A computational analysis of protein-protein interaction networks in neurodegenerative diseases

Abstract

Background: Recent developments have meant that network theory is making an important contribution to the topological study of biological networks, such as protein-protein interaction (PPI) networks. The identification of differentially expressed genes in DNA array experiments is a source of information regarding the molecular pathways involved in disease. Thus, considering PPI analysis and gene expression studies together may provide a better understanding of multifactorial neurodegenerative diseases such as Multiple Sclerosis (MS) and Alzheimer disease (AD). The aim of this study was to assess whether the parameters of degree and betweenness, two fundamental measures in network theory, are properties that differentiate between implicated (seed-proteins) and non-implicated nodes (neighbors) in MS and AD. We used experimentally validated PPI information to obtain the neighbors for each seed group and we studied these parameters in four networks: MS-blood network; MS-brain network; AD-blood network; and AD-brain network.

Results: Specific features of seed-proteins were revealed, whereby they displayed a lower average degree in both diseases and tissues, and a higher betweenness in AD-brain and MS-blood networks. Additionally, the heterogeneity of the processes involved indicate that these findings are not pathway specific but rather that they are spread over different pathways.

Conclusion: Our findings show differential centrality properties of proteins whose gene expression is impaired in neurodegenerative diseases.

Figure 1

Retrieval and representation of each disease network. The differentially expressed genes in MS or AD using blood or brain tissue were obtained from published DNA array studies. The corresponding protein (seed-protein) for each differentially expressed gene was identified in public databases (STRING). The network in which such proteins were embedded was built by retrieving the first neighbor of each protein in the protein-protein interaction database available at the STRING database.

Figure 2

MS-blood network. Purple nodes indicate the seed-proteins with their name. Orange nodes indicate neighboring proteins belonging to the giant component. Green nodes indicate neighbors that do not belong to the giant component. The graphs were built using Pajek software and the network files are available as .net files from the authors upon request.

Figure 3

MS-brain network. Purple nodes indicate the seed-proteins with their name. Orange nodes indicate neighbors proteins belonging to the giant component. Green nodes indicate neighbors that are not included in the giant component.

Figure 4

AD-blood network. Purple nodes indicate the seed-proteins with their name. Orange nodes indicate neighbors proteins belonging to the giant component. Green nodes indicate neighbors that are not included in the giant component.

Figure 5

AD-brain network. Purple nodes indicate the seed-proteins with their name. Orange nodes indicate neighbors proteins belonging to the giant component. Green nodes indicate neighbors that are not included in the giant component.