Self-supervised Pre-training via Multi-view Graph Information Bottleneck for Molecular Property Prediction.

Molecular representation learning has remarkably accelerated the development of drug analysis and discovery. It implements machine learning methods to encode molecule embeddings for diverse downstream drug-related tasks. Due to the scarcity of labeled molecular data, self-supervised molecular pre-training is promising as it can handle large-scale unlabeled molecular data to prompt representation learning. Although many universal graph pre-training methods have been successfully introduced into molecular learning, there still exist some limitations. Many graph augmentation methods, such as atom deletion and bond perturbation, tend to destroy the intrinsic properties and connections of molecules. In addition, identifying subgraphs that are important to specific chemical properties is also challenging for molecular learning. To address these limitations, we propose the self-supervised Molecular Graph Information Bottleneck (MGIB) model for molecular pre-training. MGIB observes molecular graphs from the atom view and the motif view, deploys a learnable graph compression process to extract the core subgraphs, and extends the graph information bottleneck into the self-supervised molecular pre-training framework. Model analysis validates the contribution of the self-supervised graph information bottleneck and illustrates the interpretability of MGIB through the extracted subgraphs. Extensive experiments involving molecular property prediction, including 7 binary classification tasks and 6 regression tasks demonstrate the effectiveness and superiority of our proposed MGIB.

Hierarchical Molecular Graph Self-Supervised Learning for property prediction.

Molecular graph representation learning has shown considerable strength in molecular analysis and drug discovery. Due to the difficulty of obtaining molecular property labels, pre-training models based on self-supervised learning has become increasingly popular in molecular representation learning. Notably, Graph Neural Networks (GNN) are employed as the backbones to encode implicit representations of molecules in most existing works. However, vanilla GNN encoders ignore chemical structural information and functions implied in molecular motifs, and obtaining the graph-level representation via the READOUT function hinders the interaction of graph and node representations. In this paper, we propose Hierarchical Molecular Graph Self-supervised Learning (HiMol), which introduces a pre-training framework to learn molecule representation for property prediction. First, we present a Hierarchical Molecular Graph Neural Network (HMGNN), which encodes motif structure and extracts node-motif-graph hierarchical molecular representations. Then, we introduce Multi-level Self-supervised Pre-training (MSP), in which corresponding multi-level generative and predictive tasks are designed as self-supervised signals of HiMol model. Finally, superior molecular property prediction results on both classification and regression tasks demonstrate the effectiveness of HiMol. Moreover, the visualization performance in the downstream dataset shows that the molecule representations learned by HiMol can capture chemical semantic information and properties.

Fermentation of soymilk by Lactobacillus acidipiscis isolated from Chinese stinky tofu capable of efficiently biotransforming isoflavone glucosides to dihydrodaidzein and dihydrogenistein.

BACKGROUND: The soy isoflavone microbial metabolites dihydrodaidzein (DHD), dihydrogenistein (DHG), equol and 5-hydroxy-equol are generally more biologically active than their precursors daidzein and genistein. Bacteria responsible for isoflavone metabolism have been isolated and identified. Fermented soymilk is a potential functional food; however, there are few lactic acid bacteria capable of metabolizing soy isoflavones. RESULTS: A newly isolated Gram-positive facultative anaerobic bacterium, which was named Lactobacillus acidipiscis HAU-FR7, was isolated from the traditional Chinese fermented soy product 'stinky tofu'. Bacterium strain HAU-FR7 can grow under aerobic conditions and can also convert most of the daidzin and genistin in soymilk into DHD and DHG, respectively. The concentrations of DHD and DHG produced were 183 and 134 µmol L-1 , respectively, after fermentation for 24 h. Strain HAU-FR7 does not produce the biogenic amines cadaverine, putrescine, histamine or tyramine, and an antibiotic susceptibility test showed that HAU-FR7 is sensitive to nine of the ten tested antibiotics, except for vancomycin. Moreover, the 1,1-diphenyl-2- picrylhydrazyl free radical scavenging capacity of soymilk fermented with HAU-FR7 was significantly higher than that of unfermented soymilk. CONCLUSION: A facultative anaerobic lactic acid bacterium, designated Lactobacillus acidipiscis HAU-FR7, is capable of reducing the soy isoflavone glucosides daidzin and genistin in soymilk to DHD and DHG efficiently, even in the presence of atmospheric oxygen. The biotransformation activity of HAU-FR7 grown in soymilk is higher than that in de Man-Rogosa-Sharpe liquid culture medium. © 2022 Society of Chemical Industry.

Immunosuppressive effects of dimethyl fumarate on dendritic cell maturation and migration: a potent protector for coronary heart disease.

Dendritic cells (DCs), as a bridge between innate and adaptive immunity, play key roles in atherogenesis, particularly in plaque rupture, the underlying pathophysiologic cause of myocardial infarction. Targeting DC functions, including maturation and migration to atherosclerotic plaques, may be a novel therapeutic approach to atherosclerotic disease. Dimethyl fumarate (DMF), an agent consisting of a combination of fumaric acid esters, in current study were found to be able to suppress DC maturation by reducing the expression of costimulatory molecules and MHC class II and by blocking cytokine secretion. In addition, DMF efficiently inhibited the migration of activated DCs in vitro and in vivo by reducing the expression of chemokine receptor 7 (CCR7). Additionally, DMF efficiently inhibited the expression of the costimulatory molecule CD86, as well as the chemokine receptor CCR7 and the C-X-C motif chemokine receptor 4 (CXCR4), in healthy donor-derived purified DCs that had been stimulated by ST-segment elevation myocardial infarction (STEMI) patient serum. This study points to the potent therapeutic value of DMF for protecting against cardiovascular disease by suppressing DC functions.

MicroRNA­140 suppresses Helicobacter pylori­positive gastric cancer growth by enhancing the antitumor immune response.

Immune checkpoint blockade is a promising therapeutic strategy against various human malignancies. MicroRNAs (miRNAs/miRs) regulate gene expression, by repressing mRNA translation or promoting its degradation. The aim of the current study was to investigate the role and molecular mechanisms of miR­140 in Helicobacter pylori (Hp)­associated gastric cancer, and to examine its relationship with immune function in gastric cancer. Gastritis tissue samples from gastritis patients, and gastric cancer tissue samples from gastric cancer patients were collected for miR­140 expression detection. miR­140 expression was detected using reverse transcription­quantitative polymerase chain reaction, and protein expression was measured by western blotting. TargetScan and dual luciferase reporter assays were used to reveal the association between miR­140 and programmed cell death­ligand 1 (PD­L1). BGC823 cell proliferation was detected by MTT assays. Ex vivo immune analysis by flow cytometry and ELISA were used to analyze immune function. It was demonstrated that miR­140 expression was significantly reduced in Hp­positive gastric cancer. PD­L1 was confirmed as a direct target of miR­140 in gastric cancer cells. In addition, PD­L1 expression was significantly increased in Hp­positive gastric cancer. Overexpression of miR­140 significantly suppressed gastric cancer cell proliferation through regulating PD­L1 expression. In vivo experiments also revealed that miR­140 markedly repressed tumor growth in the C57BL/6 mice. Furthermore, it was determined that the tumor­suppressive role of miR­140 in gastric cancer was associated with increased cytotoxic CD8+ T cell and reduced myeloid­derived suppressive and regulatory T cell infiltration. miR­140 significantly prevented mammalian target of rapamycin signaling in gastric cancer cells. Notably, these miR­140 overexpression­induced alterations were inhibited by PD­L1 plasmid. These findings indicated that miR­140 exerted an anti­gastric cancer effect by targeting immune checkpoint molecule PD­L1. Thus, miR­140 may be a promising and novel immunotherapeutic target for gastric cancer treatment.

Evaluation and accurate diagnoses of pediatric diseases using artificial intelligence.

Artificial intelligence (AI)-based methods have emerged as powerful tools to transform medical care. Although machine learning classifiers (MLCs) have already demonstrated strong performance in image-based diagnoses, analysis of diverse and massive electronic health record (EHR) data remains challenging. Here, we show that MLCs can query EHRs in a manner similar to the hypothetico-deductive reasoning used by physicians and unearth associations that previous statistical methods have not found. Our model applies an automated natural language processing system using deep learning techniques to extract clinically relevant information from EHRs. In total, 101.6 million data points from 1,362,559 pediatric patient visits presenting to a major referral center were analyzed to train and validate the framework. Our model demonstrates high diagnostic accuracy across multiple organ systems and is comparable to experienced pediatricians in diagnosing common childhood diseases. Our study provides a proof of concept for implementing an AI-based system as a means to aid physicians in tackling large amounts of data, augmenting diagnostic evaluations, and to provide clinical decision support in cases of diagnostic uncertainty or complexity. Although this impact may be most evident in areas where healthcare providers are in relative shortage, the benefits of such an AI system are likely to be universal.

Development and Validation of Nomogram to Predict Acute Kidney Injury in Patients with Acute Myocardial Infarction Treated Invasively.

To identify patients who are likely to develop contrast-induced acute kidney injury (CI-AKI) in patients with acute myocardial infarction (AMI), a nomogram was developed in AMI patients. Totally 920 patients with AMI were enrolled in our study. The discrimination and calibration of the model were validated. External validations were also carried out in a cohort of 386 AMI patients. Our results showed in the 920 eligible AMI patients, 114 patients (21.3%) developed CI-AKI in the derivation group (n = 534), while in the validation set (n = 386), 50 patients (13%) developed CI-AKI. CI-AKI model included the following six predictors: hemoglobin, contrast volume >100 ml, hypotension before procedure, eGFR, log BNP, and age. The area under the curve (AUC) was 0.775 (95% confidence interval [CI]: 0.732-0.819) in the derivation group and 0.715 (95% CI: 0.631-0.799) in the validation group. The Hosmer-Lemeshow test has a p value of 0.557, which confirms the model's goodness of fit. The AUC of the Mehran risk score was 0.556 (95% CI: 0.498-0.615) in the derivation group. The validated nomogram provided a useful predictive value for CI-AKI in patients with AMI.

Growth differentiation factor-15 levels and the risk of contrast induced nephropathy in patients with acute myocardial infarction undergoing percutaneous coronary intervention: A retrospective observation study.

AIMS: To investigate the association between growth differentiation factor-15 (GDF-15) and contrast-induced nephropathy (CIN) in patients with acute myocardial infarction (AMI) undergoing percutaneous coronary intervention (PCI). METHODS: A total of 311 patients with AMI were studied retrospectively. All patients were divided into two groups according to the occurrence of CIN after PCI. Baseline clinical data were compared between two groups. Multivariate logistic regression analysis was used to identify the risk factors for CIN. Cox regression analysis was used to identify the association between GDF-15, CIN and short-term outcome. RESULTS: There were 80 patients in CIN group (average age was 71.60 ± 13.00 years; 67.5% male) and 231 patients in non-CIN group (average age was 63.80 ± 11.70 years; 71.9%male). The concentration of GDF-15 in CIN group was higher than that of non-CIN group (1232 ± 366.6 ng/L vs. 939.20 ± 309.6 ng/L, P <0.001). According to GDF-15 quartiles, patients were divided into four groups. Multivariate logistic model indicated that the highest quartile(Q4) was significantly associated with an increased risk of CIN compared with lower level of GDF-15 (Q1, Q2 and Q3) (OR : 3.572, 1.803-7.078, P < 0.001). Of 243 patients who could calculate the ACEF risk score, area under the curve (AUC) of GDF-15 was 0.793, 95%CI: 0.729-0.856, P < 0.001, while AUC of ACEF was 0.708, 95%CI: 0.630-0.786, P < 0.001. Using 10% and 30% as arbitrary thresholds to define patients at low, intermediate, and high risk, GDF-15 achieved a net reclassification improvement (NRI) of 0.32 (95%CI: 0.123-0.518, P = 0.001) compared with the ACEF risk score. Cox regression model showed that high concentration of GDF-15 (Q4) was significantly associated with an increased risk of all-cause mortality and major adverse clinical events (MACE) (HR: 8.434, 95%CI: 2.650-26.837, P <0.001; HR: 3.562, 95%CI: 1.658-7.652, P = 0.001) compared with low level of GDF-15 (Q1, Q2 and Q3). CIN was an independent predictor of all-cause mortality and MACE in AMI patients (HR: 3.535, 95%CI: 1.135-11.005, P = 0.029; HR: 5.154, 95%CI: 2.228-11.925, P <0.001). CONCLUSION: GDF-15 levels increased in CIN group in AMI patients underwent PCI. GDF-15 was an independent risk factor for CIN in AMI patients underwent PCI. GDF-15 level and CIN are independent risk factors for all-cause mortality and MACE in short-term follow-ups.

Growth differentiation factor-15 levels and the risk of contrast induced acute kidney injury in acute myocardial infarction patients treated invasively: A propensity-score match analysis.

BACKGROUND: Growth differentiation factor-15 (GDF-15) is an emerging biomarker for risk stratification in cardiovascular disease. Contrast-induced acute kidney injury (AKI) is an important complication in patients undergoing coronary angiography (CAG) or percutaneous coronary intervention (PCI). In this retrospectively observational study, we aimed to determine the role of GDF-15 and the risk of AKI in acute myocardial infarction (AMI) patients. METHODS: The medical records of 1195 patients with AMI were reviewed. After exclusion criteria, a total of 751 eligible patients who underwent CAG or PCI were studied. Preoperative clinical parameters including GDF-15 levels were recorded. Multivariate logistic regression analysis was used to identify the risk factors of AKI. Subsequently, to reduce a potential selection bias and to balance differences between the two groups, a propensity score-matched analysis was performed. We recorded the 30-day all-cause mortality of the total study population. Kaplan-Meier analysis was performed to identify the association between short term survival in AMI patients and GDF-15 level. RESULTS: Among 751 enrolled patients, 106 patients (14.1%) developed AKI. Patients were divided into two groups: AKI group (n = 106) and non-AKI group (n = 645). GDF-15 levels were significantly higher in AKI group compared to non-AKI group (1328.2 ± 349.7 ng/L vs. 1113.0 ± 371.3 ng/L, P <0.001). Multivariate logistic regression analyses showed GDF-15 was an independent risk factor of AKI (per 1000 ng/L increase of GDF-15, OR: 3.740, 95% CI: 1.940-7.207, P < 0.001). According to GDF-15 tertiles, patients were divided into three groups. Patients in middle (OR 2.93, 95% CI: 1.46-5.89, P = 0.003) and highest GDF-15 tertile (OR 3.72, 95% CI: 1.87-7.39, P <0.001) had higher risk of AKI compared to patients in the lowest GDF-15 tertile. The propensity score-matched group set comprised of 212 patients. Multivariate logistic regression revealed that GDF-15 is still an independent risk factor for AKI after matching (per 1000 ng/L increase of GDF-15, OR: 2.395, 95% CI: 1.020-5.626, P = 0.045). Based on the Kaplan-Meier analysis, the risk of 30-day all-cause mortality increased in higher GDF-15 tertiles log rank chi-square: 29.895, P <0.001). CONCLUSION: This suggests that preoperative plasma GDF-15 is an independent risk factor of AKI in AMI patients underwent CAG or PCI. GDF-15 and AKI are associated with poor short term survival of AMI patients.

Polyprenols mitigate cognitive dysfunction and neuropathology in the APP/PS1 mouse.

Alzheimer's disease (AD) is a very common neurodegenerative disorder in the elderly and brings considerable financial and social problems worldwide. In this study, polyprenols were firstly evaluated the effects on the cognitive deficits and neuropathology in APP/PS1 mice model of AD. At 3 months old, the APP/PS1 mice were divided into model group; polyprenols low, middle, and high dosage group; and positive drug group. Age-matched wild-type mice were chosen in control group. The administration by oral gavage lasted 6 months. Polyprenols treatment significantly improved cognitive impairment of double transgenic mice compared with vehicle control treatment in behavioral tests. In addition, immunohistochemistry and enzyme-linked immunosorbent assay showed that there were significantly reductions in neuritic plaques and the level of hyperphosphorylated tau in brain of polyprenols-treated mice. Furthermore, we found that polyprenols treatment reduced the apoptotic cells in brain sections of 9-month-old APP/PS1 mice. These results reveal that polyprenols exert neuroprotective effects in APP/PS1 mice and could represent an effective treatment for AD.

Melatonin ameliorates Aß1-42 -induced Alzheimer's cognitive deficits in mouse model.

OBJECTIVES: The objective of this study was to evaluate whether melatonin could ameliorate cognitive function in Aß1-42 -induced mouse model and its underlying mechanisms. METHODS: Series behaviour tests were performed to demonstrate the amelioration of cognitive function of the Alzheimer's disease (AD) mice induced by Aß1-42 . Additionally, enzyme-linked immunosorbent assay was applied to detect the expression of Aß1-42 , BACE1 and p-tau protein in the brain of the AD mice. JC-1 was performed to investigate the role in alleviating mitochondrial damage by melatonin in vitro. Western blot was used to detect the expression of melatonin on apoptosis-related factors caspase-3 and Bcl-2, as well as the expressions of GSK-3ß and PP2A to further determine the mechanisms of melatonin on the expression of p-tau protein. KEY FINDINGS: Melatonin significantly ameliorated the cognitive function and mitochondrial damage in AD mice, reduced the expression levels of GSK-3ß, caspase-3, Aß1-42 , BACE1, p-tau protein and increased the expressions of PP2A and Bcl-2. CONCLUSION: From the overall results, we concluded that melatonin alleviated the mitochondrial damage effectively and decreased the expressions of the p-tau and some key proteins of apoptosis, leading to the improvement of cognitive function of the mice induced by Aß1-42 .

The ameliorative effects and underlying mechanisms of dopamine D1-like receptor agonist SKF38393 on Aß1-42-induced cognitive impairment.

Alzheimer's disease (AD) is an age-related neurodegenerative disease characterized by extracellular amyloid plaques and intracellular neurofibrillary tangles. It is the most common form of human cognitive decline and dementia. In this study, we aim to systematically investigate the ameliorative effects of dopamine D1-like receptor agonist SKF38393 on cognitive dysfunction and explore its underlying mechanisms. The Aß1-42 was injected intracerebroventricularly to establish cognitive disorder model. Then, a series of behavior tests were used. In order to further study the mechanisms, some relevant protein was assessed by ELISA method and Western blot. The results in behavior tests revealed that SKF38393 significantly ameliorated all the test indexes compared with the model mice. Then SKF38393 increased phosphorylation of cAMP response element binding protein (CREB) and expression of Bcl-2 in Western blot analyses. Furthermore, in ELISA assay, SKF38393 significantly increased the brain-derived neurotrophic factor (BDNF) levels and reduced the ß-site APP cleaving enzyme1 (BACE1) and Aß1-42 levels in hippocampus and cortex of mice. However, compared with SKF38393-H, all these results were significantly reversed by the dopamine D1 receptor antagonist SCH23390. These results indicated that SKF38393 could ameliorate Aß1-42-induced cognitive dysfunction in mice, which may be related to D1 receptor activation. It leads to the phosphorylation of CREB, which promote the expression of BDNF, Bcl-2 and decrease the expression of Aß1-42 of mice. Our findings suggest that dopamine D1-like receptor may be a potential target for the treatment of AD and its agonists may become a novel drug in the future.

The clinical implications of increased cyclophilin A levels in patients with acute coronary syndromes.

BACKGROUND: Cyclophilin A is a secreted molecule that has a physiological and pathological role in cardiovascular diseases. However, limited information is available on the relationship between cyclophilin A concentration and acute coronary syndromes (ACS). We investigated whether cyclophilin A concentration is related to the stability of coronary atherosclerotic plaque in patients with ACS. METHODS: This study included normal controls (n=50), patients with stable angina (SA) (n=60) and patients with ACS, including unstable angina (UA) (n=60) and acute myocardial infarction (AMI) (n=90). Serum soluble cyclophilin A, matrix metalloproteinase 9 (MMP-9), MMP-3 and C-reactive protein concentrations (CRP) were measured. All coronary stenosis were assessed by angiographic coronary stenosis morphology. RESULTS: Serum cyclophilin A concentration in ACS (UA and AMI ) subjects were significantly higher than those in patients with SA and controls (p<0.05). Serum cyclophilin A correlated positively with serum MMP-3 and MMP-9 and CRP in ACS patients(r(1)=0.69, r(2)=0.52, r(3)=0.49 p<0.0001), but not in control. Furthermore, the increased cyclophilin A concentrations was associated with the number of complex coronary stenoses (r(1)=0.63, p<0.0001), but not smooth lesions or stenosis severity, in coronary artery disease patients. Logistic regression analysis also demonstrated that serum cyclophilin A concentration was an independent predictor factor for ACS( OR, 2.721, 95% CI 1.563-4.042, p=0.001). CONCLUSION: Patients with ACS showed that increased concentrations of cyclophilin A may be a valuable marker for predicting the severity of ACS.