An interaction network predicted from public data as a discovery tool: application to the Hsp90 molecular chaperone machine

Abstract

Understanding the functions of proteins requires information about their protein-protein interactions (PPI). The collective effort of the scientific community generates far more data on any given protein than individual experimental approaches. The latter are often too limited to reveal an interactome comprehensively. We developed a workflow for parallel mining of all major PPI databases, containing data from several model organisms, and to integrate data from the literature for a protein of interest. We applied this novel approach to build the PPI network of the human Hsp90 molecular chaperone machine (Hsp90Int) for which previous efforts have yielded limited and poorly overlapping sets of interactors. We demonstrate the power of the Hsp90Int database as a discovery tool by validating the prediction that the Hsp90 co-chaperone Aha1 is involved in nucleocytoplasmic transport. Thus, we both describe how to build a custom database and introduce a powerful new resource for the scientific community.

Figure 1. Workflow for the construction of the interactome of the human Hsp90 molecular chaperone machine (Hsp90Int).

(A) PPIs from online databases were compiled and edited in order to obtain consistent and uniform data. These data were enriched with manually curated information from the literature. (B) Physical interactomes were constructed and visualized with Cytoscape for the indicated species. (C) Components of the Hsp90 molecular chaperone machine were used as query proteins (colored nodes) to retrieve the corresponding PPI network for each organism. (D) HomoloGene IDs were assigned to the interactors in each network to identify human orthologs and new interolog interactions. (E) The complete sets of interologs were merged with the human network into the complete human Hsp90Int. The red and blue colors represent human and interolog interactions, respectively.

Figure 2. Visualization of Hsp90Int.

The full interactome containing 1'150 nodes and 8'892 edges is presented here as a zoomable PPI network with query proteins but not their interactors being colored (see inset). The sizes of the nodes reflect their level of association with partners inside the network (degree). The origins of the interactions are indicated by the color of the edge line: interologs and human interactions are represented with blue dashed and red full lines, respectively. Chaperone, core molecular chaperones such as Hsp90α and Hsp90β; TPR, tetratricopeptide repeats; CS, “CHORD and Sgt1 domain”.

Figure 3. Topological characteristics of Hsp90Int.

(A) Schematic representation summarizing the topological characteristics of Hsp90Int compared to randomly selected networks. (B) Mean values of the graph measures calculated for Hsp90Int (red dots) with 1'150 nodes and 8'892 edges and for the control networks (black boxplots). (C) Node distributions for mean degree and clustering coefficient.

Figure 4. Functional map of Hsp90Int.

In this functionally grouped GO annotation network, nodes represent a GO term (biological process) significantly overrepresented in a group of proteins from the interactome (at least 8). The node sizes represent the term enrichment significance. Edges connect GO terms that share common sets of proteins present in Hsp90Int. Similar functional terms are grouped in colored regions and labeled with a representative name. For clarity, the original ClueGO output was edited with Adobe Illustrator. For example, some terms are highlighted in red to guide the viewer.

Figure 5. A functional map of the Aha1-Hsp90 subset.

The Aha1-focused PPI network (see Figure S1) was functionally grouped into a GO annotation network. The proteins in the node “nucleocytoplasmic transport” are listed as an inset. Other map details are as in Figure 4.

Figure 6. Experimental validation of the involvement of Aha1 in nucleocytoplasmic transport.

(A) PPI map of the proteins (red nodes) in the GO module “nucleocytoplasmic transport” of the Aha1-Hsp90 PPI subset of Figure 5. It contains potential and known (dashed and full line edges, respectively) Aha1 interactions. The Hsp90 client protein GR was also integrated into the predicted PPI (upper panel). The co-immunoprecipitation assays of panel B allowed the experimental confirmation and new demonstration of PPIs as indicated by red and green edges, respectively (lower panel). (B) Co-immunoprecipitation experiment demonstrating interactions between Aha1 or exportin-1 (XPO1) and components of the GO module “nucleocytoplasmic transport” shown in panel A. Flag-tagged Aha1, exportin-1, and GPR30 as an unrelated control protein were exogenously expressed in 293T cells. IPO4, importin-4; KPNA5, importin-α6. (C) and (D) Nuclear localization of GR in mouse fibroblasts with and without Aha1. Panel C shows representative micrographs of the localization of Tom.GR with our without treatment with dexamethasone (Dex) for 40 min, and an immunoblot on the right verifying the absence of Aha1 in the Aha1-null fibroblasts. Panel D shows the nuclear accumulation of Tom.GR over time in wild-type (▴) and Aha1-null (▾) cells. Nuclear localization of GR was initiated at time zero with the addition of 10 nM dexamethasone. Data points are the means with standard errors of three independent experiments where ∼100 cells were counted for each time point. *, significantly different with p<0.005.