Novel Genetic Variants of Anaplasma phagocytophilum, Anaplasma bovis, Anaplasma centrale, and a Novel Ehrlichia sp. in Wild Deer and Ticks on Two Major Islands in Japan

ABSTRACT Wild deer are one of the important natural reservoir hosts of several species of Ehrlichia and Anaplasma that cause human ehrlichiosis or anaplasmosis in the United States and Europe. The primary aim of the present study was to determine whether and what species of Ehrlichia and Anaplasma naturally infect deer in Japan. Blood samples obtained from wild deer on two major Japanese islands, Hokkaido and Honshu, were tested for the presence of Ehrlichia and Anaplasma by PCR assays and sequencing of the 16S rRNA genes, major outer membrane protein p44 genes, and groESL. DNA representing four species and two genera of Ehrlichia and Anaplasma was identified in 33 of 126 wild deer (26%). DNA sequence analysis revealed novel strains of Anaplasma phagocytophilum, a novel Ehrlichia sp., Anaplasma centrale, and Anaplasma bovis in the blood samples from deer. None of these have been found previously in deer. The new Ehrlichia sp., A. bovis, and A. centrale were also detected in Hemaphysalis longicornis ticks from Honshu Island. These results suggest that enzootic cycles of Ehrlichia and Anaplasma species distinct from those found in the United States or Europe have been established in wild deer and ticks in Japan.

infected wild deer in Asia. These findings identify four species of Ehrlichia and Anaplasma, including a novel Ehrlichia species and novel A. phagocytophilum strains. Analysis of specimens from Hemaphysalis longicornis ticks associated with deer revealed the presence of three of these four species, implying the established deer-tick enzootic cycle.

MATERIALS AND METHODS
Deer and tick specimens. Sera were collected from 126 wild deer for examination: samples from 79 deer (Cervus nippon yesoensis) from Hokkaido Island, northern Japan, collected during 1975, 1989, and 1991; and samples from 47 deer (Cervus nippon nippon) from Shimane Prefecture on Honshu Island, southwestern Japan, collected during 2001 and 2002. Both locations were within the Piedmont physiographic region and were approximately 1,500 km apart. Sixty-eight Hemaphysalis longicornis ticks were collected by flagging in Shimane Prefecture, Japan, in 1999. Twenty-five H. longicornis ticks were removed from deer (Cervus nippon nippon) in Nara Prefecture, Honshu Island, Japan, in 2004.
PCR. DNA from the deer sera and homogenates of ticks was extracted using the QIAamp DNA Blood Mini Kit (QIAGEN, Valencia, CA). Extracted DNA was used as the template for nested PCR amplification of 16S rRNA genes. The primer pair EC9 and EC12A (Table 1) used for the first PCR amplifies all known Anaplasma and Ehrlichia species. DNA from E. muris strain AS145 and A. phagocytophilum strain HZ were used as positive controls, and doubly distilled H 2 O was used as the negative control. The primers for the second-round PCR were specific for E. muris, Ehrlichia strain HF, 'Candidatus Neoehrlichia mikurensis,' A. phagocytophilum, A. centrale, and A. bovis ( Table 1). The A. phagocytophilum p44 gene was amplified by nested or single-step PCR using primer pairs p3709-p4257 and p3761-p4183 (Table 1) (30). The initial amplification consisted of 40 cycles, each cycle consisting of 30 s at 94°C, 30 s at 52°C, and 1 min at 72°C. For the nested PCR amplification, 1 l of the product from the first amplification was used for amplification with specific primers in a 25-l reaction mixture; the amplification consisted of 40 cycles, each of 1 min at 94°C, 1 min at 55°C, and 1 min at 72°C. DNA sequencing and data analysis. Sequences of every PCR amplicon of 16S rRNA genes were determined. All sequences shown in this study had the primer region sequences removed before analysis. Longer (Ͼ1,300-bp) 16S rRNA gene sequences were determined for seven representative sequences (ES34P-L from deer on Hokkaido; SS15E-L, SS24B-L, SS33P-L, and SS40C-L from deer in Shimane; TS37 from a tick in Shimane; and NR07 from a tick in Nara) and were compared to sequences of known Anaplasma and Ehrlichia spp. CLUSTAL W (version 1.6.5) and TreeView (version 1.65) on the DDBJ website and GENETYX-WIN (version 7.0.6; Software Development Co., Tokyo, Japan) were used for the sequence analysis and phylogram construction.
Nucleotide sequence accession numbers. The 16S rRNA gene sequences of ES34P-L, SS33P-L, SS40C-L, SS24B-L, SS15E-L, TS37, and NR07 have been deposited in the GenBank data library under accession numbers AB196720, AB196721, AB211164, AB211163, AB211162, AB074459, and AB196475, respectively. The groESL sequence of TS37 has been deposited in the GenBank data library under accession number AB074462. Sixteen different p44 sequences have been deposited in the GenBank data library under accession numbers DQ020144 to DQ020159.

RESULTS
Deer from two Japanese islands were tested for infection with Anaplasma and Ehrlichia species by PCR-based analysis of 16S rRNA genes. PCR products were sequenced to determine the pathogen identity. As shown in Table 2  To define the pathogen identity, almost full-length 16S rRNA sequences were determined for seven representative strains from deer and from ticks. The phylogenic analysis used 1,332-bp sequences of all seven new strains determined in this study and previously sequenced Anaplasma and Ehrlichia spp. (Fig. 1). Of the Hokkaido deer, 10% (8/79) were infected with A. phagocytophilum, and the 16S rRNA sequences (579 bp) of the amplicons from all eight infected deer were 99.9 to 100% identical to each other. The longer representative sequence (ES34P-L; 1,338 bp) was most closely related (99.3% identical; 10 bp differed of 1,338 bp compared) to Anaplasma sp. SA1076 from a dog in South Africa (GenBank accession no. AY570539) (21). This sequence was 98.7% identical (1,321 of 1,338 bp compared) to the A. phagocytophilum strain 'HGE agent' (where HGE is a designation for human granulocytic ehrlichiosis) CAHU-HGE2 from a human patient in northern California (GenBank accession no. AF093789). Of the Shimane deer, 34% (16/47) were infected with A. phagocytophilum, and the 16S rRNA sequences (579 bp) of the amplicons from all 16 infected deer were 99.9 to 100% identical to each other. The longer 16S rRNA sequence (SS33P-L; 1,399 bp) from the representative specimen was most closely related (99.4% identical; 1,373 of 1,381 bp compared) to Anaplasma sp. SA1076 from a dog in South Africa (Gen-Bank accession no. AY570539). It also was the next most closely related (98.4% identical; 21 bp differed of 1,402 bp compared) sequence to that of A. phagocytophilum strain LGE from a llama (Llama llama) and llama-associated ticks (Ixodes pacificus) (GenBank accession no. AF241532) (4). The sequence identity between ES34P-L from the Hokkaido deer and SS33P-L from the Shimane deer was 99.6% ( Fig. 1; Table 3), implying geographic segregation of strains between the two islands.
Using the p44-specific primer pairs shown in Table 1 (30) for PCR, followed by sequencing of the amplicons, p44 sequences of approximately 370 bp including primer regions (330 to 410 bp) were obtained from deer blood specimens. We obtained nine, four, and three different p44 paralog sequences from deer SS14, SS33, and SS40, respectively. The sequences of the primer regions were removed, and the deduced P44 amino acid sequences were compared. These sequences were characterized by the central hypervariable regions flanked by N-and C-terminal-conserved regions (30). The P44 sequences from Japanese deer samples contained the identical amino acids C, C, W, and P found in the P44 USA HZ and P44 UK A. phagocytophilum consensus amino acid sequences (Fig. 2) (12,30). However, all p44 genes of Japanese samples had relatively shorter regions, delineated by two conserved cysteines (ϳ12 amino acid residues), in the hypervariable region than those of the P44 USA HZ stains available at GenBank and the P44 UK strains (ϳ25 to 30 amino acid residues). When 16 deduced amino acid sequences from Japanese deer were compared, ϳ63% (10/16) of Japanese deer P44 sequences had Ͼ50% identity, and 37% (6/16) had 50 to 30% identity. Phylogenetic analyses showed that the 16 sequences clustered with P44 se-  quences of A. phagocytophilum from the P44 USA HZ and the P44 UK consensus sequences (Fig. 3). These P44 sequences constituted a separate cluster from the Msp2 sequences of A. phagocytophilum and from the Msp1, Msp2, Msp3, and Msp5 sequences of A. marginale (Fig. 3).
The 16S rRNA sequences from A. centrale (426 bp) from 1 Hokkaido deer, 14 Shimane deer, and two Hemaphysalis longicornis ticks removed from deer in Nara Prefecture on Honshu Island, Japan, in 2004 were 99.9 to 100% identical to each other, despite the geographic separation (two separate islands) and up to 13 years of time span between collections (1989 on Hokkaido and 2001 and 2002 in Shimane). The longer representative sequence (SS40C-L; 1,361 bp) was obtained from one of the Shimane deer; it was 99.9% identical to that of A. centrale (GenBank accession no. AF283007) previously detected in the cattle from Aomori Prefecture on northern Honshu Island, Japan (22).
Furthermore, the 16S rRNA sequences (550 bp) of A. bovis from five Hokkaido deer, six Shimane deer, and one Hemaphysalis longicornis tick removed from deer in Nara Prefecture were 99.9 to 100% identical, despite geographic separation and up to 13   FIG. 2. Comparison between the consensus-deduced amino acid sequences of p44 genes from the 16 Japanese deer (SS) obtained in this study with consensus sequences of p44 paralogs from strains from the United States (USA HZ) and the United Kingdom (UK) (11,27). Underlined letters indicate the absolutely conserved amino acids from Japan, the United States, and the United Kingdom within the hypervariable region. Dashes show sequence gaps. The region delineated by two cysteines (underlined with dots) is shorter in Japanese strains than in strains from the United States and the United Kingdom.  (54). Of 33 infected deer, 14 (42%) deer had concurrent infections with one or more Ehrlichia and Anaplasma species, where 8 deer were coinfected with two species and 6 deer were coinfected with three different species (Table 3). Thirty-three percent (11/33) of deer had concurrent infections with more than two species of Anaplasma and four deer had concurrent infections with three Anaplasma species. The three Ehrlichia-infected deer had concurrent infections with one of three Anaplasma species.

DISCUSSION
Findings from the present study show that the wild deer residing in two islands in Japan are naturally infected with four Anaplasma and Ehrlichia spp. and suggest that they may play a role in the enzootic maintenance of Anaplasma and Ehrlichia spp. in the region. White-tailed deer (Odocoileus virginianus) in the United States are known to be infected with A. phagocytophilum (3,6,31). Wild deer (Cervus elaphus, Capreolus capreolus, and Odocoileus virginianus) infected with A. phagocytophilum have also been reported in Europe (8,33,40,45). Thus, Japan is the third geographic region where a high prevalence of infection of wild deer with A. phagocytophilum is found. What is unique about A. phagocytophilum in Japan is that the 16S rRNA gene sequences were divergent from any previously reported A. phagocytophilum sequences from deer or other mammals in Europe or the United States or from ticks from Asia. The 16S rRNA gene sequence of A. phagocytophilum was found in ticks from China and Korea (11,27). A. phagocytophilum 16S rRNA gene sequences from Hokkaido and Shimane deer were 99.3% and 98.9% identical, respectively, to the Korean tick strain (GenBank accession no. AF470699; 926 bp) and were 98.5% and 98.0% identical, respectively, to the Chinese tick strain (GenBank accession no. AY079425; 919 bp). Surprisingly, the bacterium closest to A. phagocytophilum from Japanese deer, as determined by 16S rRNA gene sequence comparison, was Anaplasma sp. SA1076 from a dog in South Africa (21). Therefore, it is possible that this A. phagocytophilum strain infects domestic dogs in Japan.
The p44 gene of A. phagocytophilum encodes the immunodominant major outer membrane protein P44 (66), and multiple p44 homologs have been found in every A. phagocytophilum strain in the United States and England examined so far. Although the hypervariable regions are quite diverse, phylogenetic analyses of these p44 sequences are possible (12,30,66). To further define Japanese A. phagocytophilum strains, p44 gene sequences were determined. Sequences of 16 distinct p44 genes were obtained from three infected Japanese deer. The p44 locus consists of a central hypervariable region and 5Ј and 3Ј conserved regions (30,66). The Japanese deer p44 genes were quite unique compared to all known p44 sequences, but retained the conserved p44 group-specific deduced amino acids observed within the hypervariable region of all sequenced p44 genes. Thus, the P44 major surface protein structure of the A. phagocytophilum strain from Japanese deer had a unique feature compared with those of A. phagocytophilum strains found in the United States and England. This p44 sequence difference also may explain why serological testing using the recombinant P44 protein of an A. phagocytophilum strain from the United States (55) could not detect most of the infected deer in Japan (data not shown).
In Japan, A. phagocytophilum has not been detected in ticks. However, A. phagocytophilum was found in Ixodes persulcatus from China (11) and in Hemaphysalis longicornis ticks from Korea (27). I. persulcatus is present on Hokkaido but has not been found in Shimane Prefecture (65). H. longicornis ticks are found on Honshu Island in Japan (65), but this tick has not been noted on Hokkaido. Thus, on Hokkaido and Honshu Islands, different species of I. persulcatus and H. longicornis ticks, respectively, may serve as vectors for A. phagocytophilum transmission between mammals.
To our knowledge, the present work is the first to document infection of deer with A. centrale or A. bovis, although the seroprevalence of Anaplasma marginale among deer in the United States and Mexico has been previously reported (26,35); infection with an A. marginale-A. ovis-like agent in roe deer (Capreolus capreolus) in Spain has been previously reported as well (14,40). The sequence of 16S rRNA from deer in Shimane Prefecture was 99.9% identical to that of A. centrale from cattle in Aomori, Japan (22); 98.6% identical to that of A. marginale from cattle in the United States (strain Virginia; GenBank accession no. AF309866) (unpublished data); and 98.5% identical to that of A. centrale from cattle in Europe (GenBank accession no. AF318944) (5). Thus, wild deer may serve as the reservoir for economically important anaplasmosis of cattle caused by several Anaplasma species in the United States, Europe, and Japan. The Rhipicephalus simus tick is considered to be a vector of A. centrale in Africa (48), but in other geographic regions, a vector tick species has not been identified. Hokkaido and Shimane regions are not known to be inhabited by Rhipicephalus sp. ticks (65). In the present study, we found that the H. longicornis tick is the potential vector of A. bovis and A. centrale. Furthermore, A. bovis from an H. longicornis tick collected in Korea was reported (27).
White-tailed deer (Odocoileus virginianus) in the United States are known to be infected with Ehrlichia chaffeensis, the agent of human monocytic ehrlichiosis, and with Ehrlichia ewingii, the agent of human granulocytic ehrlichiosis (3,31,63). In the present study, the sequence (SS15E-L; 1,332 bp) detected in the deer from Shimane Prefecture was 99.9% identical to that of Ehrlichia sp. strain TS37 from H. longicornis ticks from Shimane Prefecture. This result suggests that H. longicornis ticks serve as vectors for the mammalian transmission of Ehrlichia sp. strain TS37.
Both 16S rRNA and groEL sequences of TS37 were distinct from any known Ehrlichia species and phylogenetically close to those of E. muris, E. ewingii, and E. chaffeensis. We propose to name this Ehrlichia species 'Candidatus Ehrlichia shimanensis.' All three Ehrlichia-positive deer were coinfected with Anaplasma species. Similar coinfections of deer with Ehrlichia and Anaplasma species, including E. ewingii, E. chaffeensis, or the WTD agent (an Anaplasma sp.) in Missouri or E. chaffeensis, A. phagocytophilum, or the WTD agent in Georgia were previously reported (3,31). The WTD agent was not detected in Japanese deer in the present study. The 16S rRNA gene sequences from Japanese deer had only very low levels of identity (97%) to those of the WTD agent.
Several deer were infected with three Anaplasma species, namely A. centrale, A. phagocytophilum, and A. bovis; these agents infect erythrocytes, granulocytes, and monocytes, respectively (Table 3). To our knowledge, this is the first report of concurrent infection of any animal with three Anaplasma species. The results suggest the absence of cross protection among these Anaplasma species, illustrate the potential difficulty in diagnosing the deer infection by stained blood smear and/or serological test, and support the usefulness of molecular diagnosis of Anaplasma and Ehrlichia infection.
Serum specimens are a more convenient source than wholeblood specimens for retrospective analyses of infection, since many well-preserved archival and clinical specimens are available. Although less DNA from obligatory intracellular bacteria can be recovered from serum than from whole blood, previous studies have indicated the utility of human and deer serum specimens for the nested-PCR amplification of A. phagocytophilum DNA (36). The present study showed that serum from deer blood is a good source for DNA from Ehrlichia and Anaplasma species.
This present study suggests that enzootic cycles of several Ehrlichia and Anaplasma species between ticks and wild deer are established in Japan. VOL. 72, 2006 ANAPLASMA AND EHRLICHIA SPP. IN DEER AND TICKS 1107