different results regarding tick exposure history as reported by owners. For example, tick infestation has not been described in any of the dogs examined for granulocytic anaplasmosis in western Washington, USA [9].
In a study in Sweden, exposure to ticks was observed in 13 of 14 dogs examined [40]. In Germany, a similar result  was described, and tick infestation was observed in 80% of 18 dogs naturally infected with A. phagocytophilum [42].
In the state of Rio de Janeiro, the species of Amblyomma that most frequently infest dogs are: A. cajennense, Amblyomma aureolatum and A. ovale in that order [47,48]. These tick species are most often found parasitizing dogs in rural areas near secondary forests, a variable that was associated (OR = 3:51, CI: 1.52-16.32) with the presence of A. phagocytophilum DNA in dogs. The current study showed that A. cajennense, Amblyomma dubitatum and A. aureolatum were found only in dogs that had access to forest areas. Tick species parasitizing dogs in Brazil are diverse because there are many different ecosystems in the country. The environmental characteristics and the diversity of host species present in each area are key factors for the variety and abundance of the species of ticks that infest dogs. This is the first report of A. phagocytophilum infection in R. sanguineus sensu lato and A. cajennense adult ticks in Brazil. In this study, despite the epidemiological evidence suggesting the genus Amblyomma as a possible vector of A. phagocytophilum, R. sanguineus sensu lato exhibited a higher absolute number of positive ticks. This result is most likely associated with the small number of ticks from the genus Amblyomma analyzed by qPCR. Anaplasma phagocytophilum has been detected by PCR in R. turanicus in Italy with a frequency of 1.8% [6], a higher rate was observed in the present study (2.58%). The only report of A. phagocytophilum infection in ticks from the genus Amblyomma was described in Poland with Amblyomma flavomaculatum, parasiting lizards imported from Africa [49].
The rate of infection in R. sanguineus sensu lato male ticks was higher than that observed in female ticks; this result was also reported in a study in Jilin province, China and Tunisia, Africa [19,50]. Nevertheless, in a study conducted in Poland [51], the number of female ticks positive for A. phagocytophilum (45.7%, n = 79/173) was higher than that of male ticks (4.5%, n = 9/202) and nymphs (0.9%, n = 3/319). Further studies using a larger number of ticks at different seasons and in different locations and habitats are necessary to understand the distribution and variation in the rate of A. phagocytophilum infection in ticks.

Conclusions
In the evaluated region, the epidemiology of A. phagocytophilum infection includes large sized dogs; animals which have access to forest areas; dogs in contact with other animals; animals that live in poor hygienic conditions; and, particularly, dogs infested with Amblyomma sp. ticks.