Machine-Learning in Drug Discovery

Virtual high-throughput screening (vHTS) can be used in drug discovery to replace experimental HTS or to preselect a subset of compounds for screening to reduce costs and time. The two main approaches to vHTS are structure-based or ligand-based screening. Structure-based screening does not need any known active molecules, but does need either a crystal structure or good homology model of the target protein. Ligand-based screening, on the other hand, does not need any knowledge about the target protein but identifies common patterns and features of known active molecules by creating a 2D fingerprint or 3D pharmacophore.

Scientists at Imperial College London/Equinox Pharma have developed a new rule-based vHTS methodology (INDDEx) that exploits logic-based machine learning to enhance performance in ligand-based screening.  INDDEx (Investigational Novel Drug Discovery by Example) is particularly good at identifying actives that are structurally distinct from the training set, making it useful for scaffold-hopping.

INDDEx learns easily interpretable qualitative logic rules from active ligands. These rules – in the form of ‘an active molecule requires fragment A and fragment B separated by a distance in Angstroms’ or ‘an active molecule requires the presence of fragment C’ or ‘an active molecule must NOT contain fragment D’ – give an insight into chemistry, relate molecular substructure to activity and can be used to guide the next steps of drug design chemistry. These qualitative rules are then weighted using Support Vector Machines (SVMs) to produce QSAR rules that can be used to generate novel in silico hits.

INDDEx (Investigational Novel Drug Discovery by Example)


INDDEx has been shown to be a powerful new approach to virtual screening whose strength lies in learning topological descriptors of multiple active compounds although, when considering scaffold hopping in isolation, INDDEx performs well even when there are small numbers of active molecules to learn from. One very attractive feature of INDDEx is that the rules that are produced can be readily understood and used by medicinal chemists. The technology has been extensively validated and shown to outperform comparable approaches (J. Phys. Chem. B 2012, 116, 6732-6739). In a study between Equinox and Imperial on sirtuin 2 (an NAD-dependent histone deacetylase) , INDDEx combined with structure-based docking was able to learn from only eight actives and identify a chemically novel hit that was experimentally validated to have an IC50 of 0.6 µM.

INDDEx has wide-scale applications including  rescuing failed programmes, directing hit-to-lead programmes and scaffold-hopping. Furthermore, INDDEx has the potential to derive rules for off-target activities such as the hERG receptor.

If you would like to find out more about INDDEx and Equinox Pharma, please contact us.

SkelGen – A Newly Available Tool for Computational Drug Design

drug designWith the rapidly growing body of biostructural information, structure-based drug design has increased in importance and a variety of computational methods have found a place in the drug discovery toolkit.

The de novo design program, SkelGen, was developed by De Novo Pharmaceuticals based on research begun in the Department of Pharmacology at the University of Cambridge. SkelGen constructs candidate ligands by assembling small molecular fragments within a protein target such as an enzyme or receptor (usually derived from X-ray crystal data). When growing a ligand, SkelGen uses information coded in the fragments and within its algorithm to favour synthetically tractable molecules. SkelGen is able to explore around one trillion low molecular weight, drug-like molecules using a default set of 1600 fragments. Since the accessible chemical space is so large, the majority of designed molecules are novel and patentable.

Whilst SkelGen can be run with minimal input, it also permits extensive control by the end-user, allowing the scientist to incorporate prior knowledge and insights into the drug design process. As well as completely de novo design, molecule generation can also be started from a user-defined fragment (for example, a low-affinity molecule identified by fragment-screening). SkelGen can also be used for scaffold hopping (chemotype switching) and focused library design.

Until recently SkelGen was only accessible through collaborations with De Novo Pharmaceuticals but is now available under both academic and commercial licenses. With these new licensing models, SkelGen can be a cost-effective (and accessible) tool for all scientists engaged in drug design. If you would like to find out more about SkelGen, please contact us.

Targeting Orphan Receptors for Multiple Sclerosis

ROR-alpha ligand binding domain
Crystal structure of the ligand-binding domain of RORα complexed with cholesterol sulfate. PDB ID=1S0X
Scientists at the Scripps Research Institute have reported on compounds that are able to suppress severity and disease progression in animal models of multiple sclerosis. The compounds, exemplified by SR1001, act by selectively suppressing a subset of T-helper cells characterised by their production of interleukin-17 (TH17 cells). TH17 cells have been implicated in a variety of autoimmune diseases including rheumatoid arthritis, multiple sclerosis, inflammatory bowel disease and systemic lupus erythematosus.

SR1001 selectively binds to two orphan nuclear receptors: retinoic acid receptor-related orphan receptors α and γt (RORα and RORγt). These receptors have indispensible roles in the development and function of TH17 cells, providing a mechanism for modulating one component of the immune system without general immunosuppression. The team reports that SR1001 induces a conformational change in the receptors that results in their reduced affinity for co-activators and increased affinity for co-repressors. The net result is inhibition of the receptors’ transcriptional activity.

SR1001 blocked the development of murine TH17 cells and inhibited cytokine production by differentiated murine and human TH17 cells. Although a drug is some way off, the team suggests that the results demonstrate the feasibility of targeting TH17 cells and the potential of such an approach for the treatment of autoimmune diseases.

The study is published in Nature.

Exploiting Bioenergetic Differences to Stop GVHD

Image: Flickr - Junichiro Aoyama
Bz-423, a mitochondrial F1F0-ATP synthase inhibitor, that has previously shown promise for the treatment of autoimmune disorders such as lupus, arthritis and psoriasis has now been shown to halt the progression of established graft-versus-host disease (GVHD) in mouse models of allogeneic bone marrow transplantation. GVHD, in which functional immune cells in the transplanted marrow mount an immunological attack on the recipient, is a common complication of allogeneic bone marrow transplantation that can cause severe organ damage and even be life-threatening. As well as demonstrating that Bz-423 can reduce GVHD clinical scores and improve survival in mice, researchers at the University of Michigan and the University of Florida have shed new light on the metabolism of alloreactive T cells.

Bz-423 structure
Resting lymphocytes meet their minimal demand for ATP using low rates of oxidative phosphorylation. Upon activation, normal lymphocytes meet the increased demand for ATP by dramatically increasing their rate of aerobic glycolysis which also helps to maintain high levels of antioxidants in activated cells. Data from the present study suggest that alloreactive T cells, on the other hand, rely heavily on increased oxidative phosphorylation to generate more ATP. This difference in energy generation is also seen in pathogenic T cells involved in autoimmune diseases and provides a mechanistic basis for the specific elimination of pathogenic cells by Bz-423, whilst preserving normal immune function.

The study is published in Science Translational Medicine.

Lycera Corporation has a program to develop orally bioavailable F1F0-ATP synthase inhibitors that exploit bioenergetic abnormalities in pathologically activated lymphocytes and result in the selective silencing of these cells.

The Right Tool for the Job?

Image: Flickr - Janne Moren
Tool compounds are used to explore the role of a specific protein in a biological context and – it goes without saying – that to obtain meaningful results in a complex situation, the tool compound should have appropriate potency and selectivity. The family of phospholipase C (PLC) enzymes play important regulatory roles and a small molecule inhibitor, U73122, has been extensively used to provide evidence for the involvement of PLCs in many cellular pathways. Recent reports, however, have questioned the selectivity of U73122 and scientists at the University of North Carolina and GlaxoSmithKline have now discovered that, even in its interaction with PLCs, U73122 may not be quite what it seems. When the team explored the effects of U73122 on human PLCs in cell-free micellar systems, they found that the compound actually increased the enzymatic activity of a number of isoforms in a concentration- and time-dependent manner. At micromolar concentrations, U713122 increased the activity of PLCβ3 by up to eight-fold, that of PLCγ1 by more than ten-fold, and that of PLCβ2 by around two-fold; PLCδ1 was neither activated nor inhibited.

U73122 structure
Activation of PLCβ3 was attenuated by competing nucleophiles, suggesting that activation involves covalent modification of the protein by the reactive maleimide group of U73122; the analogous succinimide, U73343, was not effective as an activator. Involvement of specific cysteine residues in the protein was demonstrated by LC/MS/MS experiments. Although N-ethyl maleimide (NEM) itself did not activate PLCβ3, excess NEM attenuated the U73122-mediated activation in a concentration-dependent manner. The authors propose an activation model in which U73122 irreversibly binds to multiple cysteine residues on PLCβ3 and acts as either a lipid anchor or interfacial recognition site for the enzyme, facilitating adsorption to the substrate interface (i.e. the micelle surface). The protein-linked U73122 increases the rate of lipase activity by keeping the enzyme in close proximity to substrate which is held in the membrane.

The study, which is published in the Journal of Biological Chemistry, provides strong evidence that U73122 activates PLC enzymes in cell-free systems, in contrast to its ‘established’ role as a specific inhibitor of this family. The authors suggest that U73122 may have opposing effects on cytosolic and membrane-bound enzymes and/or may modify other cellular nucleophiles and advise great caution when forming hypotheses based on the observed effects of U73122 in cellular systems.

Stressful ‘Lifestyle’ is Bad for Neurons

Image: Flickr - greenmelinda
Parkinson’s disease is characterised by loss of dopaminergic neurons in the area of the midbrain known as the substantia nigra. Although mitochondrial stress – an accumulation of damaging superoxide and free radicals – is believed to be the cause of cell death, it is not understood why this subset of neurons is especially vulnerable.

Researchers at Northwestern University have now suggested a possible answer: these neurons have an inherently stressful ‘lifestyle’. The cells in the substantia nigra act as pacemakers, releasing rhythmic bursts of dopamine. This activity is accompanied by an influx of calcium ions which must then be pumped back out of the cell in an energy-demanding process. The inflow of calcium ions is not essential for pacemaking activity so, if the energy needed to pump calcium ions out of the cell is adding extra stress, blocking the influx of calcium should help to alleviate this. Using mice engineered to express a redox-sensitive fluorescent protein in their mitochondria, the team showed that the opening of L-type calcium channels during normal pacemaking activity created an oxidant stress that was specific to dopaminergic cells of the substantia nigra. The oxidative stress, in turn, caused a defensive mild mitochondrial depolarization or uncoupling.

Although most cases of Parkinson’s disease have no known genetic cause, loss-of-function DJ-1 (PARK7) mutations can cause early-onset Parkinson’s disease in humans and transgenic mice lacking DJ-1 also show damage to dopaminergic cells in the substantia nigra. Knocking out DJ-1 down-regulates expression of two uncoupling proteins and increases oxidation of mitochondrial matrix proteins in dopaminergic neurons of the substantia nigra. Treatment of the transgenic animals with the L-type calcium channel blocker, isradipine, was found to protect the dopaminergic cells of the substantia nigra from oxidative damage.

The study, which is published in the journal Nature, builds on previous studies linking calcium channel blockade with protective effects in Parkinson’s disease.

A clinical trial is currently underway to examine the safety, tolerability and efficacy of isradipine – which is already approved for the treatment of high blood pressure – in patients with Parkinson’s disease. The hope is that the drug will slow disease progression and allow a broader window for existing symptomatic treatments.

Targeting the Tangles

Image: Flickr - Cory Doctorow
Although the relative importance of β-amyloid plaques and tau protein tangles in the progression of Alzheimer’s disease has been the subject of much debate, early emphasis was placed on the development of drugs to block production of β-amyloid. Although such compounds were shown to improve cognition in transgenic mice, unfortunately results from clinical trials have been more equivocal. Focus is now shifting to therapies that target tau pathology and, in a recent study, researchers from the University of Pennsylvania have identified a compound that reduced cognitive deficits in mutant human tau transgenic mice.

In healthy nerve cells, tau proteins interact with tubulin to stabilize axonal microtubules and promote tubulin assembly into microtubules. In Alzheimer’s disease and other ‘tauopathies’, hyperphosphorylated and misfolded tau proteins form insoluble neurofibrillary tangles that deplete levels of soluble tau and lead to destabilization of the microtubules and neuronal dysfunction. The team had previously proposed using microtubule-stabilising anti-cancer taxanes such as paclitaxel to treat tauopathies, but these do not penetrate the blood-brain barrier sufficiently well. The Penn team has now shown that once weekly treatment of tau transgenic mice with the brain-penetrant microtubule-stabilising agent, epothilone D, for three months significantly improved microtubule density and axonal integrity and also reduced cognitive deficits without notable side-effects.

The study, which is published in the Journal of Neuroscience, suggests that brain-penetrant microtubule-stabilising drugs could provide a new strategy for treating Alzheimer’s disease.

11β-HSD1 Inhibitor Improves Spatial Memory in Ageing Mice

Image: Wikimedia Commons - Rama
Inhibitors of 11β-hydroxysteroid dehydrogenase type 1 (11β-HSD1), a microsomal enzyme that converts cortisone into cortisol, are being developed to treat diabetes and metabolic disorders and now, in a study supported by the Wellcome Trust, researchers at the University of Edinburgh have shown that such compounds may also help to reduce – or even reverse – age-related memory loss. Such memory loss has been linked to increased activity of 11β-HSD1 and higher levels of glucocorticoids in the hippocamus, an area of the brain associated with memory.

Ageing mice display deficits in memory and learning similar to those experienced by some elderly people and life-long partial deficiency of 11β-HSD1 prevents this decline in transgenic mice. More surprisingly, improvements in memory – as judged by performance in a Y maze – were seen in mice after only ten days treatment with a selective 11β-HSD1 inhibitor, UE1961.

The team had previously shown that a non-selective 11β-HSD1 inhibitor, carbenoxolone, improves memory in healthy elderly men and in patients with type II diabetes after only one month of treatment. They now hope to complete preclinical assessment of the new compound and begin clinical trials within a year. The study is published in the Journal of Neuroscience.

New (Host) Target for Treating HCV

Water droplet
Image: Flickr - Jack Newton
Hepatitis C virus (HCV) is one of the most important causes of chronic liver disease and infection can lead ultimately to cirrhosis and liver cancer. Current standard-of-care treatment – a combination of pegylated α-interferon and ribavirin – is unable to clear the virus in all patients and new antiviral agents designed to inhibit specific viral enzymes such as the protease, helicase and polymerase are being developed.

DGAT-1 inhibitor
DGAT-1 inhibitor
Researchers led by a team at the Gladstone Institute of Virology and Immunology (GIVI) have now identified a human enzyme that is also needed for viral infectivity, a discovery that may offer a new strategy for treatment. The enzyme, diacylglycerol acyltransferase 1 (DGAT1), is one of two DGAT enzymes that catalyse the final step in triglyceride synthesis. HCV infection is closely tied to lipid metabolism and the Gladstone team showed that infection and replication is severely impaired in liver cells that lack DGAT1 activity: either RNAi-mediated knockdown of DGAT1 or treatment with a DGAT1 inhibitor was effective in limiting production of infectious viral particles. The team went on to show that DGAT1 interacts with the viral nucleocapsid core protein and is required for the trafficking of the core protein to lipid droplets. Knockdown of the other enzyme involved in triglyceride synthesis, DGAT2, had no effect on viral replication.

DGAT1 inhibitors are already being developed as treatments for type II diabetes and obesity and the new study, which is published in Nature Medicine, suggests that they may also be useful for treating HCV infection.

2-AG – Not Such Eternal Bliss?

Ananda at the first Buddhist council
Ananda at the First Buddhist Council - Image: Wikipedia - Sacca
The cannabinoid receptors, CB1 and CB2, were first identified as the receptors for the active components of Cannabis sativa. The endogenous ligands for the cannabinoid receptors, anandamide (from ananda, a Sanskrit word meaning ‘bliss’) and 2-arachidonoylglycerol (2-AG), exert most of their analgesic affects through binding to the main cannabinoid receptor in the brain, CB1. Attempts to prepare CB1 agonists that mimic the analgesic effects of the main component of cannabis, Δ9-tetrahydrocannabinol (THC) but have reduced potential for memory impairment, locomotor dysfunction, and possibly addiction have met with limited success and a recent alternative approach has been to devise ways of preventing the breakdown of anandamide and 2-AG. Signalling by the endocannabinoids is terminated by enzymatic hydrolysis which, for anandamide, is mediated by fatty acid amide hydrolase (FAAH) and, for 2-arachidonoylglycerol, by monoacylglycerol lipase (MAGL).

Inhibitors of both FAAH and MAGL produce analgesic effects but scientists at the Scripps Research Institute, Virginia Commonwealth University and the Medical College of Wisconsin have now shown that the effects of blocking MAGL are short-lived compared with the effects of blocking FAAH. Although a single injection of the MAGL inhibitor, JZL184, reduced pain in the mouse, after six consecutive daily injections, the effect disappeared. The chronically treated animals also became less sensitive to THC or the synthetic CB1 agonist, WIN55,212-2, and showed characteristic drug withdrawal symptoms when treated with the CB1 blocker, rimonabant.

Further tests showed that CB1 receptors were down-regulated in some brain areas of mice treated for 6 days with JZL184. Genetic disruption of MAGL also resulted in elevated levels of 2-AG and desensitised the CB1 signalling system, suggesting that chronic inhibition of MAGL may not provide effective analgesia. In contrast, chronic treatment with a FAAH inhibitor to boost anandamide levels did not lead to desensitisation of the CB1 system.

Although the team do not yet understand why chronic administration of FAAH inhibitors and MAGL inhibitors should produce such strikingly different results, the study, which is published in Nature Neuroscience, suggests that complete blockade of MAGL may not provide effective analgesia over the longer term.