Protein Phase Transitions in Health and Disease

Our lab aims at understanding how genetic changes between individuals can or cannot result in disease by quantifying the impact mutations have on protein aggregation and toxicity. 

We are particularly interested in amino acid sequences that can adopt different conformations and undergo a process of self-assembly which results in distinct physical states.

The aggregation of proteins into insoluble amyloid fibrils is a key process in the pathogenesis of a number of neurodegenerative conditions, such as Parkinson’s disease or Amyotrophic Lateral Sclerosis. However, examples of functional amyloid are also widespread in nature, especially across bacteria and fungi. Our work aims at systematically deciphering the sequence-dependencies of the process of aggregation in both functional and pathological contexts.

Recently, it has become clear that proteins can also self-assemble into a more dynamic and reversible state through a process of liquid de-mixing which is thought to contribute to the organization of the intracellular space. However, also for proteins undergoing liquid de-mixing, the balance between function and dysfunction is far from clear. It is also unknown if, in vivo, liquid de-mixed states are precursors of insoluble amyloid-like states, and to which extent proteins are structured once in the liquid state.

How we do it

In order to understand how mutations affect these delicate equilibria and to elucidate when and why a sequence starts aggregating or becomes toxic for the cell, our lab integrates experimental and computational approaches in different model systems. Recently, we have developed massively parallel approaches based on Deep Mutational Scanning (DMS) to quantify the toxicity or the aggregation propensity of hundreds of thousands of protein sequences in vivo. We believe that by portraying the full landscape of the effects of mutations in a specific protein domain we can reach a more systematic and comprehensive understanding of the determinants of amyloid formation and toxicity. 

We are also interested in developing similar high-throughput strategies to measure in vivo the effect of mutations on the physical state the proteins acquire upon mutation (diffuse, liquid de-mixed, insoluble) and to study the interactions between mutations to report on the conformations proteins adopt as they self-assemble. Overall, the exhaustive datasets we are generating will give mechanistic insights on the process of protein aggregation, while also reporting on specific conformations and mechanisms leading to cellular toxicity.  We also aim at using the datasets we generate to develop novel predictors of protein aggregation.

We focus on all classical amyloids, such as the amyloid-beta peptide, the main component of the plaques found in Alzheimer’s disease patients, but also on functional amyloids and on a less characterized part of the human proteome which is able to undergo liquid de-mixing: prion-like domains. Just like all disordered protein regions, prion-like domains are particularly difficult to study in vitro. In this perspective, in vivo approaches such as the ones we develop, can provide a unique opportunity to investigate these sequences in a systematic way.

Map of the effect of mutations on toxicity of the TDP-43 Prion-like Domain.

Percentage of substitutions and insertions increasing or decreasing amyloid formation of the Amyloid-Beta peptide, visualized on the cross-section of ex-vivo fibrils (7Q4M).

• Thermo MaxQ 8000

• Priyanka Narayan
  NIH-NIDDK
• Xavier Salvatella
  IRB, Barcelona
• Fran Supek
  IRB, Barcelona
• Ben Lehner
  CRG, Barcelona
• Luke McAlary /Justin Yerbury
  University of Wollongong, Australia