The project I have been focusing for the last three years was the development of a computational method to simulate protein aggregation into amyloids. As further described below, protein aggregation is a complex process leading to the accumulation and deposition of amyloid fibrils causing several and uncurable diseases. The atomistic detail of this process remains elusive and molecular dynamics could complement experimental techniques. However, the large size of the systems and long timescales of the process make difficult to apply such methods. Following the works from Camilloni and Sutto, I developed a structure-based (SB) approach to protein aggregation. During the first year of the PhD, I focused on the initial proof of concept of using two structures in a single SB model. The first version was based on a software called SMOG which creates SB models using a protein structure as input. In the first year I developed a python script to merge different SMOG models into one and perform MD simulations using GROMACS. The method was developed using a short peptide derived from the Transthyretin protein. The first attempt was to coarse grain the representation of the TTR peptide mapping a single bead per residue on the Cα. However, such simplified representation was not suitable to simulate protein aggregation. Afterward, I used SMOG in a semi-atomistic representation to define our model called multi-GO, demonstrating the feasibility of our method, although several optimizations were required. The first optimization was to develop a python code to parametrize the LJ potential instead of SMOG. As a result, bonded parameters were optimized along with the non-bonded. We based bonded parameters on GROMOS54a7, and the improved geometry greatly improved our model. Moreover, I added the possibility to learn the LJ potential from MD simulations. This new version was named multi-eGO. The results obtained in the second year were published in the first article on PNAS. In the third year I applied this method on the full length Aβ42 protein, as the knowledge required to parametrize multi-eGO is available. Testing full-length proteins highlighted a major issue in multi-eGO related to an unbalance between LJ potential of local and long-range contacts. During the third year I was able improve the description of proteins by inserting a prior information describing the local geometry. As the Aβ42 protein was properly parametrized, I was able to extend multi-eGO functionalities by adding a small molecule in the simulation. I performed this task at Cambridge University hosted by prof. Michele Vendruscolo, proving the possibility of small molecules parametrization in multi-eGO. Altogether, during those three years we developed a new model able to qualitatively simulate protein aggregation. Although several optimizations are necessary, multi-eGO could become an additional tool to simulate protein aggregation.

BUILDING A MODEL TO SIMULATE PROTEIN AGGREGATION AT ATOMISTIC DETAIL / E. Scalone ; tutor: C. Camilloni ; coordinatore: R. Mantovani. Dipartimento di Bioscienze, 2023 Mar 23. 35. ciclo, Anno Accademico 2022.

BUILDING A MODEL TO SIMULATE PROTEIN AGGREGATION AT ATOMISTIC DETAIL.

E. Scalone
2023

Abstract

The project I have been focusing for the last three years was the development of a computational method to simulate protein aggregation into amyloids. As further described below, protein aggregation is a complex process leading to the accumulation and deposition of amyloid fibrils causing several and uncurable diseases. The atomistic detail of this process remains elusive and molecular dynamics could complement experimental techniques. However, the large size of the systems and long timescales of the process make difficult to apply such methods. Following the works from Camilloni and Sutto, I developed a structure-based (SB) approach to protein aggregation. During the first year of the PhD, I focused on the initial proof of concept of using two structures in a single SB model. The first version was based on a software called SMOG which creates SB models using a protein structure as input. In the first year I developed a python script to merge different SMOG models into one and perform MD simulations using GROMACS. The method was developed using a short peptide derived from the Transthyretin protein. The first attempt was to coarse grain the representation of the TTR peptide mapping a single bead per residue on the Cα. However, such simplified representation was not suitable to simulate protein aggregation. Afterward, I used SMOG in a semi-atomistic representation to define our model called multi-GO, demonstrating the feasibility of our method, although several optimizations were required. The first optimization was to develop a python code to parametrize the LJ potential instead of SMOG. As a result, bonded parameters were optimized along with the non-bonded. We based bonded parameters on GROMOS54a7, and the improved geometry greatly improved our model. Moreover, I added the possibility to learn the LJ potential from MD simulations. This new version was named multi-eGO. The results obtained in the second year were published in the first article on PNAS. In the third year I applied this method on the full length Aβ42 protein, as the knowledge required to parametrize multi-eGO is available. Testing full-length proteins highlighted a major issue in multi-eGO related to an unbalance between LJ potential of local and long-range contacts. During the third year I was able improve the description of proteins by inserting a prior information describing the local geometry. As the Aβ42 protein was properly parametrized, I was able to extend multi-eGO functionalities by adding a small molecule in the simulation. I performed this task at Cambridge University hosted by prof. Michele Vendruscolo, proving the possibility of small molecules parametrization in multi-eGO. Altogether, during those three years we developed a new model able to qualitatively simulate protein aggregation. Although several optimizations are necessary, multi-eGO could become an additional tool to simulate protein aggregation.
23-mar-2023
Settore BIO/10 - Biochimica
Settore FIS/07 - Fisica Applicata(Beni Culturali, Ambientali, Biol.e Medicin)
CAMILLONI, CARLO
MANTOVANI, ROBERTO
Doctoral Thesis
BUILDING A MODEL TO SIMULATE PROTEIN AGGREGATION AT ATOMISTIC DETAIL / E. Scalone ; tutor: C. Camilloni ; coordinatore: R. Mantovani. Dipartimento di Bioscienze, 2023 Mar 23. 35. ciclo, Anno Accademico 2022.
File in questo prodotto:
File Dimensione Formato  
phd_unimi_R12491.pdf

Open Access dal 24/03/2024

Tipologia: Post-print, accepted manuscript ecc. (versione accettata dall'editore)
Dimensione 31.01 MB
Formato Adobe PDF
31.01 MB Adobe PDF Visualizza/Apri
Pubblicazioni consigliate

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/2434/955595
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact