Distinct Host Species Correlate with Anaplasma phagocytophilum ankA Gene Clusters^,

Abstract

Anaplasma phagocytophilum is a Gram-negative, tick-transmitted, obligate intracellular bacterium that elicits acute febrile diseases in humans and domestic animals. In contrast to the United States, human granulocytic anaplasmosis seems to be a rare disease in Europe despite the initial recognition of A. phagocytophilum as the causative agent of tick-borne fever in European sheep and cattle. Considerable strain variation has been suggested to occur within this species, because isolates from humans and animals differed in their pathogenicity for heterologous hosts. In order to explain host preference and epidemiological diversity, molecular characterization of A. phagocytophilum strains has been undertaken. Most often the 16S rRNA gene was used, but it might be not informative enough to delineate distinct genotypes of A. phagocytophilum. Previously, we have shown that A. phagocytophilum strains infecting Ixodes ricinus ticks are highly diverse in their ankA genes. Therefore, we sequenced the 16S rRNA and ankA genes of 194 A. phagocytophilum strains from humans and several animal species. Whereas the phylogenetic analysis using 16S rRNA gene sequences was not meaningful, we showed that distinct host species correlate with A. phagocytophilum ankA gene clusters.

INTRODUCTION

Anaplasma phagocytophilum is a Gram-negative, tick-transmitted, obligate intracellular bacterium and the causative agent of acute febrile diseases in humans and animals. Its main tick vectors are Ixodes scapularis and I. pacificus in North America, as well as I. ricinus in Europe (3). Major species that develop overt clinical disease after an A. phagocytophilum infection are humans, sheep, cattle, dogs, horses, and cats (6, 15, 16, 37, 58, 75). In 2001, due to a taxonomic reclassification, the organisms infecting ruminants, horses, and humans, formerly known as Ehrlichia phagocytophila, E. equi, and “agent of human granulocytic ehrlichiosis,” were united into a single species termed A. phagocytophilum (19). However, evidence exists that within this species considerable strain variation does occur (10, 75), since A. phagocytophilum isolates of human, equine, bovine, and ovine origin differed in their pathogenicities for heterologous hosts (4, 22, 30, 47, 48, 55, 59).

A. phagocytophilum is not transmitted transovarially in ticks (42). Therefore, the organism is thought to be maintained in reservoir hosts. For the United States, the white-footed mouse, Peromyscus leucopus, is considered as a well-established reservoir species (49, 66). In contrast, the role of the white-tailed deer, Odocoileus virginianus, has been questioned (38, 40) because it might harbor A. phagocytophilum strains not infectious for humans. In Europe, A. phagocytophilum has been detected in wildlife such as roe deer, red deer, and rodents (1, 35, 36, 42, 44), but their specific function in the transmission cycle of A. phagocytophilum is not completely understood (8).

Differences exist in the epidemiology of granulocytic anaplasmosis between North America and Europe. Whereas in 2008, 1,009 cases of human granulocytic anaplasmosis (HGA) were reported to the Centers for Disease Control and Prevention (13), this illness seems to be rare in Europe (7, 57). In contrast, tick-borne fever of sheep and cattle elicited by A. phagocytophilum is common in Europe (58, 74) but has not been reported to date in the United States (75). Granulocytic anaplasmosis of dogs, horses, and cats occurs on both continents (10, 75).

In order to explain host preference and epidemiological diversity, molecular characterization of A. phagocytophilum strains has been undertaken. Most often the 16S rRNA gene was used, and a different pathogenic potential of distinct 16S rRNA gene variants of A. phagocytophilum has been suggested (40, 60). However, 16S rRNA gene sequences were shown to not be informative enough to delineate distinct genotypes of A. phagocytophilum (8, 9, 12, 71). Other genes that were used for typing include the groESL heat shock operon (64), the msp4 gene encoding one of the major surface proteins (17), and the ankA gene (39).

The AnkA protein contains several ankyrin repeats that are thought to mediate protein-protein interactions (51). Although the exact function of AnkA has not been determined yet (50), its C-terminal end has been shown to be secreted by the VirB/VirD4-dependent type IV secretion system (T4SS) of Agrobacterium tumefaciens (34) and the Dot/Icm type T4SS of Legionella pneumophila (29). After translocation, AnkA is tyrosine phosphorylated by the host cell tyrosine kinases Src (31) and Abl (34). After phosphorylation, AnkA binds via SH2 domains to the host cell phosphatase SHP-1 (31), thus probably disturbing host cell signaling (52).

We have previously shown that A. phagocytophilum strains infecting I. ricinus ticks are highly diverse in their ankA genes (71); therefore, in the present study we sequenced the complete open reading frames (ORFs) of 194 samples derived from different host species in order to determine whether distinct host species correlate with ankA gene clusters.

MATERIALS AND METHODS

Samples.

EDTA-anticoagulated blood and/or spleen samples from 31 roe deer from Germany and 26 European bison from Poland shot during the hunting seasons 2003 and 2004, as well as 289 blood samples from sheep from Germany collected for Q-fever surveillance in 2006, were analyzed for A. phagocytophilum by using a 16S rRNA gene-based PCR (41). A 497-bp fragment of the 16S rRNA gene and the complete ORF of the ankA gene of A. phagocytophilum were amplified and sequenced from 55 infected samples from the above-mentioned animals and from a further 139 A. phagocytophilum-positive specimens from different host species of various geographic origins. In all, 113 samples were collected from clinically diseased individuals (43 dogs, 42 sheep, 12 humans, 10 horses, 4 cows, and 2 cats). A further 81 samples originated from A. phagocytophilum positive, but apparently healthy animals (44 roe deer, 15 European bison, 11 red deer, and 11 sheep). Table S1 in the supplemental material presents the host species and geographic origins of the total 194 A. phagocytophilum-positive specimens.

PCR analyses and sequencing.

DNA was extracted from blood, spleen, or A. phagocytophilum-infected HL60 cells (human samples from the United States) by using a QIAamp DNA minikit (Qiagen, Hilden, Germany). In general, 1 to 2 μl of DNA was used as a template in a 50-μl reaction mixture containing 50 mM KCl, 20 mM Tris-HCl (pH 8.4), 2 mM MgCl₂, 0.2 mM deoxynucleoside triphosphates, 0.4 μM concentrations of each primer, and 0.2 μl (1 U) of Taq DNA polymerase (Invitrogen, Karlsruhe, Germany). PCRs were performed by using a GeneAmp PCR System 9700 (Applied Biosystems, Darmstadt, Germany) under the following conditions: an initial denaturation at 94°C for 3 min, followed by 40 cycles of denaturation at 94°C for 30 s, annealing at the predicted melting temperature of the primers (−4°C) for 30 s, and extension at 72°C for 30 s per amplification of 500 bp; with a final extension at 72°C for 10 min. PCR amplification of roe deer, bison, and sheep DNA was carried out to control for the presence of PCR inhibitors. Nested PCR amplification and sequencing of the A. phagocytophilum 16S rRNA gene (41, 71) and of the ankA gene clusters I (39) and II (39, 71) were performed as described previously. Nested PCR amplification and sequencing of the ankA gene clusters III and IV were achieved as shown in Table S2 in the supplemental material. Nucleotide sequences of primers (Metabion, Martinsried, Germany) are summarized in Table S3 in the supplemental material. Nested PCR products were directly sequenced bidirectionally using a 3130 Genetic Analyzer (Applied Biosystems) and a BigDye Terminator v3.1 cycle sequencing kit (Applied Biosystems).

Data analysis.

Sequences were edited and assembled with the SeqMan program of the DNASTAR package (Lasergene, Madison, WI). For phylogenetic analysis of the 16S rRNA or ankA gene sequences, the program MEGA 4.1 (65) was used. Sequences were aligned by using CLUSTAL W, applying the IUB matrix for nucleotide sequences and the Gonnet matrix for protein sequences, respectively. Tree construction was achieved by the neighbor-joining method with the complete deletion option using the Jukes-Cantor matrix for nucleotide sequences and the PAM (Dayhoff) matrix for protein sequences, respectively. Bootstrap analysis was conducted with 1,000 replicates. Average distances within and between ankA gene clusters were computed using the same parameters as for the tree construction. Protein sequences were analyzed for Pfam domain matches (http://pfam.sanger.ac.uk/) and for tyrosine kinase group phosphorylation sites (http://scansite.mit.edu/). Wallace coefficients (11) for host species, geographic origin, 16S rRNA gene variant, and ankA gene cluster were calculated by using Ridom EpiCompare (http://www.ridom.de/epicompare/).

Accession numbers.

GenBank nucleotide accession numbers of 16S rRNA and ankA gene sequences are shown in Table S4 in the supplemental material.

RESULTS

Infection rate.

In all 31 roe deer samples from Germany and all 26 European bison samples from Poland, roe deer and bison DNA was detectable. In contrast, sheep DNA could be amplified only in 255 of 289 sheep specimens from Germany. A total of 29 roe deer, 15 bison, and 11 sheep were found to be positive for A. phagocytophilum by 16S rRNA gene-based PCR. Thus, in Germany the infection rates of roe deer and sheep were 94% (29/31) and 4% (11/255), respectively. In Poland, 58% (15/26) of the European bison were A. phagocytophilum positive.

16S rRNA gene sequences.

A total of 11 different 16S rRNA gene variants were detected in the 194 samples analyzed, whereas 12 sequences from 7 roe deer, 3 sheep, and 2 dogs contained ambiguous nucleotides, indicating a multiple infection with various 16S rRNA genotypes. The identities of the 16S rRNA gene sequences found in the present study to GenBank entries are shown in Table 1. Nine of the 16S rRNA gene variants had been detected earlier, whereas two of them were first discovered here. The two most abundant are 100% identical to GenBank accession numbers U02521 and M73220 and have been frequently reported from diseased humans and animals, as well as from potential reservoir hosts and ticks in Europe and North America. In contrast, others such as the 16S rRNA gene variants with accession numbers AF136712, AF481852, AJ242783, AF384214, and AY281785 have been detected only in Europe thus far. One human sample from Europe was found to be infected with an A. phagocytophilum strain whose partial 16S rRNA gene sequence was identical to accession number AF136712. This variant has been detected previously in dogs, cats, and I. ricinus ticks in Europe.

Table 1.

Identities of the detected 16S rRNA gene variants to GenBank entries^a

Accession no.	No. of samples	Host(s)	Source or reference
U02521	53	24 dogs, 11 humans, 10 horses, 6 sheep, 1 cat, 1 red deer	15
M73220	50	31 sheep, 15 bison, 3 cows, 1 red deer	2
AF136712	19	17 dogs, 1 human, 1 cat	5
AF481852	19	10 sheep, 8 red deer, 1 roe deer	44
AJ242784	15	14 roe deer, 1 cow	73
AJ242783	12	12 roe deer	73
AF384214	7	7 roe deer	36
AY281785	3	2 roe deer, 1 sheep	71
GU236611	2	2 sheep	This study
AF172166	1	1 red deer	14
GU236538	1	1 roe deer	This study

^aThe 12 sequences containing ambiguous nucleotides are not included in this table.

To address the question whether 16S rRNA gene sequences allow a meaningful characterization of A. phagocytophilum strains, we performed a phylogenetic analysis using the neighbor-joining method. As shown in Fig. 1a, all branches obtained were only supported by low bootstrap values.

Fig. 1.

Phylogenetic tree of the 16S rRNA (a) and ankA (b) gene sequences inferred using the neighbor-joining method. Bootstrap values are shown next to the nodes of the trees. The scale bar indicates the number of nucleotide substitutions per site. (a) Sequences with ambiguous nucleotides were not included. The final data set contained 497 positions. (b) Only bootstrap values of major branches are shown. The final data set contained 2,948 positions. Symbols: ♦, dog; □, human; ▾, sheep; ■, horse; ●, bison; ▵, red deer; ○, cow; ◇, cat; ▴, roe deer.

ankA sequences.

Similar to the 16S rRNA gene sequences, ambiguous nucleotides were observed in ankA sequences from 24 roe deer, 4 red deer, 1 cow, 4 bison, and 9 sheep. Most often, the ambiguous nucleotides were detected at positions where in other ankA sequences either one of the two possible nucleotides was found. The deletion/insertion of 3 bp up to 21 bp was frequently observed comparing the ankA ORFs. Two samples, one from a red deer and one from a bison, contained two variants, one with an insertion of 3 bp and one without it (AAA at position 2197 and GGA at position 3598, respectively). Only the variants without the insertions were submitted to GenBank (GU236721 and GU236744).

The ankA gene sequences separated into four different gene clusters (Fig. 1). Two of them, clusters I and II, have been found previously in samples from humans, animals, and ticks (39, 71), whereas ankA gene clusters III and IV were first detected during the present study. The ORF lengths were as follows: cluster I, 3.6 to 3.8 kb; cluster II, 3.6 to 3.7 kb; cluster III, 3.7 kb; and cluster IV, 3.5 to 4.6 kb. Within their group sequences belonging to clusters I, II, III, and IV were on average 99.0%, 99.5%, 99.3%, and 97.6% identical to each other at the nucleotide level and 97.6%, 98.8%, 98.5%, and 94.7% similar at the protein level, respectively. Identities and similarities between the different ankA gene clusters are shown in Table 2. Of note, protein similarities were lower than nucleotide identities, indicating that most nucleotide exchanges were nonsynonymous. Double infections with strains harboring sequences belonging to different ankA gene clusters were detected in four animals (roe deer 11, clusters II and III; roe deer 13, clusters II and IV; roe deer 16, clusters II and III; and red deer 707, clusters III and IV).

Table 2.

Average identities at the nucleotide level and average similarities at the protein level between the different ankA gene clusters

Cluster	Avg % identity or avg % similaritya
	Cluster I	Cluster II	Cluster III	Cluster IV
I		84.8	75.0	70.0
II	76.8		83.3	73.6
III	59.5	70.2		67.0
IV	58.6	62.4	49.4

^aThe average identities at the nucleotide level (in Roman type) and average similarities at the protein level (in italics) between the different ankA gene clusters are presented.

Analyzing the protein sequences, it became obvious that 9 dog sequences contained a 25-amino-acid direct repeat beginning at position 709. Similarly, in four European bison samples a degenerated direct repeat of 35 amino acids was detected in the C-terminal end of the protein. A search against the Pfam domain database revealed that, as expected, all AnkA sequences detected during the present study contained ankyrin repeats. Furthermore, all of them showed a putative T4SS signal sequence (51) in line with the fact that AnkA is translocated by heterologous T4SSs (29, 34). AnkA sequences of all four clusters contained at their C-terminal end several tyrosine kinase phosphorylation sites predicted by Scansite (http://scansite.mit.edu/). Some of these motifs have been shown also experimentally to be phosphorylated by the host cell kinase Src (31). However, only sequences belonging to clusters I and IV showed an extensive multiplication of the tyrosine phosphorylation sites due to the duplication of degenerated direct repeats (Fig. S1 in the supplemental material).

To address the question of whether the separation into four ankA gene clusters was also supported by phylogenetic analysis, we constructed a neighbor-joining tree. As shown in Fig. 1b, all major branches referring to clusters I to IV were supported by bootstrap values of at least 98%. Using AnkA protein sequences, similar results were obtained (data not shown). Sequences from dogs, humans, horses, and cats were found exclusively in cluster I. Human samples from the United States grouped together but were clearly a part of cluster I. Ruminant sequences were much more diverse and belonged to clusters I, II, III, and IV. However, except for one red deer, samples from sheep, European bison, cows, and red deer were found exclusively in clusters I and IV, whereas clusters II and III were restricted to roe deer (Fig. 1b). The geographic origin had no obvious impact on the separation of the sequences, although the analysis was hindered by the fact that the number of samples per country was unbalanced. Calculation of Wallace coefficients (11) for host species, geographic origin, 16S rRNA gene variant, and ankA gene cluster revealed a correlation between host species and ankA gene cluster, as well as between 16S rRNA gene variant and ankA gene cluster (Table 3).

Table 3.

Wallace coefficients for host species, geographic origin, 16S rRNA gene variant, and ankA gene cluster

Parameter	Wallace coefficient
	Host species	Geographic origin	16S rRNA gene variant	ankA gene cluster
Host species		0.567	0.449	0.728
Geographic origin	0.435		0.247	0.473
16S rRNA gene variant	0.499	0.358		0.686
ankA gene cluster	0.348	0.295	0.292

DISCUSSION

In Germany, clinically symptomatic granulocytic anaplasmosis has been detected in horses (72) and dogs (32, 33), whereas no convincing report of HGA exists to date. However, the presence of A. phagocytophilum in I. ricinus ticks (5, 21) and rodents (26) as potential reservoir hosts has been demonstrated. As part of the present study, we showed for the first time that in Germany sheep, roe deer, and red deer were also infected with A. phagocytophilum. However, in none of the animals were obvious clinical signs apparent. The infection rate of 4% in German sheep was somewhat lower than those reported from other countries such as the United Kingdom (38%), Italy (18%), and Turkey (18%) (25, 43, 67). In contrast, 94% of the German roe deer investigated were positive for A. phagocytophilum, whereas the infection rate in neighboring European states was between 18 and 88% (18, 36, 44, 46, 53, 56, 77).

The molecular characterization of the 194 A. phagocytophilum strains analyzed here showed that one European HGA patient was infected with an A. phagocytophilum strain, whose partial 16S rRNA gene sequence was identical to GenBank accession number AF136712. This strain has been found previously in dogs, cats, and I. ricinus ticks in Europe and is different from the variant with accession number U02521, which is usually regarded as pathogenic for humans (40). In line with the infection of two Californian patients with another 16S rRNA gene variant (23), this finding does not support the hypothesis that only strains with the prototype sequence U02521 are pathogenic in humans. Furthermore, according to earlier studies (8, 9, 12, 71), the phylogeny inferred from 16S rRNA gene sequences was not meaningful, indicating that this gene alone is inappropriate for delineating A. phagocytophilum genotypes.

However, upon analyzing the ankA gene sequences we found them 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. With the exception of 4 of 44 samples, roe deer sequences were restricted to clusters II and III. A high identity of major surface protein 2 (msp2) pseudogene sequences from humans, dogs, and horses has been described previously (24), which together with our findings is in line with the fact that horses and dogs are susceptible to infection with human-associated strains (4, 48, 55). Using the groESL heat shock operon for phylogenetic analysis, it was demonstrated that sequences from humans, dogs, horses, sheep, and red deer clustered together, whereas most of the roe deer samples formed a separate branch (53). In our study ankA gene clusters II and III contained the predominant majority of roe deer strains. It seems therefore unlikely that they are relevant reservoirs for granulocytic anaplasmosis in humans and domestic animals. In contrast, red deer could serve as such reservoir hosts as their A. phagocytophilum strains belonged to the respective ankA gene clusters. Furthermore, it has been shown that sheep were susceptible to a red deer-derived isolate (61). Although the exact function of AnkA has not been determined yet, it is suggested to be a T4SS-translocated effector protein of A. phagocytophilum (51) and might therefore be involved in host adaptation, as has been shown for the T4SSs of Bartonella spp. (54, 70).

Analysis of the AnkA protein sequences showed multiple deletions, insertions, and degenerated direct repeats. Degenerated repeats were most obvious in a region at the C-terminal end containing putative tyrosine kinase phosphorylation sites. A similar finding was made previously for Helicobacter pylori, where the T4SS-translocated effector protein CagA differed in the number of repeats containing its tyrosine phosphorylation sites (27). Furthermore, it has been shown that the biological activity of CagA in vitro depended on the quantity and structure of these motifs (28) and that the risk of human gastric cancer might correlate with the repeat number (76). Clearly, it remains to be demonstrated whether A. phagocytophilum strains expressing distinct AnkA proteins will differ biologically. However, it is tempting to speculate that they might do so, because in other prokaryotes repeat polymorphisms are associated with environmental adaptation, functional diversification, modification of niche tropism, antigenic variation, and immune evasion (68, 69).

Recombination is one important mechanism leading to the generation of repeats (68). Typically, recombination between repeat loci requires the existences of multiple sites in the genome. For ankA this might not be the case, because it is a single copy gene (20). However, superinfection of sheep (63), dogs (45), roe deer (77), and red deer (77) with distinct 16S rRNA gene variants of A. phagocytophilum has been reported. This is in line with our study, where multiple samples were found to be doubly infected, even with strains belonging to different ankA gene clusters. Although superinfection has been demonstrated also experimentally in vivo (62), it remains to be shown whether intergenomic recombination occurs in A. phagocytophilum.

Since even genetically homologous strains might be highly diverse in a T4SS effector protein and might show an extensive repeat polymorphism, it remains to be investigated whether the phylogeny inferred using the ankA gene is also reflected by the underlying overall genotype. Furthermore, if recombination indeed occurs, ankA would not be a reliable marker for phylogenetic analysis. To address this question, we recently developed a multilocus sequence typing scheme for A. phagocytophilum and are currently analyzing the strains examined here.

Given the homology of the investigated human-associated isolates from the United States with our European samples from humans, dogs, and horses, it seems unlikely that the epidemiological differences between the continents are caused mainly by strain divergence. Other explanations might be disease awareness in Europe, as well as variation of vector transmissibility, which will require further investigations.