A Francisella-like endosymbiont in the Gulf Coast tick evolved from a mammalian pathogen

Ticks (order Ixodida) vector pathogenic bacteria that cause diseases in humans and other mammals. They also contain bacteria that are closely related to pathogens but function as endosymbionts that provide nutrients that are missing from mammalian blood—their sole food source. For instance, mammalian pathogens such as Coxiella burnetii and Francisella tularensis, as well as Coxiella-like and Francisella-like endosymbionts (CLEs and FLEs, respectively) occur in ticks worldwide. However, it is not clear whether the pathogens evolved from symbionts or symbionts from pathogens. Recent studies have indicated that C. burnetii likely originated from a tick-associated ancestor, but the origins of FLEs are not clear. In this study, we sequenced the genome of an FLE, termed FLE-Am, present in the Gulf Coast tick, Amblyomma maculatum. We show that FLE-Am likely evolved from a pathogenic strain of Francisella, indicating that tick endosymbionts can evolve from mammalian pathogens. Although the genome of FLE-Am is almost the same size as the genomes of pathogenic Francisella strains, about one-third of its protein-coding genes contain inactivating mutations. The relatively low coding capacity and extensive metabolic capabilities indicate that FLE-Am transitioned recently to its current endosymbiotic lifestyle and likely replaced an ancient endosymbiont with degraded functionality.

Scientific RepoRts | 6:33670 | DOI: 10.1038/srep33670 that are pathogenic to fish. Our multi-protein tree clarifies earlier phylogenetic trees based on partial FLE-Am 16S rRNA sequences that showed it to be a sister taxon of F. tularensis 11 . Virulence genes have been inactivated in FLE-AM. As shown recently for C. burnetii, mammalian pathogens could evolve from non-pathogenic ancestors by acquiring virulence genes 12 ; conversely, avirulent symbionts could arise from pathogenic ancestors by losing virulence genes, but no clear examples of this process have been documented. To identify the evolutionary relationship between pathogenic Francisella and FLE-Am, we examined the FLE-Am genome for the presence of virulence genes described in F. tularensis and F. novicida 13,14 . We discovered that FLE-Am contained pseudogenized versions of several virulence genes, including genes for a Type VI Secretion System present on a pathogenicity island in F. tularensis and for Type 4 pili that are critical to mammalian infection (Fig. 2, Supplementary Table S2). Collectively, our data denotes that the ancestor of FLE-Am was most likely a mammalian pathogen that contained functional versions of virulence genes. The absence of intact secretion and effector gene systems in FLE-Am suggests that it is avirulent to humans despite its presence in the salivary glands and saliva of A. maculatum 8 . Intriguingly, salivary glands of other Amblyomma species such as A. americanum contain CLEs 15 . Although its functional significance is not understood, being secreted in saliva could facilitate the exchange of endosymbionts between ticks while co-feeding on the same host 16 .  (Table 1), and a significant portion (~ 33%) of its protein-coding genes contains inactivating mutations (Fig. 2  in long-term endosymbionts such as a CLE in A. americanum (CLEAA) 3 , and the presence of large number of pseudogenes imply that the bacterium is in the initial stages of reductive evolution, as superfluous genes are first pseudogenized and then ultimately deleted from the genome when a bacterium shifts from a free-living to a host-associated lifestyle. Additionally, endosymbiont genomes tend to be much more A+ T biased than closely related pathogens or environmental bacteria; however, the nucleotide composition (G+ C%) of FLE-Am is very  Because bacteria with extensive metabolic proficiency are known to replace ancient endosymbionts with reduced metabolic prowess 9 , we compared the metabolic pathways present in FLE-Am to that of CLEAA, the cofactor provisioning endosymbiont found in A. americanum 3 . As shown in Fig. 3, the metabolic capability of FLE-Am is much more extensive than in CLEAA. For instance, FLE-Am can produce heme in addition to cofactors (except thiamine) synthesized by CLEAA. Furthermore, while both FLE-Am and CLEAA share the ability to produce aspartate from pyruvate, and to metabolize it into ATP, only the FLE synthesizes cysteine, threonine, tyrosine, tryptophan, phenylalanine, and serine from pyruvate, and can metabolize glutamate, glutamine, and asparagine into ATP. Of these amino acids, FLE most crucially provides a reliable source of the amino acid cysteine, which is found in very low concentrations in bovine blood 1 . FLE-Am can also synthesize glutamine from glutamic acid and ammonia, thus recycling cellular ammonia waste to useful products. In sum, our data indicate that the superior biosynthetic capability of FLE-Am confers a selective advantage, which could have led to FLE-Am recently replacing an ancestral symbiont (e.g. CLE) with reduced metabolic capacity in A. maculatum.
Symbiosis with bacteria allows arthropods to thrive on nutrient-poor diets 2 . However, the absolute reliance on a symbiont could eventually become disadvantageous to the host because massive gene loss and accumulation of mildly deleterious mutations will corrode the symbiont's metabolic capability 7,9 . One solution to escaping this symbiosis "rabbit hole" is to supplement or replace the ancient symbiont with new bacteria acquired from the environment 9 . This process has occurred in several lineages of insects (e.g. ref. 17). In A. maculatum, FLE-Am seems to have recently replaced an older endosymbiont, whose identity is unknown, but based on the current bacterial prevalence data it was most probably a CLE or a Rickettsia 12 . FLE-Am likely arose by the domestication of a mammalian pathogen that was vectored by the tick. This process could occur rapidly because the bacterium doesn't have to learn anew how to circumvent the tick's immune response, and there is no need to attenuate bacterial virulence towards the tick 18 . Additionally, because genes that promote pathogenesis through amino acid scavenging (e.g., FTT 0968c, xasA), macrophage survival (e.g., carA, carB, bioF), and intracellular replication (aroE1, purMCD, purL, and purF) 13,14 are retained in FLE-Am (Fig. 1), they could be key to its endosymbiotic lifestyle. Further evolutionary, functional and genomic studies of FLEs and CLEs from a wide array of soft and hard ticks will help us to better understand how pathogenic Francisella evolved into endosymbionts that supports the blood-dependent lifestyle of ticks, and how tick-associated Coxiella evolved into pathogenic C. burnetii 12 (Fig. 4).

Methods
Sequencing and bacterial identification. DNA was extracted from a female A. maculatum procured from Oklahoma State University Tick Rearing Facility, as described previously 3 , and was sequenced using Illumina Hi-Seq 2500 (100 cycles, paired-end) at OHSU MPSSR, yielding approximately 180 million read pairs. Low confidence reads were removed, and the identify of bacteria present in the tick microbiome were determined utilizing ≥ 1 kb contigs binned using MEGAN 19 .
Genome Assembly. Trimmed reads were assembled into contigs using IDBA-UD 20 , and by comparing them to Francisella genome sequences, FLE contigs were identified. All trimmed reads were mapped to these contigs and then reassembled with IDBA-UD and IDBA-hybrid into a final set of seven contigs (Supplementary Table S3), which were submitted to NCBI (Accession: LVCE00000000). The completeness of the assembled genome was examined using a single-copy gene database 21 , and as a control, an identical single-copy gene analysis was performed on F. tularensis subsp. tularensis SCHU S4 genome (NC_006570.2) ( Table 1).
Genome Annotation. FLE-Am contigs were annotated using NCBI Prokaryotic Genome Annotation Pipeline, and pseudogenes were verified manually. Protein-coding genes were binned into categories based on their role in primary metabolism, amino acid and nucleic acid synthesis, or vitamin and cofactor metabolism, and then subsequently compared to CLEAA to illustrate differences in their metabolic capabilities. A list of genes critical to the pathogenicity of F. tularensis 13,14 was used to identify both functional and pseudogenized versions of virulence genes present in FLE-Am (Supplementary Table S2). Intact protein-coding genes present in FLE-Am but not in F. tularensis SCHU-S4 are provided in Supplementary Table S4.
Phylogenetic Analysis. In addition to FLE-Am, we included all 44 fully sequenced Francisella genomes to generate a robust phylogenetic tree. Using reciprocal BLASTP, a subset of 442 orthologous genes (Supplementary Dataset 2) that were conserved in all genomes was identified. Nucleotide sequences were aligned using Clustal Omega 22 , ambiguously aligned regions were removed using Gblocks 23 , and jModelTest2 was used to select the GTR+ I+ G (General Time Reversible plus Invariant sites plus Gamma distribution) evolution model 24 . Using concatenated sequences, Maximum Likelihood trees with 1,000 bootstrap replicates were constructed using RAxML 25 . Bayesian trees with a chain length of 500,000 and a burn-in fraction of 25% and sampling every 100 trees were constructed using MrBayes 26 . Mammalian pathogens could evolve from tick-associated bacteria by acquiring virulence genes (e.g., Coxiella burnetii). Conversely, evolution of tick endosymbionts from mammalian pathogens is associated with loss of virulence genes (e.g., FLE-Am).