From May through October 2008, we trapped small mammals at 5 sites in southern Sweden (Figure 1). Animals were captured with live traps and released after sampling. Our primary objective was to obtain samples from bank voles, Myodes glareolus, but other species were also sampled when captured. A total of 829 animals of 7 species were caught (Table). Approximately 20 µL of blood was taken from each animal. A nested PCR was performed with primers for Anaplasmataceae specific for the 16S rRNA gene (10).

All amplified fragments were sequenced, and a BLAST (http://blast.ncbi.nlm.nih.gov/Blast.cgi) search showed that 68 animals were infected by Candidatus N. mikurensis. The primers were chosen to be specific for bacteria belonging to the families Rickettsiaceae and Anaplasmataceae (10), but they also amplified Bartonella spp. under the given PCR conditions. In total, 35.5% of the animals were infected by Bartonella spp. Double infections with both Candidatus N. mikurensis and Bartonella spp., as indicated by double peaks on the sequencing chromatogram, occurred in 12 cases.

To further characterize the obtained Candidatus N. mikurensis, we sequenced 1,426 bp of the 16S rRNA and 1,233 bp of the groEL gene (1). The obtained 16S rRNA Candidatus N. mikurensis sequences in this study were identical (1,426/1,426 bp) to sequences obtained from human patients in Germany and Switzerland (2,3). The groEL sequence was identical to the isolate from Germany (1,233/1,233 bp) but differed slightly from the isolate from Switzerland (1,072/1,084 bp, 98.9% pairwise identity). A phylogenetic network containing unique 16s rRNA sequences from Candidatus N. mikurensis available at the National Center for Biotechnology Information was made with the program Network 4.5.1.6 (www.fluxus-engineering.com) by using the median-joining algorithm (11) (Figure 2).

The prevalence of Candidatus N. mikurensis at the 5 sites ranged from 0% to 12.5%, and Bartonella spp. occurred in 33.7% to 45.0% of the animals (Figure 1). The prevalence of Candidatus N. mikurensis and Bartonella spp. in each host species is given in the Table. Candidatus N. mikurensis occurred in all 4 rodent species, but not in the shrews; the difference in prevalence between rodents and shrews was statistically significant (p = 0.011, by Fisher exact test). Bartonella spp. were found in all rodent species and the common shrew.

In the present field survey of Anaplasmataceae in Swedish rodents, we have amplified 16s rRNA and groEL sequences identified as Candidatus N. mikurensis identical to sequences obtained from humans. This organism has been amplified from humans with febrile illness on ≥4 occasions (1–3), demonstrating that Candidatus N. mikurensis can cause human infections, at least occasionally.

Apart from the first human case, to our knowledge, Candidatus N. mikurensis has not been detected in Sweden. However, we identified Candidatus N. mikurensis at 4 of 5 sites, indicating that this organism is widespread in southern Sweden. Identical or similar sequences have been detected in Ixodes ricinus ticks from several European countries (3–6,13), showing that it is distributed over a large area of Europe. In contrast, we found no evidence of infection by any other Anaplasmataceae. I. ricinus is the dominating tick species in Sweden, and animals at all our study sites were heavily infested with larvae of this species and occasionally with nymphs (M. Andersson and L. Råberg, unpub. data). Reservoir hosts are essential for several tick-borne pathogens that lack the capacity for transovarial transmission from female ticks to larvae, such as A. phagocytophilum (14). Approximately 70% of the tick larvae in Sweden engorge on small mammals, such as voles, mice, and shrews (15).

The prevalence of Candidatus N. mikurensis was similar in all investigated rodent species in our study with a mean value of 8.8%. These results are consistent with several studies in Japan, which also found Candidatus N. mikurensis in several different rodent species (7,10). This finding indicates that rodents play a role in the natural cycle of Candidatus N. mikurensis in Europe and that rodents are likely to be competent reservoir hosts. In contrast, the closely related Candidatus N. lotori, found in the United States, seems to use raccoons rather than rodents as hosts (12). Whether the European variant of Candidatus Neoehrlichia is capable of infecting animals other than rodents and humans remains to be investigated. We found the organism was completely absent in shrews, which suggests that they might not be competent hosts for these bacteria.

We conclude that Candidatus N. mikurensis is geographically widespread in southern Sweden and that several rodent species, the main source of blood meals for larvae of I. ricinus in Sweden (15), can be infected. The relatively high prevalence in common rodent species suggests that the risk for humans and domestic animals to encounter infected I. ricinus nymphs and adults is substantial. Thus, Candidatus N. mikurensis should be considered when diagnosing bacterial infections associated with tick bites.