Multidrug Co-Crystals Leading to Improved and Effective Therapeutics in Drug Development

In the last ten years, research has been focused on alternative therapeutic strategies for drug development and one of them is co-crystals. A review of the development, production and future of co-crystals has recenlty been published by Thipparaboina R. and co-workers .

The definition of co-crystals given by the FDA is “solids that are crystalline materials composed of two or more molecules in the same crystal lattice”, usually the interactions between molecules are weak, having non-colavent interactions. The discovery of the first co-crystal structure was quinhydrone complex synthesise by Friedrich Wohler in 1844, he found that this co-crystal was composed of a 1:1 ratio of quinone and hydroquinone. In drug development a co-crystal or multidrug co-crystal (MDC) is an active pharmaceutical ingredient (API) with a neutral compound in the same crystal lattice with non-ionic interactions between the two.

The interest in co-crystals for the pharmaceutical industry has increased in the last ten years, thanks to the development of multidrug co-crystals, for example the recent success of the phase II clinical trial of Celecoxib and Tramadol drugs by ESTEVE and Muldipharma Laboratories GmbH for the treatment of acute pain. There is a large list of existing multidrug co-crystals, their interactions improve solubility and bioavailability in therapeutic treatments. Some examples are, Ethenzamide and Gentisic acid which increase the solubility and dissolution rate (Srinivasulu et al., 2009); Meloxicam and Aspirin that significantly increase bioavailability (Cheney et al., 2011).  Co-crystals give the opportunity to treat a specific group of patients with one drug. The multidrug co-crystal Sildenafil and Aspirin presented dual therapeutic effects, treating erectile dysfunctions in cardiovascular complication patients (Zegarac et al., 2014). The following table lists the multidrug co-crystals developed so far.



The production of multidrug co-crystals is similar to normal crystallography procedures, however, the success depends on several factors and a deep understanding of both components to co-crystalize. Important factors to consider in the production of MDC are temperature, presence of impurities, rates of evaporation, differential solubility, solvent properties, supersaturation, cooling, etc. Scaling up the production of MDC is feasible and successful, several techniques such as spherical co-crystallization, spray-drying technologies, solvent crystallization, sonic crystallization and others have been used for the production of co-crystals (Fig1).



In 2013 the Food and Drug Administration (FDA) released a regulatory guideline for the classification of pharmaceutical co-crystals and soon after, in 2014, the European Medicines Agency (EMA) released a paper on the use of co-crystals in pharmaceutical research, however there are no many regulatory guides for the pharmaceutical market. At the moment there are few marketed co-crystal products, such as Entresto (Sacubitril-Valsartan), approved by the FDA in 2015, for the treatment of heart failure, and Lexapro (Escitalopram Oxalate), approved in 2009, for the treatment of major depressive and anxiety disorders. For the pharmaceutical industry it could be quite challenging to patent MDC and their method of production. In addition, every patent office requires different criteria for co-crystals registration. Every year the number of patents granted to multidrug co-crystals by the European Patent Office (EPO) and the United States Patent Office (USPTO) has increased. The patents available currently in MDC are listed in table 2.


It is a challenge for pharmaceutical research and industry to find appropriate drug co-crystals combinations for the therapeutic and pharmaceutical market. However, several factors need to be considered for the production of MDC, such as incompatibility between compounds, impurities, differential solubility and dose variability. The latter is an important factor to be considered for optimal success of MDC. For example, the common stoichiometry of co-crystals is a 1:1 ratio, however some dose range between drugs are variable and the slight increase of dose of one can modify or cause undesirable side effects to the patient. The development of validated predicted models for MDC is required to avoid pitfalls.

On the other hand, novel treatments for psychiatric disorders and neuropathic pain can be approached with the used of lithium and magnesium salts to form multidrug co-crystals. Also nanotechnology can be used for the production of nano-crystals for drug delivery or clinical applications.

The pharmaceutical industry needs be aware of the different considerations in terms of multidrug co-crystals such as, predicted models, FDA regulations, patents, safety and bioavailability. Although challenging multidrug co-crystals could be a novel approach for developing an effective therapeutic, however the commercial success hasn’t emerged rapidly. There needs to be further investment in research for the development of multidrug co-crystals.

Blog written by Thalia Carreno Velazquez




Cheney, M.L. et al. (2011) Coformer selection in pharmaceutical cocrystal development: a case study of a meloxicam aspirin cocrystal that exhibits enhanced solubility and pharmacokinetics. J. Pharm. Sci. 100, 2172–2181

Srinivasulu, A. et al. (2009) Trimorphs of a pharmaceutical cocrystal involving two active pharmaceutical ingredients: potential relevance to combination drugs. CrystEngComm 11, 1823–1827.

Thipparaboina, R. et al. (2016) Multidrug co-crystals: towards the development of effective therapeutic hybrids. Drug Discov Today 21, 481-490.

Zegarac, M. et al. (2014) A sildenafil cocrystal based on acetylsalicylic acid exhibits an enhanced intrinsic dissolution rate. CrystEngComm 16, 32–35

Chemical and Biological Therapeutic Approaches to Neurological Disorders Symposium

On Monday 18th April, the 3rd symposium on Chemical and Biological Therapeutic Approaches to Neurological Disorders took place at Burlington House in London. Dr Paul Beswick and I (Tristan Reuillon) represented the Sussex Drug Discovery Centre (SDDC) at this one day conference, organised by the Royal Society of Chemistry. Paul gave a talk on the use of structural biology for the design of ligands for glutamate ionotropic receptors, an approach which has been and is being used on different projects at the SDDC, while I presented a poster on recent developments in the field of AMPA receptor positive allosteric modulators.

Some of the leading researchers in the field of neuroscience were presenting, such as Dr Eric Karran, former head of research at Alzheimer Research UK now at AbbVie or Prof John Hardy from UCL, recent winner of the Breakthrough Prize in Life Sciences for his pioneering research into the genetic causes of Alzheimer’s disease (AD). Dementia was the major focus of the symposium, with Dr Eric Karran introducing the statistics on AD and giving a detailed overview of the different theories believed to underlie AD (amyloid-beta (Aβ) and tau pathologies). According to Dr Karran the readouts of some critical clinical trials on AD drugs within the next two years will be extremely important to understand if the drug discovery efforts have been heading in the right direction and to guide further the current research on dementia. Prof John Hardy presented the genetic causes behind AD and amyloid deposition, with an emphasis on some specific proteins, such as TREM2, which represent very attractive drug discovery targets. Prof Nigel Hooper from the University of Manchester presented research focussed on Aβ, trying to identify what forms of Aβ oligomers and fibrils are neurotoxic and trying to link the alpha-secretase ADAM10 with Aβ production. Finally Dr Suchira Bose from Eli Lilly gave an in-depth analysis of the tau pathophysiology and different modulations of this physiological pathway which could lead to novel therapeutic approaches to AD.

Other neurological disorders were also discussed during the symposium. Dr Hasane Ratni from Roche presented the discovery of RG7800, a drug currently tested in phase II clinical trials for the treatment of Spinal Muscular Atrophy, a rare neurodegenerative disease affecting mainly children. RG7800 acts as a SMN2 splicing modifier. Dr Richard Mead from the University of Sheffield talked about his current research on Motor Neuron Disease, also termed Amyotrophic Lateral Sclerosis, with a focus on the NRF2-ARE pathway, an indicator and modulator of oxidative stress in neurodegeneration. Different attempts to identify activators of this pathway, such as apomorphine, were discussed. The presentation of Dr Paul Beswick on glutamate potentiators was centred on the identification of novel drugs to treat the cognitive dysfunction associated with Schizophrenia, a major symptom, for which there is a clear unmet medical need. Finally, Prof Kristian Stromgaard from the University of Copenhagen, presented a few drug discovery approaches that his group has undertaken to disrupt protein-protein interactions in the CNS, such as the PSD-95-NMDA interaction. Owning to the lack of success in identifying small molecule hits, his research has focussed on peptidomimetics, which are surprisingly brain penetrant and are currently in preclinical development.

I found this symposium extremely interesting, with some fantastic and innovative research being disclosed, and would highly recommend it for anyone interested in neuroscience research. I hope to have given you through this blog article a flavour of the different topics which were discussed on that day and maybe tempted you to attend the 4th symposium in this series which will take place next year.


Blog written by Tristan Reuillon

The pharmaceutical industry: ‘doing the right things and doing the things right’?

The pharmaceutical industry is constantly under pressure to satisfy demands of the healthcare sector and to ensure business survival in terms of return on R&D investment. The main objective is to deliver drug candidates of the highest possible quality to decrease the risk of failure in clinical phases during the development process. The authors of this review have analysed the current key strategies emerging from the different R&D approaches to achieve this objective.

A notable change that has occurred in the pharmaceutical industry over the last few decades is the move from the drug discovery process that was exclusively conducted within pharmaceutical organizations to the current open innovation model, characterised by strong collaborations among pharmaceutical industries, academia and other industries with the development of a new and sustainable funding model with public and academic participations.

The focus of current R&D business relies on a costly, risky and time-consuming strategy to find first-in-class medicines, based on the discovery of new targets, with the aim to treat diseases with as-yet no treatment or to deliver more efficacious drugs in the pipeline than those currently on offer. This addiction to blockbuster drugs can be partly explained by the pharmaceutical industry needs to produce sustainable revenues and returns on investment for shareholders.

However, projects based on new targets have less probability to reach the market than those based on known targets, and the later entrants – where research is based on known targets – displace the first-in-class products in the market place, even if the first entrant has the exclusivity free of competition during a period of time. The focus on first-in-class products based on new targets alone is not enough to solve the current productivity gap. These days, pharmaceutical companies producing such first-in-class drug candidates often develop a ‘follow-on’ drug based on the more validated (with the first drug) target to reduce the risk of obtaining no drug approval. Getting the right balance between first-in-class products based on new targets and second-in-class products, as well between target-based and phenotypic approaches, constitutes a lifeline offering the chance to resolve the drug pipeline attrition.

Failures in clinical trials have soared over the past 20 years, with attrition rates between 1990 and 2010 increasing for Phase I from 33% to 46%, for Phase II from 43% to 66% and for Phase III from 20% to 30%. The main reasons for failure (Fig. 1) at Phase II are insufficient efficacy (51%), safety concerns (19%) and strategic issues (29%); and reasons for Phase III failures are predominantly insufficient efficacy (66%) and safety concerns (21%). The failures during early stages are less costly than those at Phase III; therefore implementing strategies to identify them as early on in the process as possible is absolutely crucial.

irina 1

Figure 1. Main reasons for clinical failures by Phase based on 410 drugs that entered human testing between 2000 and 2009. The main failures in Phase II and III studies are efficacy issues, 54% and 2%, respectively. Safety issues represent about one-third of all the 410 drugs analysed in Phase I and Phase III studies, versus 17% of all Phase II studies.

A new model is emerging in pharmacological research called polypharmacology, which describes the activity of compounds at multiple targets. The aim of multi-targeted approach is to avoid adverse side effects (safety parameters), and to improve therapeutic efficacy, prevent drug resistance or reduce therapeutic-target-related adverse side-effects (efficacy parameters). Generally, multitarget drugs – in combination or not – are more efficacious than single-target drugs, for instance in oncology and against viral infection. Furthermore, rather than a one-target therapy, polypharmacological modulation of a network of targets is actually required in the treatment of many multigenic diseases, as in the case of multikinase inhibitors that can block multiple targets in parallel signaling pathways and thereby prevent drug resistance caused by mutations or expression changes.

The failures due to pharmacokinetics/bioavailability issues currently account only to 1% for Phase II, reflecting the quality of the research process. The key points to consider in the drug discovery process are the right pharmacologic target and the right chemical lead. Reduction of timelines and cost (R&D) are strongly related to the high quality of science. The quality of the bioactive molecule can be evaluated using several criteria such as e-ADME profile including PK/PD behaviours, metrics, drug-like concept. The improvement of the quality of the target and bioactive molecule decrease the probability of failure in clinical trials.

The authors proposed (Fig. 2) a simple overview for new hit, lead and drug optimization process using the space concept strategy and several metrics to qualify these different chemical entities, based on the Lipinski and Hopkins concept of navigation and exploration of the chemical space.

The chemical space can be subdivided into four clusters associated with several specific chemical and physicochemical properties or topological descriptors as recognition patterns such as MW, clogP, number of hydrogen acceptors (NHA) and number of hydrogen donors (NHD) to define these discrete areas. Thus, druglike chemical space (Ro5, oral route, grey cube), lead-like chemical space (Ro4, green cube), fragment chemical space (Ro3, bright blue cube) and one mauve cube dedicated to the Ro50 for transdermal drugs can be represented. Other spaces dedicated to other administration routes can also be used (mauve cube). Inside the Ro5 (clinical candidates), Ro50 (clinical candidates) and Ro4 (leads), the use of the ligand efficiency (LE) as a simple indices or metric to quantify the molecular quality of the different chemical entity types, for selection and optimization inside each cluster. Other indices can be also used such as LLE, BEI/SEI, SIHE or QED. As shown, the optimization navigation process between each cluster can be performed using LE, LLE or BEI/SEI. Based on these metrics, construction of more ‘druggable’ libraries (from fragments, leads or drugs) can be developed. The overlapping of the drug-likeness chemical space continuum (Ro5, grey cube) and the 3D ‘target classes’ (block-cylinder, also called target space) including individually, for instance, PPI space, kinase space, G-protein-coupled receptor (GPCR) space, etc., defined an overlap volume (green rays, truncated cylinder) for which all the compounds (virtual or real) within this space are druggable; the anti-overlap area corresponds to the poorly druggable compounds. The same parameters used to define the boundaries of druggable compounds (e.g. Ro5) can be used to define a specific target space including drugs.


irina 2

Figure 2. Simple overview for new hit, lead and drug optimization process using the space concept strategy (real and virtual) and metrics to qualify chemical entities.

And to conclude, the authors stress the valid point made earlier by Elebring that ‘too much process thinking in drug discovery, such as Ro5, Ro3, etc., can block enthusiasm, creativity, intuition, innovation and serendipity’. Future success in drug discovery must therefore depend on achieving the correct balance between ‘doing the right things and doing the things right’.


Blog written by Irina Chuckowree

TMEM16/ANO: a pore of 2 halves

TMEM16 proteins are found in all eukaryotes, with the family consisting of ten genes (TMEM16A-K, missing out I).  Following TMEM16A and B being discovered as the elusive calcium activated chloride channels (CaCC) in 2008 by 3 independent groups (1, 2, 3) it was fully expected that all the other members of the family would also be CaCC’s because of their high sequence similarity.  However, this was not the case and TMEM16C-K have been shown to be phospholipid scramblases (4).  So how do TMEM16 proteins on the one hand scramble plasma membrane phospholipids, and on the other hand operate as anion channels?

In an attempt to answer this question Whitlock and Hartzell in their review (5) put forward a novel hypothesis: –

  • TMEM16A and B evolved from an ancestral scramblase.
  • TMEM16A pore shares structural similarity to an ancestral TMEM16 lipid channel.
  • The TMEM16A Cl selective pore is formed not of pure protein, as ion channel dogma is conceived at the moment, but is composed partly of lipids.

The authors propose that, TMEM16A protein stabilizes a non-bilayer phase in the membrane so that the 2 leaflets are continuous where they interact with the protein.  The lipid head groups would then provide a hydrophilic environment forming half of the pore and ions could move across the membrane in the ‘aqueous channel’ formed between the protein and the lipid head groups.  In putting forward this idea they have extrapolated Pomorki and Menon’s proposed mechanism for scramblases (6), in which a hydrophobic furrow would allow the phospholipid head groups to translocate from one side of the membrane to another while the acyl chains remain in the hydrophobic phase of the membrane, and suggest TMEM16A has evolved to conduct Cl ions but not lipids using the same conduction pathway.

They draw their evidence from the recent X-ray crystal structure of fungal nhTMEM16 provided by Brunner et al. (4) and then using their homology models show TMEM16A has a hydrophilic furrow very similar to that of nhTMEM16 with the exception of a patch of hydrophobic amino acids at the cytoplasm end of the furrow.  They suggest this patch might explain why TMEM16A is not a scramblase but an ion channel, because it forms a barrier to the movement of the hydrophilic head groups of phospholipids entering the furrow.

This is strengthened with reference to mutagenesis experiments where a chimeric protein was made with 35 amino acids from the lipid scrambling domain of TMEM16F and substituted into TMEM16A which them became a scramblase (7).  Suggesting this hypothesis might well be plausible, rather than as the title of Whitlock and Hartzell’s review suggests a poor idea, however, it needs further investigation to prove it right or wrong.

However, functional details remain obscure for most TMEM16 paralogs, making it unclear if they produce ion channels or not.  If this novel hypothesis is proven, it will have a major impact on our perception of the dogma of proteinaceous ion channels and the discovery of further TMEM16A inhibitors. It might also explain why TMEM16A inhibitors are, as the authors suggest, somewhat weird and not the ‘classical’ channel blockers, as it is unclear whether the current TMEM16 inhibitors block by acting in the permeation pathway or allosterically.


Blog written by Roy Fox





Insight into the Pharmacokinetics of CNS Drugs: the Species-Independent Brain Tissue Binding Phenomenon

Neuroscience is one of the field of expertise of the Sussex Drug Discovery Centre, and as a medicinal chemist novice to this area of drug discovery, I decided to extend my knowledge of drug delivery to the brain and pharmacokinetics for central nervous system (CNS) pharmacological agents. A significant amount of extremely interesting literature covers neuropharmacokinetics, and I rapidly came across a number of essential principals such as the importance of high blood brain barrier (BBB) permeability or the complexity of predicting pharmacokinetics/pharmacodynamics (PK/PD) relationships for CNS drugs. Among these dozens of publications, one scientific article1 from a group of DMPK scientists at Pfizer particularly attracted my attention, since it was reporting that brain tissue binding is species-independent, something that I would not have expected to be true. Indeed, I was familiar with the well-known phenomenon of interspecies differences in plasma protein binding and would have expected the same for brain tissue binding. As a result, I got extremely curious in finding out why the story was different with brain tissues and how the team at Pfizer conducted their study to draw this conclusion.

Before getting into the details of this study, let’s first set the scene to refresh one’s knowledge of pharmacokinetics and make sure that we are all on the same page. One of the most important, well accepted and widely applied concept in drug discovery and development, used to establish PK/PD models among other things, is the free drug hypothesis. This concept stipulates that in vivo efficacy is solely determined by the free (unbound) drug concentration at the site of action rather than total (bound and unbound) drug concentration, and that only the free drug is able to distribute from the systemic circulation across membranes to tissues. This notion which is key to understand the action of all drugs becomes even more essential in the context of CNS pharmacological agents. Indeed, the CNS is an extravascular compartment separated from the systemic circulation by the BBB and only free drug from the plasma can cross this biological barrier. The capacity of a drug to diffuse through the BBB is referred to as CNS permeability or CNS penetration, and is controlled and influenced by a number of physicochemical properties. Once in the CNS, the drug undergoes binding to the brain tissues and only a portion of the free drug in the plasma is free in the CNS to exert a pharmacological effect.

A number of techniques have been developed and used to determine the unbound drug concentration in the brain, such as in vivo microdialysis, cerebrospinal fluid sampling, and a combined measurement of brain tissue binding and brain distribution from plasma and brain concentration time courses. The latest approach is routinely used in pharmaceutical research since it requires less compound-specific optimisation and provides a direct measurement in the compartment of interest. In addition, it is a less challenging, more reproducible and higher throughput technique than the other methods. The fraction unbound (fu) of drugs in brain tissues can be obtained from two different biological systems: brain homogenates or brain slices. Although brain slices are considered more physiologically relevant, brain homogenates are more widely used, since they are easier to handle and store than slices, and can be readily obtain from commercial suppliers. Both systems afford good correlation with each other, indicating that brain tissue binding is primarily governed by nonspecific binding to lipophilic components rather than binding to intact structural elements.

Similarly to plasma protein binding, brain tissue binding has been routinely determined in multiple species to account for any potential species dependence. However from 2007, a few studies using a limited number of drugs or a limited number of species indicated that brain tissue binding might be species-independent. Pfizer decided to conduct the present study to confirm the species-independent phenomenon of brain tissue binding and determine if it was not limited to certain species, certain classes of pharmaceutical agents or a certain nature of brain binding. With this objective, a large number of structurally diverse compounds, covering a wide range of physicochemical properties (logD from -1.43 to 6.01, MWt from 151 to 823 g/mol, tPSA from 12 to 220 Å2) and brain binding characteristics (fu from 0.0005 to 0.5) in multiple animal species, was selected. The brain tissue binding of these compounds was assessed in seven species and strains (Wistar Han rat, Sprague-Dawley rat, CD-1 mouse, Hartley guinea pig, beagle dog, cynomolgus monkey and human), commonly used in neuroscience drug discovery.

Rigorous statistical orthogonal regression was used to analyse the concordance of the data between species, strain and within strain. The results unambiguously demonstrate that the drugs unbound fractions were strongly correlated (R2 ranging from 0.93 to 0.99) across the various species and strains tested. Importantly, the cross-species/strain correlations were extremely similar to the interassay correlation with the same species. Statistical analysis were performed and indicated that no correction was required for the extrapolation of fraction unbound from one species to the other species/strains. These results suggest that determination of brain tissue binding in a single species can replace multispecies determinations. The authors argue that the difference in composition of the brain relative to the plasma is a likely cause of the species-independent nature of brain tissue binding. Indeed, the brain has a much higher lipid contents (11% lipid and 7.9% protein versus 0.65% lipid and 18% protein in the plasma) which could account for the predominant non-specific binding of drugs to brain tissues. The absence of species-specific brain proteins selectively binding to drugs could be another potential explanation.

As stated earlier, the results from this study allowed the DMPK scientists at Pfizer to conclude that measuring brain tissue binding is essential for drugs intended to be used as CNS pharmacological agents, but that a single species measurement is sufficient. In addition, these results demonstrated that brain tissue binding variations can be ruled out to explain observed interspecies differences in the behaviour of CNS drugs.

I personally found the scientific paper1 describing the above study extremely enlightening and hope that this blog article will be of interest to other medicinal chemists or scientists working in the field of neuroscience.

Blog written by Tristan Reuillon



  1. Drug Metabolism and Disposition 2011, 39, 1270-1277, Species Independence in Brain Tissue Binding Using Brain Homogenates (doi: 10.1124/dmd.111.038778)


The Importance of Good Communication in Drug Development

A commentary published in Cell last month discusses how the quality of communication between and within institutions influences how successfully promising biological findings are translated into novel drug therapies.1

In the commentary it is suggested that most drug development programs which are ultimately successful proceed via a standard approach (Figure 1). In this figure the authors describe the steps involved in a typical drug development program and highlight key communication challenges associated with each.


Figure 1. Communication Challenges Associated with the Process of Drug Development1

The first major challenge identified is regarding candidate therapeutic targets and lead compounds being “oversold” or published with insufficient technical review such that there is a high attrition rate when results cannot be replicated or confirmed by other researchers. A study is given as an example in which Amgen scientists attempt to confirm results from dozens of cancer research publications and only succeed in replicating the findings in six of 53 cases.2 Similar stories are all too familiar, and while the causes are still being widely discussed, the communication failure is very evident.

Attempts to address this issue have begun with changes to the manuscript review process. Some journals are employing a checklist for authors, intended to document the rigor with which experiments are conducted, e.g. Nature: Reporting Checklist For Life Sciences Articles, and other editors have expressed interest in having statisticians consult on certain manuscripts. As the authors point the finger particularly at academic research, they suggest that industry-based scientists with technical expertise be engaged, both in reviewing drug discovery reports and in consulting throughout projects. Although they acknowledge that nowadays experts with relevant knowledge are often embedded within academic institutions. In addition they suggest that editors should insist that reports include an honest communication of any limitations of novel targets or inhibitors but emphasise that this should not preclude publication.

The second challenge is associated with how drug discovery projects are executed. Typically the teams responsible for driving projects are comprised of an array of scientists from biology, medicinal chemistry, biochemistry, structural biology, etc. Although they share a common goal because each scientist is an expert trained in their specific discipline they often speak in different “languages”. The authors highlight that teams are required to make many pivotal decisions and suggest that these can only be well informed where experts avoid use of jargon such that all team members can understand and examine the vital information. They also propose that it is as important to communicate experimental failure as it is success. Appropriate scrutiny early in the day should translate into an improved success rate. Given as an example is an analysis of AstraZeneca’s drug development pipeline, which revealed that 40% of the projects classified as efficacy failures in Phase II were associated with a failure to link the molecular target to the intended disease indication.3

Thirdly the authors highlight a challenge regarding selecting the correct patient population as drugs enter clinical development. The PARP inhibitor, olaparib, is used as an example. Despite olaparib originally being shown to confer synthetic lethality in cancers with BRCA1 and BRCA2 mutations,4 initial clinical development utilised a larger, heterogeneous population and the inhibitor failed to demonstrate good efficacy, regrettably delaying the approval of the drug. Improved communication could have prevented this. To address the issue, the National Centre for Advancing Translational Science  has been established within the National Institutes of Health (NIH) to “transform the translational process so that new treatments and cures for disease can be delivered to patients faster”. It is suggested that educating future researchers and physicians in interdisciplinary science to improve communication will prove key to the success of precision medicine.

Finally the authors note that the biomedical community must better communicate with the public in attempt to explain the disappointing success rate of expensive drug development programs.

While disease biology is poorly understood it is inevitable that clinical experimentation will remain unpredictable. However, this commentary convincingly suggests that improved communication throughout the drug discovery process will improve efficiency and increase the likelihood of success.

Blog written by Katie Duffell


  1. Settleman, J. & Cohen, R. L. Communication in Drug Development : ‘“ Translating ”’ Scientific Discovery. Cell 164, 1101–1104 (2016). doi:10.1016/j.cell.2016.02.050
  2. Begley, C. G. & Ellis, L. M. Drug development: Raise standards for preclinical cancer research. Nature 483, 531–533 (2012). doi:10.1038/483531a
  3. Cook, D., Brown, D., Alexander, R., March, R. & Morgan, P. Lessons learned from the fate of AstraZeneca’s drug pipeline: a five-dimensional framework. Nat. Rev. Drug Discov. 13, 419–431 (2014). doi:10.1038/nrd4309
  4. Bryant, H. E. et al. Specific killing of BRCA2-deficient tumours with inhibitors of poly(ADP-ribose) polymerase. Nature 434, 913–918 (2005). doi:10.1038/nature03443


Screening cascade targeting PPIs

Screening cascades need to be designed specifically for the target chosen and for the biological activity for the target. Projects targeting the interactions between two proteins (protein-protein interaction [PPI]) targets frequently have no enzyme activity that can be measured and therefore the screening cascade needs to employ different non-enzymatic methods to measure compound efficacy. Whilst there are many methods around a recent paper published exemplifies an excellent screening cascade for targeting a PPI (1).

In this paper the group were targeting a DNA-damage repair complex involved in the Fanconi anaemia pathway, a mechanism involved in tumour resistance. Specifically the interaction of FA complementation M group (FANCM) and the RecQ-mediated genome instability protein (RMI) complex.

The screening cascade used a fluorescence polarisation (FP) assay as the primary screen, followed by hit confirmation in an alpha screen-based proximity assay to demonstrate inhibition of the PPI. Surface Plasmon Resonance (SPR) and Isothermal titration calorimetry (ITC) were then used to demonstrate compound binding to the RMI core complex.

The group was specifically targeting the interaction within the FANCM complex, between RMI and MM2 and the FP assay was designed to measure this interaction. Therefore a standard format (figure 1) with the labelled MM2 protein having a Kd <5 nM. The assay was miniaturised to 384 format and the screen produced Z` of 0.53. The alpha screen assay (figure 2) was a proximity-based assay with MM2 bound to the streptavidin donor bead and the RMI complex bound to the acceptor bead. The Z` for the alpha screen assay (0.75) was better than the FP assay. Despite this the FP assay was chosen for the initial single point screen rather than the alpha screen, presumably on cost grounds.


Figure 1 Schematic of the fluorescence polarisation (FP)


Figure 2 Schematic of the alpha screen proximity assay

The alpha screen was therefore used as the counter screen and interestingly on the hit published there was a 10 fold discrepancy in the IC50 from the FP assay (450 uM) vs. alpha screen (36 uM). Direct interaction of this hit to the RMI complex was confirmed by both SPR and ITC that produced a consistent Kd of 7.8 and 3.4 uM respectively.

What is particularly nice about this cascade is that three of the screens, the FP assay, alpha screen and SPR can all be run in high throughput therefore although the publication only talks about a pilot screen of 74,000 compounds it could easily be scaled up to many more compounds. The only lower-throughput assay is the ITC, which is at the end of the cascade.


Blog writted by Trevor Askwith


  1. Voter AF, Manthei K a, Keck JL. A High-Throughput Screening Strategy to Identify Protein-Protein Interaction Inhibitors That Block the Fanconi Anemia DNA Repair Pathway. J. Biomol. Screen. (March 8, 2016). doi: 10.1177/1087057116635503.

More tales from the cystic fibrosis pig

The cystic fibrosis (CF) pig generated by the Welsh Group in Iowa continues to deliver insight into the mechanistic basis of CF lung disease. Perhaps the most salient of the lessons that this model has provided to date is the importance of airway pH and its critical influence on the early bacterial colonisation of the lungs. The latest instalment of the story highlights the role of the non-gastric proton pump, ATP12A, in the airway and how in the absence of normal CFTR function, it drives acidification and the compromised ability of the lung to defend itself (Science. 2016 351(6272): 503-507).

Prior to the generation of the CFTR-/- pig it had been appreciated that the airway mucosa of CF patients was relatively acidic when compared to healthy controls although the putative implications for the disease were unknown. Early studies with the pig highlighted that within hours of birth, the bacterial load in the airways was higher in the CF animals (Sci Transl Med. 2010 2(29):29ra31). Follow-up studies demonstrated that the airway surface liquid (ASL) of the CF pigs was lacking a bactericidal capacity that was evident in the littermate controls. It transpired that the acidic pH of the ASL attenuated the activity of a number of innate defence molecules including lysozyme and lactoferrin and this likely conferred the susceptibility to early bacterial colonisation (Nature. 2012 487(7405):109-13).

Studies in the pig and also primary human airway epithelial cells are now demonstrating that in the absence of CFTR-mediated bicarbonate secretion, it is H+ secretion via the non-gastric H+/K+ ATPase (ATP12A) that leads to an unchecked acidification of the mucosa. Inhibition of apical ATP12A with oubain partially alkalinised the ASL and restored the anti-bacterial activity of mucosal innate defence proteins in cultured human and porcine airway epithelial cells. Furthermore, adenoviral-mediated expression of Atp12a in mouse airway epithelium induced a ‘CF-like’ acidic / immunocompromised phenotype. Of note, endogenous expression of Atp12a in mouse airways was significantly lower than observed in human and pig, a finding that may contribute to the protection of the various CF mouse models of cystic fibrosis, all of which fail to develop any signs of lung disease.

ATP12A is a cAMP activated H+ pump. To this end, Welsh and colleagues raise the intriguing question as to the risk-benefit balance of using agents that can elevate cAMP in the lung epithelium. Inhaled β2-receptor agonists will unquestionably relax airway smooth muscle and offer the potential to bronchodilate the airways. However, could there also be a downside to this elevation in cAMP in terms of an acidification of the ASL and increased susceptibility to infection? Of note, similar theoretical risks regarding the use of cAMP-elevating agents in CF have been previously raised in respect to their potential to elevate ENaC function and therefore further dehydrate the airway mucosa (Science. 1983 221(4615):1067-70).

ATP12A does therefore offer the potential to be a new drug target whereby inhibitors would be predicted to boost innate defence in the CF lung. In view of the low airway pH in additional respiratory diseases that are associated with increased infection risk, could this mechanism and therapeutic benefit be extended beyond cystic fibrosis?


Blog written by Henrey Danahay

Filtering promiscuous compounds in early drug discovery: is it a good idea?

This is the question posed by the authors of a recent Drug Discovery Today article (Senger et al, Drug Discovery Today, article in press ). On first reading the question the instant answer from most practising medicinal chemists is surely yes, however the authors make some very good points that eliminating certain groups such as indoles from screening collection is somewhat short sighted or just simply wrong. On the other hand the authors do make some rather surprising and possibly naïve statement such as the recommendation to include quinone structures in phenotypic screening collections, an experienced medicinal chemist would argue there is simply no point testing them as they will be active anyway!

The article highlights the difference between classical biochemical screening with the objective of identifying hits which are active against a single target and phenotypic screening where the objective is to identify functionally active compounds irrespective of their mechanism of action. The authors propose that the now popular PAINS filters (Nature 2014, 513, 481-483) are appropriate for biochemical screening but should not be applied to compound collections used for phenotypic screening. They highlight a number of examples of structural types which are present in known drugs but would be removed by applying PAINS filters for example quinones (table 1). It is true that there are many quinone containing drugs on the market, but it is difficult to agree with the recommendation that even in phenotypic screens such compounds should be included. The compounds simply work by scavenging electrons and given the number of marketed drugs that contain quinones there is little need for more. Perhaps the only justification is to include marketed drugs in phenotypic screening collections with a view to future repositioning.

.pault Table1

A second set of problematic compounds highlighted as potentially useful are acyl hydrazones. Most medicinal chemists would not agree with the recommendation to include these in  screening collections as whilst there are examples of drugs (fig 1) which contain this group such functionality brings with it so many potential issues during development (genetoxicity, chemical instability, metabolism to a reactive hydrazine and many more) that they are best avoided in the first place.



Despite the recommendation to consider unattractive groups such as quinones and acyl hydrazones in phenotypic assays the third set of compounds highlighted by the authors is very sensible and prompts the question are filters in some cases too strict? The authors highlight indoles are privileged scaffolds. Indoles occur widely in marketed drugs they are chemically inert and generally safe. There are reports that certain indoles can form reactive metabolites – but this is true of any scaffold and is certainly not a reason to remove them from screening collections.

This is an interesting and thought provoking article and whilst it makes some interesting suggestions rightly challenges current thinking on the composition of screening collections. Drug discovery organisations are very risk averse these days some would say too risk averse. Whilst there are certain functionalities that should never appear, removing too many structural types can limit diversity and ultimately lead to a reduced chance of success in all modes of screening.

Blog wiritten by Paul Beswick

Tackling cancer: How new updates for old techniques could be the future of drug screening

An overwhelming proportion of drug development in both academia and industry is focused on identifying novel drug targets for the treatment of cancer. The Sussex Drug Discovery Centre is no exception; our oncology portfolio is constantly expanding and changing in order to be at the cutting edge of research for a disease that has an economic cost of £15.8bn a year in the UK alone, and is responsible for almost 15% of human deaths worldwide.

Typically, the drug development process begins with identifying inhibitors of a therapeutic target, which involves screening small molecules against a biological component involved in the disease pathway – often a specific enzyme or receptor. Because cancers are caused by mutations to genes that regulate cell growth and differentiation, resulting in unregulated cell division, most targets in cancer research are transcription factors or their corresponding gene products. While many drugs have entered clinical development this way – such as those targeting the bromodomain reader BRD4, histone deacetylase (HDAC) enzymes, and DNA methylation – challenges remain in identifying drugs that do more than simply indiscriminately kill cells, and that are able to target key transcription factors with molecular features that are considered ‘undruggable’.

In a revolutionary new study, researchers at the University of North Carolina developed and tested a new high-throughput screening technique that makes it possible to test potential drug compounds against an altered transcription factor previously thought to be ‘undruggable’ that is present in most Ewing sarcomas – a highly malignant bone and soft tissue tumour that affects children and young adults. Ewing sarcoma is characterized by a chromosomal translocation that produces the chimeric transcription factor EWSR1-FLI1; its corresponding gene product then localizes to specific regions of DNA, causing the chromatin to unwind. This creates a unique, disease-specific chromatin signature resulting from nucleosome depletion, making EWSR1-FLI1 an attractive drug target despite the fact that the mechanism for its action remains unknown. In this study, the researchers adapted formaldehyde-assisted isolation of regulatory elements (FAIRE), a biochemical assay for the identification of nucleosome-depleted regions of the genome, to a miniaturized and automated form that allowed high-throughput testing of the effects of compounds on proteins that regulate chromatin.

FAIRE in its standard format is dependent on organic extraction with a mixture of phenol and chloroform; in high-throughput (HT)-FAIRE this phase was substituted with a column-based approach to allow for the switch to automation, and performed similarly when compared side-by-side with standard FAIRE. Then, using a custom library of 640 small molecules, a screen was performed against a cell model of Ewing sarcoma to see if any of the molecules could restore normal chromatin structure. This was defined by a reduction in the FAIRE signal, indicative of a decrease in chromatin accessibility as the EWSR1-FLI1-mediated effects are reversed and the chromatin re-folds. Through the screen it was found that HDAC inhibitors were particularly effective at restoring chromatin structure, and while this class of compounds had been previously identified at a potential treatment for Ewing sarcoma, the screen also elicited novel compounds that were active in the cell model.

These results are undoubtedly remarkable, and highlight the advantages of a technique such as HT-FAIRE over the target-based screening approach that is the staple of modern drug discovery. By targeting a specific characteristic of Ewing sarcoma, HT-FAIRE removes the need for lengthy enzyme assay optimization or a comprehensive understanding of the molecular mechanisms underpinning the disease. If this method can identify potential drugs in one specific type of cancer, it may be possible to apply this technique to other cancers, and potentially reform drug discovery in the process.

Blog written by Chloe Koulouris