DREAMM: a web-based server for drugging protein-membrane interfaces as a novel workflow for targeted drug design

Chatzigoulas, Alexios; Cournia, Zoe

doi:10.1093/bioinformatics/btac680

Cited by 10 publications

(6 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Current tools that predict membrane-interaction sites in proteins (DREAMM ( Chatzigoulas and Cournia 2022a , b ), PPM3 ( Lomize et al 2022 ), and MODA ( Kufareva et al 2014 )) produce output on the single-residue level, whereas PMIpred quantifies membrane-binding regions ; averaging the free energy over the local surrounding of a given amino acid. This complicates a one-to-one comparison with these qualitative prediction methods on the single-residue level (e.g.…”

Section: Resultsmentioning

confidence: 99%

“…Previously developed methods include data-informed classifiers that predict which amino acids in a protein structure are membrane-interacting (e.g. MODA ( Kufareva et al 2014 ) and DREAMM ( Chatzigoulas and Cournia 2022a , b )) or describe the general protein orientation with respect to a (curved) membrane (e.g. PPM3 ( Lomize et al 2022 )).…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

PMIpred: a physics-informed web server for quantitative protein–membrane interaction prediction

van Hilten,

Verwei,

Methorst

et al. 2024

Bioinformatics

View full text Add to dashboard Cite

Motivation Many membrane peripheral proteins have evolved to transiently interact with the surface of (curved) lipid bilayers. Currently, methods to quantitatively predict sensing and binding free energies for protein sequences or structures are lacking, and such tools could greatly benefit the discovery of membrane-interacting motifs, as well as their de novo design. Results Here, we trained a transformer neural network model on molecular dynamics data for >50,000 peptides that is able to accurately predict the (relative) membrane-binding free energy for any given amino acid sequence. Using this information, our physics-informed model is able to classify a peptide’s membrane-associative activity as either non-binding, curvature sensing, or membrane binding. Moreover, this method can be applied to detect membrane-interaction regions in a wide variety of proteins, with comparable predictive performance as state-of-the-art data-driven tools like DREAMM, PPM3, and MODA, but with a wider applicability regarding protein diversity, and the added feature to distinguish curvature sensing from general membrane binding. Availability We made these tools available as a web server, coined Protein-Membrane Interaction predictor (PMIpred), which can be accessed at https://pmipred.fkt.physik.tu-dortmund.de.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

PMIpred: a physics-informed web server for quantitative protein–membrane interaction prediction

van Hilten,

Verwei,

Methorst

et al. 2024

Bioinformatics

View full text Add to dashboard Cite

show abstract

“…This value includes residues corresponding to L335, G339, N343, T345, V367, N370-S375, N437, N439, N440, K444-G447, Y449, N450, L455, F456, T470, I472, A475-P479, N481-F490, L492-S494, F496, and Q498-Q506 of the wild-type SARS-CoV-2 sequence. For cross-validation of the membrane binding residues identified by MODA we used the DREAMM [ 29 , 30 ] and PPM3 programs [ 32 , 33 ], the latter with a planar plasma membrane simulation and extracellular N-terminus. Further confirmation of membrane attraction was obtained from the dipole moments of closed spike trimer structures using the Protein Dipole Moments Server ( , accessed on 15 December 2022) [ 43 ].…”

Section: Methodsmentioning

confidence: 99%

“…The membrane docking sites of spike trimers can be predicted by tools, including DREAMM [ 29 , 30 ], Ez-3D [ 31 ], positioning of proteins in membranes (PPM) [ 32 , 33 ], and the membrane optimal docking area (MODA) program [ 34 ]. The latter is trained to identify lipid binding surfaces and calculates membrane binding propensities for each residue in a protein structure.…”

Section: Introductionmentioning

confidence: 99%

SARS-CoV-2 Omicron Subvariants Balance Host Cell Membrane, Receptor, and Antibody Docking via an Overlapping Target Site

Overduin

Bhat

Kervin

2023

Viruses

View full text Add to dashboard Cite

Variants of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) are emerging rapidly and offer surfaces that are optimized for recognition of host cell membranes while also evading antibodies arising from vaccinations and previous infections. Host cell infection is a multi-step process in which spike heads engage lipid bilayers and one or more angiotensin-converting enzyme 2 (ACE-2) receptors. Here, the membrane binding surfaces of Omicron subvariants are compared using cryo-electron microscopy (cEM) structures of spike trimers from BA.2, BA.2.12.1, BA.2.13, BA.2.75, BA.3, BA.4, and BA.5 viruses. Despite significant differences around mutated sites, they all maintain strong membrane binding propensities that first appeared in BA.1. Both their closed and open states retain elevated membrane docking capacities, although the presence of more closed than open states diminishes opportunities to bind receptors while enhancing membrane engagement. The electrostatic dipoles are generally conserved. However, the BA.2.75 spike dipole is compromised, and its ACE-2 affinity is increased, and BA.3 exhibits the opposite pattern. We propose that balancing the functional imperatives of a stable, readily cleavable spike that engages both lipid bilayers and receptors while avoiding host defenses underlies betacoronavirus evolution. This provides predictive criteria for rationalizing future pandemic waves and COVID-19 transmissibility while illuminating critical sites and strategies for simultaneously combating multiple variants.

show abstract

“…Protein–membrane interactions play a significant role in protein function although the protein–membrane interface is usually not known. ,− When using structures generated by ML folding models to predict protein–membrane interactions, extra caution is required as the predicted structures often contain unstructured regions with low pLDDT scores (pLDDT score <70), which can impact or bias the performance of protein–membrane interaction prediction tools (see Figure ). Therefore, when attempting to predict protein–membrane interactions from an AF2-generated model, one should consider removing regions with low predictability scores from the calculation.…”

Section: Membrane Proteinsmentioning

confidence: 99%

A Perspective on the Prospective Use of AI in Protein Structure Prediction

Versini,

Sritharan,

Aykac Fas

et al. 2023

J. Chem. Inf. Model.

Self Cite

View full text Add to dashboard Cite

AlphaFold2 (AF2) and RoseTTaFold (RF) have revolutionized structural biology, serving as highly reliable and effective methods for predicting protein structures. This article explores their impact and limitations, focusing on their integration into experimental pipelines and their application in diverse protein classes, including membrane proteins, intrinsically disordered proteins (IDPs), and oligomers. In experimental pipelines, AF2 models help X-ray crystallography in resolving the phase problem, while complementarity with mass spectrometry and NMR data enhances structure determination and protein flexibility prediction. Predicting the structure of membrane proteins remains challenging for both AF2 and RF due to difficulties in capturing conformational ensembles and interactions with the membrane. Improvements in incorporating membrane-specific features and predicting the structural effect of mutations are crucial. For intrinsically disordered proteins, AF2's confidence score (pLDDT) serves as a competitive disorder predictor, but integrative approaches including molecular dynamics (MD) simulations or hydrophobic cluster analyses are advocated for accurate dynamics representation. AF2 and RF show promising results for oligomeric models, outperforming traditional docking methods, with AlphaFold-Multimer showing improved performance. However, some caveats remain in particular for membrane proteins. Real-life examples demonstrate AF2's predictive capabilities in unknown protein structures, but models should be evaluated for their agreement with experimental data. Furthermore, AF2 models can be used complementarily with MD simulations. In this Perspective, we propose a "wish list" for improving deep-learning-based protein folding prediction models, including using experimental data as constraints and modifying models with binding partners or post-translational modifications. Additionally, a meta-tool for ranking and suggesting composite models is suggested, driving future advancements in this rapidly evolving field.

show abstract

DREAMM: a web-based server for drugging protein-membrane interfaces as a novel workflow for targeted drug design

Cited by 10 publications

References 34 publications

PMIpred: a physics-informed web server for quantitative protein–membrane interaction prediction

PMIpred: a physics-informed web server for quantitative protein–membrane interaction prediction

SARS-CoV-2 Omicron Subvariants Balance Host Cell Membrane, Receptor, and Antibody Docking via an Overlapping Target Site

A Perspective on the Prospective Use of AI in Protein Structure Prediction

Contact Info

Product

Resources

About