Introduction

Symbiosis has for long been recognized as one of the crucial drivers of evolution. In insects, numerous symbiotic relationships, mainly with bacteria, enabled the hosts to exploit various environments and/or life strategies, and supposedly started adaptive radiations in some groups. The mechanisms of such evolutionary processes include for example contribution to the host immunity, modification of the reproductive strategy, or provision of essential compounds to the hosts relying on nutritionally compromised resources. Blood feeding (hematophagous) insects provide an illuminating example of a life strategy shift coupled with symbiosis. Since blood meal lacks some of the B vitamins, hematophagous insects rely on their supply by symbiotic bacteria. The relationships between bacteria and hematophagous insect displays considerable degree of variability spanning from less intimate associations with entire gut microbial community, e.g. triatomine bugs [1, 2], to highly specialized interactions with few or single obligate symbiont(s), e.g. lice, bed bugs, tsetse flies, louse flies and bat flies [37]. With the recent advancement of genomic approaches and genetic manipulations, symbioses in these insect groups, often important disease vectors, have become of a high interest.

Here we describe fundamental biological characteristics and genome properties of the obligate symbiont of a deer ked, Lipoptena cervi (Hippoboscidae). In comparison to multipartite symbiotic systems of closely related hosts from families Hippoboscidae (i.e. Melophagus ovinus [7]) and Glossinidae (i.e. Glossina sp. [6]), Lipoptena cervi harbors a single unaccompanied obligate symbiont from the genus Arsenophonus . The genome of Candidatus Arsenophonus lipopteni has been sequenced for two reasons. The first was to extend our knowledge on occurrence and genomics of the obligate symbionts across the spectrum of hematophagous hosts involved in strictly bilateral symbiosis, e.g. bed bugs [5], head lice [4], leaches [8, 9]. This is a necessary prerequisite for the future analysis of the origins and evolution of this kind of symbioses. In addition, we intend to use the sequence in a broader comparative framework focused on evolution of bacterial symbiosis, particularly on its role in B vitamin provision to various ecological types of the hosts.

Organism information

Classification and features

Ca. Arsenophonus lipopteni has an obligate association with its host, L. cervi , and is therefore uncultivable. In order to localize the bacteria within the host, Fluorescent In Situ Hybridization and Transmission Electron Microscopy was performed on dissected gut tissue as described in detail in [7]. For FISH, the tissue was fixed and hybridized in tubes with eubacterial (EUB338, Flc-GCTGCCTCCCGTAGGA; [10]) and Ca. Arsenophonus lipopteni specific probes (ArL, Cy3-CTGACTAACGCTTGCACC; this study). The later was designed in a variable region of 16S rRNA gene taking the target sequence accessibility into account [11].

The distribution of Ca. Arsenophonus lipopteni (Fig. 1) in the host body closely resembles that of Ca. Arsenophonus melophagi and Wigglesworthia glossinidia T , the obligate symbionts of the blood sucking flies Melophagus ovinus and Glossina sp., respectively [7]. Highly pleomorphic cells of the Gram negative non-sporulating bacteria from the family Enterobacteriales are primarily found in the modified part of the gut wall (bacteriome) formed by the specialized enlarged epithelial cells (bacteriocytes, Fig. 1c, 1d). Additional key features of Ca. Arsenophonus lipopteni are provided as a standardized summary in Table 1.

Fig. 1
figure 1

Visualization of the bacteria in the host tissue using FISH and TEM (D). The symbiotic organ (bacteriome) localized in the midgut section harboring Ca. Arsenophonus lipopteni targeted with green (Flc) labeled eubacterial probe (a) and red (Cy3) labeled specific probe (b). Detail of the host cells (bacteriocytes) filled with the symbionts (c). The blue signal is DAPI stained DNA. Four cells of Ca. Arsenophonus lipopteni under TEM (d). The white arrow points to bacterial outer membrane and the red bordered arrow shows the cytoplasmatic cell membrane

Table 1 Classification and general features of Ca. Arsenophpnus lipopteni

Apart from the functional characterization, the genome sequence of Ca. Arsenophonus lipopteni was also utilized to assess the relationship of this bacterium to other Arsenophonus symbionts. Since the sequence compositional shift compromises phylogenetic usage of 16S rDNA, leading to topological artifacts with long branched symbiotic taxa clustering together [12], we carried out a phylogenetic analysis of a multi-gene matrix and used advanced Bayesian approaches. The matrix was generated for all available Arsenophonus genomes (incl. Ca. Riesia pediculicola ), five other symbionts, eight non-symbiotic members of Enterobacteriaceae , and two outgroups. A set of 70 orthologous genes was determined as an intersection of COGs shared by these bacteria (generated using the MicrobesOnline database; [13]) with “SICO” gene list [14]. The genes were retrieved from the finished assembly using Blastp searches [15] and processed as described previously [7]. The resulting matrix contained 22618 unequivocally aligned positions. Phylobayes [16], a tool specifically developed to overcome the difficulty with heterogeneous composition of sequences, was used for the tree reconstruction. The analysis was run in 2 chains under the GTR + CAT model with amino acids recoded according to the Dayhoff6 option. When the convergence was not reached after 20,000 cycles, the program was stopped and majority rule consensus was calculated after discarding 4,000 cycles burn-in.

The results confirm Ca. Arsenophonus lipopteni membership in the genus Arsenophonus . All Arsenophonus species (including Ca. Rieisa pediculicola) formed a well-supported monophyletic branch clustering as a sister group to Providencia (Fig. 2). Despite the length of branches for the obligate symbionts with highly modified genomes, this arrangement was assigned a high posterior probability. Although the six included Arsenophonus lineages certainly do not form a monophyletic group within the known Arsenophonus spectrum [17], the results indicate that Ca. Arsenophonus lipopteni evolved independently from Ca. Arsenophonus melophagi housed in related Hippoboscidae host [7].

Fig. 2
figure 2

Phylogenomic reconstruction of Ca. Arsenophonus lipopteni position. The length of the double crossed branches was scaled to 1:4. The numbers indicate posterior probability for each node. The “N-S” in the brackets following the taxon name designates the non-symbiotic bacteria included into the dataset

Genome sequencing information

Genome project history

The host specimens Lipoptena cervi were collected from wild populations during summer 2010 in the Czech Republic. Finished genome sequence has been deposited in GenBank under acc. No. CP013920 on January 11, 2016. A summary on the sequencing project is provided in Table 2.

Table 2 Project information

Growth conditions and genomic DNA preparation

Since the bacterium is uncultivable, the host tissue was used for DNA extraction. The gut tissue containing the symbiotic organs were dissected from 6 flies in 1× phosphate buffered saline, homogenized with a sterile mortar and pestle and extracted using QiaAmp DNA Micro Kit (QIAGEN, United Kingdom). The DNA quality was assessed using Agilent 2100 Bioanalyzer (Agilent Technologies).

Genome sequencing and assembly

The paired end 100 bp long reads were generated on one lane of Illumina HiSeq2000 run at Yale Center for Genome Analysis. A5 assembly pipeline with the default settings was used to assemble the reads [18]. Of the 109,640 resulting contigs, the longest contig (836,730 bp) with 40× fold coverage formed a circular molecule with 99 bp overlap at the ends. This contig corresponds to the Ca. Arsenophonus lipopteni genome. Pilon v1.12 [19] was used to check assembly quality and to improve base calls and small indels.

Genome annotation

The finished genome was annotated using a combination of following tools: RAST [20], PGAP, and Prokka v1.10 [21]. The annotation was then manually curated and checked for the presence of gene remnants. The final annotation is available in GenBank (CP013920). Metabolic pathways were reconstructed in the RAST server [20] and gene absence was verified using BlastP searches. Proteins were assigned to the clusters of orthologous groups using COGnitor [22], and the presence of signal peptides was detected using SignalP [23]. Pfam domains were predicted using HMMER [24] against the Pfam-A database [25]. Transmembrane predictions were done using TMHMM Server v. 2.0. The search for CRISPR repeats was performed in Geneious [26].

Genome properties

The finished genome consists of 836,724 nucleotides in a single circular chomosome with a low GC content of 24.9 %. The total number of predicted functional genes (667) relative to the genome size implies a lower coding density (75.8 %). The average gene length of 1,001 bp however does not suggest that the genome underwent an extreme economization typical for the obligate symbionts, e.g. Buchnera aphidicola str. Cc, Ca. Sulcia muelleri , Ca. Carsonella ruddii , Ca. Zinderia insecticolla [27]. Over 99 % of protein coding genes have been assigned to particular COGs and Pfam domains (Tables 3 and 4). Signal peptides and transmembrane helices have been identified for 5 and 124 protein coding genes respectively (Table 3). The noncoding RNA genes consist of tmRNA, RNAseP, Alpha RBS, cspA, 35 tRNAs, and 3 rRNA genes (altogether 42 RNA genes). The three ribosomal genes are however not organized into a single operon, a phenomenon previously described for at least 9 unrelated bacterial clades, including gammaproteobacterial symbionts of the genus Buchnera and Candidatus Blochmannia, and attributed to their host-dependent lifestyle [28].

Table 3 Statistics for finished genome assebly of Ca. Arsenophonus lipopteni
Table 4 Number of protein coding genes assigned to the COG categories

The genome properties described above coupled with 16 pseudogenes identified in the genome suggest rather recent establishment of the obligate symbiosis resulting in significant but recent gene/function loss without removal of presently non-coding regions. Regarding the coding capacity for B vitamins and related cofactors, the genome of Ca. Arsenophonus lipopteni appears to be highly economized. Similar to Ca. Arsenophonus melophagi, the bacteria cannot synthesize thiamine (B1), niacin (B3), panthothenic acid (B5) and folic acid (B9). In addition, the genome does not code for heme biosynthesis. Other basic genome characteristics are summarized in Table 3.

Conclusions

Compared to the closely related complex symbiotic system in the sheep ked, Melophagus ovinus , Lipoptena cervi appears to maintain symbiosis exclusively with Ca. Arsenophonus lipopteni . The growing number of genome sequences available for the symbionts and the hematophagous hosts involved in strictly bilateral symbiosis (e.g. [29, 30]) will help elucidating some common requirements on B vitamins, or possibly highlight diverse needs of insects digesting blood of various vertebrates. Ca. Arsenophonus lipopteni possesses complete gene sets for biosynthesis of three B vitamins, riboflavin (B2), pyridoxine (B6) and biotin (B7). While the metabolic capacity is directly assessed from genomic data, the presence of any vitamin efflux systems cannot be easily elucidated due to yet poorly understood mechanisms for vitamin export [31]. However, based on recent findings from other hematophagous systems, it has become more clear that the nutritional interaction does not rely on biosynthesis of all B vitamins as originally suggested by Puchta [32]. For instance, similar to all the other Arsenophonus genomes, biosynthetic capacity for thiamin is compromised in Ca. Arsenophonus lipopteni . The genome however possesses ABC thiamin transporter genes (thiP, thiQ, tbpA) implying the content of thiamin or thiamin pyrophosphate, compared to e.g. biotin or riboflavin, in the host blood might be sufficient for the insect metabolic needs (Novakova, unpublished data). Within the spectrum of known Arsenophonus genomes ranging from 0.57 Mb of Ca. Riesia pediculicola to 3.5 Mb of A. nasoniae , representing various symbiotic types, the genomic sequence of Ca. Arsenophonus lipopteni clearly reflects characteristics common for obligate mutualists. Furthermore, the phylogenetic reconstruction suggests an independent origin of this obligate association within Hippoboscidae.