A Dance with Protein Assemblies : Analysis, Structure Prediction, and Design
Författare
Summary, in English
Protein assemblies are some of the most complex molecular machines in nature. They facilitate many cellular functions, from DNA replication to molecular motion, energy production, and even the production of other proteins. In a series of 3 papers, we analyzed the structure, developed structure prediction tools, and design tools, for different protein assemblies. Many of the studies were centered around viral protein capsids. Viral capsids are protein coats found inside viruses that contain and protect the viral genome.
In one paper, we studied the interfaces of these capids and their energy landscapes. We found that they differ from regular homomers in terms of the amino acid composition and size, but not in the quality of interactions. This contradicts existing experimental and theoretical studies that suggest that the interactions are weak. We hypothesise that the occlusion by our models of electrostatic and entropic contributions might be at play.
In another paper, we developed methods to predict large cubic symmetrical protein assemblies, such as viral capsids, from sequence. This method is based upon AlphaFold, a new AI tool that has revolutionized protein structure prediction. We found that we can predict up to 50% of the structures of these assemblies. The method can quickly elucidate the structure of many relevant proteins for humans, and for understanding structures relevant to disease, such as the structures of viral capsids.
In the final paper, we developed tools to design capsid-like proteins called cages – structures that can be used for drug delivery and vaccine design. A fundamental problem in designing cage structures is achieving different architectures and low porosity, goals that are important for vaccine design and the delivery of small drug molecules. By explicitly modelling the shapes of the subunits in the cage and matching the shapes with proteins from structural databases, we find that we can create structures with many different sizes, shapes, and porosities - including low porosities. While waiting for experimental validation, the design strategy described in the paper must be extended, and more designs must be tested.
In one paper, we studied the interfaces of these capids and their energy landscapes. We found that they differ from regular homomers in terms of the amino acid composition and size, but not in the quality of interactions. This contradicts existing experimental and theoretical studies that suggest that the interactions are weak. We hypothesise that the occlusion by our models of electrostatic and entropic contributions might be at play.
In another paper, we developed methods to predict large cubic symmetrical protein assemblies, such as viral capsids, from sequence. This method is based upon AlphaFold, a new AI tool that has revolutionized protein structure prediction. We found that we can predict up to 50% of the structures of these assemblies. The method can quickly elucidate the structure of many relevant proteins for humans, and for understanding structures relevant to disease, such as the structures of viral capsids.
In the final paper, we developed tools to design capsid-like proteins called cages – structures that can be used for drug delivery and vaccine design. A fundamental problem in designing cage structures is achieving different architectures and low porosity, goals that are important for vaccine design and the delivery of small drug molecules. By explicitly modelling the shapes of the subunits in the cage and matching the shapes with proteins from structural databases, we find that we can create structures with many different sizes, shapes, and porosities - including low porosities. While waiting for experimental validation, the design strategy described in the paper must be extended, and more designs must be tested.
Avdelning/ar
Publiceringsår
2023-11-24
Språk
Engelska
Fulltext
- Available as PDF - 4 MB
- Available as PDF - 3 MB
- Download statistics
Dokumenttyp
Doktorsavhandling
Förlag
Lund University
Ämne
- Biochemistry and Molecular Biology
Nyckelord
- protein design
- protein structure prediction
- virus assembly
Status
Published
Handledare
ISBN/ISSN/Övrigt
- ISBN: 978-91-7422-991-2
- ISBN: 978-91-7422-990-5
Försvarsdatum
24 november 2023
Försvarstid
13:00
Försvarsplats
K:B
Opponent
- Björn Wallner (Professor)