The rational lottery
“Rational drug design” sounds like a perfectly sensible method in the quest to understand small molecules and proteins, and identify the compounds which could lead to cures for killer diseases. But in the history of translational medicine, fashions come and go, so who is to say which approach will produce results faster? Serendipity is also a significant factor. Professor Malcolm Walkinshaw believes there is a place for many different approaches (including serendipity), and his experience…
“Rational drug design” sounds like a perfectly sensible method in the quest to understand small molecules and proteins, and identify the compounds which could lead to cures for killer diseases. But in the history of translational medicine, fashions come and go, so who is to say which approach will produce results faster? Serendipity is also a significant factor. Professor Malcolm Walkinshaw believes there is a place for many different approaches (including serendipity), and his experience in industry and academia has taught him that these different methods complement each other – for very good reasons...
Things have changed a bit since Professor Malcolm Walkinshaw was a young man studying chemistry at the University of Edinburgh. Now the Director of the Centre for Translational and Chemical Biology in Edinburgh, Walkinshaw is well known for his work in a number of areas, including structure-based drug design, database development and virtual screening, but in the early 1970s, the technology was not quite as advanced.
For his PhD, Walkinshaw used X-ray crystallography to study the structures of small molecule sugars, and this laid the foundations of his gradual move into structural biology. His post-doctoral work involved the use of X-ray fibre diffraction methods to study polysaccharides and DNA structures, to understand how molecules recognise and communicate with each other, “using basic techniques like crystallography to understand, in atomic detail, the building blocks of biological macromolecules and begin to explain the energetics and thermodynamics of their interactions. The hope was that these insights and principles could be applied to bigger biological systems.” This in turn led to his later work in protein production and characterisation, studying how proteins unfold, bind and interact, as well as studies of protein-protein interactions.
Walkinshaw's early research involved a lot of complex and tedious calculations, but in those days to use a computer – which was housed in its own special room – you had to load three or four trayfuls of punchcards (used to compile the program and read the data) into the computer, then wait 24 hours to get the result. The next day, a van would arrive with the answer, and if a single punchcard had been bent or damaged in some way, you would have to run the program all over again.
“It was primitive computing,” says Walkinshaw. “Today, the process is infinitely faster – it only takes a couple of minutes on a laptop. And as a result, the pace of research has increased.”
The technology used to generate data has also advanced out of all recognition. For example, in the past, to “solve” a crystal structure, the diffraction patterns were collected on photographic film. The films were then developed and scanned in order to calculate the 3D protein structure. Such a process took weeks and months, but nowadays, all this is done in real-time to generate a complete dataset in just a few minutes, and processes that used to take up to a year can be completed in a day.
As Walkinshaw's career progressed, various other trends also emerged. “By the late 70s,” he says, “it was clear that protein crystallography was going to be the major method to answer questions about molecular recognition in biological systems.”
Walkinshaw has always been interested in classifying the properties of molecular structures to find out why the molecules behave the way they do and this later led to his focus on rational drug design. According to Walkinshaw, in fields like biochemistry and translational medicine, fashions come and go. Through the years, rational drug design has fallen in and out of favour, but in its latest incarnation it is healthier than ever. In the late 1970s and 1980s, the focus was on protein crystal structures and the challenge was to design small drug-like molecules using a structural (or ‘rational’) approach. The development of molecular graphics also occurred around the same time and opened up new possibilities of being able to visualise molecules in three dimensions. The availability of numbers of protein structures coupled with the newly available computational and molecular graphics techniques allowed structure-based design to take centre stage.
This emphasis on rational drug design steered the research of the big pharmaceutical companies for a number of years but then began to lose favour to the newer combinatorial chemistry and screening approaches. In the mid-1990s, says Walkinshaw, the use of protein crystallography and structure-based design was not producing drugs as fast as pharma desired, and the fashionable new approach was high-throughput screening.
The big pharmaceutical companies were producing and cataloguing millions of compounds and looking for the one in a million that may eventually produce a working drug. But in recent years, huge advances in technologies to solve 3D structures of proteins, along with much more powerful computers, have also improved the hit-rates in structure-based or ‘rational’ approaches. “Rational drug design is alive and well,” says Walkinshaw, “and it works best in conjunction with other complementary approaches including biophysical studies, sophisticated computing techniques and high-throughput screening.”
Walkinshaw describes the scale of the drug-discovery problem by comparing the number of possible drug-like molecules that could be synthesised to the number of molecules in all the world's oceans – something in the region of 1050 – against that we only have about five to 10 million molecules available for screening, or as Walkinshaw puts it, “literally a drop in the ocean.”
But we are making progress, he adds. Of the 30,000 proteins in the human body, we can now visualise 3D structures for about two-thirds of the total. Many of these proteins are potential drug targets and using computational techniques to screen integrated databases containing the details of millions of potential drug-like compounds, we can carry out ‘virtual screening’ tests more quickly than ever before. For example, once a target protein has been selected we can now use programs to identify a binding pocket and study its charge and its shape and use that information to identify a potential inhibitor, sampling up to one million compounds every two hours. Finding a compound which “docks” with a particular protein is never simple, adds Walkinshaw. “And you should always treat the results with healthy scepticism,” he explains. “Predictions are seldom completely correct but they point you in the right direction for getting a hit. Our suite of virtual screening programs is now producing a 30 per cent success rate which is very encouraging.”
The Walkinshaw lab currently works on a range of medically important targets. Of particular interest are antiparasitic targets, for example, in the search to cure parasitic diseases like leishmaniasis and sleeping sickness. The computational and structural work is being carried out in Edinburgh and the biological and screening studies in collaborators' laboratories in Glasgow, Brussels and Maryland. Walkinshaw explains that in all living organisims there are 10 steps involved in the conversion of glucose to pyruvate (a process called glycolysis). The aim is to intervene in that process by knocking out a pathway in such a way that you kill the parasite, not the host. “If we could hit a few of these targets, it would be even better,” says Walkinshaw. “This could lead to the development of an effective 'cocktail' approach for drug therapy.”
Lighthouse to ligands
Walkinshaw's career so far has followed a few twists and turns and has now come “almost full-circle.” While still a student, he worked as a lighthouse keeper in his summer holidays and was tempted to go full-time after he got his degree. He also considered joining the merchant navy. Like many biologists today, his first degree was in chemistry, but he was persuaded to change tack and enter the “now fashionable field of structural biology,” focusing on sugar structures. He then won a fellowship in Wolfram Saenger's lab in Goettingen and “managed to solve my first protein structure, a snake neurotoxin – one of the relatively few structures available at that time,” before returning to a full-time post in the Chemistry Department back home in Edinburgh, focusing again on small molecule crystallography, with an emphasis on molecular recognition and biological activity.
By the mid 80's the larger pharmaceutical companies were convincing themselves (prematurely, he says) that “rational drug design” was a fast and efficient way to new and more specific drug molecules. And in 1985, his next port of call was the Swiss pharma company Sandoz (now merged with Ciba to become Novartis), when rational drug design based on 3D molecular structures was becoming all the rage. For the next 10 years he built up a multi-disciplinary team using crystallography, NMR and modelling in what is now called “structure-based ligand design.”
One of his major contributions during this period was helping in the development of a drug called cyclosporin, an immunosuppressant drug widely used in organ transplants, now also being used for HIV/AIDS and Hepatitis. Cyclosporin binds to a ubiquitous protein called cyclophilin, a member of a family of 18 such proteins found in our bodies, and according to Walkinshaw, “the role of cyclophilin is still a mystery we are only beginning to understand.”
His home town exerted its pull yet again, and he came back to take up a new Chair in Structural Biochemistry in Edinburgh in 1995. His work since then has taken “a mainly medical slant,” including the study of protein structures that provide targets for drugs against parasitic infections. In collaboration with his colleague Paul Taylor, he has also developed a database mining program called LIDAEUS (Ligand Discovery At Edinburgh UniverSity) which searches small molecule databases for potential inhibitors, to “provide a starting point for combinatorial synthetic approaches.” Walkinshaw also says his career has gone almost full circle, collaborating with the chemistry department where he got his first degree – for example, a project to generate compound libraries based on database hits to produce some genuine “lead” molecules for protein targets.
Fifteen years ago, it was unusual to leave industry for academia, but Walkinshaw today observes more researchers following a similar path. In many ways, the aims of industry and academia can be extremely different, but one thing is the same: whether the objective is science or profit, and whether or not you use rational drug design, serendipity will always have a role to play, combined with the passion which Walkinshaw believes is essential to success in any branch of science or business.