Comparative molecular analyses of Borrelia burgdorferi sensu stricto strains B31 and N40D10/E9 and determination of their pathogenicity

Abstract

Background: Lyme disease in the United States is caused primarily by B. burgdorferi sensu stricto while other species are also prevalent in Europe. Genetic techniques have identified several chromosomal and plasmid-borne regulatory and virulence factors involved in Lyme pathogenesis. B31 and N40 are two widely studied strains of B. burgdorferi, which belong to two different 16 S-23 S rRNA spacer types (RST) and outer surface protein C (OspC) allelic groups. However, the presence of several known virulence factors in N40 has not been investigated. This is the first comprehensive study that compared these two strains both in vitro and using the mouse model of infection.

Results: Phylogenetic analyses predict B31 to be more infectious. However, our studies here indicate that N40D10/E9 is more infectious than the B31 strain at lower doses of inoculation in the susceptible C3H mice. Based-upon a careful analyses of known adhesins of these strains, it is predicted that the absence of a known fibronectin-glycosaminoglycan binding adhesin, bbk32, in the N40 strain could at least partially be responsible for reduction in its binding to Vero cells in vitro. Nevertheless, this difference does not affect the infectivity of N40D10/E9 strain. The genes encoding known regulatory and virulence factors critical for pathogenesis were detected in both strains. Differences in the protein profiles of these B. burgdorferi strains in vitro suggest that the novel, differentially expressed molecules may affect infectivity of B. burgdorferi. Further exacerbation of these molecular differences in vivo could affect the pathogenesis of spirochete strains.

Conclusion: Based upon the studies here, it can be predicted that N40D10/E9 disseminated infection at lower doses may be enhanced by its lower binding to epithelial cells at the site of inoculation due to the absence of BBK32. We suggest that complete molecular analyses of virulence factors followed by their evaluation using the mouse infection model should form the basis of determining infectivity and pathogenicity of different strains rather than simple phylogenetic group analyses. This study further emphasizes a need to investigate multiple invasive strains of B. burgdorferi to fully appreciate the pathogenic mechanisms that contribute to Lyme disease manifestations.

Figure 1

Binding of B. burgdorferi strains B31 (A and C) and N40D10/E9 (B and D) to both Vero (epithelial) cells and EA.hy926 (endothelial) cells is mediated by heparan sulfate. (A) and (B). Vero cell monolayers were pretreated with the buffer alone (Mock), or with the GAG lyases, heparinase I (HI) to remove heparan sulfate or chondroitinase ABC (Chon. ABC), to cleave chondroitin sulfates from the cell surfaces. Binding of B31 (A) to the Vero cells was significantly higher than that of the N40D10/E9 (B) strain. Although inhibition of binding of both N40D10/E9 and B31 was significant, reduction in binding was more pronounced by N40D10/E9 than B31 when Vero cells were treated with HI (p < 0.05). (C) and (D). EA.hy926 cell monolayers were mock-treated, or pretreated with HI or Chon. ABC enzymes. Removal of heparan sulfate from EA.hy926 cells eliminated the binding of both B31 and N40D10/E9 strains to these cells. The experiments were repeated at least three times using four replicates for each treatment. Each value represents the mean ± SD of quadruplicate samples. Asterisks indicate significant reduction (p < 0.05) in binding percentage compared to mock-treated cells as determined by t-test for pairwise comparison of samples with unequal variance.

Figure 2

Binding of B. burgdorferi strains B31 and N40D10/E9 to C6 glioma and T/C-28a2 chondrocyte cell monolayers was significantly reduced on pretreating these cells with chondroitinase ABC but remain unaffected on their pretreatment with heparinase I. The experiments were repeated at least three times using four replicates for each treatment. Each value represents the mean ± SD of quadruplicate samples. Asterisks indicate significant reduction (p < 0.05) in binding percentage relative to mock-treated cells as determined by t-test for pairwise comparison of samples with unequal variance.

Figure 3

The gene homologous to the bbk32 was not detected in N40D10/E9 strain by PCR and Southern hybridization. (A). Using the primers in Additional Table 1, complete genes bb0153, bb0184 (csrA), bb0219, bb0268, bb0383 (bmpA), bb0647 (bosR), bba25 (dbpB), bba68 (cspA), bbd14, bbd18, bbe22 (pncA), bbg02, bbh06 (cspZ), bbj09 (ospD), bbk17 (adeC), bbu06, and partial vlsE1 gene (using internal conserved primers) were amplified by PCR using B31 and N40D10/E9 strains genomic DNA as template. The bbk32 gene was amplified from B31 genomic DNA, however, PCR product was not detected in the N40D10/E9 strain. (B) Southern blot of EcoR1-digested genomic DNA of both strains (top) was hybridized with the probe prepared using the bbk32 PCR product from B31. An approximately 1.8 kb size fragment was detected only in B31, as expected, but not in the N40D10/E9 genomic DNA containing lane.

Figure 4

Two-dimensional gel electrophoresis of B31 and N40D10/E9 strains total proteins. Polypeptide spots with increased relative abundance (more than 1.7 fold increase) in B31 versus N40D10/E9 are outlined in blue while spots with decreased relative abundance (more than 1.7 fold decrease) in B31 versus N40 are outlined in red. Several of these spots were sent for MALDI-MS analysis.

Figure 5

Tibiotarsal joint inflammation in C3H mice inoculated with the N40D10/E9 and B31 strains. C3H mice were inoculated with a different inoculum dose (10, 10², 10³, 10⁴ spirochetes) of B31 or N40D10/E9 strains. (A) Two weeks after injection, severe tibiotarsal joint swelling was evident only in mice infected with 10³ or 10⁴ of B31. (B) However, severe tibiotarsal joint swelling could be observed in mice infected with 10, 10², 10³ or 10⁴ of N40D10/E9.