Exploring mechanisms of human disease through structurally resolved protein interactome networks

Abstract

The study of the molecular basis of human disease has gained increasing attention over the past decade. With significant improvements in sequencing efficiency and throughput, a wealth of genotypic data has become available. However the translation of this information into concrete advances in diagnostic and clinical setups has proved far more challenging. Two major reasons for this are the lack of functional annotation for genomic variants and the complex nature of genotype-to-phenotype relationships. One fundamental approach to bypass these issues is to examine the effects of genetic variation at the level of proteins as they are directly involved in carrying out biological functions. Within the cell, proteins function by interacting with other proteins as a part of an underlying interactome network. This network can be determined using interactome mapping - a combination of high-throughput experimental toolkits and curation from small-scale studies. Integrating structural information from co-crystals with the network allows generation of a structurally resolved network. Within the context of this network, the structural principles of disease mutations can be examined and used to generate reliable mechanistic hypotheses regarding disease pathogenesis.

Fig. 1

Growth of genomic data and our understanding of pathogenesis (A) accumulation of dbSNP data, HGMD mutations, disease genes and drug targets over the past 12 years (number of dbSNP variations: ftp://ftp.ncbi.nlm.nih.gov/snp/organisms/human_9606/chr_rpts/; number of HGMD mutations: http://www.hgmd.cf.ac.uk/ac/hahaha.php; number of disease genes: ftp://ftp.eimb.ru/omim/; number of FDA-approved drugs: http://www.fda.gov/AboutFDA/WhatWeDo/History/ProductRegulation/SummaryofNDAApprovalsReceipts1938tothepresent). (B) Distribution of OMIM pheno-type entries by knowledge of molecular basis (http://www.omim.org/statistics/entry).

Fig. 2

Schematic representations of high-throughput assays used to generate binary interactome networks. (A) Yeast two-hybrid (Y2H). (B) Protein fragment complementation assays (PCA). (C) Luminescence-based mammalian interactome mapping (LUMIER). (D) Well-based nucleic acid programmable protein array (wNAPPA). (E) Mammalian protein–protein interaction trap (MAPPIT). (F) A high-quality reference human binary interactome comprising B40 000 interactions generated from several large-scale interactome mapping efforts and thousands of small-scale studies.

Fig. 3

Structurally resolved interactome networks and human disease. (A) Construction of a structurally resolved interactome network onto which disease mutations are mapped. (B) Percentage of mutation pairs on two proteins that cause the same disease. (C) Percentage of mutation pairs on the same protein that cause different diseases. (D) A higher resolution of the guilt-by-association principle – mutations at different structural loci on the same protein that cause different diseases [(B) and (C) are adapted from ref. 54].

Fig. 4

Functional consequences of human disease mutations. (A) Illustration of the interface of the CBS homodimer as obtained from a co-crystal and the location of mutations that do/do not disrupt the interaction. (B) Illustration of the predicted interface of the MLH1–PMS2 interaction and the location of mutations that do/do not disrupt the interaction. (C) Schematic representation of changes caused by disease mutations to the interactome network. (D) Summary of the pipeline used to construct 3D interactome networks to understand disease pathogenesis.