IDPs: Less Disordered and More Ordered than Expected

Main Text

In this issue of the Biophysical Journal, Sólyom et al. (9) provide a telling example of structural preformation in intrinsically disordered proteins (IDPs) and the relevance of distinctly folded and compact conformational substates for molecular recognition processes.

IDPs are challenging the established structural biology structure-function paradigm and thus mandate a suitable theoretical framework and concepts to properly address the subtle interdependence between protein structure and dynamics (1). In contrast to stably folded globular proteins, IDPs sample larger conformational spaces made up of loosely folded as well as compact conformational substates. In this respect they share features with the unfolded state of proteins that is also not a featureless structural ensemble but retains substantial structural ordering (2). The enormous reduction of conformational space results from the existence of predefined basic structural motifs that are combined in a combinatorial (pseudo)-repetitive fashion to build larger structural segments (2,3). The surprisingly small number of structural motifs is a consequence of the steric constraints of the natural amino acids linked together via peptide bonds; it can be further rationalized by the geometry of the polypeptide chain, which resembles a disk or a tube of nonvanishing thickness (4). Although significant sequence differences exist between folded and intrinsically disordered proteins (5,6) there is compelling evidence for a common and universal principle for protein structure (pre)-formation resulting from geometry and symmetry rather than the specific sequence of amino acids. In this view, structural presculpting results from fundamental geometric properties of the protein backbone while specific side-chain interactions (contacts) are relevant for the thermodynamics of the system by determining the populations (free energies) of individual structures.

It is also enlightening to discuss predefined protein structural motifs in the context of natural evolution. The theory of facilitated variation explains how evolution proceeds by employing new combinations of so-called core processes (i.e., biochemical transformations or cell biological and developmental modules) (7). Crucial properties of core processes are modularity, robustness, adaptability, capacity to engage in weak regulatory linkages, and exploratory behavior. Interestingly, these properties are paralleled by protein structural motifs (8). Modularity is realized in the hierarchical arrangement of basic second structure elements while robustness and adaptability are reflected in the conservation of structure and the functional variability of protein structural elements. Weak regulatory linkages are seen in cooperative stabilizations of small structural building blocks and allosteric regulations. Exploratory behavior results from the distribution of (thermally accessible) conformational states in the ensemble of proteins in general, and intrinsically disordered proteins in particular. Thus, facilitated variation acts on two molecular levels—the generation of novel protein structures built from fundamental motifs (molecular) and the integration of protein molecules into larger interaction networks (supramolecular). In both cases, structural preformation and conformational bias is essential for the generation of structural and phenotypic variation.

The article by Sólyom et al. (9) describes an illustrative example of this structural preformation in a disordered protein domain. The authors present a detailed and thorough nuclear magnetic resonance (NMR) analysis of the C-terminal intrinsically disordered region of the NS5A protein from hepatitis C virus, a multifunctional protein that is essential for viral replication and particle assembly by binding to and recruiting numerous protein binding partners. NS5A consists of a folded N-terminal domain D1 and two intrinsically disordered domains D2 and D3 separated by two low complexity regions (LCS I and LCS II). The N-terminal amphipathic helix anchors NS5A to host ER membranes. The globular D1 domain is made up of a RNA binding groove; the intrinsically disordered regions are involved in RNA replication (D2) and viral particle assembly (D3), and house interaction sites for the viral RNA polymerase NS5B and CypA (D2). Due to this intricate involvement of NS5A in the virus life cycle, it was recognized as a promising pharmacological target for the treatment of HCV. Here they report NMR data for the entire disordered region (191–447) from genotype 1b, both as a separate domain and in the context of the full-length protein. Using sophisticated NMR experiments, they could convincingly demonstrate the prevalence of local secondary structures and electrostatically driven long-range interactions in the disordered domains that are altered by both binding to the SH3 domain of the tumor suppressor Bin1 (also suspected to be involved in hepatocellular carcinoma in HCV infected liver cells) and phosphorylation by CK2.

In their studies, the authors predominantly exploit the sensitive dependence of NMR chemical-shift data on structural features, in contrast to other IDP NMR studies that often rely on paramagnetic relaxation enhancement (PRE) data. Although PRE measurements provide unique information about (also transient) structural features in IDPs in solution, it cannot be ruled out that the introduction of paramagnetic spin labels via chemical modification of Cys side chains might slightly perturb the conformational ensemble. The fact that the authors base their argumentation entirely on chemical-shift data adds to the credibility of their findings. The analysis of chemical-shift data obtained for residues located in the disordered domain NS5A-D2 in isolation and in the context of NS5A-D2D3 revealed significant and specific residue-residue contacts in this disordered domain, and provides evidence for the existence of transiently formed compact substates. Most importantly, chemical-shift changes were observed for the same residues upon CK2 phosphorylation and binding to Bin1-SH3. Particularly noteworthy is their remarkable observation that residue segments already involved in the structural compaction of apo-NS5A-D2D3 also undergo (local and long-range) structural alterations upon phosphorylation and/or protein binding. This provides strong evidence for structural preformation in this IDP, and suggests that protein-interaction events in NS5A-D2D3 largely proceed via conformation selection.

The findings of Sólyom et al. (9) can thus best be summarized as follows: The intrinsically disordered domain NS5A-D2D3 exists as an ensemble of substates differing in terms of local second structure content and overall compaction of the polypeptide chain. This structural preformation is relevant for accommodating binding partners and can be allosterically regulated by posttranslational modification. Interestingly, by employing state-of-the-art NMR spin relaxation measurements, the authors could demonstrate alterations of the conformational dynamics. Upon interaction with Bin1-SH3, the structural compaction of NS5A-D2D3 is reduced by expelling the C-terminal D3 domain from the core, and this structural and dynamical adaptation changes the accessibility to other protein-binding partners.

The detailed insight into the molecular details of structural preformation in IDPs is also relevant for the popular partitioning of proteins into order and disorder. NMR has provided compelling evidence for the existence of sparsely populated (high-energy) states in globular, well-folded proteins and their relevance to enzymatic activities (10). In view of the limited set of conceivable structural domains established a priori, the differences between ordered and disordered proteins are therefore rather in the geometry of the energy landscape and different distributions of thermally accessible substates than in the structures of the interconverting conformational states. To appropriately grasp the dynamic conformational ensembles of fluctuating biological polypeptides, more sophisticated approaches are clearly needed. An attempt to overcome the binary order-disorder descriptor was the metastructure approach, which provides quantitative, per-residue information about compactness and local secondary structure entirely based on primary sequence information (11). The metastructure parameters, for example, can be employed in sequence alignments for the identification of structural domains. Structural models can be generated by extracting distance constraints obtained from pairwise alignments with templates from the PDB database. A preliminary application to the disordered NS5A-D2D3 domain is shown in Fig. 1 suggesting that the obtained low-resolution structural model nicely reproduces the experimentally observed structural preformation and its dependence on posttranslational modifications. Similar alignment approaches can be used to generally identify putative structural motifs, and will provide valuable information about structural preformation in IDPs. It is foreseeable that the identified structural building blocks can subsequently be used to improve the quality of the obtained conformational ensembles using experimental constraints (e.g., PRE, chemical shifts, residual dipolar coupling, small-angle x-ray scattering).

Figure 1.

Structural preformation in IDPs. Sólyom et al. (9) explore the conformational features of the intrinsically disordered region (D2D3) of the NS5A protein from hepatitis C virus and demonstrate that although NS5A-D2D3 is largely unfolded in solution, a significant number of compact structures exist in their conformational ensembles. Low-resolution structural models for (orange) apo and (blue) phosphorylated NS5A-D2D3 were obtained from pairwise metastructure alignments to sequences taken from the PDB database (11) and indicate structural compaction upon phosphorylation. For the generation of the structural model, phosphorylation was taken into account by replacing Ser and Thr with Asp and Glu residues, respectively. (Green surfaces) Phosphorylated residues (C-terminal region 401, 408, 429–437); (red surfaces) residues that are affected due to long-range interactions (290–310 and PPP-region) (9). N- and C-termini of NS5A-D2D3 are also indicated (apo: C, N and phosphorylated: C′, N′). The low-resolution structural model reproduces the observed structural compaction (N′- and C′-termini are closer in the phosphorylated state, blue), and this indicates spatial proximity between phosphorylated residues (green) and regions displaying chemical shift changes (red). To see this figure in color, go online.

The article by Sólyom et al. (9) convincingly demonstrates that even in seemingly random-coil-like disordered proteins, NMR spectroscopy can reveal hidden structural preformation. However, a number of more sophisticated NMR experiments and computational tools will still be needed to properly describe correlated structural fluctuations and cooperative folding in intrinsically disordered proteins. Arguably, IDPs might be another example for a seemingly tangled skein of structures concealing Henri Poincaré’s “hidden simplicity”. The future will tell.

Editor: H. Jane Dyson.