Development of a generic Ehrlichia FRET-qPCR and investigation of ehrlichioses in domestic ruminants on five Caribbean islands

The Ehrlichia are obligate intracellular Gram-negative tick-borne bacteria that are important human and animal pathogens. There is a need for assays to rapidly and reliably detect and differentiate the five generally recognized species into groups in a single reaction: E. canis, E. chaffeensis, E. ewingii, E. muris and E. ruminantium. We developed primers and probes against the 16S rRNA gene to enable us to reliably detect the five major Ehrlichia spp. in a single FRET-qPCR. We tested the Ehrlichia FRET-qPCR on reference strains and on DNA from the blood of domestic ruminants from five Caribbean islands. The Ehrlichia present were determined using melting point analysis and by sequencing the Ehrlichia FRET-qPCR products as well as those of a nested PCR against the citrate synthase gene (gltA). Our Ehrlichia FRET-qPCR was negative for the closely related Anaplasma marginale and A. phagocytophilum but gave positive reactions with reference strains of the most generally recognized species and with other less characterized Ehrlichia of domestic ruminants, mainly E. ovina, the Panola Mountain Ehrlichia, and Ehrlichia sp. BOV2010. Melting point analysis revealed 4 distinct groups: E. ruminantium (Tm ~55.8 °C); E. chaffeensis and E. ewingii (Tm ~57.7 °C); E. canis, E. muris, E. ovina and Ehrlichia sp. BOV 2010 (Tm ~62.0 °C); and the Panola Mountain Ehrlichia (Tm ~65.5 °C). The detection limit of the FRET-qPCR was ~ 5 gene copies in a reaction and the sequences of the FRET-qPCR products were as expected. With DNA from domestic ruminants from the Caribbean we found 12.2 % (134/1,101) positive: cattle (76/385; 19.7 %), sheep (45/340; 13.2 %) and goats (13/376; 3.5 %). Melting point analysis and sequencing of the FRET-qPCR and nested PCR gltA products showed the Ehrlichia we detected were E. canis or very closely related organisms. In a single reaction, our Ehrlichia FRET-qPCR can detect the Ehrlichia spp. we studied and differentiate them into four groups. Domestic ruminants in the Caribbean are not uncommonly exposed to Ehrlichia, possibly E. canis or very closely related organisms.


Background
Ehrlichia are obligate intracellular Gram-negative tickborne bacteria that are important animal and human pathogens. There are five generally recognized Ehrlichia spp., mainly E. canis, E. chaffeensis, E. ewingii, E. muris and E. ruminantium [1,2]. Ehrlichia ruminantium is the most important in domestic ruminants, where it causes heartwater, an acute disease associated with very high mortality (up to 90 %) and extensive economic losses [3]. Although various serological tests for E. ruminantium have been described, in particular ELISAs detecting antibodies to the organism's major antigenic protein (MAP), inappropriate positive results are not uncommon, probably due to cross-reactivity with other tickborne Ehrlichia spp. [4][5][6][7][8][9][10]. A number of such Ehrlichia that might be responsible for the serological crossreactivity have been described in domestic ruminants, including E. ovina in a sheep from Turkey [4], E. chaffeensis in goats and cattle in the USA [11,12], the Panola Mountain Ehrlichia in goats in the USA [9], and Ehrlichia sp. BOV2010/Ehrlichia sp. UFMT-BV in cattle in the Americas [13,14]. There are also other Ehrlichia that have been reported in domestic ruminants, but stocks are not readily available and their taxonomic status is yet to be confirmed [15]. These include E. ondiri [15], Ehrlichia sp. Omatjenne [16], Ehrlichia sp. Germishuys [16] and an Ehrlichia sp. from Zimbabwe [7].
In a recent study in the Caribbean, inappropriate positive MAP-1B ELISA results for E. ruminantium were reported for domestic ruminants from four of the seven islands studied [10]. These inappropriate positive reactions were thought to be due to infections with other Ehrlichia spp., and the presence of these organisms made serological testing for E. ruminantium unreliable in the Caribbean, as has been shown to be the case in Africa [17]. Being able to reliably detect E. ruminantium is important as it is not only a serious threat to local livestock production, but also to animals on the American mainland [18]. To further investigate ehrlichioses in domestic ruminants in the Caribbean, we developed a generic Ehrlichia FRET-qPCR that would enable us in a single reaction to specifically and reliably detect the major Ehrlichia spp. and differentiate them into groups. The development and validation of this PCR and its use to screen domestic ruminants in the Caribbean for Ehrlichia spp. is described below.

Blood samples
Whole blood samples (n = 1,101) in EDTA were collected from apparently healthy domestic ruminants (cattle, sheep and goats) on Montserrat (n = 77), St. Kitts (n = 373), Grenada (n = 140), Nevis (n = 262) and Dominica (n = 249) as described previously [10,19] ( Table 1). Aliquots of 200 μl were frozen at −20°C until DNA was extracted for PCR studies. Ethical Approval: All work in this study was reviewed and approved by the Institutional Animal Care and Use Committee of Ross University School of Veterinary Medicine. Owners of the animals provided consent for blood samples to be collected.

Ehrlichia strains
As positive controls we used the five major Ehrlichia spp., mainly E. ruminantium, E. canis, E. chaffeensis, E. ewingii and E. muris. We also tested Ehrlichia that were available to us and have been previously reported to occur in ruminants, mainly E. ovina [4], Ehrlichia sp. BOV2010 [13] and the Panola Mountain Ehrlichia [9]. We used DNA extracted in previous studies from E. ruminantium [10] and E. canis [20], DNA extracted as described below from tissue cultures of E. canis (Oklahoma) and E. chaffeensis (Arkansas) (supplied by Gregory Dasch, Centers for Disease Control, Atlanta), from blood stabilates (E. ovina and Ehrlichia sp. BOV2010), and from an Amblyomma variegatum positive for the Panola Mountain Ehrlichia by PCR (unpublished data). We also used plasmids that were created to contain an appropriate portion of the 16S rRNA gene of E. ewingii and E. muris using the pIDTSMART cloning vector (Integrated DNA Technologies, Coralville, IA, USA) and linearization with HindIII (Promega, Madison, WI, USA).
To test the specificity of our PCR, we tested DNAs extracted from blood of cattle verified to be infected with A. marginale (identical nucleotide 16S rRNA sequences with CP006847) and A. phagocytophilum (identical 16S rRNA sequences with KJ782389).  (Fig. 1). The sequences were aligned and regions were identified for primers and probes based on the conserved and variable areas of the alignments. The forward primer (5′-GAGGATTTTATCTTTGTATTGTAGCTA AC-3′), reverse primer (5′-TGTAAGGTCCAGCCGA ACTGACT-3′) and fluorescein probe (5′-ACGCGAA AAACCTTACCACTTTTTGAC-6-FAM-3′) we selected had identical sequences in all the Ehrlichia. The LCRed 640 probe (5′-LCRred640-GAAGGTCGTATCCCTCTTA ACAGG-phos-3′) was identical to the Panola Mountain Ehrlichia but had one nucleotide mismatch with E. canis, E. muris, E. ovina and Ehrlichia sp. BOV 2010, two mismatches with E. ewingii and E. chaffeensis, and three mismatches with E. ruminantium (Fig. 1). In contrast, the primers and probes had multiple mismatches  with Anaplasma spp. and other related bacteria (Fig. 1). When we used the BLAST to compare the primers and probes we developed against all sequences available on GenBank, we found they reliably detected the Ehrlichia spp. against which they were designed and that the nucleotide polymorphisms we used in the probes for the different species were highly conserved.

Thermal cycling and melting curve analysis
High-resolution melting curve analysis following PCR was performed on a Roche Light-Cycler 480-II platform as described before [21,22].   (Fig. 2). The T m value is influenced not only by nucleotide mismatches but also the types of nucleotides and GC percentage of the probes.

Sensitivity
For quantitative standards we used amplified DNA of E. canis identified in a previous study [20]. These E. canis DNA amplification products were confirmed by nucleotide sequencing (GenScript, Nanjing, Jiangsu, China) before being gel purified with a QIAquick Gel Extraction Kit (Qiagen, Valencia, CA) and quantified using the PicoGreen DNA fluorescence assay (Molecular Probes, Eugene, OR). The molarity of the E. canis DNA was estimated using the calculated molecular mass of the amplicons [23] and dilutions made to give solutions containing 10,000, 1,000, 100, 10, and 1 gene copies/μl in T 10 E 0.1 buffer which were used as quantitative standards.

Specificity
The specificity of the positive control PCRs with the five widely recognized Ehrlichia spp. (DNAs of E. canis, E. chaffeensis and E. ruminantium, and plasmids representing E. ewingii and E. muris) were confirmed by electrophoresis of amplicons through 1.5 % MetaPhor agarose gels, purification using the QIAquick PCR Purification Kit (Qiagen, Valencia, CA, USA) and sequencing of both DNA strands using the appropriate forward and reverse primers (GenScript, Jiangsu, Nanjing, China). No reaction products were obtained when our generic Ehrlichia FRET-qPCR was performed with DNAs of A. marginale or A. phagocytophilum.
Nested PCR for the citrate synthase gene of Ehrlichia To amplify the citrate synthase gene (gltA) of Ehrlichia spp., we carried out nested PCRs (outside primers: EHRCS-131F and EHRCS-1226R, and inside primers: EHRCS-754F and EHRCS-879R, which amplify 1,108 and 126 bp sections of the gene, respectively) as described previously [9]. The PCR products we obtained were verified by gel electrophoresis, purified using the QIAquick PCR Purification Kit (Qiagen, Valencia, CA, USA) and sequenced (GenScript, Jiangsu, Nanjing, China).

Development of a generic Ehrlichia FRET-qPCR
The generic Ehrlichia FRET-qPCR we established produced amplicons with each of the five well recognized Ehrlichia spp. that we tested, mainly E. ruminantium, E. chaffeensis, E. ewingii, E. canis and E. muris. The FRET-qPCR was also positive with DNA from E. ovina, the Panola Mountain Ehrlichia, and Ehrlichia sp. BOV 2010. Sequences of the amplification products were as expected for each organism (results not shown). No products were obtained when the generic Ehrlichia FRET-qPCR was performed with DNA from A. marginale and   (Fig. 2). No reaction products or melting peaks were found with the positive DNAs of A. marginale and A. phagocytophilum. When we tested around 300 copies of E. ruminantium, E. chaffeensis, Ehrlichia sp. BOV 2010 and the Panola Mountain Ehrlichia/μL in a single reaction, the Ehrlichia FRET-qPCR revealed 4 distinct melting curves with temperatures identical to those found with individual FRET-qPCRs of the agents.
With the quantitative standards developed using purified E. canis DNA, we determined that the detection limit of the generic Ehrlichia FRET-qPCR was~5 copies of the 16S rRNA gene per PCR.
The 1,015 bp gltA sequence we obtained for E. ovina and deposited in GenBank (KP719095) was 99.9 % (2 mismatches) identical to that of E. canis from Italy (AY647155).  The Gene Accession numbers for representing Ehrlichia spp. in this

Discussion
The generic Ehrlichia FRET-qPCR we developed proved to be both specific and sensitive in detecting Ehrlichia spp. in controlled experiments. In a single reaction it reliably detected the five commonly recognized Ehrlichia spp. we used in our experiments as well as less well characterized Ehrlichia which have been found in domestic ruminants and are available for study [26]. The specificity of the PCR was shown by its failure to detect representatives of the closely related Anaplasma genus, A. marginale and A. phagocytophilum, and the fact that all the positive reaction products had sequences that were closest to Ehrlichia spp. When tested against dilutions of E. canis, the sensitivity of the generic Ehrlichia FRET-qPCR was high, detecting as few as 5 copies of the 16S rRNA gene in a reaction [27].
The 16S rRNA gene we detected in our generic Ehrlichia FRET-qPCR is a common target for PCRs for Ehrlichia spp. as its nucleotide sequence is highly conserved in the genus. By systematically aligning the sequences of the main Ehrlichia spp. and closely related organisms, we were able to identify a highly conserved region of the 16S rRNA gene against which we developed specific primers that only amplified Ehrlichia spp. and not organisms from related genera. Further, the region of the 16S rRNA gene we selected for our LCRed 640 probe had nucleotide mismatches between the major Ehrlichia spp. which enabled us to differentiate the organisms into groups by melting point analysis (Fig. 1).
When we tested our generic Ehrlichia FRET-qPCR against known Ehrlichia spp. it detected all the organisms in a single reaction and also differentiated the species to a large extent. Using high-resolution melting point analysis we were able to clearly differentiate E. ruminantium, the Panola Mountain Ehrlichia, a group containing E. chaffeensis and E. ewingii, and a group containing E. muris as well as E. canis and organisms closely related to it. The groupings appear to be largely serendipitous, rather than of taxonomic significance, as molecular studies have shown E. muris is more closely related to E. chaffeensis than to E. canis [26]. Similarly, E. ewingii is closer to E. canis [27] or the Panola Mountain Ehrlichia [9] than to E. chaffeensis.
When we applied our generic Ehrlichia FRET-qPCR to DNA from whole blood collected from domestic ruminants from five Caribbean islands, we identified relatively high prevalences of infections (12 %) with Ehrlichia spp. that were not E. ruminantium. Of note is the fact these generic Ehrlichia FRET-qPCR positive animals had previously tested negative for antibodies to E. ruminantium in a MAP-1B ELISA [20]. This test not only detects antibodies to E. ruminantium [6] but also to the Panola Mountain Ehrlichia [9], and E. canis and E. chaffeensis [6]. It seems unlikely, then, that the animals we found positive in our generic Ehrlichia FRET-qPCR had been infected with these agents. While there are no data for Ehrlichia sp. BOV2010 and Ehrlichia sp. UMFG-EF, sera from animals infected with E. ovina do not give positive MAP-1B ELISA reactions [6] and these, or closely related organisms, seem most likely to have been detected by our generic Ehrlichia FRET-qPCR in the seronegative animals. Also of note is that none of the animals that had previously been found to be positive in MAP-1B ELISAs [10] were positive in our generic Ehrlichia FRET-qPCR. Most of these animals, however, were only very weakly positive in the MAP-1B ELISA suggesting they had residual antibody titers following clearance of infections, or that the infecting Ehrlichia spp. did not generate a substantial humoral response. Further studies are underway in our laboratories to clarify the position.
Melting point analysis and sequencing suggested that the Ehrlichia we identified with our generic Ehrlichia FRET-qPCR, utilizing the 16S rRNA gene, were E. canis or closely related organisms. The 16S rRNA gene is highly conserved in E canis, being 99.4-100 % identical between strains [28,29], and hence a reliable way of identifying isolates. Although we sequenced only a relatively short segment of the gene (210 bp), the Ehrlichia spp. we identified with our generic Ehrlichia FRET-qPCR had 100 % homology with E. canis sequences in GenBank.
We found only relatively small numbers of organisms in the blood samples we studied (average copy number 231, median 9.5) with our generic Ehrlichia FRET-qPCR, most likely because we were detecting chronic subclinical infections but also perhaps because we were detecting infections in accidental and unsuitable hosts. The low copy numbers in our samples were also evident from the results of our gltA gene PCRs where we only found positive results after nesting. The gltA gene has also been shown to be highly conserved in E. canis (over 99 %) [30], but it has greater interspecies variability than the 16S rRNA gene which might make it more useful for differentiating Ehrlichia species [9,31]. The sequences we obtained for our nested gltA, however, were consistent with the 16S rRNA gene findings that the organisms present in the Caribbean domestic ruminants we studied were E. canis or closely related species.
We would note that, because of low copy numbers in our samples, the sequences we obtained from our nested gltA PCR were with the internal primers and thus relatively short (126 bp). These internal primers, however, amplify a hypervariable region of the gltA which enables accurate discrimination of species and strains. When we compared the sequences we obtained with others in GenBank we found that, consistent with comparisons of our 16S rRNA gene sequences, the organisms present in the Caribbean domestic ruminants were closest to E. canis or closely related organisms.
Although E. canis is best known as a very common dog pathogen around the world, including in the Caribbean [20,32], infections have also been described in humans [33] and cats [34]. There is a growing belief that E. canis has a wider host range than previously thought [1,14], and our findings are largely consistent with this idea. Of further note is that a number of Ehrlichia that appear to be closely related to E. canis, possibly even strains of this organism, have been reported in domestic ruminants. E. ovina (AF318946) was first recovered from a sheep in Turkey and subsequently caused illness in splenectomized Dutch sheep [4]. More recently it has been found to have an identical 16S rRNA sequence to E. canis in dogs from Turkey (Kutahya strain; AY621071) [35] and Venezuela (VHE strain; AF373612) [36]. In our study, E. ovina had a T m and 16S rRNA sequence identical to that of the Oklahoma strain of E. canis (NR_118741) and of local Caribbean strains we found. Further, E. ovina reacted with primers for the gltA of Ehrlichia spp. and produced a 1015 bp sequence that contained only 2 mismatches with E. canis from Italy (AY647155). These findings provide further support for the proposal that E. ovina is a strain of E. canis [1,34].
Recent studies have identified 3 novel cattle-related strains of Ehrlichia: in Canada, the Ehrlichia sp. BOV 2010 [13]; and in Brazil, the Ehrlichia sp. UFMG-EV in Rhipicephalus microplus hemolymph [37] and the Ehrlichia sp. UFMT-BV in cattle [1]. Molecular studies have shown these organisms are very similar to one another and that they probably evolved from a highly divergent and variable clade within E. canis [38]. The phenotypic and genotypic differences the strains have with E. canis have been ascribed to the organisms adapting to their new hosts, ruminants, and their new tick vectors. More detailed genomic and transmission studies might provide justification for the organism being classified as a distinct species, E. mineirensis [38].

Conclusions
In conclusion, the Ehrlichia FRET-qPCR we developed proved sensitive and specific in detecting the most recognized Ehrlichia spp. of ruminants in a single reaction. Further, using melting point analysis we could differentiate the organisms into four groups comprising E. ruminantium; E. chaffeensis and E. ewingii; E. canis and closely related organisms such as E. ovina and Ehrlichia sp. BOV2010/Ehrlichia sp. UFMG-EV; and the Panola Mountain Ehrlichia. When we used the generic Ehrlichia FRET-qPCR on DNA from the blood of Caribbean domestic ruminants, we found a relatively high percentage (12.2 %) were positive. Melting point analysis showed the Ehrlichia in the Caribbean domestic ruminants were most similar to organisms in the group comprising E. canis and closely related species.