Bacterial endosymbionts of insects play a central role in upgrading the diet of their hosts. In certain cases, such as aphids and tsetse flies, endosymbionts complement the metabolic capacity of hosts living on nutrient-deficient diets, while the bacteria harbored by omnivorous carpenter ants are involved in nitrogen recycling. In this study, we describe the genome sequence and inferred metabolism of Blattabacterium strain Bge, the primary Flavobacteria endosymbiont of the omnivorous German cockroach Blattella germanica. Through comparative genomics with other insect endosymbionts and free-living Flavobacteria we reveal that Blattabacterium strain Bge shares the same distribution of functional gene categories only with Blochmannia strains, the primary Gamma-Proteobacteria endosymbiont of carpenter ants. This is a remarkable example of evolutionary convergence during the symbiotic process, involving very distant phylogenetic bacterial taxa within hosts feeding on similar diets. Despite this similarity, different nitrogen economy strategies have emerged in each case. Both bacterial endosymbionts code for urease but display different metabolic functions: Blochmannia strains produce ammonia from dietary urea and then use it as a source of nitrogen, whereas Blattabacterium strain Bge codes for the complete urea cycle that, in combination with urease, produces ammonia as an end product. Not only does the cockroach endosymbiont play an essential role in nutrient supply to the host, but also in the catabolic use of amino acids and nitrogen excretion, as strongly suggested by the stoichiometric analysis of the inferred metabolic network. Here, we explain the metabolic reasons underlying the enigmatic return of cockroaches to the ancestral ammonotelic state.

Many insect species have established long-term symbiotic relationships with intracellular bacteria. Symbiosis with bacteria has provided insects with novel ecological capabilities, which have allowed them colonize previously unexplored niches. Despite its importance to the understanding of the emergence of biological complexity, the evolution of symbiotic relationships remains hitherto a mystery in evolutionary biology. In this study, we contribute to the investigation of the evolutionary leaps enabled by mutualistic symbioses by sequencing the genome of Blattabacterium cuenoti, primary endosymbiont of the omnivorous cockroach Blatta orientalis, and one of the most ancient symbiotic associations. We perform comparative analyses between the Blattabacterium cuenoti genome and that of previously sequenced endosymbionts, namely those from the omnivorous hosts the Blattella germanica (Blattelidae) and Periplaneta americana (Blattidae), and the endosymbionts harbored by two wood-feeding hosts, the subsocial cockroach Cryptocercus punctulatus (Cryptocercidae) and the termite Mastotermes darwiniensis (Termitidae). Our study shows a remarkable evolutionary stasis of this symbiotic system throughout the evolutionary history of cockroaches and the deepest branching termite M. darwiniensis, in terms of not only chromosome architecture but also gene content, as revealed by the striking conservation of the Blattabacterium core genome. Importantly, the architecture of central metabolic network inferred from the endosymbiont genomes was established very early in Blattabacterium evolutionary history and could be an outcome of the essential role played by this endosymbiont in the host’s nitrogen economy.

Supplementary figure legends.Supplementary Fig. 1 Number of antibiotic resistance genes (ARG) according to the phylogenetic group among E. coli ST410 isolates. The horizontal lines in the boxes represent the median number of ARG associated to the three clades: Basal (n=22); FQR clade (n=92) and OXA-181 subclade (n=41). The box boundaries represent the first and third quartiles of the distribution and box-plot whiskers span 1.5 times the interquartile range of the distribution. Outliers are denoted as black dots outside whiskers. Statistical significances were tested with a one-sided Wilcoxon rank-sum test. ***, P < 0.001. Supplementary Fig. 2. Recombination events in isolates of the E. coli ST410 FQR clade. Phylogeny of the E. coli ST410 FQR clade (left). On the right side, the recombined blocks identified by Gubbins 1 are highlighted in red when they occurred in an internal branch and affect more than one isolate and in blue when they occurred in a terminal branch affecting a single isolate. The core genome of E. coli ST410 FQR strains is represented above the figure and annotated genes correspond to the limits of major recombination events. Supplementary Fig. 3. Phylogeny and mutations in non-redundant CP-Ec isolates of ST167, ST101 and ST156. ML phylogenies were estimated as for Fig. 1, using 5,843, 21,810, and 18,746 non-recombinant SNPs for a. Ec ST167, rooted with the ST10 strain MG1655 (NC_00913) b. Ec ST101, and c. Ec ST156 respectively, both rooted with the ST1128 strain IAI1 (NC_011741). 14 distantly related non-CP-Ec ST156 isolates have been removed from the phylogenetic analysis. Branch tips indicates the presence and the type of carbapenemase according to the figure key on the left. On the right side of the tree are represented, from left to right: blaCTX-M ESBL, gyrA and parC QRDR mutations, mutations in the ftsI gene, and genetic events affecting ompF and ompC according to the figure key at the bottom. SNPs in the dcw region are represented by small vertical red bars. Genes from the dcw locus are indicated by arrows and the ftsI gene in red. Black arrowheads indicate isolates used as reference for SNPs mapping in the dcw gene cluster.