Protein–Protein Interactions and Genetic Disease

Abstract

The advent of high‐throughput experiments to measure protein–protein interactions has created a flood of proteomic information parallel to the influx of data caused by the advent of next‐generation sequencing technologies. The creation of whole organism protein interaction maps has opened new avenues of predicting genetic disease. Advances in network science now allow the association of genes with disease directly from the characteristics of the protein map without reference to the characteristics of the gene itself. However, none of these techniques has reached the ‘black box’ level and require careful consideration of the systematic errors in both the underlying experimental data and the computational methods to give reliable results. Here, we review the main methods to characterise protein interactions in vitro and in vivo , the methods by which protein networks are constructed and the characteristics of the major protein interaction databases, and the techniques used to predict the functional impact of mutations on protein interaction networks. Key Concepts Protein–protein interactions are the driving force for cellular responses and disruption of protein–protein interfaces. The sequence conservation of functionally important residues across protein interfaces allows the prediction of disease‐associated mutations. Disease‐associated mutations often affect protein–protein binding affinities that can be measured with high accuracy in vitro using purified proteins by a variety of biophysical techniques. High‐throughput techniques to measure protein interactions in vivo are prone to both high false‐positive and high false‐negative rates. Each method of identifying protein interactions has systematic errors associated with it that can be partly corrected by cross validating the results with two techniques. Functional associations between genes can be inferred by bioinformatic approaches, but the results do not necessarily imply a physical interaction between proteins. Protein–protein interaction databases may include both complexes between proteins and purely functional associations. This distinction must be kept in mind when analysing results from protein–protein interaction databases. Disease‐associated genes tend to be clustered together in protein–protein interaction networks. It is possible to predict new genes for a disease by a random walk across the protein interaction network from genes already known to be associated with the disease. Currently, protein–protein interaction surfaces are mostly considered undruggable by small molecules. This belief is changing with new concepts in drug design.

Protein-protein interactions of the nicotinamide/nicotinate mononucleotide adenylyltransferase of Leishmania braziliensis

“…Several web servers are available for the development of such analyses 11 . In this study, we used an integrative approach for obtaining the physical and functional associations to create a PPI network; accordingly, we used the STRING web resource, which manages diverse sources of information facilitating a comparative analysis 12 …”

Recent advances in automated protein design and its future challenges

“…Many diseases are either related to inappropriate protein – protein interactions or the absence of them [77–80]. Many cell surface receptors and their effector proteins are involved in signaling pathways, where activity is regulated by a small molecular stimulus.…”