AUSTRALIAN FRONTIERS OF SCIENCE, 2005
Walter and Eliza Hall Institute of Medical Research, Melbourne, 12-13 April
Biomolecular interactions in development and disease
Dr Jacqui Matthews, Viertel Senior Medical Research Fellow, University of Sydney
What I am going to
be talking about is some regulatory proteins in disease and development, and
you will see that my work has a lot of tie-ins with the type of work that Jamie
Rossjohn is doing and has been talking about.
I am going to introduce my topic by bringing up another side of the immune system.
(Click on image for a larger version)
This photograph shows a little girl that you can’t see very clearly. And you can’t see her very clearly because she is inside a bubble. She has almost no functional immune system whatsoever. So she has none of the T-cells and B-cells that allow the immune system to function.
She has SCIDS, a specific immune deficiency which is usually caused by a defect in a single gene. Because it is a single-gene disease and it is also something that affects blood cells, which are a fairly accessible cell pool, it was thought that it would be a fantastic target for gene therapy – to be able to actually come in, replace those genes that are defective, and get healthy, functional human beings again.
Probably about five years ago now, people set about doing some limited gene therapy trials with children that had SCIDS. And it was working incredibly well. Most of the children were either recovering at least partially or fully from this immune disease.
But a few years into the trial, it was found that several of the children developed leukaemia symptoms.
(Click on image for a larger version)
They developed the same symptoms and it was the same cause in each of these cases: a protein called LMO2 was being overexpressed or overproduced in the T-cells as a result of the gene therapy treatment itself.
It turns out that LMO2 actually causes a number of other childhood leukaemias. In fact, that is how the protein was first discovered, because it was overproduced in leukaemia T-cells.
So how does LMO2 work, and can we try and manipulate the function of this protein to see if we can actually stop disease, can treat T-cell leukaemias in a fairly specific manner?
(Click on image for a larger version)
Well, most of the time LMO2 is a good protein. We actually need to have it around, because if we don’t have it we have no blood cells whatsoever. It is absolutely required for the normal development of blood cells. But it is not normally found in T-cells. So here we have got a situation where a gene is being switched on at the wrong time and in the wrong place, and everything goes haywire. It is also required for blood vessel remodelling – so it is a part of normal developmental growth, where you need to develop new blood systems – and also in cancer, where solid tumours require new blood vessels to feed them and to supply them with nutrients.
(Click on image for a larger version)
LMO2 is not an enzyme, it doesn’t catalyse any specific reaction, it doesn’t directly bind to DNA. All it really does is to bind to other proteins, so it is much like the receptor proteins that Jamie was talking about. And it turns out that what it does is to nucleate protein DNA complexes that regulate gene expression, so they switch on or down-regulate or switch off the expression of specific genes.
It has a number of partner proteins that I am going to be talking about quite a lot through this talk. I am actually going to not really refer to them by name but by colour, because in one sense it doesn’t really matter what their names are. But I will just point out that one of these proteins, this yellow protein Tal1, is also implicated in many different T-cell leukaemias, and LMO2 and Tal1 are overproduced in most T-cell leukaemias. So they are proteins that should not be there, and they are doing something wrong.
But these complexes also are found in normal blood cells, and that is where they are doing the job that they are supposed to do. So we have got this situation where we are trying to find out what happens in normal systems and also what is happening in cancerous situations, to see if we can control what is going on there.
(Click on image for a larger version)
One of the things we are trying to do in my lab is to find out how these different molecules come together, how these different proteins interact with each other and interact with DNA.
(Click on image for a larger version)
We want to know several things. How many copies of each component are there – are there just one, or two, or a dozen? Which are the key components of the complex – which are the really tight binders, which are the things that we want to be targeting if we want to try and block a specific interaction, to be able to treat disease? And what do they look like – what can we learn about their structure that will enable us to try and block specific interactions in order to treat disease?
(Click on image for a larger version)
Next, can we block assembly? And what if we block specific protein-protein interactions? Unlike in Jamie’s case, where we want to enhance a particular interaction, here we usually want to block a specific interaction.
(Click on image for a larger version)
So the first part of this question is really: how do these complexes assemble? How do we get from the individual components to a nice active complex?
(Click on image for a larger version)
One of the first things that we have to do in our lab is actually to break things down a bit. We often end up with proteins that are very big, and they would be hard to handle in a laboratory situation if we kept the whole protein there. We actually can’t do anything with the whole protein. So one of the first steps we have to take is to break it down into pieces that we can manage and to find out what is going on.
So we basically chop up our proteins into different segments, or domains, and we try and find out which of the domains interact with one another in pair-wise and higher order complexes.
(Click on image for a larger version)
So we might use something like the yeast-two hybrid assay, which is just an assay system that is set up in yeast cells; it is a modified yeast strain. And if we are trying to look at a pair-wise interaction between two different proteins, if those proteins interact – which would be the case here – we get cell growth; if they don’t interact, we don’t get cell growth. So it is a kind of a quick on/off assay for seeing what is going on.
And so here you will see that in this case we have got a couple of things going on. We’ve got cell growth and we’ve got things turning blue; in some cases where there is interaction occurring you’ve got growth and blue-colour production, in other cases there is none. So it is a nice simple assay to see what is going on.
(Click on image for a larger version)
And when you do those and other similar experiments you can find out that there are specific domains in these proteins that are involved in the complex formation, and other domains are doing other things.
(Click on image for a larger version)
We want to try and measure the interactions. What we have just done is a kind of an on/off type of system; we want to get some better idea of what is going on in interactions. What we tend to have to do is to use a whole range of different techniques to enable us to look at all the different components of the interactions, some of which are low-technology applications, which often lead you towards the answers you are after, and some of which are high technology. A low-technology one, to me, would be like this gel-shift assay, which is very similar to the assay that Trevor showed you earlier. He showed a native gel; this is the same thing. It is a native gel, but we are looking at binding to DNA, shown here as free DNA, or complexes involving DNA being shifted to various stages.
This can give us some information about what is really going on in binding, confirming interactions and seeing what is going on.
(Click on image for a larger version)
Or we can use much higher-tech information like ITC, isothermal titration calorimetry. This is where we take advantage of the fact that any time you have an interaction or, essentially, any process you will either give off or take up heat. So what we have is an instrument that is exquisitely sensitive in terms of measuring heat changes. And if you titrate one protein or one component into another, you generate heat and you are able to generate binding curves and work out quite a lot about what is going on in the interaction. It is a method very well suited to measuring weak or moderate binding affinity interactions.
(Click on image for a larger version)
It is not suited to all purposes. You need a lot of protein for this sort of experiment. It is not good for very high-affinity binding. So we might have to move to something else, which would be Biacore/SPR, which has its own advantages and disadvantages. So we use a massive range of techniques to try and see what is going on, to try and get the answers to everything.
(Click on image for a larger version)
When we have used these different techniques, what we have found with this system is that our red protein E2A liked to form a dimer, it liked to bind to DNA on its own, but when you have the yellow protein Tal1, which also forms a dimer on its own, they preferentially form a heterodimer between the two things and they bind to DNA as well. And other components bind with different affinities. LMO2 prefers to bind to Tal1 in a heterodimer rather than as a homodimer. So we are working out the different combinations and the different ways these things interact. LMO2:Idb1 is our highest-affinity binder, so this is one of the regions that we have been focusing on quite a lot.
(Click on image for a larger version)
We also want to know what these complexes look like, and again we tend to use a range of different methods. Where possible we use very accurate methods that give us high resolution, to find out the type of information that Jamie was talking about; at other times we have to use more approximate methods because the accurate versions just won’t work for us in that situation.
 |
 |
 |
| (Click on images for larger versions) | ||
So we use NMR a lot – it is a solution-state method, and it has been very useful in enabling us to solve smaller structures of complexes. We also use crystallography where we possibly can, because crystallography is faster (when it works) and we have been able to solve structures of much larger complexes as well. So we can see in very high detail what is going on in some of our situations.
(Click on image for a larger version)
In other cases we haven’t yet been able to solve structures and we have to turn to things like homology methods to see what is going on. The red protein shown here is similar to some other proteins for which the structure is known, and applying a few other experimental methods we can assume that our red protein is going to have a structure somewhat like this. It is an alpha-helical protein and we can double check that it actually has alpha-helical structure by using methods such as circular dichroism, where you get a double minimum which is typical of alpha-helical structure. So we can learn lots of different information by using different methods.
(Click on image for a larger version)
So now we are going from something that looks like this – a blob, a very schematic structure, just a rough idea of what is going on.
(Click on image for a larger version)
Now we have a much better idea of what everything looks like and how everything packs together. We don’t know everything yet; there are still things we don’t know. We haven’t yet been able to solve the structure of the whole complex being formed together, and that’s something for which it is going to be much more useful to have synchrotron-type radiation around, to get ready access to that type of experiment, to deal with these larger complexes.
(Click on image for a larger version)
By learning some of this information, how the different proteins come together and what they look like, we can start to understand a few things about how these different complexes work. In these T-cell leukaemias that are induced by LMO2 and Tal1, what is happening is that we are forming some complexes that are probably switching on some gene patterns that aren’t supposed to be switched on in T-cells. The other thing we are doing is preventing this red E2A protein from doing its normal job. It can normally bind to DNA on its own, and it can activate the transcription or the gene expression of certain types – so, actually switching off other complexes. By having these proteins appearing at the wrong time, we are just mucking up what should be happening in the cell, and the carefully regulated processes are all going wrong.
(Click on image for a larger version)
So, can we block assembly? And, if we block a particular interaction, will that be able to help us treat diseases like T-cell leukaemias?
As I said, the LMO2:Idb1 interaction is the tightest one we know about, and LMO2 is the one that we know most about – the structure between those two proteins is the one that we have at highest resolution at the moment, so we know a lot more about it. So I will tell you some of the stuff we have been doing to try and develop inhibitors to see if blocking those interactions will be useful in treating disease.
(Click on image for a larger version)
Here you can see our structure. We have got the LMO2 protein coming along the bottom, we have got a yellow protein – which used to be orange in my other schematics but you will just have to live with that at the moment – and you might be able to see that we have actually got a very long interface going on. It is very broad. So it is quite important for us to see if we can find one of those hotspots of binding that Jamie was just talking about. Hotspots are things that drug designers like to target, because they are the bits that you actually want to block from forming. It might not be so important to block part of the interaction that doesn’t really do much, I suppose; it’s just there, holding things together.
(Click on image for a larger version)
So again what we do is to make a whole series of mutants, and we test them for their ability to bind. If they no longer bind, we can say that they are probably important in binding, and by doing this we have been able to work out some of the binding hotspots.
(Click on image for a larger version)
This very pale blue region at the left-hand end is the most important region of our LMO:ldb interaction, our blue-yellow interaction here. And some of these other residues are also important for binding but they’re not nearly as important as those.
So what we are trying to do is to use this wild-type or native sequence, which comes from a much larger protein, and produce some peptides that bind to the blue protein much more effectively than the wild-type protein would, so that we have got some very effective inhibitory peptides for use.
(Click on image for a larger version)
The way that we are going about doing this is to use ‘phage display selection’. We take part of the peptide I showed you in the previous slide and we randomise it. All the rest of it is still able to bind, so it has got some intrinsic binding activity there. And what we are trying to find is tighter binders.
(Click on image for a larger version)
We do this in a very random way so we don’t actually know what we are dealing with, and we are using phage display, where we take a biological entity – the phage particle – which has protein displayed on the outside, including our randomised peptides, and it has got DNA on the inside. Now, we can perform some binding steps, where we are going through some cycles of binding and washing things away, and by doing this we find the members of the library that have the tightest binding affinity for our target protein, which is our blue LMO2 protein that I have been talking about all along.
Once we have found those, we can use DNA sequencing methods to find out the identity of those peptides.
(Click on image for a larger version)
We did this against LMO2, and we have come up with a family of peptides, some of which came up several times, which bind effectively. We can actually get some extra information from this sort of study, showing that some of these residues are essential for binding and others are not. We can change them around.
(Click on image for a larger version)
All of those peptides that we selected all bind much more tightly than the wild-type peptide does. So this is our wild-type binder on the right, and on the left we have all of the other peptides that bind about tenfold more strongly than the wild-type version. So it is a nice method for not having to go in and rationally design things but actually being able to use combinatorial methods to get good binders.
 |
 |
| (Click on images for larger versions) | |
This is still a bit of a work in progress. What we are doing at the moment is to target other parts of that peptide in the same way, randomising different regions, and trying to select out features all the way along this sequence, in order to generate one final affinity-matured peptide that should bind to our LMO2 protein with very high affinity. We are hoping that this will be a useful lead compound for T-cell leukaemia treatment.
(Click on image for a larger version)
I am just going to leave you at that point. We haven’t taken this much further at the moment, but what I hope I have just been able to show you is that normal proteins can cause disease if they are produced at the wrong place and at the wrong time, and that by understanding how different proteins and other biomolecules can come together, then the idea is that we might be able to develop reagents and be able to modulate those interactions and, hopefully, treat disease.
(Click on image for a larger version)
I will just acknowledge some of the main people who did the work here. Daniel, Janet and Ann have done various aspects of this project. These are two groups which Joel Mackay and I run, together. I would also like to acknowledge the X-ray crystallographers in our department, who have helped with X-ray crystallography; Jane Visvader and Geoff Lindeman at the WEHI, who are our main collaborators on the biological front; and various funding sources who have given us lots of money for doing this.
