The mammalian genome and phenome

Chris Goodnow talks about defining the mammalian phenome and how this may pan out in the future.

There are approximately 30,000 genes in the mammalian genome -- only twice that of a fruit fly or roundworm.

If we take the simple view of a gene as conceived by Garrod, Beadle and Tatum -- where one gene makes one protein makes one biochemical product -- then a doubling of genes can't support the massive expansion in cellular complexity and behaviour exhibited by mammals.

However, we already know enough about mammalian genes and proteins to know that they rarely work this simply. Most biochemical processes in a cell involve combinations of 10 or 20 interacting proteins, as illustrated by signalosomes, ribosomes, proteasomes, nuclear pores, transcriptosomes, etc.

We also know that individual mammalian genes and proteins are usually multi-tasking: the same gene and protein is used in combination with different partners in different cells types to produce distinct cellular traits and activity. The signalling enzymes, adapters and transcription factors that act in different blood cells are lovely illustrations of this point.

Combinatorial specificity

Combinatorial specificity is at the heart of the mammalian genome-phenome link. Additional functions have been progressively grafted onto ancient proteins, as illustrated by the condensin proteins that coordinate chromosome segregation at mitosis but do a whole lot of extra tasks as you move up the phylogenetic tree.

The average mammalian protein has nearly twice as many domains and sites for forming partnerships as the average invertebrate protein. If you double the number of proteins encoded in the genome, and double the average set of combinatorial partners, the potential complexity of the mammalian phenome is astronomically larger than an invertebrate.

Indeed the real challenge lies in constraining that complexity, which is presumably the function of epigenetic closing off of many genes into heterochromatin in most cells of the body.

Specifying cellular complexity in this combinatorial manner has two fundamentally important practical implications.

Natural variation

First, we won't decipher the combinatorial code simply by knocking out proteins altogether. We need tools that can reveal what happens when you interfere with individual domains and interacting sites of proteins and genes.

Natural variation of individual DNA nucleotides, or controlled versions of this process performed in laboratory mice as pioneered at the Australian Phenomics Facility, provides just such a tool.

We have many examples now where human and mouse point variants in a gene illuminate dramatic roles for genes that are obscured by complete knockouts. These point variants provide a way to move from the Garrod/Beadle way of conceiving genes towards the quantitative trait variants that are key to natural human variation in disease susceptibility.

Targeting protein combinations

Second, drug development today is still focused on single proteins and targets, to some extent searching for target proteins that behave like Garrod and Beadle's beasts, with a single function.

Even when we understand the combinatorial code, there remains a great challenge of targeting protein combinations with small molecules. When we reach that goal, disease treatment or prevention will be on par with the specificity of the root causes of disease.

Christopher C Goodnow FAA, is director of the Australian Phenomics Facility and Professor of Immunology and Genetics, John Curtin School of Medical Research, The Australian National University.