Steps on the road to the Holy Grail

Following nearly two decades of hard work, a team of scientists headed by CSIRO's Dr Colin Ward published a major breakthrough in structural biology last September in the journal Nature - the molecular structure of the insulin receptor.

Solving this structure provides a huge leap in our understanding of this very important receptor on the cell surface and how it interacts with insulin to mediate signalling. Ultimately, it may also lead to novel pharmaceuticals in the fight against diabetes.

Ward will talk about this personally satisfying structural conquest at the 32nd Lorne conference on Protein Structure and Function, being held next week.

Ward actually began his science career as a parasite biochemist. He did his undergraduate and doctoral studies at the University of New South Wales in the mid-1960s, gaining the university medal, before pursuing a postdoctoral fellowship in Boston. He returned to a position with CSIRO in 1970, and is now an honorary fellow of CSIRO Molecular and Health Technologies in Melbourne.

For nearly two decades, Ward and colleagues from CSIRO and elsewhere have concentrated on the structure and function of the insulin and epidermal growth factor receptor families, important potential therapeutic targets in diabetes and cancer. This work has produced many significant scientific breakthroughs along the way, with the most recent perhaps the sweetest, according to Ward.

Since 1990, Ward has wanted to understand the molecular structure of the insulin receptor (IR). Although the structure of insulin was solved in the late 1960s, the equally clinically important insulin receptor, a tyrosine kinase-type receptor, has proven extremely intractable to structural determination.

The receptor DNA sequence was published in 1985, and the structure of the intracellular kinase domain reported in 1994, but Ward's group, the pharmaceutical company Novo Nordisk, and several other top groups worldwide have been trying hard to solve the structure of the IR ectodomain for 16 or so years.

The right conditions

Ward says the difficulty in solving this structure lay in its extremely large size and multiple domains - in vivo it exists as a dimer of 2 x 920-amino acid monomers linked by a disulphide bridge.

The protein is also heavily glycosylated, which adds around 60 kDa to the molecule. In fact, the successful purification came from a cell line that produces a functionally intact IR that is far less glycosylated, alleviating one problem at least.

"The big problem was getting the right conditions for producing crystals," Ward says. "The process of solving the structure was actually straightforward."

The crystallisation success came with a complex of the receptor ectodomain bound to four monoclonal antibody fragments, which saw the receptor suspended in its upside down V conformation by the antibody complex, "sort of like a hang-glider", as Ward terms it.

It has been known for some time that the insulin receptor on the cell surface exists as a preformed dimer that is inactive for phosphorylation and signalling until bound by insulin. Based on a wealth of biochemical and earlier structural studies, Ward and others predicted that the initiation of signalling must involve only a subtle rearrangement in the receptor domains.

Ward knew a lot and suspected even more about the receptor after solving half of the ectodomain structure a few years back, but as he explained, "solving of the complete ectodomain structure now shows us exactly how the domains are organised and so we can start to work out what the activating movement might entail."

Unexpected conformations

In his invited presentation at Lorne, Ward will discuss the key points about the insulin receptor revealed by the published structure and discuss implications of these findings for future research.

In summary, the structure revealed an unexpected folded-over conformation that places the ligand-binding regions in juxtaposition.

This arrangement is very different from previous models. It shows the two leucine-rich (L1) domains (conserved across other such receptor families) to be on opposite sides of the dimer, too far apart to allow insulin to bind both simultaneously. It therefore appears that there are major differences at the two regions of the receptor governing ligand specificity.

Ward describes the insulin receptor as "schizophrenic" in that it uses two different receptor motifs to bind insulin.

"Initially the classical receptor L1 site binds to the ligand as seen with other tyrosine-kinase receptors, but then the second domain with the fibronectin loops comes in and interacts with a different face of insulin, like a cytokine receptor," he says.

Another group earlier postulated that two surfaces on insulin participated in binding with the additional interaction, providing stability and perhaps allowing the subtle change necessary to activate the complex for signalling.

"The really exciting thing is that our structure now completely fits with these findings."

The structure also supported a plethora of mutation studies and alanine-scanning experiments in the early 1990s that identified many of the binding sites for insulin on the receptor, Ward says.

"Solving the entire ectodomain 3D structure now shows that all of the important sites are on one face of the molecule and all the non-binding sites are on the other, outward face."

Ward says he finds it particularly satisfying to be able to draw on and confirm much of the biochemical data that has gone before.

The Holy Grail

So we now know the complete structure of insulin and the insulin receptor. What next?

"We now want to desperately know what the receptor bound to insulin looks like," he says, and that is what his group is now pursuing.

That information is what will launch future drug discovery programs, but Ward believes it will prove a very difficult prospect for many reasons. For one thing, the soluble ectodomain used for the crystallisation studies, although functional, has a slightly different conformation than it does in vivo, and some of the antibody fragments used to obtain crystals bind competitively with insulin.

Fortunately an awful lot is known about both proteins and Ward's team has a few ideas for how they might get around the difficulties, such as the use of mutants that assume more amenable conformations. However, he is not expecting to achieve this next major breakthrough any time soon, although as he says it would be nice and he is forever hopeful.

Facilities such as the new Bio21 Collaborative Crystallisation Centre in Melbourne, of which Ward is the chair, and the Melbourne-based Australian Synchrotron, due to come online early next year, will help the quest enormously, even if it is only to avoid 'burnout' of those poor souls who fly to Germany with their crystals intact and fingers crossed to do a 24-hour synchrotron shift then fly home.

Ward has described the possibility of designing drugs and chemicals that successfully mimic insulin as the "Holy Grail" of this field that many people see as unattainable. However, he relishes such a challenge and will certainly keep working to prove the naysayers wrong.

"The findings just published in Nature clearly are pointing the way in which we should be attacking such a difficult problem."