Discussion
The expansion of vector-borne pathogens is inherently linked to their ability to infect and transmit through reservoir host and vector populations. This fact can be observed as the current major etiological agents of human Lyme borreliosis (LB) in Eurasia (B. afzelii ,B. bavariensis , and B. garinii ) are vectored mainly by two different tick species: I. persulcatus in Asia, and I. ricinus in Europe. This means that each of these genospecies has, at least once during its evolution, successfully invaded a novel tick species and consequently a local population of vertebrate hosts. Yet how this invasion occurred and in which order is currently not known due to a lack of data. Here we report a reconstructed phylogeny of 142 Eurasian isolates belonging to the genospecies B. afzelii , B. bavariensis , and B. garinii . All three genospecies appear to share an Asian origin, suggesting a repeated expansion into Europe in relation to successfully invading a novel tick vector, I. ricinus . However, all three genospecies display unique sub-structuring which could be linked to ecological variability in their specific reservoir hosts. The results further show that the observed bottleneck in European B. bavariensis isolates argued to be in connection to invading I. ricinus (Becker et al., 2020; Gatzmann et al., 2015; Gabriele Margos et al., 2019), is not shared by the other two genospecies. This all suggests that Eurasian Lyme borreliosis agents were all capable of geographic expansion through vector shifts but differ in their capacity as emergent pathogens in relation to potential, future expansions into novel transmission cycles.
Borrelia bavariensis was already argued to have an Asian origin due to the deep branching observed between European and Asian isolates and that the majority of diversity exists in the Asian population (Becker et al., 2020; Gatzmann et al., 2015; Gabriele Margos et al., 2019). This finding is supported by the expanded analysis reported here. One point that warrants consideration is that all European B. bavariensis isolates come from humans and that the observed pattern in diversity could potentially be an artifact of sampling only a low diversity sub-set of European B. bavariensis better adapted to human infection. As our study includes European human isolates for two additional genospecies, we were able to disprove that this pattern is due to sampling bias through showing that human B. garinii orB. afzelii isolates do not display a similar reduction in genetic diversity (Text S2). In addition to this, a search of theBorrelia MLST (multiple locus sequencing typing) database (Gabriele Margos et al., 2008) shows eight B. bavariensis samples coming from I. ricinus DNA which do not differ from patient isolates on the eight MLST loci, which can roughly proxy the full chromosome diversity (Figure S3). These data support that the observed pattern in B. bavariensis is genuine. Compared to B. bavariensis , no research has focused on the geographic origin ofB. afzelii or B. garinii . Previous work raised the hypothesis of an Asian origin for B. afzelii but based on very few samples (Takano et al., 2011), whereas for B. garinii only partial structuring between continents was previously reported (Ana Cláudia Norte et al., 2020). Here though, we show that both B. afzelii and B. garinii are characterized by a basal node which splits a fully Asian clade from a clade of mixed geographic origin, suggesting for the first time that all three of these pathogenic genospecies originate in Asia and that through successful colonization of I. ricinus were able to expand into Europe. MDS clustering based on plasmid profiles further supported this by suggesting that the plasmid profiles present in Europe are a subset of available profiles present in Asia (Figure 3A). This could further show that the European population stems from the Asian population. Borrelia afzelii was the only genospecies which showed a step-wise colonization pattern from far-east Asia through Russia and into Europe (Figure S1) which has been observed in other tick-borne pathogens (Kovalev & Mukhacheva, 2014) suggesting differences in migratory patterns between species. We further hypothesized about which genospecies colonized Europe first through calculating absolute divergence (DXY ) and Tajima’s D (Tajima, 1989). Borrelia bavariensis shows the highest DXY suggesting that this colonization is the oldest of the three with B. afzelii then the youngest with the lowest value of DXY . Additionally, as expected from bacterial populations (Gatzmann et al., 2015; Tajima, 1989), Tajima’s D values are consistent with population expansion (negative Tajima’s D ) but the European expansion for each species is younger (more negative values). The magnitude of difference in Tajima’s D mirrors that of DXY withB. bavariensis showing the lowest difference in Tajima’s D(less recent) and B. afzelii showing the highest difference (most recent).
It is apparent from our analysis that, after the colonization of Europe, each genospecies experienced variable levels of gene flow which we argue can be related back to their host associations. The fact that B. garinii showed little to no geographic structuring is in accordance with previous results (Pär Comstedt et al., 2009, 2011; Ana Cláudia Norte et al., 2020). Borrelia garinii utilizes birds as reservoir hosts and exists in overlapping terrestrial and marine transmission cycles, where it is vectored by different tick species (terrestrial:I. ricinus and I. persulcatus ; marine: I. uriae ) (Pär Comstedt et al., 2006, 2011; Gómez-Díaz et al., 2011; A. C. Norte, Ramos, Gern, Núncio, & Lopes de Carvalho, 2013) (Figure 1). The lack of geographic structure observed in B. garinii is thought to be a result of this, as birds could aid in the migration of this genospecies (Pär Comstedt et al., 2009, 2011; Ana Cláudia Norte et al., 2020). This would explain why we cannot differentiate between distinct geographic locations. This pattern for B. garinii was already observed on a European level (Ana Cláudia Norte et al., 2020), but we are now able to show that it occurs over the whole distribution range of the genospecies. Borrelia afzelii and B. bavariensisdisplayed structured populations in our analysis. Within-continent structuring for European B. afzelii was previously attributed to utilizing rodents as reservoir hosts (Gallais et al., 2018; Vollmer et al., 2011), which we now propose to also occur in Asian B. afzelii populations. Even though our analysis does show some level of migration is possible along the geographic scale of this project as one Hokkaido isolate does cluster within the Honshu clade (Figure 2). AsB. bavariensis also associates with rodents (Gabriele Margos et al., 2009, 2013), we would expect to also observe geographic structuring. As previously reported, there does not appear to be gene flow between the European and Asian populations suggesting genetic isolation (high FST andDXY ; Table 1), but within Asia B. bavariensis is not structured as expected for a rodent adapted genospecies (Gabriele Margos et al., 2009, 2013). Our analysis builds upon previous work which observed migration between Asian regions (i.e. Japan, China, Russia) (Becker et al., 2020), but by further adding randomly sampled isolates from distinct Japanese islands: Honshu and Hokkaido. Unlike B. afzelii , where we observe lower migration between the islands (FST =0.379; Table S2),B. bavariensis isolates do not seem to have the same barrier to migration (FST =0.057; Table S2). This brings forward the question, what mechanism(s) could result in this unexpected migration of Asian B. bavariensis isolates? One suggestion could be that Asian B. bavariensis utilize secondary hosts besides rodents which increase effective dispersal rate. Recently, B. bavariensis DNA was found far afield of its Eurasian range in seabird associated ticks (I. uriae ) in Canada (Munro et al., 2017). As there are similarities in the structuring of Asian B. bavariensisto B. garinii from our results (low FST , high π , AMOVA with low σ due to geography; Table 1 & 2), it could be that in rare cases B. bavariensis may successfully transmit through avian hosts although rodent adapted. This fact had been previously observed where rodent-associated genospecies (i.e. B. afzelii ) appeared to transmit through avian hosts in Europe (Heylen, Matthysen, Fonville, & Sprong, 2014). Although the extent of transmission appears to be different between B. bavariensis andB. afzelii based on our analyses. Until 2009 (Gabriele Margos et al., 2009), B. bavariensis was considered a sub-type of B. garinii which utilized rodents as reservoir hosts (Masuzawa, 2004; Takano et al., 2011). This association with rodents was experimentally shown for two isolates (PBi, European; NT29, Asian) where they were exposed to rodent or avian immune sera and were susceptible to avian sera only (Kurtenbach et al., 2002, 1998). In this case, as in many studies, immune serum resistance is taken as a proxy of reservoir host associations (Kurtenbach et al., 2002, 1998). This result was used to support that B. bavariensis is not able to transmit through avian hosts. As the Asian population is quite diverse (Becker et al., 2020; Gatzmann et al., 2015) it is possible that a single isolate will not be representative for the entire population. Previous work did indeed suggest that similar genotypes of B. bavariensis (described as rodent adapted B. garinii ) which were isolated from infected mice in Japan shared unique sequence components to a bird isolated strain from the Korean Peninsula, suggesting that B. bavariensis could spread from mainland Asia to Japan through migratory birds (Ishiguro, Takada, & Masuzawa, 2005), as we are proposing. Additionally, a study based on restriction fragment length polymorphism (RFLP) analysis described a novel RFLP type (type IVa) (Nakao, Miyamoto, & Fukunaga, 1994) which is now known to belong to B. bavariensis (Dr. Minoru Nakao & Dr. Hiroki Kawabata, personal communication). The isolates belonging to this RFLP type were isolated from rodents, humans, but also birds (Nakao et al., 1994). Whole genome sequencing of these isolates would allow us to confirm if these bird isolates truly belong toB. bavariensis .
The results presented here suggest some answers to how LB spirochetes (B. afzelii , B. bavariensis , and B. garinii ) expanded across Eurasia, through showing that all currently known pathogenic Eurasian Borrelia genospecies expanded into Europe from an ancestral Asian population through successful colonizing a novel tick vector (I. ricinus ). Recently, B. garinii was found in I. uriae ticks in seabird colonies along the Atlantic coast of North America (Pär Comstedt et al., 2009, 2011; Smith et al., 2006). AsB. garinii was shown to be rarely transmitted through the North American tick vector (I. scapularis ) in lab based studies (Eisen, 2020) and here we show that B. garinii expanded into Europe through colonization of I. ricinus , potentially another expansion into the North American transmission cycle is possible if other requirements, such as reservoir host availability, are met. Outside of this, we further observed that post-colonization gene flow appears to relate to host association and were able to make further testable hypotheses regarding the ecology of the populations. Our analysis provides novel information to the spread of LB-causing spirochetes across Eurasia with applications to how adaptation to novel vector species can facilitate geographic expansion and thus potentially aid in the spread of emergent human pathogens.