Continuing the open innovation theme from our last newsletter, the SGC have recently been talking about their latest initiatives to enable open access to key data which should boost the efficiency of drug discovery efforts in the UK and beyond. In a barnstorming talk at the RSC/SCI Med Chem Symposium in Cambridge this month, Professor Chas Bountra of the SGC presented a vision for a future open innovation model which could radically enhance our current approach to pharmaceutical research and development.
The SGC is a PPP (public -private-partnership) comprising several public, charitable and private funders, including 8 large pharmaceutical companies, and is based at the Universities of Oxford and Toronto. They work closely with a network of over 250 academic labs across the world, and ensure public access to all data and reagents generated.
Already the SGC has an impressive track record. Since they started depositing crystal structures in the PDB back in 2005, they have now solved more kinase structures than academia and overtook pharma on this metric back in 2008. Indeed, they have solved a significant portion of all novel human protein structures currently available, at one point generating up to 25% of the yearly global output. They have pioneered structural biology in several new target classes such bromodomains. As well as making all crystallographic data publicly available, the SGC generate proteins, assays, antibodies and high-potency/high-specificity, cell-permeable inhibitors to aid target validation, all placed in the public domain without any patent issues. The SGC always ensure that everything generated is available to all, without restrictions on use and access. As a result, due to unencumbered access, the SGC outputs have advanced science in many areas, led to new proprietary programmes in pharma and the establishment of new biotechs.
Now, however, they are hoping to go a stage further. Given that Phase II is still very much the ‘graveyard’ for many potential drugs, with over 90% of candidates against new targets failing at Phase IIa or IIb, Prof Bountra and colleagues are proposing a new PPP to help validate pioneer targets in patients. The new consortium would bring together academic and industrial scientists, with active participation from patient groups and regulators, and would cover the full range of activities from lead identification through to Phase IIa. Throughout the process, the groups involved would make reagents publicly available, and would be encouraged to publish data. After Phase IIa readout, data on targets found to be mechanistically invalidated would be published quickly, whilst more promising programmes would be advanced to develop novel molecules (using a proprietary development model, where appropriate). The project has already been initiated in cancer, and plans are underway to begin studies in neuro-psychiatry.
As the face of the drug discovery landscape has changed so profoundly in the last decade, so the open innovation concept continues to evolve to embrace and enhance efforts in this area. This is a bold, dynamic concept from the SGC, which, if successful, could radically benefit UK drug discovery research.
* I would like to acknowledge the considerable assistance of Dr Lee Wenhwa of the SGC in the preparation of this article
In the beginning – a very long time ago it seems – there was bacteriorhodpsin. The only available GPCR crystal structure, upon which all early homology models were based. Then, in 2000, along came the structure of bovine rhodopsin – the first mammalian GPCR to be crystallized. This helped modeling a little, but let’s face it, there were still an awful lot of issues with structural modeling of GPCRs. Then, in 2007 new technologies were utilized to stabilize GPCR structures, including binding of a monoclonal antibody to the third intracellular loop, or insertion of T4 lysozyme to constrain this flexible loop, which opened the gates to a flurry of family A GPCR structures. A third route to GPCR stabilization was also being pioneered around this time by Heptares, whereby specific point mutations were introduced to the GPCR protein construct to thermally stabilize the protein. In recent years, with these various technologies in hand, both agonist and antagonist-bound structures of family A GPCRs have become available, and the true complexity of the conformational states adopted by agonist, antagonist, inverse agonist and partial agonist ligands started to unfold.
Now it would seem that the next major chapter in GPCR structural elucidation is upon us, heralded by the publication of the first Family B GPCR structures in July. Structures of the glucagon receptor (PDB id 4L6R.pdb, crystallized in the presence of an antagonist, but with no ligand visible in the electron density) from a multi-national academic/industrial partnership including the Scripps Institute and Novo Nordisk(1), and the human corticotrophin-releasing factor 1 receptor structure (PDB id 4K5Y.pdb) from Heptares(2), crystallised with antagonist CP-376395, has uncovered yet more structural surprises, revealing a binding site which is substantially displaced compared with the known Family A structures. The Family B transmembrane domain appears to be conformationally distinct from the corresponding Family A domains elucidated to date, adopting a more V-shaped shape in the Family B structures.
This is a significant step forward in our understanding of GPCR structure and function, and should provide a model for other family B structures.
Image taken from (2), showing the site of the corticotrophin-releasing factor 1 receptor structure antagonist CP-376395, compared with the antagonist binding site observed for Family A GPCR structures to date.
1. Siu, F. Y. et al. Structure of the human glucagon class B G-protein-coupled receptor. Nature 499, 444–449 (2013)
2. Hollenstein, K. et al. Structure of class B GPCR corticotropinreleasing factor receptor 1. Nature 499, 438–443 (2013)
NMR 3D structural information in drug design
An account 1 of a new NMR method to determine high definition unbound conformations of ligands and their dynamic motion, discusses application of the approach to streptomycin. Two major conformational families are observed the most populated of which corresponds to the crystallographic conformation in complex with the 30S ribosomal subunit. The method can be applied in physiologically relevant solvents and is independent of molecular modelling using multiple datasets, much greater quantities of data than previous NMR approaches and a dynamic model during refinement. The authors argue that, in the absence of a target crystal structure the unbound conformation – particularly of high affinity ligands, can be used to deduce the target site binding pocket shape and, at least to some extent, electrostatics of interaction. Clearly the data can be used to help improve predictions of conformational constraint as well – which does call on the molecular modelling for support. The C4X team have also just reported the solution structure of an antagonist for a class B GPCR, CRF 2.
1. C. D. Blundell et al Bioorg. Med. Chem. Lett., 2013, 21, 4976.
2. C4X Discovery press release 20th Aug 2013
Attrition in Phase II and III
A new analysis of failure rates in PhII and PhIII for 2011-2012 1 compares with similar analyses for PhII (2008 – 2010) 2 and PhIII (2007-2010) 3. Between 2011 and 2012 there have been 148 failures in PhII and PhIII/submission for which 105 had reported reason for failure. In PhII failure is dominated by lack of efficacy (59%) and safety(22% a figure that includes those compounds with an inadequate TI) slightly higher than 2008 – 2010. However strategic considerations showed a reduction to 16% (still high in my view) compared with a staggering 29% in 2008 – 2010 (effect of mergers and “right-sizing”?). In contrast for PhIII failures due to efficacy has declined from two thirds to about half although failure due to safety issues has increased. In a trend analysis of companies accounting for two thirds of global R&D expenditure it appears that PhII successes is still running at below 20% although there is apparently a modest 7% increase in PhIII success rate. Still if this is a sign that better decisions are being around the output from (relatively) cheap PhII trials prior to entering PhIII that has got to be considered progress.
1. J. Arrowsmith and P. Miller Nature Reviews Drug Discovery 2013, 12, 569 doi:10.1038/nrd4090
2. J. Arrowsmith Nature Reviews Drug Discovery 2011, 10, 328
3. J. Arrowsmith Nature Reviews Drug Discovery 2011, 10, 87 doi:10.1038/nrd3375
FDA approvals for first half of 2013
Analysis of FDA drug approvals over the first six months of the year has just appeared 1 with 13 novel compounds approved. So is this time to start worrying over low numbers of approvals again – remember 39 approvals for 2012. Answer is no – at least not yet – at the half way point last year only 14 drugs had been approved, apparently the FDA tends to approve more drugs in the second half of the year than the first. Five of the approved compounds are predicted to achieve (multi)billion dollar sales by 2018. From a scientific perspective its good to see ado-trastuzumab emtansine a second antibody drug conjugate approved; antisense representation with mipomersen; the first sodium-dependent glucose co-transporter 2 (SGLT2) inhibitor to be approved in the United States canagliflozin and the first MAPK/ERK kinase inhibitor trametinib. Oncology dominates the target indications as might be expected with COPD, diabetes, hypercholesterolaemia and imaging/contrast agents also represented. The full list of approvals is here. At some point there should be a surge of products coming through from the FDA breakthrough categorisation although that (presumed) surge will mean a dipin approvals in subsequent years.
1. Nature Reviews Drug Discovery 2013, 12, 568 doi:10.1038/nrd4097
Somatic mutations of RAS are present in one-third of all human cancers which can lead to aberrant activation of downstream signaling pathways involving RAF/MEK/ERK kinases. There has been little success in trying to identify direct RAS binding inhibitors to block downstream aberrant signalling effects of mutants 1 nor have efforts so far proved successful in blocking obligate prenylation via farnesyl transferase inhibition due to shunt mechanisms kicking in. Now reported 2, however, is a strategy to block the farnesylated RAS from reaching cell membranes by competing for the PDEδ carrier protein used by the RAS to reach the membrane. The team identified a benzimidazole (1) that bound to the PDEδ that prenylated RAS binds to. Interestingly crystallographic studies revealed that two molecules of the benzimidazole bound to the PDE pocket Fig. 1 which allowed the development of more potent compounds by a simple dimerisation although the dimerisation doesn’t look like its optimal based on the drop in ligand efficiency of the dimeric molecule Deltarasin (2). Certainly it looks like there will be plenty of room for further optimisation. Despite these comments (2) does show activity in cell based assays consistent with its proposed mechanism of action and reduction in the rate of growth of tumours in a Panc-Tu mouse xenograft model 10mg/kg bid.
A second interesting approach recently out 3 is use of Cotylenin A (3), a plant growth regulator, to stabilise RAF/14-3-3 interactions by binding to inhibitory 14-3-3 interaction sites but not activating sites. Cotylenin-A on its own is inactive in RAS mutant models but combination with an anti-EGFR antibody shows synergistic effects in vitro and in vivo. Cotylenin-A in conjunction with rapamycin has previously been reported 4 to cooperatively inhibit tumour cells through induction of G2.
Finally another new development 5 is the identification via virtual screening against a pocket in RAS identified by the authors that inhibits effector binding. Compounds e.g. (4) “inhibit both anchorage-dependent and -independent growth and induce apoptosis of H-rasG12V–transformed NIH 3T3 cells, which is accompanied by down-regulation of downstream molecules such as MEK/ERK, Akt, and RalA” Furthermore compounds were active via oral administrattion on a xenograft of human colon carcinoma SW480 cells carrying the K-rasG12V gene. I have to confess I am not totally enamoured of the structures but it establishes an interesting precedent which could lead to new directions perhaps 1 will need rewriting in due course!
1. W. Wang et al Bioorg. Med. Chem. Lett. 2012, 22, 5766
2. G. Zimmermann et al Nature 2013, 497, 638
3. S. Kasper et al, ACS Chem. Biol. 2013, Article ASAP DOI: 10.1021/cb4003464
4. T. Kasukabe et al, Cancer Sci. 2008, 99, 1693. doi: 10.1111/j.1349-7006.2008.00867.x
5. F. Shima et al, Proc. Natl Acad. Sci. 2013 29 Apr doi:10.1073/pnas.1217730110
Informatics Manager, RedX Pharma, Wakefield, UK
Application Scientist, Chemical Computing Group, Cambridge, UK or Cologne, Germany
Chemometrician, GSK, Ware, Herts, UK