Plasmid content is generally homogenous between genospecies
Borrelia afzelii and B. bavariensis both differed significantly in plasmid numbers between Europe and Asia, with EuropeanB. afzelii (two-sided unpaired t-test, p = 0.03) and AsianB. bavariensis (p < 0.001) having significantly more plasmids in comparison to the other geographic population (Figure 3A). Borrelia garinii isolates did not differ in overall plasmid number between Asia and Europe (p = 0.08) but had significantly fewer plasmids in comparison to both B. afzelii populations (Asian, p = 0.003; European, p < 0.001) and to Asian B. bavariensis (p < 0.001) (Figure 3A). When we look at the absolute plasmid number for a population, defined as the number of unique plasmid types present in at least one isolate from the population, only European B. bavariensis showed a lower absolute plasmid number (black circle; Figure 3A) in comparison to the other species such as B. garinii . Borrelia garinii also had on average lower plasmid numbers per isolate (comparable to EuropeanB. bavariensis ), but the absolute number of plasmid types present in the population (i.e. diversity of plasmid types) is comparable toB. afzelii and Asian B. bavariensis (Figure 3A).
Based on the plasmid presence/absence matrix for all samples (n = 136), we further ran a multi-dimensional scaling (MDS) analysis to test if plasmid content corresponds to factors such as continent (i.e. vector) or genospecies. Plasmid content appears more homogenous between Asian isolates (Figure 3B) versus European isolates, which display clusters based on genospecies (Figure 3B). This could result from European isolates representing a subset of available plasmid combinations which are all present in the Asian populations. Even so, no plasmid types were more frequently associated with factors such as genospecies or geography (Figure S2). It is important however to note, as Borrelia can lose plasmids due to long-term culturing (G. Margos et al., 2017). Many human isolates (Table S1) have been potentially kept in culture longer suggesting that the current plasmid results could be biased even though sequencing focused on low passage isolates (<10 passages). Tick isolates did indeed have on average higher plasmid content (Table S6; mean: 1.19; 95% CI: 0.16, 2.22) suggesting that this bias may be present for our human isolates.