INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The members of the genus Ehrlichia are obligate, intracellular bacteria within the order Rickettsiales. Although ehrlichia infections of veterinary importance were first described in 1935, the first case of human ehrlichiosis in the United States was reported in 1987 (17). The human pathogen was subsequently identified as Ehrlichia chaffeensis (1). In 1994, a second ehrlichia infection of humans was reported and has been referred to as human granulocytic ehrlichiosis (HGE) owing to the proclivity of the agent to infect neutrophils (5). The majority of HGE cases have been diagnosed in the Northeastern and Upper Midwestern areas of the United States, although a limited number of cases have been reported in Europe and in northern California (2, 4, 9-12, 22, 26).

The genus Ehrlichia has been divided into three genogroups based on analysis of 16S rRNA sequences, and the HGE agent is a member of the Ehrlichia phagocytophila genogroup (29). E. phagocytophila is the etiologic agent of tick-borne fever in ruminants in Europe, and the E. phagocytophila genogroup also includes Ehrlichia equi, the agent of equine granulocytic ehrlichiosis. These three granulocytic ehrlichiae (GE) are very closely related based on numerous criteria including the high degree of homology of the 16S rRNA and groE DNA sequences and may represent strains of a single species (5, 14, 25). However, the analysis of genetic elements less conserved than the 16S and groE sequences is needed to accurately assess the phylogenetic relationship of these agents and the degree of variability at the subspecies level. Recently, several genes that may encode structural proteins were identified from an HGE agent (USG3 strain) expression library and were recognized by serum from humans and canines infected with GE (24). One of these is referred to as the ank gene, owing to a series of repeats within the predicted 131.2-kDa protein product that are similar to the repeats within the human erythrocyte ankyrin protein (24). We have PCR amplified and sequenced the complete ank gene open reading frame and 28 bp 5' of the initiating methionine and 13 bp downstream from the termination codon. Samples examined were the HGE agent from 10 confirmed cases (3 from New York, 4 from Wisconsin, 1 from Sweden, and 2 from Slovenia), a canine granulocytic ehrlichiosis agent from a Minnesota dog, E. equi, and E. phagocytophila from a Swedish cow.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Samples and sample preparation. All samples, except for the USG3 strain, were obtained as EDTA blood specimens. Human samples were from HGE cases confirmed by seroconversion or by amplification and DNA sequencing of the 16S rRNA gene. The Swedish sample was from a 32-year-old previously healthy woman who presented at an outpatient clinic in Ronneby (Blekinge County, southern Sweden), with a 5-day history of fever (38 to 39°C), chills, headache, and myalgia. The patient worked part-time as a farmer, and she had recently been involved in the hunting and handling of a slaughtered roebuck. She also reported that she had sustained five or six tick bites during the month prior to admission. The patient's acute-phase blood showed a reciprocal titer of 1:160 and was PCR positive when tested at the Centers for Disease Control and Prevention using the 16S rRNA gene as the target (data not shown).

PCR analysis. Both nested and direct PCR protocols were used. Direct PCR amplifications consisted of 40 cycles with each cycle including a 30-s denaturation at 94°C, a 30-s annealing at 55°C, and a 1-min extension at 72°C. The 40 cycles were preceded by a 2-min denaturation at 95°C and followed by a 5-min extension at 72°C. PCR amplifications were performed in a Perkin-Elmer 9600 thermal cycler (Perkin-Elmer, Applied Biosystems Division, Foster City, Calif.), and reagents were from the GeneAmp PCR Kit with AmpliTaq DNA polymerase (Perkin-Elmer). Primary reactions used 5 µl of purified DNA as template in a total volume of 50 µl. Amplifications contained 200 µM (each) deoxynucleotide triphosphates (dATP, dCTP, dGTP, and dTTP), 1.25 U of Taq polymerase, and 0.5 µM (each) primer. Reaction products were subsequently maintained at 4°C until analyzed by agarose gel electrophoresis or used as template for nested reactions.

DNA sequencing and data analysis. DNA sequencing reactions used fluorescently labeled dideoxynucleotide technology (dye terminator cycle sequencing ready reaction kit; Perkin-Elmer). Sequencing reaction products were separated, and data were collected using an ABI 377 automated DNA sequencer (Perkin-Elmer). The sequence was fully determined for both strands of each DNA template to ensure maximum accuracy of the data. Sequences were edited and assembled using the Staden software programs (6) and analyzed using the Wisconsin Sequence Analysis Package (Genetics Computer Group, Madison, Wis.) (8).

Nucleotide sequence accession numbers. GenBank accession numbers for the ank gene sequences are as follows: EE (E. equi), AF100882; NY1, AF100883; NY2, AF100884; NY3, AF100885; Sl-HG1, AF100886; Sl-HG2, AF100887; Sw-HG, AF100888; EP (E. phagocytophila), AF100889; WI1, AF100890; WI2, AF100891; WI3, AF100892; WI4, AF100893; and MN-dog, AF100894.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

PCR amplification. The complete ank gene, including 28 bp upstream of the start codon and 16 bp downstream from the stop codon, was amplified by using a combination of multiple primers and a nested PCR strategy. PCR primers were initially designed based on the ank gene sequence of the USG3 strain (24). Additional sequencing of the ank gene from this strain revealed that the open reading frame consists of 3,696 and not 2,244 bp as originally reported (24). The ank gene was divided into seven overlapping regions of 550 to 600 bp, and primer sets were designed to specifically amplify each of these regions. The locations of primers used for PCR and sequencing are shown in Fig. 1. A nested PCR strategy was used to amplify the ank gene for two reasons. First, the initial direct PCR amplification attempts resulted in little or no product with many of the templates (data not shown) because many of the clinical samples used in this study contained a low concentration of the ehrlichial agents. Second, several samples were limited by the volume of the sample. Therefore, a nested protocol that allowed both increased sensitivity and conservation of limited amounts of samples was developed. The primary reactions used primers U7 and 1R1 to amplify the 5' portion of the gene and primers 1F and 4R1 for the 3' region. The products of these reactions were used as templates for each of the specific nested or heminested reactions for the seven regions. The downstream regions were subsequently amplified independently by using primary reactions with 4F2 and D2 followed by nested reactions using 4F3 and D1.

View larger version (17K): [in this window] [in a new window]   FIG. 1.   Primers used for amplification of the ank gene and topography of the deduced Ank protein. The rectangle represents the coding region in 5'-to-3' orientation. The length indicated for the open reading frame (3,696 bp) is for the USG3 strain. The location and orientation of the primers used for PCR amplification and DNA sequencing are shown relative to the ank gene coding region. The products of each primer pair are shown as dashed lines. Superimposed on the coding region are the ankyrin-repeat region and the region of the coding sequence containing the three types of repetitive elements in the deduced Ank protein for the USG3 strain of the HGE agent.

DNA sequence analysis. The samples that were used for PCR amplification and sequencing of the ank gene are shown in Table 1. The total number of base pairs sequenced from each specimen and the predicted number of amino acids encoded within the gene are also shown in Table 1. The 28 bp upstream of the predicted initiating codon and 16 bp downstream of the stop codon were identical in length for each sample used in this study. Each of the three human samples from New York showed an ank coding region equal in length to that of the USG3 reference strain (3,696 bp), and the nucleotide sequences for two of the New York samples (NY1 and NY2) were identical to the nucleotide sequence of the USG3 strain. Likewise, each of the samples from the Upper Midwest (WI1, WI2, WI3, WI4, and MN-dog) showed sequences of equal size (ank coding region of 3,693 bp), and the E. equi and E. phagocytophila sequences were of equal size (3,615 bp). In contrast, each of the three European HGE sequences showed unique sizes: 3,618 bp for the Swedish HGE agent and 3,669 and 3,720 bp for the Slovenian samples. A feature unique to the Northeastern U.S. samples (New York and USG3) is a 3-bp insertion (TTT) at nucleotide position 2470 that is absent in all other samples examined in this study. This single codon insertion is also responsible for the 3-bp difference in size between the New York (3,696-bp) and the Upper Midwest (3,693-bp) samples.

Protein sequence analysis. The predicted amino acid sequences were also aligned, and the percent identities are shown in Table 2. Similar to the nucleotide sequence results, the amino acid sequences clearly could be classified as belonging in one of two groups, either North American or European, and the North American group could be further subdivided into either Upper Midwest or Northeastern samples. Interestingly, the percent identities between the groups were lower for the amino acid sequences than for the nucleotide sequences, indicating that many of the nucleotide changes represent nonsilent mutations that result in protein sequence variations. Amino acid homology between the Upper Midwest and Northeastern samples ranged from 96.34 to 96.43%. In contrast, the amino acid sequences of each of the four samples from the Northeast were identical, and only a single amino acid difference was noted among the Upper Midwest samples (99.92% identity). The homology within the European HGE group was also very high and ranged from 99.92 to 100%. The Swedish E. phagocytophila sequence (EP) showed lower homology, with a range from 92.61 to 93.61% identity to members of the North American group and from 95.44 to 95.52% identity to the European HGE group. The highest percent identity noted between members of the European and North American groups was 95.27% (Sw-HG and WI2).

Repetitive elements. The sequence of the ank gene for the USG3 strain showed two copies of an 81-bp (27-amino-acid) repeat beginning at nucleotide position 2828 (23). This repeat was also found in the ank gene sequence for each of the North American HGE agent samples and the Minnesota canine sample. However, only a single copy of the 81-bp sequence was found in the E. equi sequence, resulting in an 81-bp deletion in the E. equi ank gene relative to the other North American samples that were tested. Similarly, a single copy of the 81-bp element was present in each of the European samples. The Slovenian ank gene sequences also showed a variable number of copies of a 51-bp (17-amino-acid) repetitive element located directly adjacent to and downstream from the 81-bp element. Whereas each of the North American, Swedish HGE (Sw-HG), and E. phagocytophila (EP) samples showed a single copy of this 51-bp sequence, the Slovenian HGE samples (Sl-HG1 and Sl-HG2) showed two and three copies of the element, respectively. Two 11-amino-acid repeats were conserved in all samples examined. The arrangement of these repetitive elements (represented by the number of amino acids) within the ank gene for each sample examined is shown in Fig. 2.

View larger version (23K): [in this window] [in a new window]   FIG. 2.   Variable number of repetitive elements within the ank gene. The number of 81-bp (27-amino-acid) and 51-bp (17-residue) repeats and the arrangement of the repeats are shown for the GE examined. N.A., North American; E. phag., E. phagocytophila.

Phylogenetic analysis. The ank gene sequences were used as a phylogenetic tool to assess the relationship of the GE that were examined in this study. Samples with sequences that were identical were removed from the analysis but are shown in the phylogram in Fig. 3. These results separate the North American samples into two clades, one including the New York HGE and USG3 samples and another with the Wisconsin HGE and Minnesota dog samples. Likewise, the phylogram effectively positions the four European samples in a separate clade. The position of the E. equi sequence in the Upper Midwest clade is supported by 72 bootstrap replicates.

View larger version (14K): [in this window] [in a new window]   FIG. 3.   Phylogenetic analysis of the GE based on the ank gene DNA sequences using maximum parsimony as the optimality criterion. The number of bootstrap replicates (from a total of 100) that were in agreement are shown on each branch. Branches without bootstrap values represent polytomies that could not be resolved because the sequences were too similar.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Members of the GE genogroup have been shown to cause disease in numerous vertebrate species including dogs, cattle, sheep, horses, and humans. To date, analysis of the members of this group at the molecular level has consisted primarily of examination of the DNA sequence of the 16S rRNA gene. These analyses have shown very little difference within the 16S rRNA sequences, suggesting that the HGE agent, E. equi, and E. phagocytophila represent strains of a single species. Additionally, the 16S rRNA sequences determined for the HGE agent from confirmed human cases from both North America (Upper Midwest and Northeast) and Europe, including those used in this study, have been identical, suggesting that the agent causing human disease in these areas may represent a single strain. Recently, two human cases were reported from northern California that showed 16S rRNA sequences differing from the HGE agent by 1 bp and identical to the E. equi sequence, suggesting that an E. equi-like strain may be causing infections in northern California (10). In contrast to the 16S rRNA data, studies using DNA sequence and Western blot analyses of the major antigenic proteins have suggested a high degree of variability among HGE isolates from both the Northeastern and Upper Midwestern United States (13, 32, 33). Although analysis of the 16S rRNA gene sequence is a powerful tool for identifying novel agents and for determining relatedness at the genus and species level, the strong conservation of the 16S rRNA makes it less than ideal for differentiation of closely related species or subspecies. Sequences of genetic elements more variable than the highly conserved ribosomal subunits and housekeeping genes, such as genes encoding structural proteins, are often needed to make these determinations. Recently, the sequences of several genes encoded by the GE strain USG3 were determined, one of these being the ank gene. Analysis of the predicted peptide sequence suggested that the ank gene may encode a structural protein, and in this study the complete sequence of the ank gene was determined for 13 members of the E. phagocytophila genogroup including samples from 10 confirmed HGE cases.

Phylogenetic analysis of the ank gene sequences separated the HGE agent into three distinct clades representing the U.S. Northeast and Upper Midwest and Europe. Analysis of the ank gene sequence thereby represents the first epidemiologic tool able to differentiate the HGE agent based on geographic origin of the sample and the first to indicate that the etiologic agents of HGE present in each of these locations are evolving independently. The sequence conservation noted for the HGE agent sequences within each of the three clades was remarkably high, with both nucleotide and amino acid identities of >99.9%. This was in contrast to nucleotide and amino acid sequence homologies among members of the three clades that were significantly lower (nucleotide sequence identity of <98.2%; amino acid identity of <96.5%). Numerous identical sequences were noted for members within the same clade, including three of the four samples from the U.S. Northeast. Likewise, three of the four HGE agent samples from the Upper Midwest were identical. The samples from the Northeast were also collected in three different years, 1993 (USG3), 1994 (NY2), and 1996 (NY1 and NY3), suggesting that there is minimal year-to-year variation within this population. Furthermore, PCR amplification and DNA sequencing of a 500-bp region of the ank gene from additional samples collected from confirmed HGE cases in Minnesota, Wisconsin, and Connecticut support the results described herein and the conclusion that two distinct clonal populations of the HGE agent exist in the United States, defined by geographic location, Northeast and Upper Midwest (data not shown).

The USG3 strain was isolated from a canine infected by allowing Ixodes scapularis ticks collected from Westchester County, N.Y., and Montgomery County, Pa., to feed. The isolation of this strain has raised questions concerning the authenticity of the USG3 strain as a representative of the HGE agent and the efficacy of using this strain as an antigen for serodiagnostic purposes (19, 31). However, the sequence determined for the ank gene of the USG3 strain was identical to that of two of the samples from confirmed HGE cases in New York, suggesting that the USG3 strain likely represents a strain of the HGE agent. Similarly, the sequence determined for the Minnesota dog sample was identical to three of the four Wisconsin HGE samples, suggesting that a single strain, or a very closely related one, is capable of infecting and causing clinical illness in both human and canine species. In addition, portions of at least six genes other than the 16S rRNA, groE, and p44 have been sequenced from the USG3 strain and were identical to those from a New York human isolate (C. I. Murphy et al., unpublished observations).

The ank gene sequences were determined and analyzed for single representatives of E. equi and E. phagocytophila and were found to be no more closely related to each other than to the North American and European HGE agent sequences. In fact, the lowest homology found between any two amino acid sequences (92.37%) was between the predicted E. equi and E. phagocytophila proteins. While the E. phagocytophila nucleotide and amino acid sequences were clearly more closely related to the European HGE agent sequences than to those of any of the North American samples, the E. equi sequences showed highest homology to those from the Upper Midwest samples. Additional representatives of both of these species need to be examined before drawing any conclusions, particularly for E. phagocytophila, where previous studies of the groE heat shock operon sequences showed significant variability between E. phagocytophila strains relative to the high conservation noted for the 16S rRNA gene (25). However, a partial sequence of the ank gene from a Swiss E. phagocytophila sample showed >99% identity to the Swedish E. phagocytophila examined in this study and suggests a close relationship between European strains of E. phagocytophila (data not shown).

One of the more distinctive features within the ank gene sequence is a region containing a variable number of direct repeats with no ankyrin homology located in the last third of the gene. The result of these repeats at the protein level is a 27-amino-acid sequence that is repeated twice in each of the North American samples, with the exception of E. equi. The European samples and E. equi show a single copy of the 27-amino-acid element, suggesting that this organism may be more closely related to European GE than are the other North American GE. In fact, the number and arrangement of repetitive elements for E. equi are identical to those noted for both the Swedish HGE agent and E. phagocytophila. However, the partial sequence of the ank gene for the BDS strain of E. equi (GenBank accession no. AF047897) shows a repeat structure identical to that of the North American HGE agents and suggests that the number of copies of the 27-amino-acid repeat in E. equi strains can be variable. Directly adjacent to and downstream from the 27-amino-acid unit is a 17-amino-acid element that is present as a single copy in each of the North American HGE samples, E. equi, the Swedish HGE agent, and E. phagocytophila. Only the European HGE samples from Slovenia show multiple copies of the 17-amino-acid element, with either two or three copies present, and thereby appear to represent a feature unique to the HGE agent in Slovenia.

A comparison of the members of the three clades of the HGE agent (U.S. Northeast and Upper Midwest and Europe) showed that the degree of homology is not uniform across the length of the ank gene or the predicted protein. Analysis of the amino-terminal region of the protein sequence for the members of each clade showed a similar degree of identity among the three clades (92.0 to 94.7%). In contrast, the carboxyl-terminal region of the protein is very highly conserved between the members of the two North American clades (>99.7% identity), while the European clade members show only 94.3% identity to each of the North American clades. These data suggest that there are constraints on the evolution of the carboxyl-terminal portion of the protein in the North American strains that are not being exerted on the more variable amino-terminal region. It is the amino-terminal region of the protein that contains the ankyrin-like repeats from which the name of the gene is derived. Ankyrin-related proteins have been described for bacteria, plants, and animals, and the most common function attributed to ankyrin-like repeats involves protein-protein interactions (3, 16). The diversity within the amino-terminal region of the HGE agent Ank protein indicates that, although the length of the amino-terminal region is conserved, the content is flexible. The fact that there are between 14 and 18 copies of ankyrin-like repeats may allow for minor variations within the repeats to have little or no effect on the overall function of the protein, thereby allowing more diversity within this region of the protein.

The life cycle of the HGE agent involves a complex interaction between the natural host(s) or reservoirs and vectors that progress through multiple life stages and may transmit infections to humans. Additionally, tick species that rarely bite humans, including nidicolous members of the genus Ixodes, such as Ixodes trianguliceps in Europe, are likely involved as vectors for maintaining enzootic infectious cycles (20). This requires that the agent have the ability to adapt to these multiple environments and environmental pressures. The differences noted for the Ank protein among members of the three clades may reflect adaptation to differences between the vectors and/or reservoirs of the agent. While the presumed vector, I. scapularis ticks, and the major reservoir, Peromyscus leucopus, are present in both the Upper Midwestern and Northeastern regions of the United States, the genetic and biologic diversity between these populations living in distinct geographic locations has not been fully explored. There are also potentially significant differences between the vectors and reservoirs that are involved in the life cycles of the agents in North America and Europe. Whereas I. scapularis is the primary vector in the United States, a different species of Ixodid tick, Ixodes ricinus, has been suggested as the vector in Europe (7, 15, 18, 20-23, 28). Whereas P. leucopus has been shown to be a competent natural reservoir in the United States (27, 30), the corresponding reservoir in Europe has not been defined although Ogden et al. (20) identified GE closely related to the HGE agent in Apodemus sylvaticus wood mice and Clethrionomys glareolus bank voles in the United Kingdom, and Liz et al. (15) found the highest prevalence of GE in C. glareolus in Switzerland. Resolution of the mechanisms that are driving the evolution of the ank gene will require additional studies addressing the expression of the gene and characterization of the functional properties of the encoded protein, as well as a better understanding of the vectors and reservoirs of the HGE agent in the United States and Europe. Whereas the present study focused on the examination of the ank gene amplified from confirmed human cases and suggests that this gene provides an excellent epidemiologic tool for differentiating HGE agent isolates, the diversity noted within the gene should prove useful for examining the heterogeneity of GE in veterinary and arthropod populations.