Search

Drug repurposing aims to find new uses for already existing and approved drugs. Normally this has occurred through serendipity while high throughput screening of FDA approved drug libraries has been responsible for increasing its popularity. Computational repurposing is a newer credible approach to drug discovery that has been steadily building a foundation with applications and examples of experimental validation. One particular area of computational repurposing is using machine learning. We now provide a brief overview of recent developments in drug repurposing using machine learning alongside other computational approaches for comparison. In addition, we will highlight several applications of computational drug repurposing using machine learning for cancer indications targeting kinase inhibitors, Alzheimer’s disease as well as COVID-19 from many different groups as well as our own. This overview should serve to rebalance the ‘repurposing hype’ of the past year, set expectations of what is possible as well as suggest areas for future development.

Machine learning methods are attracting considerable attention from the pharmaceutical industry for use in drug discovery and applications beyond. In recent studies, we and others have applied multiple machine learning algorithms and modeling metrics and, in some cases, compared molecular descriptors to build models for individual targets or properties on a relatively small scale. Several research groups have used large numbers of datasets from public databases such as ChEMBL in order to evaluate machine learning methods of interest to them. The largest of these types of studies used on the order of 1400 datasets. We have now extracted well over 5000 datasets from CHEMBL for use with the ECFP6 fingerprint and in comparison of our proprietary software Assay Central with random forest, k-nearest neighbors, support vector classification, naïve Bayesian, AdaBoosted decision trees, and deep neural networks (three layers). Model performance was assessed using an array of fivefold cross-validation metrics including area-under-the-curve, F1 score, Cohen's kappa, and Matthews correlation coefficient. Based on ranked normalized scores for the metrics or datasets, all methods appeared comparable, while the distance from the top indicated that Assay Central and support vector classification were comparable. Unlike prior studies which have placed considerable emphasis on deep neural networks (deep learning), no advantage was seen in this case. If anything, Assay Central may have been at a slight advantage as the activity cutoff for each of the over 5000 datasets representing over 570,000 unique compounds was based on Assay Central performance, although support vector classification seems to be a strong competitor. We also applied Assay Central to perform prospective predictions for the toxicity targets PXR and hERG to further validate these models. This work appears to be the largest scale comparison of these machine learning algorithms to date. Future studies will likely evaluate additional databases, descriptors, and machine learning algorithms and further refine the methods for evaluating and comparing such models.

Ultraviolet-visible (UV-Vis) absorption spectra are routinely collected as part of high-performance liquid chromatography (HPLC) analysis systems and can be used to identify chemical reaction products by comparison to the reference spectra. Here, we present UV-adVISor as a new computational tool for predicting the UV-Vis spectra from a molecule's structure alone. UV-Vis prediction was approached as a sequence-to-sequence problem. We utilized Long-Short Term Memory and attention-based neural networks with Extended Connectivity Fingerprint Diameter 6 or molecule SMILES to generate predictive models for the UV spectra. We have produced two spectrum datasets (dataset I, N = 949, and dataset II, N = 2222) using different compound collections and spectrum acquisition methods to train, validate, and test our models. We evaluated the prediction accuracy of the complete spectra by the correspondence of wavelengths of absorbance maxima and with a series of statistical measures (the best test set median model parameters are in parentheses for model II), including RMSE (0.064), R2 (0.71), and dynamic time warping (DTW, 0.194) of the entire spectrum curve. Scrambling molecule structures with the experimental spectra during training resulted in a degraded R2, confirming the utility of the approaches for prediction. UV-adVISor is able to provide fast and accurate predictions for libraries of compounds.

Neurons express overlapping homeostatic mechanisms to regulate synaptic function and network properties in response to perturbations of neuronal activity. Endocannabinoids (eCBs) are bioactive lipids synthesized in the post-synaptic compartments to regulate synaptic transmission, plasticity, and neuronal excitability primarily through retrograde activation of pre- synaptic cannabinoid receptor type 1 (CB1). The eCB system is well-situated to regulate neuronal network properties and coordinate pre- and post-synaptic activity. However, the role of the eCB system in homeostatic adaptations to neuronal hyperactivity is unknown. To address this issue, we used western blot and targeted lipidomics to measure adaptations in eCB system to bicuculline (BCC)-induced chronic hyperexcitation in mature (>DIV21) cultured rat cortical neurons, and used multielectrode array recording and live-cell imaging of glutamate dynamics to test the effects of pharmacological manipulations of eCB on network activities. We show that BCC-induced chronic hyperexcitation triggers homeostatic downscaling and a coordinated adaptation to enhance tonic eCB signaling. Hyperexcitation triggers first the downregulation of fatty acid amide hydrolase (FAAH), the lipase that degrades the eCB anandamide, then an accumulation of anandamide and related metabolites, and finally a delayed upregulation of surface and total CB1. Additionally, we show that BCC-induced downregulation of surface AMPA-type glutamate receptors (AMPARs) and upregulation of CB1 occur through independent mechanisms. Finally, we show that endocannabinoids support baseline network activities before and after downscaling and is engaged to suppress network activities during adaptation to hyperexcitation. We discuss the implications of our findings in the context of downscaling and homeostatic regulation of oscillatory network activities.Significance statementNeurons are remarkably resilient to perturbations in network activities thanks to the expression of overlapping homeostatic adaptations. In response to network hyperactivity or silencing, neurons respond through regulating excitatory and inhibitory post-synaptic neurotransmitter receptors density, probability of pre-synaptic neurotransmitter release, and/or membrane excitability. The endocannabinoid system is a prominent signaling pathway at many synapses that is known to be involved in multiple forms of short- and long-term synaptic plasticity. Here we find that components of the endocannabinoid system are upregulated in response to chronic hyperexcitation of cultured cortical neurons, and that endocannabinoid signaling is required to maintain network activity but also suppresses network events during hyperexcitation. This work supports a novel tonic homeostatic function for the endocannabinoid system in neurons.

Drug discovery is a multi-stage process, often beginning with the identification of active molecules from a high-throughput screen or machine learning model. Once structure activity relationship trends become well established, identifying new analogs with better properties is important. Synthesizing these new compounds is a logical next step, and is key to research groups that have a synthetic chemistry team or external collaborators. Generative machine learning models have become widely adopted to generate new molecules and explore molecular space, with the goal of discovering novel compounds with desires properties. These generative models have been composed from recurrent neural networks (RNNs), Variational Autoencoders (VAEs), and Generative Adversarial Networks (GANs) and are often combined with transfer learning or scoring of physicochemical properties to steer generative design. While these generative models have proven useful in generating new molecular libraries, often they are not capable of addressing a wide variety of potential problems, and often converge into similar molecular space when combined with a scoring function for desired properties. In addition, generated compounds are often not synthetically feasible, reducing their capabilities outside of virtual composition and limiting their usefulness in real-world scenarios. Here we introduce a suite of automated tools called MegaSyn representing 3 components: a new hill-climb algorithm which makes use of SMILES-based RNN generative models, analog generation software, and retrosynthetic analysis coupled with fragment analysis to score molecules for their synthetic feasibility. We now describe the development and testing of this suite of tools and propose how they might be used to optimize molecules or prioritize promising lead compounds using test case examples.

Timelapse TIRF microscopy of pH-sensitive GFP (pHluorin) attached to vesicle SNARE proteins is an effective method to visualize single vesicle exocytic events in cell culture. To perform an unbiased, efficient identification and analysis of such events, a computer-vision based approach was developed and implemented in MATLAB. The analysis pipeline consists of a cell segmentation and exocytic-event identification algorithm. The computer-vision approach includes tools for investigating multiple parameters of single events, including the half-life of fluorescence decay and peak ΔF/F, as well as whole-cell analysis of the frequency of exocytosis. These and other parameters of fusion are used in a classification approach to distinguish distinct fusion modes. Here a newly built GUI performs the analysis pipeline from start to finish. Further adaptation of Ripley's K function in R Studio is used to distinguish between clustered, dispersed, or random occurrence of fusion events in both space and time.

Ultraviolet-visible (UV-Vis) absorption spectra are routinely collected as part of high-performance liquid chromatography (HPLC) analysis systems and can be used to identify chemical reaction products by comparison to reference spectra. Here, we present UV-adVISor as a new computational tool for predicting UV-Vis spectra from a molecule’s structure alone. UV-Vis prediction was approached as a sequence-to-sequence problem. We utilized Long-Short Term Memory and attention-based neural networks with Extended Connectivity Fingerprint diameter 6 or molecule SMILES to generate predictive models for UV-spectra. We have produced two spectrum datasets (Dataset I, N = 949 and Dataset II, N = 2222) using different compound collections and spectrum acquisition methods to train, validate, and test our models. We evaluated the prediction accuracy of the complete spectra by the correspondence of wavelengths of absorbance maxima and with a series of statistical measures (the best test set median model parameters are in parentheses for Model II), including RMSE (0.064), R2 (0.71), and dynamic time warping (DTW, 0.194) of the entire spectrum curve. Scrambling molecule structures with experimental spectra during training resulted in a degraded R2, confirming the utility of the approaches for prediction. UV-adVISor is able to provide fast and accurate predictions for libraries of compounds.

Familial dysautonomia (FD) is an autonomic and sensory neuropathy caused by a mutation in the splice donor site of intron 20 of the ELP1 gene. Variable skipping of exon 20 leads to a tissue-specific reduction in the level of ELP1 protein. We have shown that the plant cytokinin kinetin is able to increase cellular ELP1 protein levels in vivo and in vitro through correction of ELP1 splicing. Studies in FD patients determined that kinetin is not a practical therapy due to low potency and rapid elimination. To identify molecules with improved potency and efficacy, we developed a cell-based luciferase splicing assay by inserting renilla (Rluc) and firefly (Fluc) luciferase reporters into our previously well-characterized ELP1 minigene construct. Evaluation of the Fluc/Rluc signal ratio enables a fast and accurate way to measure exon 20 inclusion. Further, we developed a secondary assay that measures ELP1 splicing in FD patient-derived fibroblasts. Here we demonstrate the quality and reproducibility of our screening method. Development and implementation of this screening platform has allowed us to efficiently screen for new compounds that robustly and specifically enhance ELP1 pre-mRNA splicing.

The ability to calculate whether small molecules will cross the blood-brain barrier (BBB) is an important task for companies working in neuroscience drug discovery. For a decade, scientists have relied on relatively simplistic rules such as Pfizer's central nervous system multiparameter optimization models (CNS-MPO) for guidance during the drug selection process. In parallel, there has been a continued development of more sophisticated machine learning models that utilize different molecular descriptors and algorithms; however, these models represent a "black box" and are generally less interpretable. In both cases, these methods predict the ability of small molecules to cross the BBB using the molecular structure information on its own without in vitro or in vivo data. We describe here the implementation of two versions of Pfizer's algorithm (Pf-MPO.v1 and Pf-MPO.v2) and compare it with a Bayesian machine learning model of BBB penetration trained on a data set of 2296 active and inactive compounds using extended connectivity fingerprint descriptors. The predictive ability of these approaches was compared with 40 known CNS active drugs initially used by Pfizer as their positive set for validation of the Pf-MPO.v1 score. 37/40 (92.5%) compounds were predicted as active by the Bayesian model, while only 30/40 (75%) received a desirable Pf-MPO.v1 score ≥4 and 33/40 (82.5%) received a desirable Pf-MPO.v2 score ≥4, suggesting the Bayesian model is more accurate than MPO algorithms. This also indicates machine learning models are more flexible and have better predictive power for BBB penetration than simple rule sets that require multiple, accurate descriptor calculations. Our machine learning model statistics are comparable to recent published studies. We describe the implications of these findings and how machine learning may have a role alongside more interpretable methods.

Rare diseases impact hundreds of millions of individuals worldwide. However, few therapies exist to treat the rare disease population because financial resources are limited, the number of patients affected is low, bioactivity data is often nonexistent, and very few animal models exist to support preclinical development efforts. Sialidosis is an ultrarare lysosomal storage disorder in which mutations in the NEU1 gene result in the deficiency of the lysosomal enzyme sialidase-1. This enzyme catalyzes the removal of sialic acid moieties from glycoproteins and glycolipids. Therefore, the defective or deficient protein leads to the buildup of sialylated glycoproteins as well as several characteristic symptoms of sialidosis including visual impairment, ataxia, hepatomegaly, dysostosis multiplex, and developmental delay. In this study, we used a bibliometric tool to generate links between lysosomal storage disease (LSD) targets and existing bioactivity data that could be curated in order to build machine learning models and screen compounds in silico. We focused on sialidase as an example, and we used the data curated from the literature to build a Bayesian model which was then used to score compound libraries and rank these molecules for in vitro testing. Two compounds were identified from in vitro testing using microscale thermophoresis, namely sulfameter (K d 2.15 ± 1.02 μM) and mexenone (K d 8.88 ± 4.02 μM), which validated our approach to identifying new molecules binding to this protein, which could represent possible drug candidates that can be evaluated further as potential chaperones for this ultrarare lysosomal disease for which there is currently no treatment. Combining bibliometric and machine learning approaches has the ability to assist in curating small molecule data and model building, respectively, for rare disease drug discovery. This approach also has the capability to identify new compounds that are potential drug candidates.

Machine learning methods are attracting considerable attention from the pharmaceutical industry for use in drug discovery and applications beyond. In recent studies we have applied multiple machine learning algorithms, modeling metrics and in some cases compared molecular descriptors to build models for individual targets or properties on a relatively small scale. Several research groups have used large numbers of datasets from public databases such as ChEMBL in order to evaluate machine learning methods of interest to them. The largest of these types of studies used on the order of 1400 datasets. We have now extracted well over 5000 datasets from CHEMBL for use with the ECFP6 fingerprint and comparison of our proprietary software Assay CentralTM with random forest, k-Nearest Neighbors, support vector classification, naïve Bayesian, AdaBoosted decision trees, and deep neural networks (3 levels). Model performance was assessed using an array of five-fold cross-validation metrics including area-under-the-curve, F1 score, Cohen’s kappa and Matthews correlation coefficient. Based on ranked normalized scores for the metrics or datasets all methods appeared comparable while the distance from the top indicated Assay CentralTM and support vector classification were comparable. Unlike prior studies which have placed considerable emphasis on deep neural networks (deep learning), no advantage was seen in this case where minimal tuning was performed of any of the methods. If anything, Assay CentralTM may have been at a slight advantage as the activity cutoff for each of the over 5000 datasets representing over 570,000 unique compounds was based on Assay CentralTMperformance, but support vector classification seems to be a strong competitor. We also apply Assay CentralTM to prospective predictions for PXR and hERG to further validate these models. This work currently appears to be the largest comparison of machine learning algorithms to date. Future studies will likely evaluate additional databases, descriptors and algorithms, as well as further refining methods for evaluating and comparing models.

Boyer et al. describe reversible nondegradative ubiquitination of the actin polymerase VASP, mediated by a pair of E3 ubiquitin ligases TRIM9 and TRIM67. VASP regulation is necessary for responses of filopodia, growth cones, and axons to the guidance cue netrin-1., Appropriate axon guidance is necessary to form accurate neuronal connections. Axon guidance cues that stimulate cytoskeletal reorganization within the growth cone direct axon navigation. Filopodia at the growth cone periphery have long been considered sensors for axon guidance cues, yet how they respond to extracellular cues remains ill defined. Our previous work found that the filopodial actin polymerase VASP and consequently filopodial stability are negatively regulated via nondegradative TRIM9-dependent ubiquitination. Appropriate VASP ubiquitination and deubiquitination are required for axon turning in response to the guidance cue netrin-1. Here we show that the TRIM9-related protein TRIM67 outcompetes TRIM9 for interacting with VASP and antagonizes TRIM9-dependent VASP ubiquitination. The surprising antagonistic roles of two closely related E3 ubiquitin ligases are required for netrin-1–dependent filopodial responses, axon turning and branching, and fiber tract formation. We suggest a novel model in which coordinated regulation of VASP ubiquitination by a pair of interfering ligases is a critical element of VASP dynamics, filopodial stability, and axon guidance.

Vinculin (Vcn) is a ubiquitously expressed cytoskeletal protein that links transmembrane receptors to actin filaments, and plays a key role in regulating cell adhesion, motility, and force transmission. Metavinculin (MVcn) is a Vcn splice isoform that contains an additional exon encoding a 68-residue insert within the actin binding tail domain. MVcn is selectively expressed at sub-stoichiometic amounts relative to Vcn in smooth and cardiac muscle cells. Mutations in the MVcn insert are linked to various cardiomyopathies. In vitro analysis has previously shown that while both proteins can engage filamentous (F)-actin, only Vcn can promote F-actin bundling. Moreover, we and others have shown that MVcn can negatively regulate Vcn-mediated F-actin bundling in vitro. To investigate functional differences between MVcn and Vcn, we stably expressed either Vcn or MVcn in Vcn-null mouse embryonic fibroblasts. While both MVcn and Vcn were observed at FAs, MVcn-expressing cells had larger but fewer focal adhesions per cell compared to Vcn-expressing cells. MVcn-expressing cells migrated faster and exhibited greater persistence compared to Vcn-expressing cells, even though Vcn-containing FAs assembled and disassembled faster. Magnetic tweezer measurements on Vcn-expressing cells show a typical cell stiffening phenotype in response to externally applied force; however, this was absent in Vcn-null and MVcn-expressing cells. Our findings that MVcn expression leads to larger but fewer FAs per cell, in conjunction with the inability of MVcn to bundle F-actin in vitro and rescue the cell stiffening response, are consistent with our previous findings of actin bundling deficient Vcn variants, suggesting that deficient actin-bundling may account for some of the differences between Vcn and MVcn.

Appropriate axon guidance is necessary to form accurate neuronal connections. Guidance cues stimulate reorganization of the cytoskeleton within the distal growth cone at the tip of the extending axon. Filopodia at the periphery of the growth cone have long been considered sensors for axon guidance cues, yet how they perceive and respond to extracellular cues remains ill-defined. Our previous work found that the filopodial actin polymerase VASP is regulated via TRIM9-dependent nondegradative ubiquitination, and that appropriate VASP ubiquitination and deubiquitination are required for axon turning in response to the guidance cue netrin-1. Here we show that the TRIM9-related protein TRIM67 antagonizes VASP ubiquitination by outcompeting the TRIM9:VASP interaction. This antagonistic role is required for netrin-1 dependent filopodial responses, axon turning and branching, and fiber tract formation. We suggest a novel model that coordinated regulation of nondegradative VASP ubiquitination by a pair of ligases is a critical element of axon guidance.

SUMMARY Neuronal morphogenesis involves dramatic plasma membrane expansion, fueled by soluble N-ethylmaleimide-sensitive factor attachment protein eceptors (SNARE)-mediated exocytosis. Distinct fusion modes described at synapses include full-vesicle fusion (FVF) and kiss-and-run fusion (KNR). During FVF, lumenal cargo is secreted and vesicle membrane incorporates into the plasma membrane. During KNR, a transient fusion pore secretes cargo but closes without membrane addition. In contrast, fusion modes are not described in developing neurons. Here, we resolve individual exocytic events in developing murine cortical neurons and use classification tools to identify four distinguishable fusion modes: two FVF-like modes that insert membrane material and two KNR-like modes that do not. Discrete fluorescence profiles suggest distinct behavior of the fusion pore. Simulations and experiments agree that FVF-like exocytosis provides sufficient membrane material for morphogenesis. We find the E3 ubiquitin ligase TRIM67 promotes FVF-like exocytosis in part by limiting incorporation of the Qb/Qc SNARE SNAP47 into SNARE complexes and, thus, SNAP47 involvement in exocytosis., Graphical Abstract, In Brief Urbina et al. identify four exocytic modes in developing neurons: KNRd, KNRi, FVFd, FVFi. Simulations and experiments suggest that FVFi and FVFd provide material for plasma membrane expansion. Deletion of Trim67 decreases FVFi and FVFd while reducing surface area. SNAP47 incorporation into SNARE complexes alters fusion pore behavior, increasing KNRd.

Urbina et al. use a new computer-vision image analysis tool and extended clustering statistics to demonstrate that the spatiotemporal distribution of constitutive VAMP2-mediated exocytosis is dynamic in developing neurons. The exocytosis pattern is modified by both developmental time and the guidance cue netrin-1, regulated differentially in the soma and neurites, and distinct from exocytosis in nonneuronal cells., Neurite elongation and branching in developing neurons requires plasmalemma expansion, hypothesized to occur primarily via exocytosis. We posited that exocytosis in developing neurons and nonneuronal cells would exhibit distinct spatiotemporal organization. We exploited total internal reflection fluorescence microscopy to image vesicle-associated membrane protein (VAMP)–pHluorin—mediated exocytosis in mouse embryonic cortical neurons and interphase melanoma cells, and developed computer-vision software and statistical tools to uncover spatiotemporal aspects of exocytosis. Vesicle fusion behavior differed between vesicle types, cell types, developmental stages, and extracellular environments. Experiment-based mathematical calculations indicated that VAMP2-mediated vesicle fusion supplied excess material for the plasma membrane expansion that occurred early in neuronal morphogenesis, which was balanced by clathrin-mediated endocytosis. Spatial statistics uncovered distinct spatiotemporal regulation of exocytosis in the soma and neurites of developing neurons that was modulated by developmental stage, exposure to the guidance cue netrin-1, and the brain-enriched ubiquitin ligase tripartite motif 9. In melanoma cells, exocytosis occurred less frequently, with distinct spatial clustering patterns.

Morphological elongation of developing neurons requires plasmalemma expansion, hypothesized to occur primarily via exocytosis. We posited that exocytosis in developing neurons and non-neuronal cells exhibit distinct spatiotemporal organization. We exploited TIRF microscopy to image VAMP-pHluorin mediated exocytosis in murine embryonic cortical neurons and interphase melanoma cells, and developed computer-vision software and statistical tools to uncover spatiotemporal aspects of exocytosis. Vesicle fusion behavior differed between vesicle types, cell types, developmental stage, and extracellular environment. Experiment-based mathematical calculations indicated that VAMP2-mediated vesicle fusion supplied excess material for the plasma membrane expansion that occurred early in neuronal morphogenesis, which was balanced in part by clathrin-mediated endocytosis. Spatial statistics uncovered distinct spatiotemporal regulation of exocytosis in the soma and neurites of developing neurons that was modulated by developmental stage, exposure to the guidance cue netrin-1, and the brain-enriched ubiquitin ligase TRIM9. In melanoma cells, exocytosis occurred less frequently with distinct spatial clustering patterns.

In the presence of netrin, tripartite motif protein 9 (TRIM9) promotes deleted in colorectal cancer (DCC) clustering, but TRIM9-dependent ubiquitination of DCC is reduced. Loss of ubiquitination promotes an interaction between DCC and FAK and FAK activation. FAK activation is required for the progression from SNARE assembly to exocytic vesicle fusion, which supplies membrane material for axon branching., Extracellular netrin-1 and its receptor deleted in colorectal cancer (DCC) promote axon branching in developing cortical neurons. Netrin-dependent morphogenesis is preceded by multimerization of DCC, activation of FAK and Src family kinases, and increases in exocytic vesicle fusion, yet how these occurrences are linked is unknown. Here we demonstrate that tripartite motif protein 9 (TRIM9)-dependent ubiquitination of DCC blocks the interaction with and phosphorylation of FAK. Upon netrin-1 stimulation TRIM9 promotes DCC multimerization, but TRIM9-dependent ubiquitination of DCC is reduced, which promotes an interaction with FAK and subsequent FAK activation. We found that inhibition of FAK activity blocks elevated frequencies of exocytosis in vitro and elevated axon branching in vitro and in vivo. Although FAK inhibition decreased soluble N-ethylmaleimide attachment protein receptor (SNARE)-mediated exocytosis, assembled SNARE complexes and vesicles adjacent to the plasma membrane increased, suggesting a novel role for FAK in the progression from assembled SNARE complexes to vesicle fusion in developing murine neurons.