Modeling trophic dependencies and exchanges among insects’ bacterial symbionts in a host-simulated environment

Individual organisms are linked to their communities and ecosystems via metabolic activities. Metabolic exchanges and co-dependencies have long been suggested to have a pivotal role in determining community structure. Metabolic interactions with bacteria have been key drivers in the evolution of sap-feeding insects, enabling complementation of their deprived nutrition. The sap-feeding whitefly Bemisia tabaci (Hemiptera: Aleyrodidae) harbors an obligatory symbiotic bacterium, as well as varying combinations of facultative symbionts. We took advantage of the well-defined bacterial community in B. tabaci as a case study for a comprehensive and systematic survey of metabolic interactions within the bacterial community and their associations with documented frequency of bacterial combinations. We first reconstructed the metabolic networks of five common B. tabaci symbionts (Portiera, Rickettsia, Hamiltonella, Cardinium and Wolbachia), and then used network analysis approaches to predict: (1) species-specific metabolic capacities in a simulated bacteriocyte-like environment; (2) metabolic capacities of the corresponding species’ combinations, and (3) dependencies of each species on different media components. The automatic-based predictions for metabolic capacities of the symbionts in the host environment were in general agreement with previously reported genome analyses, each focused on the single-species level. The analysis suggested several previously un-reported routes for complementary interactions. Highly abundant symbiont combinations were found to have the potential to produce a diverse set of complementary metabolites, in comparison to un-detected combinations. No clear association was detected between metabolic codependencies and co-occurrence patterns. The findings indicate a potential key role for metabolic exchanges as key determinants shaping community structure in this system.

). The interactions formed between the most frequent symbionts -the obligatory  Table   1 8 7 2). In comparison, the lowest number of complementary metabolites was predicted for  Wolbachia, and Rickettsia-Wolbachia and Rickettsia-Cardinium (red combinations in Table 2 1 9 7 and Fig. 2). Cardinium-Portiera combination is classified together with Hamiltonella-1 9 8 Wolbachia and not with the other Portiera associated combinations. Metabolites common to 1 9 9 the Portiera-associated combinations included amino-acyl transferases and many primary precursors of methionine and purine/thiamine (Table S4); all potential interactions have been 2 0 3 previously suggested for Hamiltonella (39), but not for Wolbachia. to the fact that most of these metabolites are not common but rather interaction-specific: were mostly involved in butanoate and amino sugar metabolism (Table S4). Finally, non-2 1 0 occurring combinations typically led to a low number of potential complementary 2 1 1 metabolites and were clustered. Under the assumption that highly similar metabolic demands may hint at resource 2 1 4 competition and potentially lead to exclusion of the less fit competitor, the extent to which 2 1 5 symbiont combinations rely on common resources was assessed. Scores were evaluated using 2 1 6 NetCmpt, which provides predictions for the degree of effective metabolic overlap between 2 1 7 pairs of bacterial species, ranging between 0 (no overlap) and 1 (complete overlap) (26).

1 8
Scores are a-symmetrical whereas the effect of interactions on pair members is likely to differ 2 1 9 (i.e., one of the species is likely to be more affected than its potential competitor). The score presence of Wolbachia and Rickettsia (Table 2). Overall, pairwise scores were relatively low,  bacterial communities (4). Notably, no significant difference was obsereved in the level of 2 2 7 metabolic overlap between occurring versus non-occurring combinations ( Table 2). Since resource overlap is thought to determine community structure only under limited 2 2 9 carrying capacity of the habitat (48), we further simulated species-specific growth in the 2 3 0 bacteriocyte-like environment, rather than considering the generic optimal environment 2 3 1 assumed by the NetCmpt tool. We estimated the specific qualitative effect of each metabolite 2 3 2 on growth capacity following iterative removal of one component at a time. As expected, 2 3 3 Portiera exhibited the most differentiated dependency profile of all symbionts (Fig. 3). In the 2 3 4 specific bacteriocyte simulated environment, Portiera relied uniquely on D-ribose 5- well as on L-homocysteine for methionine production. Metabolite dependencies that were 2 3 7 common to more than a single symbiont included dependencies on the amino acids L-  Hence, co-dependency might lead to a mutually exclusive distribution pattern, as suggested 2 4 0 for Wolbachia and Rickettsia (34).

4 1
In addition, common dependencies on NAD+ (Hamiltonella, Wolbachia and Rickettsia) and ATP, its activation requires thiamine diphosphate, which was not present in our bacteriocyte 2 4 7 environment. In our simulations, Wolbachia was the only symbiont that could produce Rickettsia from B. tabaci is part of the R. bellii group that includes many pathogenic  and L-alanyl-tRNAs, Table S4) (33). Although these losses are assumed to reflect the  Complementary interactions also lead to the potential synthesis of secondary metabolites nutritional sources (65) (Table S4). Rickettsia and Hamiltonella-Wolbachia are relatively diverse (Fig. 2), suggesting a biotype-  While the analysis suggested an association between high-complementation and frequent co-3 4 7 occurrence no such indication was detected for competitive interactions (Table 2). One cysteine is a non-essential amino acid that can be supplied by the host and is found in the 3 6 0 phloem, it is the main sulfur source required for Fe-S protein biogenesis (68). In addition, 3 6 1 common dependencies in NAD+ and ATP which reflect the energy-production pathways of 3 6 2 the corresponding symbionts can have a strong influence on symbiont co-occurrences. For translocases are also present, indicating to a parasitic past (33, 70, 71). In contrast, it seems  Putative pseudogenes for all re-annotated genomes were predicted using GenePrimp (77).

0 4
Manual inspection was performed for all candidate pseudogenes that had an assigned each genome is indicated in Table 1. The final EC lists are provided in Table S5. reactions. Although, the closest organisms with a well-known and defined bacterciocyte 4 2 3 environment are aphids, we decided not to use the information generated for this organisms, that would otherwise be masked by alternative host-symbiont routes. These compounds were 4 3 2 termed "source metabolites" (detailed in Table S2) and were used as starting points for genomes, leading to the construction of niche-specific networks. Complementation was predicted through a three-stage model (1) constructing a combined set  Table S4. PCA for the vectors of synergistic metabolites was carried out using R software (81).

4
Prediction of co-dependencies in source metabolites 4 4 5 The competition scores for each pair of symbionts were calculated by the network-based tool of essential metabolites was determined (e.g., amino acids, nucleic acid and co-factors, each iteration, the number of essential metabolites that could not be produced following the 4 5 3 removal of a source metabolite was recorded. The procedure is illustrated in Fig. S3. The study was supported by the Israel Science Foundation, grant no. 1481/13. We thank Genoscope (http://www.genoscope.cns.fr) for providing the RAW data for 4 6 0 assembly of the Wolbachia genome.  Evol 4:226-232. Metabolic modeling of a mutualistic microbial community. Mol Syst Biol 3:92-104. Metabolic dependencies drive species co-occurrence in diverse microbial communities Springer Netherlands.  Alien Species in the Yunnan Province ( China ). J Insect Sci 6:1-8. aleyrodidarum" BT-QVLC, an obligate symbiont that supplies amino acids and insights into genome reduction, symbiont motility, and its settlement in Bemisia tabaci.