Membrane biology has from its earliest days strived towards a single, unifying model. That goal has been approached incrementally over the past century by various scientists.
In 1925, membranes were imagined as lipid bilayers by Gorter and Grendel, but this pioneering theory did not include proteins. Danielli and Davson in 1935 added a film of protein to either side of the bilayer. 37 years later, Singer and Nicolson proposed that there are integral proteins spanning both leaflets of a lipid bilayer and peripheral proteins attached to either side. They also suggested that proteins and lipids could move laterally and were randomly mixed. Researchers soon noticed that membrane proteins and lipids are not randomly mixed, but for over 50 years, no model became accepted for explaining this.
During my undergraduate studies at the University of Alberta, my mentor Professor Michael Overduin and I resolved this problem and introduced a new, truly comprehensive model. We proposed that the membrane is a network of structural and functional zones, like a graph with interacting nodes or topological space with overlapping open sets.
In some zones, there is a film of proteins covering a lipid bilayer as Danielli and Davson suggested. In other zones, there are no proteins and only a lipid bilayer, as Gorter and Grendel imagined 100 years ago. And in yet further zones, there is no lipid bilayer or the fluidity is very low. For this reason, I consider zonation to be a foundational concept for membranes. All these zones assemble and interact through an intricate code, so we call our model the proteolipid code.
The proteolipid code is gigantic in terms of mass, volume, and components. There are often hundreds or thousands of different lipid and protein types in a single membrane. Each protein has a ‘fingerprint’ of lipids - its own unique distribution - and we consider proteins and their fingerprints to be one type of zone. These zones cluster to form another zone called a protein island, which often depends on fingerprint-encoded compatibility. Voids, which are devoid of protein, are a third type of zone and are thermodynamically coupled to protein-containing zones.
All zones contain unique attributes. For example, any of these zones can recruit proteins from outside the membrane with lipid ‘codons’ to cause events like endocytosis - a process where substances are brought into the cell. Other lipid functionalities exist as novel predictions of the model, such as lipid glue and lipid antagonists competing for fingerprint sites to control protein clustering.
The proteolipid code has generally been well-received, and I am transitioning away from experiments to spend more time building it. I believe we need to create a database of zones to make membranes tractable to machine learning. This will eventually allow us to predict the structures and dynamics of membranes with high accuracy and detail. I expect that in due course, the proteolipid code will be widely accepted and come to be viewed with similar esteem to the genetic code.
Learn more about the proteolipid code online on the BMC Biology website: https://bmcbiol.biomedcentral.com/articles/10.1186/s12915-024-01849-6