Drug–targeted therapies are promising approaches to treating tumors, and research on receptor–ligand interactions for discovering high-affinity targeted drugs has been accelerating drug development. This study presents a mechanism-driven deep learning-based computational model to learn double drug sequences, protein sequences, and drug graphs to project drug–target affinities (DTAs), which was termed the DoubleSG-DTA. We deployed lightweight graph isomorphism networks to aggregate drug graph representations and discriminate between molecular structures, and stacked multilayer squeeze-and-excitation networks to selectively enhance spatial features of drug and protein sequences. What is more, cross-multi-head attentions were constructed to further model the non-covalent molecular docking behavior. The multiple cross-validation experimental evaluations on various datasets indicated that DoubleSG-DTA consistently outperformed all previously reported works. To showcase the value of DoubleSG-DTA, we applied it to generate promising hit compounds of Non-Small Cell Lung Cancer harboring $EGF{R}^{T790M}$ mutation from natural products, which were consistent with reported laboratory studies. Afterward, we further investigated the interpretability of the graph-based “black box” model and highlighted the active structures that contributed the most. DoubleSG-DTA thus provides a powerful and interpretable framework that extrapolates for potential chemicals to modulate the systemic response to disease.

MicroRNA is expected to be a novel therapeutic tool for tumors. Gap junctions facilitate the transfer of microRNA, which exerts biological effects on tumor cells. However, the length of microRNA that can pass through certain gap junctions composed of specific connexin remains unknown. To address this question, the present study investigated the permeability of gap junctions composed of various connexins, including connexin 43, connexin 32 or connexin 37, to microRNAs consisting of 18‐27 nucleotides in glioma cells and cervical cancer cells. Results indicated that all of the microRNAs were able to be transferred from donor glioma cells to neighboring cells through the connexin 43 composed gap junction, but not the gap junctions composed of connexin 32 or connexin 37, in cervical cancer cells. Downregulation of the function of gap junctions comprising connexin 43 by pharmacological inhibition and shRNA significantly decreased the transfer of these microRNAs. In contrast, gap junction enhancers and overexpression of connexin 43 effectively increased these transfers. In glioma cells, cell proliferation was inhibited by microRNA‐34a. Additionally, these effects of microRNA‐34a were significantly enhanced by overexpression of connexin 43 in U251 cells, indicating that gap junctions play an important role in the antitumor effect of microRNA by transfer of microRNA to neighboring cells. Our data are the first to clarify the pattern of microRNA transmission through gap junctions and provide novel insights to show that antitumor microRNAs should be combined with connexin 43 or a connexin 43 enhancer, not connexin 32 or connexin 37, in order to improve the therapeutic effect.

Background: Anti-N-methyl-D-aspartate receptor (NMDAR) encephalitis is a severe autoimmune neuropsychiatric disease. Brain access of anti-NMDAR autoantibody through the blood–brain barrier (BBB) is essential for pathogenesis. Most previous animal models limit the investigation of etiologies of BBB damage in patients. Methods: In this study, we established a novel humanized mouse model of anti-NMDAR encephalitis by intraperitoneal injection of patients’ peripheral blood mononuclear cells (PBMCs) into BALB/c Rag2^−/−Il2rg^−/−Sirpα^NODFlk2^−/− mice. Results: We found that engraftment of patients’ PBMCs not only produced potent anti-GluN1 autoantibodies, but also disrupted BBB integrity to allow brain access of autoantibodies, resulting in a hyperactive locomotor phenotype, anxiety- and depressive-like behaviors, cognitive deficits, as well as functional changes in corresponding brain regions. Transcriptome analysis suggested an exaggerated immune response and impaired neurotransmission in the mouse model and highlighted Il-1β as a hub gene implicated in pathological changes. We further demonstrated that Il-1β was produced by endothelial cells and disrupted BBB by repressing tight junction proteins. Treatment with Anakinra, an Il-1 receptor antagonist, ameliorated BBB damage and neuropsychiatric behaviors. Conclusions: Our study provided a novel and clinically more relevant humanized mouse model of anti-NMDAR encephalitis and revealed an intrinsic pathogenic property of the patient’s lymphocytes.

Abnormal expression or distribution of connexin 32 (Cx32) is associated with hepatocarcinogenesis, but the role of Cx32 and the underlying mechanisms are still unclear. The expression level of Cx32 in 96 hepatocellular carcinoma (HCC) specimens was determined using western blotting and immunohistochemistry. The correlation between Cx32 expression and clinicopathological parameters was analyzed. The cell apoptosis rate was examined using flow cytometry and western blotting. The role of Cx32 in the Src kinase and epidermal growth factor receptor (EGFR) signaling pathways was measured by quantitative real-time PCR, western blotting and coimmunoprecipitation (CO-IP). The effect of Cx32 overexpression on the streptonigrin (SN)-induced tumor growth suppression and apoptosis was assessed in nude mice. Our study showed that overexpressed Cx32 accumulated in the cytoplasm and that Cx32-containing gap junctions (GJs) were nearly absent in HCC specimens. Upregulated Cx32 expression was highly correlated with advanced tumor-node-metastasis (TNM) stage and poor tumor differentiation and was an independent predictive marker for poor prognosis in HCC. Overexpression of Cx32 significantly inhibited SN-induced apoptosis by activating the EGFR signaling pathway in vitro and in vivo. Moreover, the expression levels of Cx32 and EGFR were positively correlated in HCC specimens. The CO-IP experiments demonstrated that Cx32 could bind to Src kinase, and the western blotting results revealed that Cx32 increased the levels of EGFR and p-EGFR by upregulating Src expression. The present study demonstrated that overexpressed and internalized Cx32 was associated with advanced TNM stage and poor tumor differentiation and predicted poor prognosis in HCC. Cx32 facilitated HCC progression by blocking chemotherapy-induced apoptosis in vitro and in vivo via interacting with Src and thus promoting the phosphorylation of EGFR, subsequently activating the EGFR signaling pathway. Cx32 may be a potential biomarker and a new therapeutic target for HCC.

Necroptosis is an alternative form of programmed cell death that generally occurs under apoptosis-deficient conditions. Our previous work showed that connexin32 promotes the malignant progress of hepatocellular carcinoma by enhancing the ability of resisting apoptosis in vivo and in vitro. Whether triggering necroptosis is a promising strategy to eliminate the apoptosis-resistance hepatocellular carcinoma cells with high connexin32 expression remains unknown. In this study, we found that connexin32 expression was positively correlated with the expression of necroptosis protein biomarkers in human hepatocellular carcinoma specimens, cell lines and xenograft model. Treatment with shikonin, a well-used necroptosis inducer, markedly caused necroptosis in hepatocellular carinoma cells. Interestingly, overexpressed connexin32 exacerbated shikonin-induced necroptosis, but downregulation of connexin32 alleviated necroptosis in vitro and in vivo. Mechanistically, Cx32 was demonstrated to bind to Src and promoted Src-mediated caspase 8 phosphorylation and inactivation, which ultimately reducing the activated caspase 8-mediated proteolysis of receptor-interacting serine-threonine protein kinase 1/3, the key molecular for necroptosis activation. In conclusion, we demonstrated that Cx32 contributed to the activation of necroptosis in hepatocellular carcinoma cells via binding to Src and then mediating the inactivation of caspase 8. The present study suggested that necroptosis inducers may be more favorable than apoptosis inducers to eliminate hepatocellular carcinoma cells with high expression of connexin32.