Implementation of an AI-assisted fragment-generator in an open-source platform.

We recently reported a deep learning model to facilitate fragment library design, which is critical for efficient hit identification. However, our model was implemented in Python. We have now created an implementation in the KNIME graphical pipelining environment which we hope will allow experimentation by users with limited programming knowledge.

Fragment-based drug discovery-the importance of high-quality molecule libraries.

Fragment-based drug discovery (FBDD) is now established as a complementary approach to high-throughput screening (HTS). Contrary to HTS, where large libraries of drug-like molecules are screened, FBDD screens involve smaller and less complex molecules which, despite a low affinity to protein targets, display more 'atom-efficient' binding interactions than larger molecules. Fragment hits can, therefore, serve as a more efficient start point for subsequent optimisation, particularly for hard-to-drug targets. Since the number of possible molecules increases exponentially with molecular size, small fragment libraries allow for a proportionately greater coverage of their respective 'chemical space' compared with larger HTS libraries comprising larger molecules. However, good library design is essential to ensure optimal chemical and pharmacophore diversity, molecular complexity, and physicochemical characteristics. In this review, we describe our views on fragment library design, and on what constitutes a good fragment from a medicinal and computational chemistry perspective. We highlight emerging chemical and computational technologies in FBDD and discuss strategies for optimising fragment hits. The impact of novel FBDD approaches is already being felt, with the recent approval of the covalent KRASG12C inhibitor sotorasib highlighting the utility of FBDD against targets that were long considered undruggable.

Stabilization of the RAS:PDE6D Complex Is a Novel Strategy to Inhibit RAS Signaling.

RAS is a major anticancer drug target which requires membrane localization to activate downstream signal transduction. The direct inhibition of RAS has proven to be challenging. Here, we present a novel strategy for targeting RAS by stabilizing its interaction with the prenyl-binding protein PDE6D and disrupting its localization. Using rationally designed RAS point mutations, we were able to stabilize the RAS:PDE6D complex by increasing the affinity of RAS for PDE6D, which resulted in the redirection of RAS to the cytoplasm and the primary cilium and inhibition of oncogenic RAS/ERK signaling. We developed an SPR fragment screening and identified fragments that bind at the KRAS:PDE6D interface, as shown through cocrystal structures. Finally, we show that the stoichiometric ratios of KRAS:PDE6D vary in different cell lines, suggesting that the impact of this strategy might be cell-type-dependent. This study forms the foundation from which a potential anticancer small-molecule RAS:PDE6D complex stabilizer could be developed.

Automated Generation of Novel Fragments Using Screening Data, a Dual SMILES Autoencoder, Transfer Learning and Syntax Correction.

Fragment-based hit identification (FBHI) allows proportionately greater coverage of chemical space using fewer molecules than traditional high-throughput screening approaches. However, effectively exploiting this advantage is highly dependent on the library design. Solubility, stability, chemical complexity, chemical/shape diversity, and synthetic tractability for fragment elaboration are all critical aspects, and molecule design remains a time-consuming task for computational and medicinal chemists. Artificial neural networks have attracted considerable attention in automated de novo design applications and could also prove useful for fragment library design. Chemical autoencoders are neural networks consisting of encoder and decoder parts, which respectively compress and decompress molecular representations. The decoder is applied to samples drawn from the space of compressed representations to generate novel molecules that can be scored for properties of interest. Here, we report an autoencoder model using a recurrent neural network architecture, which was trained using 486,565 fragments curated from commercial sources, to simultaneously reconstruct both SMILES and chemical fingerprints. To explore its utility in fragment design, we applied transfer learning to the fingerprint decoder layers to train a classifier using 66 frequent hitter fragments identified from our screening campaigns. Using a particle swarm optimization sampling approach, we compare the performance of this "dual" model to an architecture encoding SMILES only. The dual model produced valid SMILES with improved features, considering a range of properties including aromatic ring counts, heavy atom count, synthetic accessibility, and a new fragment complexity score we term Feature Complexity (FeCo). Additionally, we demonstrate that generative performance is further enhanced by use of a simple syntax-correction procedure during training, in which invalid and undesirable SMILES are spiked into the training set. Finally, we used the syntax-corrected model to generate a library of novel candidate privileged fragments.

Re-introducing non-optimal synonymous codons into codon-optimized constructs enhances soluble recovery of recombinant proteins from Escherichia coli.

Gene synthesis services have largely superseded traditional PCR methods for the generation of cDNAs destined for bacterial expression vectors. This, in turn, has increased the application of codon-optimized cDNAs where codons rarely used by Escherchia coli are replaced with common synonymous codons to accelerate translation of the target. A markedly accelerated rate of expression often results in a significant uplift in the levels of target protein but a substantial proportion of the enhanced yield can partition to the insoluble fraction rendering a significant portion of the gains unavailable for native purification. We have assessed several expression attenuation strategies for their utility in the manipulation of the soluble fraction towards higher levels of soluble target recovery from codon optimized systems. Using a set of human small GTPases as a case study, we compare the degeneration of the T7 promoter sequence, the use of alternative translational start codons and the manipulation of synonymous codon usage. Degeneration of both the T7 promoter and the translational start codon merely depressed overall expression and did not increase the percentage of product recovered in native purification of the soluble fraction. However, the selective introduction of rare non-optimal codons back into the codon-optimized sequence resulted in significantly elevated recovery of soluble targets. We propose that slowing the rate of extension during translation using a small number of rare codons allows more time for the co-translational folding of the nascent polypeptide. This increases the proportion of the target recovered in the soluble fraction by immobilized metal affinity chromatography (IMAC). Thus, a "de-optimization" of codon-optimized cDNAs, to attenuate or pause the translation process, may prove a useful strategy for improved recombinant protein production.

Structure-based design, synthesis and biological evaluation of a novel series of isoquinolone and pyrazolo[4,3-c]pyridine inhibitors of fascin 1 as potential anti-metastatic agents.

Fascin is an actin binding and bundling protein that is not expressed in normal epithelial tissues but overexpressed in a variety of invasive epithelial tumors. It has a critical role in cancer cell metastasis by promoting cell migration and invasion. Here we report the crystal structures of fascin in complex with a series of novel and potent inhibitors. Structure-based elaboration of these compounds enabled the development of a series with nanomolar affinities for fascin, good physicochemical properties and the ability to inhibit fascin-mediated bundling of filamentous actin. These compounds provide promising starting points for fascin-targeted anti-metastatic therapies.

A Novel Small-Molecule Inhibitor of MRCK Prevents Radiation-Driven Invasion in Glioblastoma.

Glioblastoma (GBM) is an aggressive and incurable primary brain tumor that causes severe neurologic, cognitive, and psychologic symptoms. Symptoms are caused and exacerbated by the infiltrative properties of GBM cells, which enable them to pervade the healthy brain and disrupt normal function. Recent research has indicated that although radiotherapy (RT) remains the most effective component of multimodality therapy for patients with GBM, it can provoke a more infiltrative phenotype in GBM cells that survive treatment. Here, we demonstrate an essential role of the actin-myosin regulatory kinase myotonic dystrophy kinase-related CDC42-binding kinase (MRCK) in mediating the proinvasive effects of radiation. MRCK-mediated invasion occurred via downstream signaling to effector molecules MYPT1 and MLC2. MRCK was activated by clinically relevant doses per fraction of radiation, and this activation was concomitant with an increase in GBM cell motility and invasion. Furthermore, ablation of MRCK activity either by RNAi or by inhibition with the novel small-molecule inhibitor BDP-9066 prevented radiation-driven increases in motility both in vitro and in a clinically relevant orthotopic xenograft model of GBM. Crucially, treatment with BDP-9066 in combination with RT significantly increased survival in this model and markedly reduced infiltration of the contralateral cerebral hemisphere.Significance: An effective new strategy for the treatment of glioblastoma uses a novel, anti-invasive chemotherapeutic to prevent infiltration of the normal brain by glioblastoma cells.Cancer Res; 78(22); 6509-22. ©2018 AACR.

Discovery of Potent and Selective MRCK Inhibitors with Therapeutic Effect on Skin Cancer.

The myotonic dystrophy-related Cdc42-binding kinases MRCKα and MRCKß contribute to the regulation of actin-myosin cytoskeleton organization and dynamics, acting in concert with the Rho-associated coiled-coil kinases ROCK1 and ROCK2. The absence of highly potent and selective MRCK inhibitors has resulted in relatively little knowledge of the potential roles of these kinases in cancer. Here, we report the discovery of the azaindole compounds BDP8900 and BDP9066 as potent and selective MRCK inhibitors that reduce substrate phosphorylation, leading to morphologic changes in cancer cells along with inhibition of their motility and invasive character. In over 750 human cancer cell lines tested, BDP8900 and BDP9066 displayed consistent antiproliferative effects with greatest activity in hematologic cancer cells. Mass spectrometry identified MRCKα S1003 as an autophosphorylation site, enabling development of a phosphorylation-sensitive antibody tool to report on MRCKα status in tumor specimens. In a two-stage chemical carcinogenesis model of murine squamous cell carcinoma, topical treatments reduced MRCKα S1003 autophosphorylation and skin papilloma outgrowth. In parallel work, we validated a phospho-selective antibody with the capability to monitor drug pharmacodynamics. Taken together, our findings establish an important oncogenic role for MRCK in cancer, and they offer an initial preclinical proof of concept for MRCK inhibition as a valid therapeutic strategy.Significance: The development of selective small-molecule inhibitors of the Cdc42-binding MRCK kinases reveals their essential roles in cancer cell viability, migration, and invasive character. Cancer Res; 78(8); 2096-114. ©2018 AACR.

A fully automated procedure for the parallel, multidimensional purification and nucleotide loading of the human GTPases KRas, Rac1 and RalB.

Small GTPases regulate many key cellular processes and their role in human disease validates many proteins in this class as desirable targets for therapeutic intervention. Reliable recombinant production of GTPases, often in the active GTP loaded state, is a prerequisite for the prosecution of drug discovery efforts. The preparation of these active forms can be complex and often constricts the supply to the reagent intensive techniques used in structure base drug discovery. We have established a fully automated, multidimensional protein purification strategy for the parallel production of the catalytic G-domains of KRas, Rac1 and RalB GTPases in the active form. This method incorporates a four step chromatography purification with TEV protease-mediated affinity tag cleavage and a conditioning step that achieves the activation of the GTPase by exchanging GDP for the non-hydrolyzable GTP analogue GMPPnP. We also demonstrate that an automated method is efficient at loading of KRas with mantGDP for application in a SOS1 catalysed fluorescent nucleotide exchange assay. In comparison to more conventional manual workflows the automated method offers marked advantages in method run time and operator workload. This reduces the bottleneck in protein production while generating products that are highly purified and effectively loaded with nucleotide analogues.

Identification of a Selective G1-Phase Benzimidazolone Inhibitor by a Senescence-Targeted Virtual Screen Using Artificial Neural Networks.

Cellular senescence is a barrier to tumorigenesis in normal cells, and tumor cells undergo senescence responses to genotoxic stimuli, which is a potential target phenotype for cancer therapy. However, in this setting, mixed-mode responses are common with apoptosis the dominant effect. Hence, more selective senescence inducers are required. Here we report a machine learning-based in silico screen to identify potential senescence agonists. We built profiles of differentially affected biological process networks from expression data obtained under induced telomere dysfunction conditions in colorectal cancer cells and matched these to a panel of 17 protein targets with confirmatory screening data in PubChem. We trained a neural network using 3517 compounds identified as active or inactive against these targets. The resulting classification model was used to screen a virtual library of ~2M lead-like compounds. One hundred and forty-seven virtual hits were acquired for validation in growth inhibition and senescence-associated ß-galactosidase assays. Among the found hits, a benzimidazolone compound, CB-20903630, had low micromolar IC50 for growth inhibition of HCT116 cells and selectively induced senescence-associated ß-galactosidase activity in the entire treated cell population without cytotoxicity or apoptosis induction. Growth suppression was mediated by G1 blockade involving increased p21 expression and suppressed cyclin B1, CDK1, and CDC25C. In addition, the compound inhibited growth of multicellular spheroids and caused severe retardation of population kinetics in long-term treatments. Preliminary structure-activity and structure clustering analyses are reported, and expression analysis of CB-20903630 against other cell cycle suppressor compounds suggested a PI3K/AKT-inhibitor-like profile in normal cells, with different pathways affected in cancer cells.

A novel small-molecule MRCK inhibitor blocks cancer cell invasion.

BACKGROUND: The myotonic dystrophy kinase-related CDC42-binding kinases MRCKα and MRCKß regulate actin-myosin contractility and have been implicated in cancer metastasis. Along with the related ROCK1 and ROCK2 kinases, the MRCK proteins initiate signalling events that lead to contractile force generation which powers cancer cell motility and invasion. A potential strategy for cancer therapy is to reduce metastasis by blocking MRCK activity, either alone or in combination with ROCK inhibition. However, to date no potent small molecule inhibitors have been developed with selectivity towards MRCK. RESULTS: Screening a kinase-focused small molecule chemical library resulted in the identification of compounds with inhibitory activity towards MRCK. Medicinal chemistry combined with in vitro enzyme profiling led to the discovery of 4-chloro-1-(4-piperidyl)-N-[5-(2-pyridyl)-1H-pyrazol-4-yl]pyrazole-3-carboxamide (BDP00005290; abbreviated as BDP5290) as a potent MRCK inhibitor. X-ray crystallography of the MRCKß kinase domain in complex with BDP5290 revealed how this ligand interacts with the nucleotide binding pocket. BDP5290 demonstrated marked selectivity for MRCKß over ROCK1 or ROCK2 for inhibition of myosin II light chain (MLC) phosphorylation in cells. While BDP5290 was able to block MLC phosphorylation at both cytoplasmic actin stress fibres and peripheral cortical actin bundles, the ROCK selective inhibitor Y27632 primarily reduced MLC phosphorylation on stress fibres. BDP5290 was also more effective at reducing MDA-MB-231 breast cancer cell invasion through Matrigel than Y27632. Finally, the ability of human SCC12 squamous cell carcinoma cells to invade a three-dimensional collagen matrix was strongly inhibited by 2 µM BDP5290 but not the identical concentration of Y27632, despite equivalent inhibition of MLC phosphorylation. CONCLUSIONS: BDP5290 is a potent MRCK inhibitor with activity in cells, resulting in reduced MLC phosphorylation, cell motility and tumour cell invasion. The discovery of this compound will enable further investigations into the biological activities of MRCK proteins and their contributions to cancer progression.

The first intramolecular silene Diels-Alder reactions.

The synthesis of silaheterocycles through the first examples of an intramolecular silene Diels-Alder reaction is described.

Fragment-based hit identification: thinking in 3D.

The identification of high-quality hits during the early phases of drug discovery is essential if projects are to have a realistic chance of progressing into clinical development and delivering marketed drugs. As the pharmaceutical industry goes through unprecedented change, there are increasing opportunities to collaborate via pre-competitive networks to marshal multifunctional resources and knowledge to drive impactful, innovative science. The 3D Fragment Consortium is developing fragment-screening libraries with enhanced 3D characteristics and evaluating their effect on the quality of fragment-based hit identification (FBHI) projects.

Using fragment-based technologies to target protein-protein interactions.

Whilst fragment-based screening has found significant utility in aiding the discovery of high quality hits against a range of targets, the use of this technology in the protein-protein interaction inhibitor field is very much in its infancy. This review aims to highlight the key technologies used to identify fragment hits, such as NMR, SPR, X-ray crystallography and biochemical screening, the fragment-based protein-protein interaction case studies reported to date and, more importantly, the potential of this methodology in unearthing high quality hit molecules in this critical area of drug discovery. In addition, we also discuss some of the key aspects of fragment library design, the composition of a high quality library and suggest ways in which future, more structurally diverse fragments which occupy different regions of chemical space to the vast majority of current fragment libraries may be selected.

The design and synthesis of novel, potent and orally bioavailable N-aryl piperazine-1-carboxamide CCR2 antagonists with very high hERG selectivity.

A novel N-aryl piperazine-1-carboxamide series of human CCR2 chemokine receptor antagonists was discovered. Early analogues were potent at CCR2 but also inhibited the hERG cardiac ion channel. Structural modifications which decreased lipophilicity and basicity resulted in the identification of a sub-series with an improved margin over hERG. The pharmacological and pharmacokinetic properties of the lead compound from this series, N-(3,4-dichlorophenyl)-4-[(2R)-4-isopropylpiperazine-2-carbonyl]piperazine-1-carboxamide, are described.

Acyl polysilanes: new acyl anion equivalents for additions to electron-deficient alkenes.

Silenes, generated through thermolysis of acylpolysilanes, add to alpha,beta-unsaturated esters to form cyclobutanes and silylsubstituted cyclopropanes in moderate yields. Upon Si-C bond oxidation the cyclopropanes are converted directly to 1,4-dicarbonyl compounds, thus demonstrating the formal acyl anion chemistry of acyl polysilanes.

One-pot multistep Bohlmann-Rahtz heteroannulation reactions: synthesis of dimethyl sulfomycinamate.

[reaction: see text] The synthesis of dimethyl sulfomycinamate, the acidic methanolysis product of the sulfomycin family of thiopeptide antibiotics, from methyl 2-oxo-4-(trimethylsilyl)but-3-ynoate is achieved in a 2,3,6-trisubstituted pyridine synthesis that proceeds with total regiocontrol in 13 steps by the Bohlmann-Rahtz heteroannulation of a 1-(oxazol-4-yl)enamine or in 12 steps and 9% yield by three-component cyclocondensation with N-[3-oxo-3-(oxazol-4-yl)propanoyl]serine and ammonia in ethanol.

Structure-guided design of pyrazolo[1,5-a]pyrimidines as inhibitors of human cyclin-dependent kinase 2.

The protein structure guided design of a series of pyrazolo[1,5-a]pyrimidines with high potency for human cyclin-dependent kinase 2 (CDK2) is described. Some examples were shown to inhibit the growth of human colon tumour cells, were equipotent for CDK1 and were selective against GSK-3beta and other kinases.

Structure-based drug design targeting an inactive RNA conformation: exploiting the flexibility of HIV-1 TAR RNA.

The targeting of RNA for the design of novel anti-viral compounds represents an area of vast potential. We have used NMR and computational methods to model the interaction of a series of synthetic inhibitors of the in vitro RNA binding activities of a peptide derived from the transcriptional activator protein, Tat, from human immunodeficiency virus type 1. Inhibition has been measured through the monitering of fluorescence resonance energy transfer between fluorescently labeled peptide and RNA components. A series of compounds containing a bi-aryl heterocycle as one of the three substituents on a benzylic scaffold, induce a novel, inactive TAR conformation by stacking between base-pairs at the site of a three-base bulge within TAR. The development of this series resulted in an enhancement in potency (with Ki < 100 nM in an in vitro assay) and the removal of problematic guanidinium moieties. Ligands from this series can act as inhibitors of Tat-induced transcription in a cell-free system. This study validates the drug design strategy of using a ligand to target the RNA receptor in a non-functional conformation.

Rational design of inhibitors of HIV-1 TAR RNA through the stabilisation of electrostatic "hot spots".

The targeting of RNA for the design of novel anti-viral compounds has until now proceeded largely without incorporating direct input from structure-based design methodology, partly because of lack of structural data, and complications arising from substrate flexibility. We propose a paradigm to explain the physical mechanism for ligand-induced refolding of trans-activation response element (TAR RNA) from human immunodeficiency virus 1 (HIV-1). Based upon Poisson-Boltzmann analysis of the TAR structure, as bound by a peptide derived from the transcriptional activator protein, Tat, our hypothesis shows that two specific electrostatic interactions are necessary to stabilise the conformation. This result contradicts the belief that a single argininamide residue is responsible for stabilising the TAR fold, as well as the conventional wisdom that electrostatic interactions with RNA are non-specific or dominated by phosphates. We test this hypothesis by using NMR and computational methods to model the interaction of a series of novel inhibitors of the in vitro RNA-binding activities for a peptide derived from Tat. A subset of inhibitors, including the bis-guanidine compound rbt203 and its analogues, induce a conformation in TAR similar to that brought about by the protein. Comparison of the interactions of two of these ligands with the RNA and structure-activity relationships observed within the compound series, confirm the importance of the two specific electrostatic interactions in the stabilisation of the Tat-bound RNA conformation. This work illustrates how the use of medicinal chemistry and structural analysis can provide a rational basis for prediction of ligand-induced conformational change, a necessary step towards the application of structure-based methods in the design of novel RNA or protein-binding drugs.