Modeling conformational states of proteins with AlphaFold

Sala, Davide; Engelberger, Felipe; Mchaourab, Hassane S.; Meiler, Jens

doi:10.1016/j.sbi.2023.102645

Cited by 90 publications

(54 citation statements)

References 49 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…On the contrary, the use of ligand 1 H CS information for refinement purposes remains largely untapped, with a foreseeable applicability for the verification of docking poses derived from X-ray and cryo-EM models. Most intriguing, however, is the use of ligand 1 H CS information in conjunction with ever more present structural models derived from AlphaFold [34] and other machine-learning based approaches to generate accurate and experimentally verified ensembles of protein-ligand complexes that closely resemble the native solution state.…”

Section: Discussionmentioning

confidence: 99%

Ligand ¹H NMR Chemical Shifts as Accurate Reporters for Protein‐Ligand Binding Interfaces in Solution**

Platzer,

Ptaszek,

Böttcher

et al. 2023

ChemPhysChem

View full text Add to dashboard Cite

The availability of high‐resolution 3D structural information is crucial for investigating guest‐host systems across a wide range of fields. In the context of drug discovery, the information is routinely used to establish and validate structure‐activity relationships, grow initial hits from screening campaigns, and to guide molecular docking. For the generation of protein‐ligand complex structural information, X‐ray crystallography is the experimental method of choice, however, with limited information on protein flexibility. An experimentally verified structural model of the binding interface in the native solution‐state would support medicinal chemists in their molecular design decisions. Here we demonstrate that protein‐bound ligand 1H NMR chemical shifts are highly sensitive and accurate probes for the immediate chemical environment of protein‐ligand interfaces. By comparing the experimental ligand 1H chemical shift values with those computed from the X‐ray structure using quantum mechanics methodology, we identify significant disagreements for parts of the ligand between the two experimental techniques. We show that quantum mechanics/molecular mechanics (QM/MM) molecular dynamics (MD) ensembles can be used to refine initial X‐ray co‐crystal structures resulting in a better agreement with experimental 1H ligand chemical shift values. Overall, our findings highlight the usefulness of ligand 1H NMR chemical shift information in combination with a QM/MM MD workflow for generating protein‐ligand ensembles that accurately reproduce solution structural data.

show abstract

Section: Discussionmentioning

confidence: 99%

Ligand ¹H NMR Chemical Shifts as Accurate Reporters for Protein‐Ligand Binding Interfaces in Solution**

Platzer,

Ptaszek,

Böttcher

et al. 2023

ChemPhysChem

View full text Add to dashboard Cite

show abstract

“…Molecular dynamics simulations can be used to study protein dynamics and transitions between different conformational states . However, using molecular dynamics to simulate the transition trajectories between conformational states may require a huge computational cost, which is a challenge for large proteins . Jens Meiler et al suggested that reducing the depth of input MSAs by stochastic subsampling can enable AlphaFold2 on sample multiple alternative models .…”

Section: Challenges For Protein Structure Prediction Methodsmentioning

confidence: 99%

“…In addition, Gustavo Parisi et al found that the backbone flexibility associated with apo-holo pairs of conformations is negatively correlated with pLDDT, indicating that the local structural changes linked to ligand binding transitions can be inferred from pLDDT scores . Other AlphaFold2 derived methods for modeling multiple conformational states can be found in the reference . Recently, Chen Song et al proposed a method to explore the alternative conformation of a known protein structure based on predicted contact map .…”

Section: Challenges For Protein Structure Prediction Methodsmentioning

confidence: 99%

“…However, this presents a challenge due to the scarcity of multiple state structure information on proteins in the PDB which hampers the learning or training of deep learning methods. A more challenging task is to capture conformational changes induced by external factors, because their coevolutionary signals are weak . Moreover, it is crucial to develop model quality assessment methods that can effectively evaluate the accuracy and stability of multiple state structures within an ensemble.…”

Section: Challenges For Protein Structure Prediction Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Recent Advances and Challenges in Protein Structure Prediction

Peng,

Liang,

Xia

et al. 2023

J. Chem. Inf. Model.

View full text Add to dashboard Cite

Artificial intelligence has made significant advances in the field of protein structure prediction in recent years. In particular, DeepMind's end-to-end model, AlphaFold2, has demonstrated the capability to predict three-dimensional structures of numerous unknown proteins with accuracy levels comparable to those of experimental methods. This breakthrough has opened up new possibilities for understanding protein structure and function as well as accelerating drug discovery and other applications in the field of biology and medicine. Despite the remarkable achievements of artificial intelligence in the field, there are still some challenges and limitations. In this Review, we discuss the recent progress and some of the challenges in protein structure prediction. These challenges include predicting multidomain protein structures, protein complex structures, multiple conformational states of proteins, and protein folding pathways. Furthermore, we highlight directions in which further improvements can be conducted.

show abstract

“…128,130−133 ML folding models can hardly predict these conformational changes as they are trained on static structures and evolutionary information, although attempts have been made more recently. 134 This limitation for predicting conformational ensembles also extends to predicting protein−protein multimer structures and/or when proteins are embedded or contact the cell membrane. Coarse-grained enhanced sampling MD simulations of rhodopsin dimerization have revealed the adoption of multiple dimer interfaces, which are not predicted by AF2.…”

Section: Proteins/domainsmentioning

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.

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

Modeling conformational states of proteins with AlphaFold

Cited by 90 publications

References 49 publications

Ligand ¹H NMR Chemical Shifts as Accurate Reporters for Protein‐Ligand Binding Interfaces in Solution**

Ligand ¹H NMR Chemical Shifts as Accurate Reporters for Protein‐Ligand Binding Interfaces in Solution**

Recent Advances and Challenges in Protein Structure Prediction

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

Contact Info

Product

Resources

About

Modeling conformational states of proteins with AlphaFold

Cited by 90 publications

References 49 publications

Ligand 1H NMR Chemical Shifts as Accurate Reporters for Protein‐Ligand Binding Interfaces in Solution**

Ligand 1H NMR Chemical Shifts as Accurate Reporters for Protein‐Ligand Binding Interfaces in Solution**

Recent Advances and Challenges in Protein Structure Prediction

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

Contact Info

Product

Resources

About

Ligand ¹H NMR Chemical Shifts as Accurate Reporters for Protein‐Ligand Binding Interfaces in Solution**

Ligand ¹H NMR Chemical Shifts as Accurate Reporters for Protein‐Ligand Binding Interfaces in Solution**