Search for: All records

Abstract Staphylococcus aureus has evolved mechanisms to cope with low iron (Fe) availability in host tissues. Staphylococcus aureus uses the ferric uptake transcriptional regulator (Fur) to sense titers of cytosolic Fe. Upon Fe depletion, apo-Fur relieves transcriptional repression of genes utilized for Fe uptake. We demonstrate that an S. aureus Δfur mutant has decreased expression of acnA, which codes for the Fe-dependent enzyme aconitase. This prevents the Δfur mutant from growing with amino acids as sole carbon and energy sources. We used a suppressor screen to exploit this phenotype and determined that a mutation that decreases the transcription of isrR, which produces a regulatory RNA, increased acnA expression, thereby enabling growth. Directed mutation of bases predicted to facilitate the interaction between the acnA transcript and IsrR, decreased the ability of IsrR to control acnA expression in vivo and IsrR bound to the acnA transcript in vitro. IsrR also bound transcripts coding the alternate tricarboxylic acid cycle proteins sdhC, mqo, citZ and citM. Whole-cell metal analyses suggest that IsrR promotes Fe uptake and increases intracellular Fe not ligated by macromolecules. Lastly, we determined that Fur and IsrR promote infection using murine skin and acute pneumonia models. 

Networks or graphs provide a natural and generic way for modeling rich structured data. Recent research on graph analysis has been focused on representation learning, of which the goal is to encode the network structures into distributed embedding vectors, so as to enable various downstream applications through off-the-shelf machine learning. However, existing methods mostly focus on node-level embedding, which is insufficient for subgraph analysis. Moreover, their leverage of network structures through path sampling or neighborhood preserving is implicit and coarse. Network motifs allow graph analysis in a finer granularity, but existing methods based on motif matching are limited to enumerated simple motifs and do not leverage node labels and supervision. In this paper, we develop NEST, a novel hierarchical network embedding method combining motif filtering and convolutional neural networks. Motif-based filtering enables NEST to capture exact small structures within networks, and convolution over the filtered embedding allows it to fully explore complex substructures and their combinations. NEST can be trivially applied to any domain and provide insight into particular network functional blocks. Extensive experiments on protein function prediction, drug toxicity prediction and social network community identification have demonstrated its effectiveness and efficiency.