Plasmid diversity and phylogenetic consistency in the Lyme disease agent Borrelia burgdorferi

Abstract

Background: Bacteria from the genus Borrelia are known to harbor numerous linear and circular plasmids. We report here a comparative analysis of the nucleotide sequences of 236 plasmids present in fourteen independent isolates of the Lyme disease agent B. burgdorferi.

Results: We have sequenced the genomes of 14 B. burgdorferi sensu stricto isolates that carry a total of 236 plasmids. These individual isolates carry between seven and 23 plasmids. Their chromosomes, the cp26 and cp32 circular plasmids, as well as the lp54 linear plasmid, are quite evolutionarily stable; however, the remaining plasmids have undergone numerous non-homologous and often duplicative recombination events. We identify 32 different putative plasmid compatibility types among the 236 plasmids, of which 15 are (usually) circular and 17 are linear. Because of past rearrangements, any given gene, even though it might be universally present in these isolates, is often found on different linear plasmid compatibility types in different isolates. For example, the arp gene and the vls cassette region are present on plasmids of four and five different compatibility types, respectively, in different isolates. A majority of the plasmid types have more than one organizationally different subtype, and the number of such variants ranges from one to eight among the 18 linear plasmid types. In spite of this substantial organizational diversity, the plasmids are not so variable that every isolate has a novel version of every plasmid (i.e., there appears to be a limited number of extant plasmid subtypes).

Conclusions: Although there have been many past recombination events, both homologous and nonhomologous, among the plasmids, particular organizational variants of these plasmids correlate with particular chromosomal genotypes, suggesting that there has not been rapid horizontal transfer of whole linear plasmids among B. burgdorferi lineages. We argue that plasmid rearrangements are essentially non-revertable and are present at a frequency of only about 0.65% that of single nucleotide changes, making rearrangement-derived novel junctions (mosaic boundaries) ideal phylogenetic markers in the study of B. burgdorferi population structure and plasmid evolution and exchange.

Keywords: B. burgdorferi; Genome rearrangement; Linear plasmid; Plasmid.

Fig. 1

Sequenced B. burgdorferi sensu stricto plasmids. The 14 completely sequenced genomes shown as columns where shaded cells indicate the presence of a plasmid. For the linear plasmids, different cell colors and Roman numerals indicate subtypes; similar colors and numerals in different plasmids (lines in table) do not imply any relationship. For the circular plasmids light yellow shading marks the cp32 plasmids with large deletions, and Roman numerals denote the subtype of the cp9 and cp26 plasmids. No subtypes were defined for the cp32s, so a “ + ” indicates a cp32 of that compatibility group is present. “I or II” denotes plasmids in which unsequenced terminal regions preclude discrimination between subtype I and II; “fused” indicates that the two so indicated plasmids in affected strains appear to be covalently joined; “∆”, indicates the presence of a large deletion relative to other cp32s, and the associated letters indicate different deletion endpoints (see Additional file 1: Fig. S4 below); “inv” indicates a large inversion is present relative to the other cp32s; and parentheses (…) enclose the one cp9 and three cp32 plasmids whose sequences were not closed. (a) rRNA IGS1 (intergenic spacer number one) nomenclature according to Travinsky et al. [88]. (b) Related chromosomal SNP (single nucleotide polymorphism) groups according to Mongodin et al. [48]. (c) Defined in text and Fig. 6a below. (d) OspC type nomenclature of Barbour and Travinsky [79]. (e) Previously named cp32-2 AND cp32-7 have the same compatibility; we use cp32-7 to represent this group. (f) These plasmids are known to have been present in the isolate before sequencing, but were lost in the culture whose DNA was used for sequencing. (g) The lp32-3 plasmid likely has the same compatibility type as cp32-3 (see text)

Fig. 2

Maps of lp28-8, lp28-9 and lp32-3. The yellow bars represent the plasmids, and the arrows within them indicate most open reading frames in these plasmids; blue arrows are putative plasmid replication/maintenence/compatibility genes, green denotes vls cassettes and red are other genes. Black “X”s on arrows indicate pseudogenes. PFams [17, 19] are indicated in the boxes above, and putative function is noted in red text above the map. An asterisk (*) marks the putative lipoprotein gene at ~9 kbp of lp28-8 that is a homolog of JD1_j07 which encodes a closely related putative lipoprotein; these are ~75% identical to plasmid-encoded proteins PGP088 and BAPKO_6042 of B. garinii PBi and B. afzelii PKo, respectively; this gene has no homolog in the B. burgdorferi type strain B31. The blue bars below note some of the best similarities to other B. burgdorferi linear plasmids

Fig. 3

Tree of cp9 PFam57 proteins. A maximum likelihood tree created by RAxML using the PROTGAMMAWAG model [61] of the B. burgdorferi cp9 PFam57 proteins is shown with bootstrap values (out of 100 trials) above the lines. A fractional distance bar is shown in the lower left, and strain names followed by the protein locus_tags are shown at the right of each branch. The two strain WI91-23 cp9 plasmids are designated #A and #B (see Additional file 1: Figure S4). A neighbor-joining tree created by ClustalX [96] also places the two WI91-23 in different major branches (not shown)

Fig. 4

Plasmid lp17 organizational subtypes. Plasmids are represented as in Fig. 2, with different background colors indicating regions of nonhomologous DNA. The organizational subtypes (see text) are indicated by Roman numerals on the right, and isolates that carry each type are indicated above the maps. Some paralogous gene families are indicated in boxes above each map [17, 19], and black bars below indicate some of the similarities to other B. burgdorferi linear plasmids

Fig. 5

Plasmid lp28-1 organizational subtypes. Plasmids are represented as described in Fig. 4, and green shading between maps connects homologous sequence sections

Fig. 6

B. Burgdorferi chromosomal right end extensions. a Aligned chromosomal right end maps drawn to scale. The different colors represent very high similarity to the indicated plasmids. The black and yellow circles indicate 263 bp and 324 bp overlapping deletions, respectively, relative to the B31 chromosome. The right-end extension names and strains that carry them (in parentheses) are indicated at the right. The thin red lines with a gray circle between the L-2 and M-2 and M-3 and M-4 chromosomes indicate a large putative deletion and inversion, respectively. b. Evolutionary tree compatible with terminal replacements and rearrangements. Major rearrangements, assumed to have occurred on the chromosome and not on plasmids before they recombined with the chromosome (see text), are denoted by the small colored circles: blue, lp28-1 subtype II addition; red, lp28-1 subtype I addition; green, lp28-5 subtype I addition; black, 263 bp deletion in lp28-1 type II sequence; orange, lp28-7 addition; gray, 7 kbp inversion; purple, 12 kbp deletion; pink, addition of lp21 sequences; and yellow, enlargement of 263 bp deletion (above) to 324 bp. The right end extension names and strains that carry them (in parentheses) are indicated at the tips of the branches, and the chromosomal SNP types (see text) are given at the right. Strains 297, Sh-2-82 and CA382 were not included in the SNP analysis but fit in the SNP groups according to the facts that, like B31 in SNP group A, CA382 has a type A ospC (A. Barbour, personal communication); 297 and Sh-2-82 are both ospC type K, and 297 has lp54 and cp26 plasmids that are most closely related to the SNP group D strains (see text) [48]

Fig. 7

Comparison of linear plasmid content of B. burgdorferi isolates. Values outside parentheses are the number of plasmid subtypes in common in the pairwise comparisons (uncertainty in some cells is the result of uncertainty of the subtype of a few linear plasmids due to incomplete terminal sequences - see legend to Fig. 1). Values in parentheses are the number of plasmids of the same PFam32 compatibility type that are different subtypes in the pairwise comparison (assuming plasmids of uncertain subtype are not the same in the strain pair being compared). Thick lines separate isolates with different chromosomal SNP types (see Fig. 6 and text). Cell color indicates relatedness of linear plasmid contents as follows: pink, number of plasmids of same subtype/number of plasmids of different subtype is ≥4; yellow, same as pink only values ≥1 and <4; blue, same a pink only values <1