BindingSite-AugmentedDTA: enabling a next-generation pipeline for interpretable prediction models in drug repurposing.

While research into drug-target interaction (DTI) prediction is fairly mature, generalizability and interpretability are not always addressed in the existing works in this field. In this paper, we propose a deep learning (DL)-based framework, called BindingSite-AugmentedDTA, which improves drug-target affinity (DTA) predictions by reducing the search space of potential-binding sites of the protein, thus making the binding affinity prediction more efficient and accurate. Our BindingSite-AugmentedDTA is highly generalizable as it can be integrated with any DL-based regression model, while it significantly improves their prediction performance. Also, unlike many existing models, our model is highly interpretable due to its architecture and self-attention mechanism, which can provide a deeper understanding of its underlying prediction mechanism by mapping attention weights back to protein-binding sites. The computational results confirm that our framework can enhance the prediction performance of seven state-of-the-art DTA prediction algorithms in terms of four widely used evaluation metrics, including concordance index, mean squared error, modified squared correlation coefficient ($r^2_m$) and the area under the precision curve. We also contribute to three benchmark drug-traget interaction datasets by including additional information on 3D structure of all proteins contained in those datasets, which include the two most commonly used datasets, namely Kiba and Davis, as well as the data from IDG-DREAM drug-kinase binding prediction challenge. Furthermore, we experimentally validate the practical potential of our proposed framework through in-lab experiments. The relatively high agreement between computationally predicted and experimentally observed binding interactions supports the potential of our framework as the next-generation pipeline for prediction models in drug repurposing.

AttentionSiteDTI: an interpretable graph-based model for drug-target interaction prediction using NLP sentence-level relation classification.

In this study, we introduce an interpretable graph-based deep learning prediction model, AttentionSiteDTI, which utilizes protein binding sites along with a self-attention mechanism to address the problem of drug-target interaction prediction. Our proposed model is inspired by sentence classification models in the field of Natural Language Processing, where the drug-target complex is treated as a sentence with relational meaning between its biochemical entities a.k.a. protein pockets and drug molecule. AttentionSiteDTI enables interpretability by identifying the protein binding sites that contribute the most toward the drug-target interaction. Results on three benchmark datasets show improved performance compared with the current state-of-the-art models. More significantly, unlike previous studies, our model shows superior performance, when tested on new proteins (i.e. high generalizability). Through multidisciplinary collaboration, we further experimentally evaluate the practical potential of our proposed approach. To achieve this, we first computationally predict the binding interactions between some candidate compounds and a target protein, then experimentally validate the binding interactions for these pairs in the laboratory. The high agreement between the computationally predicted and experimentally observed (measured) drug-target interactions illustrates the potential of our method as an effective pre-screening tool in drug repurposing applications.

CNP-miR146a improves outcomes in a two-hit acute- and ventilator-induced lung injury model.

Acute respiratory distress syndrome (ARDS) has high mortality (~40 %) and requires the lifesaving intervention of mechanical ventilation. A variety of systemic inflammatory insults can progress to ARDS, and the inflamed and injured lung is susceptible to ventilator-induced lung injury (VILI). Strategies to mitigate the inflammatory response while restoring pulmonary function are limited, thus we sought to determine if treatment with CNP-miR146a, a conjugate of novel free radical scavenging cerium oxide nanoparticles (CNP) to the anti-inflammatory microRNA (miR)-146a, would protect murine lungs from acute lung injury (ALI) induced with intratracheal endotoxin and subsequent VILI. Lung injury severity and treatment efficacy were evaluated via lung mechanical function, relative gene expression of inflammatory biomarkers, and lung morphometry (stereology). CNP-miR146a reduced the severity of ALI and slowed the progression of VILI, evidenced by improvements in inflammatory biomarkers, atelectasis, gas volumes in the parenchymal airspaces, and the stiffness of the pulmonary system.

Broad-Spectrum, Potent, and Durable Ceria Nanoparticles Inactivate RNA Virus Infectivity by Targeting Virion Surfaces and Disrupting Virus-Receptor Interactions.

There is intense interest in developing long-lasting, potent, and broad-spectrum antiviral disinfectants. Ceria nanoparticles (CNPs) can undergo surface redox reactions (Ce3+ ↔ Ce4+) to generate ROS without requiring an external driving force. Here, we tested the mechanism behind our prior finding of potent inactivation of enveloped and non-enveloped RNA viruses by silver-modified CNPs, AgCNP1 and AgCNP2. Treatment of human respiratory viruses, coronavirus OC43 and parainfluenza virus type 5 (PIV5) with AgCNP1 and 2, respectively, prevented virus interactions with host cell receptors and resulted in virion aggregation. Rhinovirus 14 (RV14) mutants were selected to be resistant to inactivation by AgCNP2. Sequence analysis of the resistant virus genomes predicted two amino acid changes in surface-located residues D91V and F177L within capsid protein VP1. Consistent with the regenerative properties of CNPs, surface-applied AgCNP1 and 2 inactivated a wide range of structurally diverse viruses, including enveloped (OC43, SARS-CoV-2, and PIV5) and non-enveloped RNA viruses (RV14 and feline calicivirus; FCV). Remarkably, a single application of AgCNP1 and 2 potently inactivated up to four sequential rounds of virus challenge. Our results show broad-spectrum and long-lasting anti-viral activity of AgCNP nanoparticles, due to targeting of viral surface proteins to disrupt interactions with cellular receptors.

Lung function improves after delayed treatment with CNP-miR146a following acute lung injury.

Acute respiratory distress syndrome (ARDS) is a highly morbid pulmonary disease characterized by hypoxic respiratory failure. Its pathogenesis is characterized by unrestrained oxidative stress and inflammation, with long-term sequelae of pulmonary fibrosis and diminished lung function. Unfortunately, prior therapeutic ARDS trials have failed and therapy is limited to supportive measures. Free radical scavenging cerium oxide nanoparticles (CNP) conjugated to the anti-inflammatory microRNA-146a (miR146a), termed CNP-miR146a, have been shown to prevent acute lung injury in a pre-clinical model. In this study, we evaluated the potential of delayed treatment with CNP-miR146a at three or seven days after injury to rescue the lung from acute injury. We found that intratracheal CNP-miR146a administered three days after injury lowers pulmonary leukocyte infiltration, reduce inflammation and oxidative stress, lower pro-fibrotic gene expression and collagen deposition in the lung, and ultimately improve pulmonary function.

Cerium oxide nanoparticle conjugation to microRNA-146a mechanism of correction for impaired diabetic wound healing.

Diabetic wounds represent a significant healthcare burden and are characterized by impaired wound healing due to increased oxidative stress and persistent inflammation. We have shown that CNP-miR146a synthesized by the conjugation of cerium oxide nanoparticles (CNP) to microRNA (miR)-146a improves diabetic wound healing. CNP are divalent metal oxides that act as free radical scavenger, while miR146a inhibits the pro-inflammatory NFκB pathway, so CNP-miR146a has a synergistic role in modulating both oxidative stress and inflammation. In this study, we define the mechanism(s) by which CNP-miR146a improves diabetic wound healing by examining immunohistochemical and gene expression analysis of markers of inflammation, oxidative stress, fibrosis, and angiogenesis. We have found that intradermal injection of CNP-miR146a increases wound collagen, enhances angiogenesis, and lowers inflammation and oxidative stress, ultimately promoting faster closure of diabetic wounds.

DNA-Modified Plasmonic Sensor for the Direct Detection of Virus Biomarkers from the Blood.

The rapid spread of viral infections demands early detection strategies to minimize proliferation of the disease. Here, we demonstrate a plasmonic biosensor to detect Dengue virus, which was chosen as a model, via its nonstructural protein NS1 biomarker. The sensor is functionalized with a synthetic single-stranded DNA oligonucleotide and provides high affinity toward NS1 protein present in the virus genome. We demonstrate the detection of NS1 protein at a concentration of 0.1-10 µg/mL in bovine blood using an on-chip microfluidic plasma separator integrated with the plasmonic sensor which covers the clinical threshold of 0.6 µg/mL of high risk of developing Dengue hemorrhagic fever. The conceptual and practical demonstration shows the translation feasibility of these microfluidic optical biosensors for early detection of a wide range of viral infections, providing a rapid clinical diagnosis of infectious diseases directly from minimally processed biological samples at point of care locations.

UnbiasedDTI: Mitigating Real-World Bias of Drug-Target Interaction Prediction by Using Deep Ensemble-Balanced Learning.

Drug-target interaction (DTI) prediction through in vitro methods is expensive and time-consuming. On the other hand, computational methods can save time and money while enhancing drug discovery efficiency. Most of the computational methods frame DTI prediction as a binary classification task. One important challenge is that the number of negative interactions in all DTI-related datasets is far greater than the number of positive interactions, leading to the class imbalance problem. As a result, a classifier is trained biased towards the majority class (negative class), whereas the minority class (interacting pairs) is of interest. This class imbalance problem is not widely taken into account in DTI prediction studies, and the few previous studies considering balancing in DTI do not focus on the imbalance issue itself. Additionally, they do not benefit from deep learning models and experimental validation. In this study, we propose a computational framework along with experimental validations to predict drug-target interaction using an ensemble of deep learning models to address the class imbalance problem in the DTI domain. The objective of this paper is to mitigate the bias in the prediction of DTI by focusing on the impact of balancing and maintaining other involved parameters at a constant value. Our analysis shows that the proposed model outperforms unbalanced models with the same architecture trained on the BindingDB both computationally and experimentally. These findings demonstrate the significance of balancing, which reduces the bias towards the negative class and leads to better performance. It is important to note that leaning on computational results without experimentally validating them and by relying solely on AUROC and AUPRC metrics is not credible, particularly when the testing set remains unbalanced.

Enzyme-Free Plasmonic Biosensor for Direct Detection of Neurotransmitter Dopamine from Whole Blood.

Complex biological fluids without pretreatment, separation, or purification impose stringent limitations on the practical deployment of label-free plasmonic biosensors for advanced assays needed in point of care applications. In this work, we present an enzyme-free plasmonic neurotransmitter dopamine biosensor integrated with a microfluidic plasma separator. This integrated device allows the in-line separation of plasma directly from the bloodstream and channels it to the active detection area, where inorganic cerium oxide nanoparticles function as local selective dopamine binding sites through strong surface redox reaction. A thorough understanding and engineering of the nanoparticles is carried out to maximize its dopamine sensitivity and selectivity. We obtain detection of dopamine at 100 fM concentration in simulated body fluid and 1 nM directly from blood without any prior sample preparation. The detection selectivity is found to be at least five-times higher compared to the common interfering species. This demonstration shows the feasibility of the practical implementation of the proposed plasmonic system in detection of variety of biomarkers directly from the complex biological fluids.

An unexpected phase transformation of ceria nanoparticles in aqueous media.

Cerium oxide Nanoparticles (CNPs) are of significant interest to the scientific community due to their wide spread applications in a variety of fields. It is proposed that size dependent variations in the extent of Ce3+ and Ce4+ oxidation states of cerium in CNPs determines the performance of CNPs in application environments. To obtain greater molecular and structural understanding of chemical state transformations previously reported for ceria ≈ 3 nm nanoparticles (CNPs) in response to changing ambient conditions, microXRD and Raman measurements were carried out for various solution conditions. The particles were observed to undergo a reversible transformation from a defective ceria structure to a non-ceria amorphous oxy-hydroxide/peroxide phase in response to the addition of 30% hydrogen peroxide. For CNPs made up of ~8 nm crystallites, a partial transformation was observed and no transformation was observed for CNPs made up of ~ 40 nm crystallites. This observation of differences in size dependent transition behavior may help explain the benefits of using smaller CNPs in applications requiring regenerative behavior.

Temporal Distribution Patterns of Alexa Fluor 647-Conjugated CeNPs in the Mouse Retina After a Single Intravitreal Injection.

Intravitreal (IVT) injection of ophthalmic therapeutics is the most widely used drug delivery route to the posterior segment of the eye. We employed this method to deliver our inorganic, catalytic antioxidant, cerium oxide nanoparticles (CeNPs), to rodent models of retinal degeneration. A single IVT of CeNPs delays disease progression. Even though we have shown that our synthesized CeNPs are retained in the retina for over a year, we still do not know which cell types in the retina preferentially take up these nanoparticles. In this study, we examined the temporal and spatial distribution of fluorescently labeled CeNPs in retinal sections after IVT. We detected elevated fluorescent signals in all the layers where retinal neurons and glia reside and retinal pigment epithelium (RPE) up to 90 days post injection. Additionally, we found that free fluorochrome accumulated in retinal vasculature instead of retinal cells. These data suggested that CeNP-conjugation mediated the targeting of the fluorochrome to retinal cells. We propose that CeNPs can be deployed as ophthalmic carriers to the retina.

High-Throughput, Protein-Targeted Biomolecular Detection Using Frequency-Domain Faraday Rotation Spectroscopy.

A clinically relevant magneto-optical technique (fd-FRS, frequency-domain Faraday rotation spectroscopy) for characterizing proteins using antibody-functionalized magnetic nanoparticles (MNPs) is demonstrated. This technique distinguishes between the Faraday rotation of the solvent, iron oxide core, and functionalization layers of polyethylene glycol polymers (spacer) and model antibody-antigen complexes (anti-BSA/BSA, bovine serum albumin). A detection sensitivity of ≈10 pg mL-1 and broad detection range of 10 pg mL-1 ≲ cBSA ≲ 100 µg mL-1 are observed. Combining this technique with predictive analyte binding models quantifies (within an order of magnitude) the number of active binding sites on functionalized MNPs. Comparative enzyme-linked immunosorbent assay (ELISA) studies are conducted, reproducing the manufacturer advertised BSA ELISA detection limits from 1 ng mL-1 ≲ cBSA ≲ 500 ng mL-1 . In addition to the increased sensitivity, broader detection range, and similar specificity, fd-FRS can be conducted in less than ≈30 min, compared to ≈4 h with ELISA. Thus, fd-FRS is shown to be a sensitive optical technique with potential to become an efficient diagnostic in the chemical and biomolecular sciences.

Tissue deposition and toxicological effects of commercially significant rare earth oxide nanomaterials: Material and physical properties.

Rare earth oxide (REO) materials are found naturally in earth's crust and at the nanoscale these REO nanoparticles exhibit unique thermal, electrical, and physicochemical properties. REO nanoparticles are widely used in different industrial sectors for ceramics, glass polishing, metallurgy, lasers, and magnets. Recently, some of these REO nanoparticles have been identified for their potential application in medicine, including therapy, imaging, and diagnostics. Concurrent research into the REO nanomaterials' toxicities has also raised concern for their environmental impacts. The correlation of REO nanoparticles mediated toxicity with their physiochemical properties can help to design nanoparticles with minimal effect on the environment and living organisms. In vitro assay revealed toxicity toward Human squamous epithelial cell line (CCL30) and Human umbilical vascular endothelial cells (HUVEC) at a concentration of 100 µM and higher. In vivo results showed, with the exception of CeO2 and Gd2 O3 , most of the naoparticles did not clear or had minimum clearance (10-20%) from the system. Elevated levels of alanine transferase were seen for animals given each different nanoparticle, however the increases were not significant for CeO2 and Dy2 O3 . Nephrotoxicity was only seen in case of Dy2 O3 and Gd2 O3 . Lastly, histological examination revealed presence of swollen hepatocytes which further confirms toxicity of the commercial REO nanomaterials. The in vivo toxicity is mainly due to excessive tissue deposition (70-90%) due to the commercial REO nanoparticles' poor physical properties (shape, stability, and extent of agglomeration). Therefore, optimization of nanoparticles physical properties is very important. © 2016 Wiley Periodicals, Inc. Environ Toxicol 32: 904-917, 2017.

Non-toxic retention of nanoceria in murine eyes.

PURPOSE: We have shown that cerium oxide nanoparticles (nanoceria), with unique characteristics and catalytic activities, are retained in the retina for more than 1 year after a single intravitreal injection and can be potentially used for the treatment of a variety of eye diseases. The objective of this study is to determine whether the retention of nanoceria in the eye causes inflammation or adverse side effects. METHODS: Wild-type (C57BL/6J) mice at P30 were intravitreally injected with several concentrations of nanoceria. The health of the photoreceptors was assessed by analyzing the expression of photoreceptor-specific genes, and the retinal structure and function. The effect of nanoceria was investigated by analyzing of the vascular system, the expression of inflammatory cytokines, and cellular infiltration into the eye. RESULTS: Our data showed that there were no changes in the retinal structure or function, or cytokine gene expression following a single intravitreal injection of nanoceria. CONCLUSIONS: Nanoceria, at doses ranging from 17.2 ng to 1720 ng per eye, do not cause any damage to the retinal structure and function by 30 days post injection. No cellular infiltration and no increases in inflammatory responses were found in the eyes. Our data indicate that nanoceria are safe to use for treatment of a variety of eye diseases.

Folic acid tagged nanoceria as a novel therapeutic agent in ovarian cancer.

BACKGROUND: Nanomedicine is a very promising field and nanomedical drugs have recently been used as therapeutic agents against cancer. In a previous study, we showed that Nanoceria (NCe), nanoparticles of cerium oxide, significantly inhibited production of reactive oxygen species, cell migration and invasion of ovarian cancer cells in vitro, without affecting cell proliferation and significantly reduced tumor growth in an ovarian cancer xenograft nude model. Increased expression of folate receptor-α, an isoform of membrane-bound folate receptors, has been described in ovarian cancer. To enable NCe to specifically target ovarian cancer cells, we conjugated nanoceria to folic acid (NCe-FA). Our aim was to investigate the pre-clinical efficacy of NCe-FA alone and in combination with Cisplatin. METHODS: Ovarian cancer cell lines were treated with NCe or NCe-FA. Cell viability was assessed by MTT and colony forming units. In vivo studies were carried in A2780 generated mouse xenografts treated with 0.1 mg/Kg NCe, 0.1 mg/Kg; NCe-FA and cisplatinum, 4 mg/Kg by intra-peritoneal injections. Tumor weights and burden scores were determined. Immunohistochemistry and toxicity assays were used to evaluate treatment effects. RESULTS: We show that folic acid conjugation of NCe increased the cellular NCe internalization and inhibited cell proliferation. Mice treated with NCe-FA had a lower tumor burden compared to NCe, without any vital organ toxicity. Combination of NCe-FA with cisplatinum decreased the tumor burden more significantly. Moreover, NCe-FA was also effective in reducing proliferation and angiogenesis in the xenograft mouse model. CONCLUSION: Thus, specific targeting of ovarian cancer cells by NCe-FA holds great potential as an effective therapeutic alone or in combination with standard chemotherapy.

Redox-Sensitive Cerium Oxide Nanoparticles Protect Human Keratinocytes from Oxidative Stress Induced by Glutathione Depletion.

Cerium oxide nanoparticles (CeNPs) have gathered much attention in the biomedical field due to its unique antioxidant property. It can protect cells and tissues from oxidative stress induced damage due to its autoregenerative redox cycle. Our study explores the antioxidant and antigenotoxic behavior of PEGylated CeNPs toward oxidative insult produced by buthionine sulfoximine (BSO) in human keratinocytes (HaCaT cells). BSO inhibits the γ-glutamylcysteinesynthetase (γ-GCS) enzyme and thus acts as a glutathione (GSH) depleting agent to modulate the cellular redox potential. GSH is a natural ROS scavenger present in the mammalian cells, and its depletion causes generation of reactive oxygen species (ROS). In this study, we challenged HaCaT cells (keratinocytes) with BSO to alter the redox potential within the cell and monitored toxicity, ROS generation, and nuclear fragmentation. We also followed changes in expressions of related proteins and genes. We found that PEGylated CeNPs can protect HaCaT cells from BSO-induced oxidative damage. BSO-exposed cells, preincubated with PEGylated CeNPs, showed better cell survival and significant decrease in the intracellular levels of ROS. We also observed decrease in lactate dehydrogenase (LDH) release and nuclear fragmentation in CeNP-treated cells that were challenged with BSO as compared to treatment with BSO alone. Exposure of HaCaT cells with BSO leads to altered expression of antioxidant genes and proteins, i.e., thioredoxin reductase (TrxR) and peroxiredoxin 6 (Prx6) whereas, in our study, pretreatment of PEGylated CeNPs reduces the need for induction of genes that produce enzymes involved in the defense against oxidative stress. Since, growing evidence argued the involvement of ROS in mediating death of mammalian cells in several ailments, our finding reinforces the use of PEGylated CeNPs as a potent pharmacological agent under the lower cellular GSH/GSSG ratios for the treatment of diseases mediated by free radicals.

The change in antioxidant properties of dextran-coated redox active nanoparticles due to synergetic photoreduction-oxidation.

Nanoparticles have proven to be novel material with resourceful applications in the field of nanomedicine. Cerium oxide nanoparticles (CNPs) coated with dextran (Dex-CNPs) have been shown to exhibit anticancer properties which is attributed to the change in oxidation states mediated at the oxygen vacancies on the surface of CNPs. In this study, the extreme sensitivity of Dex-CNPs to visible light is demonstrated using room light with a clear indication of synergetic phenomenon of photoreduction of CNPs in the presence of dextran which undergoes simultaneous oxidation. The phenomenon was further confirmed through a systematic time-based expedited study using a high intensity visible light source. The physiochemical changes of Dex-CNPs such as dispersion stability, pH, surface chemistry, antioxidant property, cytotoxicity and the surrounding microenvironment of Dex-CNPs were significantly altered on exposure to visible light, thereby affecting the biological response. Given the significance of nanoparticles which are widely researched nanomaterials, in different fields of nanotechnology and biomedicine, this study demonstrates the significant changes in physiochemical properties of Dex-CNPs with light. The photoreduction of Dex-CNPs affects its bifunctional applications in cancer therapy and thereby this study puts forward the necessity to preserve and sustain their properties through proper storage.

Aqueous medium induced optical transitions in cerium oxide nanoparticles.

Experimental and theoretical investigations were performed to investigate the effect of water on optical properties of nanoceria as a function of Ce(3+) concentration. Theoretical studies based on density functional plane-wave calculations reveal that the indirect optical transitions in bare ceria nanoparticles are red-shifted with an increase in the concentration of Ce(3+). However, ceria nanoparticles model with adsorbed water molecules show a blue shift in the indirect optical spectra under identical conditions. Direct optical transitions are almost independent of Ce(3+) concentration but show a pronounced blue shift in the aqueous environment relative to the bare nanoparticles. The theoretical study is consistent with our experimental observation in difference of shift behaviour in bare and aqueous suspended ceria nanoparticles. This change from red- to blue-shift in indirect optical transitions is associated with the polarization effect of water molecules on f-electron states.

Mechanical properties of mesoporous ceria nanoarchitectures.

Architectural constructs are engineered to impart desirable mechanical properties facilitating bridges spanning a thousand meters and buildings nearly 1 km in height. However, do the same 'engineering-rules' translate to the nanoscale, where the architectural features are less than 0.0001 mm in size? Here, we calculate the mechanical properties of a porous ceramic functional material, ceria, as a function of its nanoarchitecture using molecular dynamics simulation and predict its yield strength to be almost two orders of magnitude higher than the parent bulk material. In particular, we generate models of nanoporous ceria with either a hexagonal or cubic array of one-dimensional pores and simulate their responses to mechanical load. We find that the mechanical properties are critically dependent upon the orientation between the crystal structure (symmetry, direction) and the pore structure (symmetry, direction).