INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Ehrlichia spp. and related bacteria such as Cowdria and Anaplasma spp. are obligatory intracellular bacteria with a tropism for hematopoietic cells (27). These bacteria have major outer membrane proteins (OMPs) which are encoded by a multigene family that is estimated to occupy ca. 1 to 2% of the genome in some species (20, 21, 22, 26, 30, 33). In the organisms studied thus far these OMPs were shown to be immunoprotective in animals and to induce proinflammatory cytokines by leukocytes in vitro (13, 14, 17, 21, 23, 28). However, the gene locus, organization, expression, and function of these multigene families were not known well.

Canine and human monocytic ehrlichioses are tick-borne zoonoses caused by infection with Ehrlichia canis and E. chaffeensis, respectively. Canine monocytic ehrlichiosis (CME) was first described in 1935 (7) and now occurs worldwide, especially in tropical and subtropical regions (12). Human monocytic ehrlichiosis (HME) was first recognized in the United States in 1987, and it was assumed that E. canis caused this disease because of the positive reaction of the patient's serum to E. canis antigen (15). In 1990, an organism was isolated from a patient with HME (5) and classified as a new species, E. chaffeensis, since the 16S rRNA gene sequence was 1.8% divergent from that of E. canis (1). Since this first report, HME has been increasingly recognized in the United States. Serologic evidence suggests the disease also occurs in Europe, Africa, and Mexico (28). These two Ehrlichia species are transmitted primarily by different species of ticks: E. canis by the brown dog tick, Rhipicephalus sanguineus (10), and E. chaffeensis by the Lone Star tick, Amblyomma americanum (2). The natural reservoirs of these two organisms are also different: canidae are only known reservoirs of E. canis, whereas white-tailed deer (Odocoileus virginianus) is a natural reservoir of E. chaffeensis (8). E. canis and E. chaffeensis cause a febrile systemic illness that is often severe and even be fatal in dogs and humans, respectively. However, E. chaffeensis causes only a mild febrile response with no hematological abnormalities in experimentally infected dogs (6). Conversely, an E. canis-like agent (a new strain of E. canis) causes asymptomatic chronic infection in humans (25).

We originally identified orthologous and paralogous genes encoding immunodominant and immunocross-reactive 30- to 32-kDa major surface proteins in E. chaffeensis (seven genes) and E. canis (three genes) and showed that they belong to a polymorphic multigene family, termed omp-1 for E. chaffeensis and p30 for E. canis (20, 21). In our subsequent review (28) and several meeting presentations, at least 16 omp-1 or 22 p30 genes at a single locus and the tandem gene arrangement were shown. One of the E. chaffeensis genes, p28, was overexpressed in Escherichia coli (21). Immunization with the recombinant P28 protein protected mice from infection with E. chaffeensis, suggesting that P28 is a potential vaccinogen (21). In order to understand the role of the multigene family in the host-specific pathogenesis of the monocytic ehrlichiosis (ME) agents and in the interplay between ehrlichiae and their diverse host environments and in order to explore the most protective OMP protein against ehrlichial infection, it is essential to characterize the OMP multigene family and their gene organization and expression in these two species.

Reddy et al. (26) published five and two members of the multigene families in E. chaffeensis and E. canis, respectively (four genes in E. chaffeensis and one gene in E. canis are the same genes identified by us [20, 21]). McBride et al. (16) published one member of the E. canis multigene family linking four p30 genes previously identified by Ohashi et al. (20) and Reddy et al. (26). Recently, Yu et al. (31) using a rapid adapter primer PCR method, assembled 14 members of the omp family in E. chaffeensis and linked them to seven paralogs which had been previously described by us (21) and Reddy et al. (26). Yu et al. (31) described the locus of the gene cluster to be downstream of a gene (clpx) homologous to an ATP-dependent Clp protease ATP-binding subunit, but the region downstream from an omp gene at the 3' end of the gene cluster was not described. The relationship of the multigene family between CME and HME agents, however, remains unclear. Furthermore, the transcriptional pattern of these tandemly arranged multigenes is rather unclear. Reddy et al. (26) showed that only one gene at 3' end of four genes examined by reverse transcription-PCR (RT-PCR) was transcribed by E. chaffeensis in DH82 canine monocytic cell line, whereas McBride et al. (16) found by using RT-PCR that all five members of the multigene family examined were monocistronically transcribed by E. canis in DH82 cells. Yu et al. (31) found that, based on RT-PCR analyses, 6 of 10 tested genes of E. chaffeensis were transcribed in DH82 cells and that none of these genes was cotranscribed. Since DNA template control using the same sets of primers were not shown in any of these three studies, it is not known whether the primer pairs selected produce sufficient intensities of bands for a given amount of template DNA.

In the present study, we were especially careful to avoid sequencing errors due to repetitive elements present among homologous omp genes. We report here for the first time a 28-kb region including 22 paralogs of the multigene family in E. canis (termed the omp cluster) and a 6.9-kb region including 3 paralogs. We also describe a 27-kb omp cluster including 22 paralogs in E. chaffeensis, which was distinct from the previous report by Yu et al. (31) with regard to the gene number and genetic locus. We further examined the transcription and cotranscription of 28 genes including 22 omp paralogs of E. canis using 28 sets of gene-specific primer pairs and another set of primers specific to the adjacent genes flanking 27 intergenic spaces. The DNA template control was included for reactions using every primer pair.

(Part of this study was presented at the 99th and 100th General Meetings of the American Society for Microbiology [Chicago, Ill., 30 May to 1 June 1999, and Los Angeles, Calif., 22 to 24 May 2000].)

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Organisms, culture, and purification. E. canis Oklahoma and E. chaffeensis Arkansas were propagated in DH82 canine macrophage cell line at 37°C in Dulbecco modified Eagle's medium containing 10% heat-inactivated fetal bovine serum and purified by Percoll density gradient centrifugation (21) or Sephacryl S-1000 column chromatography (29).

Genomic Southern blot analysis. Genomic DNAs were extracted from purified ehrlichiae and digested with restriction enzymes. DNA probes (A to F in Fig. 1) was labeled with [-³²P]dATP by the random primer method with a kit (Amersham Pharmacia Biotech, Piscataway, N.J.). Southern hybridization procedure was as described elsewhere (21), and a hybridization temperature of 65°C (high-stringency condition) was used to avoid the binding of the probes to other homologous genes. The membrane was exposed to Hyperfilm (Amersham Pharmacia Biotech).

View larger version (33K): [in this window] [in a new window]   FIG. 1.   Schematic representation of gene organization of the p30 and omp-1 multigene families of ME agents. Genes are represented as open boxes, with arrows indicating their orientation. (A) E. canis omp cluster (28.2 kb) and another locus (6.9 kb). The LA-PCR amplicon or the recombinant clones are shown below in a diagram of gene organization. (B) E. chaffeensis omp cluster and its restriction map. Fragments hybridized by genomic Southern blot analysis with probes A to E in Fig. 2 and the cloned fragments are shown by lines with identification numbers under the map. Bars indicate the repetitive regions ,  (subregions 1 and 2), or , that were identified by the dot plot analysis in Fig. 3. E, EcoRI; H, HindIII; C, ClaI; P, PstI; S, SpeI; X, XbaI.

Cloning and sequencing of overlapping DNA fragments of ehrlichiae. The DNA fragments, which were detected by genomic Southern blot analysis, were inserted into pBluescript II KS(+) vector or pMW119 vector. The recombinant plasmids were introduced into E. coli DH5. The ehrlichial DNA fragments were isolated by the colony hybridization method with the labeled probes used in the genomic Southern blotting. For the amplification of large DNAs (~9.0 kb), we used a Long and Accurate PCR kit (LA-PCR; Takara Biomedical, Ohtsu, Japan) using primer pairs of 36 nucleotides (nt) with BamHI sites. The PCR condition were 35 cycles consisting of 20 s of denaturation at 98°C, 10 s of annealing at 56°C, and 15 min of extension at 68°C. The amplicons obtained were digested with BamHI, and they were cloned by using the same plasmid vectors as those described above. For sequencing, multiple small fragments (~2 kb) were generated and subcloned from these large fragments based on the restriction digestion analysis. The overlapping areas were further confirmed by PCR using primers flanking the junctions of the fragments. Sequencing was performed with universal or suitable synthetic primers by the dideoxy chain termination method.

Preparation of cDNA. Total RNA was prepared from 5 × 10⁶ E. canis-infected DH82 cells using the RNeasy Mini Kit according to the manufacturer's instruction (Qiagen, Valencia, Calif.). A 5-µg RNA sample was treated with 10 U of RNase-free DNase I (Epicentre, Madison, Wis.) in Tris-HCl buffer (pH 7.5) containing 10 mM MgCl₂ and 1 U of RNase inhibitor for 30 min at 37°C. The RNA was repurified by using the same kit to remove the DNase I. For cDNA synthesis, the isolated RNA (2.5 µg) was reverse transcribed in 20 µl of the reaction mixture using 200 ng of random hexamer primers and Superscript II (Life Technologies, Inc., Gaithersburg, Md.) at 42°C for 50 min.

DNA-PCR and RT-PCR. Twenty-eight primer pairs for the amplification of 28 genes were designed based on the gene-specific region and the target size. Twenty-nine primer pairs for the amplification of 27 sets of two adjacent genes connected by their intergenic spaces were designed based on other gene-specific regions and target size. One long space was examined with three pairs of primers which amplify three overlapping fragments covering the intergenic space and the flanking genes. The nucleotide positions and the sizes of each amplicon are shown in Table 1. A PCR reaction mixture (50 µl) contained 5 ng of genomic DNA from purified E. canis for DNA-PCR or 0.5 µl of the cDNA products for reverse transcription-PCR (RT-PCR), 10 pmol each of gene-specific primer, a 0.2 mM concentration of deoxynucleoside triphosphate mixture, 2.5 U of Taq polymerase, and 1.5 mM MgCl₂. The PCR conditions were 1 cycle of 3 min of denaturation at 94°C, followed by 28 cycles consisting of 1 min of denaturation at 94°C, 1 min of annealing at 54°C, and 1 min of extension at 72°C. RT-PCR without reverse transcriptase was carried out using respective primer pairs to rule out the contamination of DNA in the RNA preparation (negative control). The amplicons were electrophoresed and visualized by ethidium bromide staining.

Sequence analysis and GenBank accession numbers. Binary comparison and dot plot analysis of the entire DNA sequences of the gene cluster were performed by using the GAP program from the Genetics Computer Group (Madison, Wis.) and Omiga 2.0 (Oxford Molecular Group Inc., Hunt Valley, Md.). Amino acid sequence alignments were performed by using CLUSTAL V. The phylogenetic analysis was carried out using PHYLIP software version 3.5.7 (9). The phylogram was constructed using neighbor-joining method with the Kimura formula, and 1,000 bootstrap replications were conducted to evaluate the reliability of the tree. The DNA sequences reported here have been assigned the GenBank accession numbers U72291 for the E. chaffeensis 27-kb region, AF078553 for the E. canis 28-kb region, and AF324792 for the E. canis 6.9-kb region.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Assembly and confirmation of the omp gene clusters of E. canis and E. chaffeensis by cloning overlapping fragments. The cloning strategy is shown in Fig. 1. We previously cloned two HindIII DNA fragments of E. canis in pECA9.0 and pECA3.8 and three genes of the p30 multigene family (p30, p30-1, and p30a) were identified within these fragments by partial sequencing (Fig. 1A) (21). In the current study these two DNA fragments were completely sequenced and linked by a PCR using the genomic DNA as a template. Three additional overlapping DNAs of 9.0, 2.1, and 5.2 kb were obtained by LA-PCR with three primer pairs (Fig. 1A) designed based on DNA sequence of the E. chaffeensis omp cluster as described below. The BamHI digestion of the 9.0-kb amplicon generated two DNA fragments of 4.6 and 4.4 kb (Fig. 1A). All fragments except for the 5.2-kb DNA were cloned into the pBluescript vector. The 5.2-kb fragment could be cloned only when the pMW119 vector, a low-copy-number plasmid, was used. These recombinant plasmids were designated pLA-PCRB4.6, pLA-PCRB4.4, pLA-PCR2.1, and pLA-PCR5.2. An additional HindIII fragment of 6.9 kb hybridized to the p30a gene probe in genomic Southern blot analysis (20) was also cloned from E. canis genomic DNA by the colony hybridization method, and the clone was designated pECA6.9. As previously shown (20), since the band intensities of two loci hybridized with the p30a probe were almost identical, these two loci may be located in a single E. canis genome.

View larger version (79K): [in this window] [in a new window]   FIG. 2.   Genomic Southern blot analysis of E. chaffeensis omp cluster. The number under each lane represents the hybridized fragment shown in Fig. 1B. The numbers with a double or single underline show the fragments cloned in our previous studies (20, 21) and in the present study, respectively. The locations of probes A to F are shown on the restriction map in Fig. 1B.

Comparative analysis of the 28- and 6.9-kb omp loci of E. canis and the 27-kb omp locus of E. chaffeensis. The omp locus of 28,254 bp in E. canis had 28 protein-coding genes (Fig. 1A and Table 1), and the G+C content was 29.36% (AT-rich). By sequence identity analysis, 22 of the 28 genes were found to belong to the p30 multigene family. These 22 paralogs were homologous, but not identical, and were tandemly arranged except for one gene with an opposite orientation. One of six remaining genes had a high similarity with secA, which is known to be required for OMP transport (3), and another gene (u1 in Fig. 1A) was similar to hypothetical transcriptional regulators of Sinorhizobium meliloti ORF3 and R. prowazekii RP497. No gene homologous to the four remaining genes (u2 to u5 in Fig. 1A) was found by database search. Another locus of 6,913 bp had five protein-coding genes, and the G+C content was 29.46%. A 3.1-kb DNA consisting of a set of three identical genes (p30-10, p30-4, and p30a) and two intergenic spaces, identical to those in the 28-kb locus, was repeated within this locus (Fig. 1A). Two other genes in this locus were purK encoding a phosphoribosylaminoimidazole carboxylase and u6 of unknown function.

View larger version (19K): [in this window] [in a new window]   FIG. 3.   Dot plot analysis of the omp cluster in E. canis (A), E. chaffeensis (B), and between the two species (C). The repetitive regions consisting of multiple homologous DNA segments were analyzed using the dot plot program with Omiga 2.0 software. The window cutoff was set to 30 bp and at an 80% minimum percentile score. The bars show three repetitive regions (, , and ).

Properties and relationships of proteins encoded by the multigene families. A total of 44 distinct genes in the p30 and omp-1 multigene families encoded 264- to 316-amino-acid proteins with molecular masses of 30,624 to 35,241 Da (Table 1). The putative signal peptides consisting of 25- to 31-amino-acid residues, one of which was determined based on the N-terminal amino acid sequence analysis of the native and mature protein (21), were also found in all proteins at the N termini. The molecular masses of the predicted mature proteins were 27,304 to 32,940 Da in 44 proteins. The estimated isoelectric points of proteins predicted from open reading frames were 4.59 to 9.43, and the 5' half of each omp cluster had basic proteins, and the 3' half had acidic proteins (Table 2). The protein sequence identities of all paralogs of the omp genes in E. chaffeensis and E. canis were 19.1 to 82.7% and 19.3 to 71.8%, respectively. We defined orthologs based on gene locations and protein sequence identities due to nonsegregation of omp genes into two Ehrlichia spp. in  and 1 regions (Fig. 4). The identities of all orthologs of omp genes between the two species were in the range of 45.5 to 79.3% (Table 2). The omp cluster of E. canis appears to lack a protein closely related to P28-2 of E. chaffeensis, because the sequence identities of P28-2 with 22 E. canis proteins were uniformly low (21.5 to 29.4%). The identities of the partial sequence of two orthologs (U1-UN1 and SecA) at both ends of the cluster were high: 83.1 and 94.3%, respectively (Table 2), whereas four orthologs (U2-UN2, U3-UN3, U4-UN4, and U5-UN5) of unknown function had the low sequence identities of 21.6 to 45.5%.

View larger version (35K): [in this window] [in a new window]   FIG. 4.   Phylogram of OMPs of E. canis and E. chaffeensis. A total of 24 OMPs were segregated between two species (marked as "ortholog"), but 20 remaining proteins were not. The tree was constructed using the neighbor-joining (NEIGHBOR program from PHYLIP) method based on the alignment generated with CLUSTAL V, and 1,000 bootstrap replications were performed. The nodes supported by bootstrap values of >75% are indicated with an open circle symbol. The OMPs encoded by three repetitive regions in Fig. 3 are indicated by , 1, 2, and . The OMPs encoded by nonrepetitive regions were marked with a "."

View larger version (29K): [in this window] [in a new window]   FIG. 5.   Schematic diagram of diversity of the protein structure of OMPs. The symbols "" to "" represent proteins grouped in Fig. 3 and 4. The black and white areas show the HV and conserved sequences, respectively, among proteins in each group.

Transcription of the multigene family in E. canis. The RT-PCR was performed with 28 cycles of the linear range amplification using primer pairs specific to each of all 28 genes within the 28-kb locus in E. canis cultured in DH82 cells. Primer positions and amplicon sizes are shown in Table 1. DNA template controls are shown for each primer pair to demonstrate the ability of the primer to amplify the target sequence (Fig. 6A). Without reverse transcriptase, no amplicon was detected in RT-PCR analyses using any of primer pairs, indicating the absence of contamination of genomic DNA in the RNA preparation (data not shown). RT-PCR products for all 22 genes in the OMP locus were detected, indicating that all of these genes were transcriptionally active (Fig. 6A).

View larger version (52K): [in this window] [in a new window]   FIG. 6.   (A) RT-PCR analysis of 28 genes in the E. canis omp cluster. (B) RT-PCR analysis of 27 intergenic spaces and flanking genes on both sides in the E. canis omp cluster. The transcripts of the longest intergenic space and the flanking P30-20 and U4 were analyzed by three overlapping RT-PCRs (A, B, and C). DNA (5 ng of genomic DNA from purified E. canis) template control shows the intensity of the band detected with each pair of primers. Each amplicon with each primer pair was detected as a single band on the agarose gel stained with ethidium bromide. RT, RT-PCR. Amplified regions are shown in Table 1. The numbers on the right indicate the respective amplicon sizes.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

This study for the first time demonstrated: (i) the conservation of the omp multigene cluster structure and the genetic locus between CME and HME agents, (ii) the presence of three repetitive regions (, , and ) within the omp cluster and of phylogenetically promiscuous relationships among orthologs and paralogs within the repetitive regions, and (iii) cotranscription of intergenic spaces with flanking genes on both sides at 5'-end half and other regions of the omp cluster. There were several significant differences in the clustered genes of E. chaffeensis between the study by Yu et al. (31) and the current study: (i) 12 omp paralogs were identified upstream of omp-1A by Yu et al., but 13 paralogs were identified in the corresponding region in the present study; (ii) the clustered genes were found to be located downstream from the clpx gene at the 5' end by Yu et al., whereas in the present study these genes were arranged downstream from the un1 gene homologous to the hypothetical transcription regulator gene at 5' end and upstream from secA at the 3' end (the latter region was not identified by Yu et al.) (iii) a DNA probe corresponding to 3'-end omp-1W to 5'-end omp-1Y hybridized to 17.6- and 5.3-kb ClaI genomic DNA fragments according to Yu et al., but the probe can hybridize with only 17.6-kb ClaI-DNA in the present study. It is unlikely that these differences were generated by mutations within the omp cluster in E. chaffeensis (not strain differences, since the same strain Arkansas was used) maintained in two different laboratories, since the locus or gene numbers were conserved between the two different species: E. canis and E. chaffeensis. The cloning strategy used here was completely different from that of Yu et al. (31). Yu et al. assembled the sequence of the DNA fragments by using a rapid adapter primer PCR method. We used a combination of genomic Southern blot analysis, colony hybridization, and/or LA-PCR to avoid errors of sequence assembly due to the presence of repetitive sequences between homologous genes. If we had not used this strategy, for example, we might have misassembled the DNAs between the two (28- and 6.9-kb) loci in the E. canis genome. Although by the current genomic Southern blot analysis of E. chaffeensis, at least seven omp-1 genes were confirmed to be a single copy, for the remaining genes it is not known whether additional gene copies such as the E. canis 6.9-kb segment exist.

Multigene families encoding a major surface protein responsible for pathogenesis, such as antigenic variation, have been well documented in several bacteria and protozoa. For example, for the human pathogens Neisseria gonorrhoeae and N. meningitidis, pil has a gene organization similar to that of the multigene families of monocytic ehrlichiae (11, 24). PilE is the main structural pilin component and is expressed from the intact pilE gene. Several silent pilin genes (pilS) are tandemly arranged and contain only the 3' region of the gene without a start codon or transcriptional signals. These silent regions can be moved into the expression site by a recombination event, generating phenotypic changes by the expression of different PilE sequences. However, in the case of monocytic ehrlichiae, all genes of the multigene families had universal start codons and all were functionally active. The gene expressions appear to be regulated primarily by polycistronic event(s) on the 5'-end half of the omp cluster and monocistronic events on the 3'-end half. Therefore, it seems to be important to retain the unique gene organization for specific regulation of the expression of paralogs within the omp cluster rather than to modulate directly the gene arrangement of multigenes by rapid recombination events. Indeed, there is no direct evidence of the recombinational gene exchange responsible for such rapid antigenic variation in Ehrlichia spp. to date. E. canis often causes persistent infection in dogs, which may lead to severe chronic illness or death (27). Probably, E. canis in persistent infections escapes from the host immune surveillance by modulating the expression levels of individual genes in the multigene family.

The current transcriptional analysis provide more complete and different results from previous results, which examined 4 to 10 paralogs (16, 26, 31). Using RT-PCR, Yu et al. showed that six genes of E. chaffeensis, including genes in the region with the shortest intergenic spaces, namely, omp-1S, omp-1H, and omp-1Z, are monocistronically transcribed (31). In the present study, the orthologous genes of E. canis (p30-8, p30-7, and p30-6) were cotranscribed with the adjacent genes on both sides. In contrast to Reddy et al. (26), who reported only a single omp transcription in 4 genes of E. chaffeensis, in the current study all of the 28 genes were transcribed by E. canis in DH82 cells. The transcriptional initiation site of each omp gene has not been determined experimentally, but we previously identified putative promoters in the long intergenic spaces between the genes at the 3' end of the cluster (21). Thus, these genes may be independently regulated by their own promoter. Although repetitive sequences and the likely presence of long transcripts make it difficult, determination of the transcript size and a quantitative transcriptional analysis will be needed to further analyze the transcription mechanisms of these genes.

Recently, we reported the features of the p44 multigene family (orthologs of p30s and omp-1) of the human granulocytic ehrlichiosis (HGE) agent (32). The 16S rRNA sequence analysis showed 7.5 to 7.8% divergence between the HGE agent and ME agents, suggesting that they shared a common ancestor approximately 390 million years ago, while E. canis and E. chaffeensis were estimated to have diverged approximately 90 million years ago based on 1.8% divergence of the 16S rRNA gene sequences (1, 4, 18, 19). Our previous studies showed that the multiple p44 paralogs are dispersed widely throughout the HGE agent genome (rather than making a large gene cluster) (32) and that at least total 20 paralogs of the HGE agent were transcriptionally active under different conditions (unpublished data). Consequently, the following process may be inferred in the evolution of the HGE agent and ME agent. (i) The common ancestor of the HGE agent and ME agents gained (an) original major surface antigen gene(s). (ii) Thereafter, it began to produce the multiple paralogs. The HGE and ME agent lineages independently developed respective gene organizations between 90 to 390 million years ago. The HGE agent acquired the mainly trans-regulation system for the gene expression of genome-distributed paralogs. The ME agents developed the mainly cis-regulation system for gene expression of the clustered paralogs during this time. (iii) Eventually, the ancient omp cluster was vertically inherited by ME agent lineages (e.g., E. canis and E. chaffeensis). The mutations within respective omp clusters have been accumulated after divergence into the lineages of these two species, but the gene organization and the genetic locus have been retained with minor modification in each ME agent, probably for the preservation of the unique gene expression mechanism. It is important to take the genetic divergence and promiscuous relationships among OMP paralogs and orthologs (e.g., P30-6 is more closely related to OMP-1H rather than to OMP-1Z) into consideration in designing PCR primers and in future phylogenic analyses of the omp cluster among ME agents. The present results provide the essential information upon which to build functional and immunologic studies on OMPs in vitro and in animal models of ehrlichiosis.