This work presents the first screening aimed at detecting DNA from tick-associated Midichloriaceae bacteria in mammalians. A total of 156 blood samples derived from horses, cattle, sheep and dogs at risk of tick bite were analyzed by PCR, using a previously described PCR protocol targeted on the 16S rDNA gene of M. mitochondrii and related bacteria . Blood samples from 30 cattle and 20 dogs that are regarded as having no or limited risk of tick bite were also included. Details on the different groups of animals examined are listed in Table 1.
The quality of the DNAs extracted was checked by PCR using universal mammalian primers targeted on the mitochondrial 12S rRNA gene that led to positive amplification from all samples. Midichloria-targeted primers led to positive amplification from 26 of the animals at risk of tick bite, and from none of the control animals (Table 1). The animal species that presented the higher prevalence of positive animals was the dog, where 20 out of 86 animals were positive. Eighteen of these dogs were hosted in a kennel in the Pantelleria island (part of the Sicily region), where a dense population of the hard tick Rhipicephalus bursa has been recorded (Epis S., personal observation), while two dogs where hosted in two kennels in the Molise and in the Campania regions (Italy). The PCR products obtained from all Midichloria-positive animals (five horses, one sheep, twenty dogs) were recovered from the gel, purified and sequenced by ABI technology; the obtained sequences were manually corrected, compared with the databases using Blast, and included into different alignments with homologous sequences, for phylogenetic analyses. All of the sequences generated gave the best scores toward 16S rDNA sequences from Midichloriaceae bacteria. Phylogenetic analyses further confirmed that the sequenced PCR products derived from bacteria closely related with M. mitochondrii. In particular, in phylogenetic analyses including all of the sequences available from Midichloriaceae bacteria, the sequences here obtained from mammalians were placed into cluster 2 of this family (result not shown), which corresponds to the genus Midichloria. Figure 1 presents an example of a phylogenetic tree obtained including representatives from the two main genera of the Midichloriaceae, Midichloria and Lariskella: the gene fragments generated from the blood samples here examined cluster with those of Midichloria bacteria previously detected in ticks (trees generated using different settings/algorithms presented the same placement for Midichloria sequences derived from blood samples). It has already been shown that Midichloria bacteria harbored by different tick species are variable at the level of the 16S rDNA .
The tree in Figure 1 shows that the novel sequences from mammalians are in some cases identical to those obtained from ticks: sequences from three horses to those obtained from three different tick species (Ixodes uriae, Haemaphysalis punctata and Rhipicephalus turanicus; sequences published in ref. ); sequences from eight dogs from the Pantelleria island to those obtained from R. bursa; sequences from two dogs from the Molise and Campania regions to those obtained from I. ricinus. Other 16S rDNA gene sequences obtained from the examined mammalians differ from those so far generated from ticks (showing from 2 to 5 nucleotide substitutions from the closest 16S rDNA Midichloria sequence from ticks). This result is not surprising, considering that only a minimal proportion of the ticks present in Italy have been screened for Midichloria bacteria ; the overall molecular diversity of these bacteria is thus still to be determined. In summary, we can affirm that DNA from bacteria attributable to the genus Midichloria can be detected in the blood of different animal species, and, in at least a few of the cases here examined, we can reasonably hypothesize that the origin of the detected DNA could be traced to ticks. For example, R. bursa, I. ricinus and R. turanicus could be involved in the transmission of Midichloria bacteria (or their DNA) to dogs and horses, also considering the host spectrum of these tick species and/or their distribution in the areas where the blood samples were collected [19, 20] (e.g. the case of R. bursa in the Pantelleria island and of I. ricinus in the Molise and Campania regions).
The above results prompted us to develop an ELISA test to screen sera for indirect signs of infection by M. mitochondrii, focusing the study on the dog; this test was based on a recombinant antigen from M. mitochondrii, the flagellar protein rFliD [11, 18]. We emphasize that we do not expect cross-reactivity of this antigen toward tick-associated spirochetes, Rickettsiaceae or Anaplasmataceae; rather, cross-reactivity with other bacteria from the family Midichloriaceae cannot be excluded (see discussion below). Using this test we analyzed 218 dog sera collected from 16 kennels located in southern Italy (Table 2), and the sera from the 20 experimental dogs included as controls. The cut-off of the test was determined on the basis of results obtained from the sera from the control dogs, and positioned at 0.25 O.D. As shown in Table 2, the average O.D. values for IgG antibodies reacting with rFliD is above/equal to the cut-off in K1-6. In these six kennels the percentage of positive dogs ranges from 22.2% to 86.6%. In the remaining 10 kennels, the average O.D. value was below the 0.25 cut-off: dogs from four of these were all negative (K13-16), while some positive dogs were recorded in the remaining six, as indicated by the O.D. maximum value (K7-12; Table 2).
Considering the whole population of the dogs examined from the 16 kennels, seroprevalence is 26.6%, which is significantly different from the 5% seroprevalence of control dogs (Student t-test, P < 0.05). The 218 dogs from the kennels can generally be assumed to be at risk of tick bite, while the 20 experimental dogs can be assumed to have no risk of tick parasitism. Considering the above information, the results here reported (i.e. 26.6% seroprevalence for M. mitochondrii in dogs at risk of tick bite, and 5% in dogs at no risk) are congruent with the idea that antigens from M. mitochondrii (or from closely related bacteria) are inoculated into animals during the tick blood meal. As for the differences in the seroprevalence in dogs from the different kennels (Table 2), this could derive from management/sanitary differences among kennels, as well as from their geographic and environmental location, in relation with tick distribution in Italy. For example, in the rural environments of the Molise region, dense populations of I. ricinus have been recorded (Rinaldi L., unpublished observations); this can explain the 86.6% seroprevalence recorded in K1 (see Table 2), also considering that positivity for M. mitochondrii is 100% in all the life stages of I. ricinus, excluding adult males . In addition, all of the kennels displaying an average O.D. above the cut-off (Table 2, K1-6) are located in rural areas, where tick presence has been recorded [21, 22] (Rinaldi L., unpublished observations). On the other hand, the presence of ticks is more sporadic in the urban and suburban environments of Napoli, Salerno and Avellino counties, where kennels displaying lower O.D./seroprevalence values (K7-16) are located [21, 22]. For only three of the kennels that were included in the serological screening we could collect, for a subsample of the dogs, whole blood for DNA extraction/PCR screening, in addition to sera samples (K1, K3 and K14; see Tables 1 and 2). In the case of K1 and K3, one PCR positive dog was detected in each kennel (K1, one out of four; K3, one out of seven; see Table 1); the O.D. values for these two PCR-positive dogs were respectively 0.67 and 0.37, i.e. both of them can be classified as seropositive for M. mitochondrii. Being based on a subsample, the above results do not, however, allow us to estimate any correlation between serological and PCR positivity. It is anyway interesting that the two kennels where serological positivity is higher present PCR-positive dogs (K1 and K3), while no positive dog was detected in a kennel where all of the animals are seronegative for M. mitochondrii (K14).
The antigen used for the above ELISA screening (i.e. rFliD) is a component of the flagellum of M. mitochondrii from I. ricinus. This bacterium is rather peculiar in that it is the sole Rickettsiales so far shown to possess a flagellar structure: the well-established pathogenic Rickettsiales form the genera Rickettsia, Ehrlichia and Anaplasma do not have flagella . We would thus exclude that results of the above serological screening derived from cross-reactivity with antigens from other Rickettsiales. On the other hand, Borrelia burgdorferi sensu lato, the main pathogen transmitted by Ixodes ticks, does possess immunogenic flagella. However, published results show that, in humans exposed to tick bite, a high proportion of the subjects positive to M. mitochondrii are negative to B. burgdorferi (and vice-versa), indicating the absence of immunological cross-reactivity among the FliD proteins of Midichloria and Borrelia bacteria . On the other hand, cross-reactivity of rFliD could be expected toward other bacteria from the genus Midichloria, under the reasonable assumption that other bacteria from this genus possess the flagellar gene fliD.
The work reported here presents two lines of evidence that suggest that Midichloria bacteria circulate in mammalians: 1) direct signs of their presence, i.e. the detection of DNA gene fragments in blood samples from different animal species, that cluster in the Midichloria bacterial genus; 2) indirect signs of their presence, i.e. the detection of anti-Midichloria antibodies in dog sera. These results do not of course allow us to infer any conclusion about whether Midichloria bacteria replicate in mammalian hosts: ticks (or other arthropods) might simply inoculate Midichloria (or DNA/proteins from these bacteria), in an amount sufficient for PCR detection and to stimulate antibody production. However, considering the amount of blood in the animals here examined (e.g. horses), we believe it would be unlikely that bacterial DNA inoculated by a few ticks, and diluted into liters of blood, could then be detected by PCR, by the analysis of the DNA extracted from 100 microliters of blood, in the absence of a multiplication in the animal host. We are thus more prone to consider the possibility that Midichloria bacteria multiply in the mammalian host. As for the antibody response toward the rFliD Midichloria antigen, repeated inoculations of antigens from Midichloria by several ticks, as it might occur in animals, could be sufficient to stimulate an antibody response. On the other hand, the seropositivity for Midichloria recorded in humans parasitized by ticks is suggestive of a replication of these bacteria, considering that single ticks had generally been removed from the examined subjects, normally after a very short blood meal .