INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Potomac horse fever (PHF), also called equine monocytic ehrlichiosis, is caused by a monocytotropic rickettsia, Ehrlichia risticii. PHF has been identified primarily by serological means in most of the United States, as well as in Canada and some European countries. PHF is characterized by fever, depression, anorexia, diarrhea, dehydration, and leukopenia. Laminitis and colic are complications in a significant number of cases. Fatalities may result if severely affected horses are not treated promptly with fluid, electrolytes, and antibiotic therapy (23).

Since PHF started being recognized around 1978 along the Potomac River in Maryland and Virginia, it has continued to be a significant problem for horse owners in the United States, while the mode of transmission and maintenance cycle in nature has eluded researchers. The need to reveal the transmission of ehrlichial infection in horses in nature is great, as there are consistent reports of vaccine failures in the field (2, 4, 14, 16).

Prior to the isolation of E. risticii, it was discovered that horses affected with PHF develop antibodies that are reactive to Ehrlichia sennetsu (9). E. sennetsu is the agent of human sennetsu ehrlichiosis, discovered in Japan in the 1950s (6, 13). Epidemiological studies suggested that E. sennetsu infection is acquired by eating raw gray mullet fish infested with the metacercaria stage of trematodes harboring the organism (8), and feeding volunteers with raw gray mullet induced clinical signs of sennetsu fever, although reisolation of the organism from these patients was not attempted (8). Western immunoblot analysis, indirect fluorescent-antibody assay, and immunoferritin labeling demonstrated that E. risticii and E. sennetsu are closely related (26). E. sennetsu was found to be infectious yet not pathogenic to the horse, and inoculation of E. sennetsu protected horses from the development of PHF upon challenge with E. risticii (26).

Two rickettsiae have been isolated by culture from trematodes: the SF agent, an ehrlichial species isolated from the metacercariae of the trematode Stellantchasmus falcatus (7, 8, 30), which is parasitic in gray mullet, and Neorickettsia helminthoeca, which infects the trematode Nanophyetus salmincola and is the agent of salmon poisoning disease of the dog (17). Comparison of the 16S rRNA gene of several E. risticii isolates with those of N. helminthoeca, E. sennetsu, and the SF agent indicates that these are closely related to one another. This suggests that E. risticii is not related to the tick-borne species of Ehrlichia but rather shares a common ancestry with helminth-borne species of rickettsiae (17, 22, 29, 30). Western immunoblot analysis revealed a strong antigenic cross-reactivity among E. risticii, E. sennetsu, the SF agent, and N. helminthoeca, but less or no antigenic cross-reactivity between these organisms and other members of Ehrlichia spp. (20, 30). Taken together, these studies suggest that E. risticii is transmitted by a trematode.

Recently, Barlough et al. reported the presence of E. risticii DNA in freshwater operculate snails (Pleuroceridae: Juga spp.) collected from stream waters on a PHF-enzootic pasture in northern California (Siskiyou County) (1). Reubel et al. further extended these observations by showing E. risticii DNA in secretions containing virgulate trematodes released from Juga spp. (19). Furthermore, Pusterla et al. showed that by subcutaneous inoculation of trematode stages from freshwater snails (Juga yrekaensis) into horses, E. risticii could be isolated from the horse by using P388D1 cells (18).

In the present study, E. risticii DNA is shown to be consistently present throughout the summer and early fall in the cercariae and/or rediae of virgulate trematodes infesting the Pleuroceridae snail Elimia livescens in Ohio. The DNA sequences of E. risticii from the trematodes in Ohio were compared with those of E. risticii from California snails or trematodes and with those from E. risticii from the blood of horses in various states including those newly sequenced in the current report, as well as with that of the SF agent. Previously, laboratory mice have been shown to be able to be experimentally infected with E. risticii (21, 24, 25, 27). Therefore, in order to demonstrate infectious E. risticii is present in the trematodes, mice were inoculated with homogenized PCR-positive trematode specimens. We demonstrate here that E. risticii from trematodes can establish infection and be serially passaged in mice.

(A part of this paper was presented at the 99th General Meeting of the American Society for Microbiology, Chicago, Ill., 30 May to 3 June 1999 [M. Kanter, J. Mott, N. Ohashi, B. Fried, S. Reed, Y. Lin, and Y. Rikihisa, Abstr. 99th Gen. Meet. Am. Soc. Microbiol., abstr. D/B-122, p. 233, 1999].)

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Snail collection. Freshwater snails were collected from the Walhonding River, located in Warsaw, Ohio (Coshocton County), between June and October 1998. This region had at least six cases of laboratory-confirmed PHF in 1997 and 1998. In total, more than 400 snails were collected by hand on seven different dates (Table 1) from a rocky shoreline. Snails were collected the day prior to use and were kept in dechlorinated tap water at 4°C overnight. The snails were segregated by size, and only the larger snails (greater than 1 cm) were used.

Processing of snails. Snail shells were broken with a mallet, and the shell was separated from the body using forceps. A small amount of distilled water was added, and the snails were minced in the digestive gland-gonad (DGG) complex using forceps. DGG complexes were then observed for the presence of trematodes under a dissection microscope. If cercariae or rediae were observed, approximately 10 mg of trematode tissue from the DGG complex of each snail was removed, centrifuged to remove as much snail tissue as possible, and used for testing by PCR. The dissected snail was then kept at 4°C until the results of nested PCR analyses were known (9 to 15 h).

Isolation of DNA from trematodes. DNA extraction was performed using a QIAamp tissue kit (Qiagen, Inc., Chatsworth, Calif.) according to the manufacturer's protocol, except that pelleted trematode-snail tissue (approximately 10 mg) was lysed in 1 mg of proteinase K (GIBCO-BRL, Gaithersburg, Md.) in 180 µl of buffer ATL from the kit for 3 h. The DNA concentration was determined on the basis of readings of A₂₆₀ and A₂₈₀ with a GenQuant II spectrophotometer (Pharmacia Biotech, Uppsala, Sweden).

Detection of E. risticii 16S rDNA. DNAs (approximately 1 µg) extracted from DGG complexes were used as templates in 50-µl reaction mixtures containing 5 µl of 10× PCR buffer, 2 µl of 50 mM MgCl₂, 1 µl of 10 mM deoxynucleoside triphosphate (dNTP) mixture, and 1.25 U of Taq polymerase (GIBCO-BRL). ER5-3, which is specific for the 16S rRNA gene of E. risticii, and ER3-2, which is specific for the 16S rRNA gene of all Ehrlichia spp. (2), were used as the external primer pair (8 pmol of each). Samples were amplified by an initial denaturing period of 5 min at 94°C, followed by denaturation (94°C, 1 min), annealing (60°C, 1 min), and extension (72°C, 1 min) for 30 cycles in a DNA Thermal Cycler 480 (Perkin-Elmer, Foster City, Calif.). Final extension followed at 72°C for 7 min. In a second PCR, 1 µl of the PCR product of the first reaction was added as the template to a 50-µl reaction mixture containing 5 µl of 10× PCR buffer, 2 µl of 50 mM MgCl₂, 1 µl of 10 mM dNTP mixture, 1.25 U of Taq polymerase, and 8 pmol of each primer. E.ris 1 (5'-GGAATCAGGGCTGCTTGCAGCCT-3'; forward primer) and E.ris 2 (5'-TGTGGGTACCGTCATTATCTTCCCCACT-3'; reverse primer), which are specific for E. risticii, were used in the nested reactions. PCR conditions for the nested reaction were identical to those for the first reaction. PCR products were electrophoresed in a 1.5% agarose gel, stained with ethidium bromide, and photographed by using a still-video documentation system, Gel Print 2000I (Biophotonics, Ann Arbor, Mich.).

Amplification of 16S rRNA gene. Overlapping nested and seminested PCR products were used to obtain nearly complete 16S rRNA gene sequences. The 5' end of the gene was amplified using primers ER5-3 and ER3-2. These primers were then removed by using a Centri-Sep spin column (Princeton Separations, Adelphia, N.J.) according to the manufacturer's suggested method. Seminested PCR was accomplished using primers ER5-3 and ER130-113 (5'-AAGTTCCCACGCGTTACG-3'). The central portion of the gene was amplified by primers ER200-219 (5'-TTGCTATCAGATAGGCCCGC-3') and ER1111-1092 (5'-TTCCTTAAAGTTCCCGGCCG-3') and primers ER783-806 (5'-TTAAAAGTGGGTTATTTTATCTGC-3') and ER1498-1479 (5'-AAAGGAGGTAATCCAGCCGC-3'). The products (1 µl) were then used as the templates in nested reactions using primers ER383-402 (5'-CGCATGAGTGATGAAGGCCC-3') and ER1015-995 (5'-AGCCATGCAACACCTGTGTTG-3') and primers ER932-952 (5'-CTTACCATACCTTGACATGTG-3') and ER1452-1434 (5'-GACTTAACCCCAGTCACCC-3'), respectively. The 3' end of the gene was amplified in the same manner as the 5' end, except that different primers were used. The first-round primers used were ER783-806 and ER1498-1479, and the seminested PCR primers used were ER1199-1220 (5'-AACTACAATGAGCTAGCTACAC-3') and ER1498-1479. All PCR reactions were accomplished using 50-µl reaction mixtures containing 5 µl of 10× PCR buffer, 2 µl of 50 mM MgCl₂, 1 µl of 10 mM dNTP mixture, 1.25 U of Taq polymerase, and 8 pmol of each appropriate primer. For seminested PCR reactions, 3 µl of template was used. Temperature cycling conditions were the same as those used for the detection of the 16S rRNA gene of E. risticii from DGG complexes, except that the reaction using primers ER383-402 and ER1015-995 had an annealing temperature of 64°C. The primer positions were designated on the basis of the 16S rRNA gene sequence of the type strain.

Amplification of 51-kDa antigen gene. A portion of the 51-kDa antigen gene was amplified using DNAs (approximately 1 µg from the S21 trematode pool [collected 11 August 1998] and 500 ng of purified organisms from the OV and Ohio 081 isolates) as described previously (1).

Sequencing of 16S rRNA gene and 51-kDa antigen gene. Amplified PCR products of the 16S rRNA gene and the 51-kDa antigen gene were cloned in the PCRII vector of a TA cloning kit (Invitrogen Co., San Diego, Calif.) as described by the manufacturer. Recombinant plasmids were purified using the Concert Rapid Miniprep system (GIBCO-BRL) and were sequenced by a dideoxy termination method with a 373A DNA sequencer (Applied Biosystems).

Sequence analysis. DNA and amino acid sequence analyses and phylogenetic studies were performed as described previously (15).

Trematode preparation and mouse inoculation. The PCR-positive trematode pools from each snail were washed three times with distilled sterile water and homogenized in 0.5 ml of RPMI 1640 (GIBCO-BRL) in a microtube with a motor-driven pestle (Kontes, Vineland, N.J.) on ice. Effective homogenization was verified under a light microscope. The three homogenized trematode pools were then inoculated intraperitoneally into three female CF-1 mice (4 weeks old; Harlan Sprague Dawley, Indianapolis, Ind.).

Mouse passage. Mice that were inoculated with trematode homogenate were sacrificed on day 11 postinoculation (p.i.). Detection of E. risticii in these mice was done by PCR, as described previously, using 1 µg of DNA extracted from the liver. The spleen was cut in half, and the capsule of the spleen was removed. The remaining spleen tissue was teased with forceps. The released cells were then resuspended in 0.5 ml of RPMI 1640, and the preparation from each mouse was inoculated into new CF-1 mice (first passage). On day 9 p.i., first-passage mice were sacrificed. Testing of these mice for E. risticii as well as homogenized spleen passage (second passage) was done in the same manner as described for the first passage. On day 15 p.i. second-passage mice were sacrificed and tested for the presence of E. risticii.

Nucleotide sequence accession numbers. The E. risticii sources (explained in footnotes of Tables 2 and 5) and the GenBank database accession numbers for the 16S rRNA nucleotide sequences used for comparison in this study are as follows: E. risticii Illinois (type strain), M21290; Buck, AF036648; Bunn, AF036649; Danny, AF036650; Doc, AF036651; Dr. Pepper, AF036652; Eclipse, AF036653; Juga, AF036654; KLSN, AF036655; Mostly Memories, AF036656; Ms. Annie, AF036657; SF agent, U34280; SHSN-1, AF037210; SHSN-2, AF037211; Tate, AF036658; and Thorenberg, AF036659. The sequences for the OV and Ohio 081 strains and the SRC agent were derived from published sources (10, 29). The sequences for the S22 and S6 trematode pools (collected on 11 August 1998 and 27 October 1998, respectively) have been assigned accession numbers AY005439 and AY005441, respectively. The accession numbers of the 51-kDa antigen gene sequences are as follows: E. risticii 90-12 strain, U85784; E. risticii 25-D strain, U85785; Doc, AF036671; Dr. Pepper, AF036672; Eclipse, AF036673; Juga, AF036674; Ms. Annie, AF036675; SHSN-1, AF037215; SHSN-2, AF037216; "Shasta River Crud" (SRC) agent, AF037217; and Thorenberg, AF036676. The sequences of the 8/11/98 S21 trematode pool and the OV and Ohio 081 strains have been assigned accession numbers AY005440, AY005442, and AY005443, respectively.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Microscopic examination of trematodes. The snails found infested with E. risticii PCR-positive trematodes were operculate snails of the Pleuroceridae family and were identified as E. livescens by John B. Burch, University of Michigan (Fig. 1). Upon dissection of these snails, cercariae and sporocysts were observed in the DGG complex of 50 to 80% of the snails (Table 1). The presence of trematodes was related to the size of the snail. The larger the snail (longer than 1 cm), the more probable was the presence of trematodes. The majority of cercariae were identified as virgulate xhiphidiocercaria based on their morphology. Each cercaria had a thin-walled excretory vesicle, a virgula organ for secreting mucous, ventral and oral suckers, and a stylet located in the oral sucker; eyespots were not seen. Each cercariae had a simple tail that was not forked, was not greater in length than the body, and was without finfolds. A cercaria from an E. risticii PCR-positive trematode pool is shown in Fig. 2. Further information on the life cycle of this trematode is necessary to identify the species.

View larger version (134K): [in this window] [in a new window]   FIG. 1.   Pleuroceridae: E. livescens snails harboring trematodes collected from the Walhonding River in Warsaw, Ohio (Coshocton County). The entire length of the ruler on the left is 2 cm.

View larger version (84K): [in this window] [in a new window]   FIG. 2.   Light microscope picture of a virgulate cercaria collected from a PCR-positive DGG complex of an E. livescens snail. Magnification, ×168.

Sequence analysis of 16S rRNA genes. Nearly complete (1,455- and 1,342-bp, respectively) 16S rRNA gene sequences of E. risticii from two pools of trematodes, S22 and S6, were obtained. Comparison of these sequences to 10 almost complete and 9 partial GenBank-accessible sequences (base 39 to 563 in the type strain) of E. risticii isolated from the blood of naturally infected horses, snails (SHSN-1 and SHSN-2), or trematodes (Juga) in California or an E. risticii strain isolated by culture from the blood of infected horses is shown in Table 2. Trematode pool S22 was completely identical to the type strain, while sequence S6 had 1 nucleotide transition at position 131 (in reference to the type strain). Only the OV isolate, which was isolated by culture from the blood of a horse in Kentucky (2), and the partial sequence from the blood from the infected horse Doc had the same transition at this position.

Amino acid sequence analysis of 51-kDa protein. The 51-kDa protein is a unique antigenic protein cloned from the E. risticii genome and shows no homology with other known proteins (27). The amino acid sequence of the 51-kDa antigen of S21 was compared to the corresponding region from 13 other E. risticii sequences (Table 5). Included in this comparison were the new sequences of the OV and Ohio 081 horse-blood culture isolates, determined in this study. The sequence of S21 was closest to the 90-12 strain, which was isolated by culture from the blood of a horse in Maryland, differing by 4 amino acids at the 3' end. The Ohio 081 strain was the most divergent from other E. risticii sequences. Compared to the 90-12 strain, the OV strain had 1 amino acid deletion whereas the Ohio 081 strain had 7 amino acid deletions in this segment.

Phylogenetic analysis of 51-kDa amino acid sequence. A phylogram (cladogram) constructed using the unweighted pair-group method of analysis is shown in Fig. 3. The sequence of the trematode pool reported in this paper is in a cluster with the sequences of the 90-12 strain, the 25-D strain, Eclipse, and the newly reported OV strain. A separate cluster contains the sequences from snails (Juga, SHSN-1, and SHSN-2) as well as sequences from Dr. Pepper, Ms. Annie, Thorenberg, and Doc. The sequence of the Ohio 081 strain is clearly distinct from the other ehrlichial sequences compared.

View larger version (17K): [in this window] [in a new window]   FIG. 3.   Phylogram of partial amino acid sequences of 51-kDa antigens of the trematode pool S21 from Ohio, of sequences obtained from the trematode or snails from California, and of sequence obtained from the blood of naturally infected horses. The evolutionary distance values were determined by the method of Kimura, and the tree was constructed by the unweighted pair-group method analysis using the PHYLIP software package (5). Bar, 1% divergence in amino acid sequences. 8/11/98, collection date for S21 trematode pool was 11 August 1998.

Detection of E. risticii infection of mice. In order to demonstrate that infectious E. risticii was present in PCR-positive trematodes, mice were inoculated with PCR-positive trematodes. Three trematode pools (S9-S12, S6-S8, and S19-S22 [all collected 27 October 1998]) were inoculated into separate mice. Upon successive passage, one mouse passage line (from trematode pool S9-S12) was positive by PCR in mice through two passages and the initial trematode inoculation, but S19-S22 and S6-S8 were positive by PCR on only the first mouse passage and only the initial trematode homogenate inoculation, respectively. Data on the mouse passage study is summarized in Table 7. Mice did not show any clinical signs.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The present study showed that throughout the summer and early autumn months (June to October), a particular type of snail in Ohio is infested with a particular type of trematode, which has consistent E. risticii infection rates. The months correspond to the occurrence of PHF in Ohio, and these trematodes appear to be the persistent source of E. risticii. Based on the high trematode infestation rate of the snail, the definitive host of this trematode must be quite abundant in this locality.

Elimia and Juga spp., which are found to be infested with E. risticii-positive trematodes in Ohio and California, respectively, belong to the Goniobasis family (3). The Goniobasis family includes Oxytrema silicula, which is the intermediate host of N. salmincola, the trematode that serves as reservoir and vector of N. helminthoeca (12). N. salmincola is also a virgulate trematode like those found in Juga and Elimia spp. In addition, numerous Elimia laqueata snails were collected from a Kentucky farm, where E. risticii was isolated from the blood of three horses (2). Thus, it appears that there are broad geographic regions of the United States in which trematodes that use Goniobasis snails as an intermediate host may be commonly infected with Ehrlichia and Neorickettsia spp. It was reported that E. risticii DNA was also found in a pool of lymnaeid snails (Stagnicola sp.) from Klamath Falls, Oregon, which is north of the Siskiyou County site (19). The SF agent, which is 99% related to the type strain of E. risticii in its 16S rRNA gene sequence, parasitizes the S. falcatus trematode. Stellantochasmus formosanum, a related trematode species, is found in the family Thiaridae (Melaniidae) snail (11). It is thus possible that additional types of snails harbor trematodes infected with E. risticii.

The virgulate trematodes described in the current study and found in California (19) had a tail, distinct from N. salmincola, which has a short stubby tail (12). Trematodes found in Ohio lack tail finfolds, whereas trematodes found in California are described to have tail finfolds (19). Whether multiple species of trematodes carry E. risticii remains to be determined.

The 16S rRNA gene sequence comparison clearly showed the sequences found in the trematodes in Ohio are those of E. risticii. The 51-kDa antigen gene, which by GenBank query is a unique protein limited to the E. risticii group, is of greater use in comparing closely related strains. By analysis of the majority of E. risticii sequences available, it appears that E. risticii population diversity is limited by geographic constraints, though more samples would be necessary to confirm this. For the most part, those sequences detected in the eastern section of the United States (from Maryland to Kentucky) show similar sequences in both the 16S rRNA gene and the 51-kDa protein amino acid sequences, while those in the western United States fall into another group. The majority of strains isolated by culture from the blood of horses in Ohio have 16S rRNA gene sequences identical to that of the type strain (29). However, 16S rRNA gene sequences of the Ohio 081 strain and the Bunn isolate dramatically differ by 10 and 14 bases, respectively, when compared to the type strain. In addition, the Ohio 081 strain showed extreme divergence in the 51-kDa protein amino acid sequence, though this was expected, as the Ohio 081 strain did not react with monoclonal antibodies to the type strain of E. risticii, and the Western immunoblot profile was distinct from that of the type strain as well as those of other strains examined (2). The basis for these phylogenetically distant strains of E. risticii is unknown, but specific snails and/or trematodes, which carry these strains, may be involved. Genetic comparison of strains from Ohio and elsewhere shows no difference in those from the blood of horses and those from snails or trematodes. This has significant consequences, as it appears that strains similar to those that affect horses are also present in trematodes.

Mice inoculated with the PCR-positive trematodes did not show any clinical signs or pathologic changes in the current study. The number of viable E. risticii present in the trematodes may be too few to cause illness, since the development of clinical signs and pathologic changes in mice upon inoculation of E. risticii is dose dependent (25). Alternatively, the E. risticii strain present in the trematodes in Ohio may not be virulent in mice. Previously, members of our group found that strain Ohio 380, isolated from a horse in Ohio, can infect mice but does not cause disease (14). The disappearance from mouse liver of E. risticii detectable by PCR shown in Table 7 may be due to either of the above-mentioned reasons.

By compiling information on additional strains of E. risticii in trematodes in diverse geographic regions, the process of spontaneous evolution and host immunity- or environment-induced selection of E. risticii strains may become evident. Future studies will provide much needed information upon which further ecological, epidemiological, and pathogenesis studies can be built in order to develop preventive strategies for limiting exposure, thus limiting the use of ineffective vaccines and dependence on antibiotic therapy.