Search

The phenotypic transformation of vascular smooth muscle cells (VSMCs) plays a key role in the pathological process of atherosclerosis (AS), and TRPM7 is involved in this process. In this study, we verified whether circRNAs participate in the phenotypic transformation of VSMCs by regulating TRPM7 in AS. The RNA-sequencing data of atherosclerosis were downloaded and analysed from the GEO database. Only hsa_circ_0021155 related to TRPM7 was differentially expressed in AS. circRNA distribution and expression were observed via FISH and PCR. CCK8, scratch test and Transwell assay were used to observe the proliferation and migration of cells. Western blot was performed to examine changes in α-actin, calponin, SMMHC and TRPM7 proteins. The expression of hsa_circ_0021155 against has-miR-4459/miR-3689c was verified via PCR. The ceRNA relationship of TPRM7-miR4459-circ0021155 was verified via dual luciferase assay, and the effects of miR4459 mimic/inhibitor on the proliferation of cells were further observed. The expression of hsa_circ_0021155 and OX-LDL was increased in VSMCs. hsa_circ_0021155 promoted the expression of TRPM7 and inhibited the protein expression of α-actin, calponin and SMMHC. In addition, it promoted the proliferation and migration of cells and inhibited the expression of miR-3689c and miR-4459 but did not affect miR-4756-5p. The dual luciferase assay showed that circ0021155-miR4459-TRPM7 mRNA was highly compatible and could be mutually regulated by a ceRNA network. In conclusion, hsa_circ_0021155 regulates the proliferation, migration and phenotype transformation of VSMCs induced by OX-LDL via the miR-4459/TRPM7 axis. hsa_circ_0021155 and TRPM7 may offer novel therapeutic targets for atherosclerosis.

BackgroundIn diverse eukaryotes, DNA methylation is deposited within the coding regions of many genes, termed gene body methylation (GbM). Whereas the role of DNA methylation on the silencing of transposons and repetitive DNA is well understood, gene body methylation is not associated with transcriptional repression, and its biological importance has remained unclear.ResultsWe report a newly discovered type of GbM in plants, which is under constitutive addition and removal by dynamic methylation modifiers in all cells, including the germline. Methylation at Dynamic GbM genes is removed by the DRDD demethylation pathway, and added by an unknown source of de novo methylation, most likely the maintenance methyltransferase MET1. We show that the Dynamic GbM state is present at homologous genes across divergent lineages (>100 million years), indicating evolutionary conservation. Furthermore, we demonstrate that Dynamic GbM is tightly associated with the presence of a promoter/regulatory chromatin state within the gene body, in contrast to other gene body methylated genes. Lastly, we find Dynamic GbM is associated with enhanced gene expression plasticity across development and diverse physiological conditions, whereas stably methylated GbM genes exhibit reduced plasticity.ConclusionsThis study proposes a new model for GbM in regulating gene expression plasticity, including a newly discovered type of GbM in which increased gene expression plasticity is associated with the activity of DNA methylation writers and erasers and the enrichment of a regulatory chromatin state. We propose that Dynamic GbM is a hallmark of genes that can navigate a greater landscape of transcriptional states by maintaining regulatory flexibility.

In recent years, convolutional neural networks (CNNs) have been successfully applied in the field of image processing, and have been deployed to a variety of artificial intelligence systems. However, such neural models are still considered to be “black box” for most tasks. Two of fundamental issues underlying this problem are as follows: 1. What type of knowledge learned by a neural network in a task was not understandable and predictable? 2. The decision made by a neural model was generally not evaluable. Like neural coding in the brain, some neurons only participated in encoding a particular task. Inspired by this, in this paper, we propose a method to modify traditional CNN models into understandable CNNs, to clarify the information coding in high conv-layers of CNNs and further evaluate the decisions made by a neural model. In our understandable CNN models, each neural node (feature map) in a selected conv-layer was assigned to participate in encoding only one class in a classification task. Our models use the same training data as ordinary models without the need for additional annotations for supervision. We applied our method to the ResNet and DenseNet models. The experiments showed that new models can learn the information coding mode that we expected in an image-recognition task, and, using the pre-assigned coding mode, we can interpret why a neural model makes a right or wrong decision, which decisions are credible, and which are not.

Pentatricopeptide repeat (PPR) proteins were identified as site-specific recognition factors for RNA editing in plant mitochondria and plastids. In this study, we characterized maize (Zea mays) kernel mutant defective kernel 53 (dek53), which has an embryo lethal and collapsed endosperm phenotype. Dek53 encodes an E-subgroup PPR protein, which possesses a short PLS repeat region of only seven repeats. Subcellular localization analysis indicated that DEK53 is localized in the mitochondrion. Strand- and transcript-specific RNA-seq analysis showed that the dek53 mutation affected C-to-U RNA editing at more than 60 mitochondrial C targets. Biochemical analysis of mitochondrial protein complexes revealed a significant reduction in the assembly of mitochondrial complex III in dek53. Transmission electron microscopic examination showed severe morphological defects of mitochondria in dek53 endosperm cells. In addition, yeast two-hybrid and luciferase complementation imaging assays indicated that DEK53 can interact with the mitochondrion-targeted non-PPR RNA editing factor ZmMORF1, suggesting that DEK53 might be a functional component of the organellar RNA editosome.

Recently, the deep neural network has achieved considerable success in the field of machine learning. However, for most tasks, the current neural network models are still considered as the “black-box” structure, essentially because the knowledge a neural network learns from data is often unpredictable and unexplainable. Similar to neural coding in the brain, one neuron may be simultaneously involved in encoding one or even several tasks. Information may be transmitted between neurons along a particular neural-path (neural circuit) to end neurons that encode a specific decision. Inspired by this, we aim to explore the reason for the performance of a neural model based on a neural-path. First, for a trained neural model, we quantify the neural-path in which the role of a neuron was assumed to control the amount of information that can be passed through; Second, we define a Euclidean distance (ED) for every two neural-paths, and by analyzing the EDs between two classes of a neural classifier, we explain the ease of prediction for some classes, whereas others are not. We performed extensive experiments for architectures of ResNet and DenseNet models on several benchmark datasets, and determined that the shorter the distance between the neural-paths of two classes, the easier it is to make mistakes. Finally, we proposed a method for controlling the formation of a “neural-path” to build a partially understandable neural model. For the new ResNet model, each feature map in redirect layers was assigned to participate in encoding only one class. The feasibility of the method is also verified through experiments.

Edge detection, a basic task in the field of computer vision, is an important preprocessing operation for the recognition and understanding of a visual scene. In conventional models, the edge image generated is ambiguous, and the edge lines are also very thick, which typically necessitates the use of non-maximum suppression (NMS) and morphological thinning operations to generate clear and thin edge images. In this paper, we aim to propose a one-stage neural network model that can generate high-quality edge images without postprocessing. The proposed model adopts a classic encoder-decoder framework in which a pre-trained neural model is used as the encoder and a multi-feature-fusion mechanism that merges the features of each level with each other functions as a learnable decoder. Further, we propose a new loss function that addresses the pixel-level imbalance in the edge image by suppressing the false positive (FP) edge information near the true positive (TP) edge and the false negative (FN) non-edge. The results of experiments conducted on several benchmark datasets indicate that the proposed method achieves state-of-the-art results without using NMS and morphological thinning operations., Comment: 15,6,41

In recent years, the convolution neural network (CNN) has been successfully applied in numerous fields as a machine learning model. However, such neural models are still considered to be “black box” for most tasks. The fundamental issue underlying this problem is that the information and knowledge learned by a neural network during the training process are unknown and unpredictable. In this study, we attempted to design a partially understandable neural network through semantic embedding. Firstly, we selected several understandable feature extraction operators as expected information. Secondly, we embedded these operators into the hierarchical layers of a neural network at the beginning of its training process. Finally, these embedded operators were only involved in forward calculation and remained unchanged in the error back-propagation of the training process. We applied our method to the ResNet and DenseNet models in image classification tasks. In the experiments, our new models achieved almost the same performance as the original ones, but decreased performance was exhibited when shutting the embedded parts, and the features in the embedded parts were understandable. The experiments verified that the embedded parts not only contributed to the classification tasks, but also caused the neural network model to be partially understandable.

Deep neural networks (DNN) have achieved great success in machine learning due to their powerful ability to learn and present knowledge. However, models of such DNN often have massive trainable parameters, which lead to big resource burden in practice. As a result, reducing the amount of parameters and preserving its competitive performance are always critical tasks in the field of DNN. In this paper, we focused on one type of convolution neural network that has many repeated or same-structure convolutional layers. Residual net and its variants are widely used, making the deeper model easy to train. One type block of such a model contains two convolutional layers, and each block commonly has two trainable parameter layers. However, we used only one layer of trainable parameters in the block, which means that the two convolutional layers in one block shared one layer of trainable parameters. We performed extensive experiments for different architectures of the Residual Net with trainable parameter sharing on the CIFAR-10, CIFAR-100, and ImageNet datasets. We found that the model with trainable parameter sharing can obtain fewer errors on the training datasets and had a very close recognition accuracy (within 0.5%), compared to the original models. The parameters of the new model were reduced by more than 1/3 of the total of the original.

Photosensitive Al2O3-resin slurries with high solid loading, low viscosity used for stereolithography based additive manufacturing were prepared in this paper. The dispersion behavior and rheological behavior of the Al2O3-resin slurries were studied by rheology observation and sedimentation tests. The dispersant type, concentration and solid loading had significant effects on the rheological behavior and stability of the photosensitive Al2O3-resin slurries. A long term stability and homogeneity slurry was obtained when the dispersant and concentration were KOS110 and 5 wt%, respectively. The Al2O3 slurry prepared with a high solid loading up to 60 vol%, low viscosity of 15.4 Pa s at 200 s−1 was chosen for stereolithography based additive manufacturing.

In this paper, MoSi2, MoSi2-20 vol% (ZrB2-20 vol% SiC) and MoSi2-40 vol% (ZrB2-20 vol% SiC) ceramics were prepared using pressureless sintering. The oxidation behaviors of these MoSi2-(ZrB2-SiC) ceramics were investigated at 1600 °C for different soaking time of 60, 180 and 300 min, respectively. The oxidation behaviors of the MoSi2-(ZrB2-SiC) ceramics were studied through weight change test, oxide layer thickness measurement, and microstructure analysis. Further investigation of the oxidation behaviors of the MoSi2-(ZrB2-SiC) ceramics was conducted at a higher temperature of 1800 °C for 10 min. The microstructure evolution of the ceramics was also analyzed. It was finally found that the oxidation resistance of MoSi2 was improved by adding ZrB2-SiC additives, and the MoSi2-20 vol% (ZrB2-20 vol% SiC) ceramic exhibited the optimal oxidation resistance behavior at elevated temperatures. From this study, it is believe that it can give some fundamental understanding and promote the engineering application of MoSi2-based ceramics at high temperatures.

Maize (Zea mays) is a leading cereal crop in the world. The maize kernel is the storage organ and the harvest portion of this crop and is closely related to its yield and quality. The development of maize kernel is initiated by the double fertilization event, leading to the formation of a diploid embryo and a triploid endosperm. The embryo and endosperm are then undergone independent developmental programs, resulting in a mature maize kernel which is comprised of a persistent endosperm, a large embryo, and a maternal pericarp. Due to the well-characterized morphogenesis and powerful genetics, maize kernel has long been an excellent model for the study of cereal kernel development. In recent years, with the release of the maize reference genome and the development of new genomic technologies, there has been an explosive expansion of new knowledge for maize kernel development. In this review, we overviewed recent progress in the study of maize kernel development, with an emphasis on genetic mapping of kernel traits, transcriptome analysis during kernel development, functional gene cloning of kernel mutants, and genetic engineering of kernel traits.

Recent breakthroughs in transcriptome analysis and gene characterization have provided valuable resources and information about the maize endosperm developmental program. The high temporal-resolution transcriptome analysis has yielded unprecedented access to information about the genetic control of seed development. Detailed spatial transcriptome analysis using laser-capture microdissection has revealed the expression patterns of specific populations of genes in the four major endosperm compartments: the basal endosperm transfer layer (BETL), aleurone layer (AL), starchy endosperm (SE), and embryo-surrounding region (ESR). Although the overall picture of the transcriptional regulatory network of endosperm development remains fragmentary, there have been some exciting advances, such as the identification of OPAQUE11 (O11) as a central hub of the maize endosperm regulatory network connecting endosperm development, nutrient metabolism, and stress responses, and the discovery that the endosperm adjacent to scutellum (EAS) serves as a dynamic interface for endosperm-embryo crosstalk. In addition, several genes that function in BETL development, AL differentiation, and the endosperm cell cycle have been identified, such as ZmSWEET4c, Thk1, and Dek15, respectively. Here, we focus on current advances in understanding the molecular factors involved in BETL, AL, SE, ESR, and EAS development, including the specific transcriptional regulatory networks that function in each compartment during endosperm development.

Methods like low-temperature liquid nitrogen gas adsorption and scanning electron microscope were used in the study to investigate pore and fissure structures of three coal samples of different metamorphic degrees. And the findings show that the volume of nitrogen adsorbed by coal samples of metamorphic degree differs noticeably, that surface fissures and internal pores of samples stay various. With metamorphic degree increasing, the specific surface area of coal body decrease slowly at first and then increase sharply. But the pore diameter of samples increases initially and decreases afterwards, the proportion of micropore diameter rises, and fissures develop, then the number of large and medium pores reduces, while transition holes increases. Combination of low-temperature liquid nitrogen gas adsorption and scanning electron microscope can better characterize the pore and fissure structures of coal, which provides the study of coal gas adsorption-desorption and diffusion and coal mine safety production with a theoretical foundation.

Mitochondrial group II introns require the participation of numerous nucleus-encoded general and specific factors to achieve efficient splicing in vivo. Pentatricopeptide repeat (PPR) proteins have been implicated in assisting group II intron splicing. Here, we identified and characterized a new maize seed mutant, defective kernel 37 (dek37), which has significantly delayed endosperm and embryo development. Dek37 encodes a classic P-type PPR protein that targets mitochondria. The dek37 mutation causes no detectable DEK37 protein in mutant seeds. Mitochondrial transcripts analysis indicated that dek37 mutation decreases splicing efficiency of mitochondrial nad2 intron 1, leading to reduced assembly and NADH dehydrogenase activity of complex I. Transmission Electron Microscopy (TEM) revealed severe morphological defects of mitochondria in dek37. Transcriptome analysis of dek37 endosperm indicated enhanced expression in the alternative respiratory pathway and extensive differentially expressed genes related to mitochondrial function. These results indicated that Dek37 is involved in cis-splicing of mitochondrial nad2 intron 1 and is required for complex I assembly, mitochondrial function, and seed development in maize.

Background/Aims: Cyclin D1 (CCND1) is frequently overexpressed in malignant gliomas. We have previously shown ectopic overexpression of CCND1 in human malignant gliomas cell lines. Methods: Quantitative reverse transcriptase PCR (qRT-PCR) and Western Blot (WB) was performed to investigate the expression of CCND1 in glioma tissues and cell lines. The biological function of CCND1 was also investigated through knockdown and overexpression of BCYRN1 in vitro. Results: Here we reported that CCND1 expression was positively associated with the pathological grade and proliferative activity of astrocytomas, as the lowest expression was found in normal brain tissue (N = 3) whereas the highest expression was in high-grade glioma tissue (N = 25). Additionally, we found that the expression level of CCND1 was associated with IC50 values in malignant glioma cell lines. Forced inhibition of CCND1 increased temozolomide efficacy in U251 and SHG-44 cells. After CCND1 overexpression, the temozolomide efficacy decreased in U251 and SHG-44 cells. Colony survival assay and apoptosis analysis confirmed that CCND1 inhibition renders cells more sensitive to temozolomide treatment and temozolomide-induced apoptosis in U251 and SHG-44 cells. Inhibition of P-gp (MDR1) by Tariquidar overcomes the effects of CCND1 overexpression on inhibiting temozolomide-induced apoptosis. Inhibition of CCND1 inhibited cell growth in vitro and in vivo significantly more effectively after temozolomide treatments than single temozolomide treatments. Finally, inhibition of CCND1 in glioma cells reduced tumor volume in a murine model. Conclusion: Taken together, these data indicate that CCND1 overexpression upregulate P-gp and induces chemoresistance in human malignant gliomas cells and that inhibition of CCND1 may be an effective means of overcoming CCND1 associated chemoresistance in human malignant glioma cells.

Sphingosine kinase 1 (Sphk1), a key enzyme responsible for phosphorylating sphingosine into sphingosine1-phosphate (S1P), plays an important role in mediating post-stroke neuroinflammation. However, the pathway and mechanism of the Sphk1-mediated inflammatory response remains unknown. In this study, we found that suppression of Sphk1 decreased IL17 production and relieved neuronal damage induced by microglia in cerebral ischemia reperfusion (IR) or in an in vitro oxygen-glucose deprivation reperfusion (OGDR) system. Inhibition of Sphk1 with an inhibitor or siRNA decreased tumor necrosis factor receptor-associated factor 2 (TRAF2) and nuclear factor-kappa B (NF-κB) sequentially in microglia in response to IR or OGDR. Moreover, we also found that after suppression of TRAF2 or NF-κB by siRNA in microglia, reductions in the downstream molecules NF-κB and IL-17 and in neuronal apoptosis were observed in response to OGDR. Taken together, we hypothesize that Sphk1, TRAF2 and NF-κB form an axis that leads to increased IL-17 and neuronal apoptosis. This axis may be a potential therapeutic target to control neuroinflammation in brain IR.