Cyano-Hydrol green derivatives: Expanding the 9-cyanopyronin-based resonance Raman vibrational palette.

9-cyanopyronin is a promising scaffold that exploits resonance Raman enhancement to enable sensitive, highly multiplexed biological imaging. Here, we developed cyano-Hydrol Green (CN-HG) derivatives as resonance Raman scaffolds to expand the color palette of 9-cyanopyronins. CN-HG derivatives exhibit sufficiently long wavelength absorption to produce strong resonance Raman enhancement for near-infrared (NIR) excitation, and their nitrile peaks are shifted to a lower frequency than those of 9-cyanopyronins. The fluorescence of CN-HG derivatives is strongly quenched due to the lack of the 10th atom, unlike pyronin derivatives, and this enabled us to detect spontaneous Raman spectra with high signal-to-noise ratios. CN-HG derivatives are powerful candidates for high performance vibrational imaging.

Selective recognition between aromatics and aliphatics by cage-shaped borates supported by a machine learning approach.

Selective recognition between hydrocarbon moieties is a longstanding issue. Although we developed a π-pocket Lewis acid catalyst with high selectivity for aromatic aldehydes over aliphatic ones, a general strategy for catalyst design remains elusive. As an approach that transfers the molecular recognition based on multiple cooperative non-covalent interactions within the π-pocket to a rational catalyst design, herein, we demonstrate Lewis acid catalysts showing improved selectivity through the support of an ensemble algorithm with random forest, Ada Boost, and XG Boost as a machine learning (ML) approach. Using 7963 explanatory variables extracted from model hetero-Diels-Alder reactions, the ensemble algorithm predicted the chemoselectivity of unlearned catalysts. Experiments confirmed the prediction. The proposed catalyst shows the highest selective recognition, reminiscing enzymatic catalytic activity. Additionally, a SHapley Additive exPlanations (SHAP) method suggested that the selectivity originates from the polarizability and three-dimensional size of the catalyst. This insight leads to rational design guidelines for Lewis acid catalysts with dispersion forces.

Developing a GNN-based AI model to predict mitochondrial toxicity using the bagging method.

Mitochondrial toxicity has been implicated in the development of various toxicities, including hepatotoxicity. Therefore, mitochondrial toxicity has become a major screening factor in the early discovery phase of drug development. Several models have been developed to predict mitochondrial toxicity based on chemical structures. However, they only provide a binary classification of positive or negative results and do not provide the substructures that contribute to a positive decision. Therefore, we developed an artificial intelligence (AI) model to predict mitochondrial toxicity and visualize structural alerts. To construct the model, we used the open-source software library kMoL, which employs a graph neural network approach that allows learning from chemical structure data. We also utilized the integrated gradient method, which enables the visualization of substructures that contribute to positive results. The dataset used to construct the AI model exhibited a significant imbalance, with significantly more negative than positive data. To address this, we employed the bagging method, which resulted in a model with high predictive performance, as evidenced by an F1 score of 0.839. This model can also be used to visualize substructures that contribute to mitochondrial toxicity using the integrated gradient method. Our AI model predicts mitochondrial toxicity based on chemical structures and may contribute to screening mitochondrial toxicity in the early stages of drug discovery.

Interpretable machine learning-based individual analysis of acute kidney injury in immune checkpoint inhibitor therapy.

BACKGROUND: Acute kidney injury (AKI) is a critical complication of immune checkpoint inhibitor therapy. Since the etiology of AKI in patients undergoing cancer therapy varies, clarifying underlying causes in individual cases is critical for optimal cancer treatment. Although it is essential to individually analyze immune checkpoint inhibitor-treated patients for underlying pathologies for each AKI episode, these analyses have not been realized. Herein, we aimed to individually clarify the underlying causes of AKI in immune checkpoint inhibitor-treated patients using a new clustering approach with Shapley Additive exPlanations (SHAP). METHODS: We developed a gradient-boosting decision tree-based machine learning model continuously predicting AKI within 7 days, using the medical records of 616 immune checkpoint inhibitor-treated patients. The temporal changes in individual predictive reasoning in AKI prediction models represented the key features contributing to each AKI prediction and clustered AKI patients based on the features with high predictive contribution quantified in time series by SHAP. We searched for common clinical backgrounds of AKI patients in each cluster, compared with annotation by three nephrologists. RESULTS: One hundred and twelve patients (18.2%) had at least one AKI episode. They were clustered per the key feature, and their SHAP value patterns, and the nephrologists assessed the clusters' clinical relevance. Receiver operating characteristic analysis revealed that the area under the curve was 0.880. Patients with AKI were categorized into four clusters with significant prognostic differences (p = 0.010). The leading causes of AKI for each cluster, such as hypovolemia, drug-related, and cancer cachexia, were all clinically interpretable, which conventional approaches cannot obtain. CONCLUSION: Our results suggest that the clustering method of individual predictive reasoning in machine learning models can be applied to infer clinically critical factors for developing each episode of AKI among patients with multiple AKI risk factors, such as immune checkpoint inhibitor-treated patients.

Modular Design Platform for Activatable Fluorescence Probes Targeting Carboxypeptidases Based on ProTide Chemistry.

Carboxypeptidases (CPs) are a family of hydrolases that cleave one or more amino acids from the C-terminal of peptides or proteins and play indispensable roles in various physiological and pathological processes. However, only a few highly activatable fluorescence probes for CPs have been reported, and there is a need for a flexibly tunable molecular design platform to afford a range of fluorescence probes for CPs for biological and medical research. Here, we focused on the unique activation mechanism of ProTide-based prodrugs and established a modular design platform for CP-targeting florescence probes based on ProTide chemistry. In this design, probe properties such as fluorescence emission wavelength, reactivity/stability, and target CP can be readily tuned and optimized by changing the four probe modules: the fluorophore, the substituent on the phosphorus atom, the linker amino acid at the P1 position, and the substrate amino acid at the P1' position. In particular, switching the linker amino acid at position P1 enabled us to precisely optimize the reactivity for target CPs. As a proof-of-concept, we constructed probes for carboxypeptidase M (CPM) and prostate-specific membrane antigen (also known as glutamate carboxypeptidase II). The developed probes were applicable for the imaging of CP activities in live cells and in clinical specimens from patients. This design strategy should be useful in studying CP-related biological and pathological phenomena.

VGAE-MCTS: A New Molecular Generative Model Combining the Variational Graph Auto-Encoder and Monte Carlo Tree Search.

Molecular generation is crucial for advancing drug discovery, materials science, and chemical exploration. It expedites the search for new drug candidates, facilitates tailored material creation, and enhances our understanding of molecular diversity. By employing artificial intelligence techniques such as molecular generative models based on molecular graphs, researchers have tackled the challenge of identifying efficient molecules with desired properties. Here, we propose a new molecular generative model combining a graph-based deep neural network and a reinforcement learning technique. We evaluated the validity, novelty, and optimized physicochemical properties of the generated molecules. Importantly, the model explored uncharted regions of chemical space, allowing for the efficient discovery and design of new molecules. This innovative approach has considerable potential to revolutionize drug discovery, materials science, and chemical research for accelerating scientific innovation. By leveraging advanced techniques and exploring previously unexplored chemical spaces, this study offers promising prospects for the efficient discovery and design of new molecules in the field of drug development.

Network-based prediction approach for cancer-specific driver missense mutations using a graph neural network.

BACKGROUND: In cancer genomic medicine, finding driver mutations involved in cancer development and tumor growth is crucial. Machine-learning methods to predict driver missense mutations have been developed because variants are frequently detected by genomic sequencing. However, even though the abnormalities in molecular networks are associated with cancer, many of these methods focus on individual variants and do not consider molecular networks. Here we propose a new network-based method, Net-DMPred, to predict driver missense mutations considering molecular networks. Net-DMPred consists of the graph part and the prediction part. In the graph part, molecular networks are learned by a graph neural network (GNN). The prediction part learns whether variants are driver variants using features of individual variants combined with the graph features learned in the graph part. RESULTS: Net-DMPred, which considers molecular networks, performed better than conventional methods. Furthermore, the prediction performance differed by the molecular network structure used in learning, suggesting that it is important to consider not only the local network related to cancer but also the large-scale network in living organisms. CONCLUSIONS: We propose a network-based machine learning method, Net-DMPred, for predicting cancer driver missense mutations. Our method enables us to consider the entire graph architecture representing the molecular network because it uses GNN. Net-DMPred is expected to detect driver mutations from a lot of missense mutations that are not known to be associated with cancer.

Individual health-disease phase diagrams for disease prevention based on machine learning.

Early disease detection and prevention methods based on effective interventions are gaining attention worldwide. Progress in precision medicine has revealed that substantial heterogeneity exists in health data at the individual level and that complex health factors are involved in chronic disease development. Machine-learning techniques have enabled precise personal-level disease prediction by capturing individual differences in multivariate data. However, it is challenging to identify what aspects should be improved for disease prevention based on future disease-onset prediction because of the complex relationships among multiple biomarkers. Here, we present a health-disease phase diagram (HDPD) that represents an individual's health state by visualizing the future-onset boundary values of multiple biomarkers that fluctuate early in the disease progression process. In HDPDs, future-onset predictions are represented by perturbing multiple biomarker values while accounting for dependencies among variables. We constructed HDPDs for 11 diseases using longitudinal health checkup cohort data of 3,238 individuals, comprising 3,215 measurement items and genetic data. The improvement of biomarker values to the non-onset region in HDPD remarkably prevented future disease onset in 7 out of 11 diseases. HDPDs can represent individual physiological states in the onset process and be used as intervention goals for disease prevention.

Improving Compound-Protein Interaction Prediction by Self-Training with Augmenting Negative Samples.

Identifying compound-protein interactions (CPIs) is crucial for drug discovery. Since experimentally validating CPIs is often time-consuming and costly, computational approaches are expected to facilitate the process. Rapid growths of available CPI databases have accelerated the development of many machine-learning methods for CPI predictions. However, their performance, particularly their generalizability against external data, often suffers from a data imbalance attributed to the lack of experimentally validated inactive (negative) samples. In this study, we developed a self-training method for augmenting both credible and informative negative samples to improve the performance of models impaired by data imbalances. The constructed model demonstrated higher performance than those constructed with other conventional methods for solving data imbalances, and the improvement was prominent for external datasets. Moreover, examination of the prediction score thresholds for pseudo-labeling during self-training revealed that augmenting the samples with ambiguous prediction scores is beneficial for constructing a model with high generalizability. The present study provides guidelines for improving CPI predictions on real-world data, thus facilitating drug discovery.

Super-resolution vibrational imaging based on photoswitchable Raman probe.

Super-resolution vibrational microscopy is promising to increase the degree of multiplexing of nanometer-scale biological imaging because of the narrower spectral linewidth of molecular vibration compared to fluorescence. However, current techniques of super-resolution vibrational microscopy suffer from various limitations including the need for cell fixation, high power loading, or complicated detection schemes. Here, we present reversible saturable optical Raman transitions (RESORT) microscopy, which overcomes these limitations by using photoswitchable stimulated Raman scattering (SRS). We first describe a bright photoswitchable Raman probe (DAE620) and validate its signal activation and depletion characteristics when exposed to low-power (microwatt level) continuous-wave laser light. By harnessing the SRS signal depletion of DAE620 through a donut-shaped beam, we demonstrate super-resolution vibrational imaging of mammalian cells with excellent chemical specificity and spatial resolution beyond the optical diffraction limit. Our results indicate RESORT microscopy to be an effective tool with high potential for multiplexed super-resolution imaging of live cells.

Activatable Raman Probes Utilizing Enzyme-Induced Aggregate Formation for Selective Ex Vivo Imaging.

Detecting multiple enzyme activities simultaneously with high spatial specificity is a promising strategy to investigate complex biological phenomena, and Raman imaging would be an excellent tool for this purpose due to its high multiplexing capabilities. We previously developed activatable Raman probes based on 9CN-pyronins, but specific visualization of cells with target enzyme activities proved difficult due to leakage of the hydrolysis products from the target cells after activation. Here, focusing on rhodol bearing a nitrile group at the position of 9 (9CN-rhodol), we established a novel mechanism for Raman signal activation based on a combination of aggregate formation (to increase local dye concentration) and the resonant Raman effect along with the bathochromic shift of the absorption, and utilized it to develop Raman probes. We selected the 9CN-rhodol derivative 9CN-JCR as offering a suitable combination of increased stimulated Raman scattering (SRS) signal intensity and high aggregate-forming ability, resulting in good retention in target cells after probe activation. By using isotope-edited 9CN-JCR-based probes, we could simultaneously detect ß-galactosidase, γ-glutamyl transpeptidase, and dipeptidyl peptidase-4 activities in live cultured cells and distinguish cell regions expressing target enzyme activity in Drosophila wing disc and fat body ex vivo.

Development of a GCN-based model to predict in vitro phototoxicity from the chemical structure and HOMO-LUMO gap.

The interaction between sunlight and drugs can lead to phototoxicity in patients who have received such drugs. Phototoxicity assessment is a regulatory requirement globally and one of the main toxicity screening steps in the early stages of drug discovery. An in silico-in vitro approach has been utilized mainly for toxicology assessments at these stages. Although several quantitative structure-activity relationship (QSAR) models for phototoxicity have been developed, in silico technology to evaluate phototoxicity has not been well established. In this study, we attempted to develop an artificial intelligence (AI) model to predict the in vitro Neutral Red Uptake Phototoxicity Test results from a chemical structure and its derived information. To accomplish this, we utilized an open-source software library, kMoL. kMoL employs a graph convolutional neural networks (GCN) approach, which allows it to learn the data for the specified chemical structure. kMoL also utilizes the integrated gradient (IG) method, enabling it to visually display the substructures contributing to any positive results. To construct this AI model, we used only the chemical structure as a basis, then added the descriptors and the HOMO-LUMO gap, which was obtained from quantum chemical calculations. As a result, the assortment of chemical structures and the HOMO-LUMO gap produced an AI model with high discrimination performance, and an F1 score of 0.857. Additionally, our AI model could visualize the substructures involved in phototoxicity using the IG method. Our AI model can be applied as a toxicity screening method and could enhance productivity in drug development.

Rapid imaging of thymoma and thymic carcinoma with a fluorogenic probe targeting γ-glutamyltranspeptidase.

In recent years, thoracoscopic and robotic surgical procedures have increasingly replaced median sternotomy for thymoma and thymic carcinoma. In cases of partial thymectomy, the prognosis is greatly improved by ensuring a sufficient margin from the tumor, and therefore intraoperative fluorescent imaging of the tumor is especially valuable in thoracoscopic and robotic surgery, where tactile information is not available. γ-Glutamyl hydroxymethyl rhodamine green (gGlu-HMRG) has been applied for fluorescence imaging of some types of tumors in the resected tissues, and here we aimed to examine its validity for the imaging of thymoma and thymic carcinoma. 22 patients with thymoma or thymic carcinoma who underwent surgery between February 2013 and January 2021 were included in the study. Ex vivo imaging of specimens was performed, and the sensitivity and specificity of gGlu-HMRG were 77.3% and 100%, respectively. Immunohistochemistry (IHC) staining was performed to confirm expression of gGlu-HMRG's target enzyme, γ-glutamyltranspeptidase (GGT). IHC revealed high GGT expression in thymoma and thymic carcinoma in contrast to absent or low expression in normal thymic parenchyma and fat tissue. These results suggest the utility of gGlu-HMRG as a fluorescence probe for intraoperative visualization of thymomas and thymic carcinomas.

9-Cyano-10-telluriumpyronin Derivatives as Red-light-activatable Raman Probes.

Photoactivatable fluorescence probes can track the dynamics of specific cells or biomolecules with high spatiotemporal resolution, but their broad absorption and emission peaks limit the number of wavelength windows that can be employed simultaneously. In contrast, the narrower peak width of Raman signals offers more scope for simultaneous discrimination of multiple targets, and therefore a palette of photoactivatable Raman probes would enable more comprehensive investigation of biological phenomena. Herein we report 9-cyano-10-telluriumpyronin (9CN-TeP) derivatives as photoactivatable Raman probes whose stimulated Raman scattering (SRS) intensity is enhanced by photooxidation of the tellurium atom. Modification to increase the stability of the oxidation product led to a julolidine-like derivative, 9CN-diMeJTeP, which is photo-oxidized at the tellurium atom by red light irradiation to afford a sufficiently stable oxidation product with strong electronic pre-resonance, resulting in a bathochromic shift of the absorption spectrum and increased SRS intensity.

Characterizing eye-gaze positions of people with severe motor dysfunction: Novel scoring metrics using eye-tracking and video analysis.

Nonverbal communication with people who have physical disabilities is difficult. Eye-tracking technologies have recently been developed and applied to help people with physical disabilities in their communication. However, the eye-gaze patterns of people with severe motor dysfunction (SMD) have not been analyzed in detail. To clarify characterization of people with SMD, we aimed to develop gaze position-based evaluation metrics and analyze detailed eye-gaze patterns of people with SMD. We developed two new scoring metrics: (1) saliency score based on three saliency maps-spectral residual (SR), fine grained (FG), and motion (Mo); and (2) the distance score, which represents to what extent people can chase an object in a video. The evaluation was performed on 102 participants, consisting of 35 subjects with profound intellectual and multiple disabilities (PIMD; SMD with IQ < 20), 19 with severe physical disabilities (SPD; SMD with IQ ≥ 20), and 48 healthy individuals. We observed that two saliency scores (SR and FG) and the distance score showed significant differences between the PIMD/SPD and healthy groups for the entire video, whereas Mo scores did not. Moreover, the distance score was analyzed separately for each scene, where scenes were categorized into three patterns-running, explanation, and hiding-according to the behavior of the moving objects. In the SPD and healthy groups, the explanation scenes accounted for the highest percentage of all scenes with the best distance score (63.6% and 61.9%, respectively), whereas in the PIMD group, the running scenes accounted for the highest percentage (54.5%). In conclusion, the new metrics were successful in quantitatively assessing the gaze responsiveness of people with SMD, which could not be assessed using a conventional metric, gaze-acquisition time. This study is expected to expand the possibilities of nonverbal communication using eye-tracking devices for people with SMD.

Measuring the pH of Acidic Vesicles in Live Cells with an Optimized Fluorescence Lifetime Imaging Probe.

Acidification of intracellular vesicles, such as endosomes and lysosomes, is a key pathway for regulating the function of internal proteins. Most conventional methods of measuring pH are not satisfactory for quantifying the pH inside these vesicles. Here, we investigated the molecular requirements for a fluorescence probe to measure the intravesicular acidic pH in living cells by means of fluorescence lifetime imaging microscopy (FLIM). The developed probe, m-DiMeNAF488, exhibits a pH-dependent equilibrium between highly fluorescent and moderately fluorescent forms, which has distinct and detectable fluorescence lifetimes of 4.36 and 0.58 ns, respectively. The pKa(τ) value of m-DiMeNAF488 was determined to be 4.58, which would be favorable for evaluating the pH in the acidic vesicles. We were able to monitor the pH changes in phagosomes during phagocytosis by means of FLIM using m-DiMeNAF488. This probe is expected to be a useful tool for investigating acidic pH-regulated biological phenomena.

Droplet Array-Based Platform for Parallel Optical Analysis of Dynamic Extracellular Vesicle Secretion from Single Cells.

Extracellular vesicles (EVs) are essential intercellular communication tools, but the regulatory mechanisms governing heterogeneous EV secretion are still unclear due to the lack of methods for precise analysis. Monitoring the dynamics of secretion from individually isolated cells is crucial because in bulk analysis, secretion activity can be perturbed by cell-cell interactions, and a cell population rarely performs secretion in a magnitude- or duration-synchronized manner. Although various microfluidic techniques have been adopted to evaluate the abundance of single-cell-derived EVs, none can track their secretion dynamics continually for extended periods. Here, we have developed a droplet array-based method that allowed us to optically quantify the EV secretion dynamics of >300 single cells every 2 h for 36 h, which covers the cell doubling time of many cell types. The experimental results clearly show the highly heterogeneous nature of single-cell EV secretion and suggest that cell division facilitates EV secretion, showing the usefulness of this platform for discovering EV regulation machinery.

Rapid imaging of lung cancer using a red fluorescent probe to detect dipeptidyl peptidase 4 and puromycin-sensitive aminopeptidase activities.

Rapid identification of lung-cancer micro-lesions is becoming increasingly important to improve the outcome of surgery by accurately defining the tumor/normal tissue margins and detecting tiny tumors, especially for patients with low lung function and early-stage cancer. The purpose of this study is to select and validate the best red fluorescent probe for rapid diagnosis of lung cancer by screening a library of 400 red fluorescent probes based on 2-methyl silicon rhodamine (2MeSiR) as the fluorescent scaffold, as well as to identify the target enzymes that activate the selected probe, and to confirm their expression in cancer cells. The selected probe, glutamine-alanine-2-methyl silicon rhodamine (QA-2MeSiR), showed 96.3% sensitivity and 85.2% specificity for visualization of lung cancer in surgically resected specimens within 10 min. In order to further reduce the background fluorescence while retaining the same side-chain structure, we modified QA-2MeSiR to obtain glutamine-alanine-2-methoxy silicon rhodamine (QA-2OMeSiR). This probe rapidly visualized even borderline lesions. Dipeptidyl peptidase 4 and puromycin-sensitive aminopeptidase were identified as enzymes mediating the cleavage and consequent fluorescence activation of QA-2OMeSiR, and it was confirmed that both enzymes are expressed in lung cancer. QA-2OMeSiR is a promising candidate for clinical application.

Development of a fluorescent probe library enabling efficient screening of tumour-imaging probes based on discovery of biomarker enzymatic activities.

Fluorescent probes that can selectively detect tumour lesions have great potential for fluorescence imaging-guided surgery. Here, we established a library-based approach for efficient screening of probes for tumour-selective imaging based on discovery of biomarker enzymes. We constructed a combinatorial fluorescent probe library for aminopeptidases and proteases, which is composed of 380 probes with various substrate moieties. Using this probe library, we performed lysate-based in vitro screening and/or direct imaging-based ex vivo screening of freshly resected clinical specimens from lung or gastric cancer patients, and found promising probes for tumour-selective visualization. Further, we identified two target enzymes as novel biomarker enzymes for discriminating between tumour and non-tumour tissues. This library-based approach is expected to be an efficient tool to develop tumour-imaging probes and to discover new biomarker enzyme activities for various tumours and other diseases.

AI-Driven Synthetic Route Design Incorporated with Retrosynthesis Knowledge.

Computer-aided synthesis planning (CASP) aims to assist chemists in performing retrosynthetic analysis for which they utilize their experiments, intuition, and knowledge. Recent breakthroughs in machine learning (ML) techniques, including deep neural networks, have significantly improved data-driven synthetic route designs without human intervention. However, learning chemical knowledge by ML for practical synthesis planning has not yet been adequately achieved and remains a challenging problem. In this study, we developed a data-driven CASP application integrated with various portions of retrosynthesis knowledge called "ReTReK" that introduces the knowledge as adjustable parameters into the evaluation of promising search directions. The experimental results showed that ReTReK successfully searched synthetic routes based on the specified retrosynthesis knowledge, indicating that the synthetic routes searched with the knowledge were preferred to those without the knowledge. The concept of integrating retrosynthesis knowledge as adjustable parameters into a data-driven CASP application is expected to enhance the performance of both existing data-driven CASP applications and those under development.