In the context of science, the well-known adage “a picture is worth a thousand words” might well be “a model is worth a thousand datasets.” Scientific models, such as Newtonian physics or biological gene regulatory networks, are human-driven simplifications of complex phenomena that serve as surrogates for the countless experiments that validated the models. Recently, machine learning has been able to overcome the inaccuracies of approximate modeling by directly learning the entire set of nonlinear interactions from data. However, without any predetermined structure from the scientific basis behind the problem, machine learning approaches are flexible but data-expensive, requiring large databases of homogeneous labeled training data. A central challenge is reconciling data that is at odds with simplified models without requiring "big data". In this work demonstrate how a mathematical object, which we denote universal differential equations (UDEs), can be utilized as a theoretical underpinning to a diverse array of problems in scientific machine learning to yield efficient algorithms and generalized approaches. The UDE model augments scientific models with machine-learnable structures for scientifically-based learning. We show how UDEs can be utilized to discover previously unknown governing equations, accurately extrapolate beyond the original data, and accelerate model simulation, all in a time and data-efficient manner. This advance is coupled with open-source software that allows for training UDEs which incorporate physical constraints, delayed interactions, implicitly-defined events, and intrinsic stochasticity in the model. Our examples show how a diverse set of computationally-difficult modeling issues across scientific disciplines, from automatically discovering biological mechanisms to accelerating the training of physics-informed neural networks and large-eddy simulations, can all be transformed into UDE training problems that are efficiently solved by a single software methodology.

In the context of science, the well-known adage "a picture is worth a thousand words" might well be "a model is worth a thousand datasets." Scientific models, such as Newtonian physics or biological gene regulatory networks, are human-driven simplifications of complex phenomena that serve as surrogates for the countless experiments that validated the models. Recently, machine learning has been able to overcome the inaccuracies of approximate modeling by directly learning the entire set of nonlinear interactions from data. However, without any predetermined structure from the scientific basis behind the problem, machine learning approaches are flexible but data-expensive, requiring large databases of homogeneous labeled training data. A central challenge is reconciling data that is at odds with simplified models without requiring "big data". In this work we develop a new methodology, universal differential equations (UDEs), which augments scientific models with machinelearnable structures for scientifically-based learning. We show howUDEs can be utilized to discover previously unknown governing equations, accurately extrapolate beyond the original data, and accelerate model simulation, all in a time and data-efficient manner. This advance is coupled with open-source software that allows for training UDEs which incorporate physical constraints, delayed interactions, implicitly-defined events, and intrinsic stochasticity in the model. Our examples show how a diverse set of computationallydifficult modeling issues across scientific disciplines, from automatically discovering biological mechanisms to accelerating climate simulations by 15,000x, can be handled by training UDEs.

Background Emerging evidence indicates that tumor cells release a large amount of exosomes loaded with cargos during tumorigenesis. Exosome secretion is a multi-step process regulated by certain related molecules. Long non-coding RNAs (lncRNAs) play an important role in hepatocellular carcinoma (HCC) progression. However, the role of lncRNA HOTAIR in regulating exosome secretion in HCC cells remains unclear. Methods We analyzed the relationship between HOTAIR expression and exosome secretion-related genes using gene set enrichment analysis (GSEA). Nanoparticle tracking analysis was performed to validate the effect of HOTAIR on exosome secretion. The transport of multivesicular bodies (MVBs) after overexpression of HOTAIR was detected by transmission electron microscopy and confocal microscopy analysis of cluster determinant 63 (CD63) with synaptosome associated protein 23 (SNAP23). The mechanism of HOTAIR’s regulation of Ras-related protein Rab-35 (RAB35), vesicle associated membrane protein 3 (VAMP3), and SNAP23 was assessed using confocal co-localization analysis, phosphorylation assays, and rescue experiments. Results We found an enrichment of exosome secretion-related genes in the HOTAIR high expression group. HOTAIR promoted the release of exosomes by inducing MVB transport to the plasma membrane. HOTAIR regulated RAB35 expression and localization, which controlled the docking process. Moreover, HOTAIR facilitated the final step of fusion by influencing VAMP3 and SNAP23 colocalization. In addition, we validated that HOTAIR induced the phosphorylation of SNAP23 via mammalian target of rapamycin (mTOR) signaling. Conclusion Our study demonstrated a novel function of lncRNA HOTAIR in promoting exosome secretion from HCC cells and provided a new understanding of lncRNAs in tumor cell biology. Electronic supplementary material The online version of this article (10.1186/s12943-019-0990-6) contains supplementary material, which is available to authorized users.

BackgroundHypoxia-inducible factor 1α (HIF-1α) is essential in hepatocellular carcinoma (HCC) glycolysis and progression. Yes-associated protein (YAP) is a powerful regulator and is overexpressed in many cancers, including HCC. The regulatory mechanism of YAP and HIF-1α in HCC glycolysis is unknown.MethodsWe detected YAP expression in 54 matched HCC tissues and the adjacent noncancerous tissues. The relationship between YAP mRNA expression and that of HIF-1α was analyzed using The Cancer Genome Atlas HCC tissue data. We cultured HepG2 and Huh7 HCC cells under normoxic (20% O2) and hypoxic (1% O2) conditions, and measured the lactate and glucose levels, migration and invasive capability, and the molecular mechanism of HCC cell glycolysis and progression.ResultsIn this study, we detected YAP expression in 54 matched HCC tissues and the adjacent noncancerous tissues. We observed that hypoxia-induced YAP activation is crucial for accelerating HCC cell glycolysis. Hypoxia inhibited the Hippo signaling pathway and promoted YAP nuclear localization, and decreased phosphorylated YAP expression in HCC cells. YAP knockdown inhibited HCC cell glycolysis under hypoxic. Mechanistically, hypoxic stress in the HCC cells promoted YAP binding to HIF-1α in the nucleus and sustained HIF-1α protein stability to bind to PKM2 gene and directly activates PKM2 transcription to accelerate glycolysis.ConclusionsOur findings describe a new regulatory mechanism of hypoxia-mediated HCC metabolism, and YAP might be a promising therapeutic target in HCC.Electronic supplementary materialThe online version of this article (10.1186/s13046-018-0892-2) contains supplementary material, which is available to authorized users.

In contrast to normal cells, which use the aerobic oxidation of glucose as their main energy production method, cancer cells prefer to use anaerobic glycolysis to maintain their growth and survival, even under normoxic conditions. Such tumor cell metabolic reprogramming is regulated by factors such as hypoxia and the tumor microenvironment. In addition, dysregulation of certain signaling pathways also contributes to cancer metabolic reprogramming. Among them, the Hippo signaling pathway is a highly conserved tumor suppressor pathway. The core oncosuppressive kinase cascade of Hippo pathway inhibits the nuclear transcriptional co-activators YAP and TAZ, which are the downstream effectors of Hippo pathway and oncogenic factors in many solid cancers. YAP/TAZ function as key nodes of multiple signaling pathways and play multiple regulatory roles in cancer cells. However, their roles in cancer metabolic reprograming are less clear. In the present review, we examine progress in research into the regulatory mechanisms of YAP/TAZ on glucose metabolism, fatty acid metabolism, mevalonate metabolism, and glutamine metabolism in cancer cells. Determining the roles of YAP/TAZ in tumor energy metabolism, particularly in relation to the tumor microenvironment, will provide new strategies and targets for the selective therapy of metabolism-related cancers.

BackgroundAutophagy is a dynamic physiological process that can generate energy and nutrients for cell survival during stress. Autophagy can regulate the migration and invasive ability in cancer cells. However, the connection between autophagy and metabolism is unclear. Monocarboxylate transporter 1 (MCT1) plays an important role in lactic acid transport and H+ clearance in cancer cells, and Wnt/β-catenin signaling can increase cancer cell glycolysis. We investigated whether autophagy promotes glycolysis in hepatocellular carcinoma (HCC) cells by activating the Wnt/β-catenin signaling pathway, accompanied by MCT1 upregulation.MethodsAutophagic activity was evaluated using western blotting, immunoblotting, and transmission electron microscopy. The underlying mechanisms of autophagy activation on HCC cell glycolysis were studied via western blotting, and Transwell, lactate, and glucose assays. MCT1 expression was detected using quantitative reverse transcription–PCR (real-time PCR), western blotting, and immunostaining of HCC tissues and the paired adjacent tissues.ResultsAutophagy promoted HCC cell glycolysis accompanied by MCT1 upregulation. Wnt/β-catenin signaling pathway activation mediated the effect of autophagy on HCC cell glycolysis. β-Catenin downregulation inhibited the autophagy-induced glycolysis in HCC cells, and reduced MCT1 expression in the HCC cells. MCT1 was highly expressed in HCC tissues, and high MCT1 expression correlated positively with the expression of microtubule-associated protein light chain 3 (LC3).ConclusionActivation of autophagy can promote metastasis and glycolysis in HCC cells, and autophagy induces MCT1 expression by activating Wnt/β-catenin signaling. Our study describes the connection between autophagy and glucose metabolism in HCC cells and may provide a potential therapeutic target for HCC treatment.Electronic supplementary materialThe online version of this article (10.1186/s13046-018-0673-y) contains supplementary material, which is available to authorized users.

No abstract

Long noncoding RNAs (lncRNAs) has been acknowledged in tumorigenesis gradually because of the great importance in different cancers. LncRNA nuclear enriched abundant transcript 1 (NEAT1) is a novel lncRNA and has been reported to promote multiple cancer progression. However, the biological roles of NEAT1 in hepatocellular carcinoma (HCC) is not cleared nowadays. In the present research, the level of NEAT1 was found to be upregulated in HCC by The Cancer Genome Atlas. In addition, NEAT1 expression is negatively correlated with the survival rate in HCC. Further investigation revealed that NEAT1 upregulation inhibited sorafenib efficacy and promoted autophagy. We found that NEAT1 could be a sponge for microRNA‐204 (miR‐204) and inhibits its level to upregulate ATG3 expression. In addition to the above, we demonstrated that miR‐204 mimics also attenuated tumor autophagy. And rescue assays demonstrated that NEAT1 promotes HCC autophagy through modulating miR‐204/ATG3 pathway. Collectively, this study first demonstrated that a novel NEAT1/miR‐204/ATG3 signaling regulates HCC progression.