Search

Myostatin (MSTN) is an important negative regulator of muscle growth and development. In this study, we performed comparatively the proteomics and phosphorylated proteomics analyses of gluteus tissues from MSTN+/- Mongolian cattle (MG.MSTN+/-), wild type Mongolian cattle (MG.WT) and Belgian Blue cattle (BBC) using a shotgun-based tandem mass tag (TMT) 10-plex labeling method to investigate the regulation mechanism of MSTN on the growth and development of bovine skeletal muscle. A total of 1950 proteins and 713 phosphorylated peptides corresponding to 268 phosphorylated proteins were identified in MG.MSTN+/-, MG.WT and BBC. Among them, 320 differentially expressed proteins and 54 differentially altered phosphorylated peptides corresponding to 37 phosphorylated proteins were identified in MG.MSTN+/- compared with MG.WT; and 376 differentially expressed proteins and 87 differentially altered phosphorylated peptides corresponding to 46 phosphorylated proteins were identified in BBC compared with MG.WT; compared with MG.MSTN+/-, 350 differentially expressed proteins, and 6 differentially altered phosphorylated peptides corresponding to 6 phosphorylated proteins were identified in BBC. Bioinformatics analysis showed that knockdown of MSTN gene increased the expression of extracellular matrix and ribosome-related proteins, induced activation of focal adhesion, PI3K-AKT, and Ribosomal pathways. The results of proteomic analysis were verified by muscle tissue Westernblot test and in vitro MSTN gene knockdown test, and it was found that knockdown MSTN gene expression could promote the proliferation and myogenic differentiation of bovine skeletal muscle satellite cells (BSMSCs). At the same time, Co-Immunoprecipitation (CO-IP) assay showed that MSTN gene interacted with extracellular matrix related protein type I collagen α 1 (COL1A1), and knocking down the expression of COL1A1 could inhibit the activity of adhesion, PI3K-AKT and ribosome pathway, thus inhibit BSMSCs proliferation. These results suggest that the MSTN gene regulates focal adhesion, PI3K-AKT, and Ribosomal pathway through the COL1A1 gene. In general, this study provides new insights into the regulatory mechanism of MSTN involved in muscle growth and development.

Myogenesis, the process of skeletal muscle formation, is a highly coordinated multistep biological process. Accumulating evidence suggests that long non‐coding RNAs (lncRNAs) are emerging as a gatekeeper in myogenesis. Up to now, most studies on muscle development‐related lncRNAs are mainly focussed on humans and mice. In this study, a novel muscle highly expressed lncRNA, named lnc23, localized in nucleus, was found differentially expressed in different stages of embryonic development and myogenic differentiation. The knockdown and over‐expression experiments showed that lnc23 positively regulated the myogenic differentiation of bovine skeletal muscle satellite cells. Then, TMT 10‐plex labelling quantitative proteomics was performed to screen the potentially regulatory proteins of lnc23. Results indicated that lnc23 was involved in the key processes of myogenic differentiation such as cell fusion, further demonstrated that down‐regulation of lnc23 may inhibit myogenic differentiation by reducing signal transduction and cell fusion among cells. Furthermore, RNA pulldown/LC‐MS and RIP experiment illustrated that PFN1 was a binding protein of lnc23. Further, we also found that lnc23 positively regulated the protein expression of RhoA and Rac1, and PFN1 may negatively regulate myogenic differentiation and the expression of its interacting proteins RhoA and Rac1. Hence, we support that lnc23 may reduce the inhibiting effect of PFN1 on RhoA and Rac1 by binding to PFN1, thereby promoting myogenic differentiation. In short, the novel identified lnc23 promotes myogenesis of bovine skeletal muscle satellite cells via PFN1‐RhoA/Rac1.

Metric learning learns a distance metric from data and has significantly improved the classification accuracy of distance-based classifiers such as k-nearest neighbors. However, metric learning has rarely been applied to categorical data, which are prevalent in health and social sciences, but inherently difficult to classify due to high feature ambiguity and small sample size. More specifically, ambiguity arises as the boundaries between ordinal or nominal levels are not always sharply defined. In this paper, we mitigate the impact of feature ambiguity by considering the worst-case perturbation of each instance and propose to learn the Mahalanobis distance through adversarial training. The geometric interpretation shows that our method dynamically divides the instance space into three regions and exploits the information on the “adversarially vulnerable” region. This information, which has not been considered in previous methods, makes our method more suitable than them for small-sized data. Moreover, we establish the generalization bound for a general form of adversarial training. It suggests that the sample complexity rate remains at the same order as that of standard training only if the Mahalanobis distance is regularized with the elementwise 1-norm. Experiments on ordinal and mixed ordinal-and-nominal datasets demonstrate the effectiveness of the proposed method when encountering the problems of high feature ambiguity and small sample size.

Skeletal muscle, the most abundant and plasticity tissue in mammals, is essential for various functions such as movement, breathing, maintaining posture and metabolism. Myogenesis is a complex and precise process, which is regulated by the sequential expression of multiple transcription factors, and accumulating evidence have confirmed that multiple lncRNAs are involved in muscle development as the important transcriptional regulator. In this study, a novel lncRNA, named lnc403 was obtained, with a full-length 2689 bp, which had poor coding ability and was mainly expressed in the nucleus of myoblasts and myotubes. The expression of lnc403 was significantly different in the proliferation and differentiation stages of muscle cells. Then we successfully constructed lnc403 loss/gain-function cell models by transfecting silnc403 and pCDNA3.1-EGFP-lnc403 into satellite cells, respectively; and found that lnc403 inhibited skeletal muscle satellite cell differentiation but had no significant effect on cell proliferation, either in the case of lnc403 knockdown or overexpression. In order to further screen the target factors regulated by lncRNA in the process of myogenic differentiation, the RNA-pull down, mass spectrometry and bioinformatics analysis were performed. The results showed that lnc403 negatively regulated the expression of the adjacent gene Myf6 and positively regulated interaction proteins KRAS expression. The above results indicate that lnc403 affects skeletal muscle cell differentiation by affecting the expression of nearby genes and interacting proteins, implying lnc403 might participate in the bovine myoblasts differentiation through multi-pathway network regulation mode. This study provides a new perspective for further understanding of the regulation mechanism of lncRNAs on bovine myogenic process.

Unlike the white-box counterparts that are widely studied and readily accessible, adversarial examples in black-box settings are generally more Herculean on account of the difficulty of estimating gradients. Many methods achieve the task by issuing numerous queries to target classification systems, which makes the whole procedure costly and suspicious to the systems. In this paper, we aim at reducing the query complexity of black-box attacks in this category. We propose to exploit gradients of a few reference models which arguably span some promising search subspaces. Experimental results show that, in comparison with the state-of-the-arts, our method can gain up to 2x and 4x reductions in the requisite mean and medium numbers of queries with much lower failure rates even if the reference models are trained on a small and inadequate dataset disjoint to the one for training the victim model. Code and models for reproducing our results will be made publicly available., 10 pages + 3 pages supplementary material

For network architecture search (NAS), it is crucial but challenging to simultaneously guarantee both effectiveness and efficiency. Towards achieving this goal, we develop a differentiable NAS solution, where the search space includes arbitrary feed-forward network consisting of the predefined number of connections. Benefiting from a proposed ensemble Gumbel-Softmax estimator, our method optimizes both the architecture of a deep network and its parameters in the same round of backward propagation, yielding an end-to-end mechanism of searching network architectures. Extensive experiments on a variety of popular datasets strongly evidence that our method is capable of discovering high-performance architectures, while guaranteeing the requisite efficiency during searching.

Traditional clustering methods often perform clustering with low-level indiscriminative representations and ignore relationships between patterns, resulting in slight achievements in the era of deep learning. To handle this problem, we develop Deep Discriminative Clustering (DDC) that models the clustering task by investigating relationships between patterns with a deep neural network. Technically, a global constraint is introduced to adaptively estimate the relationships, and a local constraint is developed to endow the network with the capability of learning high-level discriminative representations. By iteratively training the network and estimating the relationships in a mini-batch manner, DDC theoretically converges and the trained network enables to generate a group of discriminative representations that can be treated as clustering centers for straightway clustering. Extensive experiments strongly demonstrate that DDC outperforms current methods on eight image, text and audio datasets concurrently.

Deep neural networks (DNNs) are computationally/memory-intensive and vulnerable to adversarial attacks, making them prohibitive in some real-world applications. By converting dense models into sparse ones, pruning appears to be a promising solution to reducing the computation/memory cost. This paper studies classification models, especially DNN-based ones, to demonstrate that there exists intrinsic relationships between their sparsity and adversarial robustness. Our analyses reveal, both theoretically and empirically, that nonlinear DNN-based classifiers behave differently under $l_2$ attacks from some linear ones. We further demonstrate that an appropriately higher model sparsity implies better robustness of nonlinear DNNs, whereas over-sparsified models can be more difficult to resist adversarial examples., l1 regularization on weights --> l1 regularization on activations

Despite the efficacy on a variety of computer vision tasks, deep neural networks (DNNs) are vulnerable to adversarial attacks, limiting their applications in security-critical systems. Recent works have shown the possibility of generating imperceptibly perturbed image inputs (a.k.a., adversarial examples) to fool well-trained DNN classifiers into making arbitrary predictions. To address this problem, we propose a training recipe named "deep defense". Our core idea is to integrate an adversarial perturbation-based regularizer into the classification objective, such that the obtained models learn to resist potential attacks, directly and precisely. The whole optimization problem is solved just like training a recursive network. Experimental results demonstrate that our method outperforms training with adversarial/Parseval regularizations by large margins on various datasets (including MNIST, CIFAR-10 and ImageNet) and different DNN architectures. Code and models for reproducing our results are available at https://github.com/ZiangYan/deepdefense.pytorch, Comment: Accepted by NeurIPS 2018