Utilization of protein intrinsic disorder knowledge in structural proteomics

Abstract

Intrinsically disordered proteins (IDPs) and proteins with long disordered regions are highly abundant in various proteomes. Despite their lack of well-defined ordered structure, these proteins and regions are frequently involved in crucial biological processes. Although in recent years these proteins have attracted the attention of many researchers, IDPs represent a significant challenge for structural characterization since these proteins can impact many of the processes in the structure determination pipeline. Here we investigate the effects of IDPs on the structure determination process and the utility of disorder prediction in selecting and improving proteins for structural characterization. Examination of the extent of intrinsic disorder in existing crystal structures found that relatively few protein crystal structures contain extensive regions of intrinsic disorder. Although intrinsic disorder is not the only cause of crystallization failures and many structured proteins cannot be crystallized, filtering out highly disordered proteins from structure-determination target lists is still likely to be cost effective. Therefore it is desirable to avoid highly disordered proteins from structure-determination target lists and we show that disorder prediction can be applied effectively to enrich structure determination pipelines with proteins more likely to yield crystal structures. For structural investigation of specific proteins, disorder prediction can be used to improve targets for structure determination. Finally, a framework for considering intrinsic disorder in the structure determination pipeline is proposed.

Fig. 1

Cumulative histograms of (A) missing density regions by length and (B) structures by length of the longest missing density region in any protein chain in the structure. For each, missing density regions were filtered by the fraction of residues predicted to be disordered: no filtering (threshold of 0.0), at least partially disordered (threshold of 0.3), mostly disordered (threshold of 0.7), and completely disordered (threshold of 1.0).

Fig. 2

Cumulative histograms of predicted disorder in the ordered, disordered, and X-ray with disorder datasets. Disorder is summarized for each protein in the set by three different methods: (A) the percentage of residues per-protein predicted to be disordered, (B) the longest consecutive disorder prediction, and (C) the number of residues within predicted regions of disorder of 20 residues or longer. The dotted and dashed lines in each plot represent aggressive and conservative cutoffs, respectively, and are discussed in the text.

Fig. 3

Means of the percentage of predicted disorder for targets in the TargetTrack database. Targets are grouped by the year targets were first selected. Error bars give the 95% confidence interval of the mean.

Fig. 4

Cumulative histograms for targets in the TargetTrack database for (A and C) percentage of predicted disorder and (B and D) number of residues within predicted regions of disorder of 20 residues or longer. Plots for two selection date thresholds are shown: (A and B) 2004 or earlier and (C and D) 2007 or later.

Fig. 5

Cumulative histogram for targets in the TargetTrack database select in 2004 or earlier by the linear combination of percentage of predicted disorder and number of residues within predicted regions of disorder of 20 residues or longer (equation given in text).

Fig. 6

PONDR® VL-XT predictions on the protein RluC. PONDR® VL-XT prediction output values plotted versus residue number, where the order–disorder threshold is indicated by the thin horizontal line at a score of 0.5 (A). The region of this protein that was removed by proteolysis during crystallization trials is indicated by the thick gray bar. In an alternative representation, PONDR® VL-XT output scores are indicated in the upper bar of (B) in binary form where black bars (score ≥ 0.5) indicate disorder and white bars (score < 0.5) indicate order, with the sequence location indicated by the distance from the left end of the bar. The region of the protein that was crystallized is shown by the lower bar of (B), coded white to indicate the region is ordered.

Fig. 7

Comparison of PONDR® predictions and protein fragments having 3-D structure. Proteins were chosen as representative of accurate and inaccurate PONDR® VL-XT predictions. The PONDR® VL-XT prediction for each protein is represented by the upper bar, where black disorder and white indicates order. The lower bars represent determined structures with PDB accessions as labeled, where white indicates defined coordinates and black represents missing coordinates or divergent NMR structures. Some examples are for proteins in isolation (1COK, 1TSR, 1NRC, 2U1A, 1IMX, 1HCP, 1F3M A/B, 1F3M C/D, 1F7W), other structures are for long protein fragments in complexes with other proteins (mdm2 in 1YCR, 1I4E, 1QBQ) or nucleic acid (1JK1, 2RAM), some are for homooligomers (1TGK, 1G50, 1FBU, 1ICE), and some are for short protein fragments in complexes with other proteins (1FOS, p53 in 1YCR, 3SAK, 1DT7).

Fig. 8

Intrinsic disorder-based interaction between c-Myc, Max and DNA. Both c-Myc and Max are intrinsically disordered in their unbound states (left side), but fold into coiled-coil dimer upon interaction with DNA (right side) (PDB ID: 1NKP). c-Myc, Max, and DNA are shown as red ribbon, pink ribbon, and blue surface, respectively.