Drug design: HIV Molecules made to measure HIV inhibitors have been one of the big successes of rational . Clare Sansom looks at the impact of on drug discovery DATA BANK DATA PROTEIN

50 | Chemistry World | November 2009 www.chemistryworld.org There are fashions in drug discovery, on this new virus. are over 300 residues In short as in many fields of technical Oroszlan had proved that a similar long and contain two copies of a development, and these even follow virus, the murine leukaemia virus,  HIV protease characteristic tripeptide sequence common enough patterns to be matured by proteolytic cleavage – inhibitors – now a key necessary for activity: mapped. Gartner’s ‘hype cycle’ that is, as new virus capsules bud component of multi- aspartic , threonine (or serine), describes the path of a technology from an infected host cell, drug HIV treatments and glycine – a trio generally referred from initial adoption, over the slice up its payload of polyproteins – are a prime example to as DTG. In contrast, the retroviral peak of inflated expectation and into individual, functional of structure-based drug protease is a much smaller enzyme through the consequent trough of and enzymes. Oroszlan had also design – only 99 residues in HIV – and disillusion, until reaching its ‘plateau found that the protease cleavage  HIV’s protease contains only one DTG sequence. It of productivity’, by which time it sites were resistant to all known enzymes were validated was clear from the beginning that has evolved to fit its particular niche mammalian proteases, suggesting as a potential drug target this protein could not be active as a in the market and become stable. the presence of an essential retroviral in 1985, sparking a race monomer. And structure-based drug discovery protease. to unravel the enzyme’s Mammalian aspartic proteases is a technique that has followed The sequence of HIV revealed structure such as pepsin have a bilobal the cycle exactly. It was extremely that it contained a similar protease  X-ray crystal structure, with a large cleft in which fashionable during the 1980s, after to murine leukaemia virus. In 1985, structures of the enzyme the binds and a flap that the structures of the first important Oroszlan published a paper showing, began appearing in 1989 folds down over it when it is bound. drug targets, including DNA and the through site-directed mutagenesis, –and the first drugs The two characteristic tripeptides enzyme dihydrofolate reductase, that protease was essential for HIV designed to inhibit it are symmetrically related and found were published. Disillusionment maturation and was hence a valid were licensed for use just at the bottom of the cleft, and the set in during the following decade target for antiviral therapy. At a six years later is supported by a tight with the downgrading of these time when the medical community  Structure-based drug network of hydrogen bonds that ‘rational’ approaches in favour of was struggling to cope with rapidly design has since become Tom Blundell and high throughput screening and developing HIV epidemics, that was an established drug formally from Birkbeck College, combinatorial chemistry. The extremely welcome news. discovery tool University of London, UK, later pendulum has since swung back, and termed a ‘fireman’s grip’. One structural biology is now a mature A question of clan scientist who did much of the early science that has found its niche. Enzymologists have divided the work in this field was David Davies at Yet it was during the years when proteases into broad ‘clans’, divided the US National Institutes of Health, this technique was so resolutely out by their mechanism of action and who remembers: ‘some time in the of fashion that it scored perhaps named according to the residues or late 70s I took wire models of the its most stunning success to date: chemical groups involved in that pepsin structure – all the models we rationally designed HIV protease mechanism. The HIV protease had then – to a meeting where I was inhibitors. The first such drug, sequence seemed most closely able to discuss it with Jordan Tang Roche’s saquinavir, entered the clinic related to proteins in the aspartic (who first sequenced the protein), in 1995 and, almost immediately, protease clan, a group that includes Blundell, and others. We came up life expectancy for people with the well-studied mammalian with the idea that there might have HIV and Aids began to improve. pepsin (see Gartner’s hype cycle been an ancestral Only two years after saquinavir was box p52). However, there were describes the evolution with a single DTG triplet that licensed, Roy Gulick, a prominent also significant differences. The of new technologies, was active as a dimer, and that the physician at Cornell University, New sequences of pepsin and related including drug design mammalian proteases might have York, US, involved in clinical trials of protease inhibitors, predicted that ‘we may finally have the tools to turn HIV infection into a long- term, manageable and treatable disease, much like hypertension and diabetes’. Today, there are ten drugs in this class in clinical use – the latest, Tibotec’s darunavir, was licensed in 2006 – and they are key components of multi-drug HAART [highly active anti-retroviral therapy] treatments.

Rapid response The global HIV pandemic is still less than 30 years old. The first cases of the disease that became Aids were reported to the US Centers for Disease Control and Prevention in 1981; four years later, not only had the disease agent been elucidated, but the complete genome sequence of that agent, the retrovirus later named HIV, had been determined. Scientists such as Steve Oroszlan, a researcher in viral oncology at the US National Institute, rapidly set to work www.chemistryworld.org Chemistry World | November 2009 | 51 Drug design: HIV

evolved through gene duplication. would be similar enough to that of We didn’t realise then that enzymes pepsin to be inhibited by known with this structure could still be generic aspartic protease inhibitors, found in nature.’ and that some of these might be By the late 80s, there were three good enough inhibitors to be leads groups working on crystal structures for drug design.’ Blundell, who was of retroviral proteases: Blundell’s, at independently thinking along the Birkbeck College; Alex Wlodawer’s, same lines, had started talking to

DR TIM EVANS / SCIENCE PHOTO LIBRARY LIBRARY PHOTO SCIENCE / EVANS TIM DR at the US National Cancer Institute, collaborators at in Sandwich, Frederick, Maryland; and a group UK, about testing known aspartic led by Manuel Navia at Merck, Sharp protease inhibitors against and Dohme in New Jersey, US. ‘It HIV. ‘Pfizer in the US had was very difficult to get enough pure a series of compounds HIV protease to make crystals,’ that inhibited aspartic remembers Wlodawer. ‘Our proteases that are known first structural studies were targets for drugs against on the related Rous sarcoma hypertension,’ he says. ‘We virus [RSV] protease, as that were intrigued by the idea that protein was more readily an antihypertensive might available. Our first crystal be a lead compound for drugs structure of the HIV enzyme against Aids.’ used protein that Steve Kent, from California Institute of Structure racing Technology, had produced by By 1989, the three groups were chemical synthesis.’ working against the clock, Pearl was a postdoctoral and against each other, to solve researcher at the Institute and publish their structures. The of Cancer Research (ICR) at Merck group was the first to submit Sutton in Surrey, UK, when the a manuscript on the HIV protease HIV sequence was published. structure; they sent it to Nature, ‘After obtaining my PhD with who sent it to Blundell to review. Tom Blundell, I found myself in the He realised that although the bulk fortunate position of knowing a of the structure was correct, the great deal about aspartic protease C-terminal part was inconsistent structure and mechanism, and with the models that he had been being in a lab doing state-of-the- building, and that his models seemed art work in retrovirology – a group more plausible. ‘However, knowing at the ICR had just discovered that well how medically important this CD4 was the receptor for HIV,’ he structure was, and believing it to remembers. He studied alignments be largely right, I recommended of the HIV sequence with other that Blundell, Tang and Davies had The mammalian digestive it for publication.’ He and Pearl retroviral proteases, and noticed that speculated about, and he worked pepsin is a close laid out the experimental and ‘everything needed for the aspartic with Willie Taylor at Birkbeck to relative of HIV protease predicted structures side by side in protease mechanism was completely build a structural model of a cut- an accompanying News and Views conserved in these sequences, but down half protease. ‘Our model could article. The paper came out a week it was only half there’. He started form a dimer with perfect two-fold after the Wlodawer group had to wonder whether these proteins symmetry,’ he says. ‘Furthermore, published the structure of the Rous could be the ‘ancestral half-enzyme’ we concluded that its mechanism sarcoma virus enzyme, and in the same week Irene Weber, working Clan connections with Wlodawer, published a model of the HIV enzyme based on the RSV The structural biology of the aspartic proteases protease structure. When Wlodawer, has a very long history and pepsin in particular and three months later, but to higher has a unique place in this. The first ever protein resolution, Blundell, published their x-ray diffraction pattern was obtained by John x-ray structures, they proved that Desmond Bernal and his student Dorothy the Merck group had indeed traced Crowfoot [later Hodgkin] at the University of the C-terminal end of the chain Cambridge from crystalline pepsin. This was incorrectly. published in Nature in 1934, and Bernal and The enzyme has an extremely Crowfoot wrote ‘…now that a crystalline protein elegant, economical structure, has been made to give x-ray photographs it is made up largely of b-strands; it can clear that we have the means of…arriving at far preserve perfect two-fold symmetry more detailed conclusions about protein structure if no substrate or inhibitor is bound. than previous physical or chemical methods have Yet the main interest in the structure been able to give,’ a prediction that has proved to had to be its potential role in the be, if anything, an understatement. By the mid- development of drugs against Aids. 1980s, atomic models of pepsin and several other Polish crystallographer Mariusz aspartic proteases were known. was a pioneer of structural biology Jaskolski, a long-distance member of Wlodawer’s protease group and

52 | Chemistry World | November 2009 www.chemistryworld.org first author on one of the structure based methods, particularly x-ray papers, describes HIV protease as , to identify fragment ‘an outstandingly beautiful molecule hits; and joining or growing the from an outstandingly cruel virus’. hits to form larger, more potent Fortunately, the characterisation of inhibitors. a novel member of the already well Chris Murray, Astex’s vice- studied aspartic protease family as president of discovery technology, a target gave the pharmaceutical explains how this approach has industry a head start. ‘Just about built on both the structure-based every pharma company had an HIV methods of the 80s and the high protease programme in the 1990s,’ throughput techniques of the 1990s. remembers Blundell. ‘Structure-based drug discovery is The first HIV protease inhibitors all about using a small number of were so-called mimetics, compounds in a clever way. High based on the structures of known throughput screening is based peptide substrates of the enzyme, on the premise that these clever with alterations mainly to prevent approaches are unnecessary if cleavage of the target peptide bond, enough diverse compounds can be thereby ensuring the enzyme tested. We have now come full circle; remained blocked. But make we are looking at compound quality poor drugs, and the lead compounds as well as numbers. Our drug design had to be modified to make them methods use libraries of fragments more ‘drug-like’: essentially, smaller his involvement with Astex back to Peptide mimetic that can be joined in many different and less hydrophilic. Nevertheless, the HIV protease story. ‘We soon saquinavir was the first ways, guided by structure and saquinavir was licensed for use began to realise that we had to licensed HIV protease modelling, and they need both.’ against Aids a mere six years after screen non-peptide compounds to inhibitor Astex currently has four kinase the publication of the HIV protease explore the whole possible range inhibitors and one inhibitor of structure – still a record for the of protease inhibitors,’ he says. ‘In the heat shock protein Hsp90 time ‘from bench to bedside’ for a fact, you need to explore as much in preclinical development novel drug. Pearl attributes part of of the available “chemical space” or clinical trials for cancer. this to the team effort of the groups as possible. And as estimates of the Interestingly, however, another of its involved, collaborators as well as number of potential “drug-like” programmes – in collaboration with competitors. ‘While the individual molecules range from 1020 to a a major pharmaceutical partner, contributions blur into history, this barely conceivable 10 200, even AstraZeneca – concerns a novel very tangible contribution to human a library of a million compounds aspartic protease: β-secretase, which welfare is something everyone will only scratch the surface.’ The cleaves amyloid precursor protein involved should be immensely proud technology behind Astex involves Tom Blundell produced and is one of the key enzyme targets of.’ screening with very small molecules one of the first crystal for Alzheimer’s disease. ‘This is or ‘fragments’ rather than potential structures of the HIV a membrane-bound, monomeric Reaching the plateau lead compounds; using structure- protease protease that resembles more By the time the first inhibitors than it does HIV protease, but has entered the clinic, the pendulum structural features that distinguish had swung away from structure- it from both. Tom [Blundell]’s based or ‘rational’ drug design to accumulated decades of experience the high throughput approach in with aspartic proteases and their which libraries containing thousands inhibitors have proved invaluable or even millions of unrelated in this project’, says Murray. As compounds are screened against a the population ages, Alzheimer’s potential drug target. It took another disease is becoming as high a ten years, and the onset of disillusion research priority as Aids was twenty as high throughput screening failed years ago. Will structure-based to increase the number or proportion methods play as large a part in the of candidate drugs entering the clinic, discovery of drugs that halt the for Wlodawer to see that his protease development, rather than just easing work had truly been ‘a platform the symptoms, of this devastating for the development of rational condition, as they have in the HIV design strategies.’ And several of the protease inhibitor story? original players in the HIV protease game have since made substantial Clare Sansom is a freelance science contributions to the development writer based in London and of structure-based approaches Cambridge, UK that fit the twenty-first century. In 1999, Blundell co-founded Astex Further reading Therapeutics to exploit structure-   A Wlodawer and J Vondrasek, Annu. Rev. based discovery methods; he is Biophys. Biomol. Struct., 1998, 27, 249 (DOI: 10.1146/annurev.biophys.27.1.249) still a board member and scientific   A Mastrolorenzo et al, Curr. Med. Chem., consultant for the company. 2007, 14, 2734 Blundell traces the thinking and   M Congreve, C W Murray and T L Blundell, method development that led to Drug Discovery Today, 2005, 10, 895 www.chemistryworld.org Chemistry World | November 2009 | 53