Identifying determinants of specificity in yeast protein complexes

We aim to tackle a number of important and fundamental questions in biology. Specifically, we aim to understand how proteins interact to produce biological function; we aim to understand the cellular events that take place when hybrids form between two species; we also aim to understand on a molecular level what gives rise to the differences between species. Proteins almost invariably work together to achieve biological function. Approximately 60% or proteins take part in some kind of stable assembly or 'complex'. These protein complexes play a role in the majority of cellular processes. In order to form complexes individual proteins must make contact with ('bind') to a small number of specific partners. It is the rules that control this 'specificity' for binding that we wish to investigate. Binding in complexes is, thus, the result of specific contacts in the context of proteins' three-dimensional protein structures. We propose to determine the key regions for binding (termed 'interfaces') and in these interface regions the amino acids that are most important for binding. In order to achieve this goal, we will make use of computer-based methodologies to predict which regions of an individual protein are responsible for the precise interactions between different proteins. A biological signal we will make use of is that provided by evolution - the process of change as species evolve away ('diverge') from their common ancestor - and that manifests itself as changes in amino acid sequences. As the protein structures must form correctly, and keep their function, the nature of these changes is informative for studying binding specificity. For example, if a key amino acid in one protein sequence changes, then an associated change might be necessary in the sequence of its binding partner. This coupling of change between proteins is termed 'co-evolution'. The consequence of this ongoing process in diverging species is that, at some point, proteins will become species-specific and so lose the ability to bind partners in their sister species. The primary objective of our project is to understand the limits of this process of change, i.e., at what point can partners no longer bind and which changes contribute most to the ability to bind. In order to achieve our objectives, we will combine computer (bioinformatics) and laboratory-based research strategies. These complementary approaches will be the key to our success. In yeast, hybrids can be formed by the fusion of two parent cells. When this happens, the daughter cell will contain a mixture of all of the cellular components of the two parents. This will include all genetic material and all proteins. Thus, protein complexes from one parent may bind to those of the other parent, making 'chimeric' complexes, or they may not bind resulting in incompatibilities at the molecular level. We will investigate the rules of binding by analysing both naturally occurring and artificial yeast hybrids. Specifically, we will use computational techniques to develop likely rules that will enable us to predict which proteins will form chimeric complexes, and which will not. This will allow us to understand the process of complex formation between proteins from different species and to quantify the extent to which species must be different at the molecular level before they are incompatible at the level of protein structure (our primary objective). Our results will also have implications for understanding the significance of hybrids in speciation by permitting insight at the molecular level into the process of 'hybrid vigour'; the increased success or 'fitness' of the hybrid organism (our secondary objective).

We propose to study the fundamental biological processes of (i) specificity of binding in protein-protein interactions (ii) the effects of this specificity on the fitness of hybrid species and (iii) the role of this fitness on speciation. We will iteratively apply rounds of computational prediction and experimental testing of these predictions. This will provide a powerful means to tackle these important questions. We will use several yeast species as model organisms, making use of both naturally-occurring hybrids and artificially-produced hybrids that we can make in the laboratory. When hybrids form the proteomes of the two parental species mix. If there is a high degree of sequence similarity between the two species many subunits of protein complexes can substitute for each other. We have preliminary data indicating that that this happens in a range of yeast hybrids. We will use comparative modelling methods to produce sets of yeast complexes, and use these to quantify the degree and context of sequence divergence required to change specificity. We will analyse the binding interfaces, identifying interfaces that are conserved, have functionally equivalent substitutions or have coevolved. On this basis, we will be able to predict which subunits can form chimeric complexes. In order to test these predictions will compare sequences of parent and hybrid yeast species, permitting the identification of the evolutionary origins of constituents of protein complexes. Artificially-produced yeast hybrids will be used to determine the effect of mixing the proteome immediately after hybridisation. Forcing chimeric complexes to form, using gene deletion techniques, will enable the quantification of the effects of hybridisation on fitness. By iterative prediction and testing we propose to develop a quantitative understanding both of specificity in protein complexes (objective 1) and the contribution of protein complexes to hybrid vigour and speciation (objective 2).