Allosteric Network Compiler
ANC: A rule-based framework for modelling allosteric proteins and biochemical networks
Most biochemical networks contain proteins whose function is subject to regulation by various other molecules, such as drugs, 2nd messengers, covalent modifications and other proteins. One common form of regulation is when one molecule, by interacting with a specific binding site of a protein, regulates the strength of the protein’s interaction with another molecule at a distinct, or "allosteric", site. For example, binding of an extracellular agonist by a G protein-coupled receptor can increase its affinity for intracellular ligands such as G proteins, even though the two binding sites are physically distinct. Allostery is ubiquitous in biochemical networks and occurs in many membrane receptors, ion channels, enzymes and transcription factors.
Using ANC, one can model the structure of allosteric proteins at a level of detail that accounts for their functional properties. Simple components are used as lego-like building blocks for more complex allosteric structures. You can delineate subunits and domains, binding and phosphorylation sites, and can model regulatory linkages between subunits and components (dashed lines in A, above). Modelling is both modular and hierarchical as simple structures can become components of larger, more complex structures.
Underpinning ANC is a thermodynamically-grounded modelling framework that describes allostery and cooperativity through transitions between discrete conformational states. Borrowing from classic allosteric theory, this framework assumes that ligands and other regulatory inputs interact non-cooperatively (i.e. independently) with each conformational state. Regulation arises indirectly because each ligand biases the equilibrium between the conformational states of a protein in favour of the state to which it has higher affinity.
The assumption of independence built into ANC’s thermodynamic framework can help biologists tackle the problem of regulatory complexity. Regulatory complexity arises in biochemical networks when an allosteric protein has multiple regulatory interactions. In principle, each combination of regulatory inputs can uniquely determine the strength of a protein’s interactions with downstream pathways. Thus, the number of parameters required to model allosteric and cooperative effects in networks grows combinatorially with the number of interactions and can be very large. However, if ligands are assumed to interact independently with a small number of discrete conformational states, this combinatorial growth in the dimensionality of parameter space is avoided. By reducing the number of parameters, this framework also facilitates the creation of biochemical models that have more predictive power.
Through its rule-based modelling approach, ANC also addresses the problem of combinatorial complexity. In biochemical networks, proteins dynamically form macromolecular complexes with ligands or other proteins and are often subject to covalent modifications that regulate these interactions. As the number of such interactions increases, there arises a combinatorial "explosion" in the number of possible states of the system, which challenges our ability to write down a realistic biochemical model. Fortunately, such systems can often be described with a small number of rules from which ANC can automatically infer the existence of a much larger set of biochemical species and reactions.
ANC is more than a modelling tool. Indeed, it embodies a methodological framework, which is based on well-established allosteric theory and biophysical principles and also a modular language for describing allosteric proteins and networks. In the process of creating an ANC model, your understanding of the relationship between the structure and function of a protein can improve significantly. ANC models are also straightforward to construct, making it easy to explore different hypotheses.
Modelling biochemical networks in the face of biological complexity is challenging, but computational tools such as ANC can help researchers create realistic and predictive models of cellular signalling and gene regulatory pathways. Please try ANC and let us know what you think!