Polyphasic taxonomy has been advocated to ensure well-balanced determinations of taxonomic relationships. Different genes have been proposed to classify ehrlichial agents, however, the most widely used are 16S rRNA[8, 9], groESL operon, groEL gene, gltA, dsb, gp36, and gp19.
Sequence comparison of the 16S rRNA gene is recognized as one of the most powerful and precise methods for determining the phylogenetic relationships of bacteria[8, 11, 35]. Our results were consistent with previous phylogenetic analysis of Ehrlichia spp by using the 16S rRNA gene sequences[9, 36]. In this study, our analysis of a relevant fragment of 16S rRNA sequences revealed that the novel agent found in Brazilian R. (B.) microplus ticks was closely related to E. canis [GenBank: GU810149], but was also closely related to E. chaffeensis [GenBank: AF147752] showing 98.3% and 96.9% of homology, respectively. It is worth noting that the hypervariable region 16S rRNA is well conserved in members of the same species (data not shown) and are different among members of Ehrlichia genus[8, 9]. However, our hypervariable region of 16S rRNA was different when compared with other members of Ehrlichia genus.
Since the 16S rRNA gene is known to exhibit a high level of structural conservation with a low evolutionary rate, levels of sequence divergence greater than 0.5% in comparisons with nearly complete 16S rRNA gene sequences of members of the genus Ehrlichia have been considered sufficient to classify organisms as different species[8, 35]. The levels of divergence of the 16S rRNA sequence between this novel Brazilian ehrlichial agent and the closest member of the Anaplasmataceae, E. canis was 1.7% in pairwise comparisons of 1384 base sequences (data not shown), and this level of difference should be sufficient to classify the novel ehrlichial agent as a new species of the genus Ehrlichia. Furthermore, the 16S rRNA phylogenetic tree constructed with a maximum likelihood method show that E. mineirensis (UFMG-EV) falls in a different clade separated from any previously reported Ehrlichia spp.
The genes groEL and gltA have been proposed as an alternative to 16S rRNA for the phylogenetic analysis of the Anaplasmatacaea family as they are less conserved than 16S rRNA among the family members and dsb gene has been previously used to classified members of the Ehrlichia genus. It is important to note that the spacer of the groESL operon was 95 bp in E. mineirensis (UFMG-EV), which differs from the reported for E. canis, E. chaffeensis, E. ruminantium with 93, 100 and 96 bp, respectively. The gp36 orthologs are a divergent gene in E. canis, E. chaffeensis and E. ruminantium due to their high evolutionary pressure[14, 15]. This gene has been used to differentiate new isolates of E. canis where 16S rRNA was not well suited to discriminate between E. canis isolates.
In our study the level of similarity among ehrlichial gltA and dsb were lower than that of 16S rRNA and groEL gene sequences in the genus Ehrlichia. E. canis was the closest Ehrlichia species to E. mineirensis (UFMG-EV) in all the studied genes. Similar phylogenetic relationships are observed between other members of the Ehrlichia genus – i.e., E. chaffeensis/E. muris, N. risticii/N. sennetsu and A. marginale/A. platys.
The architecture of gltA, groEL and dsb based phylogenetic trees were similar to that of the tree derived from the 16S rRNA gene sequences. However, the trees constructed from gltA and dsb show more divergence than that from the 16S rRNA and groEL gene. The difference of E. canis and E. mineirensis (UFMG-EV) was well established in all the four trees based on nucleotide sequences. E. mineirensis (UFMG-EV) was well defined, with higher bootstrap values in the gltA (100) and dsb (100) based trees than for those of the 16S rRNA (97) and groEL (93) based tree.
Based on aa homology and genomic synteny analyses, it has been determined that the mucin-like protein of Ehrlichia ruminantium, gp36 of E. canis and gp47 of E. chaffeensis are orthologs. Identity of 87.2% has been found in the pre-repeat region among geographically distant E. canis isolates. The single tandem repeat was highly conserved among isolates (TEDSVSAPA) with variations in the number of repeats[13–15] and few conservative changes in amino acid sequences. The tandem repeat genetic unit varies in length (from 27 bp – 99 bp) among the different orthologs, number of repeats (from 3.4 - 56) and the homology of the nucleotide and the aa sequence encoded in the repeat (Table7). Our sequence contains a tandem repeat that shares an extremely low homology with the gp36 orthologs reported until now ranging from 22% (E. ruminantium and E. canis) to 33% (E. chaffeensis). Doyle et al. describes gp36 and gp47 as glycoprotein sharing O-glycosylation predicted sites in the serines and threonines of the tandem repeat. It is noteworthy that the tandem repeat of our sequence does not contain threonine; nevertheless, we predicted three sites of O-glycosylation in the serines of the tandem repeat and two in threonines of the post-repeat region. Two N-glycosylation sites were found in our aa sequence. The analysis for N-glycosylation was done for E. ruminantium, E. canis and for E. chaffeensis ortholog sequences (data not shown) and potential sites of N-glycosylation were found as well for these sequences. Glycosylation plays a crucial role in the immunogenicity of these glycoproteins[14, 15]. Deglycosylation of the gp36 tandem repeat drastically reduces its immunogenicity. Both gp36 and gp47 are described as the major immunoreactive protein of E. canis and E. chaffeensis and the tandem repeats contain the major antibody epitope[14, 15]. It was found that the tandem repeat of gp36 from E. mineirensis (UFMG-EV) contain the major B cell epitope previously reported for the glycoprotein orthologs. The prediction of the 3D structure of the B cell epitopes present in the tandem repeat shows a high structural divergence among the closest gp36 orthologs in E. mineirensis (UFMG-EV), E. canis and E. chaffeensis. These structural differences may explain the results obtained by Doyle et al. in which neither gp36 nor gp47 reacted with heterologous antisera.
The C + G content of the gp36 gene of E. mineirensis (UFMG-EV) is higher than the rest of the orthologs previously reported (data not shown). The C + G content in specific genes have been used in systematics as support for the classification of organisms, and it is known that recombination significantly increases the silent C + G content of a genome in a selectively neutral manner.
Although it is well known that Babesia bovis, B. bigemina and Anaplasma marginale are the most common etiological agents transmitted by R. (B.) microplus ticks, the detection of any species of Ehrlichia in R. (B.) microplus ticks has been infrequently reported. The first two reports were in China in the Guangxi Autonomous Region in 1999 and Tibet in 2002; the second in Thailand in 2003 and the latest one in Xiamen, China in 2011. Except the isolate from Guangxi, E. canis, the rest share, based on 16S rRNA, a 99.9% of homology[36, 40] and differ from the ehrlichial species previously reported and classified as Ehrlichia spp strain Tibet. In the present study, determined by pairwise alignment, the E. mineirensis (UFMG-EV) isolated from R. (B.) microplus shares 97% of similarity with the 16S rRNA sequences of the referred species (data not shown). This is the second report of a new Ehrlichia spp isolated from R. (B.) microplus, but the first to be reported in the American continent. The identification of E. mineirensis (UFMG-EV) in R. (B.) microplus ticks suggests a potential of infection and transmission of this agent to cattle in the area where infected ticks are present.