In an effort to understand how a tick-borne pathogen adapts to the body louse, we sequenced and compared the genomes of the recurrent fever agents Borrelia recurrentis and B. duttonii. The 1,242,163–1,574,910-bp fragmented genomes of B. recurrentis and B. duttonii contain a unique 23-kb linear plasmid. This linear plasmid exhibits a large polyT track within the promoter region of an intact variable large protein gene and a telomere resolvase that is unique to Borrelia. The genome content is characterized by several repeat families, including antigenic lipoproteins. B. recurrentis exhibited a 20.4% genome size reduction and appeared to be a strain of B. duttonii, with a decaying genome, possibly due to the accumulation of genomic errors induced by the loss of recA and mutS. Accompanying this were increases in the number of impaired genes and a reduction in coding capacity, including surface-exposed lipoproteins and putative virulence factors. Analysis of the reconstructed ancestral sequence compared to B. duttonii and B. recurrentis was consistent with the accelerated evolution observed in B. recurrentis. Vector specialization of louse-borne pathogens responsible for major epidemics was associated with rapid genome reduction. The correlation between gene loss and increased virulence of B. recurrentis parallels that of Rickettsia prowazekii, with both species being genomic subsets of less-virulent strains.
Borreliae are vector-borne spirochetes that are responsible for Lyme disease and recurrent fevers. We completed the genome sequences of the tick-borne Borrelia duttonii and the louse-borne B. recurrentis. The former of these is responsible for emerging infections that mimic malaria in Africa and in travellers, and the latter is responsible for severe recurrent fever in poor African populations. Diagnostic tools for these pathogens remain poor with regard to sensitivity and specificity due, in part, to the lack of genomic sequences. In this study, we show that the genomic content of B. recurrentis is a subset of that of B. duttonii, the genes of which are undergoing a decay process. These phenomena are common to all louse-borne pathogens compared to their tick-borne counterparts. In B. recurrentis, this process may be due to the inactivation of genes encoding DNA repair mechanisms, implying the accumulation of errors in the genome. The increased virulence of B. recurrentis could not be traced back to specific virulence factors, illustrating the lack of correlation between the virulence of a pathogen and so-called virulence genes. Knowledge of these genomes will allow for the development of new molecular tools that provide a more-accurate, sensitive, and specific diagnosis of these emerging infections.
Citation: Lescot M, Audic S, Robert C, Nguyen TT, Blanc G, Cutler SJ, et al. (2008) The Genome of Borrelia recurrentis, the Agent of Deadly Louse-Borne Relapsing Fever, Is a Degraded Subset of Tick-Borne Borrelia duttonii. PLoS Genet 4(9): e1000185. https://doi.org/10.1371/journal.pgen.1000185
Editor: Paul M. Richardson, Progentech, United States of America
Received: April 15, 2008; Accepted: July 31, 2008; Published: September 12, 2008
Copyright: © 2008 Lescot et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: The study was funded by the participating laboratories.
Competing interests: The authors have declared that no competing interests exist.
Spirochetes of the genus Borrelia are bacterial pathogens responsible for relapsing fever and Lyme borreliosis. Whereas the Lyme disease agents Borrelia burgdorferi ,, Borrelia garinii , and Borrelia afzelii  are transmitted by hard ticks, the numerous relapsing fever borreliae are typically transmitted by soft ticks. Interestingly, tick-borne relapsing fever borreliae, including Borrelia duttonii, have shown extended vectorial capacity, whereas transmission of Borrelia recurrentis, which causes louse-borne relapsing fever, is restricted to Pediculus humanus ,. Besides their mode of transmission, these two highly related species of Borrelia exhibit very different epidemiological and clinical features. B. duttonii is endemic in Western Africa, where it demonstrates the highest incidence among all bacterial infections and causes up to six relapses, no mortality, and adverse perinatal outcomes . In contrast, B. recurrentis, once responsible for worldwide outbreaks, is currently limited to Ethiopia and its surrounding countries . It causes fewer relapses, but spontaneous mortality remains as high as 2–4% despite antibiotics, with patients suffering from distinctive hemorrhagic syndrome . In addition, women who develop relapsing fever during pregnancy have a high incidence of spontaneous abortion . Indeed, B. recurrentis and other louse-borne pathogens, including the typhus agent Rickettsia prowazekii  and the trench fever agent Bartonella quintana , exhibit higher virulence than their respective tick-borne relatives B. duttonii, Rickettsia conorii , and Bartonella henselae .
Borreliae are unique among bacteria in that their genome is comprised of a linear chromosome and both linear and circular plasmids . We sequenced the genomes of B. duttonii and B. recurrentis to gain new insights into the structure and evolution of the borreliae.
Genome Organization of B. duttonii and B. recurrentis
While the 1,242,163 bp B. recurrentis A1 strain genome contains only 8 linear fragments of 930,981-6,131 bp, the 1,575,296 bp B. duttonii Ly strain genome contains 17 linear fragments of 931,674-11,226 bp and one 27,476 bp circular fragment (Table 1, Figures S1 and S2, Genbank accession numbers CP000976-CP000992 for B. duttonii and CP000993-CP001000 for B. recurrentis). For each species, we designated the largest fragment as the chromosome and the smaller ones as the plasmids. The organization of the chromosome was conserved among borreliae, with spoOJ, gyrA, gyrB, dnaA, and dnaN (BDU_431-435, BRE_434-438) being clustered around the putative origin of replication near the GC/AT skew cross point (Figure 1 and Figure S3). In both species, the sole rrs operon (BDU_415-416, BDU_424, BRE_419-420, BRE_428), which is close to the putative origin of replication, was split by hpt, purA, and purB (BDU_418-420, BRE_422-424), as reported for B. hermsii and other relapsing fever borreliae , (Figure 1). We also found similarity between the B. duttonii-circular plasmid (cp) 27, B. duttonii-linear plasmid (lp) 26, and B. duttonii-lp28. In addition, colinearity was observed between B. duttonii-lp23/B. recurrentis-lp23, B. duttonii-lp11/B. recurrentis-lp37, B. duttonii-lp32/B. recurrentis-lp33, B. duttonii-lp(26,28,31,40–42,70)/B. recurrentis-lp(35,53), and B. duttonii-lp165/B. recurrentis-lp124 (Figure 2). The latter plasmid has no counterpart among Lyme group borreliae. In both species, the linear plasmid lp23, which is syntenic to the circular plasmids B. burgdorferi/B. garinii-cp26 and B. afzelii-cp27, was particularly interesting. This plasmid exhibited a large polyT track (174 nucleotides in B. duttonii and 46 in B. recurrentis) of a length not previously reported in other bacteria, although T-rich regions containing Ts in 16 of 20 positions and Ts in 18 of 20 positions have been reported in the ospAB and vmp promoters of B. burgdorferi and B. hermsii, respectively . This polyT track is located in the promoter of an intact variable large protein (vlp, BDU_13021, BRE_6020) gene situated at the telomere (Figure 3). This locus has been shown to be the site of vlp expression in B. recurrentis . Strikingly, this plasmid encodes the unique telomere resolvase (resT, BDU_13014, BRE_6013), a protein specific to Borrelia species (Figure 3) ,. In B. duttonii and B. recurrentis, lp23 lacks the celABC genes involved in the PTS cellobiose system as well as oppA compared to other Borrelia.
Insertion of hpt, purA, and purB is specific to the recurrent fever group borreliae. Duplication of 5S–23S rDNA is specific to the Lyme disease group borreliae. Variable spacing was observed between the Ala and Ile tRNAs. Specific degradation in the 5′ genomic region of spo0J was observed in B. recurrentis. Genes are colored according to their predicted functional category (Figure S1). Shaded areas correspond to regions of difference.
This figure was constructed using the NUCmer program from the MUMmer package. Red segments correspond to same strand matches, while blue segments correspond to opposite strand matches.
The large poly-T track in the promoter region of an intact vlp gene was specific to recurrent fever borreliae.
Comparative Chromosomal Gene Content (Table S1)
Aside from the variable number of copies of repeated genes (see below), a few genomic differences were found between B. duttonii and B. recurrentis. There was a difference in the number of the protein genes encoded by the chromosome (820 genes in B. duttonii and 800 in B. recurrentis). Five genes (recJ, putative membrane protein, rpsU, ftsK, bacA, BDU_257-262, BRE_256-260, BRE_261-265) were duplicated in B. recurrentis, with one copy of recJ (BRE_261) presenting a frameshift and one copy of bacA (BRE_260) containing a frameshift and partial deletion. Four genes, pantothenate permease (panF, BDU_821/826, BRE_824), pseudouridylate synthase (rluA, BDU_822/827, BRE_825), an uncharacterized conserved protein (BDU_823/828, BRE_826), and UDP-N-acetylmuramate-alanine ligase (murC, BDU_824/829, BRE_827), were duplicated in B. duttonii. An ATPase involved in chromosome partitioning (homolog to Soj, BDU_429), close to the replication origin was lacking in B. recurrentis.
An in-frame STOP codon (tga, replacing tgg in B. duttonii) was found in the B. recurrentis copy of recA (BDU_135, BRE_134), involved in the RecBCD dsDNA end repair pathway and the RecFOR ssDNA gap repair pathways  (Table S2A). We also found that mutS (BDU_101,BRE_100) and smf (BDU_300,BRE_304), genes belonging to the DNA processing DpRA family that collaborates with recA for recombination and bacterial transformation, were both impaired in B. recurrentis , with an in-frame STOP codon in smf (taa replacing caa) and a frameshift in mutS. Other impaired genes were found in B. recurrentis that are implicated in the following processes: maltose transport and metabolism (malX , BDU_119, BRE_118 and malQ, BDU_165, BRE_164, frameshifts), glycerol metabolism (glpA, BDU_244, BRE_243 and glpK, BDU_241, BRE_240, frameshifts), and adaptation to host environments (oppA1 transporter, BDU_329, BRE_333, internal STOP codon taa replacing caa). Other disrupted genes in B. recurrentis were yplQ (BDU_120, BRE_119, frameshift), encoding a hemolysin III, xylR2 (BDU_843, BRE_841, frameshift) of the xylose operon, the A subunit of an ATP-dependant Clp protease (BDU_364, BRE_368, frameshift), and an uncharacterized conserved protein (BDU_743, BRE_746, frameshift). Finally, a p35-like antigen (BDU_1), similar to the B. burgdorferi fibronectin-binding lipoprotein BBK32, was absent in B. recurrentis.
Gene Families in B. duttonii and B. recurrentis
A significant number of Borrelia genes corresponded to repeat families, including variable major proteins (Vmp) and Borrelia direct repeats (Bdr). Most of these were plasmid-borne paralogous families . To further study this phenomenon and compare different Borrelia species, we grouped together all predicted protein coding genes of B. duttonii, B. recurrentis, B. burgdorferi, B. garinii, and B. afzelii (see Materials and Methods). This analysis indicated that the most abundant families were those of the variable major proteins (vmp, including 600-bp vsp and 1000-bp vlp) , Borrelia direct repeats (Bdr), and plasmid partition proteins PF32, PF49, ppap1, and ppap2 (Table 2).
Most Vmps are encoded by linear plasmids, and only two and three copies were found at the beginning of the B. recurrentis and B. duttonii chromosome, respectively (Table S3). The vlp family genes, similar to VlsE in Lyme disease borreliae, encode lipoproteins that, as a result of antigenic variation, allow relapsing fever borreliae to escape the host immune response . B. duttonii encodes 68 vlp copies (19 with the consensus GGAGG of Ribosomal Binding Site), while B. recurrentis encodes 17 vlp copies (6 with the consensus GGAGG of Ribosomal Binding Site) (Table S3, Figure S4). Phylogeny clearly indicated that vlps are grouped into 4 subfamilies designated α, β, γ, and δ (Figure S4), as previously found for B. hermsii . The largest subfamily is γ, with 26 vlp copies in B. duttonii and 9 in B. recurrentis. While numerous vlp pseudogenes were found in both genomes, B. recurrentis showed a tendency to lose intact vlps, with one vlp every 18-kb (on average, excluding the chromosome) compared with one vlp every 9.5-kb for B. duttonii. We identified remnants of 46 vlp genes in B.duttonii and 29 in B. recurrentis. The vsp family genes are related to the lipoprotein ospC present in Lyme disease borreliae. We identified 14 vsp in B. duttonii and 10 in B. recurrentis. The ratio of intact vlp to vsp was 17/10 (1.7) in B. recurrentis and 68/14 (4.9) in B. duttonii.
The Bdr family is common to relapsing fever and Lyme disease group borreliae . In B. burgdorferi, Bdr are characterized by temperature-independent, low expression level, inner membrane-localized immunogenic proteins that are organized into 6 families (A to F). Bdr genes are found on most plasmids, except for the large B. duttonii-lp165/B. recurrentis-lp124 plasmid, which was also devoid of vlp and vsp.
In B. duttonii, putative replication and partition genes were identified on most plasmids, and were usually organized as a set of the four consecutive genes: PF32, PF49, ppap1, ppap2 (ORFe in B. burgdorferi) . In B. recurrentis, this organization was still apparent despite gene decay.
The Bmp family contains basic membrane protein genes encoding lipoproteins. These proteins are expressed in infected patients, and result from different gene rearrangements in the five borreliae (Figure S5). For instance, the protein BmpB-1 is present only in Lyme group borreliae and could thus be used as a Lyme-specific diagnostic test.
An abundant repeat family (Family 44, 14 members, Table 2) was found in B. duttonii, but not in B. recurrentis. Indeed, members of this family are located at the 5′-end of the B. duttonii-lp164 plasmid, a region that lacks a counterpart in B. recurrentis. It contains uncharacterized conserved lipoproteins that are predicted to represent 7.6% of the lipoproteins in B. duttonii.
Comparison with the Lyme Disease Group Borrelia
Genome sequencing of B. recurrentis and B. duttonii provides the opportunity to compare the gene content between relapsing fever and Lyme disease group borreliae. Whole chromosome comparison (Figure S1) shows extensive conservation of gene content and gene order. In both groups, we found an intact RecBCD system, which is important for repairing double-stranded DNA ends, but a deficient RecFOR pathway. RecF and RecR proteins are associated with RecO in the reparation of single-stranded DNA; however, RecO is absent in all borreliae, potentially leading to deficient repair of single-stranded nicks. We observed only 13 genes specific to the Lyme disease group and 17 genes specific to the relapsing fever group (excluding bmp genes, Table S2B) in the chromosomes of borreliae.
As previously observed in B. hermsii ,, chromosome-encoded genes involved in purine metabolism and salvage were similarly found in these relapsing fever borreliae, including adenylosuccinate synthase (purA, BDU_419, BRE_423), adenylosuccinate lyase (purB, BDU_420, BRE_424), and hypoxanthine phosphoribosyltransferase (hpt, BDU_422, BRE_425). They were located between the 16S and 23S ribosomal DNA. Other genes unique to the relapsing fever group borreliae included a putative adenine-specific DNA methyltransferase (BDU_467, BRE_470), a copper homeostasis protein (cutC, BDU_844, BRE_842), the sugar specific PTS family protein (nagE, BDU_838,BRE_836), a trypsin-like serine protease (BDU_797, BRE_800), an ATP-dependent helicase belonging to the DinG family (BDU_740, BRE_743), a TPR domain containing protein (BDU_737, BRE_740), a protein with similarity to a response regulator receiver (CheY) modulated serine phosphatase (BDU_523, BRE_526), glpQ (BDU_243, BRE_242), glpT (BDU_241, BRE_240), maf protein (BDU_127, BRE_126), hsp20 heat shock protein (BDU_444, BRE_447), purine salvage pathway genes including peptidyl-prolyl cis-trans isomerase (BDU_407, BRE_411), and the rec family members RecN (BDU_313, BRE_317), RecF (BDU_436, BRE_439), and RecR (BDU_465, BRE_468). Likewise, arcC (Carbamate kinase, BDU_857, BRE_855), which is involved in glutamate, arginine and proline biosynthesis are specific to relapsing fever borreliae, but was impaired in B. recurrentis. Among these genes, 16 exhibited best homologs with sequences outside of the spirochetes group. Interestingly, 5 demonstrated good homology with Fusobacterium nucleatum, as described for another spirochete, Treponema denticola .
Conversely, some genes were only found on the Lyme disease group (Table S2B), including a putative L-sorbosone dehydrogenase, two antigens S2, an oligopeptide ABC transporter (oppA-3), a methylglyoxal synthase, a lipoprotein LA7, a basic membrane protein B (bmpB-1), an inositol monophosphatase, an aldose reductase, a MATE efflux family protein, a pfs protein (pfs-2), a rep helicase, a small primase-like protein, and an Na+/H+ antiporter (nhaC-1).
In contrast to what was observed for the chromosome, the plasmid contents of the relapsing fever group were very different from that of the Lyme disease group. Only three B. duttonii plasmids (lp165, lp70 and lp23) exhibited significant synteny with B. burgdorferi plasmids (Figure S6). B. duttonii-lp165 and B. recurrentis-lp124 encoded nrdF (ribonucleoside-diphosphate reductase beta subunit, BDU_1075, BRE_1045), nrdE (ribonucleoside-diphosphate reductase alpha subunit, BDU_1076, BRE_1046), and nrdI (auxiliary protein, BDU_1077, BRE_1047) (Table S2B), all of which were previously reported in B. hermsii , but were absent in the Lyme disease group of Borrelia. Using the SpLip program  with the B. burgdorferi matrix supplied by the authors, we retrieved 171 probable and 13 possible lipoproteins in B. duttonii, 80 (11) in B. recurrentis, 111 (9) in B. burgdorferi, 45 (8) in B. garinii, and 84 (10) in B. afzelii. Relapsing fever borreliae proteomes contain a larger fraction of lipoprotein (13.63% in B. duttonii and 8.72% in B. recurrentis) than Lyme disease group borreliae (7.74% in B. afzelii, 7.32% in B. burgdorferi and 5.9% in B. garinii).
B. duttonii contained no impaired genes in its chromosome (except for two vlp pseudogenes), whereas B. recurrentis exhibits 20 impaired genes (Table S2A). This suggests that B. recurrentis evolved under more relaxed constraints (e.g. accumulated more deleterious mutations) than B. duttonii. This hypothesis was examined by analyzing the ratio of non-synonymous (Ka) to synonymous (Ks) substitution rates (denoted ω = Ka/Ks) among 773 conserved genes of the five borreliae. Based on the most suitable model of evolution (See Materials and Methods), the estimated ω ratio was nearly twice as high for the B. recurrentis branch (ωBre = 0.18) than for the B duttonii branch (ωBdu = 0.10). These results suggest that, on average, the genome of B. recurrentis tends to evolve under weaker coding sequence constraints than the genome of B. duttonii. In addition, the number of non-synonymous substitutions was higher in the B. recurrentis branch (n = 695) than in the B. duttonii branch (n = 366). This indicates that B. recurrentis proteins tend to diverge faster. To find out whether this acceleration was restricted to a specific subset of genes, we further analyzed sub-alignments comprising, on average, 10 genes. This analysis showed that ωBre calculated for the sub-alignments were not systematically higher than ωBdu (Figure 4A). This suggests that the selective constraints acting on coding sequences are, in general, not less effective in B. recurrentis than in B. duttonii. In contrast, the Ka and Ks values were almost systematically higher for B. recurrentis (Figure 4B and C). These results indicate that B. recurrentis genome is globally evolving faster that the one of B. duttonii.
Seventy-seven 2190-codon alignments derived from the initial concatenated alignment of the borrelia core set were analyzed using model 2. Only values obtained for the B. recurrentis and B. duttonii branches are presented. The dot plots show ω = Ka/ks (A), Ka (B), and Ks (C) values.
The Linear, Fragmented Genome of Borrelia
While circular chromosomes are most commonly seen in bacteria, linear chromosomes are encountered in some phylogenetically distinct species including Agrobacterium tumefaciens ,, Streptomyces species ,, and Borrelia species –. The latter are unique in that they harbor >3 linear genomic fragments, whereas the other sequenced spirochetes, Treponema , and Leptospira –, possess 1–2 circular chromosomes. This suggests that genome linearization is a recent evolutionary event in the spirochete lineage. Genome linearization of Borrelia is sustained by telomeres, terminal small inverted repeats with covalently closed hairpin ends ,. Similar features have been described for Poxvirus, African swine fever virus, Chlorella viruses, the mtDNA of yeasts and protozoa, and the Escherichia coli phage N15 –. Replication of telomeres from a bidirectional origin , produces intermediates for which the replicated telomeres comprise dimer junctions between inverted repeats of the original plasmid . Replicated telomeres are then processed by ResT, the essential B. burgdorferi cp26-encoded telomere resolvase responsible for a particular DNA breakage and reunion event that regenerates the hairpin telomeres ,,. When cp26 was deleted in B. burgdorferi cells, viability was lost . ResT acts via a catalytic mechanism analogous to that of tyrosine recombinases and type IB topoisomerases . We found ResT in relapsing fever Borrelia, in agreement with the concept of telomere-mediated genome linearization among these organisms. ResT was recently also shown to perform a reverse reaction that fuses telomeres from unrelated replicons. In the Lyme disease group, initiation of replication occurs in the central region of the linear chromosome that comprises a polar CG skew and proceeds bidirectionnaly ,. The observed parallel genome architecture suggests an identical replication mechanism among the relapsing fever group.
B. recurrentis, a Decaying Strain of B. duttonii
Previous limited phylogenetic data based on 16S rDNA  and 16S–23S intergenic spacer  raised the question of whether B. duttonii and B. recurrentis are different species . Gene content analysis showed that the genome of B. recurrentis is a subset of that of B. duttonii. The chromosomes of both species were found to be almost entirely colinear, and all B. recurrentis plasmids have a counterpart in B. duttonii. Altogether, 30 genes or gene families of B. duttonii were either absent, split, or reduced in number in B. recurrentis. In particular, a set of four consecutive genes, PF32, PF49, ppap1, and ppap2, involved in plasmid replication and partitioning were well conserved in most B. duttonii plasmids, but were damaged considerably in B. recurrentis plasmids. This suggests ongoing plasmid loss in B. recurrentis. Likewise, B. recurrentis lacks a chromosomal Soj homologue, which is involved in chromosome partitioning. Such reductive evolution may be linked to defective DNA repair in B. recurrentis. Indeed, the B. recurrentis recA gene sequence presents an in-frame STOP codon. Although compensatory mechanisms that preserve the expression of recA could not be ruled out, this finding was surprising, as recA is a ubiquitous and highly conserved gene involved in DNA repair . Impaired recA was previously reported in Spiroplasma melliferum , whereas Buchnera and Blochmania floridanus lack this gene ,. In Escherichia coli, 50% of recA mutants are viable and avoid chromosome lesions , but recA dut* (dUTPase) mutants are lethal in the presence of nfi, which encodes endonuclease V (deoxyinosine 3′ endonuclease) . Since Borrelia species lack dut, we hypothesize that the viability of B. recurrentis is maintained by the absence of nfi, as occurs in B. burgdorferi, B. garinii, and B. duttonii. We were unable to find either an ATP-dependant LigD or the DNA-end-binding-protein, Ku, involved in DNA repair by non-homologous end-joining . The lack of an intact recA and smf in B. recurrentis may explain the observed accelerated evolution of its genome compared to B. duttonii. Taken together, the genomic data and phylogenetic data suggest that B. recurrentis is actually a strain of B. duttonii.
Adaptation of Pathogens to the Body Louse Vector
Genome comparison of louse-borne bacteria with their tick-borne counterparts indicated an extensive genome size reduction of 20.4% for Borrelia spp., 18% for Bartonella spp., and 12.6% for Rickettsia spp. Among borreliae, genes that were lost included the antigenic lipoproteins vlp and vsp, genes involved in chromosome and plasmid partitioning, and genes involved in xylose and glycerate metabolism. Degradation of genes into pseudogenes within louse-borne species (128 B. henselae / 175 B. quintana; 2 B. duttonii / 20 B. recurrentis, Table S2A) suggests a progression toward the complete loss of these genes. Indeed, louse-borne species contain 21%–39% less CDSs than their tick-borne counterpart. This phenomenon is illustrated by the decreased number of repeat families from 43 in B. henselae to 11 in B. quintana , from 12 in R. conorii  to 3 in R. prowazekii , and from 54 in B. duttonii to 17 in B. recurrentis. Loss of DNA repair genes such as mutM and mutT in the typhus group R. prowazekii , and recA, mutS, and smf in B. recurrentis may contribute to a higher rate of replication error, leading to faster genome decay among these louse-borne pathogens. Genomic differences between louse-borne species and their tick-borne counterparts may correlate with their concomitant adaptation to a human host . A 4-nucleotide difference (0.26%) in the 16S rDNA sequence of B. duttonii and B. recurrentis estimates their divergence to have occurred between 6.5 and 13 million years ago . This is roughly the same as the time of the divergence of the human specific louse vector of B. recurrentis and the common ancestral primate-associated ectoparasite . We hypothesize that genome decay in louse-borne bacteria correlates with the host-specific bottleneck of the arthropod vector. Conversely, tick-transmitted organisms may adapt to diverse host populations, which is facilitated by tick feeding habits, unlike louse-borne pathogens. Such adaptation to body louse transmission is correlated with increased evolutionary rates illustrated in B. recurrentis analogous to those observed for R. prowazekii . Genome size reduction and on-going gene and function decay in louse-borne pathogens illustrate the genomic fluidity associated with adaptation of bacteria from a large environmental niche to a more restricted one ,.
Antigenic Variability and Virulence Factors
Variation in the expression of a dominant surface antigen allows borreliae to evade immune defences. This evasion increases the duration and number of recurrences of bacteremia, and thus, the likelihood of subsequent transmission . In B. recurrentis strain A1, Vlp has been shown to be the major pro-inflammatory molecule . Furthermore, expression of certain lipoproteins, for instance in Borrelia turicatae, has been shown to modulate tissue tropism. Specifically, the Bt1 and Bt2 variants are predictive of either neurotropism or spirochetemia and arthritis, respectively ,. Detailed molecular analyses revealed that the corresponding genes are arranged into silent and expressed copies on different plasmids ,. Indeed, two copies of vlp1B. recurrentis A1 were found in B. recurrentis . This gene was identified as a pseudogene in lp53 and as an active gene in lp23 (lp23_20295_21386, BRE_6020). Antigenic variation occurs either by replacing the entire open reading frame of the expressed gene with a previously silent one, or by activating a previously silent downstream gene . The likelihood of different antigenic variants being expressed appears not to be random, but is ordered in a semi-hierarchical fashion. This hierarchy depends on the sequence similarity between the upstream homology sequence located at the expression site of the variant gene and the distance separating the extragenic downstream homology sequence . To date, the absence of suitable animal models has precluded antigenic variation studies among B. recurrentis and B. duttonii; however, the genome sequence data reported here could facilitate the molecular characterization of antigenic variants in clinical samples.
In contrast to Lyme disease spirochetes (<105/ml), relapsing-fever spirochetes achieve high cell densities (>108/ml) in patients' blood, suggesting differences in the ability of both groups to either exploit or survive in blood. It has been hypothesized that the purine salvage pathways are among these differences . In particular, hypoxanthine, a primary product of purine catabolism, is exported to the outer surface of red blood cells. This could facilitate the direct uptake of hypoxanthine from red blood cells, providing a purine source for the synthesis of nucleotides by these borreliae . In addition, some researchers have suggested that differences in glycerol-3-phosphate (G3P), an important metabolic intermediate for phospholipid synthesis, acquisition pathways contribute to differences in the density of borreliae in blood . B. recurrentis has apparently inactivated glpA and glpK, indicating that two of the three G3P acquisition pathways in Borrelia have been turned-off in B. recurrentis. B. recurrentis could acquire G3P only by the hydrolysis of deacylated phospholipids from the erythrocyte membrane, in agreement with the fact that its body louse vector takes daily bloody meal in order to survive. Therefore, such a restriction would not be deleterious to B. recurrentis, but indeed exemplifies adaptation to a specific ecological niche . As GlpQ is an immunodominant antigen used to discriminate between Lyme disease and relapsing fever groups , the present genomic data may help refine the serological diagnosis of relapsing fever group borrelioses.
Genome analysis revealed that B. recurrentis encodes fewer putative virulence factors than B. duttonii, an unexpected finding given the high mortality in untreated louse-borne relapsing fever . In particular, B. recurrentis encodes a reduced proportion of major antigenic Vlp compared to Vsp lipoproteins than B. duttonii. It also lacks a hemolysin, which is present but is obviously degradated, as well as a p35-like antigen similar to the BBK32 fibronectin-binding lipoprotein of B. burgdorferi. Loss of intact glpA and glpK in B. recurrentis may limit the acquisition of glycerol-3-phosphate. It is also possible that the loss of one intact copy of bacA in B. recurrentis may cause increased virulence, as observed for Brucella abortus, in which bacA is deleted . Other genes that are critical for the environmental survival of B. recurrentis, including the broad-spectrum peptide permease OppA-1 gene  and the ClpA chaperone, were also degraded. The ClpA chaperone prepares protein substrates for degradation by ClpP , a central complex that controls the stability and activity of transcriptional regulators during cell stress Impaired ClpA may deregulate transcription during B. recurrentis infection and lead to uncontrolled expression of virulence factors. Altogether, these defects may impair environmental sensing by B. recurrentis. These findings illustrate the lack of correlation between the observed virulence and the number of virulence factors possessed by an organism . Finally, B. recurrentis illustrates the emerging concept that microbial virulence, for humans, may result from gene loss .
Materials and Methods
Isolation of Strains and Growth Conditions
B. recurrentis strain A1 isolated from an adult patient with louse-borne relapsing fever in Ethiopia  and B. duttonii strain Ly isolated from a 2-year-old girl with tick-borne relapsing fever in Tanzania  were grown on BSK-H complete medium batch number 057K4413 and 10K8402 (Sigma) at 37°C. Pulsed field gel electrophoresis (PFGE) was performed (CHEF-DRIII apparatus, Biorad) to determine the size of the genome and to analyze plasmid patterns under three different electrophoretic conditions. The samples were prepared as described previously . Small plasmids could be visualized using a linear increase in pulse times between 1 to 3 sec. at 180 V over a 10 h period. Plasmids from 145 to 23 kb were detected using a linear increase in pulse time between 3 to 10 sec. at 180 V over a 15 h period, followed by an extensive migration using a linear increase in pulse time between 50 to 150 sec. at 180 V over a 30 h period (Figure S7).
Shotgun Sequencing of B. duttonii and B. recurrentis Genomes and Sequencing Strategy
As attempts to isolate chromosome and plasmid DNA from PFEG after β-agarase treatment failed to produce sufficient DNA yield, genomic DNA was extracted from 25 ml of culture by incubation with 1% SDS-RNAseI (50 µg/ml) for 3 hours at 37°C, followed by proteinase K digestion (250 µg/ml) at 37°C overnight. After 3 phenol extractions, the DNA was precipitated with ethanol. The quality, yield, and DNA concentration were estimated by electrophoresis on agarose gels stained with ethidium bromide. Genomic DNA was sheared by mechanical fragmentation with a Hydroshear device (GeneMachines, San Carlos, California, USA) to construct plasmid libraries. After blunt end repair and BstXI adapter ligation, fragments of 2 kb, 5 kb, and 10 kb were cloned into the high copy number vector pCDNA2.1 (Invitrogen, Life Technologies) digested with BstXI. Transformations were performed using the electrocompetent E. coli strain DH10B (Invitrogen, Life Technologies). Each library was validated using 96 clones from which the insert size was estimated by agarose gel electrophoresis. Sequencing using vector-based primers was carried out using the ABI 3730 Applera sequencer. For B. duttonii, only libraries of 2 kb and 10 kb were sequenced, producing 14,719 and 10,066 reads, respectively. For B. recurrentis, three shotgun libraries of 2 kb, 5 kb, and 10 kb generated 14,794, 2,248, and 2,042 reads, respectively. Reads were analyzed and assembled into contigs using the Phred, Phrap, and Consed software packages –. Finishing was performed to verify low quality regions, to fill-in sequences by DNA walking using subcloned DNA, and to close gaps. A total of 1,034 B. duttonii specific primers and 784 B. recurrentis primers were designed. All finishing sequencing reactions were carried out on an ABI 3130 Applera sequencer.
Annotation of Borrelia recurrentis and Borrelia duttonii Sequences
An initial set of protein-coding genes was detected using self-training Markov models  and careful examination of intergenic regions to rescue additional genes. Putative protein coding genes were then validated and annotated by sequence similarity using BlastP  against the non-redundant protein database from the National Center for Biotechnology Information (NCBI) and the KEGG protein database . Putative protein coding genes were also validated by profile detection using RPSblast  and the COG database . Genes encoding tRNA were identified with tRNAscan-SE , and other RNAs were located using BlastN . Dot plots of plasmids from both species were computed using the NUCmer program from the MUMmer package .
To compare the distribution of genes in different Borrelia families, we grouped together all predicted protein coding genes for B. duttonii (this work), B. recurrentis (this work), B. burgdorferi (GenBank: NC_000948-57, NC_001318, NC_001849-57, NC_001903, NC_001904), B. garinii (GenBank: NC_006128, NC_006129, NC_006156), and B. afzelii (GenBank: NC_008273, NC_008274, NC_008277, NC_008564-69), by performing a mutual BlastP comparison of this set of genes. The resulting comparison data were submitted to a Markov Chain Clustering algorithm to regroup the genes into families . The resulting set of clustered sequences is available as Dataset S1. The same analysis was performed on the individual proteome of B. henselae, B. quintana, R. prowazekii, R. conorii, B. duttonii, and B. recurrentis to count the number of repeat families containing at least 3 members in each of these genomes.
Lipoprotein computational prediction has been the subject of a specific article  that describes the SpLip program used in the present work.
Analysis of Borrelia Evolution
The 856 proteins of the B. burgdorferi chromosome were aligned with the other Borrelia (B. duttonii, B. recurrentis, B. garinii and B. afzelii) proteomes using the BlastP program (e-value<1e-10) . We identified 773 genes that were conserved in all borreliae (borreliae core genes) using the reciprocal best Blast hit criterion. The 773 Borrelia core proteins were first aligned individually using MUSCLE . Poorly aligned regions were discarded by GBLOCKS . The resulting alignments were used as a guide to align the corresponding coding sequences on a codon basis. After cleaning up the nucleotide alignments for poorly aligned regions, the 773 multiple alignments were concatenated in a single alignment of 169,249 codons. Estimation of the ω = Ka/Ks ratio was performed using the maximum likelihood method implemented in the CODEML program . The ω ratio measures the magnitude and direction of selective pressure on coding sequence, with ω = 1, <1, and >1 indicating neutral evolution, purifying selection, and positive diversifying selection, respectively. To examine whether the ω ratio varied between the B. recurrentis and B. duttonii branches, we fitted two different models: the first model considered a single ω ratio for the 2 branches of B. recurrentis and B. duttonii (ωBre-Bdu) and a background ω ratio (ω0) averaged over the remaining branches of the borrelia phylogeny. In the second model, a specific ω ratio was considered for each of the B. recurrentis and B. duttonii branches (ωBre and ωBdu, respectively) as well as a background ω0 ratio common to the remaining branches. To determine which of the two nested models best fit the data, we compared their likelihoods using the Likelihood Ratio Test (LRT)(Table S4). The likelihood statistics – i.e. twice the log likelihood difference between the 2 models (2δlnL), can be compared to the chi square distribution with a degree of freedom equal to the difference of the number of free parameters in the two models (ddf = 1 in our analysis). The LRT test (2δlnL = 6.0) indicated that model 2 better fits the data than model 1. However, the likelihood difference between the two models is only borderline significant (P = 0.014).
Whole chromosome display of sequenced borreliae, including the recurrent fever group B. duttonii and B. recurrentis and the Lyme disease group B. burgdorferi, B. garinii, and B. afzelii. Genes are colored according to their predicted functional category. Highlighted areas correspond to regions of difference.
(9.45 MB PDF)
B. duttonii and B. recurrentis plasmids. The large B. duttonii-lp165 and B. recurrentis-lp124 plasmids, which demonstrate extensive similarity, are shown side by side, with shaded areas indicating regions of difference. Genes are colored according to their repeat-family membership (Table 2).
(1.78 MB PDF)
GC and AT skews of B. recurrentis and B. duttonii chromosomes showing reversal near the origin of replication.
(0.05 MB PDF)
(0.40 MB PDF)
Comparison of the Bmp gene family in five borreliae genomes indicates structural rearrangements in Lyme disease group borreliae. Genes are colored according to predicted functional category (Figure S1).
(0.15 MB PDF)
Dot plot showing the extensive similarity between B. duttonii and B. burgdorferi plasmids. This figure was constructed using the PROmer program from the MUMmer package. Red segments correspond to same strand matches, while blue segments correspond to opposite strand matches.
(0.07 MB PDF)
Pulse field gel electrophoresis images of B. duttonii and B. recurrentis.
(0.15 MB PDF)
List of genes which are either absent, split, or in reduced number in B. recurrentis when compared to B. duttonii.
(0.03 MB DOC)
A. Split and truncated genes on the Borrelia chromosome. B. List of genes unconserved between the five borreliae.
(0.08 MB DOC)
List of the different variable large proteins in the B. duttonii and B. recurrentis genomes. A. B. recurrentis; B. B. duttonii; C. Repartition of the Vlp genes among different classes in the two borreliae.
(0.15 MB DOC)
Parameters of the codon models used in this study.
(0.02 MB DOC)
Conceived and designed the experiments: DR. Performed the experiments: CR TTN PW AC. Analyzed the data: ML SA GB SJC JMC DR MD. Wrote the paper: ML SA GB SJC JMC DR MD. Performed the bioinformatic analyses: ML SA GB. Performed genome sequencing: PW AC.
- 1. Fraser CM, Casjens S, Huang WM, Sutton GG, Clayton R, et al. (1997) Genomic sequence of a Lyme disease spirochaete, Borrelia burgdorferi. Nature 390: 580–586.
- 2. Casjens S, Palmer N, van Vugt R, Huang WM, Stevenson B, et al. (2000) A bacterial genome in flux: the twelve linear and nine circular extrachromosomal DNAs in an infectious isolate of the Lyme disease spirochete Borrelia burgdorferi. Mol Microbiol 35: 490–516.
- 3. Glockner G, Lehmann R, Romualdi A, Pradella S, Schulte-Spechtel U, et al. (2004) Comparative analysis of the Borrelia garinii genome. Nucleic Acids Res 32: 6038–6046.
- 4. Glockner G, Schulte-Spechtel U, Schilhabel M, Felder M, Suhnel J, et al. (2006) Comparative genome analysis: selection pressure on the Borrelia vls cassettes is essential for infectivity. BMC Genomics 7: 211.
- 5. Scott J, Wright D, Cutler S (2005) Typing African relapsing fever spirochetes. Emerg Infect Dis 11: 1722–1729.
- 6. Ras NM, Lascola B, Postic D, Cutler SJ, Rodhain F, et al. (1996) Phylogenesis of relapsing fever Borrelia spp. Int J Syst Bacteriol 46: 859–865.
- 7. Vial L, Diatta G, Tall A, Ba el H, Bouganali H, et al. (2006) Incidence of tick-borne relapsing fever in west Africa: longitudinal study. Lancet 368: 37–43.
- 8. Raoult D, Roux V (1999) The body louse as a vector of reemerging human diseases. Clin Infect Dis 29: 888–911.
- 9. Southern PM, Sandford JP (1969) Relapsing fever: a clinical and microbiological review. Med 48: 129–143.
- 10. Bryceson AD, Parry EH, Perine PL, Warrell DA, Vukotich D, et al. (1970) Louse-borne relapsing fever. Q J Med 39: 129–170.
- 11. Andersson SG, Zomorodipour A, Andersson JO, Sicheritz-Ponten T, Alsmark UC, et al. (1998) The genome sequence of Rickettsia prowazekii and the origin of mitochondria. Nature 396: 133–140.
- 12. Alsmark CM, Frank AC, Karlberg EO, Legault BA, Ardell DH, et al. (2004) The louse-borne human pathogen Bartonella quintana is a genomic derivative of the zoonotic agent Bartonella henselae. Proc Natl Acad Sci U S A 101: 9716–9721.
- 13. Ogata H, Audic S, Renesto-Audiffren P, Fournier PE, Barbe V, et al. (2001) Mechanisms of evolution in Rickettsia conorii and R. prowazekii. Science 293: 2093–2098.
- 14. Barbour AG (1993) Linear DNA of Borrelia species and antigenic variation. Trends Microbiol 1: 236–239.
- 15. Barbour AG, Putteet-Driver AD, Bunikis J (2005) Horizontally acquired genes for purine salvage in Borrelia spp. causing relapsing fever. Infect Immun 73: 6165–6168.
- 16. Pettersson J, Schrumpf ME, Raffel SJ, Porcella SF, Guyard C, et al. (2007) Purine salvage pathways among Borrelia species. Infect Immun 75: 3877–3884.
- 17. Sohaskey CD, Zuckert WR, Barbour AG (1999) The extended promoters for two outer membrane lipoprotein genes of Borrelia spp. uniquely include a T-rich region. Mol Microbiol 33: 41–51.
- 18. Vidal V, Cutler S, Scragg IG, Wright DJ, Kwiatkowski D (2002) Characterisation of silent and active genes for a variable large protein of Borrelia recurrentis. BMC Infect Dis 2: 25.
- 19. Chaconas G, Stewart PE, Tilly K, Bono JL, Rosa P (2001) Telomere resolution in the Lyme disease spirochete. Embo J 20: 3229–3237.
- 20. Kobryn K, Chaconas G (2002) ResT, a telomere resolvase encoded by the Lyme disease spirochete. Mol Cell 9: 195–201.
- 21. Rocha EP, Cornet E, Michel B (2005) Comparative and evolutionary analysis of the bacterial homologous recombination systems. PLoS Genet 1: e15.
- 22. Mortier-Barriere I, Velten M, Dupaigne P, Mirouze N, Pietrement O, et al. (2007) A key presynaptic role in transformation for a widespread bacterial protein: DprA conveys incoming ssDNA to RecA. Cell 130: 824–836.
- 23. Hinnebusch BJ, Barbour AG, Restrepo BI, Schwan TG (1998) Population structure of the relapsing fever spirochete Borrelia hermsii as indicated by polymorphism of two multigene families that encode immunogenic outer surface lipoproteins. Infect Immun 66: 432–440.
- 24. Dai Q, Restrepo BI, Porcella SF, Raffel SJ, Schwan TG, et al. (2006) Antigenic variation by Borrelia hermsii occurs through recombination between extragenic repetitive elements on linear plasmids. Mol Microbiol 60: 1329–1343.
- 25. Roberts DM, Carlyon JA, Theisen M, Marconi RT (2000) The bdr gene families of the Lyme disease and relapsing fever spirochetes: potential influence on biology, pathogenesis, and evolution. Emerg Infect Dis 6: 110–122.
- 26. Seshadri R, Myers GS, Tettelin H, Eisen JA, Heidelberg JF, et al. (2004) Comparison of the genome of the oral pathogen Treponema denticola with other spirochete genomes. Proc Natl Acad Sci U S A 101: 5646–5651.
- 27. Zhong J, Skouloubris S, Dai Q, Myllykallio H, Barbour AG (2006) Function and evolution of plasmid-borne genes for pyrimidine biosynthesis in Borrelia spp. J Bacteriol 188: 909–918.
- 28. Setubal JC, Reis M, Matsunaga J, Haake DA (2006) Lipoprotein computational prediction in spirochaetal genomes. Microbiology 152: 113–121.
- 29. Goodner B, Hinkle G, Gattung S, Miller N, Blanchard M, et al. (2001) Genome sequence of the plant pathogen and biotechnology agent Agrobacterium tumefaciens C58. Science 294: 2323–2328.
- 30. Wood DW, Setubal JC, Kaul R, Monks DE, Kitajima JP, et al. (2001) The genome of the natural genetic engineer Agrobacterium tumefaciens C58. Science 294: 2317–2323.
- 31. Ikeda H, Ishikawa J, Hanamoto A, Shinose M, Kikuchi H, et al. (2003) Complete genome sequence and comparative analysis of the industrial microorganism Streptomyces avermitilis. Nat Biotechnol 21: 526–531.
- 32. Heuts DP, van Hellemond EW, Janssen DB, Fraaije MW (2007) Discovery, characterization, and kinetic analysis of an alditol oxidase from Streptomyces coelicolor. J Biol Chem 282: 20283–20291.
- 33. Fraser CM, Norris SJ, Weinstock GM, White O, Sutton GG, et al. (1998) Complete genome sequence of Treponema pallidum, the syphilis spirochete. Science 281: 375–388.
- 34. Ren SX, Fu G, Jiang XG, Zeng R, Miao YG, et al. (2003) Unique physiological and pathogenic features of Leptospira interrogans revealed by whole-genome sequencing. Nature 422: 888–893.
- 35. Bulach DM, Zuerner RL, Wilson P, Seemann T, McGrath A, et al. (2006) Genome reduction in Leptospira borgpetersenii reflects limited transmission potential. Proc Natl Acad Sci U S A 103: 14560–14565.
- 36. Picardeau M, Bulach DM, Bouchier C, Zuerner RL, Zidane N, et al. (2008) Genome Sequence of the Saprophyte Leptospira biflexa Provides Insights into the Evolution of Leptospira and the Pathogenesis of Leptospirosis. PLoS ONE 3: e1607.
- 37. Hinnebusch J, Tilly K (1993) Linear plasmids and chromosomes in bacteria. Mol Microbiol 10: 917–922.
- 38. Casjens S (1999) Evolution of the linear DNA replicons of the Borrelia spirochetes. Curr Opin Microbiol 2: 529–534.
- 39. Nosek J, Kosa P, Tomaska L (2006) On the origin of telomeres: a glimpse at the pre-telomerase world. Bioessays 28: 182–190.
- 40. Picardeau M, Lobry JR, Hinnebusch BJ (1999) Physical mapping of an origin of bidirectional replication at the centre of the Borrelia burgdorferi linear chromosome. Mol Microbiol 32: 437–445.
- 41. Beaurepaire C, Chaconas G (2005) Mapping of essential replication functions of the linear plasmid lp17 of B. burgdorferi by targeted deletion walking. Mol Microbiol 57: 132–142.
- 42. Byram R, Stewart PE, Rosa P (2004) The essential nature of the ubiquitous 26-kilobase circular replicon of Borrelia burgdorferi. J Bacteriol 186: 3561–3569.
- 43. Tourand Y, Lee L, Chaconas G (2007) Telomere resolution by Borrelia burgdorferi rest through the collaborative efforts of tethered DNA binding domains. Mol Microbiol 64: 580–590.
- 44. Jewett MW, Byram R, Bestor A, Tilly K, Lawrence K, et al. (2007) Genetic basis for retention of a critical virulence plasmid of Borrelia burgdorferi. Mol Microbiol 66: 975–990.
- 45. Bankhead T, Chaconas G (2004) Mixing active-site components: a recipe for the unique enzymatic activity of a telomere resolvase. Proc Natl Acad Sci U S A 101: 13768–13773.
- 46. Picardeau M, Lobry JR, Hinnebusch BJ (2000) Analyzing DNA strand compositional asymmetry to identify candidate replication origins of Borrelia burgdorferi linear and circular plasmids. Genome Res 10: 1594–1604.
- 47. Cutler SJ, Scott JC, Wright DJM (2008) Phylogenetic origins of Borrelia recurrentis. Int J Med Microbiol. S1438-4221(07)00197-X.
- 48. Marais A, Bove JM, Renaudin J (1996) Characterization of the recA gene regions of Spiroplasma citri and Spiroplasma melliferum. J Bacteriol 178: 7003–7009.
- 49. Gil R, Silva FJ, Zientz E, Delmotte F, Gonzalez-Candelas F, et al. (2003) The genome sequence of Blochmannia floridanus: comparative analysis of reduced genomes. Proc Natl Acad Sci U S A 100: 9388–9393.
- 50. Klasson L, Andersson SG (2006) Strong asymmetric mutation bias in endosymbiont genomes coincide with loss of genes for replication restart pathways. Mol Biol Evol 23: 1031–1039.
- 51. Bradshaw JS, Kuzminov A (2003) RdgB acts to avoid chromosome fragmentation in Escherichia coli. Mol Microbiol 48: 1711–1725.
- 52. Kouzminova EA, Kuzminov A (2004) Chromosomal fragmentation in dUTPase-deficient mutants of Escherichia coli and its recombinational repair. Mol Microbiol 51: 1279–1295.
- 53. Shuman S, Glickman MS (2007) Bacterial DNA repair by non-homologous end joining. Nat Rev Microbiol 5: 852–861.
- 54. Blanc G, Ogata H, Robert C, Audic S, Suhre K, et al. (2007) Reductive Genome Evolution from the Mother of Rickettsia. PLoS Genet 3: e14.
- 55. Ochman H, Elwyn S, Moran NA (1999) Calibrating bacterial evolution. Proc Natl Acad Sci U S A 96: 12638–12643.
- 56. Reed DL, Light JE, Allen JM, Kirchman JJ (2007) Pair of lice lost or parasites regained: the evolutionary history of anthropoid primate lice. BMC Biol 5: 7.
- 57. Ahmed N, Dobrindt U, Hacker J, Hasnain SE (2008) Genomic fluidity and pathogenic bacteria: applications in diagnostics, epidemiology and intervention. Nat Rev Microbiol 6: 387–394.
- 58. Pallen MJ, Wren BW (2007) Bacterial pathogenomics. Nature 449: 835–842.
- 59. Vidal V, Scragg IG, Cutler SJ, Rockett KA, Fekade D, et al. (1998) Variable major lipoprotein is a principal TNF-inducing factor of louse-borne relapsing fever. Nat Med 4: 1416–1420.
- 60. Pennington PM, Cadavid D, Barbour AG (1999) Characterization of VspB of Borrelia turicatae, a major outer membrane protein expressed in blood and tissues of mice. Infect Immun 67: 4637–4645.
- 61. Cadavid D, Pachner AR, Estanislao L, Patalapati R, Barbour AG (2001) Isogenic serotypes of Borrelia turicatae show different localization in the brain and skin of mice. Infect Immun 69: 3389–3397.
- 62. Plasterk RH, Simon MI, Barbour AG (1985) Transposition of structural genes to an expression sequence on a linear plasmid causes antigenic variation in the bacterium Borrelia hermsii. Nature 318: 257–263.
- 63. Kitten T, Barbour AG (1990) Juxtaposition of expressed variable antigen genes with a conserved telomere in the bacterium Borrelia hermsii. Proc Natl Acad Sci U S A 87: 6077–6081.
- 64. Barbour AG, Burman N, Carter CJ, Kitten T, Bergstrom S (1991) Variable antigen genes of the relapsing fever agent Borrelia hermsii are activated by promoter addition. Mol Microbiol 5: 489–493.
- 65. Barbour AG, Dai Q, Restrepo BI, Stoenner HG, Frank SA (2006) Pathogen escape from host immunity by a genome program for antigenic variation. Proc Natl Acad Sci U S A 103: 18290–18295.
- 66. Schwan TG, Battisti JM, Porcella SF, Raffel SJ, Schrumpf ME, et al. (2003) Glycerol-3-phosphate acquisition in spirochetes: distribution and biological activity of glycerophosphodiester phosphodiesterase (GlpQ) among Borrelia species. J Bacteriol 185: 1346–56.
- 67. Cutler SJ, Fekade D, Hussein K, Knox KA, Melka A, et al. (1994) Successful in-vitro cultivation of Borrelia recurrentis. Lancet 343: 242.
- 68. Schwan TG, Schrumpf ME, Hinnebusch BJ, Anderson DE, Konkel ME (1996) GlpQ: an antigen for serological discrimination between relapsing fever and Lyme borreliosis. J Clin Microbiol 34: 2483–92.
- 69. Cutler SJ (2001) Molecular biology of the relapsing fever borrelia. In: Sussman R, editor. Molecular medical microbiology. Oxford: Academic Press. pp. 2093–2113.
- 70. Parent MA, Goenka R, Murphy E, Levier K, Carreiro N, et al. (2007) Brucella abortus bacA mutant induces greater pro-inflammatory cytokines than the wild-type parent strain. Microbes Infect 9: 55–62.
- 71. Wang XG, Kidder JM, Scagliotti JP, Klempner MS, Noring R, et al. (2004) Analysis of differences in the functional properties of the substrate binding proteins of the Borrelia burgdorferi oligopeptide permease (Opp) operon. J Bacteriol 186: 51–60.
- 72. Frees D, Savijoki K, Varmanen P, Ingmer H (2007) Clp ATPases and ClpP proteolytic complexes regulate vital biological processes in low GC, Gram-positive bacteria. Mol Microbiol 63: 1285–1295.
- 73. Audic S, Robert C, Campagna B, Parinello H, Claverie JM, et al. (2007) Genome analysis of Minibacterium massiliensis highlights the convergent evolution of water-living bacteria. PLoS Genet 3: e138.
- 74. Cutler SJ, Akintunde CO, Moss J, Fukunaga M, Kurtenbach K, et al. (1999) Successful in vitro cultivation of Borrelia duttonii and its comparison with Borrelia recurrentis. Int J Syst Bacteriol 49: 1793–1799.
- 75. Ogata H, Renesto P, Audic S, Robert C, Blanc G, et al. (2005) The genome sequence of Rickettsia felis identifies the first putative conjugative plasmid in an obligate intracellular parasite. PLoS Biol 3: e248.
- 76. Ewing B, Green P (1998) Base-calling of automated sequencer traces using phred. II. Error probabilities. Genome Res 8: 186–194.
- 77. Ewing B, Hillier L, Wendl MC, Green P (1998) Base-calling of automated sequencer traces using phred. I. Accuracy assessment. Genome Res 8: 175–185.
- 78. Gordon D, Desmarais C, Green P (2001) Automated finishing with autofinish. Genome Res 11: 614–625.
- 79. Audic S, Claverie JM (1998) Self-identification of protein-coding regions in microbial genomes. Proc Natl Acad Sci U S A 95: 10026–10031.
- 80. Altschul SF, Madden TL, Schaffer AA, Zhang J, Zhang Z, et al. (1997) Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Res 25: 3389–3402.
- 81. Kanehisa M, Goto S, Kawashima S, Okuno Y, Hattori M (2004) The KEGG resource for deciphering the genome. Nucleic Acids Res 32: D277–280.
- 82. Tatusov RL, Fedorova ND, Jackson JD, Jacobs AR, Kiryutin B, et al. (2003) The COG database: an updated version includes eukaryotes. BMC Bioinformatics 4: 41.
- 83. Lowe TM, Eddy SR (1997) tRNAscan-SE: a program for improved detection of transfer RNA genes in genomic sequence. Nucleic Acids Res 25: 955–964.
- 84. Kurtz S, Phillippy A, Delcher AL, Smoot M, Shumway M, et al. (2004) Versatile and open software for comparing large genomes. Genome Biol 5: R12.
- 85. Enright AJ, Van Dongen S, Ouzounis CA (2002) An efficient algorithm for large-scale detection of protein families. Nucleic Acids Res 30: 1575–1584.
- 86. Edgar RC (2004) MUSCLE: multiple sequence alignment with high accuracy and high throughput. Nucleic Acids Res 32: 1792–1797.
- 87. Castresana J (2000) Selection of conserved blocks from multiple alignments for their use in phylogenetic analysis. Mol Biol Evol 17: 540–552.
- 88. Yang Z (1997) PAML: a program package for phylogenetic analysis by maximum likelihood. CABIOS 13: 555–556.
- 89. Guindon S, Gascuel O (2003) A simple, fast, and accurate algorithm to estimate large phylogenies by maximum likelihood. Syst Biol 52: 696–704.