An intrinsically disordered proteins community for ELIXIR

Introduction

Intrinsically disordered regions (IDRs), protein segments that lack persistent secondary or tertiary structure (Chouard, 2011; Dyson & Wright, 2005; Tompa, 2011), are predicted to cover almost a third of the residues in eukaryotic proteomes (Pancsa & Tompa, 2012; Xue et al., 2012). IDRs play a central role in cell regulation and contribute significantly to the cellular complexity of higher eukaryotes (Dunker et al., 2008; Dyson & Wright, 2005; Forman-Kay & Mittag, 2013; Gouw et al., 2018; Mitrea & Kriwacki, 2016; Schad et al., 2018; Tompa, 2005; Van Roey et al., 2014). They represent a major source of protein diversity and versatility on the level of organisms and during evolution (Babu et al., 2012; Buljan et al., 2012; Davey et al., 2015; Light et al., 2013; Weatheritt & Gibson, 2012). In the human proteome, IDRs are expected to contain up to one hundred thousand interaction interfaces and a million sites of post-translational modification (Tompa et al., 2014). However, to date, only a small fraction of these functional modules have been characterised (Gouw et al., 2018; Schad et al., 2018).

IDR-mediated interactions are commonly found in dynamic and transient complexes that underlie enzyme inhibition, signal transduction and liquid-liquid phase transition (Borgia et al., 2018; Ivarsson & Jemth, 2019; Kriwacki et al., 1996; Martin & Mittag, 2018; Mitrea & Kriwacki, 2016; Olsen et al., 2017; Scott & Pawson, 2009; Wright & Dyson, 2015). IDR-mediated interactions are also major determinants of protein regulation, and are so far known to affect: (i) the post-translational modification state of a protein by acting as docking sites to recruit modifying enzymes; (ii) the cellular half-life of a protein by recruiting E3 ubiquitin ligases resulting in ubiquitin-dependent proteasomal degradation of the protein; and (iii) the localisation of a protein by acting as signals that target proteins to specific subcellular locations (Beltrao et al., 2012; Davey & Morgan, 2016; Gouw et al., 2018; Guharoy et al., 2016; Iakoucheva et al., 2004; Mészáros et al., 2017; Van Roey et al., 2014). Many functions of IDRs are directly associated with their structural attributes, without directly contributing to binding events that result in complex formation; for example, IDRs can act as entropic springs, flexible linkers or spacers (Tompa, 2005; van der Lee et al., 2014), or fly-casting regions to capture binding partners (Shoemaker et al., 2000). Furthermore, IDRs are subject to extensive pre- and post-translational regulation to modulate protein function in response to cellular stimuli (Bah & Forman-Kay, 2016; Csizmok & Forman-Kay, 2018; Van Roey et al., 2013; Van Roey et al., 2012; Weatheritt et al., 2012).

As a result of the fundamental regulatory functions performed by IDRs, and the cell-state conditionality of these regulatory processes, IDRs encode many of critical steps of a protein life-cycle from the ribosome to the proteasome. Consequently, IDRs play a key role in many human diseases, including cancer (p53), Alzheimer’s disease (Aβ, Tau) or Parkinson’s disease (α-synuclein) (Shigemitsu & Hiroaki, 2018; Uversky et al., 2008) and human IDR interfaces are often mimicked by pathogens to hijack host pathways and deregulate the cell (Davey et al., 2011; Dyson & Wright, 2018; Kruse et al., 2019; Via et al., 2015; Xue et al., 2012). Given their therapeutic relevance, IDP-mediated interactions are now seen as potential drug targets (Corbi-Verge & Kim, 2016). Therefore, a better understanding of their structure and function will help to develop new strategies to fight human diseases.

IDP research spans several experimental fields studying protein structure and function including structural biology, biophysics, biochemistry, cell biology, proteomics, comparative genomics, systems biology, synthetic biology and pharmacology (Figure 1A) (Blikstad & Ivarsson, 2015; Corbi-Verge & Kim, 2016; Felli & Pierattelli, 2015; Forman-Kay & Mittag, 2013; Plitzko et al., 2017). On a structural level, IDRs do not adopt a single stable highly populated structure, instead, they are structurally heterogeneous, continuously sample a wide ensemble of conformations, with preferentially sampled intramolecular contacts driving local transient secondary structure and compaction of conformations (Davey, 2019; Dyson & Wright, 2005; Forman-Kay & Mittag, 2013; Holehouse & Pappu, 2018). Therefore, IDRs are more easily described by probabilistic models than the intuitive visual representations of structures of folded protein regions. Classical methods for structural characterisation of a protein, such as X-Ray crystallography, are unable to capture the dynamic structures of IDRs. Instead, the structural aspects of the IDRs are studied by a range of biophysical methods including Nuclear Magnetic Resonance (NMR), Small-angle X-ray scattering (SAXS), circular dichroism (CD) or Förster resonance energy transfer (FRET) (Felli & Pierattelli, 2015; Fuertes et al., 2017; Holmstrom et al., 2018; Plitzko et al., 2017; Tompa, 2011; Tolchard et al., 2018). The results of these complementary methods can be integrated to build a model of IDR structure and dynamics.

Figure 1.

(A) Key experimental methods used across the three major research focuses of the IDP field: structure, interactome and function. (B) The growth in structural IDP data curated by the DisProt database (Piovesan et al., 2017). DisProt was established over a decade ago in the USA, and recently brought to Europe after years of inactivity and completely re-annotated, this explains the lag in the curation. A huge amount of IDP/IDR data in the IDP literature remain uncurated and the vast majority of IDP regions remain to be characterised (Pancsa & Tompa, 2012; Xue et al., 2012). (C) The growth in functional IDP data curated by the Eukaryotic Linear Motif (ELM) resource (Gouw et al., 2018). Similar to the structural IDR data, a huge amount of functional IDP/IDR data in the IDP literature remains uncurated and the vast majority of functional modules in IDPs remain to be characterised (Tompa et al., 2014).

The functional aspects of IDPs are studied by a battery of well-established cell biology and biophysical approaches (Gibson et al., 2015) (Figure 1). Many of the approaches were developed for protein interaction elucidation, for example, by coupling mutagenesis to affinity-purification. However, interactions mediated by IDRs are often defined by their low affinity and cell-state dependent conditionality, two properties that are not always amenable to the available experimental protein interaction detection methods (Gibson et al., 2015; Van Roey et al., 2014). Therefore, there are a growing number of approaches that have been specifically designed to characterise low affinity IDR-mediated interactions such as peptide arrays, proteomic phage display (ProP-PD) and peptides attached to Microspheres with Ratiometric Barcode Lanthanide Encoding (MRBLE-pep) (Blikstad & Ivarsson, 2015; Davey et al., 2017; Nguyen et al., 2017; Volkmer, 2009). As IDRs often adopt a dominant conformation in their bound state, many IDR-containing interfaces can be studied in complex with the structured binding partners using structural approaches including NMR spectroscopy and X-Ray Crystallography (Bonetti et al., 2018; Fuxreiter, 2018; Iešmantavicius et al., 2014; Schad et al., 2018). Distinct IDR functionalities are also studied using isolation, deletion or mutagenesis of a functional module combined with bespoke assays to characterise phase separation, protein localisation, stability and post-translational modification (Gibson et al., 2015; Gouw et al., 2018; Martin & Mittag, 2018).

The computational IDP field tackles several distinct research tasks including: (i) the prediction of IDRs (Cilia et al., 2014; Dosztányi et al., 2005), (ii) the identification of functional modules in IDRs (Dosztányi et al., 2009; Edwards & Palopoli, 2015; Krystkowiak & Davey, 2017; Neduva & Russell, 2006), (iii) docking of peptides that are intrinsically disordered in their unbound state (Raveh et al., 2011; Trabuco et al., 2012), (iv) force field development and molecular simulations for IDR structure (Best, 2017; Chong et al., 2017; Huang & MacKerell, 2018; Stanley et al., 2015), (v) in silico inhibitor design and development for IDRs (Baggett & Nath, 2018; Santofimia-Castaño et al., 2019; Yu et al., 2016) and (vi) the processing of the data produced by experimental structural and functional analyses of IDRs (Bernadó et al., 2007; Franke et al., 2017; Nodet et al., 2009; Ozenne et al., 2012) (Table 1). The members of the European IDP community are major contributors across these computational IDP fields (Figure 2).

Figure 2. Representative core software (Table 1, marked with a diamond) and resources (Table 2) for the IDP field developed and hosted in Europe.

Europe hosts many of the key IDP resources (Table 2). DisProt is the largest database of manually curated experimental data on IDP structure (Piovesan et al., 2017; Sickmeier et al., 2007). MobiDB and D²P² are central resources of pre-computed structural prediction of IDPs (Piovesan et al., 2018; Oates et al., 2013). ELM is a manually curated database of binding regions residing in IDPs (Gouw et al., 2018). SASBDB and PCDDB are central resources for Small-angle X-ray Scattering (SAXS) and Circular Dichroism (CD) data (Valentini et al., 2015; Whitmore et al., 2017). PED³ is a database of conformational ensembles from Nuclear Magnetic Resonance (NMR) and SAXS data, and Molecular Dynamics (MD) simulations (Varadi et al., 2014). It is important to note that the vast majority of the computational frameworks developed by the IDP field had to be created from scratch given the unique structural and functional aspects of IDRs. For example, the analysis of IDP structure drew predominantly from polymer biophysics rather than the existing structural biology of folded proteins (Holehouse & Pappu, 2018; Milles et al., 2018; Schuler et al., 2016).

The ELIXIR IDP Community has grown out of the NGP-NET COST Action (Non-globular proteins: From sequence to structure, function and application in molecular physiopathology), a scientific cooperation funded in 2015 under a Horizon 2020 EU Framework Programme. The NGP-NET community spanned 30 different countries, plus EMBL-EBI and EMBL Heidelberg. NGP-NET held a series of thematic workshops on IDPs to drive the development of computational resources and community standards. A strategic workshop titled “Intrinsically Disordered Proteins in Core Data Resources” was organised by NGP-NET at the EBI campus, Hinxton, the UK on June 1–2, 2017 to discuss the integration of IDP-related data and computational resources into the ELIXIR framework. A major outcome of this workshop was the recognition that IDP annotations are significantly underrepresented in the ELIXIR Core Data Resources (CDRs) (Durinx et al., 2016). These initial insights evolved into a comprehensive plan to develop the IDP field and integrate key IDP resources and tools into the CDRs. That plan formed the basis for the ELIXIR IDP Community proposal that was submitted and presented at the ELIXIR Head of Nodes meeting in Basel, Switzerland on September 13th, 2017. After completing the ELIXIR Community application process, including the creation of this white paper, the IDP Community proposal was accepted by the ELIXIR Head of Nodes on May 14th, 2019.

In parallel with these developments, the IDP research community has developed closer integration with ELIXIR activities. A follow up strategic meeting, “The 2^nd Workshop on Intrinsically Disordered Proteins in Core Data Resources”, was held in Prague on March 13–14, 2019 and brought together data producers, database developers and a representative of the ELIXIR Data Platform to discuss the integration of IDP data into the CDRs; and a member of the IDP research community attended the ELIXIR Interoperability Platform Face To Face Meeting on 1–2 April 2019. Furthermore, several actions have already taken place within the ELIXIR framework to tackle time-sensitive priorities including two implementation studies, “Implementation study for the integration of ELIXIR-IIB in ELIXIR Data Curation activities” and “Integration and standardisation of intrinsically disordered protein data”, focussing on interoperability. An additional, fundamental development associated with these implementation studies was the initialisation of the HUPO (Human Proteome Organisation) Proteomics Standards Initiative (PSI) Intrinsic Disorder workgroup.

The goal of the IDP Community in ELIXIR is to support the development of standards, tools and resources to accelerate the identification, analysis and functional characterisation of intrinsically disordered regions. Here, we introduce the major areas of priority for the development of an e-infrastructure that will allow the community to realise these goals while supporting the needs of the generators, users and consumers of IDP data.

Identification of community challenges

A strategic workshop titled “An intrinsically disordered protein user community proposal for ELIXIR” was held on October 31st, 2018 at the University of Padua, Italy. The meeting was attended by participants from 13 ELIXIR nodes: Belgium, Cyprus, Czech Republic, EBI, Germany, Hungary, Ireland, Israel, Italy, Netherlands, Spain, Switzerland, and the United Kingdom. ELIXIR members in both Greece and Sweden were also interested, but could not attend. The ELIXIR Hub team was represented by John Hancock, the ELIXIR Communities and Services Coordinator and representatives from the Interoperability, Tools and Training ELIXIR platforms. Members of the 3DBioInfo and Proteomics ELIXIR Communities, the Instruct-ERIC structural biology research infrastructure, and the ELIXIR CDRs of high relevance to the IDP research, namely PDBe (Mir et al., 2018), IMEx/IntAct (Orchard et al., 2014) and UniProt (UniProt Consortium, 2019), were also present. During the meeting, the following topics were identified as key areas of priority for the ELIXIR IDP Community.

Alignment with ELIXIR Activities

The community challenges identified are already well-aligned with on-going ELIXIR platform goals and activities. In particular:

Data Platform: The task proposed in Priority Area 3 “Integration with ELIXIR Core Data Resources” aligns with the goals of the Data Platform. The community is already in the process of integrating IDP annotation into the ELIXIR CDRs as part of two ELIXIR implementation studies, “Implementation study for the integration of ELIXIR-IIB in ELIXIR Data Curation activities” and “Integration and standardisation of intrinsically disordered protein data”. The recent integration of MobiDB into InterPro, PDBe and UniProt can be taken as a blueprint for the further integration of IDP information into ELIXIR CDRs. Furthermore, initial discussions have taken place to plan the integration of additional sources of IDP data. A separate task, which can benefit from ongoing Data Platform activities, is the development of an IDP curation framework as described in Priority Area 2 “Automated and community-driven curation”. This aligns with ELIXIR Data Platform Task 3 “Scalable curation” of the 2019–2023 ELIXIR Programme.

The distributed community curation for IDP resources will benefit from stronger interaction with Europe PMC, with the SciLite framework providing literature mining. This is especially important where data is scattered throughout the extensive corpus of biological literature and not properly indexed. Together with an automated curation triage step selecting relevant papers, this can boost the productivity of community curation while ensuring high-quality annotation of the IDP literature. Initial efforts have started with the design of a dedicated curation-support prototype, which has been used by DisProt curators since 2018. The service capitalises on the neXtA5 platform (Mottin et al., 2017), designed by the Text Mining group of the Swiss Institute of Bioinformatics. The demonstrator (http://candy.hesge.ch/disprotGUI/) is able to rank articles based on a scoring function, which prioritises articles with a high density of IDP-related concepts. Thanks to the curation-support tool, it will be possible to obtain a high quality curated benchmark, with the aim to evolve the current system from a triage system to a more accurate binary classifier, being able to select not only relevant papers but also to highlight short passages of text in full-text papers likely to support the expert curation.

Tools Platform: The tasks proposed in Priority Area 4 “Standardisation, benchmarking and indexing of computational tools” fall under the objectives of the Tools Platform. A major goal of Priority Area 4 is to increase productivity by developing reusable experimental workflows for commonly used IDP analysis pipelines built on top of ELIXIR CDRs. The Tools Platform can advise the ELIXIR IDP Community on the development of such workflows using the Common Workflow Language in collaboration with the ELIXIR Interoperability platform. The IDP analysis tool benchmarking by the CAID initiative is complementary to the OpenEBench benchmarking platform. OpenEBench represents an information hub for the distribution of reference datasets, the application of metrics and comparative assessments and the distribution of benchmarking results. CAID can also drive prototype development for software containers and reusable experimental workflows for the community. The outcomes could then be generalised for other applications. The bio.tools service registry can be a comprehensive census of the available software for IDP analysis and the BioContainers and Bioconda platforms can assist in the generation of software containers and their registry across different technologies to facilitate access and use of software by the IDP communities and beyond.

Interoperability Platform: Several of the identified priorities will benefit from the input of the Interoperability Platform. The main interaction with the platform will be to develop FAIR-compliant standards and guidelines for reporting, data exchange and retrieval of data on IDP structure and function (Wilkinson et al., 2016). The recent development of a HUPO PSI-ID workgroup and a draft MIADE standard are key steps towards this goal. IDP resources themselves will benefit from becoming FAIR and adhering and contributing to the further development of standards, controlled vocabularies and ontologies. These advances are of paramount importance to improve the dissemination of IDP data. FAIR-compliant data will also aid in the FAIRification of analysis tools where reusable computational workflows can make programmatic calls to retrieve reusable data from an integrated centralised source. The Interoperability Platform service framework has already defined the various key activities to support the “FAIRification” of data which can be applied to IDP data. The effort to make IDP data FAIR can exploit the various key services that are offered by the Interoperability Platform such as those identified as recommended practices through the ELIXIR Recommended Interoperability Resources (RIRs), and the platform mission-critical initiatives (i.e., Bioschemas, and Common Workflow Language). As this is an emerging community with new data resources in development, deploying these recommended practices in new IDP databases will highlight the IDP Community as a use case example in the mission to make data FAIR-at-source. This work can be supported by the Interoperability Platform under their remit to provide support for interoperability for ELIXIR Communities.

Compute Platform: Work being carried out in the Compute Platform can be leveraged in two main ways. Community curation requires authentication for a wide range of participants, which can benefit from the federated ELIXIR AAI approach for curator login. This could also facilitate the attribution of credit to curators. A second element would be to use the ELIXIR distributed computing infrastructure including identity and access management, data integration and container deployment across a range of appropriate compute endpoints as required, both for updating large-scale databases and to run the CAID experiment. A range of new Compute Platform activities are underway to enable an integrated approach to the deployment of relevant workflows and containers. The IDP Community can take advantage of these activities to deploy IDP-related workflows and containers.

Training Platform: The IDP Community is focussed on the development of the next generation of IDP researchers. Extensive training has been funded and carried out by the COST Action NGP-NET and additional training schools are planned as part of the MSCA-RISE-funded IDPfun project. The Italian ELIXIR node, which has been actively providing training, will organise a yearly training course on IDP resources. The input of the ELIXIR Training Platform will be indispensable to facilitate the development of a core set of training materials, its dissemination and inclusion in ELIXIR training courses. These materials will comprise introductory and advanced lessons on databases access, in silico analysis of IDPs (e.g. IDP prediction, IDP docking, MD simulations and IDP-specific primary sequence analysis techniques), and computational processing of experimental data (e.g. NMR spectra, phage display). There will also be a requirement for the development of training material related to IDP interoperability to train experimentalists, software developers, database developers and curators in topics such as IDP standards, IDP controlled vocabularies and Common Workflow Language (CWL) for IDPs. This aligns with the Interoperability Platform’s training and outreach task in the 2019–2023 ELIXIR Programme to collaborate with Training Platform members to provide Interoperability training. In collaboration with the Training Platform, the IDP Community will be able to adopt and implement a wide range of best practices and guidelines available through the ELIXIR Training Toolkit, which aims to provide a comprehensive reference resource for developing training capacity, rolling out new training programs, as well as expanding existing ones. All materials and activities developed for IDP training will be shared through TeSS for further dissemination. Additionally, train-the-trainer activities will be planned to increase the size of the IDP trainers pool and build additional training capacity.

Alignment with other communities

Conclusions

Recent years have seen a rapid growth of interest in IDPs. This has coincided with significant advances in the in vivo, in vitro and in silico methods to study the structure and function of IDPs. Unfortunately, the basic requirements for the organisation and dissemination of the tools and data produced by the field have not advanced in step with these developments. The field is on the cusp of an era where the high-throughput characterisation of the extensive intrinsically disordered regions (IDRs) of proteins is possible. This roadmap will provide the foundation that supports this data explosion and provide a solid platform for the future biological research of IDPs.

Data availability

No data are associated with this article.