Science, People & Politics, issue 3, Volume i, Volume II,
published 4th May, 2009 by Gavaghan Communications.
synchrotrons and drug design:
The how and why of the structure of proteins.
by Liz Carpenter*
BIBLIOGRAPHY AND PHOTOGRAPHS OF SAMPLES AND EQUIPMENT
If you really want to understand how the molecules in your cells work you need to understand their three-dimensional structures. It is all very well to say this molecule performs this chemical reaction or that function in the cell but to understand how you need to know where all the atoms are and how each part of the molecule moves during the process. Thus armed you can see how cells capture light to make stores of energy, say, and how cells consume energy, how DNA and proteins are made, how muscles work and how hormones function. You can understand how small molecules and drugs penetrate cells and are expelled. All these processes are fundamental to understanding how cells work and, of course, to understanding how to control the situation when things go wrong.
So what are these molecules we are interested in? They might be as simple as a salt molecule, which is just sodium and chloride atoms. Or they may be as complex as a large protein molecule with several thousand carbon, nitrogen, oxygen, hydrogen and sulphur atoms arranged to form a long string of amino acids, with short branches called side chains hanging off the string.
There are 20 possible side chains, each with a defining atomic groups in the side chain known as a functional group. The order of these side chains decides both the three-dimensional structure of the protein and the function. The side chains make the protein fold into a ball with a precisely defined structure and every atom is held in electronic balance and has an exact place in the structure. It would be great if we could just read off the sequence of a protein, encoded in the DNA, and use a computer to say how the protein folds up and what function it performs. Sadly, up to now, we don't have the computational methods to work out protein structure and function purely by computation, so we have to do experiments.
One of the experimenters more powerful techniques for elucidating protein structure is X-ray crystallography. All you need to do is get your protein nice and pure, persuade it to form an ordered crystal, where all the molecules are lined up in a neat array, one next to the other, then you zap your crystal with a very intense beam of X-rays and out comes a diffraction pattern. The diffraction pattern is just a lot of very weak secondary X-ray beams coming out of your crystal, which you shine onto an X-ray detector, and you can measure the intensity of these weak beams. From the position and strength of the diffraction pattern you can then work out where all the atoms have to be in the molecule to give the pattern, and there you go, you have your structure.
Sounds simple doesn't it?
It isn't, but certainly it is possible for many important biological molecules. The first and most difficult problem is to make your protein. We usually modify bacteria, yeast or insect cells by adding the gene for our target protein to the host cells and commandeering the host cells' protein-making machinery to overproduce large quantities of the protein we want. This often works better than purifying proteins from their normal environment because cells often produce only small amounts of each protein. We have to grow large amounts, say 50 litres of liquid cultures of the host cells, with our protein produced inside. For one recently solved project we grew enough 50-litre cultures of protein-producing E.coli bacteria over the years to fill a swimming pool.
Once the cultures exist we have to break open the cells, extract the protein, often in the presence of detergents, if the protein that interests you is very oily. Then you need to purify your protein, removing all the other proteins that are part of the host cells. These days we do this by modifying the gene for our target protein to include a series of amino acids with affinity for a resin, then we can pull the protein out of the cell soup because it is the only one which binds to that resin. In the old days we had sieving columns and charge-binding columns and precipitation with salts and many other tricks, but the affinity resins that bind and extract only your target protein have made life a lot easier.
For the next stage of the process we need to persuade the protein to form a crystal, an esoteric process but essential for X-ray crystallography because a single molecule in an X-ray beam would give a very small signal and would die very quickly. By lining up the protein molecules in an ordered way we can get them to act as an amplifier, providing a stronger signal, and, if you are lucky, some of the molecules will survive until the end of the experiment.
Growing crystals is not a trivial process. Sadly we do not yet have the magic ingredient that causes all proteins to fall into a perfect neat array, so we are left with empirical testing of many hundreds of possible crystallisation agents. This was all done by hand a few years ago, but now we have commercial screen kits with carefully optimized potential crystallisation conditions, robots which prepare crystallisation plates, and experiments which are done on the 100-nanolitre scale so that we can get 10 or 100 times more experiments out of the small amount of material we can prepare. Our lab at the Diamond light source has a robot which can prepare up to 2000 crystallisation drops without human intervention in 2 or 3 hours, whereas before a person would have spent a day or two performing this process. The robot has the added advantage that it does not get bored and make mistakes in pipetting solutions, so it is more accurate.
When it comes to identifying crystals in a crystallisation drop the human being wins. Our robot can take pictures of the crystallisation drops in the crystallisation plates, but it is no good at finding crystals, which can be any shape and size and are often buried in precipitation. A human being still has to spend hours looking at images on the computer, or down a microscope.
These crystals can be between 10 and 500 microns long and handling them is still something that requires a human being with steady hands. We extract the crystals from the crystallisation plate by carefully fishing out the crystal with a tiny fibre loop on the end of a metal cap, dropping the crystal into some preserving cryoprotectant solution, then fishing it out again with the fibre loop and plunging it into liquid nitrogen. With the right 'cryoprotectant' solution, careful handling and a good dose of luck, you will end up with cryocooled crystal which will diffract as well as a crystal at room temperature and will survive 10 to 100 times longer in a powerful X-ray beam.
At this point in the process you have your crystal frozen in a fibre loop and sitting in a little plastic vial in a vat of liquid nitrogen. Then you need some X-rays. The stronger the better. X-ray generators have been around for around 100 years. There are small X-ray sources you can install in your own lab. We have one, about the size of an office water-dispenser, for doing initial checks on crystals. But for really good quality X-ray diffraction patterns we use the amazing facilities at particle accelerators called synchrotrons. There is one in Oxfordshire - the Diamond Light Source. Diamond is a £250-million research complex and the largest investment in science in Britain for 30 years. Of course it is not just for our work, we use four of the existing beam lines. Others are used for a huge variety of science, from physics, to chemistry, engineering, biology and even archaeology.
At the heart of this system is a circular tube about 4 cm across and about 1 km long, which carries a series of packages of electrons moving at huge speed in a 1-km circle. Electrons normally fly in a straight line, so to make them go around in a circle there are magnets providing a magnetic field which forces the electrons to change direction to go around the ring. When the electrons are change direction they emit some of the energy in the form of X-rays, which comes off the ring at a tangent. All we have to do for our experiment is to place our frozen protein crystal inside a lead lined room in a beam of these X-rays and the X-rays will interact with the electrons around the atoms in the proteins and be diffracted. This gives a diffraction pattern, recorded as a series of spots on an image. We rotate the crystal in the X-ray beam and record a whole series of 90 to 180 images from a crystal and this is our dataset.
Often for difficult cases we will need to test a large number of crystals, many of which will not diffract well or will decay rapidly in the X-ray beam. We need to test, optimize the crystallisation, test again and finally find the best conditions to get a good diffraction pattern. Then we often need to modify the crystals and collect more data until we can finally resolve the structure. For one recent project we tested more than 2000 crystals, all frozen by hand, then collected over 100 datasets before the structure could be resolved. This was an extreme case though, involving particularly difficult proteins called membrane proteins, of which I shall say more later.
In easier cases it is sometimes possible to grow crystals of a soluble protein in initial screening conditions, collect several datasets from a single crystal, then solve the structure on your portable on the train on the way back from the synchrotron. Somehow, though, it often turns out that the easy ones aren't the ones that give you the most information about the biology of cells and the molecules you really want to understand are the ones that are most resistant. If you really want to know the structure and you have enough funding you will persevere and with hard work and determination most structures can be solved.
Once you have enough good quality datasets, there is a certain amount of time on the computer to calculate first an electron density map, which is a three dimensional representation of the location of all the electrons in the structure, then the positions of the atoms, which will be where the electrons are in the map. The time for this process varies from a few hours to a few years, depending on the quality of the data and the maps, but finally you will be in a position to see the structure of your protein, the most thrilling moment in any crystallographers research career. From the structure you can see how DNA is cut by proteins, how drugs are thrown out of cells, how proteins are made and how mutations can lead to cancers and heart disease.
Just to make life more interesting my group works on some particularly difficult, but scientifically and medically important proteins. These are the membrane proteins, the molecules that are embedded in lipid bilayers which form the surface of our cells. This oily, lipid bilayer, which is what the cells needs to hold the cell together with its contents protected from the outside environment, makes it very difficult for most molecules to get in and out of your cells.
The problem then for the cell is that some molecules do have to go into and out of the cell: eg, sugars, amino acids, salts, water and some proteins. Without food and water the cell would die. So the cell has protein molecules embedded in this lipid bilayer, which bind and transport carefully selected molecules, so that the cell takes up and removes only those molecules that it wants and needs. This process has to be carefully controlled, or the cell would die and so the proteins that are embedded in the lipid bilayer are carefully tuned to transport only the right molecules in the amount needed by the cell. Other membrane proteins relay signals from hormones into cells. They bind a hormone and perform a chemical reaction inside the cell, which sets off a chain of reactions which signals to the cell to change how it is working.
Not surprisingly when these processes go wrong, it is pretty disastrous for the cell and the organism. Problems with membrane proteins are linked to many diseases, from heart disease and cancer, diabetes, neurological diseases, depression and cystic fibrosis. Resistance to drugs is often also caused by membrane proteins, which either throw cancer drugs out of cancer cells or remove antibacterial drugs from the bacteria, making them resistant to treatment. By controlling the proteins that move molecules into and out of cells we can produce drugs for many of these serious diseases. In fact these proteins are already the sites of action of many commercial drugs, eg calcium channel blockers for heart disease.
Many of these drugs were discovered first and found to bind to membrane proteins afterwards, but it is vital that we understand how they interact, so we can design new and better drugs, with fewer side effects. In many cases both the effectiveness of a drug and the side effects caused by the drug are specific to the individual. As we enter the era when it will be practical to rapidly sequence the relevant genes for each individual, we will need to understand at the molecular level how the differences observed in the genes will affect the structure and function of the proteins where the drugs bind.
Illustration by Helen Gavaghan BSc (hons)©
*Liz Carpenter is a research fellow at Imperial College, London and a group leader in the Membrane Protein Laboratory at Diamond. So Iwata, who holds the David Blow chair of biophysics at Imperial College, London, directs the laboratory. The group has solved 12 of the total of 170 known membrane protein structures. The research team also trains membrane protein scientists who want to learn about crystallography and crystallographers who want to learn about working with membrane proteins. Some 25 research groups in Europe use the laboratory, which is a collaboration including Imperial College, the Wellcome Trust and Diamond Light Source Ltd. For more about the team visit http://www.diamond.ac.uk/Science/MPL/aboutus.html