Bacterial protein meta-interactomes predict cross-species interactions and protein function

Abstract

Background: Protein-protein interactions (PPIs) can offer compelling evidence for protein function, especially when viewed in the context of proteome-wide interactomes. Bacteria have been popular subjects of interactome studies: more than six different bacterial species have been the subjects of comprehensive interactome studies while several more have had substantial segments of their proteomes screened for interactions. The protein interactomes of several bacterial species have been completed, including several from prominent human pathogens. The availability of interactome data has brought challenges, as these large data sets are difficult to compare across species, limiting their usefulness for broad studies of microbial genetics and evolution.

Results: In this study, we use more than 52,000 unique protein-protein interactions (PPIs) across 349 different bacterial species and strains to determine their conservation across data sets and taxonomic groups. When proteins are collapsed into orthologous groups (OGs) the resulting meta-interactome still includes more than 43,000 interactions, about 14,000 of which involve proteins of unknown function. While conserved interactions provide support for protein function in their respective species data, we found only 429 PPIs (~1% of the available data) conserved in two or more species, rendering any cross-species interactome comparison immediately useful. The meta-interactome serves as a model for predicting interactions, protein functions, and even full interactome sizes for species with limited to no experimentally observed PPI, including Bacillus subtilis and Salmonella enterica which are predicted to have up to 18,000 and 31,000 PPIs, respectively.

Conclusions: In the course of this work, we have assembled cross-species interactome comparisons that will allow interactomics researchers to anticipate the structures of yet-unexplored microbial interactomes and to focus on well-conserved yet uncharacterized interactors for further study. Such conserved interactions should provide evidence for important but yet-uncharacterized aspects of bacterial physiology and may provide targets for anti-microbial therapies.

Keywords: Genome evolution; Interactome; Networks; Protein interactions.

Fig. 1

Citation analysis of the bacterial interactome literature. Publication counts include all papers that cite at least one of a total of 11 published bacterial interactome studies (as of August 2015)

Fig. 2

Consensus meta-interactome. a A breakdown of source species of the meta-interactome shows that PPIs in E. coli or C. jejuni contributed to more than half of the total set of interactions in the meta-interactome. b We defined the pool of interactions between proteins in different bacteria as the meta-interactome. To account for homologous proteins, we considered groups of orthologous proteins (OG) as nodes in a consensus meta-interactome. In particular, we weighted links between OGs by the underlying number of observed interactions between proteins in such groups. c The main component of the consensus meta-interactome pools 88.9% of all OGs. Our graphical depiction suggests that the majority of OGs consist of one protein, while such groups are mostly linked by one underlying PPI. d More quantitatively, we found that the majority of OGs in the consensus meta-interactome indeed have only one protein while a minority of groups includes many proteins. e The distributions of the number of PPI that connect proteins in different OGs (f) as well as the number of neighboring OGs decay as a power-laws

Fig. 3

Conserved and cross-functional interactions in the consensus meta-interactome. a Counts of PPIs in the consensus meta-interactome network. N_species indicates the number of distinct bacterial species contributing the interaction; a value of 1 denotes an interaction observed for a single species only. For each count, subsets denote how many PPIs involve two, one, or zero interactors of known function (as both, one, and none, respectively). b Significant connections between functional classes are mediated by the underlying PPIs in the consensus meta-interactome. For each class combination we calculated a Z-score that reflects the significance of the interaction density between classes and class coverage. While interactions mostly appear between the same classes, we also observe that most functional cross-talk emanates from OGs with translational as well as posttranslational functions

Fig. 4

The consensus meta-interactome improves functional predictions. a Predicting the functions of sampled OGs we observed that the addition of the consensus meta-interactome allowed for better functional prediction (P < 10^-50, Student’s t-test). Connecting functionally annotated orthologous groups (OG) if they harbored interacting proteins of E. coli we randomly sampled 20% of all OGs 1000 times and utilized the remainder to predict the functions of the sampled OGs. As a measure of the prediction quality we calculated the area under the ROC curve. In comparison, we augmented the underlying E. coli specific network of OGs with remaining links in the underlying consensus meta-interactome. b We obtained similar results when we considered OGs that were initially connected by interactions between proteins of C. jejuni. In the inset of (c) we calculated the fraction of correctly predicted functions of OGs as a function of the degree in the underlying OG networks of E. coli and C. jejuni, suggesting that increased number of links corresponds to elevated levels of prediction accuracy. Assessing the impact of the consensus meta-interactome on the accuracy of predicted functions of OGs, we observed that the functional prediction for OGs with low degree was improved

Fig. 5

Functional prediction of uncharacterized orthologous groups. Functional similarity of interacting orthologous groups in the network. Each orthologous group (specifically, a node in the network) occupies a single row in the heatmap. A node’s degree in the consensus meta-interactome is shown on the right. Each column is a single functional category

Fig. 6

Predictions of maximal interactome size. Based on the consensus meta-interactome we show the upper bounds of predicted interactome size (in number of PPI) as a function of proteome size. Each point corresponds to the Uniprot reference proteome of a single species (see Methods for strain identities and text for details)

Fig. 7

The NDH-1 complex as an example of conserved interactions. a An NDH component interaction network from multiple species. Each node in this network corresponds to a single orthologous group and is labeled with the corresponding group member in cyanobacteria (i.e., the sources of most of the PPI observed for NDH complex members). Groups are colored as in Part B; groups in gray have predicted accessory functions. Interactions between any proteins in two groups are shown as edges. Edges are colored as noted in the Key. b A model of the NDH-1MS complex in cyanobacteria. Figure after [54]. Each box corresponds to a protein or group of proteins; those labeled with a single letter are Ndh proteins. Dotted lines indicate alternate complex forms. See [54] for further details