Three tick species were collected during this study, all of which had been reported from the northern Balkan region before . The majority of the ticks collected were I. ricinus which is the main vector for Lyme borreliosis in Europe. The infection rates of I. ricinus were similar to those reported in earlier studies in Croatia and other European countries [31–33] but much higher and lower infection rates were also reported from a neighbouring country [34, 35]. Seven out of nine B. burgdorferi sl genospecies known to occur in Europe were identified and only B. bissettii and B. kurtenbachii were not found . As in most European countries B. afzelii was the most common genospecies identified.
In EM skin biopsies three different B. burgdorferi sl genospecies were detected and the previously recognized overrepresentation of B. afzelii in the skin condition erythema migrans was confirmed [4, 6]. More importantly, a strong correlation between certain IGS types within the B. afzelii cluster and EM were found.
In earlier studies, certain ospC types of B. burgdorferi ss have been shown to be associated with higher probabilities of invasiveness and human disease [8–10, 36]. Lagal and co-workers found similar associations for B. afzelii and B. garinii strains in Europe but did not compare the frequency of the different ospC types in clinical samples to those in the tick population . We expanded these observations to EM in Croatia, where a broad range of B. burgdorferi sl is present and supported our findings with substantial tick data. Correlations of EM with certain types of the plasmid-encoded marker ospC were evident from our datasets. The expectation that this gene, due to its role in early infection, might show a stronger correlation to EM than IGS, could not be confirmed .
IGS type determination predicted the isolates propensity to cause EM better than genospecies determination alone. This is illustrated by the example that both the isolates with the highest and lowest propensity to cause EM belonged to B. afzelii genospecies. Significant positive and negative association with EM were more often found for IGS- than for ospC-typing (Figure 2) suggesting that IGS predicts EM formation better than ospC. This would imply that at least some of the genes responsible for EM formation are located on the chromosome or other non-exchangeable genomic elements.
As pointed out by Dykhuizen and co-workers it is important to use tick data alongside patient data to correctly estimate the propensity of strains to cause infection . This was done in the present study by sampling ticks around the main study site, from which the majority of patient samples were obtained. As a geographic bias could not be excluded, analyses were done for the whole study population or the Slavonski Brod study population alone. When only Slavonski Brod samples were taken into account, fewer IGS and ospC types were significantly more or less often found in patients than expected than when the whole study population was used. However, even when only isolates from this site were analysed one ospC and two IGS types were significantly more often found in patients than expected. Four patients from Čakovec and Rijeka study areas had B. burgdorferi sl sequences which were not found in any of the ticks can be explained by the lack of tick data from those study areas. These data highlight the importance of tick sampling representative for a certain study population. As Lyme borreliosis can be acquired by bites from adult as well as nymphal I. ricinus, a further potential bias lies in the testing of adult ticks only. Adult ticks might have obtained B. burgdorferi sl during feeding as nymphs. Hosts that fed nymphs and larvae might slightly differ and thus adult ticks might be infested with a different subpopulation of B. burgdorferi sl than nymphs.
It is unlikely that the propensity to cause EM or the invasiveness of Borrelia strains is determined by a single virulence factor alone. We rather hypothesize that a combination of different virulence factors encoded on the chromosome as well as on several plasmids of Borrelia determine the pathogenicity and invasiveness of different strains. Recently, a study showed that expression of ospC is regulated by a gene on a different plasmid . This emphasizes that the virulence of a certain strain, but also its fitness in a certain ecosystem, is a product of a complex system of genes interacting with each other and that it might be influenced by horizontal gene transfer. Even with the limited number of samples (n = 99) for which both genetic markers were available, five ospC types were found in more than one genospecies. This shows that substantial horizontal gene transfer takes place between different genospecies. Plasmid transfer has been reported before and shows that the apparent clonality seen in North American B. burgdorferi ss is lacking in European genospecies [10, 39–41].
The current study shows that different Borrelia IGS and ospC types exhibit different propensities to cause EM. We cannot exclude, however, that the genotypes involved in EM differ by location and in time. Furthermore, B. burgdorferi sl with IGS and ospC types identified to have a low propensity to cause EM here, could omit this early manifestation of Lyme borreliosis and directly cause disseminated manifestations of the disease. Besides the genetic constitution of B. burgdorferi sl, other explanations for the variability in the clinical manifestations of Lyme borreliosis are the (genetic) constitution of the human host [42, 43], co-infections with other pathogens [44–46] and the bacterial load of ticks. However, the current study only focuses on genetic aspects of B. burgdorferi sl.
In future studies other sample types, such as cerebrospinal fluid or arthritis isolates, could be analysed in a similar way as was done here. The two markers used here are not directly responsible for EM formation and efforts should be undertaken to identify virulence factors that determine clinical manifestations. Once virulence genes are identified, risk assessment for Lyme borreliosis could be improved by including only human pathogenic B. burgdorferi sl into the risk assessment. In future, this information might help in the design of ecological interventions to reduce Lyme borreliosis risk, the improvement of serology-based diagnosis by use of more specific antigens or for the development of specialised vaccines targeting virulent B. burgdorferi sl strains.