structure – Drug Discovery Opinion

Allosteric PDE4 Inhibitors with Reduced Potential for Side Effects

Crystal structure of PDE4D catalytic domain (tetramer) complexed with Rolipram - pdb ID = 1Q9M. Data from Emerald Biostructures are not yet released in the pdb.

The cyclic nucleotide phosphodiesterases (PDEs) are important regulators of signal transduction and selective inhibitors of the different subtypes have great clinical potential. PDE4 inhibitors are expected to be beneficial in the treatment of inflammatory and respiratory diseases such as asthma and COPD as well as CNS disorders including schizophrenia, depression, and Alzheimer’s disease but their potential has so far been limited by the incidence of side effects, particularly emesis. The emetic response is mediated in part by a brainstem noradrenergic pathway and, for non-CNS indications, can be reduced by limiting distribution of inhibitors to the brain. Active site directed PDE4 inhibitors completely inhibit enzyme activity at high concentrations but researchers at Emerald Biostructures (formerly deCODE biostructures) have now identified allosteric small molecule modulators of PDE4 with reduced potential for side effects. The four PDE4 variants (PDE4A, B, C, and D) all contain signature regulatory domains called upstream conserved regions 1 and 2 (UCR1 and UCR2). UCR2 is needed for high-affinity binding of the PDE4 inhibitor rolipram and X-ray crystallographic structures revealed that small molecule inhibitors bind to UCR2, thereby controlling access to the active site. The team used the structural data together with supporting mutational data to design PDE4 allosteric modulators that only partially inhibit cAMP hydrolysis. The modulators were shown to be potent in cellular assays as well as in vivo cognition tests and to have greatly reduced potential for emesis in several species. The authors hope that their work will lead to the identification of PDE4 modulators with reduced potential for emesis that can be used to treat disorders where brain distribution is needed. The study is published in the December 27^th advance online issue of Nature Biotechnology.

Now iSee

Ever had difficulty following a structural biology paper? Well, the Structural Genomics Consortium (SGC) and PLoSone have got together to provide a collection of articles available in enhanced versions utilising the iSee 3D visualisation platform. The collection, entitled ‘Structural Biology and Human Health: Medically Relevant Proteins from the SGC’, contains a series of articles documenting many of the novel protein structures determined by the SGC and work to further characterise their function. The iSee platform enables the reader to manipulate three-dimensional images within the article, as well as presenting pre-determined scenes from links in the text.

The iSee concept has been implemented using Molsoft’s ICM software, allowing structural data to be packaged and viewed in a browser (either online, through use of a free browser plugin, or offline with a standalone version). Plugins are available for all the major browsers and operating systems. Details of the approach and technical implementation are reviewed in ‘SGC – Structural Biology and Human Health: A New Approach to Publishing Structural Biology Results’ and ‘A New Method for Publishing Three-Dimensional Content’. If you want to try it out, take a look at ‘Crystal Structure of the ATPase Domain of the Human AAA+ Protein Paraplegin/SPG7’ and follow the link to the enhanced version.

The SGC is a not-for-profit organisation that aims to solve the three-dimensional structures of proteins of medical relevance and place them into the public domain without encumbrance or restriction. It is driving the concept of ‘open-source science’ to enable drug discovery by promoting pre-competitive structural biology and medicinal chemistry. In 2008 the SGC contributed 20.5% of novel structures released by the PDB.

Strengthening Bonds

The formation of strengthening crosslinks in the triple-helical structure of collagen is crucial to its structure and function, but the exact nature of the crosslinks in collagen IV – which is found predominantly in basement membranes and forms supramolecular networks that influence cell adhesion, migration and differentiation – have been particularly difficult to establish. The extracellular networks are formed by oligomerisation of triple-helical subunits by end-to-end associations and by intertwining of triple helices. At the C-terminus, two subunits are covalently linked via their trimeric non-collagenous domains to form a hexamer structure which confers additional strength to the matrix. The crosslinks were initially thought to be disulfide bonds but, using a combination of high resolution mass spectrometry and NMR spectroscopy, researchers at Vanderbilt University Medical Center have now shown that the linkages are actually sulfilimine (-S=N-) bonds. Lys²¹¹ was found to be modified to hydroxylysine (Hyl²¹¹) and a double bond shown to connect the sulfur atom of Met⁹³ to the nitrogen atom of Hyl²¹¹. This is the first time that a sulfilimine bond has been reported in a native biomolecule. The authors speculate that the sulfilimine crosslink may have arisen relatively early in evolution as an adaptation to the mechanical stresses on more complex organisms and they have now begun a search for the enzyme that carries out this transformation. Abnormalities in collagen IV have been linked to perinatal cerebral haemorrhage and the rare neurological disease, porencephaly, as well as to the autoimmune conditions, Alport syndrome and Goodpasture’s syndrome.

The study is published in the journal Science.

Targeting the Riboswitch

Recent research has shown that gene expression can be regulated at the level of mRNA by riboswitches. A riboswitch is an aptamer region on an mRNA molecule that can specifically bind a small effector molecule, causing changes in the structure of the expression platform and so regulating the activity of the mRNA. Riboswitches most usually switch off the ability of mRNA to carry out protein synthesis but can also switch it on. A variety of riboswitch classes have been identified, with most of the examples being discovered in bacteria including E. coli and streptococcus as well as bacteria causing anthrax, gonorrhoea, meningitis and dysentry.

Researchers at the University of Rochester Medical Center have now solved the crystal structure of the smallest known riboswitch, the preQ₁ riboswitch. The preQ₁ riboswitch controls the ability of bacteria to produce queuosine (Q), a molecule which enables accurate gene expression by overcoming an inbuilt defect in the mRNA-ribosome-tRNA system known as tRNA wobble, and which is essential for the survival of many important pathological bacteria. The preQ₁ riboswitch ‘senses’ the level of preQ₁, a precursor to Q. If too much preQ₁ is present, genes responsible for producing preQ₁, or for its transport, are shut down. The preQ₁ precursor, known as preQ₀, has the same effect in reducing production of Q. One gene that is regulated by the preQ₁ riboswitch is that which codes for the enzyme queF, which converts preQ₀ into preQ₁.

The structure of the preQ₁ riboswitch from Thermoanaerobacter tengcongensis complexed with preQ₀ shows preQ₀ bound in a buried pocket. The structure also reveals how the first base of the mRNA ribosome binding site binds to a loop of the riboswitch, and how the loop end of the preQ₁ riboswitch aptamer domain binds to preQ₀. Binding of the preQ₁ aptamer loop to the first base in the ribosome binding site was found to be mediated by a standard G to C base pairing interaction. The preQ₁ aptamer (34 nucleotides) is about 2.5-fold shorter than functionally related riboswitches that recognize similar metabolites.

An understanding of how bacterial species sequester their ribosome binding sites using divergent preQ₁ riboswitch aptamers could lead to the design of a new class of antibiotics. There is evidence that some existing antibiotics act – in part at least – by targeting riboswitches and, since riboswitches have not yet been found in human cells, the hope is that antibiotics acting on riboswitches will have a low propensity for side effects. The study is published in the Journal of Biological Chemistry.

P-Glycoprotein – A Close-up View

P-glycoprotein (Pgp), with its ability to transport a wide range of xenobiotic compounds – including drug molecules – across cell membranes, is the bane of medicinal chemists. Pgp, which probably evolved as a defense mechanism against toxic substances, is an ATP-dependent integral membrane protein that is particularly highly expressed in cells of the gut and kidney, and also in capillary endothelial cells that make up the blood-brain barrier. This means that as well as preventing absorption of drugs from the gut after oral dosing, Pgp can limit entry of drugs into the brain. Expression of Pgp by tumour cells also results in decreased accumulation of anti-cancer drugs in the cells, and contributes to multi-drug resistance in chemotherapy.

A team led by scientists at the Scripps Research Institute have now succeeded in solving the X-ray crystallographic structure of murine Pgp, a result that they hope will help chemists to design more effective drugs. The 3.8Å structure of the apo protein revealed an internal cavity of ca 6000Å³, with a 30Å separation of the two nucleotide-binding domains. Two additional structures with bound inhibitors showed that hydrophobic and aromatic amino acids form distinct drug-binding sites which are capable of stereo-recognition. The apo and drug-bound Pgp structures are open to the cytoplasm and the inner leaflet of the lipid bilayer for drug entry, representing initial stages of the transport cycle. The overall structure of Pgp is very similar to that of the bacterial protein, MsbA, which transports lipids out of bacteria, suggesting that Pgp may work in a similar way. In the bacterial transporter, binding of ATP changes the accessibility of the carrier from cytoplasmic (inward) facing to extracellular (outward) facing so that substances caught inside the cavity are ejected from the cell. The cavities of both transporters are lined with hydrophobic amino acids, but Pgp contains a larger number of highly varied amino acids which perhaps explains its broader substrate specificity.

The study, which is published in full in the journal Science, should help chemists to better understand, if not tame, the beast.

Mycobacterium tuberculosis NAD+ Synthetase Characterised

crystal structure of mycobacterium tuberculosis NAD synthetase The World Health Organisation estimates that one third of the world’s population is latently infected with Mycobacterium tuberculosis (MTB), and that ten per cent of infected individuals will develop active disease. Current treatments for tuberculosis are effective only during active infection, and the emergence of drug-resistant strains of MTB is compromising the efficacy of existing drugs. Nicotinamide adenine dinucleotide (NAD⁺) synthetase is an attractive target for control of MTB since the enzyme is essential for survival of both active and latent mycobacteria, but drug discovery efforts against this enzyme have so far been hampered by a lack of structural information.

University of Maryland scientists have now characterised the structure and mechanism of the MTB synthetase and hope that their work will facilitate the discovery of new drugs that will be able to combat both latent and active MTB infections. All living cells need the coenzyme, NAD⁺, which regulates many physiological processes including redox reactions. A number of biochemical pathways are able to synthesise NAD⁺ but, since MTB has only two pathways – both involving NAD⁺ synthetase – and humans have pathways that avoid NAD⁺ synthetase, an inhibitor of this enzyme should be effective against MTB but have minimal side effects.

The MTB NAD⁺ synthetase is a multifunctional enzyme which catalyzes the ATP-dependent formation of NAD⁺ at the synthetase domain using ammonia derived from L-glutamine in the glutaminase domain. The study, which is published in full in Nature Structural & Molecular Biology, revealed a homooctameric subunit organization, suggesting a tight dependence of catalysis on the quaternary structure.

Just Another Kinase Structure?

The Janus family of protein tyrosine kinases is comprised of four members: JAK-1, JAK-2, JAK-3 and Tyk-2. These kinases provide membrane proximal signalling through association with type 1 and type 2 cytokine receptors, phosphorylating and activating Signal Transducers and Activators of Transcription proteins (STATs) in response to cytokine binding. Expression of JAK-3 is predominantly restricted to haematopoietic cells, whilst the other family members are ubiquitously expressed. Inhibitors of these kinases have received interest since aberrant JAK activity has been associated with a variety of hematopoietic malignancies, cardiovascular diseases and immune-related disorders. A number of clinical studies are in progress to evaluate JAK inhibitors in transplantation, myelofibrosis, polycythaemia vera and essential thrombocythaemia.

Crystal structures of the kinase domains of JAK-2 and JAK-3 have previously been solved, enabling structure-based design of inhibitors. Now collaborators at Cytopia Research and Monash University have published the high-resolution crystal structure of the JAK-1 kinase domain. Whilst the ATP-binding sites of protein kinases are highly conserved, particularly amongst family members, the researchers have identified subtle differences surrounding the JAK1 and JAK2 ATP-binding sites. There is no doubt that developing JAK-1 or JAK-2 selective inhibitors (at the ATP-site) will be challenging, but the new data at least suggest that it could be possible. Whether it is necessary or desirable to achieve such selectivity is, as yet, unclear.

Full details are published in the Journal of Molecular Biology.

Influenza ‘Cap Snatcher’ Identified

Seasonal epidemics of the influenza virus continue to kill hundreds of thousands of people annually, and the increasing incidence of resistance to approved drugs means that there is a pressing need for new therapies. The viral polymerase is an attractive target for drug development and a newly reported high-resolution crystal structure of the polymerase PA domain should provide useful insights for inhibitor design.

In eukaryotic cells, mRNA must be capped at the 5’-end for efficient translation. Cap structures, consisting of N⁷-methylated guanine units, also assist in transport of mRNA from nucleus to cytoplasm and protect mRNA from degradation by 5′ exonucleases. The viral polymerase ‘steals’ caps from cellular mRNA in a process known as ‘cap-snatching’ and attaches them to viral mRNAs so that these can be translated into new viral proteins.

The polymerase is a heterotrimer comprising three subunits, PA, PB1 and PB2, and whilst previous studies had shown that PB2 plays a role in cap-binding, PB1 was believed to be responsible for ‘cap-snatching’. The new study, published online on February 4^th in the journal Nature, clearly shows, however, that the PA subunit contains the endonuclease active site and plays a crucial role in cleaving the cap from host mRNA. The active site, which is conserved in all influenza viruses, contains a histidine residue together with a cluster of three acidic residues that bind two manganese ions in a configuration similar to that observed in other two-metal-dependent endonucleases. Inhibition of cap-cleavage by the endonuclease would be an effective anti-viral strategy since it would effectively block synthesis of viral proteins.

X-ray Structure of Malaria Peptidase Solved

blood smear of plasmodium falciparum The malaria parasite, Plasmodium falciparum, has limited capacity for de novo amino acid synthesis and relies on degradation of host haemoglobin for a supply of these essential building blocks. Haemoglobin is first degraded into di- and tri-peptides by the action of a number of cysteine-, aspartyl-, and metallo-proteases. These small peptide fragments are then further hydrolysed to release free amino acids by the action of the metallo-exopeptidases, PfA-M1 (an alanyl aminopeptidase) and PfA-M17 (a leucine aminopeptidase).

A team from the University of Monash has recently described the X-ray crystallographic structure of truncated recombinant PfA-M1 at a resolution of 2.1 Å. Comparison of structures of PfA-M1 bound to the known inhibitors, bestatin (K_i 500nM) and hPheP[CH₂]Phe (K_i 80nM), with the native structure showed that the enzyme did not undergo any global conformational rearrangements on binding either inhibitor. It is proposed that substrate access is achieved by means of the C-terminal domain vortex, and that control of substrate hydrolysis can be achieved, and depends on, the size of this channel. hPheP[CH₂]Phe, which provides effective protection in a murine model of malaria, also inhibits PfA-M17 and the authors suggest that inhibiting both PfA-M1 and PfA-M17 may be less likely to allow the development of drug-resistant malaria. The fact that the site of action of PfA-M1 is outside the digestive vacuole, together with the comparative ease of identifying drug-like inhibitors of metallo-proteases, makes PfA-M1 an attractive target for new anti-malarial therapies.

The study is published in the February 5^th Early Edition of PNAS.
bestatin structure hPheP[CH2]Phe structure

Structure of Ebola Virus Protein Determined

Ebola virus, named after a river in the Democratic Republic of the Congo where it was first identified, causes a severe, often fatal haemorrhagic fever. There is no vaccine against or specific treatment for Ebola virus infection, but researchers at Iowa State University have now moved a step closer to finding a treatment. Using a combination of X-ray crystallography and nuclear magnetic resonance spectroscopy, the group has succeeded in determining the structure of a key portion of the viral protein, VP35, at a resolution of 1.4 Å. The Ebola VP35 protein forms part of the viral RNA polymerase complex, acts as a viral assembly factor, and also inhibits production of host interferon.

Binding of VP35 to dsRNA correlates with suppression of interferon activity and viral virulence, and the group hopes that the new structure of the interferon inhibitory domain will allow them to design drugs that bind to VP35 and block its function. The protein forms a unique fold with two clusters of basic residues, one of which is important for dsRNA binding and inhibition of interferon production. The study is published in the January 13^th issue of PNAS.