IUPred2A: context-dependent prediction of protein disorder as a function of redox state and protein binding

Abstract

The structural states of proteins include ordered globular domains as well as intrinsically disordered protein regions that exist as highly flexible conformational ensembles in isolation. Various computational tools have been developed to discriminate ordered and disordered segments based on the amino acid sequence. However, properties of IDRs can also depend on various conditions, including binding to globular protein partners or environmental factors, such as redox potential. These cases provide further challenges for the computational characterization of disordered segments. In this work we present IUPred2A, a combined web interface that allows to generate energy estimation based predictions for ordered and disordered residues by IUPred2 and for disordered binding regions by ANCHOR2. The updated web server retains the robustness of the original programs but offers several new features. While only minor bug fixes are implemented for IUPred, the next version of ANCHOR is significantly improved through a new architecture and parameters optimized on novel datasets. In addition, redox-sensitive regions can also be highlighted through a novel experimental feature. The web server offers graphical and text outputs, a RESTful interface, access to software download and extensive help, and can be accessed at a new location: http://iupred2a.elte.hu.

Figure 1.

ROC curves of four methods predicting disordered binding regions/MoRFs on the four different negative testing datasets. The upper row shows testing on verified negative data containing virtually no disordered binding regions. Negative sets of the bottom row might contain an unknown number of disordered binding regions, albeit with a significantly lower frequency compared to the positive set.

Figure 2.

The output of IUPred2 and ANCHOR2 for the oncogenic Human adenovirus C early E1A protein. Top: IUPred2 and ANCHOR2 scores are shown in red and blue. Bottom: schematic architecture of E1A. Disordered binding regions with known complex structure are shown in deep red boxes. Light red boxes correspond to known linear motifs. Grey box marks the region sufficient for interaction with UBE2I.

Figure 3.

The output of the redox-state dependent IUPred2 predictor for the N-terminal region of NOS3. Top: the coordination of Zn²⁺ by cysteines 94 and 99 from both chains in the dimeric NOS3 structure. Bottom: the output of IUPred2 using the redox state modeling option, where the estimated sensitivity of the disorder tendency is marked in purple. The plot is zoomed into the N-terminal region that can be seen in the dimeric complex (PDB: 3NOS).