Zoran Obradovic

Sequence complexity of disordered protein

Intrinsic disorder refers to segments or to whole proteins that fail to self‐fold into fixed 3D structure, with such disorder sometimes existing in the native state. Here we report data on the relationships among intrinsic disorder, sequence complexity as measured by Shannons entropy, and amino acid composition. Intrinsic disorder identified in protein crystal structures, and by nuclear magnetic resonance, circular dichroism, and prediction from amino acid sequence, all exhibit similar complexity distributions that are shifted to lower values compared to, but significantly overlapping with, the distribution for ordered proteins. Compared to sequences from ordered proteins, these variously characterized intrinsically disordered segments and proteins, and also a collection of low‐complexity sequences, typically have obviously higher levels of protein‐specific subsets of the following amino acids: R, K, E, P, and S, and lower levels of subsets of the following: C, W, Y, I, and V. The Swiss Protein database of sequences exhibits significantly higher amounts of both low‐complexity and predicted‐to‐be‐disordered segments as compared to a non‐redundant set of sequences from the Protein Data Bank, providing additional data that nature is richer in disordered and low‐complexity segments compared to the commonness of these features in the set of structurally characterized proteins. Proteins 2001;42:38–48.

Intrinsic disorder in cell-signaling and cancer-associated proteins

The number of intrinsically disordered proteins known to be involved in cell-signaling and regulation is growing rapidly. To test for a generalized involvement of intrinsic disorder in signaling and cancer, we applied a neural network predictor of natural disordered regions (PONDR VL-XT) to four protein datasets: human cancer-associated proteins (HCAP), signaling proteins (AfCS), eukaryotic proteins from SWISS-PROT (EU_SW) and non-homologous protein segments with well-defined (ordered) 3D structure (O_PDB_S25). PONDR VL-XT predicts >or=30 consecutive disordered residues for 79(+/-5)%, 66(+/-6)%, 47(+/-4)% and 13(+/-4)% of the proteins from HCAP, AfCS, EU_SW, and O_PDB_S25, respectively, indicating significantly more intrinsic disorder in cancer-associated and signaling proteins as compared to the two control sets. The disorder analysis was extended to 11 additional functionally diverse categories of human proteins from SWISS-PROT. The proteins involved in metabolism, biosynthesis, and degradation together with kinases, inhibitors, transport, G-protein coupled receptors, and membrane proteins are predicted to have at least twofold less disorder than regulatory, cancer-associated and cytoskeletal proteins. In contrast to 44.5% of the proteins from representative non-membrane categories, just 17.3% of the cancer-associated proteins had sequence alignments with structures in the Protein Data Bank covering at least 75% of their lengths. This relative lack of structural information correlated with the greater amount of predicted disorder in the HCAP dataset. A comparison of disorder predictions with the experimental structural data for a subset of the HCAP proteins indicated good agreement between prediction and observation. Our data suggest that intrinsically unstructured proteins play key roles in cell-signaling, regulation and cancer, where coupled folding and binding is a common mechanism.

DisProt: The database of disordered proteins

The Database of Protein Disorder (DisProt) links structure and function information for intrinsically disordered proteins (IDPs). Intrinsically disordered proteins do not form a fixed three-dimensional structure under physiological conditions, either in their entireties or in segments or regions. We define IDP as a protein that contains at least one experimentally determined disordered region. Although lacking fixed structure, IDPs and regions carry out important biological functions, being typically involved in regulation, signaling and control. Such functions can involve high-specificity low-affinity interactions, the multiple binding of one protein to many partners and the multiple binding of many proteins to one partner. These three features are all enabled and enhanced by protein intrinsic disorder. One of the major hindrances in the study of IDPs has been the lack of organized information. DisProt was developed to enable IDP research by collecting and organizing knowledge regarding the experimental characterization and the functional associations of IDPs. In addition to being a unique source of biological information, DisProt opens doors for a plethora of bioinformatics studies. DisProt is openly available at .

Length-dependent prediction of protein intrinsic disorder

BackgroundDue to the functional importance of intrinsically disordered proteins or protein regions, prediction of intrinsic protein disorder from amino acid sequence has become an area of active research as witnessed in the 6th experiment on Critical Assessment of Techniques for Protein Structure Prediction (CASP6). Since the initial work by Romero et al. (Identifying disordered regions in proteins from amino acid sequences, IEEE Int. Conf. Neural Netw., 1997), our group has developed several predictors optimized for long disordered regions (>30 residues) with prediction accuracy exceeding 85%. However, these predictors are less successful on short disordered regions (≤30 residues). A probable cause is a length-dependent amino acid compositions and sequence properties of disordered regions.ResultsWe proposed two new predictor models, VSL2-M1 and VSL2-M2, to address this length-dependency problem in prediction of intrinsic protein disorder. These two predictors are similar to the original VSL1 predictor used in the CASP6 experiment. In both models, two specialized predictors were first built and optimized for short (≤30 residues) and long disordered regions (>30 residues), respectively. A meta predictor was then trained to integrate the specialized predictors into the final predictor model. As the 10-fold cross-validation results showed, the VSL2 predictors achieved well-balanced prediction accuracies of 81% on both short and long disordered regions. Comparisons over the VSL2 training dataset via 10-fold cross-validation and a blind-test set of unrelated recent PDB chains indicated that VSL2 predictors were significantly more accurate than several existing predictors of intrinsic protein disorder.ConclusionThe VSL2 predictors are applicable to disordered regions of any length and can accurately identify the short disordered regions that are often misclassified by our previous disorder predictors. The success of the VSL2 predictors further confirmed the previously observed differences in amino acid compositions and sequence properties between short and long disordered regions, and justified our approaches for modelling short and long disordered regions separately. The VSL2 predictors are freely accessible for non-commercial use at http://www.ist.temple.edu/disprot/predictorVSL2.php

The protein trinity--linking function and disorder.

http://biotech.nature.com • SEPTEMBER 2001 • VOLUME 19 • nature biotechnology Interpreting function in terms of specific three-dimensional structure has dominated thinking about proteins for more than 100 years, starting with the lock-and-key proposal of Fischer1 and continuing with the equating of denaturation with loss of specific structure by Wu2 and independently at a slightly later date by Mirsky and Pauling3. This dependence of function on structure is even embedded in our language: unfolded protein and denatured protein are used interchangeably. Furthermore, the avalanche of protein three-dimensional structures determined by X-ray diffraction and by nuclear magnetic resonance (NMR)4 has diverted attention away from alternative views. Numerous counterexample proteins have surfaced over the years—proteins for which lack of three-dimensional structure is required for function. One clear example is calcineurin, a serine/threonine phosphatase that becomes activated by the binding of the Ca2+–calmodulin complex to a region that exists as a disordered ensemble5,6. The disorder spans the calmodulin binding site and is essential for calcineurin function. That is, when calmodulin binds to its target helix, the helix becomes completely surrounded7. Thus, the open, flexible disordered region of calcineurin provides the space needed by calmodulin so it can completely surround its target helix. Even though hundreds of other examples of proteins with intrinsic disorder have surfaced over the past 50 years, review articles on this topic are only just now beginning to appear8–10. Wright and Dyson8 suggested that the existence of proteins with intrinsic protein disorder calls for a reassessment of the protein–structure– function paradigm. Since amino acid sequence determines three-dimensional structure, amino acid sequence should also determine lack of three-dimensional structure. Furthermore, if intrinsic disorder provides the basis for some biological functions, then the operation of natural selection should conserve the lack of folding and thereby preserve those functions that depend on this property. If disorder is indeed encoded by the amino acid sequence, then predictors of disorder should exceed the accuracies expected by chance. Work in our group has used literature and database searches to collect a set of proteins structurally characterized to have regions of disorder, some of which were indicated by NMR to be wholly disordered under physiological conditions. Using this set of proteins with intrinsic disorder, we have set out to construct the predictors needed to test the hypothesis.

Exploiting heterogeneous sequence properties improves prediction of protein disorder

During the past few years we have investigated methods to improve predictors of intrinsically disordered regions longer than 30 consecutive residues. Experimental evidence, however, showed that these predictors were less successful on short disordered regions, as observed two years ago during the fifth Critical Assessment of Techniques for Protein Structure Prediction (CASP5). To address this shortcoming, we developed a two‐level model called VSL1 (CASP6 id: 193‐1). At the first level, VSL1 consists of two specialized predictors, one of which was optimized for long disordered regions (>30 residues) and the other for short disordered regions (≤30 residues). At the second level, a meta‐predictor was built to assign weights for combining the two first‐level predictors. As the results of the CASP6 experiment showed, this new predictor has achieved the highest accuracy yet and significantly improved performance on short disordered regions, while maintaining high performance on long disordered regions. Proteins 2005;Suppl 7:176–182.

Predicting intrinsic disorder from amino acid sequence

Blind predictions of intrinsic order and disorder were made on 42 proteins subsequently revealed to contain 9,044 ordered residues, 284 disordered residues in 26 segments of length 30 residues or less, and 281 disordered residues in 2 disordered segments of length greater than 30 residues. The accuracies of the six predictors used in this experiment ranged from 77% to 91% for the ordered regions and from 56% to 78% for the disordered segments. The average of the order and disorder predictions ranged from 73% to 77%. The prediction of disorder in the shorter segments was poor, from 25% to 66% correct, while the prediction of disorder in the longer segments was better, from 75% to 95% correct. Four of the predictors were composed of ensembles of neural networks. This enabled them to deal more efficiently with the large asymmetry in the training data through diversified sampling from the significantly larger ordered set and achieve better accuracy on ordered and long disordered regions. The exclusive use of long disordered regions for predictor training likely contributed to the disparity of the predictions on long versus short disordered regions, while averaging the output values over 61‐residue windows to eliminate short predictions of order or disorder probably contributed to the even greater disparity for three of the predictors. This experiment supports the predictability of intrinsic disorder from amino acid sequence. Proteins 2003;53:566–572.

Flavors of protein disorder

Intrinsically disordered proteins are characterized by long regions lacking 3‐D structure in their native states, yet they have been so far associated with 28 distinguishable functions. Previous studies showed that protein predictors trained on disorder from one type of protein often achieve poor accuracy on disorder of proteins of a different type, thus indicating significant differences in sequence properties among disordered proteins. Important biological problems are identifying different types, or flavors, of disorder and examining their relationships with protein function. Innovative use of computational methods is needed in addressing these problems due to relative scarcity of experimental data and background knowledge related to protein disorder. We developed an algorithm that partitions protein disorder into flavors based on competition among increasing numbers of predictors, with prediction accuracy determining both the number of distinct predictors and the partitioning of the individual proteins. Using 145 variously characterized proteins with long (>30 amino acids) disordered regions, 3 flavors, called V, C, and S, were identified by this approach, with the V subset containing 52 segments and 7743 residues, C containing 39 segments and 3402 residues, and S containing 54 segments and 5752 residues. The V, C, and S flavors were distinguishable by amino acid compositions, sequence locations, and biological function. For the sequences in SwissProt and 28 genomes, their protein functions exhibit correlations with the commonness and usage of different disorder flavors, suggesting different flavor‐function sets across these protein groups. Overall, the results herein support the flavor‐function approach as a useful complement to structural genomics as a means for automatically assigning possible functions to sequences. Proteins 2003;52:573–584.

Identification and functions of usefully disordered proteins

Publisher Summary This chapter illustrates that protein disorder is encoded by the amino acid sequence and that protein disorder is essential for many important biological functions. An ordered protein contains a single canonical set of Ramachandran angles, whereas a disordered protein or region contains an ensemble of divergent angles at any instant and these angles interconvert over time. Intrinsically disordered protein can be extended (random coil–like) or collapsed (molten globule–like). The latter type of disorder typically includes regions of fluctuating secondary structure, so disorder does not mean absence of helix or sheet. Both types of disorders have been observed in apparently native proteins. Intrinsic disorder might not be encoded by the sequence, but rather might be the result of the absence of suitable tertiary interactions. If this were the general cause of intrinsic disorder, any subset of ordered sequences and any subset of disordered sequences would likely be the same within the statistical uncertainty of the sampling. On the other hand, if intrinsic disorders were encoded by the amino acid sequence, any subset of disordered sequences would likely differ significantly from samples of ordered protein sequences.

OPTIMIZING LONG INTRINSIC DISORDER PREDICTORS WITH PROTEIN EVOLUTIONARY INFORMATION

Protein existing as an ensemble of structures, called intrinsically disordered, has been shown to be responsible for a wide variety of biological functions and to be common in nature. Here we focus on improving sequence-based predictions of long (>30 amino acid residues) regions lacking specific 3-D structure by means of four new neural-network-based Predictors Of Natural Disordered Regions (PONDRs): VL3, VL3H, VL3P, and VL3E. PONDR VL3 used several features from a previously introduced PONDR VL2, but benefitted from optimized predictor models and a slightly larger (152 vs. 145) set of disordered proteins that were cleaned of mislabeling errors found in the smaller set. PONDR VL3H utilized homologues of the disordered proteins in the training stage, while PONDR VL3P used attributes derived from sequence profiles obtained by PSI-BLAST searches. The measure of accuracy was the average between accuracies on disordered and ordered protein regions. By this measure, the 30-fold cross-validation accuracies of VL3, VL3H, and VL3P were, respectively, 83.6 +/- 1.4%, 85.3 +/- 1.4%, and 85.2 +/- 1.5%. By combining VL3H and VL3P, the resulting PONDR VL3E achieved an accuracy of 86.7 +/- 1.4%. This is a significant improvement over our previous PONDRs VLXT (71.6 +/- 1.3%) and VL2 (80.9 +/- 1.4%). The new disorder predictors with the corresponding datasets are freely accessible through the web server at http://www.ist.temple.edu/disprot.