The spirochete genus Borrelia, named after the French biologist Amédée Borrel, was originally described in 1907 by Swellengrebel [1], with B. anserina (Sakharoff 1891) Bergey et al. 1925 designated as the type species. Since then numerous species and strains have been described, and members of this genus are well recognized as the aetiological agents of Lyme borreliosis (LB) and relapsing fever (RF) in humans. Lyme borreliosis and RF genospecies have long been recognized to have different clinical, biological, and epidemiological characteristics, and phylogenetic data is concordant with this, demonstrating that these two groups are genetically similar yet distinct, and form independent monophyletic sister clades that share a common ancestor [2].

Nevertheless, LB and RF Borrelia share a common set of genetic and biological characteristics that unify these organisms as a group compared to other related spirochetes. Namely, all LB and RF Borrelia species are spirochetes with an obligate parasitic lifestyle, are transmitted between vertebrate hosts by arthropod vectors (ticks and louse), and can be transstadially transmitted within their arthropod vectors. Various vector associations of Borrelia have been found in nature, with the genus Ixodes mainly vectoring LB species while argasid ticks often vector the RF group. However, some members of the RF group are associated with hard ticks of the family Ixodidae (e.g. B. lonestari) [3], with the human body louse Pediculus humanus (B. recurrentis) [4] or the genus Ixodes (e.g. B. miyamotoi) [5]. The genus Ixodes represents an ancient genus of the family Ixodidae sharing numerous original features with argasid ticks. Both, LB and RF spirochetes are dependent on their vertebrate and arthropod hosts for the majority of their nutritional requirements, and share a unique genomic structure comprised of a single highly conserved linear chromosome and numerous extrachromosomal linear and circular plasmids that can be highly variable between strains [6–8].

Recently, a third group of Borrelia organisms has been described that are associated with reptile and echidna (Tachyglossus aculeatus) hosts, and do not phylogenetically cluster within either the RF or LB clades. Instead these novel borreliae form their own independent lineages which sit as an outgroup to, and shares a most recent common ancestor with, the RF clade [9]. This novel clade currently has two designated species, B. turcica [10] and ‘Candidatus Borrelia tachyglossi’ [9, 11, 12], and several other genetic variants that are yet to be formally taxonomically classified [13]. Known vectors for this group include hard ticks of the genera Amblyomma, Bothriocroton, and Hyalomma [10, 11, 13].

Recently Adeolu and Gupta [14] proposed to divide Borrelia into two genera to reflect the genetic and phenotypic divergence between LB and RF species, however, this proposal remains under debate [15, 16], and has not been widely utilized in the literature. The justification for this proposal was largely based on the identification of conserved signature insertions/deletions (indels) (CSIs) and conserved signature proteins (CSPs) that are differentially present in the LB or RF Borrelia genogroup, as well as average nucleotide identity (ANI) values calculated between whole genomes of 18 Borrelia species including eight LB species and ten RF species. Although it is uncontested that these differences exist between LB and RF Borrelia, we propose that the methodology used to identify these group-specific differences is subjective and has a highly limited power to delineate LB and RF Borrelia into separate genera.

The methodology employed by Adeolu and Gupta [14] reported only “CSIs that are specific for different groups within the Borrelia”, and CSPs only “if either all significant [BLAST] hits were from well-defined group of Borrelia or which involved a large increase in E-values from the last hit belonging to a particular group of Borrelia to the first hit from any other group”. This methodology specifically identifies only CSIs and CSPs that are exclusive only to one Borrelia genogroup, and precludes the detection of CSIs or CSPs that may be shared non-exclusively between both genogroups (i.e. contests the hypothesis that LB and RF belong in different genera). This data presented in isolation misrepresents the extent of genomic divergence between LB and RF Borrelia and fails to consider widespread genomic similarities between these two groups. Additionally, although ANI has previously been used to investigate prokaryote taxonomy [17], a comprehensive review of this method revealed that although ANI can accurately quantify the genetic relationships between strains belonging to the same species, it was not suitable to differentiate prokaryotic genera. This is due to significant overlapping of intergenera ANI and interspecies ANI values [18], which leads to unreliability in the method.

Alternatively, Qin et al. [18] presented a more heuristic method for delineating prokaryotic genera that measures the percentage of conserved proteins (POCP) between whole genome pairs, reasoning that the degree of protein conservation reflect both genetic and phenotypic relatedness more substantially. Qin et al. [18] demonstrated that among 235 prokaryotic species from 97 genera that POCP values had a higher predictive power than ANI to delineate genera. They showed that with few exceptions POCP values of ≥ 50% could be considered a threshold for prokaryotic genus delimitation, pending other genomic factors that influence POCP, such as large differences in genome size.

Here we investigate the validity of the proposed delineation of LB and RF Borrelia into separate genera by performing pairwise analysis of POCP values between 30 Borrelia type strain genomes (where possible), including two new Borrelia genomes from the novel reptile and echidna-associated clade, B. turcica, and ‘Candidatus Borrelia tachyglossi’, which have yet to be analyzed in this context. We also re-examine the CSIs previously used to support the delineation of LB and RF Borrelia in the genomes of B. turcica and ‘Candidatus Borrelia tachyglossi’ to establish whether these molecular markers are useful to establishing the relationship of B. turcica, and ‘Candidatus Borrelia tachyglossi’. Our analyses indicate that insufficient genomic divergence exist between LB and RF Borrelia to consider them separate genera, and that Borrelia CSIs are limited in their ability to unambiguously distinguish the taxonomic identity of B. turcica and ‘Candidatus Borrelia tachyglossi’.

In order to determine whether the 50% POCP threshold for genus delineation was appropriate for spirochete taxa, we used pairwise POCP analysis to determine the inter-genera POCP values for 84 spirochete genomes from the genera Borrelia, Brachyspira, Leptospira, Spirochaeta, and Treponema. Among all spirochete genomes investigated inter-genera POCP values ranged between 4.8% and 36.8% (mean 10.1%), indicating a low degree of protein conservation occurs between spirochete genera (S1 Table). These spirochete inter-genera POCP values are at a minimum of 13.2% lower than the 50% value determined by [18], suggesting this value is an appropriate and highly conservative threshold for spirochete genera delineation.

Compared to the low level of protein conservation measured between spirochete genera, POCP values were significantly higher among Borrelia species. Borrelia POCP values were highest among the LB genospecies, which ranged between 81.1–94.4% (mean 90.2%), while POCP values between RF species were generally lower and more variable, ranging between 65.3–93.1% (mean 81.1%) (Figs 2 and 3). Despite sharing a most recent common ancestor with RF Borrelia, B. turcica and ‘Candidatus Borrelia tachyglossi’ genomes shared higher POCP values with many LB Borrelia species compared to RF species, such as B. chilensis (78.9% and 87.7%, respectively), B. americana (77.4% and 82.2%, respectively), B. japonica (77.0% and 82.0%, respectively), and B. mayonii (76.8% and 81.42%, respectively). However, B. turcica and ‘Candidatus Borrelia tachyglossi’ genomes did have higher levels of protein conservation with B. anserina (79.1% and 88.5%, respectively) and B. miyamotoi (‘Candidatus Borrelia tachyglossi’ only: 87.7% POCP) (Figs 2 and 3). Most significantly, all Borrelia pairwise POCP values consistently remained well above the 50% POCP threshold for genus delineations proposed by Qin et al. [18], with a minimum value of 64.8% (B. crocidurae Achema vs. B. chilensis VA1), and a maximum value of 88.8% (B. miyamotoi LB-2001 vs. B. chilensis VA1) (mean value: 73.6%) (Figs 2 and 3).

The original proposal to delineate LB and RF Borrelia was largely based on the occurrence of 53 CSIs that have different forms in LB and RF genogroups [14]. It was subsequently defended by suggesting that novel Borrelia species that group with, or as an outgroup to RF Borrelia would be expected to contain RF-specific CSIs and generally none specific to the LB group [23]. An examination of these 53 CSIs in B. turcica and ‘Candidatus Borrelia tachyglossi’ genomes shows although the majority of CSIs present in these genomes correspond to the RF-specific form of the indels, 9/53 (17.0%) and 11/53 (20.8%) of the CSIs in B. turcica and ‘Candidatus Borrelia tachyglossi’, respectively, correspond to LB-specific forms (Fig 4; Table 2).

The data presented in this study very clearly demonstrate that all groups investigated, i.e. RF group spirochetes, LB group spirochetes, reptile- and echnida-associated Borrelia species belong to the same genus as values for POCP were consistently above the proposed threshold for genus delimitation. We propose to re-establish the genus Borrelia in its original form.

Emended description of the genus Borrelia (Swellengrebel 1907) (approved lists 1980)

Organisms are helical (0.2–0.3 μm by 10–35 μm). Periplasmic flagella overlap in the central region of the cell. Cells are flexible and motile with rotational and forward/backwards movement. Organisms are host-associated and microaerophilic. They are vectored by argasid ticks, prostriate ixodid ticks, metastriate ixodid ticks and the human body louse. The genome is fragmented into a linear main chromosome, and linear or circular plasmids. The G/C content of the genomic DNA is 27–32 (mol%).

Members of this genus are the causative agents of relapsing fever, Lyme borreliosis or of unknown pathogenicity.

The type species is Borrelia anserina (Sakharoff 1891) Bergey et al. 1925(Approved Lists 1980).

Description of Borrelia afzelii Canica et al 1994

The description is the same as in Canica et al 1994. DNA-DNA hybridization, RFLP of the rrs gene as well as reactivity of monoclonal antibodies differentiates B. afzelii from other Borrelia species [26, 27]. B. afzelii strains are also distinguishable from all other LB species by using Multilocus sequence analysis [24].

Type strain VS461^T (= DSM 10508^T = CIP 103469^T)

Description of Borrelia lusitaniae (Le Fleche et al. 1997, sp. nov.)

The description is the same as given in Le Fleche et al. 1997.

Type strain PotiB2 ^T (= CIP-105366 ^T, = DSM-107168^T)