Proteins are fundamental components of biological processes thus, they are often termed the
molecular machinery of life. They commonly form dimers, in a process that is often essential
for their functionality. Given the ubiquitous nature of protein regulation, many diseases are
based on malfunctioning proteins and inhibiting them by binding to the active site is a widely
chosen approach in drug development. However, due to acquired resistance mechanisms or high
off-target effects, the active site might not always be a viable approach. This work presents an
atomistic, structural investigation of dimeric proteins in the context of major disease processes,
where we provide insights into potential alternative drug targeting approaches.
In this Thesis, novel diffusion-based methods are applied to characterise the intra-structural
connectivity and signalling of protein dimers. The basis of our methods is the description of
proteins as atomistic, energy-weighted graphs, where every atom represents a node, and every
bond or interaction is encoded as a weighted edge. These graphs facilitate the study of connectivity
and signal propagation within the protein through diffusion processes on the atom
(node) and bond (edge) space. Two complementary methodologies are applied here, Markov
Transients and bond-to-bond propensities, which have been successfully used in the context
of allosteric site detection, the study of protein-protein interactions and the investigation of
allosteric signalling on an atomistic level. This work explores the extension of these methodologies
onto protein dimers and presents the investigation of allosteric mechanisms in three
disease-relevant study systems:
1. Estrogen receptor alpha (ERα) is a homodimer and the main driver in breast cancer
(BC) development and progression. Current chemotherapies based on inhibiting ERα
become ineffective when recurrent tumours develop resistance against anti-estrogens. Our
methodologies validate the molecular mechanism in ERα, and we further establish the
prevalent role of the dimer interface in the inhibition process.
2. The main protease (Mpro) of the coronavirus SARS-CoV-2 is essential for virus replication
in an early step of the viral life cycle. Since the beginning of 2020, we have seen this virus
causing a global pandemic of COVID-19, with over 285 million cases of infection and
over 5.5 million deaths by the end of 2021. To aid in combating COVID-19, we predict highly connected allosteric hotspots and provide insights into how the disruption of the
obligatory Mpro dimerisation presents a fruitful approach.
3. Cyclin-dependent kinases (CDKs) 4 and 6 are two essential cell cycle regulators that are
often associated with cancer development, and in BC, their inhibition is part of an effective
combinatorial treatment. This work contributes to understanding their activation process
in complex with D-type cyclins and sheds light on the differential inhibitor patterns seen
for CDKs.
By exploring these three systems with atomistic graph analysis, we describe intra-complex
communication essential for activation in all three proteins. We further present implications
for the respective dimer interface connectivities and how they could be a fruitful drug target.
We conclude that ERα, the SARS-CoV-2 Mpro and CDK4/6 can be disrupted over allosteric
mechanisms that include their dimer interfaces. These results provide scope for targeted drug
development and provide a valuable contribution to the ongoing efforts to find efficient treatments
for BC and COVID-19.