Identification of 19 Polymorphic Major Outer Membrane Protein Genes and Their Immunogenic Peptides in Ehrlichia ewingii for Use in a Serodiagnostic Assay

Abstract

Ehrlichia ewingii, a tick-transmitted rickettsia previously known only as a canine pathogen, was recently recognized as a human pathogen. E. ewingii has yet to be cultivated, and there is no serologic test available to diagnose E. ewingii infection. Previously, a fragment (505 bp) of a single E. ewingii gene homologous to 1 of 22 genes encoding Ehrlichia chaffeensis immunodominant major outer membrane proteins 1 (OMP-1s)/P28s was identified. The purposes of the present study were to (i) determine the E. ewingii omp-1 gene family, (ii) determine each OMP-1-specific peptide, and (iii) analyze all OMP-1 synthesized peptides for antigenicity. Using nested touchdown PCR and a primer walking strategy, we found 19 omp-1 paralogs in E. ewingii. These genes are arranged in tandem downstream of tr1 and upstream of secA in a 24-kb genomic region. Predicted molecular masses of the 19 mature E. ewingii OMP-1s range from 25.1 to 31.3 kDa, with isoelectric points of 5.03 to 9.80. Based on comparative sequence analyses among OMP-1s from E. ewingii and three other Ehrlichia spp., each E. ewingii OMP-1 oligopeptide that was predicted to be antigenic, bacterial surface exposed, unique in comparison to the other E. ewingii OMP-1s, and distinct from those of other Ehrlichia spp. was synthesized for use in an enzyme-linked immunosorbent assay. Plasmas from experimentally E. ewingii-infected dogs reacted significantly with most of the OMP-1-specific peptides, indicating that multiple OMP-1s were expressed and immunogenic in infected dogs. The results support the utility of the tailored OMP-1 peptides as E. ewingii serologic test antigens.

Ehrlichia ewingii, a granulocytotropic Ehrlichia species (1), was recognized in 1999 as a human pathogen belonging to the family Anaplasmataceae in the United States, following Ehrlichia chaffeensis and Anaplasma phagocytophilum (6). Between 1996 and 2001, approximately 10 confirmed cases of human granulocytic ehrlichiosis caused by E. ewingii infection were identified in Missouri and Oklahoma (17). The numbers of E. ewingii cases being significantly lower than the numbers of A. phagocytophilum and E. chaffeensis cases (9) may be due partially to difficulty in diagnosing E. ewingii infection. Clinical signs of patients infected with E. ewingii, such as fever, headache, myalgia, leukopenia, and thrombocytopenia, are similar to those for human monocytic ehrlichiosis caused by E. chaffeensis and human granulocytic anaplasmosis caused by A. phagocytophilum (4, 6, 31). Hence, clinical features alone cannot distinguish these causative agents.

A further complication in the diagnosis of ehrlichiosis infections is that E. ewingii and E. chaffeensis also share the same vector tick species and animal reservoirs. Experimentally, the Lone Star tick (Amblyomma americanum) is a competent vector for E. ewingii and E. chaffeensis infection (2, 10). The white-tailed deer (Odocoileus virginianus) is suspected to be and considered an important reservoir for E. ewingii and E. chaffeensis, respectively, and the dog is another possible reservoir of both Ehrlichia spp. (8, 10, 21, 48). Consequently, E. ewingii and E. chaffeensis have similar seasonal and geographic distributions. In contrast, A. phagocytophilum uses Ixodes species of ticks as vectors and wild rodents as primary reservoirs, thus having a different seasonal and geographic distribution from E. ewingii. However, in addition to human travel, there is enough crossover in both vector and reservoir distributions to cause confusion in attempting to properly distinguish E. ewingii from the other agents of ehrlichioses, warranting development of better laboratory testing.

To distinguish infections, direct tests such as culture isolation, PCR, and microscopic observation of morulae (microcolonies of ehrlichiae) are useful if blood specimens are available at acute stages of infection. However, unlike E. chaffeensis and A. phagocytophilum, which can be culture isolated in myelocytic leukemia and tick cell lines, E. ewingii is currently not cultivable. PCR tests based on the E. ewingii-specific partial sequence of a 16S rRNA gene and a partial p28-19 sequence have been reported (8, 16, 17, 21, 26, 48), yet the sensitivities and specificities of E. ewingii PCR tests of clinical specimens are unknown, as there are no other definitive tests with which to compare. The microscopic observation of morulae in Romanovsky dye-stained peripheral blood granulocytes provides definitive proof of ehrlichial infection. Unfortunately, this test cannot be used as a single diagnostic test for E. ewingii infection because it cannot distinguish E. ewingii morulae from other granulocytic agents, such as A. phagocytophilum. Furthermore, negative results from Romanovsky dye staining cannot rule out E. ewingii infection, owing to high false-negative rates caused by sample conditions and the low sensitivity of the assay. These setbacks in prior diagnostic testing necessitate an additional test to properly identify E. ewingii infection.

Given the shortcomings in direct testing, indirect testing may be the answer for identification of ehrlichioses. Since ehrlichial infections induce significant antibody titers in nonimmunocompromised patients and since nonexposed people seldom have antibodies reactive to Ehrlichia spp., serologic tests are considered the most reliable tests for confirmation of ehrlichioses, especially when ruling out the possibility of infection. Although an indirect fluorescent-antibody (IFA) test using cultivated bacteria is widely used for E. chaffeensis and A. phagocytophilum serodiagnosis (9), this assay is not applicable for the diagnosis of E. ewingii infection due to the lack of cultured bacteria. As an alternative to whole bacteria, several immunodominant proteins, including major outer membrane proteins of Ehrlichia and Anaplasma spp., have been cloned and expressed as immunoreactive fusion proteins (5, 18, 22-24, 29, 30, 46, 50, 53). We previously reported that dot immunoblots or enzyme-linked immunosorbent assays (ELISAs) of dog and human sera with the recombinant 30-kDa major outer membrane protein (rP30) of Ehrlichia canis and the recombinant 44-kDa major outer membrane protein (rP44) of A. phagocytophilum, respectively, provide diagnostic sensitivities and specificities comparable to those of their whole bacteria IFA test counterparts (29, 40, 53). In the present study, we therefore proposed to create a similar test for E. ewingii.

In order to develop a serologic test using major antigens of E. ewingii, genes encoding these proteins must first be identified. Several members of the genus Ehrlichia have immunodominant surface proteins, e.g., outer membrane protein 1 (OMP-1)/P28s in E. chaffeensis, P30s in E. canis, and Map 1 in Ehrlichia ruminantium, which have been used as serodiagnostic antigens (12, 29, 32, 41, 43, 51). These surface proteins are encoded by a polymorphic multigene family and are immunologically highly cross-reactive to each other (25, 28-30, 33, 45, 49). A partial sequence (505 bp) for a single gene of the major OMP-1 gene family called p28-19 (ortholog of E. chaffeensis p28 [30]) is known for E. ewingii (17). In the present study, we systematically identified the entire E. ewingii OMP-1 genomic locus, which contains multiple omp-1 paralogs.

While there is no small laboratory animal model for E. ewingii infection, dogs can be used as an infection model. In fact, E. ewingii was originally discovered as a granulocytic variant of Ehrlichia canis that typically infects canine monocytes (11). Canine infection with E. ewingii has been detected more frequently and in broader geographic regions than human infection (8, 16, 21, 26, 27). In the present study, therefore, plasmas from experimentally E. ewingii-, E. chaffeensis-, and E. canis-infected dogs were used to test the potential utility of the E. ewingii OMP-1 peptides as serodiagnostic antigens by ELISA.

MATERIALS AND METHODS

E. ewingii omp-1 cluster amplification, sequencing, and assembly.

An EDTA-treated whole-blood specimen (∼200 μl) collected in April 2005 from an 8-week-old male German Shepherd mixed-breed dog in Ohio was used for DNA extraction. DNA was extracted using a QIAamp blood kit (Qiagen, Valencia, CA) and used as the template for the entire amplification and sequencing process. E. ewingii infection of the dog was confirmed by PCR and sequencing of the 16S rRNA of E. ewingii as well as by observation of bacterial inclusions (morulae) in granulocytes in the blood and joint fluid smear. PCR analysis showed that the dog was negative for infection by A. phagocytophilum, E. chaffeensis, and E. canis (Q. Xiong, W. Bao, and Y. Rikihisa, unpublished data).

The omp-1 fragments were amplified first by touchdown PCR (37) with the primer pairs F1 and R7, F8 and R14, and F15 and R21 (see Table S1 in the supplemental material). The PCR mix (50 μl) included 0.5 μl template DNA corresponding to 4 μl of the original blood sample, 10 pmol of each primer, 0.2 mM deoxynucleoside triphosphate mixture, 2.5 U high-fidelity Taq polymerase (Invitrogen, Carlsbad, CA), and 1.5 mM MgCl₂. Amplification was performed with the following program: 94°C for 3 min; a gradient over 10 cycles of 94°C for 0.5 min, 64°C for 0.5 min, and 72°C for 2 min, with the annealing temperature decreased by 1°C/cycle; 35 cycles of 94°C for 0.5 min, 55°C for 0.5 min, and 68°C for 9 min; and, finally, 68°C for 9 min. Nested PCRs were performed using the first PCR products as templates, with 21 pairs of degenerate primers, creating amplicons of approximately 1,500 bp that each overlapped approximately 200 bp according to the E. chaffeensis and E. canis omp-1 clusters (Fig. 1; see Table S1 in the supplemental material). The conditions of the nested PCR were similar to those of the first PCR, except that Taq polymerase was used and the elongation step was done at 72°C for 2 min. The nested PCR products were run in a 1% agarose gel with TAE buffer (40 mM Tris-acetate, 1 mM EDTA, pH 8.0). The amplified DNA fragments were recovered from the gel with a QIAEX II gel extraction kit (Qiagen) and directly sequenced with the nested PCR primers (Fig. 1). For fragment 3, two touchdown PCRs with high-fidelity Taq polymerase were performed, using infected dog blood DNA as the template and one of the following primer pairs: forward primer P28-19F and primer R21 or a forward primer designed based on the 3′ end of fragment 2 (Specific 4F) and reverse primer P28-19R (see Table S1 in the supplemental material). Nested amplification of these two PCR products and direct sequencing were used for subsequent design of new specific primers. Direct sequences obtained ranged from 250 to 800 bp. The poly(G/C) or poly(A/T) region (Fig. 1) was cloned using a TA cloning kit (Invitrogen), and the plasmid insert was sequenced. All sequencing data were assembled using the SeqMan program of DNASTAR software (DNASTAR Inc., Madison, WI).

FIG. 1.

Strategy for E. ewingii omp-1 cluster sequencing. E. chaffeensis omp-1 and E. canis p30 were aligned to design 21 pairs of degenerate primers. (A) The OMP-1 multigene locus was divided into three fragments, each composed of seven shorter fragments. (B) The initial nested touchdown PCRs generated four specific sequences within fragments 1 and 2. (C) Two fragments were amplified by nested touchdown PCR within fragment 3, using the p28-19 sequence and degenerate primers. Specific primers were designed to close all gaps. Two poly(A/T) and poly(G/C) regions were cloned into a TA vector and sequenced. The final sequence (24,126 bp) was assembled with the SeqMan program in the DNASTAR software.

E. ewingii omp-1 cluster analysis.

Artemis (38) was used to identify the open reading frames (ORFs) in the newly obtained E. ewingii omp-1 cluster. The ORFs encoding more than 100 amino acids were searched against the NCBI GenBank database to find homologs. The Artemis comparison tool (7) was used to analyze the synteny of the E. ewingii omp-1 cluster to those of E. chaffeensis, E. canis, and E. ruminantium. To search for repeat regions in the E. ewingii omp-1 cluster and between the E. ewingii omp-1 cluster and that of E. chaffeensis, E. canis, or E. ruminantium, dot plot analysis was performed with the Java Dot Plot Alignments program (JDotter; http://athena.bioc.uvic.ca/index.php).

Phylogenetic analysis.

The deduced amino acid sequences of E. chaffeensis OMP-1s/P28s, E. canis P30s, E. ruminantium MAP1s, and E. ewingii OMP-1s were aligned using the MegAlign program of DNASTAR software. Phylogenetic analysis was then performed with PHYLIP software (version 3.66) (13). The phylogram was constructed using the neighbor-joining method with Kimura's formula, and 1,000 bootstrap replications were conducted to evaluate the reliability of the tree (13).

Peptide synthesis and peptide-pin ELISA analysis.

Peptide libraries were synthesized using noncleavable multipin synthesis technology and 9-fluorenylmethoxy carbonyl chemistry (Mimotopes Pty. Ltd., Victoria, Australia) (15). After stripping off any previously bound proteins from the peptide pins with 0.1 M sodium phosphate buffer containing 1% sodium dodecyl sulfate (pH 7.2) and 0.1% β-mercaptoethanol at 60°C for 30 min, followed by two water washes for 10 min each, nonspecific binding sites were blocked with 200 μl of 3% skim milk (Becton Dickinson and Co., Sparks, MD) in phosphate-buffered saline (PBS)-Tween 20. Blocking was carried out in 96-well plates for 1 h at room temperature. Sets of peptide-bound pins were washed once with PBS containing 0.1% (vol/vol) Tween 20 for 10 min and then incubated in the blocking solution (1:100 dilution) with plasmas from E. ewingii-, E. chaffeensis-, or E. canis-infected dogs or with preinfection dog plasma at 4°C overnight. Samples from dogs 2119, 2185, and 2405 were collected on days 206, 109, and 123, respectively, after infection with E. ewingii. Samples from dog CTUALJ (E. chaffeensis IFA titer, 1:2,560), dog 1425 (E. chaffeensis IFA titer, 1:320), and dog 3918815 (E. chaffeensis IFA titer, 1:2,560) were collected on days 41, 121, and approximately 210, respectively, after infection with E. chaffeensis. Samples from dogs 246, 433, and 688 were collected 21 days after infection with E. canis, when the IFA titers were >1:2,560. Dog 2405 serum from the indicated time points (0, 21, 28, and 123 days after infection with E. ewingii) was selected for time course assays. Preinfection plasmas from dogs 2119, 2185, 1425, 246, 433, and CTUALJ were used as negative controls. After being washed four times as described above, the peptide pins were placed in wells filled with horseradish peroxidase-labeled goat anti-dog immunoglobulin G (heavy plus light chains) (Kirkegaard & Perry Laboratories, Gaithersburg, MD) diluted at 1:1,000 in PBS-Tween 20 and incubated for 1 h at room temperature. Samples were washed four times, and then the peptide pins were incubated for 20 min at room temperature with the horseradish peroxidase substrate 2,2′-azido-di-(3-ethyl)-benzthiazoline-6-sulfonic acid (Sigma, St. Louis, MO) in 70 mM citrate buffer (pH 4.2) applied to a new plate. Absorbance values at 405 and 492 nm were measured in an ELISA plate reader (Molecular Devices, Sunnyvale, CA). Each assay was repeated at least three times. The cutoff value (optical density at 405 nm [OD₄₀₅] − OD₄₉₂) for positive reaction was set as the mean OD₄₀₅ − OD₄₉₂ + 3 standard deviations of the negative control plasma.

Nucleotide sequence accession number.

The final sequence assembled from the entire E. ewingii omp-1 locus was deposited in GenBank under accession no. EF116932.

RESULTS

E. ewingii omp-1 cluster sequencing and assembly.

Forty-two degenerate primers were initially designed based on the conserved regions of the aligned omp-1/p30 clusters of E. chaffeensis and E. canis (Fig. 1; see Table S1 in the supplemental material). To efficiently utilize the limited amount of E. ewingii DNA, the putative omp-1 cluster was divided into three overlapping fragments of approximately 9 kb, estimated based on homologous regions of E. chaffeensis and E. canis. The first touchdown PCR (37) was designed to amplify the three putative long fragments. The PCR products were then used as templates for 21 nested touchdown PCRs using degenerate primer pairs (Fig. 1; see Table S1 in the supplemental material). As a result, only four PCRs resulted in bands ranging from ∼200 to ∼1,500 bp, namely, those with the F1 and R1 (∼700 bp), F4 and R4 (∼1,000 bp), F7 and R7 (∼1,500 bp), and F11 and R11 (∼220 bp) primers. The PCR products were directly sequenced. The results showed that they belonged to the omp-1/p28/p30 family. For regions covered by fragments 1 and 2 (Fig. 1), E. ewingii-specific omp-1 primers were designed based on the four newly obtained E. ewingii omp-1 DNA sequences (see Table S1 in the supplemental material). However, because no omp-1 sequence in fragment 3 was amplified using degenerate primers, two touchdown PCRs with high-fidelity Taq polymerase were performed using infected dog blood DNA as the template and one of the following primer pairs: forward primer P28-19F, designed based on the conserved region of E. ewingii p28-19 DNA sequences (17), and primer R21 or a forward primer designed based on the 3′ end of fragment 2 and reverse primer P28-19R, designed based on the conserved region of the p28-19 DNA sequence. Nested amplification of these two PCR products and direct sequencing were used for subsequent design of new specific primers. This process was repeated for three fragments until we encountered the poly(G/C) or poly(A/T) region in fragments 2 and 3. The poly(A/T) and poly(C/G) tracts (Fig. 1) were determined by TA cloning and sequencing of 10 and 22 plasmid inserts, respectively. The poly(G) tract had 9 to 13 G's (the number of G's was distributed among the 22 sequenced clones as follows: 9 G's, 1 clone; 10 G's, 4 clones; 11 G's, 7 clones; 12 G's, 2 clones; and 13 G's, 8 clones) and was reported as having 13 G's according to SeqMan software. The poly(A) tract had 10 to 13 A's (the number of A's was distributed among the 10 sequenced clones as follows: 10 A's, 1 clone; 11 A's, 3 clones; 12 A's, 3 clones; and 13 A's, 3 clones) and was reported as having 12 A's according to SeqMan software. The predominant in-frame sequences in each region were deposited in GenBank. The final sequence assembled from the entire E. ewingii omp-1 locus contained 24,126 bp. The G+C content of the E. ewingii omp-1 cluster was 28.74%, which is similar to those of the E. canis, E. chaffeensis, and E. ruminantium clusters (29.36%, 30.95%, and 27.19%, respectively). Sequence identities of the entire E. ewingii omp-1 cluster relative to E. canis, E. chaffeensis, and E. ruminantium were 28.4%, 22.2%, and 14.8%, respectively.

Features of OMP cluster structure are conserved among Ehrlichia species.

The Artemis software analysis showed that each of the 24 ORFs encodes more than 100 amino acids from the assembled E. ewingii omp-1 DNA fragment. One of the 24 ORFs in the middle of the cluster was short (390 bp), partially overlapped with two other ORFs in the opposite orientation, and had no homolog in the GenBank database, and thus this ORF was not included in the figures or in Table 1. The 23 ORFs were numbered ORF1 to -23. These 23 genes were arranged in tandem except for three ORFs (ORF19, -20, and -21) that were in the opposite orientation. Nineteen of the 23 ORF-encoded proteins were homologous to OMP-1/P28/MAP1 of E. chaffeensis, E. canis, or E. ruminantium. The most closely related proteins to each E. ewingii OMP-1 (EeOMP-1) are listed in Table S2 in the supplemental material.

TABLE 1.

Properties of E. ewingii OMP-1 cluster^c

ORF	OMP-1 no.	Length of upstream intergenic space (bp)	Length (location [bp])	No. of aa	Predicted no. of signal peptide aa	Molecular massa (Da)	Isoelectric pointa
ORF-1 (hypothetical transcriptional regulator)	NA	NA	357 (3-359)	118	NA	13,687.9b	7.93b
ORF-2	1	244	882 (603-1484)	293	23	29,848.6	6.38
ORF-3	2	568	819 (2053-2871)	272	21	28,125.9	8.61
ORF-4 (UN2)	NA	101	558 (2973-3530)	185	32	18,000.7	9.80
ORF-5	3	10	855 (3541-4395)	284	23	29,372.6	7.07
ORF-6	4	6	882 (4405-5286)	293	27	30,194.6	9.47
ORF-7	5	24	819 (5311-6129)	272	23	28,174.6	5.39
ORF-8	6	11	900 (6141-7040)	299	25	30,194.6	9.47
ORF-9	7	26	855 (7067-7921)	284	26	28,864.7	5.43
ORF-10	8	26	732 (7948-8679)	243	28	25,098.4	7.88
ORF-11	9	11	846 (8691-9536)	281	27	28,661.4	7.24
ORF-12	10	8	843 (9545-10387)	280	25	28,476.4	6.25
ORF-13	11	8	867 (10396-11268)	290	27	29,488.9	5.90
ORF-14	12	23	897 (11292-12188)	298	26	30,036.8	5.89
ORF-15	13	16	909 (12205-13113)	302	29	31,275.1	5.31
ORF-16	14	1,343	858 (14455-15312)	285	25	28,864.7	5.03
ORF-17	15	301	837 (15614-16450)	278	25	28,862.8	5.58
ORF-18	16	335	849 (16786-17634)	282	25	28,476.0	5.41
ORF-19	17	769	819 (18404-19222)	272	25	27,782.3	6.50
ORF-20	18	539	849 (19462-20310)	282	32	28,603.4	5.08
ORF-21	19	808	849 (21119-21967)	282	23	30,505.0	7.88
ORF-22 (UN3)	NA	1,082	408 (23050-23457)	135	NA	14,800.0	3.85
ORF-23 (SecA)	NA	390	285 (23848-24126)	93	NA	10,533.3b	9.40b

^aAfter removing signal peptide.

^bNo signal peptide.

^cNA, not available; aa, amino acids.

We numbered the proteins EeOMP-1-1 to EeOMP-1-19 (Fig. 2). The sequence similarity and molecular mass of EeOMP-1-8 were lower than those of the other EeOMP-1s. There is a protein ortholog of EeOMP-1-8, UN3, with unknown function, in the E. chaffeensis and E. canis genomes. In E. ruminantium, the EeOMP-1-8 ortholog is MAP1-9. The protein encoded by the first ORF (ORF1) is homologous to a hypothetical transcriptional regulator, and the protein encoded by the last ORF (ORF23) is homologous to SecA. Proteins encoded by the other two ORFs (ORF4 and ORF22) are most homologous to two E. chaffeensis and E. canis peptides, UN2 and UN4, with unknown function, as well as two E. ruminantium peptides, UN1 and UN2, whose functions are unknown. The p28-19 505-bp sequence was a part of Eeomp-1-16 (Table 1). Intergenic spaces between omp-1 genes ranged from 6 to 1,343 bp (Table 1). In the 5′ half of each OMP cluster, 14 genes (un2 to Eeomp-1-13 in E. ewingii) were linked by short intergenic spaces ranging from 6 to 26 bp (Table 1). Eight genes in the 3′ half (Eeomp-1-14 to Eeomp-1-19) were connected by longer intergenic spaces, ranging from 301 to 808 bp. Thus, features of the OMP cluster structure were conserved among E. ewingii, E. canis, E. chaffeensis, and E. ruminantium, with the exception of the opposite orientation of three genes at the 3′ end in E. ewingii instead of one gene (E. canis and E. ruminantium) or two genes (E. chaffeensis).

FIG. 2.

Schematic representation of the organization of the E. ewingii omp-1 gene cluster. Genes are represented as boxes, with arrows indicating their orientation. omp-1s are shown with a horizontal shading pattern. Black, white, and gray boxes show tr1, unknown genes, and secA, respectively.

After removal of the signal peptide sequence, predicted molecular masses of mature E. ewingii OMP-1s ranged from 25.1 to 31.3 kDa. The predicted signal peptides ranged from 21 to 32 amino acids. The predicted isoelectric points of the mature OMP-1s were 5.03 to 9.80. Properties of the ORFs of the E. ewingii omp-1 cluster, including predicted signal peptide lengths, molecular masses of mature proteins, and isoelectric points, are shown in Table 1.

omp-1/p28/map1 gene clusters display synteny at the 5′ end.

The synteny among entire OMP-1 gene clusters of E. ewingii and three related Ehrlichia species was analyzed by the Artemis comparison tool, and the results are shown in Fig. 3. The genes at the 5′ ends of the omp-1 clusters were more highly conserved than were genes in the central region or the 3′ end (Fig. 3). Previously, we defined three repeat sequence regions, α, β, and γ, in omp-1 clusters of E. chaffeensis and E. canis (28). The dot plot analysis of the E. ewingii omp-1 cluster and the dot plot between E. ewingii and E. ruminantium revealed only β and γ repeat regions, as the α regions of these bacteria are too short (Fig. 4). The β repeat region in E. ruminantium was shorter than those of E. chaffeensis and E. canis. The dot plot analyses between E. ewingii and E. chaffeensis and between E. ewingii and E. canis showed three clear repeat regions, indicating that the α region is expanded in E. canis and E. chaffeensis.

FIG. 3.

Synteny analysis of the E. ewingii (Ee) omp-1 cluster relative to those of E. chaffeensis (Ech), E. canis (Eca), and E. ruminantium (Eru), using the Artemis comparison tool.

FIG. 4.

Dot plot analysis of the E. ewingii omp-1 cluster (A) and the E. ewingii omp-1 cluster relative to those of E. chaffeensis (B), E. canis (C), and E. ruminantium (D). Repetitive regions consisting of multiple homologous DNA segments were analyzed using the Web-based dot plot program JDotter (http://athena.bioc.uvic.ca/index.php). The window cutoff was set to the default. The α, β, and γ repetitive regions described by Ohashi et al. (28) are marked by lines.

Phylogenetic analysis of all 79 OMPs of E. ewingii, E. chaffeensis, E. canis, and E. ruminantium is shown in Fig. 5. The previously defined α and β1 regions in the E. chaffeensis omp-1 cluster (28) encoded five (P28, OMP-1F, -1D, -1C, and -1E) and four (OMP-1H, -1A, -1S, and -1Z) proteins, respectively, and the α and β1 regions in the E. canis p30 cluster encoded six (P30, P30-1, P30-2, P30-3, P30-4, and P30a) and four (P30-6, P30-5, P30-7, and P30-8) proteins, respectively. However, in E. ewingii, the α and β1 regions each encoded two proteins (EeOMP-1-15/EeOMP-1-16 and EeOMP-1-12/EeOMP-1-13, respectively). In E. ruminantium, the α region encoded only one protein (MAP1) and the β1 region encoded two proteins (MAP1-2 and MAP1-3) (Fig. 5).

FIG. 5.

Phylogram of OMPs of E. ewingii, E. chaffeensis, E. canis, and E. ruminantium. A total of 39 OMPs were segregated into 10 clusters with four or three Ehrlichia species each, but 40 remaining proteins were not. The tree was constructed using the neighbor-joining (NEIGHBOR program from PHYLIP) method, based on an alignment generated with CLUSTAL V; 1,000 bootstrap replications were performed. The nodes supported by bootstrap values of >60% are labeled. The OMPs encoded by the three repetitive regions in Fig. 4 are indicated by α, β1, and β2. Branch lengths are proportional to percent divergence. The calibration bar is on the lower left.

EeOMP-1-8 and MAP1-5 were far removed from the remaining OMP-1s, raising the possibility that they do not belong to the OMP cluster (Fig. 5). EeOMP-1-18 and EeOMP-1-19, which are encoded by genes in the reverse orientation, were clustered with P28-2, which is encoded by one of two reverse-oriented E. chaffeensis omp-1 genes. All proteins except α and β group proteins formed separate small clusters, including four proteins from each of the four Ehrlichia species. Each cluster of proteins is thus expected to share a common ancestor.

Previously reported 505-bp E. ewingii p28-1 sequences (GenBank accession numbers AF287961, AF287962, AF287963, AF287964, and AF287966) (17) were compared with corresponding sequences identified in the present study. The 505-bp region begins at bp 16918 and ends at bp 17422 in the cluster, which corresponds to 75% of omp-1-16 (i.e., from bp 133 to 637 of the 849-bp omp-1-16 gene). The E. ewingii p28-1 sequences of a Missouri canine sample and an Oklahoma human sample (17) were identical to the sequence obtained from the Ohio dog analyzed in the present study.

E. ewingii OMP-1-specific peptide ELISA.

Following our OMP-1 amino acid sequence alignment, repetitive sequence analysis, and phylogenetic analysis results, we tried to design an E. ewingii OMP-1-specific serologic test. Since OMP-1s share repetitive common or homologous amino acid sequences with OMP-1s of the same or different Ehrlichia spp., it is difficult to design recombinant proteins (>10 kDa) that provide Ehrlichia sp.-specific or gene-specific antigens. Also, it is cost- and labor-prohibitive to clone, express, and purify 19 recombinant OMP-1 proteins. Therefore, we designed 12- to 17-mer peptides specific to each of the 19 E. ewingii OMP-1s. For this purpose, extracellular loops of the 19 E. ewingii OMP-1s were first predicted using the posterior decoding method of PRED-TMBB (http://bioinformatics.biol.uoa.gr/PRED-TMBB) (3). PRED-TMBB is a Web server capable of predicting transmembrane strands and the topology of β-barrels in OMPs of gram-negative bacteria based on a hidden Markov model. The validity of these predictions is tested using nonhomologous OMPs with structures known at atomic resolution according to conditional maximum likelihood criteria (3). Relatively highly antigenic and hydrophilic 12- to 17-mer peptide fragments located within one of the extracellular loops were chosen from each of the 19 EeOMP-1 amino acid sequences, based on DNASTAR Protean analysis. Using the program BLAST, these peptide sequences were compared with the entire E. ewingii omp-1 locus and the E. chaffeensis, E. canis, and E. ruminantium genome sequences to synthesize one peptide specific to each of the 19 EeOMP-1s (Table 2).

TABLE 2.

E. ewingii OMP-1 peptide sequences used in ELISA

OMP-1 no.	Amino acid sequence	Peptide length (aaa [aa position in the protein sequence])
OMP-1-1	SCTEQEMKPAQQNGSSK	17 (151-167)
OMP-1-2	EQHFALASELDTNGNQ	16 (130-145)
OMP-1-3	VSAPSGYDDNIYAYFSI	17 (123-139)
OMP-1-4	FAIPRDTFFNNSIPY	15 (144-158)
OMP-1-5	NSSSLISSNNHYTQLY	16 (129-144)
OMP-1-6	KDNNQVQPKAHDSSSTD	17 (148-164)
OMP-1-7	KDPKDCSVKDAFRHL	15 (130-144)
OMP-1-8	TEDKYLTSEQEVNDY	15 (120-134)
OMP-1-9	ICKENDKPTPKEKKY	15 (145-159)
OMP-1-10	YRYFAIAREMNSSSNNQ	17 (137-153)
OMP-1-11	KNSGHSSIDAHR	12 (136-147)
OMP-1-12	IESDQNKFQPKNANSNS	17 (154-170)
OMP-1-13	SESSKEPQPKNPNSAGN	17 (159-175)
OMP-1-14	KYYGLFREGTPQEEEH	16 (146-161)
OMP-1-15	SRKDNANIGTTPQDKK	16 (141-156)
OMP-1-16	KIEDNQVQNKFTISNY	16 (76-91)
OMP-1-17	QFYREGSNNYKF	12 (123-134)
OMP-1-18	VQDTKSHIVDDNYR	14 (121-134)
OMP-1-19	SKQDNLNSDYVTLIN	15 (171-185)

^aaa, amino acids.

Plasmas from three dogs experimentally infected with E. ewingii and preinfection plasmas from four dogs were then tested in ELISAs containing the 19 EeOMP-1-specific peptides. Thirteen peptides (EeOMP-1-1, EeOMP-1-2, EeOMP-1-3, EeOMP-1-4, EeOMP-1-5, EeOMP-1-8, EeOMP-1-9, EeOMP-1-10, EeOMP-1-13, EeOMP-1-14, EeOMP-1-15, EeOMP-1-16, and EeOMP-1-19) were consistently recognized with plasmas from three dogs experimentally infected with E. ewingii compared with preinfection dog plasmas (Fig. 6).

FIG. 6.

ELISA analysis of E. ewingii- and E. chaffeensis-infected dogs with the 19 EeOMP-1 oligopeptides. Preinfection control and postinfection plasmas from dogs were allowed to react with the 19 synthesized EeOMP-1-specific peptides. The y axis shows the OD₄₀₅ - OD₄₉₂ values. A reaction was considered positive when the plasmas from infected dogs yielded an OD₄₀₅ - OD₄₉₂ value greater than the mean OD₄₀₅ - OD₄₉₂ value for preinfection control plasma plus 3 standard deviations (dashed line with closed triangles). Representative data from three to five assays are shown. Reactivity ratios of E. ewingii/control plasma values (EW/Cont.) and E. chaffeensis/control plasma values (ECH/Cont.) were calculated based on the averages for three E. ewingii-positive and three E. chaffeensis-positive samples, respectively, and four negative control samples. EeOMP-1 peptides that showed good sensitivity and specificity for detecting E. ewingii infection are underlined.

By peptide-pin ELISA, eight peptides (EeOMP-1-1, EeOMP-1-3, EeOMP-1-4, EeOMP-1-5, EeOMP-1-13, EeOMP-1-14, EeOMP-1-16, and EeOMP-1-19) with high sensitivities (indicated by the ratio of E. ewingii plasma reactivity to control plasma reactivity) did not cross-react with plasmas from E. chaffeensis-infected dogs (indicated by the ratio of E. chaffeensis plasma reactivity to control plasma reactivity [∼1.00]) (Fig. 6). The same ELISA was performed using three E. canis-infected dog plasmas. Based on the ratios of E. ewingii plasma reactivity to control plasma reactivity and E. canis plasma reactivity to control plasma reactivity, EeOMP-1-9, EeOMP-1-11, EeOMP-1-12, EeOMP-1-13, EeOMP-1-14, and EeOMP-1-15 showed high sensitivities and good specificities (data not shown). The temporal reactivities of E. ewingii-infected dog sera to these peptides showed that EeOMP-1-13, EeOMP-1-14, EeOMP-1-15, and EeOMP-1-16 gave higher sensitivities than the remaining peptides and gave signals above that of control dog plasma when E. ewingii PCR became positive (at day 21 postexposure) (Fig. 7). While more specimens need to be tested and peptide antigens can be modified, several of these EeOMP-1 peptides may serve as good candidate antigens for E. ewingii serodiagnosis to distinguish E. ewingii infection from both E. chaffeensis and E. canis infections.

FIG. 7.

Temporal development of antibodies to EeOMP-1 peptides in an E. ewingii-infected dog. Plasmas collected on 0, 21, 28, and 123 days postinfection were assayed using selected EeOMP-1 peptides 13, 14, 15, and 16 by ELISA. Nested PCR using E. ewingii-specific primers was negative on day 0 and positive on days 21, 28, and 123 postinfection (47). The y axis shows OD₄₀₅ - OD₄₉₂ values.

DISCUSSION

In the present study, for the first time, the entire 24-kb E. ewingii omp-1 locus, containing 19 omp-1 genes, was sequenced. Since the only available source of E. ewingii DNA was a small amount of an infected dog blood specimen, we employed touchdown PCR. This method has been used previously to amplify a small amount of fragmented Aegyptianella pullorum DNA from archival paraffin sections on glass slides (36). Incorrect base calls resulting from amplification or sequencing errors were minimized in the present study because large pools of PCR products were directly sequenced. In addition, multiple overlapping regions throughout the sequences ensured the reliability of sequencing results. So far, only a few E. ewingii genes, including the 16S rRNA gene (1, 16, 48), groESL (39), p28-19 (17), dsbA (GenBank accession no. DQ902688), gltA (GenBank accession no. DQ365879), and the disulfide oxidoreductase gene (19), have been reported. Applying a similar approach to that used here, it would be possible to obtain DNA sequences of other genomic regions to further our understanding of this uncultivable emerging zoonotic pathogen.

Because E. ewingii infects granulocytes, the distinction between E. ewingii and strains of A. phagocytophilum was unclear prior to the molecular era. However, in concordance with the 16S rRNA and groESL sequence-based classification of this bacterium (1, 39), our finding of the complete OMP-1 cluster structure flanked with tr1 and secA clearly demonstrated that E. ewingii belongs to the genus Ehrlichia. Synteny analysis suggests that the OMP clusters existed in a common ancestor of the present-day four Ehrlichia species. Furthermore, the locus appears to have been partially scrambled as species evolved. The E. ewingii OMP-1 cluster has greater synteny with monocytotropic E. chaffeensis and E. canis than with endotheliotropic E. ruminantium. Whether OMP-1s and host cell type specificity coevolved remains to be studied.

The present study revealed 19 E. ewingii OMP-1 amino acid sequences and 19 E. ewingii immunogenic amino acid sequences. Studies of E. chaffeensis have shown an important role for OMP-1s/P28s in the stimulation of the host immune response and protection of the host from infection. Immunization with recombinant P28 (one of the major OMP-1/P28 family members) protected mice from E. chaffeensis challenge (30). A monoclonal antibody against OMP-1g (P28) mediated protection of SCID mice from fatal E. chaffeensis infection (20). While antibodies against a single OMP-1 protein confer partial protection, the existence of multiple homologous surface proteins has been suspected to allow evasion of the host immune response. A recent proteomic study showed that 18 of 21 E. chaffeensis OMP-1/P28 family proteins are indeed bacterial surface exposed, supporting the idea of immunoevasion (14). The number of E. ewingii omp-1 genes found in the OMP-1 cluster (19 copies) was similar to those for E. canis (22 copies, but there is an additional locus with three p30s) (28), E. chaffeensis (22 copies), and E. ruminantium (16 copies). In addition, there is extensive diversification among omp-1 genes of E. ewingii, similar to the case for other Ehrlichia spp., supporting the hypothesis that multiple omp-1/p28 paralogs present in Ehrlichia spp. are involved in immunoavoidance. Thus, these studies suggest that incorporation of immunogenic peptides of multiple OMP-1s in a vaccine preparation would provide better protection against Ehrlichia infection than the use of a single OMP-1 in a vaccine.

Multiple OMP-1/P28 and P30 mRNAs are expressed by E. chaffeensis and E. canis during experimental infections of dogs with these bacteria (42, 44). All 22 E. chaffeensis P28 recombinant antigens were recognized by sera from two dogs experimentally infected with E. chaffeensis (52). Similarly, the present results suggest that all 19 EeOMP-1 peptides were recognized by the plasmas from three E. ewingii-infected dogs. Thus, the lack of immunological cross-reactivity of E. canis and E. chaffeensis OMP-1/P28/P30 with plasmas from human patients or dogs infected with E. ewingii in previous studies (6, 34, 35, 43) was likely due to divergence of the amino acid sequences of the E. ewingii OMP-1s from those of the E. canis and E. chaffeensis OMP-1/P28/P30 proteins expressed in cell culture. It is also most likely for E. ewingii-infected humans and dogs that different combinations of multiple OMP-1s are expressed at different stages of infection and under different immune and health statuses of animals. Considering these observations, for serodiagnosis of E. ewingii infection in both humans and animals, the use of a combination of EeOMP-1s as the antigen is expected to provide more sensitive and specific serodiagnosis than the use of a single EeOMP-1 antigen. Furthermore, all of the EeOMP-1 amino acid sequences obtained in the present study would help in optimizing peptide antigens to provide desired specificity and sensitivity to detect potentially diverse E. ewingii strains in the field. The entire EeOMP-1 DNA sequence data also obtained in the present study should help in improving diagnostic PCRs for human and dog E. ewingii infections to make this direct test more reliable for all infective species. Lastly, the sequencing and comparative analysis of all of the E. ewingii OMP-1s in the present study should greatly advance E. ewingii epidemiologic and immunologic studies.