Periodic Reporting for period 1 - InMIND (Intervention in Neurodegenerative disorders via Mechanistic INsight into liquid-like Droplets)
Okres sprawozdawczy: 2021-06-01 do 2023-05-31
Neurodegenerative diseases are a leading cause of dementia and death in the elderly population. With population aging and the lack of effective treatment, the number of individuals with dementia in EU28 is projected to double in the next 25 years. A hallmark of neurodegenerative diseases is the progressive accumulation of insoluble aggregates in the brain. While the path to the formation of these aggregates is still elusive, recent evidence suggests that they may in part originate from the dysregulation of biomolecular condensates. Biomolecular condensates are one of the ways cells compartmentalise and concentrate proteins, often together with nucleic acids, and are formed through a reversible de-mixing process which is often driven or aided by intrinsically disordered proteins and regions (collectively IDRs). In a new promising approach for therapeutic intervention, small-molecule drugs may be used to modulate the self-assembly process and material properties of condensates and prevent the formation of pathogenic aggregates.
To substantiate the feasibility of this approach and guide the design of small-molecule modulators of condensate properties, we need molecular models which are able to accurately predict the influence of solution conditions and sequence variations on the preferences of IDRs to self-associate.
In this project, we developed an unprecedentedly accurate and computationally efficient sequence-dependent model of IDRs and made use of the model to (i) interpret experimental data on the role of condensate formation and aberrant phase transitions in disease states, (ii) predict the effect of sequence variations and small molecules on the phase and rheological properties of condensates of IDRs, (iii) investigate the sequence dependence of Alzheimer's-related Amyloid-β42 self-association, and (iv) generate and analyse a database of the conformational ensembles of all the IDRs in the human proteome.
To substantiate the feasibility of this approach and guide the design of small-molecule modulators of condensate properties, we need molecular models which are able to accurately predict the influence of solution conditions and sequence variations on the preferences of IDRs to self-associate.
In this project, we developed an unprecedentedly accurate and computationally efficient sequence-dependent model of IDRs and made use of the model to (i) interpret experimental data on the role of condensate formation and aberrant phase transitions in disease states, (ii) predict the effect of sequence variations and small molecules on the phase and rheological properties of condensates of IDRs, (iii) investigate the sequence dependence of Alzheimer's-related Amyloid-β42 self-association, and (iv) generate and analyse a database of the conformational ensembles of all the IDRs in the human proteome.
To gain insights into the sequence-dependence of condensate formation, we developed CALVADOS, a residue-level model for molecular dynamics simulations of IDRs. We based our model development strategy on an automated Bayesian procedure which trains the amino-acid-specific parameters of the model against experimental data reporting on single-chain conformational properties. The training procedure results in significantly improved predictions of conformational properties, self-association, and condensate formation of diverse and unrelated protein sequences. The development, validation, and testing of the model is published in two scientific articles in PNAS (DOI: 10.1073/pnas.2111696118) and Open Research Europe (DOI: 10.12688/openreseurope.14967.2). The code to run CALVADOS simulations is freely available on Zenodo and has been used by other researchers across the world.
To identify the molecular determinants governing the phase and material properties of protein condensates, we applied CALVADOS to study condensates of the N-terminal low complexity domain of nCPEB4 (nCPEB4-LCD), an IDR involved in autism spectrum disorder. By direct comparison with biophysical experiments, we validated the accuracy of CALVADOS at capturing the influence of pH and sequence variation on the phase-separation propensity of nCPEB4-LCD. Our molecular-level insights into the network of intermolecular interactions within condensates aided the interpretation of experimental data on the role of splice variations in condensate formation and aberrant phase transitions of nCPEB4. The work is published as a preprint on bioRxiv (DOI: 10.1101/2023.03.19.532587) and the underlying data and code are available on Zenodo.
To gain insights into the effect of mutations on the rheological properties of condensates, we calculated dynamic moduli and viscosities from molecular simulations of several variants of the low complexity domain of hnRNPA1 (hnRNPA1-LCD), an IDR associated with amyotrophic lateral sclerosis. We combined the simulation data with polymer theory to provide a residue-level interpretation of the sequence dependence of the dynamic moduli of condensates of hnRNPA1-LCD and its variants.
Further, as a proof of concept for the use of CALVADOS to test the therapeutic potential of small-molecule drugs for neurodegenerative diseases, we studied the modulation of the phase properties of the protein Tau by Suramin, an anionic small-molecule drug used for treatment of African sleeping sickness. In qualitative agreement with experiments, we found that Suramin enhances the propensity of Tau to form condensates via electrostatic interactions with the positively charged residues of the proline-rich domain. Stabilising liquid-like condensates of Tau using a small-molecule drug has therapeutic potential as it may disfavour the aggregation into neurofibrillary tangles in Alzheimer’s disease.
Although CALVADOS does not account for secondary structure and hence cannot directly address the formation of irreversible aggregates, the model can be used to accurately probe the reversible formation of dimers, which is arguably the first step of the nucleation process. We explored this idea by simulating the self-association between all possible single-amino-acid substitutions of Amyloid-β42, the main protein component of amyloid plaques in Alzheimer's disease. From these simulations we could quantify the effect of mutations on the equilibrium constant of the dimerisation process and compared these computational data with experimental mutagenesis studies on Amyloid-β42 reported in the literature.
Lastly, we exploited the computational efficiency and accuracy of CALVADOS to generate a database of all the IDRs of the human proteome. Based on the confidence score assigned in AlphaFold structure predictions of the full-length proteins, we identified 29,998 IDRs, and performed CALVADOS simulations of each of these sequences. Through bioinformatics analyses, we explored (i) the relationship between the compaction of the IDRs and the biological function and cellular localisation of the proteins, (ii) the relationship between sequence descriptors and chain compaction, and (iii) the conservation of sequence and conformational properties across homologous IDRs. The work is published as preprint on bioRxiv (DOI: 10.1101/2023.05.08.539815) and a browsable database of conformational ensembles is accessible via the Electronic Research Data Archive at University of Copenhagen.
To identify the molecular determinants governing the phase and material properties of protein condensates, we applied CALVADOS to study condensates of the N-terminal low complexity domain of nCPEB4 (nCPEB4-LCD), an IDR involved in autism spectrum disorder. By direct comparison with biophysical experiments, we validated the accuracy of CALVADOS at capturing the influence of pH and sequence variation on the phase-separation propensity of nCPEB4-LCD. Our molecular-level insights into the network of intermolecular interactions within condensates aided the interpretation of experimental data on the role of splice variations in condensate formation and aberrant phase transitions of nCPEB4. The work is published as a preprint on bioRxiv (DOI: 10.1101/2023.03.19.532587) and the underlying data and code are available on Zenodo.
To gain insights into the effect of mutations on the rheological properties of condensates, we calculated dynamic moduli and viscosities from molecular simulations of several variants of the low complexity domain of hnRNPA1 (hnRNPA1-LCD), an IDR associated with amyotrophic lateral sclerosis. We combined the simulation data with polymer theory to provide a residue-level interpretation of the sequence dependence of the dynamic moduli of condensates of hnRNPA1-LCD and its variants.
Further, as a proof of concept for the use of CALVADOS to test the therapeutic potential of small-molecule drugs for neurodegenerative diseases, we studied the modulation of the phase properties of the protein Tau by Suramin, an anionic small-molecule drug used for treatment of African sleeping sickness. In qualitative agreement with experiments, we found that Suramin enhances the propensity of Tau to form condensates via electrostatic interactions with the positively charged residues of the proline-rich domain. Stabilising liquid-like condensates of Tau using a small-molecule drug has therapeutic potential as it may disfavour the aggregation into neurofibrillary tangles in Alzheimer’s disease.
Although CALVADOS does not account for secondary structure and hence cannot directly address the formation of irreversible aggregates, the model can be used to accurately probe the reversible formation of dimers, which is arguably the first step of the nucleation process. We explored this idea by simulating the self-association between all possible single-amino-acid substitutions of Amyloid-β42, the main protein component of amyloid plaques in Alzheimer's disease. From these simulations we could quantify the effect of mutations on the equilibrium constant of the dimerisation process and compared these computational data with experimental mutagenesis studies on Amyloid-β42 reported in the literature.
Lastly, we exploited the computational efficiency and accuracy of CALVADOS to generate a database of all the IDRs of the human proteome. Based on the confidence score assigned in AlphaFold structure predictions of the full-length proteins, we identified 29,998 IDRs, and performed CALVADOS simulations of each of these sequences. Through bioinformatics analyses, we explored (i) the relationship between the compaction of the IDRs and the biological function and cellular localisation of the proteins, (ii) the relationship between sequence descriptors and chain compaction, and (iii) the conservation of sequence and conformational properties across homologous IDRs. The work is published as preprint on bioRxiv (DOI: 10.1101/2023.05.08.539815) and a browsable database of conformational ensembles is accessible via the Electronic Research Data Archive at University of Copenhagen.
The CALVADOS model filled a gap in the state-of-the-art as it provides an accurate, sequence-dependent model which can be used to efficiently predict single-chain conformations as well as the formation and the material properties of condensates of IDRs.
As highlighted by the applications of the model in this project, the computational efficiency and accuracy of CALVADOS makes it possible to perform fast, large-scale simulations of multi-component condensates as well as proteome-wide analyses of conformational ensembles of IDRs. This technical achievement has the potential to enhance the innovation capacity of pharmaceutical companies working on the development of drugs targeting IDRs. Therefore, the work carried out in the fellowship contributes towards the European pharmaceutical strategy which strives to support competitiveness and innovation in the pharmaceutical industry in the EU. Potential users of the project results are researchers working with proteins in academia and in pharmaceutical companies.
As highlighted by the applications of the model in this project, the computational efficiency and accuracy of CALVADOS makes it possible to perform fast, large-scale simulations of multi-component condensates as well as proteome-wide analyses of conformational ensembles of IDRs. This technical achievement has the potential to enhance the innovation capacity of pharmaceutical companies working on the development of drugs targeting IDRs. Therefore, the work carried out in the fellowship contributes towards the European pharmaceutical strategy which strives to support competitiveness and innovation in the pharmaceutical industry in the EU. Potential users of the project results are researchers working with proteins in academia and in pharmaceutical companies.